Key words
archazolid - natural product - Stille reaction - V-ATPase - cyclooxygenase - total
synthesis.
The archazolid natural products[1] (A–F, Figure [1]) constitute a family of highly potent (subnanomolar IC50) and selective vacuolar-type ATPase (V-ATPase) inhibitors that have shown promising
activity against a number of particularly aggressive and lethal cancers including
trastuzumab-resistant breast cancer,[2] glioblastoma multiforme (GBM),[3] and T-cell acute lymphoblastic leukemia (T-ALL).[4] More recent studies indicate that the archazolids also block iron metabolism and
thereby mediate a therapeutic effect in breast cancers,[5] they modulate release of tumor-promoting cytokines,[6] and when combined with the p53 activator nutlin-3a, synergistically induce tumor
cell death.[7]
Figure 1 Structure of the archazolid natural products
All members consist of a structurally similar 24-membered macrolactone and thiazole/carbamate
side chain. Glycosylation at either the C7- or C15-hydroxyls (archazolids C and E, respectively) significantly reduces their V-ATPase
inhibitory activity,[1d] indicating that these two groups form important interactions with the enzyme. Interestingly,
these same hydroxyls are connected by a Z,Z,E-conjugated triene unique to the archazolids.
Figure 2 Reported synthetic approaches to the archazolid conjugated triene
Several synthetic strategies have been reported for the pharmacophorically relevant
conjugated triene region of the archazolids (Figure [2]). Menche’s group ultimately utilized a three-step aldol condensation after attempted
Horner–Wadsworth–Emmons reactions failed,[8] which then required a late-stage enantioselective CBS reduction to install the C15 hydroxyl. Two palladium-catalyzed cross-couplings have also been described, one a
successful, albeit low-yielding, Stille reaction[9] and the other a similar yet simpler Negishi coupling.[10] Recently, our group reported a synthesis of the archazolid triene by cross-metathesis
(CM).[11] While promising, the convergence of this CM strategy is limited, for example due
to the requirement of using a cis homodimer and issues associated with metathesis back-biting.[12c] Combined with nonideal aspects of the other synthetic approaches (e.g., low yields
or requisite late-stage manipulations), we continued to explore alternative disconnections.
Herein we report the synthesis of a C1–C23 fragment of the archazolids based on a high-yielding and convergent Stille coupling
for construction of the conjugated triene (Scheme [1]). The choice of coupling partner identity (i.e., which was the organostannane and
halide) proved critical to the success of this reaction. After removal of the protecting
groups, the V-ATPase and cyclooxygenase (COX) inhibitory activities of the resulting
tetrol 1 were then assayed. The results suggest some level of V-ATPase inhibition by this
compound but no significant interaction with COX.
Scheme 1 Novel Stille coupling-based synthesis of the archazolid triene containing compound
1
When considering alternative non-metathesis based archazolid syntheses, we nonetheless
wanted to make use of the chemistry developed for these prior approaches.[12] To that end, aldehyde 2 was identified as a suitable starting point (Scheme [2]). This compound had been previously prepared en route to dihydroarchazolid B.[12c] We argue that saturation of the C2–C3 olefin (Figure [1]) would not be expected to negatively impact archazolid biological activity, but
may simplify their synthesis and improve stability. In preparation for possible palladium-catalyzed
cross-couplings, aldehyde 2 was first converted into vinyl iodide 3 in 85% as an 8:1 mixture of Z/E isomers by 1H NMR analysis at the newly formed alkene.[13]
Scheme 2 Vinyl iodide synthesis for cross-coupling
Our ‘western hemisphere’ synthesis[12b] was adapted to produce an appropriate vinyl stannane coupling partner, accomplished
through the use of phosphonate 4 which is available in two steps from known Weinreb amide 5
[14] (Scheme [3]). Horner–Emmons olefination[15] with aldehyde 6
[16] gave ketone 7 which was then reduced and methylated as previously described.[12b] Unfortunately all attempts to couple stannane 8 and iodide 3 failed. In each, unreacted iodide 3 was observed, suggesting difficulties in the oxidative addition step of the catalytic
cycle, perhaps sterically hindered by the γ-methyl group.
Scheme 3 Vinyl stannane synthesis and first attempts at Stille coupling
Gratifyingly, switching the sense of organometallic/halide in these reactions now
led to success with the couplings. Specifically, iododestannylation[17] of 8 gave iodide 9, and iodide 3 was converted into stannane 10 by lithium–halogen exchange and trapping with Bu3SnCl (Scheme [4]).[18] These two compounds underwent very efficient coupling using Fürstner’s conditions,[19] providing 11 in 82% yield.
Scheme 4 Reagents and conditions: (a) I2, CH2Cl2, –10 °C, 70%; (b) t-BuLi then Bu3SnCl, 90%; (c) 10 (1.0 equiv), Pd(PPh3)4 (5 mol%), CuTC (2.0 equiv), [Ph2PO2][NBu4] (3.0 equiv), THF, 15 h, 82%.
We were intrigued about the possibility of a compound of type 11 inhibiting V-ATPase function. Others have commented on the utility of natural product
derived fragments for drug discovery,[20] particularly those that maintain essential pharmacophoric features. Previously we
had tested both ‘western’ (12) and ‘eastern’ (13) hemispheres of the archazolids using an Arabidopsis V-ATPase assay and found that neither displayed measurable inhibitory activity (Figure
[3]).[12b] However, these compounds lacked the important linked C7- and C15-hydroxyls present in tetrol 1.
Figure 3 Synthetic archazolid fragments; only compound 1 bears the linked C7- and C15-hydroxyls known to be important for V-ATPase inhibitory activity
As indicated in Figure [4], compound 1
[21] displayed dose-dependent growth inhibition of etiolated Arabidopsis.[22] A key component in the etiolated habit is stem elongation driven by V-ATPase-mediated
cell expansion,[23] such that monitoring seedling stem length provides a measurement for V-ATPase activity.
Previously, we demonstrated that stem elongation in Arabidopsis seedlings is inhibited by known V-ATPase inhibitors concanamycin A[24] and bafilomycin,[23] with concanamycin A exhibiting four times the potency of bafilomycin.[12b] The IC50 value for compound 1 in this assay was approximately two orders of magnitude greater than that of concanamycin
A and thus also the archazolids.[25] Nonetheless, this modest activity is significant given the major structural differences
and overall simplification of 1 relative to the natural product.
Figure 4 Select Arabidopsis V-ATPase assay results[22]
Based on a recent report by Reker et al.,[26] we may also suspect other biochemical targets for compound 1. In their study, archazolid A (ArcA) was dissected into four hypothetical fragments (e.g., ArcA-1, Figure [5]) from which potential targets were predicted. This computational exercise identified
primarily proteins associated with arachidonic acid (e.g., cyclooxygenase, COX). However
when tested, the COX-2 inhibitory activity of ArcA was weak (24 ± 6% inhibition at 10 μM). Reker et al. suggest this disconnect between
predicted hypothetical fragment activities and the actual natural product could be
due to the COX active site being buried, allowing for binding of smaller fragments
(e.g., ArcA-1 and arachidonic acid) but not ArcA.[26]
Figure 5 Structures of synthetic archazolid fragment 1, a hypothetical archazolid fragment ArcA-1 used by Reker,[26] archazolid A (ArcA), and arachidonic acid
We saw compound 1 as an opportunity to add additional experimental data to this theoretical work, representing
a fragment similar to the hypothetical fragment ArcA-1. Interestingly, compound 1 did not show a dose response of greater than 5% inhibition of COX-1 or COX-2 at concentrations
between 1 and 200 μM.[27] It is possible that compound 1 (C23) is similarly too large compared to ArcA-1 (C17) and arachidonic acid (C20) to bind COX. Alternatively, the inactivity of 1 could suggest that a carboxylic acid terminus is critical, which is known to be important
for other COX inhibitors.[28] However, this would not explain the measureable activity of ArcA which also lacks a carboxylic acid. Other factors such as binding entropies[29] might also therefore need to be considered (with ArcA being more conformationally restricted than 1) to understand the differential COX-inhibitory activity of these compounds.
In summary, we have developed an efficient synthesis of the archazolid macrolactone
(C1–C23) framework. The resulting fragment 1 displayed evidence for inhibition of the V-ATPase, in line with the importance of
properly linked C7- and C15-hydroxyls for archazolid/V-ATPase binding. Compound 1 was also assayed for COX inhibition based on previously reported predicted activities
for a structurally similar hypothetical fragment. The results showed no significant
COX-inhibitory activity for 1 (<5% inhibition at concentrations up to 200 μM) suggesting certain structural requirements
for COX binding (i.e., carboxylic acid or macrocycle). Current efforts are aimed at
advancing our understanding of archazolid structure–activity-relationships,[30] by utilizing 1 as a starting point to further probe the archazolid/V-ATPase and archazolid/COX interactions.
