Sugar/Steroid/Sugar Conjugates: Sensitivity of Lipid Binding to Sugar Structure

Article abstract of DOI:10.1021/ol030084r

(Equation presented) Three steroids, each bearing a sugar on rings A and D, have been synthesized. Their effect on the "melting" behavior of a lipid bilayer depends on whether the sugar is glucose, galactose, or mannose. Packing constraints dictate how th

Full text of DOI:10.1021/ol030084r

ORGANIC
LETTERS
2003
Vol. 5, No. 22
4041-4044
Sugar/Steroid/Sugar Conjugates:
Sensitivity of Lipid Binding to Sugar
Structure
Bessie N. A. Mbadugha† and Fredric M. Menger*
Department of Chemistry, Emory UniVersity, Atlanta, Georgia 30322
menger@emory.edu
Received July 9, 2003
ABSTRACT
Three steroids, each bearing a sugar on rings A and D, have been synthesized. Their effect on the “melting” behavior of a lipid bilayer
depends on whether the sugar is glucose, galactose, or mannose. Packing constraints dictate how the lipid bilayer responds to the sugars.
In the bark, fruit, and roots of many plants (as well as in
some marine organisms) dwell the saponins, compounds with
sugar units covalently bound to one or two sites of a steroid
or triterpene framework (the “aglycone”). Saponins are the
key ingredient in a host of traditional Chinese, African, and
Indonesian medicines.1,2 Commercial applications of saponins
can be ascribed in large measure to their polar/apolar
(“amphiphilic”) nature. These include foaming agents,
emulsifiers, and preservatives, as well as agents for combat-
ing cholesterol, microbes, fungi, viruses, tumors, and mol-
luscs.1,2 Digitoxin, a cardiac glycoside drawn below, is
perhaps the most famous member of the clan.
A (a saponin having six hexose units) without resorting to
lengthy protection-deprotection procedures.5 In 1999, Yu
and Hui published the synthesis of several saponins by a
one-pot sequential glycosylation.6 We cite the preceding work
only as a representative sample of what has been a much
more widespread synthetic effort.
Our interest focuses here on synthetic “bidesmosidic”
saponins with sugars attached to both rings A and D of a
steroid. The resulting sugar/“facial” hydrophobe/sugar com-
bination presented intriguing possibilities with regard to
colloidal properties in water. In particular, we wondered how
such a molecule would interact with a lipid bilayer, consider-
ing that (1) sugar groups are not expected to tolerate a
bilayer’s hydrophobic interior, (2) such a saponin would be
too short to span an entire membrane from one polar surface
to the other, and (3) since steroids are rigid, the saponin
molecule, which would presumably align parallel to the
bilayer’s lipid chains, cannot readily “loop” back to place
both sugars at the same membrane surface. Uncertainties in
membrane adsorption, caused by the hydrophobic and
geometric restrictions, led us to synthesize double-glycosy-
1993 was a good year for saponin synthesis. Nishizawa
reported the total synthesis of osladin, an intensely sweet
saponin.3 Danishefsky completed the synthesis of a complex
saponin, desgalactotigonin.4 And Schmidt prepared holotoxin
(2) Waller, G. R., Yamasaki, K., Eds. Saponins Used in Traditional and
Modern Medicine; Plenum Press: New York, 1996. Waller, G. R.,
Yamasaki, K., Eds. Saponins Used in Food and Agriculture; Plenum
Press: New York, 1996.
(3) Yamada, H.; Nishizawa, M. Synlett 1993, 54.
(4) Randolph, J. T.; Danishefsky, S. J. J. Am. Chem. Soc. 1993, 115,
8457.
† Current address: Department of Chemistry, Northwestern University,
Evanston, IL 60208.
(1) Hostettman, K.; Marston, A. Saponins; Cambridge University
Press: Cambridge, England, 1995.
(5) Han, X.-B.; Jiang, Z-H.; Schmidt, R. R. Liebigs Ann. Chem. 1993,
853.
(6) Yu, B.; Yu, H.; Hui, Y.; Han, X. Tetrahedron Lett. 1999, 40, 859.
10.1021/ol030084r CCC: $25.00 © 2003 American Chemical Society
Published on Web 09/30/2003

Scheme 1
lated saponins with three different sugars (glucose, galactose,
and mannose). As will be shown, their membrane binding
displayed an unexpected sensitivity to the sugar structure.
red blood cells.7 We have previously examined the effect of
the monodesmoside digitonin on giant vesicles.8 While
digitonin’s membrane behavior is well-established and
understandable given its “conventional” amphiphilic struc-
ture, the interaction mode of our present novel bidesmosides,
with their dual polar moieties, was not obvious a priori. To
develop a preliminary picture of their comparative properties,
we turned to differential scanning calorimetry (DSC), which
measures the influence of incorporated components on the
fluidity of membrane bilayers. Thin films of 1, 2, 3, or 4
plus dipalmitoylphosphatidylcholine (DPPC), in a molar ratio
of 1:10, were prepared by drying chloroform solutions in a
glass vial under a stream of N2 followed by vacuum drying
for at least 16 h. Hydration of the mixed films, achieved by
adding Milli-Q-purified water (5 mg/2 mL), was accom-
plished by manually shaking while warming with a heat gun
to produce a homogeneous opaque suspension. This suspen-
sion was vortexed for 1 min, sonicated in a water bath for l
h at 70-85 °C, and then subjected twice to a freeze-thaw
cycle (-20 to 23 °C) to provide samples suitable for DSC.
Typical runs, obtained with a Hart Scientific DSC instrument,
were carried out with a 30 min hold at 20 °C and an ensuing
scan up to 60 °C at a 5 °C/h heating rate to detect the
endothermic peak resulting from the DPPC bilayer “melting”
(i.e., a conversion from a “solid” gel phase into a more fluid
The synthesis of the glucose-glucose and galactose-
galactose saponins (1 and 2, respectively) from 5-R-andros-
tane-3-â,17-â-diol is given in Scheme l. Mannose-mannose
saponin 3, with its R-configuration at the glycosidic center,
was made in an identical fashion. Yields reported in Scheme
1 have not been optimized. Structure verifications consisted
of 1H and 13C NMR, FAB-HRMS, and EA (see Supporting
Information for full details). None of the three compounds
was soluble to the extent of 0.1 wt % in water at room
temperature.
For comparison purposes, the monodesmosidic compound
4, with a single galactose fixed to a cholesteryl aglycone,
was prepared in a 75% overall yield using the general
procedure given in Scheme 1.
Many saponins interact with cell membranes, as mani-
fested, for instance, in the capacity to induce hemolysis in
(7) Nose, M.; Amagaya, S.; Ogihara, Y. Chem. Pharm. 1989, 37.
(8) Menger, F. M.; Keiper, J. S. Angew. Chem., Int. Ed. 1998, 37, 3433.
4042
Org. Lett., Vol. 5, No. 22, 2003

liquid-crystalline phase). Duplicate scans on fresh samples
indicated that the data are reproducible.
Pure DPPC bilayers display a sharp melting peak at a Tm
) 41.3 °C. Incorporating 10 mol % cholesterol into the
DPPC bilayer broadens the melting peak while lowering the
Tm to about 39.5 °C. The DSC plots for DPPC containing
10 mol % 1-4 are given in Figure 1. Saponin 3 plus DPPC
Figure 2. Schematic models for the adsorption of bidesmosidic
saponin-like compounds into a bilayer. The shaded circles represent
polar groups, and the rectangles represent the steroidal aglycone.
Keeping the interdigitation model in mind, we can explain
the DSC differences among the three sugars with one
additional assumption: a guest molecule within a membrane
bilayer causes maximal perturbation to the motional freedom
of the phospholipid chains (and thus to their melting
behavior) when the guest and lipid chains have intimate
contact. If this is true, then one sees from Figure 3 that
Figure 1. Endothermic plots for 10 mol % 1-4 in DPPC obtained
by DSC at a scanning rate of 5 °C/h.
gives a sharp DSC peak virtually identical in position and
shape to that of pure DPPC. In contrast, the Tm of the
galactose-galactose 2 system is shifted downward to 40.5
°C, and the peak is noticeably broadened. The Tm for DPPC
plus the glucose-glucose saponin 1 is further shifted to 39.6
°C with the peak being now substantially broadened. Since
the bilayers are overwhelmingly composed of the carrier
lipid, DPPC, the seemingly small variations in peaks caused
by the saponins are in fact worthy of note. First, the shifts
in Tm for 1 and 2 are indicative of membrane incorporation.
Second, despite their general structural similarities, saponins
1-3 provided Tms having nontrivial differences. How might
these unexpected differences be explained in molecular
terms?
Assuming the saponins 1-3 did incorporate into the
vesicles, we can offer the following interpretation of the data.
Two packing modes for the bidesmosidic saponins in a
bilayer can be envisioned (Figure 2). It is possible that two
saponins line up to span a normal lipid bilayer. Alternatively,
a single saponin molecule might span a bilayer that, via
interdigitation, locally contracts to accommodate the guest.
Of the two models, we prefer the latter because it does not
demand that polar sugars embed themselves within a
hydrocarbon environment (as does the former model). The
interdigitation model is in fact not a new idea; biophysical
studies have been carried out on bilayers incorporating a
component quite similar in concept to the present saponins,
a bolaamphiphile (a double-headed surfactant with a hydro-
phobic length roughly half of the bilayer thickness).9,10 It
was proposed that the membrane narrows itself by inter-
digitating in the vicinity of the bolaamphiphile molecules.
Figure 3. Structural representation of the glucose-glucose steroid
1 (left) and the mannose-mannose steroid 3 (right) interdigitated
into a DPPC bilayer.
glucose-glucose 1 should have a greater Tm reduction than
mannose-mannose 3 (with galactose-galactose 2 being
intermediate in behavior) as is observed. Stated in another
way, even if mannose-mannose 3 causes large defects due
to its axial glycosidic linkage, the large interdefect regions
of the phospholipid manifest a rather normal DSC.
(9) Lemmich, J.; Honger, T.; Mortensen, K.; Ipsen, J. H.; Bauer, R.;
Mouritsen, O. G. Eur. Biophys. J. 1996, 25, 61.
(10) Duwe, H. P.; Kaes, J.; Sackman, E. J. Phys. France 1990, 51, 945.
Org. Lett., Vol. 5, No. 22, 2003
4043

We note that although it is tempting to account for the
lack of a Tm shift with 3 by speculating that it did not enter
the membrane bilayer, there was no visual evidence of the
water-insoluble bolaamphiphile floating in the suspension
which resembled that of the other samples analyzed. An
alternate rationale for 3’s Tm can be offered in terms of a
“canceling out” effect, whereby the decrease in Tm afforded
by perturbation of bilayer chain movement is countered by
the increase in Tm associated with contraction of the bilayer.11
Since 3 alone lacks a Tm shift, the canceling-out rationale
would seem to be an ad hoc addition to our model.
The DSC of 4, in which the melting plot almost disappears
into the baseline, is consistent with the interdigitation model.
With only a single sugar, this compound can slip its steroid
moiety between the chains without inducing a bilayer
contraction. The resulting tight packing has a dramatic and
long-range effect on the DSC.
and evaluated for their influence on phospholipid membranes.
The surprising dependence of the DSC data on the nature of
the sugar may be understandable in terms of packing
constraints arising from sugar stereochemistry. The prepara-
tion and analysis of additional analogues containing sugar
moieties, including branched saccharides and xylose deriva-
tives, may provide more insight into the nature of bidesmo-
sidic saponin-membrane interaction.
Acknowledgment. This work was supported by the
National Institutes of Health.
Supporting Information Available: Synthetic details and
1
characterization data, including H and 13C NMR, mass
spectroscopy, and elemental analysis for compounds 1-4,
as well as DSC sample preparation and procedure. This
material is available free of charge via the Internet at
http://pubs.acs.org.
In summary, three bidesmosidic saponins were synthesized
(11) Abrams, S. B.; Yager, P. Biochim. Biophys. Acta 1993, 1146, 127.
OL030084R
4044
Org. Lett., Vol. 5, No. 22, 2003