Figure 2
. (a) Synthesized analogues and GI50 values against MDA-MB231 and HME cells; (b) MDA-MB231 cells + largazole; (c) HME
cells + largazole. For details of the cellular proliferation assay see the Supporting Information.
With a route to largazole secured, it was possible to
evaluate both the original claim of differential growth
inhibition for transformed vs. nontransformed breast cells
and also the effect of variation of the side chain on
antiproliferative activity. Readily accessible analogues, such
as 3, seco-ester 12, ester 13, and ketone 14, as well as
synthetic largazole were tested against MDA-MB231 cells
and nontransformed human mammary epithelial cells (HME).
We choose the MDA-MB231 cell line in light of the data
published in the isolation paper describing largazole’s
remarkable differential activity against transformed vs.
nontranformed cells. We expected that the comparison
between tumorigenic human breast cancer cells and non-
transformed human mammary epithelial cells should be more
meaningful than the original comparison utilizing mouse
mammary epithelial cells (NMuMG). As can be seen in
Figure 2, largazole inhibits the growth of MDA-MB231 cells
with the GI50 of 71 nM. In contrast, the same compound
had little effect on proliferation of HME cells (GI50 values
of >600 nM). Even though the GI50 value of largazole on
MDA-MB231 is higher than what was observed in the
previous studies (7.7 nM), our results support the notion that
largazole preferentially acts on tumor cells and this unique
property of largazole could be exploited in antibreast cancer
therapies in the future.
terms of targeting, and also suggests a potential site for
modifications to improve hydrolytic stability.
In light of the loss of activity of the seco-ester, we set out
to study the conformation of largazole in CDCl3 by NMR
as a starting point for developing a three-dimensional picture
of largazole’s pharmacophore. NOESY spectra were col-
lected at a variety of mixing times ranging from 150 to 700
ms.12 Key transannular and long-range correlations that were
used for Monte-Carlo conformational searching are shown
in Figure 3.13 The structures generated for largazole with
these data all depict a relatively rigid and flat macrocycle,
with the thiol-ester side chain and Val residue residing on
opposite faces. In keeping with our observations above,
cleavage of the ester (or any of the amides) could be expected
to significantly change the overall molecular topology.
Although it seems not likely to be relevant to the actual active
compound, the acyl side chain assumes a position across the
periphery of the macrocycle and shows some conformational
mobility within the constraints provided.
(12) A 10.8 mg sample of largazole was dissolved in 700 µ`L of CDCl3
and was rigorously degassed. Spectra were recorded at 20 °C on a Varian
Inova 500 NMR spectrometer operating at 500.369 MHz for 1H
observation, using an inverse triple-resonance probe optimized for 1H
detection. The standard NOESY experiment (Bodenhausen, G.; Kogler,
H.; Ernst, R. R. J. Magn. Reson. 1984, 58, 370) was used, with the
addition of the use of pulsed field gradients, both to eliminate residual,
unrelaxed magnetiztiion between transients, and also in the center of
the mixing time to minimize unwanted artifacts due to incomplete phase-
cycling to eliminate COSY-type cross-peaks (Wagner, R.; Berter, S. J.
Magn. Reson. Ser. A 1996, 123, 119). The following parameters were
used: data were aquired in phase-sensitive (States-TPPI) mode, using
90° excitation pulses ) 9.0 µs, relaxation delay ) 1.25 s, spectral width
(f1 and f2) ) 4203.7 Hz, mixing times ranging from 150 to 700 ms were
used, acquisition time ) 0.244 s in T2 (1024 complex points in the
acquisition dimension), 128 increments were acquired in t1, and linear-
predicted (4×) to 512 points. The final spectrum consisted of 1024
complex points in both dimensions, after application of optimized
gaussian apodiztion functions in t1 and t2.
Of further interest is the degree to which the importance
of the thioester/thiol domain is underscored by the testing
of analogues 3, 13, and 14 in our cell proliferation assay.
These data are in full accord with the studies by Luesch and
Hong and support the notion that the cellular target of
largazole is histone deacetylase. Also of note is the lack of
activity of the seco-ester 12. This presumably reflects the
importance of the overall topology of the depsipeptide in
Org. Lett., Vol. 10, No. 16, 2008
3597