Carbohydrates mediate a wide range of biological processes, and understanding these events and how they might be influenced is a complex undertaking that requires access to pure glycoconjugates. The isolation of sufficient quantities of carbohydrates and glycolipids from biological samples remains a significant challenge that has redirected efforts toward chemical synthesis. However, progress toward complex glycoconjugate total synthesis has been slowed by the need for multiple protection and deprotection steps owing to the large number of similarly reactive hydroxyls in carbohydrates. Two methodologies, regioselective silyl exchange technology (ReSET) and glycosyl iodide glycosylation have now been integrated to streamline the synthesis of the globo series trisaccharides (globotriaose and isoglobotriaose) and α-lactosylceramide (α-LacCer). These glycoconjugates include tumor-associated carbohydrate antigens (TACAs) and immunostimulatory glycolipids that hold promise as immunotherapeutics. Beyond the utility of the step-economy syntheses afforded by this synthetic platform, the studies also reveal a unique electronic interplay between acetate and silyl ether protecting groups. Incorporation of acetates proximal to silyl ethers attenuates their reactivity while reducing undesirable side reactions. This phenomenon can be used to fine-tune the reactivity of silylated/acetylated sugar building blocks.
Carbohydrates mediate a wide range of biological processes, and understanding these events and how they might be influenced is a complex undertaking that requires access to pure glycoconjugates. The isolation of sufficient quantities of carbohydrates and glycolipids from biological samples remains a significant challenge that has redirected efforts toward chemical synthesis. However, progress toward complex glycoconjugate total synthesis has been slowed by the need for multiple protection and deprotection steps owing to the large number of similarly reactive hydroxyls in carbohydrates. Two methodologies, regioselective silyl exchange technology (ReSET) and glycosyl iodide glycosylation have now been integrated to streamline the synthesis of the globo series trisaccharides (globotriaose and isoglobotriaose) and α-lactosylceramide (α-LacCer). These glycoconjugates include tumor-associated carbohydrate antigens (TACAs) and immunostimulatory glycolipids that hold promise as immunotherapeutics. Beyond the utility of the step-economy syntheses afforded by this synthetic platform, the studies also reveal a unique electronic interplay between acetate and silyl ether protecting groups. Incorporation of acetates proximal to silyl ethers attenuates their reactivity while reducing undesirable side reactions. This phenomenon can be used to fine-tune the reactivity of silylated/acetylatedsugar building blocks.
For the past five decades,
cancer has been one of the top 10 causes
of death in the United States, and the incidence rate is gradually
increasing.[1,2] Cancer treatments and therapeutics to eradicate
the disease are being intensively investigated in an effort to improve
quality of life for those who suffer. Carbohydrate-based vaccine development
is one promising approach to this end.[3−9] Cell membranes are decorated with carbohydrates in the form of glycolipids
and glycoproteins with unique structures and aberrant glycosylation
patterns that are correlated with tumor progression and metastases.[10−14] Vaccines based on the carbohydrate epitopes of these glycoconjugates
are promising therapeutic targets, as are the more recently discovered
bacterial derived glycolipids that stimulate immune response. Limited
access to sufficient quantities of these biomolecules is a discovery
roadblock making practical synthesis of complex carbohydrates a top
priority in a recent National Academy of Sciences publication on the
future of glycoscience.[15]One major
class of mammalian glycosphingolipids (GSLs) is the globo
series tumor-associated carbohydrate antigens (TACAs).[3,4,10,14,16] Globotriaosyl ceramide (Gb3), isoglobotriaosyl
ceramide (iGb3), Gb4, Gb5, and Globo H are the prominent members of
this class (Figure 1). All of these biomolecules
share a lactose core, which is diversified by galactosylation at either
the 3′ or 4′ hydroxyls giving rise to iGb3 or Gb3, respectively.
Gb3 in turn is the core structure shared by Globo H and Gb5. Given
the centrality of the Gb3 core, its synthesis has been the focus of
numerous investigations. Nicolaou and co-workers were the first to
report the total synthesis of Gb3, and their approach remains one
of the most efficient to date.[17,18] A total of eleven steps
from commercially available lactose were required to prepare a protected
globotriaose that served as a donor for the ceramideaglycon. Seven
of those steps were focused on orthogonally protecting lactose for
4′-galactosylation. Similarly, all syntheses of globotriaose/isoglobotraose
that have engaged a lactosyl acceptor have required between 6 and
12 steps to prepare the disaccharide building block.[16−29]
Figure 1
Globo
series: A class of common TACAs.
Globo
series: A class of common TACAs.While β-linked ceramides are known TACAs, bacterial
derived
α-linked analogues also stimulate an immune response and hold
promise as vaccine adjuvants.[30−33] Isolating these glycolipids from their natural environments
is a challenging process that can lead to decomposition or even rearrangement
of the molecules of interest.[34] The lipid
composition of the ceramide is also usually obtained as an inseparable
and noncharacterizable mixture. For these reasons, chemical synthesis
of α-linked ceramides is the primary source of pure and homogeneous
materials for immunological studies. To meet this need, our group
developed one-pot syntheses of KRN7000, BbGL-II and α-GalCer
(Figure 2) and their glucosyl analogues,[35,36] while the Wang[37] and Savage[24] groups have reported syntheses of α-lactosylceramide,
α-Gb3 and α-iGb3. These novel “sugar-capped”
CD1d ligands for natural killer T (NKT) cells have been tested for
their ability to stimulate cytokine release. The results established
that the α-linkage is required for immune response and that
the oligosaccharide structures serve as immunomodulators. These findings
and biological evaluations of TACAs clearly point to an emerging area
of immunotherapeutic discovery based upon α- and β-linked
ceramides, and access to sufficient quantities of these glycoconjugates
is a critical need.
Figure 2
Examples of immunostimulatory glycolipids.
Examples of immunostimulatory glycolipids.Recently, the concept of step-economy was introduced
and has subsequently
been advocated among synthetic chemists.[38−41] Instead of the traditional linear
synthesis or tour de force total synthesis, considerations of semisynthesis,
diversity-oriented synthesis, function-oriented synthesis, and convergent
synthesis can lead to high efficiencies in the production of drug
leads. In applying this concept to glycoconjugate syntheses, eliminating
the need for multiple protection/deprotection steps is an obvious
starting point. Per-O-TMS protected monosaccharides
have proven to be especially useful starting materials for step-economy
protecting group manipulations.[42−44] We have exploited these readily
available compounds in a process coined regioselective silyl exchange
technology (ReSET).[45,46] In just two steps from free sugars,
a wide range of partially acetylated/silylated carbohydrates can be
prepared and readily transformed to either the corresponding glycosyldonor or acceptor (Figure 3). The introduction
of electron-withdrawing acetate groups to silylsugars affords bifunctional
modules that are more stable than their per-O-TMS
counterparts, yet highly reactive in glycosylation reactions. The
anomeric acetate can be converted into various leaving groups including
halides, thioethers and acetimidates. Efforts reported herein have
concentrated on integrating glycosyl iodide glycosylation with a tandem
ReSET strategy to accomplish step-economy syntheses of globo series
trisaccharides and α-lactosylceramide. The dual methodology
platform expands the current organic synthesis toolbox and provides
new insight into the electronic interplay of acetate and silyl ether
functionalities and their influence upon chemical reactivity.
Figure 3
Application
of ReSET to generate bifunctional modules for oligosaccharide
and glycoconjugate syntheses.
Application
of ReSET to generate bifunctional modules for oligosaccharide
and glycoconjugate syntheses.
Results and Discussion
Tandem ReSET to Prepare Bifunctional Lactose
Modules
ReSET begins with per-O-silylated
sugars, which
undergo selective exchange of silyl ethers for acetate protecting
groups. The reactions are typically run in pyridine with excess acetic
anhydride, and exchange is mediated by the addition of acetic acid.
Regiocontrol is correlated with acetic acid stoichiometry and microwave
reaction time. Typically, the reactions afford a mixture of products,
all of which are useful in making a library of different analogues.
For example, treating per-O-silylated lactose with
3.0 equiv of AcOH for 1.25 h leads to production of the di- and triacetylated
compounds (1 and 2, Scheme 1) in 20 and 53% yields, respectively, whereas increasing the
amount of acetic acidto 7 equiv forces the production of compound 4 having only one silyl ether at the 4′ position and
per-O-Ac lactose (5) after 3.75 h.[46] Since the lactosyl 3′ and 4′ positions
are the major glycosylation sites for globo series antigens, we attempted
to establish conditions for preparing the disilyl ether analogue (3), but that compound could not be directly obtained in ReSET
reactions of per-O-silylated lactose under any of
the conditions evaluated. Previous studies in our lab indicated that
proximal acetate groups facilitate silyl exchange;[46] thus, we reasoned that greater regiocontrol might be achieved
using compounds 1 and 2 in a tandem process.
To our delight, the reaction proceeded nicely to afford the desired
analogues, 3 and 4, in 31 and 25% yields,
respectively. With these partially acetylated building blocks (1–4) in hand, focus shifted to employing
them as either glycosyl donors or acceptors in oligosaccharide and
glycoconjugate syntheses.
Scheme 1
Tandem ReSET Affords Differentially Protected
Lactose Modules
ReSET Products as Glycosyl
Donors: α-Lactosylceramide
Synthesis
Previous attempts to form and utilize per-O-TMS-lactosyl iodide led to complex reaction mixtures due
to glycosidic bond cleavage and silyl exchange.[47] We later discovered that C-6 acetates protect glycosidic
linkages from TMSI degradation.[48,49] Consistent with these
findings, we were able to cleanly and quantitatively generate the
lactosyl iodide from both 1 and 2 (Scheme 2A).[46] Synthetic ceramide
(7)[36,50] was selected as the acceptor
in the glycosylation studies due to its biological relevance. Iodide 6 was reconstituted in dry benzene and cannulated into a mixture
of 7, tetrabutylammonium iodide (TBAI) and diisopropylethylamine
(DIPEA). After heating the reaction to 65 °C in anhydrous benzene
overnight, the reaction mixture was concentrated and subjected to
acidic methanolysis prior to per-O-acetylation for
characterization purposes (Scheme 2B). A mixture
of the per-O-acetylated isomers was isolated in 89%
yield, and the ratio of primary adduct (8) to secondary
adduct (9) was 1.5 to 1.0, respectively, as determined
by anomeric proton integration values obtained by 1H NMR.
Scheme 2
Synthesis of α-Lactosylceramide
The in situ anomerization process promoted by TBAI resulted in
the exclusive formation of α-linked glycosides; however, multiple
flash column chromatography purifications were required to separate 8 and 9. To alleviate this complication, compound 10 was prepared by selective desilylation of di-O-TMSceramide[42] according to a method
recently published by Cui and co-workers.[51] Ceramide 10 was then coupled with 6 under
the TBAI-promoted conditions (Scheme 2B). After
acidic methanolysis and peracetylation, compound 8 was
the only glycosylation product isolated. The glycosylation yield (61%)
was slightly decreased compared to the glycosylation result of ceramide 7 (89%), possibly due to the result of increased steric hindrance
from the secondary TMS ether. Nevertheless, deacetylation of 8 using NaOMe/MeOH led to fully deprotected α-lactosylceramide
(11) in quantitative yield.The synthesis of α-lactosylceramide
has several salient features:
(1) the glycosyldonor (6) can be prepared in three steps
and 72% overall from free lactose; (2) the lactosyl iodide is reactive
enough to couple with unprotected or partially protected ceramides,
yet the inter-residue glycosidic bond is stable enough to survive
glycosylation; (3) only the desired α-linkage is obtained in
good yield; (4) only TMS ether and acetate protecting groups are utilized,
allowing mild deprotection steps that are compatible with alkene and
amide functionality in the ceramide component. This methodology nicely
complements reactions of per-O-acetylatedlactose,
which afford the β-anomer due to neighboring group participation
of the C-2 acetate.[52]
ReSET Products As Acceptors: Globo Series Trisaccharide Syntheses
Having demonstrated glycosylation efficiency with partially acetylatedsilyl donors, we set our sights on exploiting these substrates as
glycosyl acceptors. Chemical syntheses of the Globo series TACAs involve
the construction of the crucial α-1,3 and α-1,4 glycosidic
linkages. Yet a simple first order disconnection of globotriaose and
isoglobotriaose at these linkages leads to 4′-OH acceptor 12, 3′,4′-di-OH acceptor 13, or
3′-OH acceptor 14 (Figure 4), which are all readily available from the tandem ReSET products 2, 3, and 4.
Figure 4
Retrosynthetic analysis
of globo series trisaccharides.
Retrosynthetic analysis
of globo series trisaccharides.To reduce the glycosylation strategy to practice, compound 4 was treated with Dowex acidic resin in MeOH to give the
4′-OH acceptor 12 in 90% yield (Scheme 3A). The long reaction time (10 h) was consistent
with earlier results in our lab indicating that the rate of protodesilylation
is attenuated with increasing numbers of acetate protecting groups.[46] In an attempt to shorten the reaction time,
we initially subjected 4 to TBAF/AcOH, but the reaction
led to unresolved trans-acetylation products. However, a report from
Ikawa and co-workers[53,54] encouraged us to explore the
deprotection of silyl ethers using Pd-catalyzed hydrogenolysis. Much
to our delight, after 30 min under 1 atm H2, 4 was transformed to 4′-OH acceptor 12 in 91%
yield. In a similar manner, either acidic methanolysis or Pd-catalyzed
hydrogenolysis was applied to 3, leading to 3′,4′-di-OH
compound 13 in 92 and 91% yields, respectively (Scheme 3B). To confirm the deprotection was the result of
hydrogenolysis and not acidic catalysis, a control experiment without
the introduction of H2 gas was conducted. No reaction was
observed after 2 h, indicating that H2 gas is required
for the Pd-catalyzed deprotection.
Scheme 3
Synthesis of Acceptors 12 and 13
In order to prepare 3′-OH acceptor 14, we looked
to the work of Lin and co-workers, who published a one-pot procedure
to selectively acetylate at the 4-position of galactose via 3,4-orthoester
formation followed by selective acidic hydrolysis.[55] The procedure showed excellent results when applied to
monosaccharides but not oligosaccharides, presumably because of solubility
issues.[56] An adapted version of the methodology
was applied to ReSET products 2 and 3 (Scheme 4). Compound 2 was first subjected to
Pd-catalyzed hydrogenolysis and then concentrated to dryness. The
resulting residue was reacted with trimethyl orthoacetate in the presence
of catalytic camphor sulfonic acid (CSA) to form the cyclic orthoester
at the 3′ and 4′ positions. After peracetylation, the
orthoester was selectively hydrolyzed affording the 3′-OH acceptor
(14) in almost quantitative yield. The four-step procedure
was carried out in less than 3 h, and only one flash column chromatography
purification was needed to obtain the target molecule. The consecutive
hydrogenolysis–orthoester formation–acidic hydrolysis
protocol was also applied to di-O-TMS compound 3, leading to the same 3′-OH acceptor 14 in almost quantitative yield (Scheme 4).
Scheme 4
Synthesis of 3′-OH Acceptor 14
With all the acceptors (12–14)
in hand, we next examined the glycosylation reactions to form the
globo series trisaccharides (Table 1). Wishing
to achieve efficient α-galactosidation of the 4′-OH acceptor
(12), we first tried using per-O-TMSgalactosyl iodide promoted by TBAI in situ anomerization,[35,36] but the major product obtained was silylated acceptor. We previously
observed similar trans-silylation complications, especially in cases
where the acceptor is hindered or unreactive.[57,58] However, the per-O-benzyl galactosyl iodide derived
from anomeric acetate 15(59) was reactive when activated with AgOTf, and the presence of benzyl
protecting groups conveniently allowed UV monitoring of the reaction
progress. Importantly, only the desired α-1,4 linkage was obtained,
which simplified the purification (Table 1,
entry 1). The same reaction conditions with 1.5 and 2.5 equiv of iodide
were applied to the glycosylation reaction with acceptor 13 to examine the relative reactivity of the two hydroxyl groups. When
1.5 equiv of iodide was used, the desired protected isoglobotraose
(17) was obtained, albeit in a lower yield of 44% (Table 1, entry 2). When 2.5 equiv of iodide was added,
not only was 17 obtained in 26% yield, but also the digalactosylated
product (18) was obtained in 31% yield (Table 1, entry 3).
Table 1
Glycosyl Iodide Glycosylation Gave
Gb3 and iGb3 Trisaccharide Scaffolds
Condition: 3.0 equiv of AgOTf,
3.0 equiv of tetramethylurea (TMU), 4 Å MS, CH2Cl2, −30 °C to rt, 24 h.
To avoid over-galactosylation,
the 3′-OH acceptor 14 was introduced. In this
case, the desired product (19) was obtained in 58% yield
(Table 1, entry 4). Side reactions, which we
attributed to acyl migration,
were also observed on the TLC plate. To minimize this possibility,
an acid scavenger (tetramethylurea, TMU)[60] was added to the AgOTf-promoted glycosylation (Table 1, entry 5). The neutralized glycosylation procedure required
higher temperature and longer time to complete; however, the target
molecule (19) was obtained in 75% yield, indicating that
the side reactions could be suppressed by the addition of weak base.Condition: 3.0 equiv of AgOTf,
3.0 equiv of tetramethylurea (TMU), 4 Å MS, CH2Cl2, −30 °C to rt, 24 h.Debenzylation of compound 16 followed
by reacetylation
led to per-O-Ac globotriaose 20 in 95%
yield (Scheme 5). Storing the trisaccharide
in the peracetylated form was desired to increase stability. Moreover,
the 1H NMR signals of compound 20 were better
resolved than the benzylated counterpart (16), making
compound characterization and quality control more reliable. Acetate
protecting groups are also preferred when generating the glycosyliodide of the trisaccharides. When compound 16 was treated
with TMSI, the reaction became messy and glycosidic bond cleavage
products were observed in crude MS analyses; another example of electron
donating protecting groups rendering the glycosidic linkage susceptible
to cleavage. In contrast, peracetylated 20 readily transformed
to the corresponding iodide (22) in situ, and subsequent
reaction leads to β-linked glycoconjugates.
Scheme 5
Six-Step
Synthesis of Globotriaose from Free Lactose
Global deprotection
of 16 continued with Pd-catalyzed
hydrogenolysis followed by deacetylation yielding globotriaose (21) in nearly quantitative yield (Scheme 5). The total synthesis of globotriaose required only four
steps from free lactose to form the globotriaose scaffold, and two
more consecutive protecting group manipulations (total six steps)
to the natural product globotriaose (21).In 1988, Nicolaou and co-workers published the first synthesis
of Gb3, which is arguably the most efficient synthesis until now.[17] The approach required seven steps to prepare
a selectively protected 4′-OH acceptor, which was glycosylated
with per-O-Bn galactosyl fluoride under the activation
of AgCl4 and SnCl2. During the process, seven
protecting groups (OAc, OBn, OBz, OPiv, SPh, F, and benzylidene) were
utilized. In contrast, the simplified RESET/glycosyl iodide synthetic
platform involves only three steps to prepare the 4′-OH glycosyl
acceptor and arrives at the globotriaose scaffold in four steps. Moreover,
the glycosyl iodides react under neutral conditions avoiding toxic
tin reagents as the promoter for α-1,4 glycosylation.Compounds 17 and 19 could be transformed
to per-O-Ac isoglobotriaose (23) in
85% yield by hydrogenolysis and peracetylation (Scheme 6). In 2009, Castillon and co-workers published an efficient
TBAI/AW-300 MS-promoted synthesis of iGb3 using per-O-Ac isoglobotriaose iodide and stannylceramide.[61] The group purchased isoglobotriaose and peracetylated it
prior to generating the iodide. In the total synthesis approach to
iGb3 reported here, both 23 and the in situ generated
iodide intercept the Castillon synthesis, providing a formal synthesis
of iGb3.
Scheme 6
Formal Synthesis of iGb3 from Compound 17
Among the published syntheses
of isoglobotriaose scaffold using
3′,4′-di-OH acceptors, galactosylation typically suffers
from low yields and poor stereoselectivity. Moreover, the 3′,4′-di-OH
acceptors require at least six or seven steps to prepare from free
lactose. When using acetylatedgalactosyl bromide[24] and trichloroacetimidate[26] donors,
the glycosylation yields ranged from 28 to 31% with inseparable α-1,3
and β-1,3 isomers. In order to eliminate side reactions and
increase donor reactivity, both Ogawa[25] and Schmidt[27] have introduced O-benzyl-protected galactosyl thiomethyl and trichloroacetimidate
donors. These reactive donors provided increased yields but did not
prevent the formation of β-linked isomers. In contrast, O-benzyl protected galactosyl iodide (Table 1, entries 2 and 3) readily reacts with acceptors 12, 13, and 14 to afford the α-linked
products exclusively and in yields ranging from 44 to 75%.Selectively
protected 3′-OH acceptors generally afford better
yields than 3′,4′-di-OH acceptors, since there is no
competition between the two hydroxyl groups. However, 6–8 steps
were required in order to prepare the 3′-OH acceptors. The
protecting group manipulations involved OAc, OBn, OPiv, OPMB, and
SPh groups, and the preparation took days to complete.[28,29] In comparison, the ReSET approach (Table 1, entry 4 and 5) required only three protecting groups (OBn, OAc,
and OTMS) and 21 h to synthesize acceptor 14. The optimized
glycosylation of compound 14 reached 75% yield when coupled
with per-O-Bn galactosyl iodide.
Conclusion
The combined ReSET/glycosyliodide glycosylation strategy provides
a step-economy platform for glycoconjugate synthesis that centers
on the conversion of per-O-TMS-lactose into selectively
protected modules with differential reactivities.[62] The orthogonally protected intermediates can be transformed
into reactive glycosyl iodides in situ and coupled with unprotected
or partially protected ceramides with high stereoselectivity in good
yields. The marriage of these powerful platforms results in the exclusive
formation of alpha-linked glycosides, which leads to increased efficiencies
during the purification process. The bifunctional nature of these
modules can be further exploited upon removal of the TMS groups yielding
glycosyl acceptors ready for sugar chain elongation. When introduced
into glycosylation reactions, the corresponding globo series tumor-associated
carbohydrate antigens can be obtained in respectable yields. The TACA
syntheses require only three protecting groups (OAc, OBn, and OTMS),
which significantly reduces the number of protection/deprotection
steps, not only in preparing the glycosyl acceptors, but also by direct
activation of the anomeric acetate or silyl ether to the glycosyliodide. These combined features characterize a versatile synthetic
platform for the rapid assembly of biologically relevant glycolipids.Beyond the time and step efficiencies of these methodologies, a
unique interplay between acetate and silyl ether protecting groups
is revealed. The acetate protecting groups help suppress side reactions
such as silyl migration and inter-residue glycosidic bond cleavage
during TMSI-promoted iodide formation. Similarly, we find that proximal
acetates significantly alter the reactivity of silyl ethers toward
protodesilylation. This phenomenon is clearly evidenced in the acidic
methanolysis of 4′-OTMS acetylatedlactose (4),
which took twice as long as analogue 3 having one less
acetate, i.e., 3′,4′-di-OTMS acetylatedlactose. We
attribute the reactivity attenuation to acetate electron withdrawing
effects, which inductively reduce the basicity of the ether and acetal
oxygen atoms. Exquisite control is afforded by acetate incorporation,
as shown in the contrasting behavior of per-O-silylated
lactose and 6,6′-di-O-Ac-per-O-silyllactose (1) under the action of TMSI. Di- and
trisaccharide substrates having no acetates undergo inter-residue
glycosidic bond cleavage, whereas incorporation of only two acetate
groups at the C-6 positions directs reactivity toward the reducing
end acetal resulting in the quantitative generation of the corresponding
glycosyl iodides. These findings offer opportunities in systems removed
from carbohydrate substrates where one could capitalize on the concept
of attenuating ether reactivity by the inductive effects of proximal
protecting groups.
Experimental Section
Procedure
for Tandem Regioselective Silyl Exchange (ReSET) of
Per-O-TMS Lactose (Scheme 1)
In a 10 mL microwave reactor vessel with a dry stir bar,
per-O-TMSlactose (500 mg, 0.54 mmol) was dissolved
in anhydrous benzene (3 mL). The solvent was removed under rotary
evaporation with argon backfilling. The azeotropic distillation was
repeated two additional times to dryness affording viscous syrup.
To the reaction flask was added anhydrous pyridine (2.0 mL/per mmol
TMS sugar: 1.1 mL), Ac2O (1.5 mL/per mmol TMS sugar: 8.1
mL), and AcOH (93 μL, 1.62 mmol, 3.0 equiv). The reaction vessel
was subjected to microwave irradiation (standard mode, 100 W, 125
°C, ∼40 psi) for 1.25 h. Once TLC showed the reaction
was complete, the reaction mixture was transferred into a 50 mL round-bottom
flask, where it was azeotroped with copious amounts of anhydrous benzene
to remove excess reagents and solvent. The crude mixture was immediately
purified by gradient flash column chromatography (EA/Hex/NEt3 = 5:85:10 to 14:85:1 to 29:70:1) to afford compound 1 (93 mg, 20%) and 2 (239 mg, 53%). Later on, compound 1 and 2 were combined. A mixture of compound 1 and 2 (194 mg, ∼0.23 mmol: calculated
on the basis of compound 2′s molecular weight;
same as the following) was then treated with dry pyridine (0.35 mL),
Ac2O (0.46 mL) and AcOH (40 μL, 0.69 mmol, ∼3.0
equiv), followed by microwave irradiation (standard mode, 100 W, 125
°C, ∼40 psi) for 1.5 h. After the starting material was
completely consumed, the reaction was azeotroped with dry benzene
to remove excess reagents and solvent. The resulting residue was immediately
purified by gradient flash column chromatography (EA/Hex = 40:60 to
60:40) to afford compound 3 (54 mg, ∼31%), compound 4 (41 mg, ∼25%) and compound 5 (42 mg,
∼26%) as white foams.
General Procedure
for α-Lactosylceramide Synthesis (Scheme 2)
In an oven-dried NMR tube, compound 2 (59
mg, 0.07 mmol) was added and dissolved in dry benzene-d6 (0.7 mL). After TMSI (12 μL, 0.08 mmol)
was introduced to the reaction vessel, the reaction was kept and 0
°C and gradually warmed to rt over 4–5 h. The reaction
was monitored by proton NMR until the corresponding iodide (6) formed in situ. Next, compound 6 was transferred
to a 25 mL pear-shape bottle, azeotroped with dry benzene (3 mL ×
3) and dried under a high vacuum for 1 h to afford the iodide as a
light yellow foam. Note: The iodide is highly reactive and
moisture sensitive. Column chromatography or aqueous workup should
be avoided. The compound 6 was next dissolved
in dry benzene (2 mL) and kept under Ar. In a separate round-bottom
flask, ceramide 7 (20 mg, 0.035 mmol), TBAI (52 mg, 0.14
mmol), DIPEA (25 mL, 0.14 mmol) and activated 4 Å molecular sieves
(80 mg) were dissolved in dry bezene (2 mL) and stirred at rt under
Ar. The iodide solution was transferred to the glycosyl acceptor solution
via cannula dropwisely over 3 min at rt. Once transferred, the reaction
mixture was gently heated to 65 °C overnight. Next, the reaction
mixture was cooled to rt, filtered through a short Celite pad and
concentrated under reduced pressure. The resulting residue was dissolved
in MeOH (5 mL) and stirred with the Dowex acidic resin (pH = 2–3)
at rt. After 2 h, the resin was filtered, and the solvent was evaporated
under reduced pressure to afford a viscous orange oil. The resulting
oil was then dissolved in CH2Cl2 (1 mL) under
Ar. To the solution was then added Ac2O (0.8 mL), NEt3 (1.2 mL) and catalytic DMAP (∼2 mg), and the mixture
was stirred at rt overnight. Next, the reaction mixture was concentrated
under reduced pressure and was purified using flash column chromatography
(EA/Hex = 50:50) to afford a mixture of isomers (compounds 8 and 9) (39 mg, 89%) as a white foam. The mixture could
be further purified by gradient flash column chromatography (EA/Hex
= 40:60 to 60:40) to separate both isomers for characterization purpose.
Mono-O-TMS protected ceramide 10 (19
mg, 0.03 mmol) could also be applied to the above reaction conditions
but using 50 °C in the glycosylation step. Followed by desilylation
and reacetylation, compound 8 (22 mg, 61%) was obtained
as the only isomer.
Method A: To a MeOH (3 mL) solution of
compound 3 (41
mg, 0.06 mmol) was added Dowex H+ resin (∼80 mg)
until the pH = 2–3. The suspension was allowed to stir at rt
for 4.5 h until the starting material was completely consumed. Next,
the Dowex acidic resin was removed via filtration, and the filtrate
was concentrated to afford a viscous oil. The resulting residue was
immediately purified by flash column chromatography (100% EA) to obtain
compound 13 (30 mg, 92%) as a white foam. Method B: To
a MeOH solution of compound 3 (115 mg, 0.16 mmol) was
added Pd(OH)2/C (20% Pd, 100 mg). The reaction mixture
was allowed to stir under H2 gas (1 atm) at rt for 0.5
h until the starting material was completely consumed. Next, the Pd-catalyst
was removed by a short plug of MeOH-packed Celite. The filtrate was
concentrated to afford a viscous oil. The resulting residue was purified
by flash column chromatography (100% EA) to obtain compound 13 (85 mg, 91%) as a white foam. The product consisted of
inseparable α/β anomeric acetates, and the major β-anomer
is reported: TLC (100% EA) R 0.36; 1H NMR (600 MHz, CDCl3) δ
5.67 (d, J = 8.8 Hz, 1H, H-1), 5.23 (appt. t, J = 8.8 Hz, 1H, H-3), 5.07 (appt. t, J =
8.8 Hz, 1H, H-2), 4.81 (appt. t, J = 8.4 Hz, 1H,
H-2′), 4.46 (d, J = 12.4 Hz, 1H, H-6a), 4.38-.34
(m, 2H, H-1′, H-6b), 4.28–4.22 (m, 1H, H-6′a),
4.22 (m, 1H, H-6′b), 3.84–3.77 (m, 3H, H-4, H-5, H-4′),
3.69–3.54 (m, 2H, H-3′, H-5′), 2.13 (s, 3H),
2.12 (s, 3H), 2.11 (s, 3H) 2.10 (s, 3H), 2.06 (s, 2H), 2.03 (s, 2H); 13C NMR (150 MHz, CDCl3) δ 171.9, 171.0, 170.4,
169.9, 169.5, 168.9, 100.5, 91.7, 75.6, 74.1, 73.7, 72.9, 72.4, 72.0,
70.4, 68.1, 62.1, 61.9, 20.9, 20.84, 20.82, 20.8, 20.76, 20.6; HRMS
(ESI-ion trap) m/z calcd for [C24H34O17 + NH+] 612.2134,
found 612.2134.
Method A: To a MeOH (3 mL) solution of
compound 2 (70 mg, 0.08 mmol) was added Pd(OH)2/C (20% Pd, 50 mg). The reaction mixture was allowed to stir under
H2 (1 atm) at rt for 0.5 h. Next, the reaction mixture
was filtered through a MeOH-packed Celite pad to remove the Pd catalyst,
and the filtrate was concentrated under reduced pressure. The resulting
residue was then dissolved in dry acetonitrile (2 mL) and treated
with trimethyl orthoacetate (32 μL, 0.25 mmol) and catalytic
camphorsulfonic acid (∼4 mg, 0.02 mmol). After 1 h, NEt3 (∼0.1 mL) was added to quench the reaction, and the
reaction mixture was concentrated under reduced pressure. The resulting
residue was then dissolved in CH2Cl2 (2 mL)
and treated with Ac2O (50 μL, 0.50 mmol), NEt3 (105 μL, 0.75 mmol) and catalytic DMAP (∼2 mg).
After 0.5 h, the solvent was evaporated, and the dry residue was treated
with 80% AcOH(aq) at rt with vigorous stirring for another
0.5 h. The reaction mixture was then azeotroped with benzene to remove
excess reagents, and the resulting residue was purified using flash
column chromatography (EA/Hex = 3/1 to 4/1) to afford compound 14 (52 mg, 95%) as a white foam. Method B: To a MeOH (3 mL)
solution of compound 3 (65 mg, 0.088 mmol) was added
Pd(OH)2/C (20% Pd, 60 mg). The reaction mixture was allowed
to stir under H2 (1 atm) at rt for 0.5 h. Next, the reaction
mixture was filtered through a MeOH-packed Celite pad to remove the
Pd catalyst, and the filtrate was concentrated under reduced pressure.
The resulting residue was then dissolved in dry ACN (2 mL) and treated
with trimethyl orthoacetate (35 μL, 0.26 mmol) and catalytic
CSA (∼4 mg, 0.02 mmol). After 1 h, NEt3 (∼0.1
mL) was added to quench the reaction, and the reaction mixture was
concentrated under reduced pressure. The resulting residue was treated
with 80% AcOH(aq) at rt with vigorous stirring for another
0.5 h. The reaction mixture was then azeotroped with benzene to remove
the excess reagents, and the resulting residue was purified using
flash column chromatography (EA/Hex = 3/1 to 4/1) to afford compound 14 (57 mg, 98%) as a white foam. The product consisted of
inseparable α/β anomeric acetates, and the major β-anomer
is reported: TLC (EA/Hex = 2/1) R 0.28; 1H NMR (600 MHz, CDCl3) δ
5.68 (d, J = 8.4 Hz, 1H, H-1), 5.30 (d, J = 2.9 Hz, 1H, H-4′), 5.24 (appt. t, J =
9.3 Hz, 1H, H-3), 5.06 (dd, J = 9.3, 8.4 Hz, 1H,
H-2), 4.85 (dd, J = 10.0, 7.9 Hz, 1H, H-2′),
4.47 (dd, J = 12.1, 1.8 Hz, 1H, H-6a), 4.42 (d, J = 7.9 Hz, 1H, H-1′), 4.20 (dd, J = 12.1, 4.9 Hz, 1H, H-6b), 4.15–4.04 (m, 2H, H-6′ab),
3.86–3.73 (m, 4H, H-4, H-5, H-3′, H-5′), 2.46
(d, J = 6.3 Hz, 1H, OH), 2.18 (s, 3H), 2.13 (s, 3H),
2.12 (s, 3H), 2.10 (s, 3H), 2.09 (s, 3H), 2.05 (s, 3H), 2.03 (s, 3H); 13C NMR (150 MHz, CDCl3) δ 171.2, 170.7, 170.5,
170.4, 169.7, 169.5, 100.6, 91.6, 75.7, 73.6, 73.1, 72.5, 71.8, 71.0,
70.4, 69.1, 61.9, 61.4, 20.84, 20.81, 20.75, 20.73, 20.67, 20.5; HRMS
(ESI-ion trap) m/z calcd for [C26H36O18 + NH4+] 654.2240, found 654.2289.
General Procedure for Constructing
Globotriaose and Isoglobotriaose
(Table 1)
To a 0 °C CH2Cl2 solution of compound 15 (conc. 0.1–0.2
M, 2.5 equiv) was added TMSI (2.8 equiv) under argon atmosphere. After
stirring for 30 min, the reaction mixture was azeotroped with anhydrous
benzene (5 mL × 3) and dried under a high vacuum system for 1
h to afford the corresponding glycosyl iodide as a light yellow foam.
The iodide was then dissolved in CH2Cl2 (1 mL)
under argon and cooled to −78 °C. In a separate flask,
AgOTf (3.2 equiv), acceptor (compound 12, 13, or 14; ∼0.10 mmol scale; 1.0 equiv) and activated
4 Å molecular sieves (∼200 mg) were allowed to stir in
a CH2Cl2 (2 mL) solution under argon atmosphere
at rt for 30 min. The acceptor solution was then cooled to −78
°C, and the cooled glycosyl iodide solution was transferred dropwisely
to the acceptor flask via cannula. After 4 h, the reaction mixture
was gradually warmed from −78 to −30 °C, diluted
with EA (10 mL), and quenched by NEt3 (∼1 mL). The
suspension was filtered through a well-packed Celite pad and washed
with EA (∼30 mL). The filtrate was washed with saturated NaHCO3(aq) (20 mL × 1) and brine (20 mL × 2), dried over
anhydrous Na2SO4, and concentrated under reduced
pressure. The resulting residue was purified by gradient flash column
chromatography (EA/Hex) to obtain corresponding oligosaccharides (compound 16, 17, 18, or 19)
as white foams.
To a MeOH solution of compound 16 (63 mg, 0.054 mmol) was added Pd(OH)2/C (20% Pd, 63 mg)
and stirred under H2 gas (1 atm) for 2 h. After the disappearance
of starting material on TLC, the Pd(OH)2/C was removed
by passing through a MeOH-packed Celite pad. The filtrate was concentrated
and redissolved in anhydrous MeOH (3 mL) followed by addition of NaOMe/MeOH
solution (25 wt %, 40 μL). Some white suspension formed after
15 min, and H2O (1.0 mL) was added to dissolve the suspension.
The reaction mixture was stirred for another 15 min and was acidified
to pH = 6 by adding Dowex H+ resin. The solution was filtered
through a plug of C18 reverse-phrase silica gel and washed
with MeOH and water to remove the resin and desalt. The filtrate was
concentrated and lyophilized to afford compound 21 (27
mg, 99%) as a white fluffy foam: TLC (EA/2-propanol/H2O
= 2:2:1) R 0.22; 1H NMR (800 MHz, D2O) δ 5.22 (d, J = 3.7 Hz, 0.4 H), 4.94 (d, J = 3.7 Hz, 1H), 4.66
(d, J = 8.0 Hz, 0.4 H), 4.50 (dd, J = 7.8, 1.7 Hz, 1H), 4.35 (s, 1H), 4.06–4.01 (m, 2H), 3.98–3.54
(m, 14H), 3.27 (appt. t, J = 8.6 Hz, 0.4 H); 13C NMR (200 MHz, D2O) δ 103.90, 103.86, 101.0,
96.4, 92.4, 79.3, 79.2, 78.0, 75.5, 74.5, 72.8, 71.5, 70.8, 69.8,
69.6, 69.22, 69.20, 61.1, 61.0, 60.7, 60.6 (Assignment matches with
literature);[64] HRMS (ESI0-ion trap) m/z calcd for [C18H32O16 + NH4+] 522.2029, found 522.2012.