Fig. 3 Flipping of NBD-GPI’s. (A) Schematic showing the predicted
50% fluorescence loss on adding dithionite to symmetrically labeled
liposomes (or inactive proteoliposomes) (left; labeled ‘Lip’), compared
with 100% loss in flippase-active proteoliposomes (‘Proteolip’) due to
flipping of NBD-GPIs.4 The looping back of the NBD-bearing acyl-chain
positions the polar NBD group at the membrane–water interface where it
can react with dithionite.6,7 (B) Fluorescence traces of assays with
liposomes (L) and proteoliposomes (P) containing NBD-GPI’s (1 or 2).
(C) Protein dependence of the extent of dithionite reduction of NBD-
GPI’s. The blue line is a fit of the data based on a model described here
(see ESI{ for details) and elsewhere.4
loss, since NBD-lipids in the inner leaflet will flip out and become
accessible to dithionite (Fig. 3A).4
Fluorescence dropped rapidly by y45% when dithionite was
added to NBD-GPI-labeled liposomes (Fig. 3B, green traces), and
was eliminated when the vesicles were detergent-permeabilized
indicating that dithionite was sufficient to reduce all NBD present
and that the NBD-GPI probes were roughly symmetrically
distributed. For proteoliposomes, dithionite caused a similarly
rapid, yet greater fluorescence loss (Fig. 3B, purple traces) that (a)
was reduced by protease treatment, (b) did not require ATP and
(c) depended on the amount of TE used for reconstitution
(Fig. 3C). Although fluorescence loss on dithionite addition is
predicted to range from y50–100% depending on the number of
flippase-equipped vesicles in the sample (which in turn depends on
the amount of TE used for reconstitution)4 (Fig. 3A), the
measured range was narrower (y45–75%; Fig. 3C), as reported
previously for other phospholipid probes.4 The reason for this is
not fully understood.
Scheme 1 Synthesis of NBD-GlcNAc-PI and NBD-GlcN-PI. Reagents
and conditions: (a) TMSOTf, CH2Cl2, 0 uC, 0.5 h, 90%, (b) NaOMe,
CH2Cl2–MeOH, rt, 24 h; then NaH, BnBr, DMF, 80%, (c) PdCl2,
NaOAc, AcOH–H2O, rt, 48 h, 64%, (d) propanedithiol, Py–H2O, Et3N, rt,
24 h; then Boc2O, rt, 12 h, 66%, (e) lipid-H-phosphonate 9, Py, Piv–Cl, rt,
0.5 h; then I2 in Py–H2O, rt, 0.5 h, 57%, (f) Pd(OH)2, MeOH–CH2Cl2–
H2O, H2, 12 h, 85%, (g) NBD-X,SE, DMF, Et3N, 2 h, rt, (h) TFA–
CH2Cl2–CH3CN, 4 : 4 : 2, rt, 2 h, 55%, (i) Ac2O, NaHCO3–MeOH, rt,
0.5 h, quant.
Our synthetic design involved three chiral building blocks: (a)
1-allyl-2,3,4,5-tetra-O-benzyl-D-myo-inositol 3 made in 7 steps
from bis-cyclohexylidene-D-inositol, (b) the 3,4,6-tri-O-acetyl-2-
azido-2-deoxy-b-D-glucosyl donor 4 prepared from tri-O-acetyl-D-
glucal by azidonitration and (c) the phosphatidyl donor 9 with a
protected terminal amine in the sn-1 acyl chain prepared from 1,2-
isopropylidene-sn-glycerol. Glycosylation of 3 with 4 gave the
The protein-dependence profile was the same for both GPI
probes (Fig. 3C). As detailed and validated elsewhere,4 at the
inflection point of the profile, i.e., the point at which the y75%
reduction plateau is first reached, each vesicle has one flippase.
This occurs at TE y 32 ml where the protein : phospholipid ratio
of the vesicles is y21 mg mmol21. Using this we estimate that the
flippase(s) responsible for transporting the GPI probes represents
y0.6% by weight of ER proteins in the TE (see below).4
Interestingly, the protein-dependence profile obtained in assays
of NBD-PC flipping was similar to that obtained for the GPI
probes (Fig. 3C and ESI{). It is possible, therefore, that the GPI
probes are transported by the ‘membrane building’ ER glycer-
ophospholipid flippase(s)4 responsible for flipping NBD-PC.
Alternatively, the two NBD-GPIs and NBD-PC may be flipped
by different flippases that are nevertheless similarly abundant in
the TE. Although our assay does not provide data on flipping
kinetics, it is clear that both GPI probes are transported rapidly,
on a time scale of y1 min (Fig. 3A), similar to that measured for
glycerophospholipid flipping in the ER.7 Thus, in our reconstituted
a-glucosaminyl(1
A 6)inositol intermediate 5 followed by
deprotection of the allyl group and replacement of the azido
group with NHBoc to enable the eventual selective coupling of the
NBD probe. Phospholipidation of 8 with 9 using H-phosphonate
chemistry gave the protected GPI intermediate 10. The next three
steps, removal of benzyls, coupling of the NBD probe, and
deprotection of the NHBoc group, provided NBD-GlcN-PI (1),
which on N-acetylation gave NBD-GlcNAc-PI (2).
We made unilamellar proteoliposomes from TE, egg phospha-
tidylcholine (egg PC) and trace amounts of NBD-GPI (1 or 2).
Protein-free liposomes were prepared by omitting TE. Using ES-
MS we found no evidence of degradation of the NBD-GPI’s
during reconstitution. Since the NBD-GPI’s are presumed to be
symmetrically distributed in the vesicle membrane,4 addition of
dithionite to liposomes should cause y50% fluorescence loss;
treatment of flippase-active proteoliposomes should yield y100%
454 | Chem. Commun., 2005, 453–455
This journal is ß The Royal Society of Chemistry 2005