Efficient chemoenzymatic enantioselective synthesis of diacylglycerols (DAG)

Article abstract of DOI:10.1016/j.tetasy.2004.07.040

A new efficient chemoenzymatic methodology for the production of 3-O-benzyl-sn-glycerol and 1,2-O-dipalmitoyl-sn-glycerol has been developed. It starts from racemic 1-O-benzylglycerol and is based on the sequential enzymatic acylation-Mitsunobu inversion-enzymatic hydrolysis, which has been performed without isolation of the intermediates. In this way a 70-75% yield of 3-O-benzyl-sn-glycerol with 94-96% ee has been obtained.

Full text of DOI:10.1016/j.tetasy.2004.07.040

Tetrahedron:
Asymmetry
Tetrahedron: Asymmetry 15 (2004) 2889–2892
Efficient chemoenzymatic enantioselective synthesis
of diacylglycerols (DAG)
Giuseppe Guanti,* Luca Banfi, Andrea Basso, Elisabetta Bevilacqua, Laura Bondanza
and Renata Riva
Dipartimento di Chimica e Chimica Industriale, via Dodecaneso 31, I-16146 Genova, Italy
Received 1 June 2004;accepted 8 July 2004
Available online 11 September 2004
Abstract—A new efficient chemoenzymatic methodology for the production of 3-O-benzyl-sn-glycerol and 1,2-O-dipalmitoyl-sn-
glycerol has been developed. It starts from racemic 1-O-benzylglycerol and is based on the sequential enzymatic acylation––Mitsu-
nobu inversion––enzymatic hydrolysis, which has been performed without isolation of the intermediates. In this way a 70–75% yield
of 3-O-benzyl-sn-glycerol with 94–96% ee has been obtained.
Ó 2004 Published by Elsevier Ltd.
1. Introduction
friendly chemoenzymatic route that avoids the use of
reagents such as NaIO4 or NaBH4. Various research
groups have described chemoenzymatic approaches to
2 or 3. Kinetic resolution has been accomplished by acyl-
ation of racemic 3 with various enzymes.2–6 The enantio-
selectivity is not very high. In order to increase the
ee, crystallisation methodologies5 or double kinetic res-
olutions4 have been devised. Lowering the temperature
helps in increasing the ee, but larger amounts of enzyme
are needed.6 Also other cyclic derivatives of glycerol
such as carbonates7 have been resolved in this way.
However, the main drawback of this approach is that
the yields are lower than 40% and that the unwanted
enantiomer has to be discarded.
In the course of a project directed to the development of
an efficient synthesis of phospholipids amenable of
scale-up, we needed to elaborate a cost-effective enantio-
selective preparation of 1,2-O-sn-diacylglycerols (DAG)
1 (RCO2H = fatty acid) (Scheme 1).
OCOR
OH
O
O
ROCO
OH
HO
OBn
OH
(R)-2
(S)-3
(S)-1
This drawback is not present in other chemoenzymatic
strategies, such as the asymmetrisation of 2-protected
glycerols 4 to give chiral derivatives 55 or the microbio-
logical reduction of an asymmetrically diprotected
dihydroxyacetone to give 6.8 In the first case, however,
conversion of the chiral adduct 5 into 2 requires several
synthetic steps, whereas in the second case the prepara-
tion of the starting ketone is rather long.
OH
O(P)
O(P)
AcO
OBn
HO
OH
HO
OCOR
(S)-6
(R)-5
4
Scheme 1.
DAGs have been previously synthesised mainly starting
from 3-O-benzyl-sn-glycerol 2, that is in turn obtained in
two steps from (S)-solketal 3.1 The typical synthesis of
3, which is also commercially available, starts from
D-mannitol. However, we wanted to use a more eco-
Va¨nttinen and Kanerva have demonstrated the possibil-
ity to overcome the main disadvantage of hydrolase cat-
alysed kinetic resolutions, that is the loss of at least 50%
of starting material, by coupling the enzymatic reaction
with a Mitsunobu esterification.3 We decided to use this
method (that was never employed before on glycerol
derivatives) for the enantioconvergent synthesis of
*
Corresponding author. Tel./fax: +39 010 353 6105;e-mail: guanti@
chimica.unige.it
0957-4166/$ - see front matter Ó 2004 Published by Elsevier Ltd.
doi:10.1016/j.tetasy.2004.07.040

2890
G. Guanti et al. / Tetrahedron: Asymmetry 15 (2004) 2889–2892
(R)-2 and, in this paper, we report the achievement of
this goal.
Although it is not possible to calculate precisely the
enantioselectivity factor E12 for the second acylation
reaction, we can compute a ÔvirtualÕ E, from these two
ee values. In the hypothesis that the first acylation is
fully non-stereoselective, and that it is complete before
the second one begins, the virtual value calculated in this
way and the ÔrealÕ one will be coincident. These assump-
tions are reasonable, since we have seen that the first
acylation affords regioselectively a nearly racemic mono-
acetate. Moreover only small quantities (1–2%) of the
monoacetate 9 (Scheme 3), acylated at the secondary
alcohol, may be detected by TLC and NMR during
the course of the reaction. These ÔvirtualÕ E value may
be useful because they give an idea of the overall selec-
tivity of the two acylation steps. They can also be used
for predicting the ees of 7 and 8 in function of the con-
version. Actually, using the appropriate equations,12
either from the two ee values or from one of them and
the measured 48% conversion of the second acylation,
we determined in all cases an E of about 33. This result
was in line with what reported by Valverde and co-
workers.10
2. Results and discussion
Racemic solketal 3 was benzylated by a modification of
the procedure of Hirt and Barner.9 This method, that em-
ploys phase-transfer catalysis and benzyl chloride as the
electrophile, is superior from a practical point of view
than the most widely used reaction with NaH and benzyl
bromide. We preferred n-Bu4NHSO4 as the catalyst,
again for economical reasons. The intermediate O-benzyl
solketal may be distilled as well, but its isolation is not
necessary, and the following step can be performed di-
rectly on the crude product. Deblocking of the cyclic acet-
al was conveniently performed by the use of a resin bound
acid catalyst (Amberlyst 15). In this way the work-up is
very easy. Racemic 2 was purified by distillation.
Resolution of racemic 2 through PFL (Fluka) or Amano
P catalysed acetylation was already reported by Valv-
erde and co-workers.10 They demonstrated that acyl-
ation of the primary hydroxy group is much faster than
that of the secondary one. While the first acylation pro-
ceeded with very low enantioselection, the second one
afforded good levels of selectivity.
Amano P
pH 7 buffer / iPr2O
OCOR
OCOR
ROCO
OBn
HO
OBn
(S)-8b,c
(R)-9b,c
Due to the high cost of PFL, we repeated these experi-
ments using Amano P lipase11 (from Pseudomonas cepa-
cia) (Scheme 2). The reactions were carried out starting
with racemic 2 in THF–hexane 1:1, employing vinyl ace-
tate as acyl donor. We did not isolate the intermediate
monoacetates, but let the reaction proceed until the
monoacetate 7 and the diacetate 8 were nearly equimo-
lar. Upon chromatography we obtained (S)-8a with 86%
ee, and (R)-7a with 80% ee. The 7:8 ratio was 48:52 (by
NMR of the crude product).
1) C15H31CO2H,
DCC, DMAP
2) H2, Pd-C
OCOC15H31
OH
OH
C
15H31OCO
HO
OBn
92%
(S)-1
(R)-2
Scheme 3.
Isolated (R)-7a (ee 80%) was then submitted to a Mitsu-
nobu reaction with acetic acid in Et2O. The reaction
turned out to be quite slow. By stopping it after one
day, only 65% conversion was measured. The ee of the
product (S)-8a was 75%, indicating a nearly complete
inversion. However, when the reaction was brought to
completion, by carrying it out for prolonged time and
through further additions of reagents, the ee of product
(S)-8a was only 44%, indicating that the reaction was
not stereoselective, probably because of intramolecular
transacylation processes.
1) BnCl, KOH,
OH
nBu4NHSO4, CH2Cl2-H2O
2) Amberlyst 15, MeOH
O
HO
OBn
O
OH
80%
(rac)-2
(rac)-3
Amano P
RCOOCH=CH2
iPr2O
OCOR
ROCO
ROCO
OBn
These unsatisfactory results prompted us to try different
acylating agents, namely vinyl propionate and vinyl buty-
rate. The enzymatic acylations were carried out in diiso-
propyl ether, which, after a preliminary screening, was
found to be the best solvent. From these reactions we
could determine enantioselectivity factors of 23 and 35,
respectively, for the propanoylation and the butyrylation.
OH
(S)-8a-c
Amano P
ROCO
OBn
+
7a-c
OH
OBn
(R)-7a-c
PPh3, DEAD
RCO2H
Et2O
The isolated monoesters (R)-7b–c were then submitted
to the Mitsunobu esterification. In these two cases, the
reactions in Et2O were found to be completely stereospe-
cific (inversion), also after long reaction times. Addition
of DMF as cosolvent had an accelerating effect, but a
slight lowering of the ees was observed. The reagents
a: R = CH3
b: R = Et
c: R = nPr
OCOR
ROCO
OBn
(S)-8a-c
Scheme 2.

G. Guanti et al. / Tetrahedron: Asymmetry 15 (2004) 2889–2892
2891
are in part consumed by a concurrent reaction that
forms diethyl N,N-diacyl hydrazinodicarboxylate;the
best procedure involves two or three further additions
of the reagents (see Section 3) in order to drive the reac-
tion to completion.
EtO2CNHNHCO2Et by filtration after the Mitsunobu
reaction. In the case of propanoylation, we obtained,
with an excellent 75% overall yield, (R)-2 with 94% ee.
In the case of butyrylation, the overall yield was slightly
lower (70%) and the ee was 96%.
The Mitsunobu reaction was then repeated on the crude
products coming from the enzymatic acylations. From
the values of E, it is possible to calculate the ideal con-
version of the second enzymatic acylation in order to get
the best ee after the Mitsunobu step, in the hypothesis
that the latter is quantitative and proceeds with com-
plete stereospecificity.3 In our case the optimum conver-
sion should be 54% (with overall expected ees after
Mitsunobu = 83% for 8b and 87% for 8c). However,
the reactions tend to become very sluggish when
approaching 50% conversion. This is clearly an advan-
tage since a careful monitoring of the reaction is not nec-
essary, but it is difficult to reach 54% conversion. The
two reactions were therefore stopped at about 47%
and the crude products submitted to the Mitsunobu
conditions, giving, after chromatography, 8b (77% yield,
82% ee) and 8c (75% yield, 83% ee). The ees are even
slightly higher than calculated, probably because the
substitution reaction was not quantitative.
While both procedures are valid, probably the one
involving the dipropionate is better suited for a bulk
synthesis, since vinyl propionate is cheaper than vinyl
butyrate and because of the unpleasant odour of butyric
acid. Although the enantioselectivity is slightly lower, a
similar level of ee may be easily reached by stopping the
final hydrolysis at a lower degree of conversion.
Finally we converted this intermediate into 1,2-dipalmi-
toyl-sn-glycerol (S)-1, following essentially the method
previously described by Martin.14 The overall yield of
this DAG from racemic solketal was 55%. The synthetic
sequence involved reactions taking place at 0–50°C,
without the need of inert atmosphere, and with low cost
and easily handled reagents. The intermediates either
did not require purification or were conveniently puri-
fied by distillation or crystallisation. Therefore we think
that this enantioselective synthesis of DAG may be con-
veniently up-scaled for an industrial production. This
will make biologically important phospholipids more
easily accessible.
Although this two step methodology was demonstrated
to be quite efficient, the ees of 8b and 8c were not yet
high enough. Since for the obtainment of (R)-2 we had
to remove in some way the two acyl groups, we reasoned
that, by performing this hydrolysis enzymatically,13 we
could further improve the ee. This Ôdouble kinetic reso-
lutionÕ strategy had been used often in the past for
increasing the optical purity.4 We first explored the
selectivity of the hydrolysis by using racemic 8b and
8c. Also in this case we did not isolate the intermediate
monoesters 9, but allowed the reaction to proceed until
the ratio of 9 to 2 was close to 50%. In this case the reac-
tion did not slow down very much at this point. The Ôvir-
tualÕ enantioselectivity factors were determined as for
the acylation reactions and turned out to be in this case
considerably lower (2.8 starting from 8b and 5.7 starting
from 8c). The hydrolysis reactions were also less regiose-
lective and we detected significant amounts of the regio-
isomeric monoacetates 7. Notwithstanding these low E
values, we calculated that, by carrying out the hydrolysis
of enriched 8b and 8c, a high ee could be obtained by
stopping the reaction at a 9:2 ratio of about 1:9.
3. Experimental
3.1. (rac)-1-O-Benzylglycerol 2
A solution of rac 2,2-dimethyl-1,3-dioxolane-4-methanol
(10mL, 80.4mmol) in CH2Cl2 (50mL) was treated
sequentially with 30% (w/v) aqueous NaOH (37.8mL,
283.5mmol), n-Bu4NHSO4 (2.730g, 8.04mmol) and
benzyl chloride (10.18mL, 88.5mmol). The mixture
was vigorously stirred at 50°C for 25h. After cooling,
it was diluted with saturated aqueous NaCl and
extracted with Et2O. After washing the organic extracts
with saturated NaCl, and drying (Na2SO4), evaporation
gave a crude oil. It was taken up in MeOH (50mL) and
treated with 4.0g of Amberlyst 15. After stirring for 5h
at rt, the suspension was filtered and the resin was washed
with AcOEt/MeOH 1:1. CaCO3 (1.0g) was added to the
filtrate and, after stirring for 10min, the mixture was
filtered again. Evaporation and distillation (128–130°C
at 0.15–0.2mbar) gave pure (rac)-2 (11.74g, 80%).
This was indeed the case. When we submitted pure 8b
(82% ee) and 8c (83% ee) to the enzymatic hydrolysis
and stopped the reactions when the ratio 9:2 was equal
to 11:89, we recovered (R)-2 with 94% and 96% ee,
respectively. The remaining 9b and 9c were also isolated
and were demonstrated to have the opposite absolute
configuration and ees = 36% and 40%, respectively.
Thus double kinetic resolution is quite helpful in
increasing the ees.
3.2. 3-O-Benzyl-sn-glycerol (R)-2
A solution of (rac)-2 (10.38g, 57.0mmol) in diisopropyl
ether (155mL) was treated with vinyl propionate
(31mL, 285mmol) and with Amano P lipase (5.43g).
After stirring at 20°C for 66h, the suspension was fil-
tered and evaporated to dryness. The ratio monoprop-
ionate 7b:dipropionate 8b was 53:47 (1H NMR). The
crude oil was taken up in dry Et2O (150mL) and treated
with triphenylphosphine (5.07g, 19.3mmol), diethyl
azodicarboxylate (DEAD) (3.0mL, 19.05mmol) and
propionic acid (1.43mL, 19.2mmol). After 17h a further
aliquot of PPh3 (2.157g, 8.2mmol), DEAD (1.30mL,
Having all the conditions set up, we performed the
whole sequence without purification of the intermedi-
ates. As described in the experimental part, we only sep-
arated the diesters 8b and 8c from most of Ph3PO and

2892
G. Guanti et al. / Tetrahedron: Asymmetry 15 (2004) 2889–2892
8.26mmol) and propionic acid (0.61mL, 8.18mmol) was
added. This operation was repeated after 22h and 37h.
After further stirring for 18h (the complete disappear-
ance of monopropionate was detected by TLC), the mix-
ture was filtered, washed with saturated aqueous
NaHCO3 and evaporated to dryness. It was taken up
in petroleum ether/Et2O 2:1 (50mL), filtered through a
sintered funnel and evaporated to dryness. The crude
product was taken up in i-Pr2O (80mL), and treated
with 1M pH7 phosphate buffer (KH2PO4–K2HPO4)
(240mL). Amano P lipase (8.2g) was added. The mix-
ture was stirred at 20°C for 34h. The mixture was satu-
rated with NaCl and filtered through a Celite cake.
Extraction with AcOEt gave, after evaporation, a crude
ated at room temperature on 10% Pd–C (500mg) at
atmospheric pressure. As soon as the reaction was com-
plete, the catalyst was removed by filtration on Celite
pad. Evaporation afforded a solid that was recrystallised
from n-hexane to give 19.98g of 1 (93%). Mp: 62.8–
64.6°C. The spectroscopic and polarimetric data were
in accordance to those reported in the literature.14,15
Acknowledgements
This work was financially supported by K2 Euticals. We
would like to thank Dr. Katharine Powles for her
exploratory work, Dr. Valeria Rocca and the late Dr.
Delia Bertoia for HPLC analyses.
1
product. H NMR indicated a ratio monopropionate
9b:diol 2 = 11:89. The crude product was chromato-
graphed on silica gel (CH2Cl2 to CH2Cl2/MeOH 97:3)
to give pure (R)-2 (7.78g, 75%). Alternatively it can also
be purified by distillation. This product had an ee of
94%. The spectroscopic and polarimetric data were in
accord with those reported in the literature. The abso-
lute configuration was also confirmed by transformation
into the diacetyl derivative (S)-8a (Ac2O, pyridine),
which had [a]D = +16.3 (c 1.64, CHCl3). Lit.10 = +14.0
(c 0.5, CHCl3).
References
1. Jurczak, J.;Pikul, S.;Bauer, T.
447–488.
Tetrahedron 1986, 42,
2. Partali, V.;Melbye, A. G.;Alvik, T.;Anthonsen, T.
Tetrahedron: Asymmetry 1992, 32, 65–72;Jungen, M.;
Gais, H.-J. Tetrahedron: Asymmetry 1999, 10, 3747–3758;
Va¨nttinen, E.;Kanerva, L. T. J. Chem. Soc., Perkin Trans.
1 1994, 3459–3463;Va¨nttinen, E.;Kanerva, L. T. Tetra-
hedron: Asymmetry 1996, 7, 3037–3046;Wang Y.-F.;
Lalonde, J. J.;Momongan, M.;Bergbreiter, D. E.;Wong,
C.-H. J. Am. Chem. Soc. 1988, 110, 7200–7205;Ladner,
W. E.;Whitesides, G. M. J. Am. Chem. Soc. 1984, 106,
7250–7251;Parmar, V.;Prasad, A. K.;Singh, P. K.;
Gupta, S. Tetrahedron: Asymmetry 1992, 3, 1395–1398.
3. Va¨nttinen, E.;Kanerva, L. T. Tetrahedron: Asymmetry
1995, 6, 1779–1786.
4. Va¨nttinen, E.;Kanerva, L. T. Tetrahedron: Asymmetry
1997, 8, 923–933.
5. Bianchi, D.;Bosetti, A.;Golini, P.;Cesti, P.;Pina, C.
Tetrahedron: Asymmetry 1997, 8, 817–819.
6. Sakai, T.;Kishimoto, T.;Tanaka, Y.;Ema, T.;Utaka, M.
Tetrahedron Lett. 1998, 39, 7881–7884.
7. Pallavicini, M.;Valoti, E.;Villa, L.;Piccolo, O.
By a similar procedure, (R)-2 was also obtained through
the dibutyrate in 71% overall yield and with an ee of 96%.
3.3. Determination of enantiomeric excesses of 7a–c,
8a–c, 2
Compounds 7a–c and 8a–c were all hydrolysed with
0.35M KOH in MeOH–H2O 3:1 to the diol 2. This
was in turn converted into the double Mosher ester,
by reaction with (R) a-methoxy-a-phenyl-a-(trifluoro-
methyl)acetyl chloride and 4-dimethylaminopyridine in
CH2Cl2. The enantiomeric excess was then determined
by 1H NMR or by HPLC (Daicel Chiralpak AD, n-hex-
ane–i-PrOH 9:1, flow: 0.5mL/min, detection: UV-DAD
(220, 230, 250nm)).
J. Org.
Chem. 1994, 59, 1751–1754;Hamaguchi, S.;Yamamura,
H.;Hasegawa, J.;Watanabe, K. Agric. Biol. Chem. 1985,
49, 1509–1511.
3.4. 3-O-Benzyl-1,2-O-dipalmitoyl-sn-glycerol (S)-1
8. Aragozzini, F.;Maconi, E.;Potenza, D.;Scolastico, C.
Synthesis 1989, 225–227.
A solution of (R)-2 (7.00g, 38.4mmol) in CH2Cl2
(150mL) was treated with palmitic acid (22.64g,
88.3mmol), and 4-dimethylaminopyridine (938mg,
7.68mmol), cooled in an ice bath, and treated with a solu-
tion of dicyclohexylcarbodiimide (18.22g, 88.3mmol) in
CH2Cl2 (100mL). The resulting suspension was stirred at
0°C for 1h and at rt overnight. After dilution with
100mL of petroleum ether, the dicyclohexylurea was
removed by filtration. The filtrate was washed with 5%
NaHCO3, with 1M HCl, and with saturated NaCl to
give, after evaporation, a solid that was triturated from
MeOH (24.81g, 96%). Mp: 64–65°C.
9. Hirt, G.;Barner, R.
1059–1084.
10. Herradon, B.;Cueto, S.;Morcuende, A.;Valverde, S.
Tetrahedron: Asymmetry 1993, 4, 845–864.
11. We wish to thank Amano for a kind gift of this enzyme.
12. Bornscheuer, U. T.;Kazlauskas, R. J. Hydrolases in
Organic Synthesis;Wiley-VCH: Weinheim, 1999.
Helv. Chim. Acta 1982, 65,
´
´
´
13. Poppe, L.;Nova k, L.;Kajta r-Peredy, M.;Sza ntay, C.
Tetrahedron: Asymmetry 1993, 4, 2211–2217.
14. Martin, S. F.;Josey, J. A.;Wong, Y.-L.;Dean, D. W. J.
Org. Chem. 1994, 59, 4805–4820.
15. Muller, H. K.;Burgold, J. Tetrahedron Lett. 1986, 42,
¨
1533–1538;Reddy, K. K.;Saady, M.;Falck, J. R.;
Withed, G. J. Org. Chem. 1995, 60, 3385–3390;Bruzik,
K. S.;Salamonczyk, G.;Stec, W. J. J. Org. Chem. 1986,
51, 2368–2370.
This solid was taken up in 300mL of 96% EtOH and
30mL of glacial acetic acid. The solution was hydrogen-