A post-PKS oxidation of the amphotericin B skeleton predicted to be critical for channel formation is not required for potent antifungal activity

Article abstract of DOI:10.1021/ja075739o

The clinically vital antimycotic agent amphotericin B represents the archetypal example of a channel-forming small molecule. The leading model for self-assembly of the amphotericin B channel predicts that the C(41) carboxylate and the C(3′) ammonium ions form intermolecular salt bridges/hydrogen bonds that are critical for stability. We herein report a flexible degradative synthesis pathway that enables the removal of either or both of these groups from amphotericin B. We further demonstrate with extensive NMR experiments that deleting these groups does not alter the conformation of the polyene macrolide skeleton. As predicted by the leading model, amphotericin B derivatives lacking the mycosamine sugar that contains the C(3′) ammonium ion are completely inactive against Saccharomyces cerevisiae. However, strikingly-and in strong contradiction with the current model-the amphotericin B derivative lacking the C(41) carboxylate is at least equipotent to the natural product. Collectively, these findings demonstrate that the leading model for the mechanism of action of amphotericin B must be significantly revised-either the C(41) carboxylate is not required for channel formation or channel formation is not required for antifungal activity. Copyright

Full text of DOI:10.1021/ja075739o

Published on Web 10/23/2007
A Post-PKS Oxidation of the Amphotericin B Skeleton Predicted to be Critical
for Channel Formation Is Not Required for Potent Antifungal Activity
Daniel S. Palacios, Thomas M. Anderson, and Martin D. Burke*
Roger Adams Laboratory, Department of Chemistry, UniVersity of Illinois at UrbanasChampaign,
Urbana, Illinois 61801
Received August 9, 2007; E-mail: burke@scs.uiuc.edu
The leading model for the antifungal action of amphotericin B
(
AmB, 1) involves its self-assembly into a membrane-spanning ion
1
channel. This natural product thus represents a potential prototype
for small molecules with the capacity to perform ion channel-like
functions in living systems. Efforts to harness this potential and/or
improve the notoriously poor therapeutic index of this clinically
2
vital antimycotic would benefit from a molecular-level understand-
ing of this channel activity.
Although the evidence that AmB can self-assemble in lipid
membranes to form discrete ion conducting channels is strong,1
the molecular architecture of this channel assemblage and its role
,3
4
in antifungal activity remain poorly understood. Despite this, the
5
leading “barrel-stave” model is an often cited textbook classic
Figure 1. (A) A bird’s eye view of the current “barrel stave” model for
the AmB channel. Salt bridges and/or hydrogen bonds (dashed lines)
between oppositely charged C(41) carboxylate and C(3′) ammonium ions
are predicted to be critical for channel stabilization. (B) These two functional
groups are installed as post-PKS modifications of the macrolide skeleton.
6
(
Figure 1A). Extensive computer modeling studies predict that this
7a
complex is stabilized by a ring of salt bridges and/or hydrogen
bonds7b,c at the channel periphery between oppositely charged C(41)
carboxylate and C(3′) ammonium ions. Conspicuously, these two
functional groups are installed biosynthetically as post-polyketide
synthase (PKS) modifications of the macrolide skeleton, that is, a
P450-mediated oxidation of the C(41) methyl group and a glycosyl
Scheme 1 a
transferase-mediated attachment of mycosamine to the C(19) alcohol
Figure 1B).8
,9
(
Many interesting studies have probed the action of AmB via
covalent modification of the C(41) carboxylic acid and/or C(3′)
amine.4 However, the self-assembly of small molecules can be
e,10
11
exquisitely sensitive to steric effects, and this phenomenon may
complicate this experimental approach. We herein report an
alternative strategy that involves synthetically deleting chemical
groups appended to the macrolide skeleton and determining the
functional consequences.12 This approach has led to the striking
observation that, contrary to the current channel model (Figure 1A),
oxidation at C(41) is not required for potent antifungal actiVity.
The synthetic manipulation of AmB is challenging because of
its sensitivity to light, oxygen, and acid as well as its minimal
solubility in most organic solvents and water. Nevertheless, we
ultimately developed a flexible degradative pathway that enables
the synthetic reversal of either13 or both of the two post-PKS
modifications predicted to be critical for self-assembly of the AmB
channel (Scheme 1). Synthesis of the novel MeAmdeB 2 (Figure
a
Reagents and conditions: (a) (i) Fmoc-succinimide, pyridine, DMF/
MeOH 2:1, 23 °C, 12 h; (ii) CSA, THF/MeOH 1:1, 0 °C, 1 h, 90% (two
steps); (b) TESOTf, 2,6-lutidine, hexanes, 0 °C, 3 h, 96%; (c) 2-thiopyridyl
chloroformate, Et3N, Et2O, 0 °C, 30 min, 91%; (d) LiBH4, Et2O, 23 °C, 2
h, 88%; (e) I2, PPh3, imidazole, THF, 0 °C, 1 h, 78%; (f) DDQ, CaCO3,
THF, 23 °C, 10 min, 67%; (g) (S)-CBS oxazaborolidine, Me2S‚BH3, CH2Cl2,
10 °C, 30 min, 6:1 dr, 79%; (h) NaBH4, DMPU, 23 °C, 6 h, 78%; (i) (i)
HF/pyridine, THF/pyridine 3:2, 0 °C, 6 h; (ii) AcOH/H2O/THF 1:1:2,
3 °C, 30 min; HPLC, 38% (two steps); (j) allyl bromide, i-Pr2NEt, DMF,
3 °C, 8 h, 86%; (k) DDQ, CaCO3, THF, 23 °C, 20 min, 65%; (l) NaBH4,
THF/MeOH 3:1, 0 °C, 30 min, >20:1 dr., 77%; (m) HF/pyridine, THF/
pyridine 5:3, 0 f 23 °C, 6 h, 56%; (n) CSA, THF/H2O 2:1, 23 °C, 5 h;
HPLC, 81%; (o) Pd(PPh3)4, thiosalicylic acid, THF, 23 °C, 13 h; HPLC,
50%; (p) HF/pyridine, THF/pyridine 5:3, 0 f 23 °C, 6.5 h, 73%; (q) NaBH4,
DMSO, 23 °C, 8 h, 58%; (r) (i) CSA, THF/H2O 2:1, 23 °C, 30 min; (ii)
piperidine, DMSO/MeOH 4:1, 23 °C, 3 h; HPLC, 56% (two steps).
1
B) commenced with the conversion of AmB into the suitably
13b
protected and more soluble nonasilylated N-Fmoc methyl ketal 7.
The C(41) carboxylic acid was then selectively reduced to the
-
corresponding primary alcohol via the intermediacy of a 2-pyridi-
nethiol ester.14 Subsequent iodination with PPh
/I
15
yielded the
3
2
2
2
desired C(41) iodomethyl derivative 8. Oxidative deglycosylation
of this intermediate using DDQ16 in the presence of CaCO
smoothly generated enone 9. Reduction of the C(19) ketone with
NaBH
/MeOH resulted in a ∼2:1 mixture of diastereomers, while
use of the (S)-CBS oxazaborolidine catalyst provided the desired
9-R isomer in a synthetically useful 6:1 dr (Supporting Informa-
tion). A subsequent reductive cleavage of the C(41) iodide was
3
4
17
1
18
achieved with NaBH
0. Global desilylation with HF/pyridine, hydrolysis of the methyl
ketal, and preparative high performance liquid chromatography
4
in DMPU to yield advanced intermediate
(HPLC) yielded diastereomerically pure MeAmdeB 2. The fun-
neling of intermediates 7 and 8 into modified versions of this
flexible pathway enabled the preparation of the remaining two
1
13804
9
J. AM. CHEM. SOC. 2007, 129, 13804-13805
10.1021/ja075739o CCC: $37.00 © 2007 American Chemical Society

C O M M U N I C A T I O N S
Rienstra, B. Wylie, V. Mainz, P. Molitor, F. Lin, F. Delaglio, and
J. Baudry for assistance with NMR studies, NMRPipe, MOE, and
Figure 1A, P. Orlean, J. Grimme, and R. Hellman for assistance
with yeast assays, and J. Morrissey, S. Smith, K. Suslick, G. Fujii,
and S.M. Chiang for assistance with liposome studies.
Supporting Information Available: Detailed synthesis, NMR, and
assay procedures, spectral data, and spectra; full citations for refs 4e,
7b, 8a,b, 10a,b,e, 13a-d, 15, 18, 19a, and 24. This material is available
free of charge via the Internet at http://pubs.acs.org.
Figure 2. Superposition of the ground-state conformation of the macro-
lactone skeletons of compounds 1-4 (or their more soluble analogues; see
Supporting Information for details).
References
(
1) Reviews: (a) Bolard, J. Biochim. Biophys. Acta 1986, 864, 257-304. (b)
Hartsel, S. D.; Hatch, C.; Ayenew, W. J. Liposome Res. 1993, 3, 377-
4
08. (c) Cereghetti, D. M.; Carreira, E. M. Synthesis 2006, 6, 914-942.
(
d) Hartsel, S.; Bolard, J. Trends Pharm. Sci. 1996, 17, 445-449.
(
2) Kontoyiannis, D. P.; Lewis, R. E. Lancet 2002, 359, 1135-1144.
(
3) (a) Andreoli, T. E.; Monahan, M. J. Gen. Physiol. 1968, 52, 300-325.
(
b) Cass, A.; Finkelstein, A.; Krespi, V. J. Gen. Physiol. 1970, 56, 100-
1
24. (c) Ermishkin, L. N.; Kasumov, K. M.; Potzeluyev, V. M. Nature
1
976, 262, 698-699. (d) Borisova, M. P.; Ermishkin, L. N.; Silberstein,
A. Y. A. Biochim. Biophys. Acta 1979, 553, 450-459. (e) Mickus, D. E.;
Figure 3. (A) Disc diffusion assay with S. cereVisiae (40 µg of compound
per disc). Similar results were achieved with C. albicans (Supporting
Information). (B) Broth dilution assays; values represent the average of
three experiments.
Levitt, D. G.; Rychnovsky, S. D. J. Am. Chem. Soc. 1992, 114, 359-
3
2
60. (f) Sykora, J.; Yilma, S.; Neely, W. C.; Vodyanoy, V. Langmuir
003, 19, 858-864.
(
4) (a) HsuChen, C. C.; Feingold, D. S. Nature 1974, 251, 656-659. (b) Chen,
W. C.; Chou, D.-L.; Feingold, D. S. Antimicrob. Agents Chemother. 1978,
13, 914-917. (c) Brajtburg, J.; Powderly, W. G.; Kobayashi, G. S.;
Medoff, G. Antimicrob. Agents Chemother. 1990, 34, 183-188. (d) Beggs,
W. H. Antimicrob. Agents Chemother. 1994, 38, 363-364. (e) Zumbuehl,
A.; et al. Angew. Chem., Int. Ed. 2004, 43, 5181-5185.
targeted derivatives AmdeB 313a and MeAmB 413b,c (Scheme 1,
Figure 1B).
Because the AmB framework is known to be quite rigid,19 we
(
5) (a) Finkelstein, A.; Holz, R. In Membranes. Lipid Bilayers and Antibiotics;
Eisenman, G., Ed.; Marcel Dekker, Inc.: New York, 1973; Vol. 2, pp
377-408. (b) De Kruijff, B.; Demel, R. A. Biochim. Biophys. Acta 1974,
postulated that the ground-state conformation would be unchanged
by these appendage deletions, thereby further facilitating the
interpretation of structure/function data generated with this ap-
proach. To confirm this hypothesis, we determined the ground-
state conformation of compounds 1-4 (or more soluble analogues;
see Supporting Information for details) using Monte Carlo methods
constrained by extensive NOESY and phase-sensitive COSY NMR
data processed using amplitude-constrained multiplet evaluation.20
As shown in Figure 2, the conformation of the macrolide skeleton
was unaltered by these appendage deletions (root-mean-square
deviation for all four compounds ) 0.081 Å).
3
39, 57-70. (c) Andreoli, T. E. Ann. N.Y. Acad. Sci. 1974, 235, 448-468.
(
6) Petty, H. R. Molecular Biology of Membranes, Structure and Function;
Plenum Press: New York, 1993.
(7) (a) Khutorsky, V. E. Biochim. Biophys. Acta 1992, 1108, 123-127. (b)
Baginski, M.; et al. Mol. Pharmacol. 1997, 52, 560-570. (c) Baginski,
M.; Resat, H.; Borowski, E. Biochim. Biophys. Acta 2002, 1567, 63-78.
(8) (a) McNamara, C. M.; et al. J. Chem. Soc. Perkin Trans. 1 1998, 83-87.
(b) Caffrey, P.; et al. Chem. Biol. 2001, 8, 713-723.
(9) To the best of our knowledge, these two post-PKS modifications are unique
to the polyene macrolides and they always co-appear.
10) (a) Ch e´ ron, M.; et al. Biochem. Pharmacol. 1988, 37, 827-836. (b)
Matsumori, N.; et al. J. Am. Chem. Soc. 2002, 124, 4180-4181. (c)
Matsumori, N.; Sawada, Y.; Murata, M. J. Am. Chem. Soc. 2005, 127,
(
The impact of deleting these functional groups on antifungal
activity against Saccharomyces cereVisiae was qualitatively evalu-
ated using a disc diffusion assay.21 As shown in Figure 3A,
1
0667-10675. (d) Matsumori, N.; Sawada, Y.; Murata, M. J. Am. Chem.
Soc. 2006, 128, 11977-11984. (e) Umegawa, Y.; et al. Tetrahedron Lett.
2007, 48, 3393-3396.
derivatives 2 and 3, both of which lack the mycosamine appendage,
(11) (a) Mathias, J. P.; Simanek, E. E.; Whitesides, G. M. J. Am. Chem. Soc.
1994, 116, 4326-4340. (b) G o¨ rbitz, C. H.; Etter, M. C. J. Am. Chem.
Soc. 1992, 114, 627-631.
were completely inactive.22 In stark contrast, and counter to the
current channel model, MeAmB 4 was found to be roughly
(12) (a) Lemieux, R. U. Acc. Chem. Res. 1996, 29, 373-380. (b) Tor, Y.
ChemBioChem 2003, 4, 998-1007. (c) Morgan, A. J.; Wang, Y. K.;
Roberts, M. F.; Miller, S. J. J. Am. Chem. Soc. 2004, 126, 15370-15371.
equipotent to the natural product. This striking result was confirmed
quantitatively in a broth dilution assay23 (MIC: AmB ) 2 µM,
(
d) Nam, J.; Shin, D.; Rew, Y.; Boger, D. L. J. Am. Chem. Soc. 2007,
MeAmB ) 1 µM) (Figure 3B). Similar results were observed in
both assays with clinically relevant Candida albicans (Figure 3).
Clearly, post-PKS oxidation of the AmB macrolide at C(41) is not
required for potent antifungal activity.
129, 8747-8755.
(13) (a) In pioneering studies, Nicolaou and co-workers reported an alternative
degradative synthesis of 3: Nicolaou, K. C.; et al. J. Am. Chem. Soc.
1
988, 110, 4660-4672. (b) A degradative synthesis of 4 is described in
the patent literature: Driver, M. J.; et al. U.S. Patent #5,204,330, 1993.
c) MacPherson, D. T.; et al. Rec. AdV. Chem. Antiinfect. Agents 1993,
(
These findings stand in strong contradiction with the current
model for the mechanism of action of AmB (Figure 1A). There
are at least two possible explanations: oxidation at C(41) may not
be required for channel formation and/or channel formation may
2
05-222. (d) An impure sample of 4 was prepared via genetic engineering
of the producer: Carmody, M.; et al. J. Biol. Chem. 2005, 280, 34420-
34426.
(
14) Corey, E. J.; Clark, D. A. Tetrahedron Lett. 1979, 20, 2875-2878.
4
(15) Garegg, P. J.; et al. J. Chem. Soc., Perkin Trans. 1 1980, 2866-2869.
not be required for antifungal activity. An extensive series of
(
16) Kennedy, R. M.; Abiko, A.; Masamune, S. Tetrahedron Lett. 1988, 29,
biophysical studies with compounds 1-4 are planned to distinguish
447-450.
between these possibilities. In preliminary studies using pyranine-
(17) Corey, E. J.; Bakshi, R. K.; Shibata, S.; Chen, C. P.; Singh, V. K. J. Am.
Chem. Soc. 1987, 109, 7925-7926.
impregnated liposomes,24 MeAmB demonstrates membrane-per-
(
18) Hutchins, R. O.; et al. J. Org. Chem. 1978, 43, 2259-2267.
meabilizing activity similar to that of AmB (Supporting Informa-
tion).
(19) (a) Sowinski, P.; et al. Magn. Reson. Chem. 1992, 30, 275-279. (b)
Matsumori, N.; Houdai, T.; Murata, M. J. Org. Chem. 2007, 72, 700-706.
(
(
(
20) Delaglio, F.; Zhengrong, W.; Bax, A. J. Magn. Reson. 2001, 149, 276-
These results also demonstrate that the deletion of appended
functional groups represents a powerful approach for probing the
still poorly understood activity of AmB. The general application
of this strategy to systematically dissect the structure/function
relationships that underlie this potentially prototypical channel-
forming small molecule is currently underway in our laboratories.
2
81.
21) NCCLS Performance Standards for Antimicrobial Disk Susceptibility
Tests, M2-A8 Approved Standard, 8th ed.; Vol. 23, Number 1, 2003.
22) To the best of our knowledge, this is the first demonstration that the sugar
moiety of AmB is required for biological activity. The relative importance
of the three polar functional groups at C(2′), C(3′), and C(4′) remains
unclear.
23) NCCLS Reference Method for Broth Dilution Antifungal Susceptibility
Testing, M27-A2, Approved standard, 2nd ed.; Vol. 22, Number 15, 2002.
24) Fujii, G.; et al. Biochemistry 1997, 36, 4959-4968.
(
Acknowledgment. We gratefully acknowledge Bristol-Myers
Squibb Company for a gift of AmB, and the NIH (GM080436),
Dreyfus Foundation, and UIUC for funding. We also thank C.
(
JA075739O
J. AM. CHEM. SOC.
9
VOL. 129, NO. 45, 2007 13805