Spiroadamantyl 1,2,4-trioxolane, 1,2,4-trioxane, and 1,2,4-trioxepane pairs: Relationship between peroxide bond iron(II) reactivity, heme alkylation efficiency, and antimalarial activity

Article abstract of DOI:10.1016/j.bmcl.2009.07.013

These data suggest that iron(II) reactivity for a set of homologous spiroadamantyl 1,2,4-trioxolane, 1,2,4-trioxane, and 1,2,4-trioxepane peroxide heterocycles is a necessary, but insufficient, property of animalarial peroxides. Heme alkylation efficiency appears to give a more accurate prediction of antimalarial activity than FeSO₄-mediated reaction rates, suggesting that antimalarial activity is not merely dependent on peroxide bond cleavage, but also on the ability of reactive intermediates to alkylate heme or other proximal targets.

Full text of DOI:10.1016/j.bmcl.2009.07.013

Bioorganic & Medicinal Chemistry Letters 19 (2009) 4542–4545
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Spiroadamantyl 1,2,4-trioxolane, 1,2,4-trioxane, and 1,2,4-trioxepane pairs:
Relationship between peroxide bond iron(II) reactivity, heme alkylation
efficiency, and antimalarial activity
Xiaofang Wang a, Darren J. Creek b, Charles E. Schiaffo c, Yuxiang Dong a, Jacques Chollet d,
Christian Scheurer d, Sergio Wittlin d, Susan A. Charman b, Patrick H. Dussault c,
James K. Wood e, Jonathan L. Vennerstrom a,
*
a College of Pharmacy, University of Nebraska Medical Center, 986025 Nebraska Medical Center, Omaha, NE, USA
b Centre for Drug Candidate Optimisation, Monash Institute of Pharmaceutical Sciences, Monash University (Parkville Campus), 381 Royal Parade, Parkville, Victoria 3052, Australia
c Department of Chemistry, University of Nebraska-Lincoln, Lincoln, NE, USA
d Swiss Tropical Institute, Socinstrasse 57, CH-4002 Basel, Switzerland
e Department of Chemistry, University of Nebraska at Omaha, Omaha, NE, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
These data suggest that iron(II) reactivity for a set of homologous spiroadamantyl 1,2,4-trioxolane, 1,2,4-
trioxane, and 1,2,4-trioxepane peroxide heterocycles is a necessary, but insufficient, property of anima-
larial peroxides. Heme alkylation efficiency appears to give a more accurate prediction of antimalarial
activity than FeSO4-mediated reaction rates, suggesting that antimalarial activity is not merely depen-
dent on peroxide bond cleavage, but also on the ability of reactive intermediates to alkylate heme or
other proximal targets.
Received 29 May 2009
Revised 20 June 2009
Accepted 2 July 2009
Available online 9 July 2009
Keywords:
Ó 2009 Elsevier Ltd. All rights reserved.
1,2,4-Trioxolane
1,2,4-Trioxane
1,2,4-Trioxepane
Peroxide
Antimalarial
Artemisinin
The semisynthetic artemisinins rapidly reduce parasite burden
and are particularly effective when used in a 3-day artemisinin
combination treatment (ACT) regimen.1 One hypothesis2–4 that ac-
counts for the antimalarial specificity of artemisinin5 is that the
peroxide bond undergoes reductive activation by heme released
by parasite hemoglobin digestion to produce carbon-centered free
radicals or carbocations that alkylate heme6 or parasite proteins.
Although the 1,2,4-trioxane heterocycle in artemisinin (Fig. 1) is
the critical pharmacophore, its presence alone7 is insufficient for
high antimalarial activity as most synthetic 1,2,4-trioxanes are less
active than artemisinin.8 The discovery9 of ozonide (1,2,4-trioxo-
lane) drug development candidate OZ277 (arterolane) ( Fig. 1),
now entering Phase III clinical trials in the form of an arterolane
maleate + piperaquine phosphate combination, prompted us to
investigate a set of structurally related 1,2,4-trioxolanes 1, 1,2,4-
trioxanes 2, and 1,2,4-trioxepanes 310 (Fig. 2) to better understand
the relationship between peroxide bond iron(II) reactivity and
antimalarial activity for these homologous peroxide heterocycles.
Trioxolanes 1b11 and 1c were obtained by Griesbaum coozonol-
ysis12 of the corresponding oxime ether/ketone13,14 pairs 4/5 and
6/7 in low yields (Scheme 1). Trioxolane 1a was similarly obtained
as previously described.15
Trioxanes 2b and 2c were prepared as shown in Scheme 2. In
the presence of catalytic p-toluenesulfonic acid (PTSA), 2b was
readily formed by condensation of b-hydroperoxy alcohol 816 with
5. Attempts to perform the corresponding reaction of b-hydroper-
oxy alcohol 916 with 5 provided only small quantities of trioxane
2c accompanied by a complex mixture of decomposition products.
H
O
O
O
O
O
O
H
NH NH2
O
O
O
OZ277 (Arterolane)
Figure 1. Artemisinin and OZ277.
Artemisinin
* Corresponding author. Tel.: +1 402 559 5362; fax: +1 402 559 9543.
E-mail address: jvenners@unmc.edu (J.L. Vennerstrom).
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.07.013

X. Wang et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4542–4545
4543
R1
R1
R2
O
R1
R2
O
O
O
O
O
O
O
1a, 2a, 3a R1,R2 = (CH2)5
1b, 2b, 3b R1 = C6H5, R2 = H
1c, 2c R1 = C6H5, R2 = CH3
R2
O
1
2
3
Figure 2. Trioxolanes 1, trioxanes 2, and trioxepanes 3.
in the presence of 10–20% CSA in CH2Cl2 afforded the desired 2c
(Scheme 2). Trioxane 2a was prepared as previously described.20
Trioxepanes 3a and 3b were prepared via parallel routes
(Scheme 3) beginning with alcohol 1321 and cinnamyl alcohol
(14). The corresponding acetates (15, 16) underwent successive
Co-mediated dioxygenation22 and chemoselective reduction to fur-
nish triethylsilyl peroxyalcohols 17 and 18.23 The peroxyalcohols
underwent HF-mediated condensation with ketone 5 to produce
3a and 3b.
N
H
OMe
OMe
O3
+
1b, 13%
1c, 37%
O
O
4
6
5
O3
+
N
Me
In vitro and in vivo antimalarial activities9 (Table 1) were mea-
sured using the chloroquine-resistant K1 and chloroquine-sensi-
tive NF54 strains of Plasmodium falciparum and Plasmodium
berghei-infected mice. Several observations arise from the antima-
larial data. First, trioxolane 1b was more than two orders of mag-
7
Scheme 1.
Characterization of the reaction products indicated that b-hydro-
peroxy alcohol 9 underwent acid-catalyzed O–O bond (pathway
I) and C–O bond (pathway II) heterolysis followed by rearrange-
ment. Formation of ketone 7 (pathway I) can be rationalized by a
Grob-type fragmentation or by migration of the hydroxymethyl
group.17 This contrasts with the phenyl migration seen in the
acid-catalyzed decomposition of cumene hydroperoxide to form
phenol and acetone.18 Aldehyde 11 may arise from the formation
and rearrangement of a relatively stable benzylic tertiary carbocat-
ion in the transition state for fragmentation, with the co-generated
hydrogen peroxide producing lactone 12 via an acid-catalyzed Bae-
yer–Villiger oxidation of adamantanone Thus, acid-catalyzed het-
erolytic peroxide bond fragmentation (pathway II) was the
dominant reaction course for 9. Therefore, we tried protecting
the hydroperoxide functional group of 9 to suppress the fragmen-
tation.19 Treatment of 9 with N,O-bis-(trimethylsilyl)acetamide
(BSA) in CH2Cl2 gave the corresponding bis(trimethylsilylated)
product 10 in high yield. Subsequent condensation of 10 with 5
nitude less potent than trioxolanes 1a and 1c. Because it has a a-
H atom, 1b may lack the necessary chemical stability24 required
for antimalarial activity. This contrasts with the good antimalarial
1) Co(acac)2
R1
OH
R1
OAc
Ac2O,
DMAP
Et3SiH, O2
R2
R2
pyridine
2) DIBAL-H
CH2Cl2
13, R1,R2= (CH2)5
14, R1=H,R2=Ph
15, 80%
16, 89%
OOSiEt3
R1
5, HF
3a, 15%
3b, 21%
R2
OH
CH2Cl2
17, 14% (2 steps)
18, 41% (2 steps)
Scheme 3.
O OH
PTSA
+
5
2b, 73%
H
OH
8
Me
H+
H
H
–H2O
O O
H+
CH2O +
7, 10% +
Pathway I
OH
O OH
Me
Me
Me
1,2-H shift
H+
+
+
5
Me
OH
OH
OH
O
H
H
9
H
H
Me
H
Pathway II
H+
O OH
+
11, 19%
O
O
BSA
OH
adamantanone
H2O2
O OTMS
CSA
12, 28%
+
2c, 64%
5
Me
OTMS
10, 91%
Scheme 2.

4544
X. Wang et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4542–4545
Table 1
Activity of 1–3 against P. falciparum in vitro and P. berghei in vivo
Compd
IC50 (ng/ml)a
Activityb (%) po
0
K1
—
NF54
—
NONE
1ac
1b
1c
0.97
400
2.9
1.4
>1000
3.4
>99.99
25
99.0
2ad
2b
2c
54
9.1
19
44
19
22
0
96
0
Figure 3. LUMO (blue) mapped on isodensity surfaces of 1a and 3a.
3a
NDe
NDe
1.6
>1000
>1000
2.8
0
NDe
98
not accessible from either direction, consistent with the low reac-
tivity of 3a with iron(II). The greater steric crowding in 3a is due to
the methylene groups of the trioxepane ring decreasing the angle
between the adamantane and cyclohexane rings. This lack of
iron-mediated reactivity for trioxepane 3a is consistent with its
lack of antimalarial activity; similar arguments have been put forth
to explain the low antimalarial activity of other sterically con-
gested 1,2,4-trioxepanes.29
3b
ARTc,f
a
Mean from n = 2–3 against chloroquine-resistant (K1) and chloroquine-sensi-
tive (NF54) strains of P. falciparum.
b
Groups of three P. berghei-infected NMRI mice were treated orally one day post-
infection with trioxolanes (100 mg/kg) dissolved or suspended in 3% ethanol and 7%
Tween 80. Antimalarial activity was measured by percent reduction in parasitemia
on day 3 post-infection. Individual measurements generally differed by less than
10%.
We have previously shown30 a good correlation between
in vitro antimalarial activity and heme alkylation efficiency for a
series of 22 trioxolanes. We measured the efficiency of heme alkyl-
ation for these trioxolanes, trioxanes and trioxepanes using these
same experimental conditions.30 Significant heme alkylation was
observed within 30 s for the active 1c (67 2%) and moderately ac-
tive 2b (55 4%), and to a slightly lesser extent for the less active
1b (50 1%) and 2a (45 1%); however the extent of heme alkyl-
ation by these compounds was far lower than previously reported
for the highly active trioxolanes 1a (86 0.2%) and OZ277
(83 2%).30 Trioxepanes 3a and 3b mediated very little heme alkyl-
ation (25 1% and 1 0.4%), in agreement with their lack of anti-
malarial activity. The investigation of heme alkylation appears to
give a more accurate prediction of antimalarial activity than inves-
tigation of iron-mediated reaction rates, suggesting that antimalar-
ial activity is not merely dependant on peroxide bond cleavage, but
also on the ability of reactive intermediates to alkylate heme or
other proximal targets. This further supports a likely role for heme
alkylation in the antimalarial mechanism of action of peroxide
antimalarials.6
c
Data from Dong et al.15
Data from Tang et al.20
d
e
Not determined.
ART = artemisinin.
f
activity of the corresponding and presumably more chemically sta-
ble trioxane 2b. Trioxanes 2a and 2c were less potent than the cor-
responding trioxolanes 1a and 1c; neither 2a nor 2c had any
activity in vivo. The less accessible peroxide bond in the trioxane
vs trioxolane heterocycle20 may explain why trioxanes 2a and 2c
are less active than their trioxolane chemical cousins 1a and 1c.
Similarly, the complete lack of activity of trioxepanes 3a and 3b
may arise from peroxide bonds that are too sterically hindered
(vide infra). These data suggest that antimalarial activity for these
spiroadamantyl trioxolanes, trioxanes, and trioxepanes depends on
a finely tuned balance between peroxide bond shielding and acces-
sibility. To provide mechanistic insight into this hypothesis, we
examined the reactivity of several of these compounds with iron(II)
in the form of FeSO4 and heme.
The reactivity of 1–3 with inorganic ferrous iron was deter-
mined using previously established standardized conditions.25 In
these experiments, pseudo-first order reaction rate constants for
1b, 2a, 2b, 3a, and 3b (0.03 mM) with FeSO4 (3 mM) in 50% aceto-
nitrile/water at 37 °C under an argon atmosphere were deter-
mined. These data (n = 3) were corrected for non-specific
degradation in iron-free controls. Under these experimental condi-
tions, the previously determined pseudo-first order rate constants
(k) for 1a and artemisinin were 0.41 0.02 hꢀ126 and
Acknowledgments
We thank Josefina Santo Tomas and Christopher Snyder for
assistance in performing the in vivo antimalarial assays. This inves-
tigation received financial support from Medicines for Malaria
Venture (MMV).
References and notes
0.054 0.006 hꢀ1 25
The weakly active 1b reacted considerably
.
more rapidly (k = 1.77 0.06 hꢀ1) with iron(II) than did 1a con-
firming our previous observations26,27 that chemically reactive
peroxides do not necessarily possess high antimalarial activities.
Trioxane 2b reacted with iron(II) at a rate (k = 0.12 0.03 hꢀ1) be-
tween that of 1a and artemisinin, whereas trioxepane 3b reacted
much more slowly (k = 0.021 0.006 hꢀ1), and neither 2a nor 3a
underwent statistically significant degradation over the 24 h time
1. White, N. J. Science 2008, 320, 330.
2. Jefford, C. W. Curr. Med. Chem. 2001, 8, 1803.
3. Meshnick, S. R. Int. J. Parasitol. 2002, 32, 1655.
4. O’Neill, P. M.; Posner, G. H. J. Med. Chem. 2004, 47, 2945.
5. Kaiser, M.; Wittlin, S.; Nehrbass-Stuedli, A.; Dong, Y.; Wang, X.; Hemphill, A.;
Matile, H.; Brun, R.; Vennerstrom, J. L. Antimicrob. Agents Chemother. 2007, 51,
2991.
6. Robert, A.; Benoit-Vical, F.; Claparols, C.; Meunier, B. Proc. Natl. Acad. Sci. U.S.A.
2005, 102, 13676.
7. Casteel, D. A.; Peri, S. P.; Gerena, L. Bioorg. Med. Chem. Lett. 1993, 3, 1707.
8. Tang, Y.; Dong, Y.; Vennerstrom, J. L. Med. Res. Rev. 2004, 24, 425.
9. Vennerstrom, J. L.; Arbe-Barnes, S.; Brun, R.; Charman, S. A.; Chiu, F. C. K.;
Chollet, J.; Dong, Y.; Dorn, A.; Hunziker, D.; Matile, H.; McIntosh, K.;
Padmanilayam, M.; Santo Tomas, J.; Scheurer, C.; Scorneaux, B.; Tang, Y.;
Urwyler, H.; Wittlin, S.; Charman, W. N. Nature 2004, 430, 900.
10. Adamantane-2-spiro-30-50-phenyl-10,20,40-trioxolane (1b):11 1H NMR (500 MHz,
CDCl3) d 1.70–1.93 (m, 9H), 2.03–2.16 (m, 4H), 2.22–2.26 (m, 1H), 6.09 (s, 1H),
7.36–7.46 (m, 3H), 7.50–7.56 (m, 2H). 13C NMR (125.7 MHz, CDCl3) d 26.5, 26.8,
34.2, 34.6, 34.9, 35.7, 35.8, 36.78, 36.8, 103.7, 113.0, 127.9, 128.5, 130.3, 132.8.
Anal. Calcd for C17H20O3: C, 74.97; H, 7.40. Found: C, 74.94; H, 7.19.
Adamantane-2-spiro-30-50-methyl-50-phenyl-10,20,40-trioxolane (1c): mp 70–
course studied (k = 0.011 0.013 hꢀ1 and 0.017 0.012 hꢀ1
,
respectively).
As has been previously demonstrated for 1a versus 2a,20 the dif-
ference in iron-mediated reactivity between 1a and 3a can be ex-
plained by considering peroxide bond accessibility of the more
reactive equatorial peroxide bond conformers28 as determined by
molecular modeling (Spartan’08, Wavefunction, Inc.). As shown
in Figure 3, the peroxide bond LUMO in 1a is easily accessible from
the cyclohexane side, whereas the peroxide bond LUMO in 3a is

X. Wang et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4542–4545
4545
72 °C. 1H NMR (500 MHz, CDCl3) d 1.72 (s, 3H), 1.56–2.20 (m, 14H), 7.28–7.38
(m, 3H), 7.50–7.56 (m, 2H). 13C NMR (125.7 MHz, CDCl3) d 25.9, 26.5, 26.9, 34.5,
34.75, 34.78, 35.3, 35.4, 36.6, 36.8, 108.7, 112.9, 125.2, 128.06, 128.11, 142.6.
Anal. Calcd for C18H22O3: C, 75.50; H, 7.74. Found: C, 75.58; H, 7.63.
Adamantane-2-spiro-30-60-phenyl-10,20,40-trioxane (2b): mp 110–111 °C. 1H
14. Dong, Y.; Vennerstrom, J. L. J. Org. Chem. 1998, 63, 8582.
15. Dong, Y.; Chollet, J.; Matile, H.; Charman, S. A.; Chiu, F. C. K.; Charman, W. N.;
Scorneaux, B.; Urwyler, H.; Santo Tomas, J.; Scheurer, C.; Snyder, C.; Dorn, A.;
Wang, X.; Karle, J. M.; Tang, Y.; Wittlin, S.; Brun, R.; Vennerstrom, J. L. J. Med.
Chem. 2005, 48, 4953.
16. Ogata, Y.; Sawaki, Y.; Shimizu, H. J. Org. Chem. 1978, 43, 1760.
17. Kagan, J.; Przybytek, J. T. Tetrahedron 1973, 29, 1163.
18. Kharasch, M. S.; Fono, A.; Nudenberg, W. J. Org. Chem. 1950, 15, 748.
19. Kim, H.-S.; Tsuchiya, K.; Shibata, Y.; Wataya, Y.; Ushigoe, Y.; Araki, M.; Nojima,
M.; McCullough, K. J. J. Chem. Soc., Perkin Trans. 1 1999, 1867.
20. Tang, Y.; Dong, Y.; Wang, X.; Sriraghavan, K.; Wood, J. K.; Vennerstrom, J. L.
J. Org. Chem. 2005, 70, 5103.
NMR (500 MHz, CDCl3)
d
1.60–2.16 (m, 13H), 3.07 (br, s, 1H), 3.83 (dd,
J1 = 12 Hz, J2 = 3.5 Hz, 1H), 4.12 (dd, J1 = 12 Hz, J2 = 10.5 Hz, 1H), 5.30 (dd,
J1 = 10.5 Hz, J2 = 3.5 Hz, 1H), 7.30–7.40 (m, 5H). 13C NMR (125.7 MHz, CDCl3) d
27.16, 27.17, 29.4, 33.0, 33.3, 33.5, 33.7, 36.3, 37.2, 62.7, 81.3, 104.5, 127.3,
128.7, 129.1, 135.3. Anal. Calcd for C18H22O3: C, 75.50; H, 7.74. Found: C, 75.76;
H, 7.60; Adamantane-2-spiro-30-60-methyl-60-phenyl-10,20,40-trioxane (2c): mp
61–62 °C, 1H NMR (500 MHz, CDCl3) d 1.44–2.04 (br m, 16H), 3.02 (br s 1H),
3.75 (br s 1H), 4.16 (br s 1H), 7.25–7.60 (m, 5H), 13C NMR (125.7 MHz, CDCl3) d
21.9 (br), 25.4 (br), 27.14, 27.17 (br), 29.7 (br), 33.2 (br), 33.3, 33.5, 36.2 (br),
37.2, 65.7 (br), 79.8 (br), 104.4, 124.7 (br), 127.6 (br), 128.3, 141.6 (br). Anal.
Calcd for C19H24O3: C, 75.97; H, 8.05. Found: C, 76.16; H, 8.24; Adamantane-2-
spiro-30-70-cyclohexyl-10,20,40-trioxepane (3a): 1H NMR (400 MHz, CDCl3) d
1.30–1.86 (20H), 1.89–2.02 (m, 3H), 2.02–2.15 (m, 2H), 2.40 (br s, 1H), 3.69 (dt,
JHa = 3.7, JHb = 3.7, 1H), 3.69 (dt, J = 12.4, 3.7, 1H), 3.90 (apparent pentet, J = 6.2,
probably dt, J = 12.2, 6.2, 1H), 13C NMR (100.6 MHz, CDCl3) d 22.0, 22.4, 26.0,
27.3, 32.9, 33.6, 33.7, 33.8, 33.9, 34.2, 34.6, 35.7, 37.5, 42.3, 58.1, 81.7, 107.5,
HRMS-FAB calcd for C18H29O3 (MH+): 293.2117; found 293.2103. Adamantane-
21. Bourhani, Z.; Malkov, A. Chem. Commun. 2005, 4592.
22. Tokuyasu, T.; Kunikawa, S.; Masuyama, A.; Nojima, M. Org. Lett. 2002, 4, 3595.
23. Jin, H.; Liu, H.; Zhang, Q.; Wu, Y. J. Org. Chem. 2005, 70, 4240.
24. Kornblum, N.; DeLaMare, H. E. J. Am. Chem. Soc. 1951, 73, 880.
25. Creek, D. J.; Chiu, F. C. K.; Prankerd, R. J.; Charman, W. N.; Charman, S. A.
J. Pharm. Sci. 2005, 94, 1820.
26. Creek, D. J.; Charman, W. N.; Chiu, F. C. K.; Prankerd, R. J.; McCullough, K. J.;
Dong, Y.; Vennerstrom, J. L.; Charman, S. A. J. Pharm. Sci. 2007, 96, 2945.
27. Wang, X.; Dong, Y.; Wittlin, S.; Creek, D.; Chollet, J.; Charman, S. A.; Santo
Tomas, J.; Scheurer, C.; Snyder, C.; Vennerstrom, J. L. J. Med. Chem. 2007, 50,
5840.
2-spiro-30-70-phenyl-10,20,40-trioxepane (3b): 1H NMR (400 MHz, CDCl3)
d
1.59–1.75 (m, 6H), 1.79–1.86 (m, 2H), 1.89–2.14 (m, 6H), 2.22–2.42 (m, 2H),
3.85 (dt, J = 12.2, 3.4, 1H), 4.13 (app t, J = 11.3, 1H), 5.14 (dd, J = 11.5, 3.5, 1H),
7.29–7.40 (m, 5H), 13C NMR (75.5 MHz, CDCl3) d 27.1, 27.2, 33.1, 33.6, 33.9,
34.1, 34.7, 35.0, 37.5, 39.4, 60.0, 76.6, 86.1, 108.6, 127.3, 128.4, 128.5, 138.5,
HRMS-FAB calc for C19H25O3 (MH+): 301.1804; found 301.1809.
11. Keul, H. Chem. Ber. 1975, 108, 1207.
28. The equatorial peroxide conformer provides greater access to the peroxide
bond from the cyclohexane side. For 1a and 3a, the energy difference between
the two chair conformers
respectively.
(D
E = Eeq ꢀ Eax
) is 0.280 and 1.71 kcal/mol,
29. Amewu, R.; Stachulski, A. V.; Berry, N. G.; Ward, S. A.; Davies, J.; Labat, G.;
Rossignol, J.-F.; O’Neill, P. M. Bioorg. Med. Chem. Lett. 2006, 16, 6124.
12. Griesbaum, K.; Liu, X.; Kassiaris, A.; Scherer, M. Liebigs Ann. Recl. 1997, 1381.
13. Konakahara, T.; Matsuki, M.; Sugimoto, S.; Sato, K. J. Chem. Soc., Perkin Trans. 1
1987, 1489.
30. Creek, D. J.; Charman, W. N.; Chiu, F. C. K.; Prankerd, R. J.; Dong, Y.;
Vennerstrom, J. L.; Charman, S. A. Antimicrob. Agents Chemother. 2008, 52, 1291.