Synthesis of a malaria candidate glycosylphosphatidylinositol (GPI) structure: A strategy for fully inositol acylated and phosphorylated GPIs

Article abstract of DOI:10.1021/ja038807p

A congener of the glycosylphosphatidylinositol (GPI) membrane anchor present on the cell surface of the malaria pathogen Plasmodium falciparum has been synthesized. This GPI is an example of a small number of such membrane anchors that carry a fatty acyl

Full text of DOI:10.1021/ja038807p

Published on Web 05/29/2004
Synthesis of a Malaria Candidate
Glycosylphosphatidylinositol (GPI) Structure: A Strategy for
Fully Inositol Acylated and Phosphorylated GPIs
Jun Lu, K. N. Jayaprakash, Urs Schlueter, and Bert Fraser-Reid*
Contribution from the Natural Products and Glycotechnology Research Institute, Inc.,
4118 Swarthmore Road, Durham, North Carolina 27707
Received September 30, 2003; E-mail: dglucose@aol.com
Abstract: A congener of the glycosylphosphatidylinositol (GPI) membrane anchor present on the cell surface
of the malaria pathogen Plasmodium falciparum has been synthesized. This GPI is an example of a small
number of such membrane anchors that carry a fatty acyl group at O-2 of the inositol. Although the acyl
group plays crucial roles in GPI biosynthesis, it rarely persits in mature molecules. Other notable examples
are the mammalian GPIs CD52 and AchE. The presence of bulky functionalities at three contiguous positions
of the inositol moiety creates a very crowded environment that poses difficulties for carrying out selective
chemical manipulations. Thus installations of the axial long-chain acyl group and neighboring phosphoglyceryl
complex were fraught with obstacles. The key solution to these obstacles in the successful synthesis of
the malarial candidate and prototype structures involved stereoelectronically controlled opening of a cyclic
ortho ester. The reaction proceeds in very good yields, the desired axial diastereomer being formed
predominantly, even more so in the case of long-chain acyl derivatives. The myoinositol precursor was
prepared from methyl R-D-glucopyranoside by the biomimetic procedure of Bender and Budhu. For the
glycan array, advantage was taken of the fact that (a) n-pentenyl ortho ester donors are rapidly and
chemospecifically activated upon treatment with ytterbium triflate and N-iodosuccinimide and (b) coupling
to an acceptor affords R-coupled product exclusively. A strategy for obtaining the GPI’s R-glucosaminide
component from the corresponding R-mannoside employed Deshong’s novel azide displacement procedure.
Thus all units of the glycan array were obtained from a â-D-manno-n-pentenyl ortho ester, this being readily
prepared from D-mannose in three easy, high-yielding steps. The “crowded environment” at positions 1
and 2, noted above, could conceivably be relieved by migration of the acyl group to the neighboring cis-
O-3-hydroxyl in the natural product. However, study of our synthetic intermediates and prototypes indicate
that the O-2 acyl group is quite stable, and that such migration does not occur readily.
workers.11 Subsequent investigations by Gowda, Gupta, and
Davidson12 have shown that the lipid residues of 1 can vary
Introduction
The insurgence of cerebral malaria that traditionally claims
2-3 million lives annually, mainly in tropical “Third World”
nations,1,2 now threatens the northern and southern temperate
zones.3,4 Attention has thereby been drawn to the causative
parasite, Plasmodium falciparum.5 This circumstance has
prompted unraveling of the mosquito’s genome,6 and develop-
ment of various drugs and vaccine candidates.7 Seminal 1992
publications by Playfair and co-workers implicated phospho-
inositide antigens,8 and subsequent studies by Schofield9 and
others10 culminated in the assignment of the glycosylphospha-
tidylinositol (GPI) candidate structure(s) 1 by Schwarz and co-
widely in structure, with profound consequences for biological
(6) Gardner, M. J.; Hall, N.; Fung, E.; White, O.; Berriman, M.; Hyman, R.
W.; Carlton, J. M.; Pain, A.; Nelson, K. E.; Bowman, S.; Paulsen, I. T.;
James, K.; Eisen, J. A.; Rutherford, K.; Salzberg, S. L.; Craig, A.; Kyes,
S.; Chan, M.-S.; Nene, V.; Shallom, S. J.; Suh, B.; Peterson, J.; Angiuoli,
S.; Pertea, M.; Allen, J.; Selengut, J.; Haft, D.; Mather, M. W.; Vaidya, A.
B.; Martin, D. M. A.; Fairlamb, A. H.; Fraunholz, M. J.; Roos, D. S.; Ralph,
S. A.; McFaddeb, G. I.; Cummings, L. M.; Subramanian, G. M.; Mungall,
C.; Venter, J. C.; Carucci, D. J.; Hoffmann, S. L.; Newbold, C.; Davis, R.
W.; Fraser, C. M.; Barrel, B. Nature 2002, 419, 498-511.
(7) Schofield, L.; Hewitt, M. C.; Evans, K.; Slomos, M.-A.; Seeberger, P.
Nature 2002, 418, 785-789. Travis, J. Sci. News 2002, 162, 99. Pasvol,
G. Trends Parasitol. 2002, 18, 524-525. Stoute, J. H.; Slaoui, M.; Heppiner,
D. G.; Momir, P.; Kester, K. E.; Desmons, P.; Wellde, B. T.; Gareon, N.;
Krzych, U.; Marchand, M.; Ballou, W. R.; Cohen, J. D. New Eng. J. Med.
1997, 336, 86-91. Simmis, P.; Nussenzweg, V. In Malaria Vaccine
deVelopment: a multi-immune response approach; Hoffman, S. L., Ed.;
ASM Press: Washington, DC, 1996; pp 15-33.
(8) Bate, C. A. W.; Taverne, J.; Playfair, J. H. L. Infect. Immun. 1992, 60,
1894-1901. Bate, C. A. W.; Taverne, J.; Bootsma, H. J.; Mason, R. C. St
H.; Skalko, N.; Gregoriadidis, G.; Playfair, J. H. L. Immunology 1992, 76,
35-41.
(1) For a brief introduction, see: Lopez-Antinano, F. J.; Schmunis, G. A.
Plasmodia of Humans. In Parasitic Protozoa; Academic Press: New York,
1993; Vol. 5, pp 146-169.
(2) MALARIA: and Obstacles and Opportunities; National Academy Press:
Washington, DC, 1991.
(3) Martin, P. H.; Lefebvre, M. G. Ambio 1996, 24, 200-209.
(4) See, for example: The New York Times, Jan 8, 1997, p 1; The Wall Street
Journal, May 7, 2001, p B12; The Globe and Mail, June 18, 1977, p 1;
Business Week, Sept 21, 1998, p 70; The Ottawa Citizen, Sept 28, 1997, p
A3.
(5) See, for example: Miller, L. H.; Good, M. F.; Milon, G. Science 1994,
264, 1878-1883.
(9) Schofield, L.; Hackett, F. J. Exp. Med. 1993, 177, 145-153.
(10) Jakobsen, P. H.; Moon, R.; Ridley, R. G.; Bate, C. A. W.; Taverne, J.;
Takacs, B.; Playfair, J. H. L.; McBride, J. S. Parasite Immunol. 1993, 15,
229-237. Kristensen, G.; Jokobsen, P. H. Exp. Parasitol. 1996, 82, 147-
154. Caro, H. N.; Sheikh, N. A.; Taverne, J.; Playfair, J. H. L.; Rademacher,
T. W. Infect. Immun. 1996, 64, 3438-3441.
9
7540
J. AM. CHEM. SOC. 2004, 126, 7540-7547
10.1021/ja038807p CCC: $27.50 © 2004 American Chemical Society

Synthesis of a Malaria Candidate GPI Structure
A R T I C L E S
pathways, reproduced in Figure 1,20 shows that crucial differ-
ences exist between both pathways. Thus as indicated by
asterisks 1 and 2, inositol acylation (via Acyl-T) necessarily
precedes the first mannosylation (via MT-1) in mammals.
However, in the case of T. brucei, the mannosylation does not
require prior inositol acylation.
Scheme 1
Inositol acylation therefore enforces a subtle distinction
between mammalian and parasitic biosynthetic pathways,
thereby presenting a window of opportunity for therapeutic
interVention. Indeed, Kinoshita has already exploited this
successfully in the case of African sleeping sickness.21
Interestingly, removal of the acyl group (asterisk 3) is also
an important biosynthetic step in that it precedes critical
processes involving “fatty acid remodeling” of the lipidation at
the sn1 and sn2 positions of the glyceryl moiety (asterisk 4).
McConville and Menon have suggested that these remodeling
procedures are specifically designed to facilitate translocation
of the growing GPI between the cytosolic and lumenal faces of
the endoplasmic reticulum.22
The glycerolipid residues are essential for anchoring the
glycoconjugate to the cell membrane, but the structural require-
ments, such as length, and functionality (e.g., acyl versus alkyl)
of these lipid residues are as yet unclear.
In view of the inositol deacylation step in Figure 1, it is not
surprising that most mature GPIs do not carry an acyl group on
inositol.23 It is therefore of interest that reacylation is sometimes
seen, notable examples being the human derived GPI-anchored
proteins CD5214 and AchE.15 Acylated modifications have also
been harvested from T. brucei procyclins.24
In light of the paramount importance of inositol acylation, it
is interesting to note that assignment of this functionality is
usually based on formation of a cyclic phosphate, e.g., 3 f 4,
upon treatment with phospholipase C (PLC)25 (Scheme 2).
However, the specter of acyl migration to the cis-related C3-
OH, e.g., 2 f 3, is worrisome since such an occurrence would
result in a positive, but compromised, PLC diagnosis. Notable
in this context is the fact that Puzo and co-workers have recently
isolated a 3-O-acyl phosphoinositide from Mycobacterium
tuberculosis extracts.26
activity.13 These observations clearly have crucial implications
for the design of therapeutic agents, and therefore justify the
development of synthetic strategies.
Accordingly, in this paper, we report the synthesis and
properties of a malarial GPI prototype, 35a, and one variant,
i.e., 35b, of the candidate structures summarized as 1 (Scheme
1).11 The approach addresses the crucially important (vide infra)
O-2 acylation of the inositol moiety, an unusual feature which
also occurs in other heavily lipidated GPIs such as CD5214 and
AchE.15 While this paper was being prepared, two important
contributions by Guo and co-workers relating to this challenging
feature, in the context of CD52, were published.16,17 Apart from
the issue of inositol acylation, the synthetic strategy described
herein also overcomes additional obstacles related to the
phosphodiacyl-glycerolipid residues, without which GPI con-
structs cannot function as “membrane anchors”.
Ferguson and co-workers have used O-2 alkyl groups as
biosynthetic probes with some success.27 However, such O-
alkylated substrates cannot be used to study phenomena such
as deacylation and migration, hence the importance of the
compounds described herein.
Background and Significance
Although GPIs have a conserved core comprising units I-V
of 1, there are significant differences in how it is assembled by
various organisms. By using cell free systems to study the most
readily available GPI, the variant surface glycoprotein (VSG)
of Trypanosoma brucei, Guther and Ferguson have estab-
lished a parasite biosynthetic pathway18 that makes it possible
to draw comparisons with the mammalian pathway deduced by
Kinoshita and co-workers.19 Ferguson’s summary of these
Results and Discussion
The representation of compound 1 in Scheme 3 emphasizes
the fact that the intersaccharide linkages are all R-D. Although
this anomeric option is favored in mannosides owing to the
(11) Gerold, P.; Diekmann-Schuppert, G. P.; Schwarz, R. T. J. Biol. Chem. 1994,
269, 2597-2606. Gerold, P.; Scholfield, L.; Blackman, M. J.; Holder, A.
A.; Schwarz, R. T. Mol. Biochem. Parasitol. 1996, 75, 131-143.
(12) Gowda, D. C.; Gupta, P.; Davidson, E. A. J. Biol. Chem. 1997, 272, 6428-
6439.
(19) Maeda, Y.; Watanabe, R.; Harris, C. L.; Hong, Y.; Ohishi, K.; Kinoshita,
K.; Kinoshita, T. EMBO J. 2001, 20, 250-261. Kinoshita, T.; Inoue, N.
Curr. Opin. Chem. Biol. 2000, 4, 632-638.
(20) Ferguson, J. Cell. Sci. 1999, 112, 2794-2809.
(21) Nagamune, K.; Noyaki, T.; Maeda, Y.; Ohishi, K.; Fukuma, T.; Hara, T.;
Schwarz, R.; Sutterlin, C.; Brun, R.; Reizman, H.; Kinoshita, T. Proc. Natl.
Acad. Sci. U.S.A. 2000, 97, 10336-10341.
(22) McConville, M. J.; Menon, A. K. Mol. Membr. Biol. 2000, 17, 1-18.
(23) McConville, M. J.; Ferguson, M. A. J. Biochem. J. 1993, 294, 305-324.
(24) Ferguson, M. A. J. Philos. Trans. R. Soc. London, B 1977, 352, 1295-
1302.
(25) Campbell, A. S. In Glycoscience: Chemistry and Biology; Fraser-Reid,
B., Tatsuta, K., Thiem, J., Eds.; Springer: Heidelberg, 2001; Vol. 2, pp
1696-1717.
(26) Nigou, J.; Gilleron, M.; Puzo, G. Biochimie 2003, 85, 153-166.
(27) Smith, T. K.; Sharma, D. K.; Crossman, A.; Dix, A.; Brimacombe, J. S.;
Ferguson, M. A. J. EMBO J. 1997, 16, 6667-6675.
(13) Gowda, D. C. Microbes Infect. 2002, 4, 983-990. Naik, R. S.; Branch, O.
H.; Woods, A. S.; Vijaykumar, M.; Perkins, D. J.; Nahlen, B. L.; Lal, A.
A.; Cotter, R. J.; Costello, C. E.; Ockenhouse, C. F.; Davidson, E. A.;
Gowda, D. C. J. Exp. Med. 2000, 192, 1563-1575.
(14) Treumann, A.; Lifely, M. R.; Schneider, P.; Ferguson, M. A. J. J. Biol.
Chem. 1995, 270, 6088-6099.
(15) Roberts, W. L.; Santikarn, S.; Reinhold, V. N.; Rosenberry, T. L. J. Biol.
Chem. 1988, 263, 18776-18784.
(16) For a recent synthesis of a GPI with the C2 acyl (but not the phospholipid),
see: Xue, J.; Shao, N.; Guo, Z. J. Org. Chem. 2003, 68, 4020-4029.
(17) Xue, J.; Guo, Z. J. Am. Chem. Soc. 2003, 125, 16334-16339.
(18) Guther, M. L. S.; Ferguson, M. A. J. EMBO J. 1995, 14, 3080-3093.
9
J. AM. CHEM. SOC. VOL. 126, NO. 24, 2004 7541

A R T I C L E S
Lu et al.
Figure 1. Parasite versus mammalian biosynthetic pathways for assembly of glycosylphosphatidyinositol membrane anchors.20
Scheme 2
kinetic anomeric effect,28 n-pentenyl ortho ester (NPOE) donors
have the advantage that R-oriented products are usually formed
exclusively.29 Furthermore, converting C2-OH into C2-NH2 with
inversion of configuration transforms an R-mannoside into an
R-glucosaminide. Thus, as indicated in the retrosynthetic plan
(Scheme 3), the entire glycan array of 1 can be derived from
the manno-orthobenzoate 5, which can be prepared from
D-mannose in three easy steps.29
For the myoinositol moiety, the route developed to compound
6 from methyl R-D-glucopyranoside by Bender and Budhu30
has been the most convenient in our hands.
Several total syntheses of GPIs have been reported,16,17,32 none
of which has had to confront phosphodiacylglycero lipidation
as well as inositol acylation. Anticipating that the latter feature
would present new challenges, we carried out model studies
on the inositol moiety by itself.33
The attractiveness of the strategy is further enhanced by recent
investigations which have shown that use of ytterbium triflate
(Yb(OTf)3) as activator for N-iodosuccinimide (NIS) induces
chemospecific reaction of NPOEs, while leaving armed and
disarmed NPG donors untouched.31 Additional valuable nuances
concerning choice of activators for use with NIS are the
incompatibility of p-methoxybenzyl groups with both boron
trifluoride etherate and tert-butyldimethylsilyl triflate, while
cyclohexylidene acetals tolerate the former, but not the latter.
Nearly 40 years ago, Angyal and Tate had shown that the
vicinal hydroxyl groups of diol 10 could be differentiated by
chemoselective reactions with alkylating and acylating agents
giving rise to 9 and 11 respectively as major products.34 The
desired axial acyl derivative therefore could not be achieved
directly from such a diol, so protection/deprotection steps
featuring tin-mediated chemistry35 were applied to give 12 and
thence 13 and 14.33 However, as is clear from Scheme 4, this
(32) See, for example: Pakari, K.; Schmidt, R. R. J. Org. Chem. 2003, 68,
1295-1308. Campbell, A. S.; Fraser-Reid, B. J. Am. Chem. Soc. 1995,
117, 10387-10388. Baeschlin, D. K.; Chaperon, A. R.; Green, L. G.; Hahn,
M. G.; Ince, S. J.; Ley, S. V. Chem. Eur. J. 2000, 6, 172-186. Ogawa, T.
Chem. ReV. 1994, 23, 397.
(33) Schlueter, U.; Lu, J.; Fraser-Reid, B. Org. Lett. 2003, 5, 255-257.
(34) Angyal, S. J.; Tate, M. E. J. Chem. Soc. 1965, 6949-6955.
(35) David, S. In PreparatiVe Carbohydrate Chemistry; Hanessian, S., Ed.;
Marcel Dekker: New York, 1997; pp 67-68.
(28) The Anomeric Effect and Related Stereoelectronic Effects at Oxygen; Kirby,
A. J., Ed.; Springer-Verlag: New York, 1983.
(29) Mach, M.; Schlueter, U.; Mathew, F.; Fraser-Reid, B.; Hazen, K. C.
Tetrahedron 2002, 58, 7345-7354.
(30) Bender, S. L.; Budhu, R. J. J. Am. Chem. Soc. 1991, 113, 9883-9885.
(31) Jayaprakash, K. N.; Radhakrishnan, K. V.; Fraser-Reid, B. Tetrahedron
Lett. 2002, 43, 6953-6955. Jayaprakash, K. N.; Fraser-Reid, B. Synlett
2004, 301-305.
9
7542 J. AM. CHEM. SOC. VOL. 126, NO. 24, 2004

Synthesis of a Malaria Candidate GPI Structure
A R T I C L E S
Scheme 3 a
a (i) NaOMe/MeOH/CH2Cl2. (ii) (a) (nBu3Sn)2O, benzene; (b) PhCH2Br/Bu4NI/benzene; (c) p-MeOC6H4CH2Cl/DMF/NaH/TBAI. (iii) PhCH2Br/NaH/
Bu4NI/DMF. (iv) TBDMSCl/THF/imidazole. (v) Ph3CCl/pyridine/Ac2O/DMAP/Et3N.
Scheme 4
strategy, which has also been explored by Guo,16 met with only
modest success.33
Removal of PMB gave acceptor 17, which coupled efficiently
(91%) with NPOE 7c (4 equiv) to give the pseudotrisaccharide
18a, which was converted into the perbenzylated counterpart
18b in the usual way. Removal of the cyclohexylidene acetal
with camphorsulfonic acid led to the corresponding diol, to
which the protocol in Scheme 4 (10 f f 14) was applied.
However, the O-1-allylation step went poorly, and additional
problems were experienced with the subsequent deallylation
using palladium chloride in water and acetic acid. Thus short-
chain models of 19 gave poor yields, and with long-chain
analogues the situation became worse in that there was gross
decomposition.
Testing was therefore carried out on a more advanced
substrate as outlined in Scheme 5. In keeping with the
retrosynthetic plan in Scheme 3, the n-pentenyl ortho ester
(NPOE) 7b was chosen as the donor. Preparation of this material
took advantage of the fact that the NPOE triol 7a undergoes
tin-mediated selective double alkylation of the C6- and C3-
OHs, thereby making the C4-OH available for p-methoxy-
benzylation. Acceptor 8 was prepared from the Bender-Budhu
product, 6, as described previously by us.36 Coupling of 7b (3
equivalents) with 8 under the agency of ytterbium triflate
(Yb(OTf)3)31 and N-iodosuccinimide (NIS) gave a pseudo-
disaccharide, 15a. The lone benzoate was replaced by triflate
under routine conditions to give 15b in 77% yield over two
steps. Deshong’s novel azide displacement37 was then employed.
Unfortunately, the yield of the glucosaminyl synthon 16 was
much lower than with our previously tested substrates.38 The
1H parameters of compound 16 (5.60 ppm, J12 ) 3.6 Hz) are
distinctive, and this knowledge would be valuable in making
assignments as the synthesis progressed.
A more reliable approach was therefore needed, and in view
of previous frustrations, we decided that the best “model” would
be the actual precursor. Accordingly, acceptor 17 was coupled
with the silylated NPOE 7d (5 equiv) to give pseudotrisaccharide
20a in 79% yield. As before, the lone benzoate was replaced
by benzyl in two steps, and desilylation of the resulting product,
20c, then gave the acceptor 20d. Coupling with previously used
NPOE 7c (3 equiv) gave pseudotetrasaccharide 21a, and
subsequent debenzoylation afforded acceptor 21b.
(36) Jia, Z. J.; Olsson, L.; Fraser-Reid, B. J. Chem. Soc., Perkin Trans. 1 1998,
631-632.
(37) Soli, E. D.; Manoso, A. E.; Satterson, M. C.; DeShong, P.; Favor, D. A.;
Hirschmann, R.; Smith, A. B., III. J. Org. Chem. 1999, 64, 3171-3177.
(38) Anilkumar, G.; Nair, L. G.; Olsson, L.; Danield, J. K.; Fraser-Reid, B.
Tetrahedron Lett. 2000, 41, 7605-7608.
Problems intensified with the next step. We had hoped to
use the 6-O-silylated NPOE 7d again, but for reasons which
are unclear, coupling of 7d with acceptor 21b failed to give
the desired pseudopentasaccharide 22.
9
J. AM. CHEM. SOC. VOL. 126, NO. 24, 2004 7543

A R T I C L E S
Lu et al.
Scheme 5 a
a (i) NIS/Yb(OTf)3/0.3 equiv/CH2Cl2/0 °C (98%). (ii) NaOMe/MeOH/CH2Cl2. (iii) Tf2O/Py/DMAP/CH2Cl2. (iv) TMSN3/TBAF/THF (39%). (v) BF3‚Et2O/
CH2Cl2 (86%). (vi) NIS/BF3‚Et2O/CH2Cl2/0 °C (91%). (vii) BnBr/NaH/Bu4NI. (viii) CSA/ethylene glycol (68%). (ix) Bu2SnO, toluene, then DMF, -10 °C
allyl bromide, CsF, 18 h (54%). (x) Alkyl-COCl, pyridine. (xi) PdCl2, HOAc, H2O, NaOAc.
Scheme 6 a
a (i) NIS/BF3‚Et2O/CH2Cl2/0 °C. (ii) NaOMe/MeOH/CH2Cl2 (82%). (iii) BnBr/NaH/Bu4NI (69%). (iv) TBAF/MS(4 Å)THF (83%). (v) Ph3CCl/pyridine/
DMAP/Et3N. (vi) TBDMSCl/THF/imidazole.
We have recently observed that acylated NPOEs frequently
give higher coupling yields than their alkylated counterparts,39
so the tribenzoylated NPOE 5 was tried. Indeed the expected
tetrabenzoate 23 was obtained in acceptable yield, paving the
way to corresponding tetrol 24a. Surprisingly, however, attempts
to selectively tritylate or silylate the latter, using the same
conditions as in Scheme 3, failed to give 24b or 24c. Silylation
(39) Fraser-Reid, B.; Lopez, J. C.; Radhakrishnan, K. V.; Mach, M.; Schlueter,
U.; Gomez, A.; Uriel, C. Can. J. Chem. 2002, 80, 1075-1087.
9
7544 J. AM. CHEM. SOC. VOL. 126, NO. 24, 2004

Synthesis of a Malaria Candidate GPI Structure
A R T I C L E S
Scheme 7 a
a (i) NIS/BF3‚Et2O/CH2Cl2/0 °C (75%). (ii) (a) NaOMe/MeOH/CH2Cl2; (b) BnBr/NaH/Bu4NI (73%). (iii) HCOOH/Et2O (60%). (iv) CbzNH(CH2)2OP(OBn)-
N(iPr)2/1H-tetrazole/CH2Cl2, -40 °C/MCPBA (60-70%). (v) CSA/ethylene glycol.
with TBDMSOTf could not be tried, because as noted above,
the cyclohexylidene ring would not have survived.
172.31, 172.58, and 172.99 ppm for the three ester carbonyl
groups. Debenzylation of 34a by atmospheric hydrogenolysis
over Pearlman’s catalyst finally gave 35a, confirmed by MALDI
(1385.7 M + H, 1407.9 M + Na) and NMR (5.54 ppm for
inositol’s H2).
Fortunately, the tritylated diacetylated NPOE 7e (8 equiv)
worked well to give 25 (Scheme 7), and deacetylation followed
by benzylation gave the elusive pseudopentasaccharide 26a in
73% overall yield. Mild cleavage of the trityl group afforded
26b, to which the phosphoethanolamine complex40 was attached
leading to 27. Removal of the cyclohexylidene was then readily
accomplished to give diol 28.
Given the poor yields for converting 18 into 19 (Scheme 4),
a different strategy for installing the axial acyl entity was needed.
The stereoelectronically controlled opening of cyclic ortho esters
pioneered by King and Albutt41 was an attractive option. In an
exploratory experiment with a short-chain model, Scheme 8,
treatment of diol 28 with trimethyl orthovalerate 29a afforded
the cyclic ortho ester 30a. This material was usually not isolated,
but was treated directly with ytterbium(III) triflate31 to give the
regioisomeric â-hydroxy esters 31a and 32a in acceptable yield
(75%), but a disappointing 1:2 ratio. However, reaction with
the readily prepared40 diacylated glycerylphosphoamidite 33a
proceeded well to give the fully lipidated glycan 34a. The latter
structure was supported by FAB (3010.4 M + Na), by signal(s)
at 5.89 ppm for the inositol’s H2, and by carbon signals at
We were encouraged by these results, but would 35a, with
its short alkyl chains, be able to mimic naturally occurring long-
chain GPI anchors in its ability to be embedded into the cell
membrane? Accordingly, O-2 myristoylation was effected via
the trimethyl ortho ester 29b, the latter being prepared from
the corresponding myristonitrile by the method of Presova and
Smrt.42 The derived cyclic ortho ester 30b was rearranged
efficiently (86%) by treatment with Yb(OTf)3, and gratifyingly
with greater regioselectivity than had been achieved with the
short-chain analogue 30a, there being a 5:1 ratio of axial (32b)
and equatorial (31b) esters.
Longer chains were also placed on the glyceryl moiety in
the form of phosphoramidite 33b, the use of which led to the
perbenzylated GPI 34b smoothly in 72% overall yield. The
structure was confirmed by MALDI (3471.01 M + Na), and
the parameters for proton (5.88 ppm) and carbon (172.34,
172.55, and 172.96 ppm) NMR spectra were in excellent
agreement with the short-chain analogue 34a.
(40) Campbell, A. S.; Fraser-Reid, B. Bioorg. Med. Chem. 1994, 2, 1209-
1219.
(41) King, J. F.; Allbutt, A. D. Can. J. Chem. 1970, 48, 1754-1759.
(42) Presova, M., Smrt, J. Collect. Czech. Chem. Commun. 1989, 54, 487-
497.
9
J. AM. CHEM. SOC. VOL. 126, NO. 24, 2004 7545

A R T I C L E S
Lu et al.
Scheme 8 a
a (i) CSA 0.5 h, CH3CN. (ii) Yb(OTf)3; CH2Cl2. (iii) 1H-Tetrazole/CH2Cl2, -40 °C/MCPBA (60-70%). (iv) CHCl3/MeOH (1:1) HOAc(trace)/Pd/C
(10%) room temperature slightly over atmospheric pressure 12 h; add H2O (0.3), room temperature 12 h.
The final global debenzylation step proved to be extremely
frustrating. Procedures that had succeeded with previous GPI
syntheses in our lab and elsewhere,32 and even with the short-
chain prototype43 of 34a, failed with the more lipophilic
substrate 34b. The presence of the myristoyl ester on inositol
of 34b clearly presented unprecedented problems. First, after
12 h of hydrogenolysis at 70 psi in chloroform-methanol-
water mixture (3:3:1) over palladium hydroxide on carbon, some
benzyl groups were still present. However, there had been partial
loss of the glyceryl acyl group(s)sbut notably the myristoyl
ester was not affected.
Second, we tried palladium on carbon (10%) in methanol:
chloroform (1:1) mixture with a trace of acetic acid using a
balloon of hydrogen for 12 h. The isolated product, upon TLC
in butanol:ethanol:water (1:1:1), indicated the presence of more
1
polar material, but H NMR showed that, while there was no
more starting material (34b), there were still some benzyl groups
present.
(43) Lu, J.; Jayaprakash, K. N.; Frasaer-Reid, B. Tetrahedron Lett. 2004, 45,
879-882.
9
7546 J. AM. CHEM. SOC. VOL. 126, NO. 24, 2004

Synthesis of a Malaria Candidate GPI Structure
A R T I C L E S
Third, the recovered marterial was redissolved in methanol:
chloroform:water (1:1:0.3) mixture and hydrogenolysis was
continued for a further 12 h. This gave and even more polar
materialswhich proved to be 35b.
Fourth, in view of the preceding observation, we tried to use
the last solvent mixture, methanol:chloroform:water (1:1:0.3),
with 34b right from the beginning. Unfortunately, the desired
product was not obtained after 12 h!
1H NMR easily gave no evidence of any acyl migration over
several months. The same conclusions were reached for the other
two substrates, based on the signal at 5.43 ppm that is assignable
to the equatorial proton R to the acyl group. This conclusion
could be reached by comparing the intensity of this signal to
that of the well-resolved anomeric protons at 5.03, 5.12, and
5.22 ppm. (As a point of interest, we have also synthesized the
diastereomer of 35a with the inositol’s C1 and C2 functionalities
reversed. The axial proton R to the acyl group comes to a much
higher field, ∼4.7 ppm.)
Finally, further prodigious experimentation revealed that the
best strategy for hydrogenolytic deprotection of 34b was to start
with methanol:chloroform (1:1) mixture and acetic acid (trace)
with balloon pressure for 12 h. At this time the experiment was
briefly interrupted, water (0.3) was added, the apparatus was
reassembled, and hydrogenolysis was continued for a further
12 h. This ensured complete debenzylation to give 34b.
The product was purified by chromatography using a Sepha-
dex G-15 column with chloroform:methanol:water (1:1:0.3) as
eluent. Electrospray mass spectroscopy of the product, 35b, in
methanol/formic acid gave m/z 1937.1 (M + 2HCOOH). The
1H NMR spectrum showed a 6-proton signal at 2.29 ppm for
the three methylene groups R to carboxyl, nine-proton over-
lapping triplets at 0.91 ppm for the three fatty acyls, and the
telling one-proton signal at 5.54 ppm for the inositol H-2.
With respect to the crucial question of migration of the
inositol’s acyl group, we have examined three substrates: (1)
O-2-pentanoylmyoinositol obtained by hydrogenolysis of 14,
(2) the fully lipidated pseudodisaccharide corresponding to units
I and II of 15, and (3) the synthetic prototype 35a. In the first,
Acknowledgment. We are grateful to the World Health
Organization, the Human Frontier Science Program Organiza-
tion, the National Science Foundation, and the Mizutani
Foundation for support. We thank Dr. George Dubay, Director
of Instrument Operations in the Department of Chemistry, Duke
University, for the mass spectroscopic and MALDI measure-
ments. Natural Products and Glycotechnology Research Institute,
Inc., is an independent, nonprofit research facility with labo-
ratories at Centennial Campus, North Carolina State University,
Raleigh, NC.
Supporting Information Available: Synthesis of 7b,e, 15a,b,
16, 17, 18b, 20a-d, 21a,b, 26a,b, 27, 28, 31a, 32a,b, 34a,b,
35a,b; NMR spectra of 18e, 19a, 22b, 24a, 24b, and 25a.This
material is available free of charge via the Internet at
http://pubs.acs.org.
JA038807P
9
J. AM. CHEM. SOC. VOL. 126, NO. 24, 2004 7547