Amphotericin B (AmB) is a chemically
complex polyene-macrolide natural product that has been used continuously since
the 1960s as a last line of defense against systemic fungal infections.
Unfortunately, AmB is plagued by an extrememly narrow therapeutic index.
Indeed, because of the often dose-limiting toxicity of the antifungal drug,
mortality rates for systemic fungal infections still hover near 50%.
Martin Burke’s laboratory at the
University of Illinois at Urbana-Champaign has taken a keen interest in the
fungicidal mode of action of AmB, which has managed to evade significant
microbial resistance for over 50 years. A better understanding of the precise
nature of AmB’s ‘resistance-refractory’ mechanism might lead to new antimicrobial
agents with reduced toxicity. The prevailing model used to interpret
biophysical studies of AmB’s role within living systems is the ‘artificial ion
channel model.’ According to this theory, AmB exists in the form of small ion
channel aggregates that are inserted into lipid bilayers with membrane
sterols oriented in between vertically-stacked molecules of AmB, arranged in a circular pore or cavity (see Figure
below, top left). The artificial channels are thought to permeabilize and ultimately kill cells. The ion channel model dictates that the strategic pathway to
improving therapeutic index is by achieving selective formation of ion channels
in yeast over human cells.
Burke has recently advanced an alternate mechanistic scenario
to explain AmB's potent fungicidal properties. In the 'sterol sponge model' (Figure above,
top right), AmB forms large extramembranous aggregates that extract the
essential sterol ergosterol from phospholipid bilayers. The coincidental extraction
of membrane cholesterol (unselectively with ergosterol) might then be primarily
responsible the observed toxicity of AmB to human cells. This insight, if
proven correct, could guide the development of novel AmB derivatives with an improved
therapeutic index. With regard to evasion of resistance, AmB may simultaneously
disrupt all of the cellular processes that depend on membrane ergosterol.
Simultaneous mutation of all requisite fungal proteins in order to alleviate
the ergosterol dependence of the pathogenic organism is then improbable and,
hence, no significant resistance emerges.
In 2014, Burke’s group presented
evidence from biophysical, cell-based experiments that uniformly supports the
new sterol sponge model. As such, an improvement in the relative binding
affinity of AmB aggregates for ergosterol over cholesterol was expected to translate
to antimicrobial agents with lower toxicity towards human cells. The Illinois
team approached this challenge by initially examining a recent crystal
structure of an AmB derivative that reveals an intramolecular salt bridge
between the C16 carboxylate and the C3’ ammonium moiety of AmB’s mycosamine
subunit (See AmB molecular structure, depicted above). They hypothesized that a
complex network of intramolecular noncovalent interactions, including the
aforementioned salt bridge, orients the mycosamine in a conformation that is
capable of forming an H-bond to the 3b-hydroxyl
of both ergosterol and cholesterol (diagrammed below). The mycosamine
C2’-hydroxyl was implicated in this network of interactions by previous studies
which demonstrated that deletion of the C2’-hydroxyl group (in semisynthetic
C2’deOAmB) leads to differential binding of ergosterol and cholesterol. A
high-resolution structure of the AmB aggregate (the functional ‘sponge’), with
and without bound sterol, would enable rational design of new synthetic
molecules. However, in its absence, a practical and highly modular semisynthetic
route to derivatize the ‘eastern’ mycosamine-bearing substructural framework of
AmB was sought. Newly generated analogs would then be evaluated for sterol
binding selectivity and correlation to selective toxicity to yeast.
A remarkably concise strategy for
AmB derivatization was developed by Burke and co-workers. The route (shown
below) involves only three reaction ‘pots’ and requires one chromatographic
separation to produce pure material on gram-scale. The exceptionally mild
oxazolidinone ring-opening by amine nucleophiles is noteworthy. The facile
nature of this transformation, which introduces a new point of diversity for
SAR studies, likely reflects inherent ring-strain residing within the
trans-fused 6/5 heterocyclic ring system of the bracketed intermediate. The
tetrahydropyran methyl ketal of the penultimate species is converted into its
corresponding hemiacetal during the course of HPLC purification, which uses
aqueous formic acid in the mobile phase. An ultracentrifugation-based membrane
isolation assay indicated that the new C16-urea derivatives exhibit binding
selectivity for ergosterol over cholesterol. Moreover, the sterol binding
selectivity was generally associated with an improvement in therapeutic index,
as indicated by a comparison of the minimum inhibitory concentration (MIC) against
S. cerevisiae with the minimum
hemolytic concentration (MHC) determined against human red blood cells. This
project is a great example of a practical and well-executed organic chemistry
strategy, applied to a long-standing pharmacological problem with profound human
clinical implications. The emerging AmB mechanistic picture is an example of a
rarely encountered pharmacological phenomenon – that is, noncovalent binding
between a small molecule drug and an endogenous small molecule. The seemingly
‘resistance-refractory’ nature of this type of mechanism renders it a highly
desirable platform for pharmacological intervention aimed at anti-infective
drug development with low tox and evasion of resistance. The work also
illuminates to the ubiquity and essential nature of sterols such as cholesterol
and ergosterol to living systems within the cells of both humans as well as
pathogenic organisms.