Isoapoptolidin: structure and activity of the ring-expanded isomer of apoptolidin.

Article abstract of DOI:10.1021/ol0266222

[formula: see text] Apoptolidin (1) is a novel oncolytic lead that induces apoptosis in transformed cell lines with exceptional selectivity. We report the isolation and characterization of a ring-expanded macrolide isomer of apoptolidin: isoapoptolidin (2). The solution conformation of isoapoptolidin is described. The rate of isomerization was measured under biologically relevant conditions and found to approach equilibrium within the time frame of most cell-based assays. Isoapoptolidin's ability to inhibit mitochondrial F0F1-ATPase is over 10-fold less than that of apoptolidin.

Full text of DOI:10.1021/ol0266222

ORGANIC
LETTERS
2002
Vol. 4, No. 22
3819-3822
Isoapoptolidin: Structure and Activity of
the Ring-Expanded Isomer of
Apoptolidin
,†
Paul A. Wender,* Aaron V. Gulledge,† Orion D. Jankowski,† and Haruo Seto‡
Department of Chemistry, Stanford UniVersity, Stanford, California 94305-5080, and
Molecular and Cellular Biosciences, UniVersity of Tokyo, Japan
wenderp@stanford.edu
Received July 27, 2002
ABSTRACT
Apoptolidin (1) is a novel oncolytic lead that induces apoptosis in transformed cell lines with exceptional selectivity. We report the isolation
and characterization of a ring-expanded macrolide isomer of apoptolidin: isoapoptolidin (2). The solution conformation of isoapoptolidin is
described. The rate of isomerization was measured under biologically relevant conditions and found to approach equilibrium within the time
frame of most cell-based assays. Isoapoptolidin’s ability to inhibit mitochondrial F0F1-ATPase is over 10-fold less than that of apoptolidin.
In 1997, Seto and co-workers1 reported the structure of
apoptolidin (1), which was isolated from Nocardiopsis sp.
in a screening program designed to identify new compounds
that selectively induce apoptosis in E1A-transformed cells.
In the National Cancer Institute’s 60 human cancer cell line
assay, apoptolidin was found to be among the top 0.1% most
selective agents of the more than 37 000 compounds analyzed
to date.2,3 The ability of apoptolidin to induce apoptosis in
transformed cells with such remarkable selectivity represents
an exciting new lead strategy for the treatment of cancer.
The Khosla group recently identified mitochondrial F0F1-
ATPase as a cellular target of apoptolidin and further
demonstrated that cell lines that were unresponsive to
treatment with apoptolidin could be sensitized by cotreatment
with either oxamate or 2-deoxyglucose.3 Despite this, the
basis for apoptolidin’s selectivity against different cell lines
is not well understood.
Apoptolidin’s important biological activity and complex
structure have made it a popular target for total synthesis.4
An alternative approach to establishing the structural basis
for apoptolidin’s activity and for identifying related clinical
leads is modification of apoptolidin itself, available in
substantial quantity (109 mg/L) through fermentation.1 As
part of our efforts in this direction, we have identified an
(2) Developmental Therapeutics Program NCI/NIH. http://dtp.nci.nih.gov
(accessed July 2002).
(3) (a) Salomon, A. R.; Voehringer, D. W.; Herzenberg, L. A.; Khosla,
C. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 14766. (b) Salomon, A. R.;
Voehringer, D. W.; Herzenberg, L. A.; Khosla, C. Chem. Biol. 2001, 8,
71. (c) Salomon, A. R.; Zhang, Y.; Seto, H.; Khosla, C. Org. Lett. 2001, 3,
57.
† Stanford University.
‡ University of Tokyo.
(1) (a) Kim, J. W.; Adachi, H.; Shin-ya, K.; Hayakawa, Y.; Seto, H.
J. Antibiotics 1997, 50, 628. (b) Hayakawa, Y.; Kim, J. W.; Adachi, H.;
Shin-ya, K.; Fujita, K.; Seto, H. J. Am. Chem. Soc. 1998, 120, 3524.
10.1021/ol0266222 CCC: $22.00 © 2002 American Chemical Society
Published on Web 10/01/2002

1
Table 1. 13C (100 MHz) and H (500 MHz) NMR Data for Isoapoptolidin in CD3OD
isoapoptolidin (2)
δH (J ) Hz)
apoptolidin (1)1
isoapoptolidin (2)
δH (J ) Hz)
apoptolidin (1)1
no.
δC
δH
no.
δC
δH
1
2
3
4
5
6
7
8
169.92
124.86
147.08
132.76
143.65
132.90
136.75
39.41
2-Me
4-Me
6-Me
8-Me
12-Me
17-OMe
22-Me
24-Me
28-OMe
1′
13.92
17.82
17.63
18.26
12.00
59.87
12.13
5.38
59.48
96.23
73.32
74.92
87.53
68.05
18.38
61.01
99.88
45.29
2.04
2.07
1.72 (1.0)
1.17 (6.4)
1.70
2.14
2.21
1.97
1.17
1.71
3.40
1.06
0.92
3.30
4.85
3.44
3.76
2.76
3.78
1.29
3.61
4.97
1.96
1.84
7.31
5.93
7.41
6.23
3.47
5.06 (9.1)
2.74
3.77 (9.0, 9.0)
5.26 (15.8, 9.0)
6.04 (15.8)
5.27
2.79
3.87
5.26
6.21
1.03 (6.7)
0.86 (6.9)
3.35
4.78 (3.8)
3.38 (9.8, 3.8)
3.71
2.70 (9.0)
3.73
1.22 (6.2)
3.56
9
83.98
10
11
12
13
14R
14â
15R
15â
16
17
18R
18â
19
20
21
22
23
24
25
26R
26â
27
28
125.77
141.01
134.05
133.80
25.49
2′
3′
4′
5′
5.48 (9.6, 4.8)
1.93
2.48
1.24
5.71
2.09
2.50
1.44
1.52
3.47
2.75
1.78
2.20
5.32
3.57
6′
33.37
4′-OMe
1′′
2′′R
2′′â
3′′
4′′
1.39
5.05 (4.1)
2.01
1.77
73.10
81.89
35.05
3.51 (10.3, 5.0, 2.3)
3.36
1.36
1.84
4.41 (11.3, 3.7, 2.1)
4.98 (3.7)
73.67
85.91
68.62
18.35
22.76
101.91
37.15
3.32
3.73
3.37
3.70
1.25
1.36
4.86
2.47
1.32
3.21
3.01
3.24
1.31
3.46
67.44
74.13
103.27
36.79
73.20
40.52
68.17
36.82
5′′
6′′
1.22 (6.2)
1.34
4.82 (9.7, 1.8)
2.43 (12.3, 5.0, 1.8)
1.28
3.16 (11.6, 9.0, 5.0)
2.95 (9.0, 9.0)
3.19
1.27 (6.2)
3.42
3′′-Me
1′′′
2′′′R
2′′′â
3′′′
4′′′
5′′′
6′′′
3′′′-OMe
1.77
3.72
1.74
2.08
3.76
1.76
3.99
1.62
1.49
3.48
3.36
4.21 (10.3, 2.5, 2.0)
1.76
1.53 (14.7, 10.7, 3.0)
3.82
3.44
81.98
77.12
73.45
18.87
57.32
76.24
77.07
isomer of apoptolidin and report herein its characterization,
solution conformation, and biological activity.
chemical shifts of most proton signals in 2 remain very
similar to those reported by Seto and co-workers for
apoptolidin.1 The same spin systems found in 1 were also
present in 2, as determined by DQF-COSY experiments.
The HMBC spectrum of 2, however, shows a strong
correlation between the carbonyl carbon (C-1) and 20-H. In
addition, the chemical shift of 20-H is shifted downfield,
from 3.57 ppm in 1 to 4.98 ppm in 2, strongly suggesting
that it is now geminal to the lactone functionality. Similarly,
19-H shifts upfield, from 5.32 ppm in 1 to 4.41 ppm in 2.
All the proton and carbon resonances have been assigned
for 2 in CD3OD and are summarized in Table 1.
Our initial attempts to isolate apoptolidin from crude cell
extracts were complicated by the presence of an isomer,
referred to herein as isoapoptolidin (2), that coeluted during
chromatography. It was found that apoptolidin can be
partially separated from this isomer, which has a slightly
lower Rf on normal-phase silica, by flash column chroma-
tography.5 Isoapoptolidin, however, remained impure, even
after subsequent attempts to isolate it using either normal-
phase or reverse-phase HPLC. Surprisingly, an enriched
solution of 2 in either chloroform or acetonitrile was found
to produce pure 2 over several hours at room temperature
as white filamentous crystals (mp 134-136 °C). These
crystals of 2, unfortunately, have thus far proven to be
unsuitable for X-ray analysis because of their poor aspect
ratio. This crystallization takes place even from very dilute
solutions of 2 but does not occur when 1 is present in
substantial amounts. Mass recovery of 1 and 2 from crude
cell extracts indicates that the two isomers are present in
roughly equal amounts.
Strong ROESY correlations between 1,3-diaxial substit-
uents on the C-21 to C-25 cyclic hemiketal in 2 indicate
that it is in a chairlike conformation. ROESY correlations
are observed from 20-H to both 22-H and 22-Me-H3,
indicating that C-20 is equatorial on the ring. Coupling
constant data for the anomeric proton signals (Table 1)
indicate that the stereochemistry of the glycosidic bonds in
2 remain the same as those found in 1. Finally, high-field
carbon chemical shift for the allylic methyl signals and the
large vicinal coupling constant across the C-10/C-11 double
The structure of 2 was determined by analysis of one- and
two-dimensional NMR spectra (500 MHz). In CD3OD, the
3820
Org. Lett., Vol. 4, No. 22, 2002

bond indicate that the olefins in 2 are all of (E)-geometry.
ROESY correlations are fully consistent with this conclusion.
To determine the solution conformation of isoapoptolidin,
representative ROESY correlations were used to evaluate the
results of a Monte Carlo conformational search using
Macromodel 7.0.6 Of the 51 ROESY correlations resolved
for the C-1 to C-26 portion of isoapoptolidin, 23 were
nontrivial and applied in the subsequent analysis. Strong
ROESY correlations are observed between 3-H and 5-H and
between 5-H and 7-H, indicating that the three methyl groups
on the triene lie on the same face of the macrolide as they
do in apoptolidin. Particularly notable are transannular
correlations between 3-H and both 18-Hâ and 15-Hâ. The
observed hypsochromic shift in the enoate chromophore from
a λmax of 319 nm in 1 to 304 nm in 2, along with a decrease
in the extinction coefficient for this absorption, from 22 800
in 1 to 19 400 in 2, suggests a decrease in the degree of
conjugation for the triene system in 2 relative to 1.
The identification of isoapoptolidin raises the possibility
that an interconversion between 1 and 2 could take place
under the conditions of biological assay or chemical isolation.
Apoptolidin has been found to be stable for up to three
months at -20 °C when dissolved in chloroform or meth-
ylene chloride. During this time, no conversion to isoapop-
tolidin can be observed, either by NMR or HPLC. Similarly,
isoapoptolidin is stable as a solution in methanol at -20 °C
for up to six months. In contrast, a dilute aqueous solution
of apoptolidin at ambient temperature can be observed to
convert to isoapoptolidin when monitored by HPLC or UV-
vis. Similarly, an aqueous solution of 2 can be observed to
convert to 1. The respective products of these conversions
can be isolated and are found to be 2 and 1. Because this
conversion goes in both directions, 1 and 2 are likely to be
in simple equilibrium.
For synthetic work and for the evaluation of biological
assays of apoptolidin and its derivatives, it is critical that
the equilibrium constant and rates of interconversion are
known under relevant conditions. To measure these kinetic
parameters, solutions of 1 and 2 in Dulbecco’s phosphate-
buffered saline8 (PBS) were prepared and incubated at 37
°C in sealed reaction vessels. Periodically, the components
present in these solutions were quantified using reverse-phase
HPLC.9 After normalization for the difference in extinction
coefficients (λobs ) 254 nm), these values were plotted
(Figure 3), and the rate constants k1 and k-1 were extracted
Figure 1. Structure and numbering scheme for isoapoptolidin.
The unusually high chemical shift of 19-H in 2 could be
attributed to a new hydrogen bond between the C-19 alcohol
and the C-25 oxygen. Significant ROESY correlations
between the 20-H and the 22-H proton suggest an arrange-
ment of the C-20/C-21 torsion that would favor this hydrogen
bond. A representative structure from the conformational
Figure 3. Kinetics toward equilibrium for the interconversion of
1 and 2 in aqueous solution where Q ) [2]/[1].
by fitting the data to the integrated rate expression for a
simple equilibrium (Table 2).
The rate parameters calculated starting from either a
solution of 1 or a solution of 2 are very similar, suggesting
(4) (a) Schuppan, J.; Wehlan, H.; Keiper, S.; Koert, U. Angew. Chem.,
Int. Ed. 2001, 40, 2063. (b) Schuppan, J.; Ziemer, B.; Koert, U. Tetrahedron
Lett. 2000, 41, 621. (c) Toshima, K.; Arita, T.; Kato, K.; Tanaka, D.;
Matsumura, S. Tetrahedron Lett. 2001, 42, 8873. (d) Nicolaou, K. C.; Li,
Y.; Fylaktakidou, K. C.; Mitchell, H. J.; Wei, H. X.; Weyershausen, B.
Angew. Chem., Int. Ed. 2001, 40, 3849. (e) Nicolaou, K. C.; Li, Y.;
Fylaktakidou, K. C.; Mitchell, H. J.; Sugita, K. Angew. Chem., Int. Ed.
2001, 40, 3854. (f) Nicolaou, K. C.; Li, Y.; Weyershausen, B.; Wei, H. X.
Chem. Commun. 2000, 307. (g) Sulikowski, G. A.; Jin, B. H.; Lee, W. M.;
Wu, B. Abstracts of Papers, 218th National Meeting of the American
Chemical Society, New Orleans, LA, Aug 22-26, 1999; American Chemical
Society: Washington, DC, 1996; 563-ORGN. (h) Sulikowski, G. A.; Jin,
B. H.; Lee, W. M.; Wu, B. Org. Lett. 2000, 2, 1439.
Figure 2. Representative solution conformation of the region from
C-1 to C-26 of 2, excluding 6-deoxy-4-O-methyl-L-glucose, ole-
androse, and olivomycose residues.
search is depicted in Figure 27 and is fully consistent with
data from the ROESY experiment.
Org. Lett., Vol. 4, No. 22, 2002
3821

solutions of 1 and 2 including all assay components were
analyzed for isomerization. After 20 min at 37 °C, less than
12% conversion was detected for either compound.
Table 2. Kinetic Parameters for the Equilibrium of 1 and 2 in
Dulbecco’s PBS at 37 °Ca
These data indicate that apoptolidin is a more potent
inhibitor of F0F1-ATPase than isoapoptolidin (Table 3). It is
Q(t) ) 0 at t ) 0
Q(t) ) ∞ at t ) 0
Table 3. Results of the F0F1-ATPase Inhibition Assay for 1
and 2
k1
k-1
Keq
0.0656 h-1
0.106 h-1
0.616
0.0626 h-1
0.0982 h-1
0.638
IC50 (µM)
apoptolidin (1)
isoapoptolidin (2)
0.7 ( 0.5
17 ( 5
a Q(t) ) [2]/[1] at time t.
not clear, however, if this difference in inhibitory activity
can be extended to infer differences in the desired biological
effect of apoptolidin: the selective induction of apoptosis.
It is critical to determine which structure is more relevant,
and to what degree, in order for an understanding of the
structural basis for activity to be meaningful or useful.
Experiments that address this issue are currently underway.
the absence of any long-lived intermediate structure. Fur-
thermore, an equilibrium constant of ∼0.6 suggests that the
isolation yields of 1 and 2 are governed by this intercon-
version. More importantly, the value of k1 suggests that under
the conditions of most cell-based assays, a significant
conversion of 1 to 2 will have occurred within several
hours.10 The presence of isoapoptolidin, therefore, must be
taken into account when interpreting the results of cell-based
assays.
Acknowledgment. This research was supported by a
grant (CA31845) from the National Cancer Institute. We
thank Professor Chaitan Khosla for his guidance with the
ATPase inhibition assay and isolation of apoptolidin.
As a preliminary assessment of the biological activity of
2, its ability to inhibit F0F1-ATPase in isolated yeast
mitochondria was measured.11 The duration of this assay is
less than 20 min, during which time the isomerization of 1
and 2 should be minimal. To ensure that this is the case,
Supporting Information Available: Experimental pro-
cedure for the ATPase inhibition assay and spectroscopic
1
data for 2, including H NMR, 13C NMR, DQF-COSY,
HSQC, HMBC, ROESY, IR, and ESMS. This material is
available free of charge via the Internet at http://pubs.acs.org.
(5) In a typical procedure, 1.0 g of cell extract was purified using a 40
cm × 2.5 cm flash column of 230-400 mesh normal-phase silica eluting
with 7% methanol in chloroform.
(6) (a) Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Liskamp, R.;
Lipton, M.; Caulfield, C.; Chang, G.; Hendrickson, T.; Still, W. C. J.
Comput. Chem. 1990, 11, 440. (b) Calculations performed using the GB/
SA water solvation model: Still, W. C.; Tempczyk, A.; Hawley, R. C.;
Hendrickson, T. J. Am. Chem. Soc. 1990, 112, 6127.
OL0266222
(9) Baseline separation can be achieved using a 150 × 4.6 mm C18/5 µ
column eluting with a water/acetonitrile gradient from 40% to 90% over
30 min.
(7) Image rendered using VMD-Visual Molecular Dynamics: Humphrey,
W.; Dalke, A.; Schulten, K. J. Mol. Graphics 1996, 14, 33.
(8) Dulbecco, R.; Vogt, M. J. Exp. Med. 1957, 106, 167.
(10) The half-life for k1 in the absence of k-1 is 10.5 h.
(11) Roberts, H.; Choo, W. M.; Murphy, M.; Marzuki, S.; Lukins, H.
B.; Linnane, A. W. FEBS Lett. 1979, 108, 501.
3822
Org. Lett., Vol. 4, No. 22, 2002