Synthesis of [15,15,15-2H3]-Dihydroartemisinic Acid and Isotope Studies Support a Mixed Mechanism in the Endoperoxide Formation to Artemisinin

Article abstract of DOI:10.1021/acs.jnatprod.1c00246

Artemisinin is the plant natural product used to treat malaria. The endoperoxide bridge of artemisinin confers its antiparasitic properties. Dihydroartemisinic acid is the biosynthetic precursor of artemisinin that was previously shown to nonenzymatically undergo endoperoxide formation to yield artemisinin. This report discloses the synthesis of [15,15,15-²H₃]-dihydroartemisinic acid and its use to determine the mechanism of endoperoxide formation. Several new observations were made: (i) Ultraviolet-C (UV-C) radiation initially accelerates artemisinin formation and subsequently promotes homolytic cleavage of the O-O bond and rearrangement of artemisinin to a different product, and (ii) dideuterated and trideuterated dihydroartemisinic acid isotopologues at C3 and C15 converted to artemisinin at a slower rate compared to nondeuterated dihydroartemisinic acid, revealing a kinetic isotope effect in the initial ene reaction toward endoperoxide formation (k_H/k_D～ 2-3). (iii) The rate of conversion from dihydroartemisinic acid to artemisinin increased with the amount of dihydroartemisinic acid, suggesting an intermolecular interaction to promote endoperoxide formation, and (iv)¹⁸O₂-labeling showed incorporation of three and four oxygen atoms from molecular oxygen into the endoperoxide bridge of artemisinin. These results reveal new insights toward understanding the mechanism of endoperoxide formation in artemisinin biosynthesis.

Full text of DOI:10.1021/acs.jnatprod.1c00246

pubs.acs.org/jnp
Article
Synthesis of [15,15,15‑2H3]‑Dihydroartemisinic Acid and Isotope
Studies Support a Mixed Mechanism in the Endoperoxide Formation
to Artemisinin
Kaitlyn Varela, Hadi D. Arman, and Francis K. Yoshimoto*
Cite This: J. Nat. Prod. 2021, 84, 1967−1984
Read Online
sı
*
Metrics & More
Article Recommendations
Supporting Information
ACCESS
ABSTRACT: Artemisinin is the plant natural product used to
treat malaria. The endoperoxide bridge of artemisinin confers its
antiparasitic properties. Dihydroartemisinic acid is the biosynthetic
precursor of artemisinin that was previously shown to nonenzy-
matically undergo endoperoxide formation to yield artemisinin.
This report discloses the synthesis of [15,15,15-2H3]-dihydroartemisinic
acid and its use to determine the mechanism of endoperoxide
formation. Several new observations were made: (i) Ultraviolet-C
(UV-C) radiation initially accelerates artemisinin formation and
subsequently promotes homolytic cleavage of the O−O bond and
rearrangement of artemisinin to a different product, and (ii) dide-
uterated and trideuterated dihydroartemisinic acid isotopologues at
C3 and C15 converted to artemisinin at a slower rate compared to nondeuterated dihydroartemisinic acid, revealing a kinetic isotope
effect in the initial ene reaction toward endoperoxide formation (kH/kD ∼ 2−3). (iii) The rate of conversion from dihydroartemisinic
acid to artemisinin increased with the amount of dihydroartemisinic acid, suggesting an intermolecular interaction to promote
endoperoxide formation, and (iv) 18O2-labeling showed incorporation of three and four oxygen atoms from molecular oxygen into the
endoperoxide bridge of artemisinin. These results reveal new insights toward understanding the mechanism of endoperoxide formation
in artemisinin biosynthesis.
shifts the signal upfield,15 and Brown and Sy showed the
changes of the deuterium signal in deuterium NMR spec-
troscopy12 by monitoring a deuterated DHAA probe, which
was injected through the roots of the dead plant, Artemisia
annua. Furthermore, studies employing mass spectrometry to
explore the endoperoxide forming mechanism are even more
scarce.16
In particular, we previously reported the rate of spontaneous
endoperoxide formation from [3,3-2H2]-dihydroartemisinic acid
(3,3-d2-DHAA, 3) to artemisinin.16 This cascade transformation
occurred when dried down samples of 3,3-d2-DHAA were left
in vials open to air. Subsequently, liquid chromatography
high-resolution mass spectrometry (LC-HRMS) was employed
to quantify this spontaneous process (Figure 2, 3 to 6, 2, and
10).16 The rate of 3,3-d2-artemisinin (6) formation from
3,3-d2-DHAA (3) was determined to be faster in the presence
of light. Additionally, an alternative mechanism was found to
ihydroartemisinic acid (DHAA)1 (Figure 1, 1) is the
Dbiosynthetic precursor of artemisinin (2), a plant
natural product used to treat malaria. The conversion from
DHAA to artemisinin is a complex process, which involves:
(i) incorporation of an oxygen molecule, (ii) C4−C5 bond
cleavage, (iii) incorporation of a second molecule of oxygen,
and (iv) polycyclization to form the endoperoxide (Figure 1).
The endoperoxide of artemisinin (2) is the pharmacophore
that triggers the production of reactive oxygen species and
leads to the death of the parasite (Plasmodium falciparum)
that causes malaria.2,3 Due to its importance in medicine,
many research groups have explored the development of new
approaches to enable cost-effective methods for artemisinin
production.4−8
Because the endoperoxide confers the biological activity of
artemisinin,9,10 there is significant interest in determining how
the endoperoxide bridge is formed in nature (Figure 1).11
Although there have been previously reported studies related
to exploring the endoperoxide formation in artemisinin bio-
synthesis through NMR spectroscopy,12−15 published reports
showing the raw data are rare. In particular, Roth and Acton
showed oxygen (18O) incorporation into dihydroartemisinic
acid hydroperoxide (see Figure 2, nondeuterated version of
4) using 13C NMR where the carbons bonded to 18O slightly
Received: March 23, 2021
Published: June 17, 2021
© 2021 American Chemical Society and
American Society of Pharmacognosy
https://doi.org/10.1021/acs.jnatprod.1c00246
J. Nat. Prod. 2021, 84, 1967−1984
1967

Journal of Natural Products
pubs.acs.org/jnp
Article
Figure 1. DHAA (1) is the biosynthetic precursor that converts to artemisinin (2). The structure of 2 shown in the figure has the carbons
numbered in the same manner as 1 for clarity. The structure of 2 in parentheses shows the official numbering of the atoms in artemisinin.
Figure 2. In the previous study,16 3,3-d2-DHAA (3) was converted to a mixture of 3-d1-artemisinin and 3,3-d2-artemisinin isotopologues (10 and 6)
through proposed path a, and nondeuterated artemisinin was converted through path b. In the presence of light, path a is dominant.
be operative with the detection of nondeuterated artemisinin
(2) from 3,3-d2-DHAA, which proceeded independent of
light (path b of Figure 2).
liquid chromatography high-resolution mass spectrometry
(LC-HRMS).
RESULTS AND DISCUSSION
The possibility of multiple mechanisms operating in the
formation of artemisinin from DHAA (Figure 2) provided the
inspiration to devise more experiments to elucidate the possible
pathways of how this complex cascade reaction occurs.
The detection of nondeuterated artemisinin (2) from 3,3-d2-
DHAA (3) suggested an alternative pathway to endoperoxide
formation involving a regioselective loss of the two C3 deuterium
atoms in the formation of artemisinin (path b, Figure 2).
Therefore, it was reasonable to synthesize [15,15,15-2H3]-
dihydroartemisinic acid (15,15,15-d3-DHAA, 11) to mecha-
nistically determine whether the C15 deuterium atoms would
also be lost in its conversion to artemisinin with and without
light (Figure 3, 11 to 16, 17, or 2). Moreover, a completely
trideuterated artemisinin isotopologue (14) could be potentially
useful as an internal standard to quantify the conversion of
nondeuterated DHAA to artemisinin (1 to 2).
■
(i) Synthesis of 15,15,15-d3-DHAA (11). In order to
deuterate the C15 position of DHAA (1), initial attempts
involved the regioselective oxidation of 1 to obtain a C15-
oxygenated intermediate (see Supporting Information, Part 12).
However, all attempted routes to directly oxidize the C15-allylic
methyl of the dehydrodecalin system had failed. Therefore,
the synthesis of 15,15,15-d3-DHAA (11) involved the cleavage
of the C4−C5 bond (Figure 4, 1 to 21) to enable the formation
of a methyl ketone intermediate, which could be oxidized at the
α-methyl position. The key deuterium incorporation step would
employ the reaction of an α,β-unsaturated carbonyl intermediate
(18) with LiAlD4 and AlCl3.16 The C15-oxygenated inter-
mediate would be derived from a Grubbs ring-closing metathesis
(RCM) reaction of a monocyclic diene precursor (19).
A Grubbs RCM has been previously employed to access the
cis-dehydrodecalin system of DHAA.17 Diene 19 would
be derived from a silyl enol ether precursor (20). The key
C15-oxygen would be incorporated through Rubottom oxidation
of silyl enol ether 20. The silyl enol ether would be accessed
from the ketoaldehyde precursor (21) through olefination at
C5 through a Wittig reaction. The ketoaldehyde intermediate
This study involved the synthesis of 15,15,15-d3-DHAA
(11) and its use to investigate the mechanism of conversion
of DHAA to artemisinin. Endoperoxide formation was tested
under various conditions. In addition, the use of 3,3-d2-DHAA
(3) was revisited under these various conditions (Supporting
Information). The formation of artemisinin was detected using
1968
https://doi.org/10.1021/acs.jnatprod.1c00246
J. Nat. Prod. 2021, 84, 1967−1984

Journal of Natural Products
pubs.acs.org/jnp
Article
Figure 3. Regioselectively deuterated compound, 15,15,15-d3-DHAA, was synthesized to determine whether any of the C15 deuterium atoms
would be lost during its spontaneous conversion to artemisinin (16, 17, or 2).
Figure 4. Retrosynthetic analysis of 15,15,15-d3-DHAA (11) from DHAA (1).
(21) would in turn be derived through oxidative cleavage of
the C4−C5 double bond of the cis-dehydrodecalin ring of
DHAA (1).
The primary alcohol of diol 26 was regioselectively
protected as acetate 27 with acetyl chloride and triethylamine
in CH2Cl2. The epimeric mixture at C4 remained unchanged;
the ratio was determined to be 54/46 through integration
of the two acetate methyl protons at δ 2.02 and 2.03. The
resulting secondary alcohol of acetate 27 was protected as the
TES ether 28 (56/44 epimeric mixture was determined by
integrating the two multiplets at δ 3.78 and 3.72, respectively).
The acetate was subsequently deprotected with MeLi to yield
primary alcohol 29. The C4 epimeric mixture was determined
to be 52/48 by integrating the C14 methyl doublets at δ 0.86
and 0.84, respectively. The resulting alcohol was treated with
Dess Martin periodinane (DMP) to yield aldehyde 30, which
was isolated as an epimeric mixture at C4 (56/44 determined
by integrating the C4 methine protons at δ 3.76 and 3.69,
respectively). Wittig olefination gave the monoalkene product
31 (51/49 epimeric mixture at C4 determined by integrating
the C15 methyl protons at δ 1.13 and 1.12, respectively).
Mild deprotection of the TES group with PPTS in MeOH
afforded secondary alcohol 32, which was determined to be a
DHAA (1) was treated with NaIO4 in the presence of a
catalytic amount of RuCl3 and H2SO4 to dihydroxylate the
C4−C5 bond and subsequently yield lactone 22 (Scheme 1).
The resulting lactone 22 was reduced to triol 23 with LiAlH4.
Both lactone 22 and triol 23 were crystallized, and the
stereochemistry was determined to be (4R, 5S) (Figure 5).
Triol 23 was subsequently treated with NaIO4 to cleave the
C4−C5 bond to give lactol 24. The resulting lactol was converted
to TBDMS-protected primary ether 25. The resulting ketoalde-
hyde 25 was reduced to diol 26 with LiAlH4. A strong reducing
agent such as LiAlH4 was required for reduction, because
when weaker hydride sources such as NaBH4 were used, the
reduction would stall at the lactol stage (data not shown).
The C4-stereocenter was carried forward as an epimeric mixture
(44/56 ratio determined by integrating the multiplets at δ 3.88
and 3.80, respectively) in the following six steps until this chiral
center was oxidized to ketone 33.
1969
https://doi.org/10.1021/acs.jnatprod.1c00246
J. Nat. Prod. 2021, 84, 1967−1984

Journal of Natural Products
pubs.acs.org/jnp
Article
a
Scheme 1. Synthesis of the Diene Intermediate 37 from DHAA (1)
a
* = carried forward from compound 26 as a diastereomeric mixture at C4 until the chiral center was converted to the ketone in compound 33.
The epimeric ratios of compounds 26−32 are described in the main text.
51/49 epimeric mixture at C4 through integrating the C16
alkene carbon signals at δ 117.4 and 117.5, respectively. The
secondary alcohol was oxidized to methyl ketone 33 with
DMP. TES triflate was used to convert methyl ketone 33 to
the silyl enol ether derivative 34. Silyl enol ether 34 was used
as the crude material for the next step; careful inspection of
the 1H NMR spectrum indicated some of the undesired
regioisomer (see Part 10 of the Supporting Information).
Of the mixture, 28% was the more substituted silyl enol ether,
while 72% was the desired silyl enol ether 34. Rubottom
oxidation of TES enol ether 34 with mCPBA incorporated the
oxygen to yield α-substituted ketone 35. Due to the mixture of
silyl enol ether regioisomers, the undesired Rubottom oxidation
product where the oxygen was incorporated at C3 was also
incorporated. By integration of the corresponding protons,
this ratio of the desired to undesired regioisomers was 81 to 19
(see Supporting Information for details). The ketone was
converted to diene 36 upon treatment with methylenetriphe-
nylphosphorane, to yield a regioisomeric mixture of 66 to 34;
the major regioisomer was the desired diene 36. The TES
group of diene 36 was cleaved with PPTS in MeOH to yield
allylic alcohol 37. This deprotection to free alcohol 37 enabled
the separation of the minor amount of the C3-alcohol through
flash column chromatography. Details of this synthesis are
included in the Supporting Information file.
With the key diene intermediate 37, treatment with Grubbs
second generation catalyst followed to give the cis-dehydrodecalin
ring system (38) with the desired oxygen incorporated at
C15 (Scheme 2). The allylic alcohol was oxidized with Dess
Martin periodinane to yield the aldehyde intermediate (39).
The aldehyde intermediate underwent Pinnick oxidation con-
ditions to yield the carboxylic acid at C15 (40), which would
be ready for deuterium incorporation. When the C15-carboxylic
acid (40) was treated with LiAlD4 and AlCl3,16 the dideutero
alcohol was obtained (41) instead of the desired trideuter-
omethyl group. The yield of this reduction was 52% on a
70 mg scale. Therefore, only LiAlD4 was used to reduce the
carboxylic acid (40) to give the dideuterated alcohol intermediate
1970
https://doi.org/10.1021/acs.jnatprod.1c00246
J. Nat. Prod. 2021, 84, 1967−1984

Journal of Natural Products
pubs.acs.org/jnp
Article
Figure 5. Crystal structures of (A) 22 and (B) 23.
Scheme 2. Synthesis of 15,15,15-d3-DHAA (11) from Diene Intermediate 37
(41) in a 45% yield. The alcohol (41) was converted to
mesylate 42 with methanesulfonyl chloride in the presence of
stoichiometric amounts of Et3N. Subsequent treatment of
crude mesylate 42 with LiAlD4 gave the desired trideuter-
omethyl intermediate (43) with a yield of 32% over two
steps. The NMR spectrum of the crude reaction mixture
showed the desired trideuteromethyl product 43 as the major
product; therefore, the low yield is likely due to loss of material
during the workup procedures for two steps. The TBDMS
group of 43 was deprotected in the presence of stoichiometric
amounts of pTsOH in CH3OH to afford primary alcohol 44.
The final desired compound was obtained by treating primary
alcohol 44 through the oxidation sequence of Dess Martin
periodinane and Pinnick conditions to yield aldehyde 45 and
15,15,15-d3-DHAA (11), respectively. The overall yield for
the 25 steps to access 15,15,15-d3-DHAA (11) from DHAA
was 0.02% (Schemes 1 and 2).
(ii) Identification of UV-C Radiation to Promote the
Conversion of DHAA to Artemisinin. In our previous
study to determine the rate of the conversion of 3,3-d2-DHAA
to artemisinin (Figure 2, 3 to 6), light was determined to
accelerate the transformation.16 However, in order to rule out
heat as a factor to enhance the rate, several other conditions
were tested to determine if heat were responsible in enhancing
the rate of transformation of DHAA to artemisinin.
A solution of DHAA (1) in CH2Cl2 was aliquoted into three
borosilicate glass vials; the solvent was evaporated by nitrogen
flow, and each vial was left in different conditions: (1) in a
1971
https://doi.org/10.1021/acs.jnatprod.1c00246
J. Nat. Prod. 2021, 84, 1967−1984

Journal of Natural Products
pubs.acs.org/jnp
Article
Figure 6. (A) Three conditions were tested to determine that UV-C light can enhance the rate of conversion of DHAA to artemisinin (1 to 2).
(B) 1H NMR spectroscopic overlay of the CDCl3 extracts of the different conditions. Top spectrum = IR lamp at ambient temperature
(purple), second row = heat at 80 °C (green), third row = UV-C light at ambient temperature (red), and bottom row = DHAA standard
(blue). Both conditions with heat (IR lamp and 80 °C) produced less artemisinin (δ 5.88) compared to the conditions with a UV-C lamp
1
within 2.5 h. H NMR spectrum obtained on a 500 MHz NMR magnet in CDCl3 solvent. Chemical shift range = δ 4.3−6.3.
box with an infrared (IR) lamp, (2) in a box with a UV-C
lamp, and (3) in an oil bath in the fume hood set to 80 °C
(Figure 6A). After 2.5 h, the contents of each vial were
The conversion of artemisinin to this rearranged product under
UV-C irradiation suggests that, although light accelerates the
conversion of DHAA to artemisinin (1 to 2), the exposure of
light for too long will result in the conversion of artemisinin to
this alternative rearranged product (Figure 7, 2 to 47). The
mechanism shows that, under UV-C light, the O−O bond of
the endoperoxide homolytically cleaves and recombines to
form the tetrahydrofuran ring, and the other oxygen radical
recombines with the carbon radical to form the ester carbonyl.
This rearranged product of artemisinin was also previously
observed when artemisinin was heated to 190 °C.18
The presence of this rearranged product (47) was con-
firmed in plant extracts of Artemisia annua supplements and
by matching the retention time and mass by LC-HRMS to
the authentic standard (see Supporting Information, Part 5).
(iv) Time Course Studies with DHAA Isotopologues
and Quantification with LC-HRMS. With the desired
15,15,15-d3-DHAA isotopologue (11) in hand (vide supra),
its conversion to artemisinin was measured through time
course assays as previously developed with 3,3-d2-DHAA16
(Supporting Information (Part 7) and summarized in Table 1).
In order to determine the effect of deuterium incorporation
with respect to the rate of artemisinin formation, the rate of
conversion of nondeuterated DHAA to artemisinin (1 to 2)
was determined using a trideuterated artemisinin isotopologue
1
dissolved in CDCl3, and H NMR spectra were acquired. The
results indicated that the UV-C lamp visibly formed arte-
misinin with the clear detection of artemisinin (the proton
with δ 5.88 chemical shift), while the other two conditions
(IR lamp and oil bath) had a smaller amount of artemisinin
(Figure 6B). These results confirm that light, not heat, is
responsible for enhancing the rate of conversion of DHAA to
artemisinin. Notably, there are other chemical shifts present
in all three conditions that are likely related to other products,
but only the UV−C lamp conditions clearly resulted in the
detection of artemisinin by 1H NMR spectroscopy within
2.5 h (Figure 6B). There is a small signal that corresponds to
artemisinin in the case of the IR radiation, but this signal is
significantly smaller compared to the one found in the conditions
involving UV-C irradiation.
(iii) Identification of a Light-Induced Rearrangement
Product of Artemisinin. In order to identify a possible
rearrangement pathway of artemisinin under UV-C light, a
solution of artemisinin (2) in CH2Cl2 was dried down in a
vial and placed under UV-C irradiation (Figure 7A). After 8
days under UV-C light, the material was purified, and a rearranged
product that lacked the endoperoxide bridge was identified (47).
1972
https://doi.org/10.1021/acs.jnatprod.1c00246
J. Nat. Prod. 2021, 84, 1967−1984

Journal of Natural Products
pubs.acs.org/jnp
Article
1
Figure 7. (A) Mechanism of light-induced rearrangement of artemisinin (2 to 47). (B) H NMR spectroscopic overlay. Top = rearranged
product. Second row = artemisinin exposed under UV-C lamp for 192 h (8 days). Third row = artemisinin exposed under UV-C lamp for 24 h.
Bottom row = artemisinin standard. The singlet at δ 5.32 is CH2Cl2.
(14) as an internal standard. The influence of the density of
DHAA on the rate was also determined by comparing the
amount of artemisinin formed with a more dense and a less
dense amount of DHAA (i.e., 100 and 10 μg of DHAA were
used for Table 1 entries 1−12 and 13−24, respectively).
In addition, three different conditions were tested that involved
exposing DHAA to (i) ambient light, (ii) UV-C light (100−
280 nm), and (iii) total darkness.
When 15,15,15-d3-DHAA (11) was exposed to air in the
dark and two sources of light, light by the window and a
UV-C light source, d3- and d2-artemisinin isotopologues were
predominantly formed (Figure 8). In our previous study with
3,3-d2-DHAA (3), d2-artemisinin (6) was formed at a 40-fold
faster rate in the presence of light. It is important to note that
the amount of light by the window may differ depending on
the climate (i.e., the time of the year). In the previous
study,16 the window experiment was performed in May, while
in the current study, the window experiment was performed in
October when there is generally less sunlight. With 15,15,15-
d3-DHAA (11) as the starting material, d3-artemisinin (14) was
formed around 3-fold faster in the presence of ambient light at
24 and 47 h (Table 1). However, at 120 and 312 h, the rates
of formation of artemisinin were almost equal in the presence
and absence of light. The results of the time course are sum-
marized in Table 1. At 24 h, the conditions that involved
shining UV-C light to DHAA formed artemisinin the fastest
(49-fold faster than conditions in the dark and 15-fold faster
than conditions with ambient light). However, after 120 and
312 h, the conditions involving UV-C radiation resulted in
less artemisinin detection compared to that of the conditions
involving ambient light and dark conditions. This decrease in
detection of artemisinin under UV-C irradiation at 120 h and
beyond could be explained by the fact that, under UV-C
irradiation, artemisinin begins to rearrange to the tetrahydrofuran
product (Figure 7, vide supra). Indeed, the ion matching the
calculated mass (m/z 331.2307) increasingly appears over
time, especially under the UV-C conditions (see Supporting
Information, Part 5). The same rearranged product is also
observed by LC-HRMS when 3,3-d2-DHAA (3) and DHAA
(1) were used as the starting material under the UV-C
irradiation conditions (see Supporting Information, Part 5).
Figure 8B shows a representative LC-HRMS trace for the 120 h
time point of 15,15,15-d3-DHAA (11) in the presence of ambient
light. Clearly, d3-artemisinin (14) is the dominant isotopologue of
artemisinin that is formed.
In order to test a density dependence on artemisinin
formation from DHAA, similar time course experiments were
performed but with less 15,15,15-d3-DHAA (10 μg of 11) in
the vials (Supporting Information (Part 7), entries 13−24 in
Table 1). The two types of samples had 100 and 10 μg of
15,15,15-d3-DHAA (11) in each vial, respectively (see Supporting
Information, Part 7). In general, the more dense samples of
1973
https://doi.org/10.1021/acs.jnatprod.1c00246
J. Nat. Prod. 2021, 84, 1967−1984

Journal of Natural Products
pubs.acs.org/jnp
Article
a
Table 1. Results of Time Course Experiments of 15,15,15-d3-DHAA (11) to d3-Artemisinin (14)
entry
condition
amount of 11
time
d3-artemisinin (μg)
fold vs dark
1
light
100 μg
100 μg
100 μg
100 μg
100 μg
100 μg
100 μg
100 μg
100 μg
100 μg
100 μg
100 μg
10 μg
24 h
47 h
3.36 × 10−3
2.01 × 10−2
8.28 × 10−2
2.81 × 10−1
1.01 × 10−3
5.82 × 10−3
6.78 × 10−2
2.52 × 10−1
4.94 × 10−2
1.61 × 10−2
1.14 × 10−2
5.01 × 10−3
8.64 × 10−5
1.28 × 10−3
7.49 × 10−3
1.89 × 10−2
1.55 × 10−5
3.3
3.5
1.2
1.1
1.0
1.0
1.0
1.0
49
2
light
3
light
120 h
312 h
24 h
4
light
5
dark
6
dark
47 h
7
dark
120 h
312 h
24 h
8
dark
9
UV-C
UV-C
UV-C
UV-C
light
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
47 h
28
120 h
312 h
24 h
0.17
0.020
5.6
light
10 μg
47 h
Not Available
light
10 μg
120 h
312 h
24 h
1.5
light
10 μg
0.73
1.0
dark
10 μg
b
dark
10 μg
47 h
Not Detected
Not Available
dark
10 μg
120 h
312 h
24 h
4.97 × 10−3
2.59 × 10−2
6.63 × 10−3
2.95 × 10−3
2.03 × 10−3
1.0
1.0
dark
10 μg
UV-C
UV-C
UV-C
UV-C
10 μg
430
Not Available
0.41
0.15
10 μg
47 h
10 μg
120 h
312 h
10 μg
3.83 × 10−3
a
b
Amount of d3-artemisinin (14) formed from 15,15,15-d3-DHAA (11) (either 100 or 10 μg in the vial). The peak was not detected = the internal
standard peak area (m/z 283 for d0-artemisinin) was also relatively smaller than the other measurements (40 000 000 vs 110 000 000).
15,15,15-d3-DHAA (100 μg of 11 in each vial) converted to
d3-artemisinin (14) at a faster rate. For instance, at 24 h in
the presence of ambient light, 3.36 ng of d3-artemisinin (14)
formed in the more dense case, while only 86.4 pg of
d3-artemisinin (14) was formed when there was less DHAA
in the vials, a 39-fold difference in formation of artemisinin.
Because more artemisinin is formed when the amount
of DHAA is increased in the vials (100 vs 10 μg, Table 1),
we propose possible intermolecular interactions between
one DHAA molecule and a second DHAA molecule that
initiates the endoperoxide forming cascade reaction (Figure 9).
Moreover, when the C15-carboxylic acid is reduced to the
alcohol, this alcohol is stable over time (i.e., there is no
spontaneous endoperoxide formation), confirming a role for
the carboxylic acid functional group14 in DHAA to initiate
the transformation to artemisinin. In order to further elaborate
on the findings that increased amounts of DHAA results in
an enhanced rate of endoperoxide formation, another set of
experiments was performed to identify optimal conditions to
convert DHAA to artemisinin (see Supporting Information,
Part 14). Among the conditions, the addition of stoichiometric
amounts of benzoic acid resulted in an enhanced rate of
endoperoxide formation. The role of the carboxylic acid in
enhancing endoperoxide formation is not completely clear,
but a possible explanation is that the carboxylic acid provides
the protons to promote polycyclization as shown in Figures 13,
16, and 17 (e.g., Figure 16, 64 to 65).
the amount of artemisinin detected was ∼1.1- and ∼1.2-fold
more in the presence of light. This observation could be
explained by the fact that, over time, a layer of artemisinin
forms on the surface of a lattice of DHAA and, in turn,
blocks exposure of oxygen to the rest of DHAA (Supporting
Information (Part 7), Figure S19). Another reason that the
“dark” conditions form similar amounts of artemisinin com-
pared to the “ambient light” at later time points (120 h and
beyond) is, under the conditions involving light, artemisinin
may be converting to the rearranged product (cf. Figure 7,
47) over time. However, the rearranged product was not
detected at the later time points (see Supporting Information,
Part 5) by LC-HRMS probably because the mass spectro-
meter was not tuned for ionization to the rearranged product
(instead, the instrument was tuned to detect artemisinin).
Nonetheless, in our previous study,16 when artemisinin was
left under ambient light by the window for 64 days, the
rearranged product was detected (see Supporting Information of
reference). On the other hand, when the same amount of
artemisinin was left in the dark for 64 days, no rearranged
product was detected.
(v) Intermolecular Competitive Kinetic Isotope
Effects. To expand on the observation of the rates of
artemisinin formation from the different DHAA isotopologues
and to determine the kinetic isotope effect (KIE), mixtures of
the different DHAA isotopologues were made, and the artemi-
sinin isotopologues were detected by LC-HRMS (Figures 10−12).
When 15,15,15-d3-DHAA (11) was mixed with nondeuterated
DHAA (1) in a 1:1 ratio, nondeuterated DHAA converted to
artemisinin at a faster rate (Figure 10). At all time points tested
(48, 125, and 312 h), the C15-trideuterated DHAA isotopologue
converted to artemisinin at a slower rate (Figure 10B). These
rates were determined by measuring the amounts of d3-artemisinin
In analyzing the amounts of artemisinin formed from
15,15,15-d3-DHAA (14 from 11) in the presence and absence
of light over time, the relative amounts of artemisinin differed
considerably in the earlier time points (i.e., Table 1, 24 and
47 h). For instance, at 24 and 47 h, artemisinin was formed
∼3-fold faster in the presence of ambient light compared to
in the absence of light (Table 1). However, at 120 and 312 h,
1974
https://doi.org/10.1021/acs.jnatprod.1c00246
J. Nat. Prod. 2021, 84, 1967−1984

Journal of Natural Products
pubs.acs.org/jnp
Article
Figure 8. (A) Various conditions (i.e., ambient light, dark, UV-C, and different amounts of 11) were tested to convert 15,15,15-d3-DHAA (11)
to artemisinin. Percentages indicate relative abundances of each artemisinin isotopologue at the 120 h time point (Supporting Information
(Part 7) for LCMS traces). (B) Representative LCMS data for 120 h time point with ambient light (no internal standard). The data of the time
course are summarized in Table 1.
and d0-artemisinin by mass spectrometry (Supporting Information,
Part 8).
15,15,15-d3-DHAA (11) in the formation to artemisinin. At 48
and 125 h, the rates of artemisinin formation were similar, but
at 312 h, 3,3-d2-DHAA (3) converted to artemisinin at a faster
rate (3-fold, see Figure 12B).
Figure 11 shows the intermolecular kinetic isotope effect
studies when DHAA was mixed in a 1:1 ratio with 3,3-d2-DHAA
(3). In all time points (48, 125, and 312 h), the nondeuterated
version of DHAA converted to artemisinin at a faster rate
compared to 3,3-d2-DHAA (3) (Figure 11B). In summary, a
kinetic isotope effect was observed in that nondeuterated DHAA
(1) converted to artemisinin at a faster rate compared to the
deuterated DHAA isotopologues (3 and 11). The KIE (kH/kD)
values at C15 ranged from 2.23 to 3.31, while the KIE (kH/kD)
values at C3 ranged from 2.05 to 2.42. The higher KIE values of
d3-DHAA could be associated with a secondary KIE.
On the other hand, Figure 12 shows the intermolecular
competitive KIE experiments between 3,3-d2-DHAA (3) and
Based on the observed kinetic isotope effects from the
different rates of conversion of DHAA isotopologues to arte-
misinin (Figures 10 and 11), the initial ene reaction between
DHAA and one molecule of oxygen is proposed to occur via
three different modes (Figure 13). Deuteration at either the
C3 or the C15 position resulted in a slower rate of conversion
to artemisinin from DHAA. Each resulting allylic hydroperoxide
intermediate (Figure 13, 49, 52, and 55) is eventually converted
to the same oxonium (53) that reacts with a second molecule of
oxygen to ultimately form artemisinin (2). The values for the
intermolecular kinetic isotope effects ranging from 2.1 to 3.3
1975
https://doi.org/10.1021/acs.jnatprod.1c00246
J. Nat. Prod. 2021, 84, 1967−1984

Journal of Natural Products
pubs.acs.org/jnp
Article
Figure 9. Alternative intermolecular interactions between DHAA (1) molecules proposed in promoting the different ene reactions shown in
Figure 13. (A) The neighboring carboxylate interacts with the hydrogen at C3. (B) The neighboring carboxylate interacts with the hydrogen at
C6. (C) The neighboring carboxylate interacts with the hydrogen at C15.
Figure 10. Kinetic isotope effect was directly measured by using a (A) 1:1 mixture of DHAA (1) and 15,15,15-d3-DHAA (11). The artemisinin
isotopologues (2 and 14) were detected by LC-HRMS at 48, 125, and 312 h. (B) Chart summarizing the results. For Figures 10−12, see
Supporting Information (Parts 8 and 11) for LC-HRMS traces. The bars on the chart in (B) are the standard deviation values of two different
data points. y-axis = areas of the artemisinin isotopologues.
(Figures 10 and 11) are reasonable when compared to a
previous study of a nitrosoarene ene reaction19 (kH/kD ∼2).
Alternative to the concerted ene reactions shown in Figure 13,
the formation of the allylic hydroperoxides can be rationalized
through a perepoxide (Figure 14, 57).20 However, in the case of
the mechanism through a perepoxide intermediate, the formation
of the perepoxide is the rate-limiting step, and subsequent
rearrangement is fast. Therefore, through this particular mech-
anism, when singlet oxygen undergoes an ene reaction, inter-
molecular kinetic isotope effect values for the ene reaction are
usually smaller21 (kH/kD ∼ 1.0 and 1.1). Similarly, a biradical
mechanism22 can also be ruled out due to the high-energy
state of a biradical intermediate compared to the lower barrier
for hydrogen atom abstraction, which would result in a small
kinetic isotope effect (kH/kD ∼1−2).
(vi) 18O2 Labeling Studies for Endoperoxide Formation
from DHAA. In order to test the oxygen incorporation
mechanism to form the endoperoxide from DHAA, 18O2
(99%, 18O) was used in the presence of 15,15,15-d3-DHAA
(11) (Figure 15). Although a previous study had determined
a mixture of 18O incorporation when DHAA hydroperoxide
was exposed to 18O2,15 there has not been a previous study
that performed the 18O2 experiment from DHAA. Surpris-
ingly, 18O was incorporated into artemisinin from DHAA, but
1976
https://doi.org/10.1021/acs.jnatprod.1c00246
J. Nat. Prod. 2021, 84, 1967−1984

Journal of Natural Products
pubs.acs.org/jnp
Article
Figure 11. (A) Kinetic isotope effect was directly measured by using a 1:1 mixture of DHAA (1) and 3,3-d2-DHAA (3). The artemisinin isotopologues
(2 and 6) were detected by LC-HRMS at 48, 125, and 312 h. (B) Chart summarizing the results. y-axis = areas of the artemisinin isotopologues.
Figure 12. (A) The kinetic isotope effect was directly measured by using a 1:1 mixture of 15,15,15-d3-DHAA (11) and 3,3-d2-DHAA (3). The
artemisinin isotopologues (14 and 6) were detected by LC-HRMS at 48, 125, and 312 h. (B) Chart summarizing the results. y-axis = areas of
the artemisinin isotopologues.
1977
https://doi.org/10.1021/acs.jnatprod.1c00246
J. Nat. Prod. 2021, 84, 1967−1984

Journal of Natural Products
pubs.acs.org/jnp
Article
Figure 13. Three possible ene reactions (at C3, C6, or C15) are illustrated that involve the interaction of DHAA (1) with a molecule of
oxygen. The resulting allylic hydroperoxides (49, 52, and 55) can converge to a common oxonium intermediate (53), which reacts with another
molecule of O2 to form artemisinin (2). A mechanism is proposed for the reaction with the second molecule of oxygen in Figure 16.
the isotopic composition was 1.0:7.2:14 for 18O2:18O3:18O4 in
the artemisinin product (Figure 15, Table 2: 71 h entry).
Two different time points were taken for the exposure
of 15,15,15-d3-DHAA (11) to 18O2, 71 h and 14 days, to
determine the effect of time and 18O incorporation (Table 2).
These experiments were performed in the presence of light
(see Supporting Information, Part 9). Notably, at the longer
time point (14 days), there was a significant amount of
artemisinin with no 18O atoms incorporated (see Supporting
Information). This lack of 18O incorporation suggests that the
oxygen (16O2) from the atmosphere was permeating through
the reaction vessel, which may not have been completely
sealed from the environment. Nonetheless, this information
was useful in that the amount of artemisinin that lacked
18O atoms directly correlated with the amount of artemisinin
with one and two 18O atoms incorporated. Table 2 summarizes
the results from the 18O2 studies.
pathways as shown above in Figure 13. The resulting oxonium
(63) can be attacked either by 18O labeled water to yield the
cyclic hemiketal intermediate (64) through pathway a in
Figure 16. The resulting hemiketal (64) can attack molecular
oxygen (18O2) to yield the peroxide (65). This peroxide (65)
can cyclize through intramolecular nucleophilic attack of the
peroxide moiety onto the ketone, which subsequently attacks
the aldehyde and cyclizes with the carboxylic acid to yield the
endoperoxide with four 18O atoms (66).
Alternatively, oxonium 63 can be attacked by the 16O water
(pathway b, Figure 16) to yield hemiketal 67. This hemiketal can
attack 18O2 to yield the peroxide (68). The oxonium 68 can
undergo intramolecular nucleophilic attack by the carboxylic acid
to yield the lactone (69). The hemiketal of 69 could dehydrate
to yield the oxonium 70 (pathway c, Figure 16), which
undergoes cyclization to yield artemisinin with three 18O atoms
(71).
A mechanism to explain the mixture of isotopologues
is shown in Figure 16. DHAA (1) undergoes an initial ene
reaction, which can yield the oxonium intermediate (Figure 16,
63). This initial ene reaction can go through multiple mechanistic
In a third pathway from intermediate 69, the hemiketal
could open to yield ketone 72 (pathway d in Figure 16). The
resulting hemiacetal (72) could dehydrate to yield oxonium 73,
which could undergo polycyclization to yield the artemisinin
1978
https://doi.org/10.1021/acs.jnatprod.1c00246
J. Nat. Prod. 2021, 84, 1967−1984

Journal of Natural Products
pubs.acs.org/jnp
Article
Figure 14. Alternative stepwise mechanism for the ene reaction to convert DHAA (1) to the different allylic hydroperoxide intermediates
(49, 52, and 55) from perepoxide 57.
isotopologue with two 18O atoms (75). Table 2 summarizes the
different ratio of artemisinin isotopologues produced from the
exposure of 15,15,15-d3-DHAA (11) to 18O2.
artemisinin isotopologues containing four and three 18O atoms
formed (Figure 15B and Table 2). These 18O labeling experi-
ments also confirm a mixed mechanism toward endoperoxide
formation in the conversion of DHAA to artemisinin (Figures 16
and 17). Collectively, these results support a mixed mechanistic
pathway when two molecules of oxygen react with DHAA to
form the endoperoxide functional group (Figure 18). In addition,
it is notable that UV-C light: (i) promotes the conversion
of DHAA to artemisinin (Figure 18, 1 to 66 and 71) and
(ii) accelerates a rearrangement of the endoperoxide bridge
of artemisinin to a tetrahydrofuran product (Figure 7, 2 to
47, and Figure 18, 66 and 71 to 82).
These observations are relevant in optimizing the light-
driven process to convert DHAA to artemisinin where there
should be enough to efficiently produce the endoperoxide
bridge but not too much to prevent the rearrangement and
loss of the endoperoxide.
An alternative to the pathway presented in Figure 16 to
obtain artemisinin with three 18O atoms (compound 71) is
presented in Figure 17. The carboxylic acid intermediate 65
could potentially undergo a cyclization process that is initiated
by the carboxylic acid moiety nucleophilically attacking the
aldehyde functional group to form the cyclic hemiacetal (76),
which results in the loss of one of the 18O atoms through a
dehydration process (76 to 77). The resulting oxonium can
undergo a cyclization to form artemisinin with three 18O atoms
incorporated (77 to 71). Other possible mechanisms are
presented in the Supporting Information (Part 9, Figures S38
and S39).
CONCLUSION
■
In conclusion, a synthesis of 15,15,15-d3-DHAA (11) was
accomplished from DHAA (1). Subsequent conversion to
artemisinin indicated artemisinin formation at a faster rate in
the presence of UV-C radiation initially (Table 1, 24 h).
However, at later time points (47, 120, and 312 h), more
artemisinin was detected in the presence of ambient light and
in darkness due to a rearrangement of artemisinin promoted
by UV-C irradiation (Figure 7). A comparison of rates of
artemisinin formation from 15,15,15-d3-DHAA (11), 3,3-d2-
DHAA (3), and DHAA showed that nondeuterated DHAA
converted to artemisinin the fastest (i.e., 1 to 2 was the
fastest), indicating a kinetic isotope effect in the initial ene
reaction with molecular oxygen (Figures 10 and 11, kH/kD ∼
2.1−3.3). This result supported multiple ene reactions
that occur at the C3-, C6-, and C15-positions of DHAA
(Figures 13 and 14). Furthermore, when 15,15,15-d3-DHAA
(11) was exposed to 18O2 (99% 18O) over time, a mixture of
EXPERIMENTAL SECTION
■
General Experimental Procedures. Melting points were
recorded on a Global Medical and Lab Solutions melting point
apparatus (India). A JASCO P-1010 polarimeter (JASCO Inc.
Easton, MD) was used for measuring optical rotations of
compounds. IR data were obtained on an FTIR instrument (Nicolet
iS50 FT-IR spectrometer, Thermo Fisher Scientific, Waltham, MA,
USA). The IR data were analyzed on OMNIC software (Thermo
Fisher Scientific). A Bruker (Billerica, MA) 300 or 500 MHz NMR
spectrometer was used to characterize compounds presented in the
synthesis. NMR data were analyzed using either Topspin software
(Bruker) or Mestrenova software (Mestrelab, Santiago de Compostela,
A Coruna, Spain). CDCl3 (Cambridge Isotope Laboratories, Tewsks-
̃
bury, MA) was used as the NMR solvent for the obtained spectra.
The chloroform peaks for the proton and 13C NMR spectra were
referenced to δ 7.26 and δ 77.16, respectively. An LTQ Orbitrap
XL mass spectrometer (Thermo) connected to an Acuity (Waters,
Milford, MA) UPLC system was used to analyze compounds.
1979
https://doi.org/10.1021/acs.jnatprod.1c00246
J. Nat. Prod. 2021, 84, 1967−1984

Journal of Natural Products
pubs.acs.org/jnp
Article
Figure 15. (A) Reaction of 18O2 with 15,15,15-d3-DHAA (11). LC-MS results for (B) 71 h with ambient light. A mechanism for
18O incorporation from molecular oxygen (18O2) is proposed in Figure 18. Extracted ion chromatogram in (B) is shown with an 8 ppm mass
tolerance window. n = number of 18O atoms incorporated into artemisinin.
Xcalibur 2.1 software (Thermo Scientific) was used to analyze liquid
chromatography mass spectrometry (LCMS) data. Prior to analysis,
the mass spectrometer was calibrated with ESI positive mode or ESI
negative mode solution (Thermo Scientific, Waltham, MA). 18O2 gas
(99% 18O) was purchased from Sigma-Aldrich (St. Louis, MO).
TLC plates (silica gel) with fluorescent indicator (254 nm) were
used to analyze compounds. The compounds were visualized with
CAM stain (Ceric Ammonium Molybdate recipe: 2.35 L of water,
120 g of ammonium molybdate, 5 g of ceric ammonium molybdate,
and 150 mL of concentrated sulfuric acid). For time course studies
with DHAA to artemisinin measurements, clear glass 2 mL screw
cap vials were used for the light conditions, and amber glass 2 mL
screw cap vials were used for the dark conditions.
harmonics, implemented in SCALE3 ABSPACK23 scaling algorithm.
The structure was solved by direct methods (OLEX224 and ShelXT)25
and refined against F2 with full-matric least-squares using the program
ShelXL.26 All non-hydrogen atoms were refined with anisotropic
displacement parameters. All H atoms, bound to carbon atoms, were
placed on calculated positions and refined using the riding model with
isotopic displacement parameters based on those of the parent atom.
The oxygen bound hydrogen atom, H2, position was determined by
electron density plot.
The crystallographic data can be obtained free of charge via
https://www.ccdc.cam.ac.uk/structures/ or from the Cambridge
Crystallographic Data Center, 12 Union Road, Cambridge CB2
1EZ, UK; fax: (+44)1223−336−033; or e-mail: deposit@ccdc.cam.
ac.uk. CCDC number 2061809.
X-ray Crystallography. X-ray Crystal Data for 1. Directly from
the commercial source, 1 was obtained as colorless blocks.
Crystallogrpahic data were collected at 100.0(1) K on a XtaLAB
Synergy/Dualflex, HyPix fitted using Cu Kα, λ = 1.541 84 Å,
followed by empirical absorption correction using spherical
C15H24O2, M
= 236.34, 0.216 × 0.171 ×
0.076 mm3,
orthorhombic, space group P212121 (no. 19), a = 6.24107(9) Å,
b = 8.21018(12) Å, c = 26.5277(5) Å, V = 1359.29(4) Å3, Z = 4,
Dc = 1.155 g cm−3, μ = 0.580 mm−1. F000 = 520, 2θmax = 152.4°,
1980
https://doi.org/10.1021/acs.jnatprod.1c00246
J. Nat. Prod. 2021, 84, 1967−1984

Journal of Natural Products
pubs.acs.org/jnp
Article
Figure 16. Proposed mechanism of oxygen incorporation into the endoperoxide bridge from DHAA (1) based on 18O2 studies. *O = 18O.
9206 reflections collected, 2578 unique (Rint = 0.0325). Final
GOOF = 1.014, R1 = 0.0304, wR2 = 0.0780, R indices based on 2454
reflections with I > 2σ(I) (refinement on F2), |Δρ|max= 0.170 e Å−3,
161 parameters, 0 restraints.
X-ray Crystal Data for 2. Directly from the commercial source,
2 was obtained as colorless blocks. Crystallogrpahic data were
collected at 100.0(1) K on a XtaLAB Synergy/Dualflex, HyPix fitted
using Cu Kα, λ = 1.541 84 Å, followed by empirical absorption
correction using spherical harmonics, implemented in SCALE3
ABSPACK23 scaling algorithm. The structure was solved by direct
methods (OLEX224 and ShelXT)25 and refined against F2 with full-
matric least-squares using the program ShelXL.26 All non-hydrogen
atoms were refined with anisotropic displacement parameters. All
H atoms, bound to carbon atoms, were placed on calculated positions
and refined using the riding model with isotopic displacement param-
eters based on those of the parent atom.
The crystallographic data can be obtained free of charge via
https://www.ccdc.cam.ac.uk/structures/ or from the Cambridge
Crystallographic Data Center, 12 Union Road, Cambridge CB2
1EZ, UK; fax: (+44)1223−336−033; or e-mail: deposit@ccdc.cam.
ac.uk. CCDC number 2061808.
Figure 17. Alternative mechanistic pathway to artemisinin with
C15H22O5, M = 282.32, 0.169 × 0.096 × 0.062 mm3,
three 18O atoms (71) from intermediate 65. *O = 18O.
orthorhombic, space group P212121 (no. 19), a = 6.3116(2) Å,
b = 23.9499(6) Å, c = 9.3103(3) Å, V = 1406.28(7) Å3, Z = 4, Dc =
1.333 g cm−3, μ = 0.820 mm−1. F000 = 608, 2θmax = 152.7°, 29 337
reflections collected, 2876 unique (Rint = 0.0392). Final GOOF =
1.099, R1 = 0.0505, wR2 = 0.1340, R indices based on 2770 reflections
1981
https://doi.org/10.1021/acs.jnatprod.1c00246
J. Nat. Prod. 2021, 84, 1967−1984

Journal of Natural Products
pubs.acs.org/jnp
Article
a
Table 2. Summary of 18O2 Incorporation into Artemisinin from d3-DHAA (11)
b
b
b
b
starting
time
18O4−artemisinin (Art) (58)
18O3−Art (59)
18O2−Art (60)
18O1−Art (61)
d3-Art (14)
d3-DHAA
71 h
1.44 × 108
255
9.44 × 106
1.00
7.31 × 107
129
6.49 × 107
6.88
1.02 × 107
18.0
3.55 × 108
37.6
5.68 × 105
1.00
2.23 × 108
23.6
4.64 × 106
8.17
1.18 × 109
125
ratio =
336 h
ratio =
d3-DHAA
a
Numbers reported are the areas under the peaks in the extracted ion chromatograms (retention time = 3.88 min corresponding to artemisinin,
b
18O4−Art (58) = m/z 294, 18O3−Art (59) = m/z 292, 18O2−Art (60) = m/z 290, 18O1−Art (61) = m/z 288, d3-Art (14) = m/z 286). All detected
artemisinin isotopologues with 18O atoms have three deuteriums.
Figure 18. Summary of the mixed mechanism of artemisinin formation from DHAA (1) supported by KIE studies to show multiple ene
reactions and 18O2 labeling studies to show multiple polycyclization pathways to form the endoperoxide (66 and 71). *O = 18O.
with I > 2σ(I) (refinement on F2), |Δρ|max= 0.283 e Å−3, 184 param-
eters, 0 restraints.
atoms were refined with anisotropic displacement parameters. All
H atoms, bound to carbon atoms, were placed on calculated positions
and refined using the riding model with isotopic displacement param-
eters based on those of the parent atom. The oxygen bound hydrogen
atom, H3, position was determined by electron density plot.
The crystallographic data can be obtained free of charge via
https://www.ccdc.cam.ac.uk/structures/ or from the Cambridge
Crystallographic Data Center, 12 Union Road, Cambridge CB2
1EZ, UK; fax: (+44)1223−336−033; or e-mail: deposit@ccdc.cam.
ac.uk. CCDC number 2061810.
X-ray Crystal Data for 22. From a solution of ethyl acetate:hexanes
(1:1, v/v), 22 was obtained as colorless platelets. Crystallogrpahic data
were collected at 98(2) K on a Rigaku AFC12/Saturn 724 CCD fitted
using Mo Kα, λ = 0.710 75 Å, followed by empirical absorption
correction using spherical harmonics, implemented in SCALE3
ABSPACK23 scaling algorithm. The structure was solved by direct
methods (OLEX224 and ShelXT)25 and refined against F2 with full-
matric least-squares using the program ShelXL.26 All non-hydrogen
1982
https://doi.org/10.1021/acs.jnatprod.1c00246
J. Nat. Prod. 2021, 84, 1967−1984

Journal of Natural Products
pubs.acs.org/jnp
Article
C15H24O3, M = 252.34, 0.50 × 0.33 × 0.10 mm3, orthorhombic,
space group P212121 (no. 19), a = 5.7974(5) Å, b = 14.3683(13) Å,
c = 16.0292(13) Å, V = 1335.2(2) Å3, Z = 4, Dc = 1.255 g cm−3,
μ = 0.085 mm−1. F000 = 552, 2θmax = 50.1°, 2507 reflections collected,
2507 unique (Rint = 0.1619). Final GOOF = 1.063, R1 = 0.0533,
wR2 = 0.1229, R indices based on 2180 reflections with I > 2σ(I)
(refinement on F2), |Δρ|max= 0.281 e Å−3, 157 parameters, 0 restraints.
X-ray Crystal Data for 23. From a solution of CDCl3 (10 mg in
0.7 mL), 23 was obtained as colorless platelets. Crystallogrpahic
data were collected at 100.0(1) K on a XtaLAB Synergy/Dualflex,
HyPix fitted using Cu Kα, λ = 1.541 84 Å, followed by empirical
absorption correction using spherical harmonics, implemented in
SCALE3 ABSPACK23 scaling algorithm. The structure was solved
by direct methods (OLEX224 and ShelXT)25 and refined against
F2 with full-matric least-squares using the program ShelXL.26 All
non-hydrogen atoms were refined with anisotropic displacement
parameters. All hydrogen atoms were placed on calculated positions
and refined using the riding model with isotopic displacement param-
eters based on those of the parent atom.
Hadi D. Arman − Department of Chemistry, The University
of Texas at San Antonio (UTSA), San Antonio, Texas
78249-0698, United States
Complete contact information is available at:
https://pubs.acs.org/10.1021/acs.jnatprod.1c00246
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This research was supported through the Bill & Melinda
Gates Foundation [OPP1188432] and the Max and Minnie
Tomerlin Voelcker Fund. Francis K. Yoshimoto PhD holds a
Voelcker Fund Young Investigator Award from the MAX
AND MINNIE TOMERLIN VOELCKER FUND. The
UTSA RISE program is also acknowledged for supporting a
research stipend for Ms. Kaitlyn Varela. NSF MRI grant for
X-ray diffraction (Award No. 1920057) and NSF MRI grant
for NMR spectrometer (Award No. 1625963) are also acknowl-
edged. This manuscript is dedicated to Dr. Trevor Laird, on the
occasion of his 75th birthday and for his guidance and support
throughout this funded research project. We are truly grateful
for the reviewers of this manuscript, who have significantly
improved the quality of this paper by suggesting additional
insightful experiments and providing their valuable, expert
opinion.
The crystallographic data can be obtained free of charge via
https://www.ccdc.cam.ac.uk/structures/ or from the Cambridge
Crystallographic Data Center, 12 Union Road, Cambridge CB2
1EZ, UK; fax: (+44)1223−336−033; or e-mail: deposit@ccdc.cam.
ac.uk. CCDC number 2061811.
C15H28O3, M = 256.37, 0.25 × 0.20 × 0.02 mm3, triclinic, space
group P1 (no. 1), a = 7.4494(4) Å, b = 9.7601(4) Å, c =
11.3665(6) Å, α = 76.475(4)°, β = 86.519(4)°, γ = 67.953(4)°, V =
744.41(7) Å3, Z = 2, Dc = 1.144 g cm−3, μ = 0.612 mm−1. F000
=
284, 2θmax = 125.0°, 11 239 reflections collected, 3514 unique
(Rint = 0.0706). Final GOOF = 1.111, R1 = 0.0677, wR2 = 0.1532, R
indices based on 3254 reflections with I > 2σ(I) (refinement on F2),
|Δρ|max= 0.345 e Å−3, 337 parameters, 7 restraints.
REFERENCES
■
(1) Wallart, T. E.; van Uden, W.; Lubberink, H. G. M.;
Woerdenbag, H. J.; Pras, N.; Quax, W. J. J. Nat. Prod. 1999, 62,
430−433.
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge at
https://pubs.acs.org/doi/10.1021/acs.jnatprod.1c00246.
■
sı
(2) Pandey, A. V.; Tekwani, B. L.; Singh, R. L.; Chauhan, V. S. J.
Biol. Chem. 1999, 274, 19383−19388.
(3) Krishna, S.; Bustamante, L.; Haynes, R. K.; Staines, H. M.
Trends Pharmacol. Sci. 2008, 29, 520−527.
LC-HRMS and NMR analysis of DHAA isotopologues;
1D and 2D NMR data of compound 47; description of
time course experiments of DHAA to artemisinin;
description of KIE experiments; description of experi-
ments involving 18O2 with DHAA; NMR spectra, high-
resolution mass spectra, and IR spectra of synthesized
compounds; HRMS data of time course experiments;
HRMS data of KIE experiments; HRMS data of 18O2
experiments; preliminary studies on the synthesis of
15,15,15-d3-DHAA; X-ray crystallography data of
compounds 1, 2, 22, and 23; and extra experiments
to determine that benzoic acid can promote the
conversion of DHAA to artemisinin (PDF)
Structure of 1 (CIF)
(4) Amara, Z.; Bellamy, J. F. B.; Horvath, R.; Miller, S. J.; Beeby,
A.; Burgard, A.; Rossen, K.; Poliakoff, M.; George, M. W. Nat.
Chem. 2015, 7, 489−495.
(5) Levesque, F.; Seeberger, P. H. Angew. Chem., Int. Ed. 2012, 51,
1706−1709.
(6) Lee, D. S.; Amara, Z.; Clark, C. A.; Xu, Z.; Kakimpa, B.;
Morvan, H. P.; Pickering, S. J.; Poliakoff, M.; George, M. W. Org.
Process Res. Dev. 2017, 21, 1042−1050.
(7) Demiray, M.; Tang, X.; Wirth, T.; Faraldos, J. A.; Allemann, R.
K. Angew. Chem., Int. Ed. 2017, 56, 4347−4350.
(8) Elsherbini, M.; Allemann, R. K.; Wirth, T. Chem. - Eur. J. 2019,
25, 12486−12490.
(9) Krishna, S.; Woodrow, C. J.; Staines, H. M.; Haynes, R. K.;
Mercereau-Puijalon, O. Trends Mol. Med. 2006, 12, 200−205.
(10) Eckstein-Ludwig, U.; Webb, R. J.; van Goethem, I. D. A.;
East, J. M.; Lee, A. G.; Kimura, M.; O’Neill, P. M.; Bray, P. G.;
Ward, S. A.; Krishna, S. Nature 2003, 424, 957−961.
(11) Bryant, L.; Flatley, B.; Patole, C.; Brown, G. D.; Cramer, R.
BMC Plant Biol. 2015, 15, 175.
Structure of 2 (CIF)
Structure of 22 (CIF)
Structure of 23 (CIF)
(12) Brown, G. D.; Sy, L.-K. Tetrahedron 2004, 60, 1139−1159.
(13) Sy, L.-K.; Brown, G. D. Tetrahedron 2002, 58, 897−908.
(14) Sy, L.-K.; Brown, G. D. Tetrahedron 2002, 58, 909−923.
(15) Acton, N.; Roth, R. J. J. Org. Chem. 1992, 57, 3610−3614.
(16) Varela, K.; Arman, H. D.; Yoshimoto, F. K. J. Nat. Prod. 2020,
83, 66−78.
AUTHOR INFORMATION
Corresponding Author
Francis K. Yoshimoto − Department of Chemistry, The
University of Texas at San Antonio (UTSA), San Antonio,
Texas 78249-0698, United States; orcid.org/0000-
0002-2308-2999; Email: francis.yoshimoto@utsa.edu
■
(17) Rej, R. K.; Acharyya, R. K.; Nanda, S. Tetrahedron 2016, 72,
4931−4937.
Authors
(18) Lin, A. J.; Klayman, D. L.; Hoch, J. M.; Silverton, J. V.;
George, C. F. J. Org. Chem. 1985, 50, 4504−4508.
(19) Adam, W.; Krebs, O.; Orfanopoulos, M.; Stratakis, M.;
Vougiokalaki, G. C. J. Org. Chem. 2003, 68, 2420−2425.
Kaitlyn Varela − Department of Chemistry, The University of
Texas at San Antonio (UTSA), San Antonio, Texas
78249-0698, United States
1983
https://doi.org/10.1021/acs.jnatprod.1c00246
J. Nat. Prod. 2021, 84, 1967−1984

Journal of Natural Products
pubs.acs.org/jnp
Article
(20) Singleton, D. A.; Hang, C.; Szymanski, M. J.; Meyer, M. P.;
Leach, A. G.; Kuwata, K. T.; Chen, J. S.; Greer, A.; Foote, C. S.;
et al. J. Am. Chem. Soc. 2003, 125, 1319−1328.
(21) Alberti, M. N.; Orfanopoulos, M. Chem. - Eur. J. 2010, 16,
9414−9421.
(22) Harding, L. B.; Goddard, W. A. J. Am. Chem. Soc. 1980, 102,
439−449.
(23) SCALE3 ABSPACKAn Oxford Diffraction program (1.0.4,
g.); Oxford Diffraction Ltd., 2005.
(24) Dolomanov, O. V.; Bourhis, L. J.; Gildea, R. J.; Howard, J. A.
K.; Puschmann, H. J. Appl. Crystallogr. 2009, 42, 339−341.
(25) Sheldrick, G. M. Acta Crystallogr., Sect. A: Found. Adv. 2015,
A71, 3−8.
(26) Sheldrick, G. M. Acta Crystallogr., Sect. A: Found. Crystallogr.
2008, A64, 112−122.
1984
https://doi.org/10.1021/acs.jnatprod.1c00246
J. Nat. Prod. 2021, 84, 1967−1984