A study of the structure and chiral selectivity of micelles of two isomeric D-glucopyranoside-based surfactants

Article abstract of DOI:10.1039/a708418h

We have synthesised the sodium salts of two isomeric surfactants that differ in the anomeric orientation of the C₁₄ hydrocarbon chain to the D-glucopyranoside fused ring structure, containing a phosphate polar head group. The inclusion of α- an

Full text of DOI:10.1039/a708418h

View Article Online / Journal Homepage / Table of Contents for this issue
A study of the structure and chiral selectivity of micelles of two
isomeric D-glucopyranoside-based surfactants
David Tickle,a Ashley George,b Kevin Jennings,b Patrick Camilleri*,b and
Anthony J. Kirbya
a University Chemical Laboratory, Lensfield Road, Cambridge, UK CB2 1EW
b SmithKline Beecham Pharmaceuticals, New Frontiers Science Park, Coldharbour Road,
The Pinnacles, Harlow, Essex, UK CM19 5AD
We have synthesised the sodium salts of two isomeric surfactants that differ in the anomeric orientation of
the C14 hydrocarbon chain to the D-glucopyranoside fused ring structure, containing a phosphate polar
head group. The inclusion of á- and â-anomers in the separation buffer, using micellar electrokinetic
capillary chromatography (MECC) shows that the latter is the superior chiral selector. The critical
micelle concentration (cmc) for both surfactants was around 0.6 mM, and the mean aggregation number
is 65 for both anomers. In an attempt to understand the possible mechanism of chiral discrimination, we
have calculated minimum energy conformations of micelles produced by these surfactants, and have
established the importance of their chiral head groups and the sodium counter-ions in the discriminating
process. We have also investigated the micro-structures of the two surfactants by transmission electron
microscopy. This technique shows that the two enantiomers aggregate very differently; whereas the
á-anomer shows randomly orientated string-like forms the â-anomer forms ordered ‘whorl’ arrays.
Major advances in the synthesis and commercial production of
OH
single enantiomers as biologically active molecules require the
P
O
O
O
development of selective and sensitive analytical methods, suit-
able for the analysis of one enantiomer in the presence of a vast
excess of the other. Until recently traditional ‘wet’ separation
techniques, in particular thin layer chromatography (TLC) and
high performance liquid chromatography (HPLC) have been
the methods of choice for the analysis of enantiomeric species.
The high separation efficiencies routinely achieved in capillary
electrophoresis (CE) have enhanced its popularity for the reso-
lution of mixtures of enantiomers related to molecules that
vary widely in their structure, lipophilicity and ionisation. This
technique is becoming increasingly effective in monitoring
enantiomeric purity of both intermediates and final drug
product.
O
O
HO
OH
1, β-anomer
OH
P
O
O
O
O
HO
HO
O
2, α-anomer
in chiral selectivity by these surfactants we have measured their
respective aggregation and critical micelle concentration (cmc)
values. We also report studies on the microstructures of the two
surfactants by molecular modelling and transmission electron
microscopy.
In this application of CE, chiral additives are added to the
separation buffer, and the mode of operation is usually free
zone electrophoresis or micellar electrokinetic capillary chrom-
atography (MECC). Chiral selectors have included naturally
occurring additives such as cyclodextrins,1–3 bile salts,4,5 pro-
teins6 and peptide antibiotics.7 We have also reported the design
and synthesis of anionic surfactants, which when used above
their critical micelle concentration (cmc) in MECC allow the
resolution of a number of structurally unrelated racemic
mixtures.8–10 The formation of non-covalent ‘diastereoisomers’
between the micelles and the enantiomers leads to chiral
discrimination if the energies of formation of these transient
species differ sufficiently.
Of the various structures of chiral surfactants we have so far
synthesised for chiral CE discrimination, glucopyranoside-
based surfactants8 have been the most successful. To obtain
information about the mechanism of chiral resolution by this
class of anionic detergents we have now prepared the two iso-
meric molecules, 1 and 2 that differ in the anomeric orientation
of the hydrocarbon chain with respect to the glucopyranoside
ring.
Experimental
Materials
Starting materials and reagents for the synthesis of surfactants
1 and 2 and intermediates were purchased from Aldrich
(Gillingham, Dorset, UK). Dansyl derivatives and buffer con-
stituents for capillary electrophoresis and phosphotungstic
acid for electron transmission microscopy were purchased from
Agar Scientific (Stanstead, UK). Tris(2,2Ј-bipyridyl)ruthenium
chloride, 9-methylanthracene and N-phenyl-1-naphthylamine
were obtained from ICN Biomedicals Ltd. (High Wycombe,
Bucks, UK). Solutions were prepared in doubly-distilled water
throughout.
Synthesis of the sodium salt of n-tetradecyl â-D-glucopyranoside
4,6-(hydrogen phosphate) 1
Tetradecyl 2,3,4,6-tetra-O-acetyl-â-D-glucopyranoside 3.
Acetobromoglucose (9.4 g, 0.23 mmol) was added to a stirred
solution containing n-tetradecanol (6.9 g, 0.32 mmol), yellow
Differences in the chiral discrimination properties of micelles
formed from the sodium salts of the α- and β-anomers have
been observed in the MECC analysis of a number of amino
acid dansyl derivatives. In an attempt to account for variations
J. Chem. Soc., Perkin Trans. 2, 1998
467

View Article Online
mercuric oxide (7.1 g, 32 mmol), mercuric bromide (0.2 g, 0.7
mmol) and Drierite (10 g) in 100 cm3 of dry dichloromethane
and the solution stirred at room temp. for 20 h under argon.
The mixture was then filtered, the solids washed with dichloro-
methane, and the filtrate combined with the washings, then
washed with 1  KBr (4 × 50 cm3) until no mercuric salts were
present in the washings. The organic layer was then washed
with water, dried (Na2SO4) and concentrated in vacuo. The resi-
due was purified by repeated column chromatography [SiO2,
light petroleum (bp 30–40 ЊC)–EtOAc; 2:1] to yield the
tetraacetyl glucopyranoside 3 (4.2 g, 34%) as needles, mp 57–
59 ЊC; Rf [SiO2, light petroleum (bp 30–40 ЊC)–EtOAc; 2:1],
MHz, (CD3)2SO] 102.9ϩ (OCHO), 76.9ϩ, 73.5ϩ, 70.2ϩ,
68.6Ϫ (OCH2R), 61.2Ϫ (CH2OH), 31.4Ϫ, 29.4Ϫ, 29.2Ϫ,
29.1Ϫ, 28.9Ϫ, 25.7Ϫ, 22.2Ϫ and 14.04ϩ; m/z (EI) 376.3 (0.5%,
Mϩ), 345.3 (20%, Mϩ Ϫ CH2OH), 197.2 (60%, OC12H25), 73.0
(100%, C4H8O) (Found: Mϩ, 376.2824, C20H40O6 requires
376.2824).
Tetradecyl â-D-glucopyranoside 4,6-(phenyl phosphate) 5. Dry
triethylamine (0.8 cm3, 9.1 mmol) was added to a stirred solu-
tion of glycopyranoside 4 (2.49 g, 6.6 mmol) in 500 cm3 of dry
dichloromethane containing 5 g of 4 Å molecular sieves and the
solution was stirred under argon for 1 h. The solution was then
cooled to 0 ЊC, phenyldichlorophosphate (1.4 g, 6.6 mmol)
added and the reaction stirred for another 3 h at room temp.
The solution was then filtered and the filtrate concentrated
in vacuo. The residue was purified by flash column chrom-
atography (SiO2, EtOAc–hexane, 1:1) to yield the two dia-
stereoisomers (4:1) of the phosphate triester 5 (1.76 g, 51.7%)
as a clear oil; Rf (SiO2, EtOAc–hexane, 1:1), 0.1; νmax/cmϪ1
0.58; νmax/cmϪ1 (CH Cl ) 1757 (C᎐O) and 1262 (CO); δ (400
᎐
2
2
H
MHz; CDCl3; J values in Hz throughout) 5.17 (1 H, t, J 9.5,
CHOAc), 5.05 (1 H, t, J 9.7, CHOAc), 4.95 (1 H, dd, J 9.5 and
8.1, CHOAc), 4.45 (1 H, d, J 8.0, OCHO), 4.23 (1 H, dd, J 12.3
and 4.3, CHAHBOAc), 4.09 (1 H, dd, J 12.3 and 2.1, CHA-
HBOAc), 3.83 (1 H, dt, J 9.5 and 6.3, OCHAHBR), 3.66 (1 H,
ddd, J 9.9, 4.4 and 2.3, CHO), 3.43 (1 H, dt, J 9.5 and 6.9,
OCHAHBR), 2.05 (3 H, s, CH3), 2.00 (3 H, s, CH3), 1.98 (3 H, s,
CH3), 1.97 (3 H, s, CH3), 1.55–1.48 (2 H, m, OCH2CH2C12H25),
1.27–1.20 (22 H, m, OC2H4C11H22CH3) and 0.84 (3 H, t, J 6.7,
CH3); δC(62.5 MHz, CDCl3)† 170.7Ϫ, 170.3Ϫ, 169.4Ϫ, 169.3Ϫ
(4 × CO), 108.8ϩ (OCHO), 72.9ϩ, 71.7ϩ, 71.3ϩ (3 ×
CHOAc), 70.2Ϫ (OCH2R), 68.5ϩ (CHO), 62.0ϩ (CH2OAc),
31.9Ϫ, 29.7Ϫ, 29.6Ϫ, 29.6Ϫ, 29.4Ϫ, 29.4Ϫ, 29.3Ϫ, 29.3Ϫ,
25.8Ϫ, 25.8Ϫ, 22.7Ϫ, 20.7ϩ, 20.6ϩ and 14.1ϩ; m/z (EI) 424.3
(10%, Mϩ Ϫ 2 × OAc), 351.2 (10%, Mϩ Ϫ C14H29), 331.1 (50%,
Mϩ Ϫ OC14H29), 55.0 (100%, C4H7) (Found: Mϩ Ϫ C4H6O4
424.2834, C24H40O6 requires 424.2825).
(CH Cl ) 3584 (OH), 1592, 1490 (C᎐C), 1316 (P᎐O) and 1205
᎐
᎐
2
2
(PO); δH(400 MHz, CDCl3) (major isomer) 7.40–7.10 (5 H,
m, Ph), 4.50–4.35 (1 H, m, CHAHBOP), 4.42 (1 H, d, J 7.8,
OCHO), 4.27 (1 H, t, J 10.4, CHOP), 4.18 (1 H, t, J 9.4,
CHOH), 3.90–3.65 (3 H, m, CHAHBOP, OCHAHBR and CHO),
3.53 (1 H, dt, J 9.2 and 7.0, OCHAHBR), 3.43 (1 H, t, J 8.5,
CHOH), 1.60–1.50 (2 H, m, OCH2CH2R), 1.30–1.20 (22 H, m,
OC2H4C11H22CH3), 0.86 (3 H, t, J 6.8, CH3); (no additional
signals due to minor isomers); δC(100 MHz, CDCl3) (major
isomer) 150.0Ϫ (Cipso), 130.1ϩ (Cmeta), 125.6ϩ (Cpara), 119.6ϩ (1
C, d, J 4.9, Cortho), 80.6ϩ (1 C, d, J 6.6, CHOP), 73.7ϩ
(CHOH), 73.3ϩ (1 C, d, J 8.5, CHOH), 70.8Ϫ (OCH2R),
69.0Ϫ (1 C, d, J 7.8, CH2OP), 66.0 (1 C, d, J 5.7, CHO), 31.9Ϫ,
29.7Ϫ, 29.6Ϫ, 29.6Ϫ, 29.4Ϫ, 29.4Ϫ, 25.9Ϫ, 22.7Ϫ and 14.2ϩ;
(additional signals due to minor isomer) 129.9ϩ (Cmeta), 125.8ϩ
(Cpara), 120.3ϩ (1 C, d, J 4.5, Cortho), 103.2ϩ (OCHO), 79.4ϩ
(CHOP), 73.9ϩ (CHOH), 73.5ϩ (CHOH), 70.7Ϫ (OCH2R),
68.6Ϫ (CH2OP) and 66.2ϩ (CHO); m/z (FAB) 515.5 (8%,
MHϩ), 175 (100%, PhOPO3H2) (Found: MHϩ, 1 515.278 50.
C26H44O8 requires 515.2774).
R2O
O
R2O
O
R1O
R1O
3 R1 = R2 = Ac
4 R1 = R2 = H
5 R1 = H, R2R2 =
OPh
O
P
R2O
R2O
R1O
Tetradecyl â-D-glucopyranoside 4,6-(hydrogen phosphate),
sodium salt 1. Aqueous sodium hydroxide (1 ; 5 cm3) was
added to a stirred solution of phosphate triester 5 (1.67 g, 3.2
mmol) in 45 cm3 THF and the solution stirred for 15 h. The
solution was concentrated in vacuo to approximately 10 cm3
and neutralized with 1  HCl. The resulting solution was then
lyophilized to yield a white powder which was placed in a
Soxhlet extractor and washed with dichloromethane for 48 h to
yield the phosphate diester 1 (1.40 g, 93.6%) as white flakes;
O
R1O
O
6 R1 = R2 = H
7 R1 = H, R2R2 =
OPh
O
P
νmax/cmϪ1 (Nujol) 3580 (OH), 1320 (P᎐O); δ (400 MHz, D O)
Tetradecyl â-D-glucopyranoside 4. A solution of sodium
methoxide in methanol (1 ; 20 cm3) was added to a stirred
solution of tetraacetyl glucopyranoside 3 (4.2 g, 11.2 mmol) in
50 cm3 of dry methanol. The resulting solution was stirred for
2 h under argon. It was then desalted by the addition of ∼30 g
of Dowex 50 W (Hϩ) resin and stirred overnight. The solution
was then filtered and the residual resin washed with methanol.
The washings were combined with the filtrate and the residue
was purified by flash column chromatography (SiO2, CH2Cl2–
MeOH, 9:1) to give the glycoside 4 (2.9 g, 99.8%) as needles,
᎐
H
2
4.44 (1 H, d, J 7.2, OCHO), 4.2–4.0 (1 H, m, CHOP), 3.93 (1 H,
t, J 9.1, CHOH), 3.9–3.2 (6 H, m, CH2OP, CHO, CHOH and
OCH2R), 1.7–1.6 (2 H, m, OCH2CH2R), 1.4–1.2 (22 H, m,
OC2H4C11H22CH3) and 0.83 (3 H, t, J 6.0, CH3); δC(100 MHz,
D2O) 106.1ϩ (OCHO), 80.4ϩ (CHOP), 76.2ϩ, 75.8ϩ
(2 × CHOH), 73.2Ϫ (OCH2R), 70.0ϩ (CHO), 69.2Ϫ (CH2OP),
34.7Ϫ, 32.8Ϫ, 32.7Ϫ, 32.3Ϫ, 32.2Ϫ, 28.6Ϫ, 25.4Ϫ and 16.5ϩ;
δP(250 MHz, D2O, H-decoupled) Ϫ142.9; δP(250 MHz, D2O,
H-coupled) Ϫ142.9 (1 P, d, J 57.8); m/z (Ϫve FAB) 437 [5%,
(M Ϫ Naϩ)Ϫ], 551 [100%, (M Ϫ Naϩ ϩ TFA)Ϫ].
mp 119–121 ЊC; Rf (SiO2, CH2Cl2–MeOH, 9:1), 0.2; νmax
/
cmϪ1 (CH2Cl2) 3410 (OH) and 1078 (CO); δH[400 MHz,
(CD3)2SO] 4.94 [1 H, d, J 5.0, CHOH (exc.)], 4.92 [1 H, d,
J 5.41, CHOH (exc.)], 4.90 [1 H, d, J 4.6, CHOH (exc.)], 4.47
[1 H, t, J 5.9, CH2OH (exc.)], 4.08 (1 H, d, J 7.8, OCHO), 3.74
(1 H, dt, J 9.4 and 6.9, OCHAHBR), 3.65 (1 H, dd, J 11.5 and
5.9, CHAHBOH), 3.5–3.35 (2 H, m, OCHAHBR and CHAHB-
OH), 3.2–3.0 (3 H, m, 2 × CHOH and CHO), 2.91 (1 H, td,
J 7.9 and 5.0, CHOH), 1.5–1.45 (2 H, m, OCH2CH2R), 1.3–1.2
(22 H, m, OC2H4C11H22CH3) and 0.85 (3 H, t, J 6.7); δC[100
Synthesis of the sodium salt of n-tetradecyl á-D glucopyranoside
4,6-(hydrogen phosphate) 2
Tetradecyl á-D-glucopyranoside 6. n-Tetradecanol (0.8 g, 3.8
mmol) was added to a stirred solution of the tetraacetyl glyco-
side 3 (2.08 g, 3.8 mmol), in 20 cm3 of dry dichloromethane
under argon. 1  tin() chloride in dichloromethane (3.8 cm3,
3.8 mmol) was added dropwise and the resulting solution was
stirred for 48 h. The solution was then diluted with dichloro-
methane (100 cm3), washed with 3  HCl (2 × 50 cm3), dried
(MgSO4) and concentrated in vacuo. Crude NMR of the resi-
due showed a product ratio of greater than 95:5 α:β. The crude
† Where the 13C NMR attached proton test was performed, ϩ signifies
a CH3 or CH and Ϫ signifies a quaternary carbon or CH2.
468
J. Chem. Soc., Perkin Trans. 2, 1998

View Article Online
product was deprotected using sodium methoxide by the same
method as the β anomer 1 and the residue purified by flash
column chromatography (SiO2,CH2Cl2–MeOH, 9:1) to give the
glucopyranoside 6 (1.07 g, 74.8%) as needles, mp 118–120 ЊC;
Rf (SiO2, CH2Cl2–MeOH, 9:1), 0.1; νmax/cmϪ1 (CH2Cl2) 3405
(OH) and 1080 (CO); δH[400 MHz, (CD3)2SO] 4.73 [1 H, d,
J 5.4, CHOH (exc.)], 4.61 [1 H, d, J 4.9, CHOH (exc.)], 4.49
[1 H, d, J 6.3, CHOH (exc.)], 4.46 (1 H, d, J 3.6, OCHO), 4.31
[1 H, t, J 6.0, CH2OH (exc.)], 3.5–3.2 (6 H, m, OCH2R,
CH2OH, CHOH and CHO), 3.03 (1 H, ddd, J 9.7, 6.2 and 3.6,
CHOH), 2.91 (1 H, td, J 9.2 and 5.5, CHOH), 1.4–1.3 (2 H, m,
OCH2CH2R), 1.2–1.1 (22 H, m, OC2H4C11H22CH3) and 0.72
(3 H, t, J 6.8, CH3); δC[100 MHz, (CD3)2SO] 98.6ϩ (OCHO),
73.4ϩ, 72.8ϩ, 72.1ϩ (3 × CHOH), 70.4ϩ (CHO), 66.9Ϫ
(OCH2R), 61.0Ϫ (CH2OH), 31.4Ϫ, 29.4Ϫ, 29.2Ϫ, 29.1Ϫ,
29.0Ϫ, 28.8Ϫ, 25.8Ϫ, 22.2Ϫ and 14.1ϩ; m/z 377.3 (40%, MHϩ)
(Found: MHϩ, 377.292 00. C20H41O6 requires 377.2902).
a wavelength of 400 nm, on excitation at 340 nm. These meas-
urements were carried out at varying concentrations of 1 and 2
up to 2 m, under the same buffer conditions as for the MECC
experiments. The cmc values of the two surfactants were
obtained from a plot of fluorescence intensity versus surfactant
concentration.
Determination of the mean aggregation number, N
The method outlined by Turro and Yekta11 was applied, using
tris(2,2Ј-bipyridyl)ruthenium chloride (D) and 9-methyl-
anthracene (Q) as the luminescent donor and quencher, respect-
ively. The Perkin-Elmer spectrofluorimeter was again used for
these measurements. The concentration of D was fixed at
7.2 × 10Ϫ5  and that of Q was varied between 0 and 10Ϫ4 .
The concentration of 1 and 2 was also kept constant at 40 m.
N was determined from the slope of a plot of ln(IO/I) versus the
ratio of the concentration of quencher to that of the surfactant.
Again the same buffer conditions were used as in the case of the
MECC and cmc measurements.
Tetradecyl á-D-glucopyranoside 4,6-(phenyl phosphate) 7. Dry
triethylamine (0.8 cm3, 9.1 mmol) was added to a stirred solu-
tion of glucopyranoside 6 (1.60 g, 4.2 mmol) in 900 cm3 of dry
dichloromethane containing 5 g of 4 Å molecular sieves and the
solution was stirred under argon for 1 h. The solution was then
cooled to 0 ЊC and phenyldichlorophosphate (1.03 g, 4.9 mmol)
was added and the reaction stirred for another 3 h at room
temp. The solution was then filtered and the filtrate was concen-
trated in vacuo. The residue was purified by flash column chro-
matography (SiO2, EtOAc–toluene, 2:1) to yield the phosphate
triester (1.06 g, 48.5%) as a clear oil; Rf (SiO2, EtOAc–toluene,
Capillary electrophoresis
MECC separations were carried out on a Beckman P/ACE
Model 2100 (Beckman, High Wycombe, Bucks, UK). A fused
silica capillary, 57 cm (effective length, 50 cm) × 50 µm (i.d.)
was used. The separation buffer consisted of 30 m sodium
dihydrogen phosphate, 10 m boric acid and 40 m of 1 or 2
adjusted to pH 7.1 using dilute sodium hydroxide. The separ-
ation voltage was 15 kV at a temperature of 30 ЊC. Analytes
were injected as aqueous solutions using high pressure and were
monitored at a constant wavelength of 254 nm.
2:1), 0.4; νmax/cmϪ1 (CH Cl ) 3563 (OH), 1593, 1490 (C᎐C),
᎐
2
2
1319 (P᎐O) and 1205 (PO); δ (400 MHz, CDCl ) 7.31 (2 H, t,
᎐
H
3
J 7.8, Hmeta), 7.24 (2 H, d, J 8.5, Hortho), 7.15 (1 H, t, J 7.3, Hpara),
4.83 (1 H, d, J 3.8, OCHO), 4.32 (1 H, ddd, J 23.7, 10.2 and 4.8,
POCHAHB), 4.22 (1 H, t, J 10.1, CHOP), 4.15–4.0 (2 H, m,
POCHAHB and CHO), 3.95 (1 H, t, J 9.0, CHOH), 3.7–3.6 [3
H, m, OCHAHBR and 2 × CHOH (br)], 3.53 (1 H, dd, J 9.3 and
3.8, OCHO), 3.44 (1 H, dt, J 9.6 and 6.8, OCHAHBR), 1.7–1.6
(2 H, m, OCH2CH2C12H25), 1.4–1.2 (22 H, m, OC2H4C11-
H22CH3) and 0.85 (3 H, t, J 6.7, CH3); δC(100 MHz, CDCl3)
150.0Ϫ (1 C, d, J 6.7, Cipso), 130.0ϩ (2 Cmeta), 125.5ϩ (Cpara),
119.7ϩ (2 C, d, J 4.8, Cortho), 98.8ϩ (OCHO), 81.1ϩ (1 C, d,
J 6.8, CHOP), 72.0ϩ (CHOH), 71.3ϩ (1 C, d, J 8.0, CHO),
69.5Ϫ (1 C, d, J 7.9, CH2OP), 69.3Ϫ (OCH2R), 62.3 (1 C, d,
J 5.7, CHOH), 31.9Ϫ, 29.7Ϫ, 29.6Ϫ, 29.5Ϫ, 29.4Ϫ, 29.4Ϫ,
26.0Ϫ, 22.7Ϫ and 14.2ϩ; δP(100 MHz, CDCl3, H-coupled)
Ϫ154.0 (1 P, d, J 56.8); m/z 514.3 (65%, Mϩ), 94 (100%, PhOH)
(Found: Mϩ, 514.2689. C26H43PO8 requires 514.2695).
Molecular modelling
Initially 55 monomers (for both the α and β anomers) were
arranged in an approximate sphere, with the aliphatic tail
methyl carbons forming a core nucleus in van der Waals con-
tact. Simulated annealing was used as the sampling procedure
to overcome many of the numerous local minima that will be
present on such a highly complex potential energy hypersurface.
The simulated annealing methodology has been shown by many
workers12,13 to be an effective tool for this type of problem,
therefore only the ensemble details will be reported here. A
molecular dynamics annealing strategy was employed utilizing:
a temperature of 700 K; a Verlet algorithm with an integration
time step of 1 fs; 5 ps of high temperature molecular dynamics
equilibration; 50 structures were recorded at 500 fs intervals.
Each stored structure was subjected to a full geometry opti-
mization to yield representative structures from which a suitable
model was chosen. In the simulation, flat-bottomed harmonic
tethers were used to overcome the initial problem of bad con-
tacts and to prevent the micelle from evaporating during the
high-temperature simulation. These parameters allowed free
movement of the terminal aliphatic methyl groups up to a dis-
tance of 8 Å from the arbitrarily defined centre of the micelle. If
this distance was exceeded then a harmonic force constant was
applied to keep them within the constraint distances.
Tetradecyl á-D-glucopyranoside 4,6-(hydrogen phosphate)
sodium salt 2. Aqueous sodium hydroxide (1 ; 5 cm3) was
added to a stirred solution of phosphate triester 7 (1.0576 g,
2.1 mmol) in 45 cm3 of THF and the solution stirred for 15 h.
The solution was concentrated in vacuo to approximately 10
cm3 and neutralized with 1  HCl. The resulting solution was
then lyophilized to yield a white powder which was placed in a
Soxhlet extractor and washed with dichloromethane for 48 h to
yield the phosphate diester 2 (0.902 g, 95.2%) as white flakes;
νmax/cmϪ1 (Nujol) 3570 (OH), 1310 (P᎐O); δ (400 MHz, D O)
These simulations were performed using the BIOSYM14 suite
of codes, namely INSIGHTII, DISCOVER3.0 and the cvff
forcefield. The lowest energy conformation that resulted from
the simulated annealing could be a plausible model for the
micelle structure within the constraints of the approximations
and it was this structure that the flexible docking studies
exploited.
The program DELPHI was used to calculate the Poisson–
Boltzmann15 electrostatic potential for each model of the α
and β micelle. This surface was subsequently contoured on to
a calculated solvent accessible Connolly surface which enabled
subtle differences to be displayed between the two micelle
structures.
᎐
H
2
4.85 (1 H, d, J 3.5, OCHO), 4.15–3.40 (8 H, m, OCH2R,
CH2OP, CHOP, 2 × CHOH and CHO), 1.7–1.6 (2 H, m,
OCH2CH2R), 1.4–1.2 (22 H, m, OC2H4C11H22CH3) and 0.9–0.8
(3 H, m, CH3); δC(100 MHz, D2O) 101.8ϩ (OCHO), 80.8ϩ
(CHOP), 74.0ϩ (CHOH), 73.6ϩ (CHO), 71.5Ϫ (OCH2R),
69.5Ϫ (CH2OP), 66.5ϩ (CHOH), 34.7Ϫ, 32.7Ϫ, 32.3Ϫ, 31.8Ϫ,
28.6Ϫ, 25.3Ϫ and 16.5ϩ; δP(250 MHz, D2O, H-coupled)
Ϫ143.0 (1 P, d, J 41.5); δP(250 MHz, D2O, H-decoupled)
Ϫ143.0; m/z (ϩve FAB) 461.4 (100%, MHϩ).
Determination of critical micelle concentration (cmc)
The cmc values for 1 and 2 were determined using a spectro-
fluorimeter (Perkin-Elmer, UK) monitoring the fluorescence
emission intensity of N-phenyl-1-naphthylamine (5 × 10Ϫ6 ) at
This paper also reports the docking studies of the represen-
tative of the dansyl derivatives, namely dansyl-Phe, onto the
J. Chem. Soc., Perkin Trans. 2, 1998
469

View Article Online
Fig. 1 Plot of the fluorescence intensity of N-phenyl-1-naphthylamine
with concentration of (a) surfactant 1 and (b) surfactant 2
proposed micelle models. The procedures used to dock the
dansyl derivative with the surface of the micelle are again
embodied in the BIOSYM software (flexible docking module).
This procedure utilizes a Monte Carlo approach, whereby the
torsional angles of the dansyl-Phe are perturbed ( 180Њ) and
this substrate then replaced at an arbitrary point on the surface
(both in Cartesian and angular space). Energy minimisation is
subsequently performed on the resulting complex. The inter-
action energy of the chiral derivative and the micelle surface is
calculated and is termed the binding energy; both the minimum
and average binding energies are calculated.
Fig. 2 Electropherograms showing the separation of the enantiomers
of dansylamino acid (DNS) derivatives using (a) surfactant 1 and (b)
surfactant 2. The numbering of the chiral mixtures shown in this figure
refers to the following derivatives: (1) DNS-valine; (2) DNS-norvaline;
(3) DNS-leucine; (4) DNS-norleucine; (5) DNS-phenylalanine; (6)
DNS-tryptophan.
MECC experiments (see later). Under these conditions N was
found to be about 65 ( 2) for both surfactants. This value is
similar to the value of 61 obtained for SDS using either the
same method or from membrane osmometry studies.16,17
The separation of a mixture of the dansyl derivatives of
six amino acid racemates is shown in Fig. 2. From these data
it is clear that, under the present conditions of separation, the
β-anomer (1) is a much superior chiral selector than the α-
anomer (2). Thus when the former surfactant is included in
the separation buffer baseline resolution is obtained for all the
dansyl derivatives. In contrast the α-anomer only fully separates
the dansyl derivative of phenylalanine.
Transmission electron microscopy
Solutions of surfactants 1 and 2 were prepared in 5 m borate
buffer (pH 8.5) at ambient temperature, at two concentrations
(4 and 40 m). Samples were negatively stained on carbon/
Formvar coated grids with sodium phosphotungstic acid (pH
7.4) for 15 seconds, blotted dry and examined in a Hitachi
H7100 transmission microscope.
Results and discussion
The fluorescence intensity of the N-phenyl-1-naphthylamine
is increased substantially as the concentration of 1 or 2 is
increased, indicating a high degree of distribution of this
hydrophobic probe within micelles [Fig. 1(a) and 1(b)] formed
from either of the two surfactants. This arises due to an increase
in the fluorescence quantum yield from the excited probe via a
reduction of radiationless mechanisms, in particular quenching
of the S1 state by molecular oxygen. The break in the plot of
the variation of fluorescence intensity with concentration of
surfactant indicates the transition from monomeric species to a
micellar form. The point of intersection of the two lines for
each surfactant gives a measure of the corresponding cmc.
The cmc values for 1 and 2 were found to be virtually identi-
cal and were determined as 0.60 and 0.57 m, respectively.
These values are of the same order of magnitude as those
determined for two other glucopyranoside surfactants contain-
ing an n-dodecyl (C12) hydrocarbon chain8 and are more than
an order of magnitude lower than that for the commonly used
achiral surfactant sodium dodecyl sulfate (SDS).
There is no obvious relationship of migration order to charge
density, even though these dansyl derivatives are negatively
charged at the pH of this analysis. If this were the case, and
assuming that all analytes carry a single negative charge
under the conditions of these experiments, then the order of
migration should be the reverse of that observed. Under these
conditions the ‘bulkiest’ derivative containing the tryptophan
moiety would migrate first as the difference between its electro-
osmotic and electrophoretic mobilities is the greatest. Con-
versely, the dansyl derivative with the smaller valine substituent
would be expected to migrate last.
In the case of micelles from either 1 or 2 migration times
are largely related to the relative hydrophobicity of the dansyl
derivatives. Thus the dansyl derivative of valine containing an
isopropyl group migrates first and the tryptophan derivative
(the most hydrophobic) migrates last. These observations might
point to the role played by the surfactant counter-ions (in this
case sodium) in the general mechanism of partitioning by
micelles. We discuss this further in the context of our molecular
modelling studies on 1 and 2 in the following paragraphs.
Comparing Fig. 2(a) and 2(b) it is also clear that the
migration times of the unresolved analytes using 2 are slightly
reduced compared to those obtained for the resolved antipodes
The measured mean aggregation number (N) of 1 and 2 was
determined using the method of Turro and Yekta11 as detailed
in the Experimental section. These measurements were carried
out using the same buffer conditions as those used in the
470
J. Chem. Soc., Perkin Trans. 2, 1998

View Article Online
Fig. 3 Calculated lowest energy structures for micelles formed by (a) surfactant 2 and (b) surfactant 1
with surfactant 1. This behaviour indicates a weaker interaction
of the dansyl amino acid derivatives with surfactant 2.
β-model micelle structures. The β-micelle appears less compact
than the α-micelle, the average distance of the phosphorus in
Not surprisingly, the orientation of the chiral head groups
(that is, the glucopyranoside moieties) with respect to each
other appears to be the most important feature leading to dif-
ferences in chiral resolution. It is also interesting that whereas
the derivative of the -amino acid migrates before the corre-
sponding -enantiomer for all six analytes using 1, the order of
migration for the phenylalanine derivative is reversed when 2 is
added to the separation buffer. In the following section we use
molecular modelling in an attempt to define the macro- and
micro-molecular properties of these micelles, to probe the
extent of counter-ion binding, and to explain the observed
trends in chiral discrimination.
the head group from the centre of the micelle being 20.94 Å
for the α- and 21.53 Å for the β-micelle. Another interesting
structural feature is the amount of solvent-accessible chiral
head group surface present for each micelle model. A solvent
accessible Connolly surface of the chiral head groups for each
micelle reports a value of 7066 Å2 for the α- and 7132 Å2 for the
β-micelle. Clearly, if the effectiveness of the micelle is depend-
ent simply on the amount of solvent accessible ‘chiral face’ then
the β-micelle is predicted to be a more effective separation
phase than the α.
The surface distribution of electrostatic potential is another
property that could enhance the effectiveness of a micelle. Fig. 4
shows the amount of electrostatic potential on the calculated
surface for each micelle model. The total charge on the calcu-
lated surface has been obtained from these plots: the β-micelles
are found to contain more negative potential on the total chiral
head group surface as compared to that calculated for the
surface on α-micelles.
The above discussion relates to the differences between the
predicted structures of the α- and β-micelles, but this does not
explain or attempt to model the effective separation of the
- and -enantiomers of the dansyl derivatives. Thus the separ-
ation of one of the dansyl derivatives, dansyl-Phe, was inves-
tigated via flexible docking of this species onto the surface
of each of the isomeric micelles. Although the MECC work
reported at this stage of involvement of molecular modelling
studies had only investigated the dansyl–β-micelle complex, the
interaction with the α-micelle was also investigated via compu-
tational means. Fig. 5(a) and 5(b) report the lowest binding
energy structure calculated for the interaction of the β-micelle
and the  and  stereomers, respectively. Clearly the Naϩ
counter-ions play an important role in both structures in chelat-
ing the carboxylic acid group (which is expected to be deproton-
ated at the experimental pH of 7.1). The remaining parts of the
dansyl derivative interact with another chiral head group. These
structural features are interesting since they demonstrate that
the chiral monomer will be of limited use as a chiral separating
Insights from molecular modelling
The lowest energy structures that the simulated annealing pro-
cedure evolved are shown in Fig. 3(a) and 3(b), respectively for
both the α- and β-micelles. In this representation, the chiral
glucopyranoside head groups are represented as solid spheres
(the Naϩ counter ions are in cyan) whereas the aliphatic tails
are displayed in the stick representation. Both models of the
micelle structure clearly display clustering of the chiral head
groups. This is due in part to the head group sharing of the Naϩ
counter-ions. Such a feature would not have been easily pre-
dicted without the modelling studies presented here. The calcu-
lated radial distribution function of the Naϩ-oxygen is found to
contain a first shell peak at 2.1 Å which is consistent with the
Naϩ being closely associated with a charged phosphate group.
The detailed sub-structure at larger distances is typical of a
more fluid-like aspect of the structure, and a detailed inspection
of the structure identifies the head group sharing of the counter
ions. This phenomenon would be solvent dependent, but the
lack of solvent in these simulations is a necessary approxim-
ation because of resource constraints. Trends identified via
isolated/gas phase simulations, are usually observed in solvent
simulations: in any case we are concerned primarily with differ-
ences between two closely similar systems.
There are discrete structural differences between the α- and
J. Chem. Soc., Perkin Trans. 2, 1998
471

View Article Online
Fig. 4 Electrostatic potential located on the chiral head groups for (a) surfactant 2 and (b) surfactant 1
Fig. 5 Lowest energy structures obtained for (a) -dansylphenylalanine and (b) -dansylphenylalanine interacting with surfactant 1
Table 1 Calculated binding energies for the interaction of dansyl-Phe
and the micelle models
agent, since there is not enough ‘chiral face’ to interact with the
dansyl-Phe molecule.
The average and lowest binding energy for these systems can
be calculated and these results are displayed in Table 1. Initially
focusing on the experimentally known system, the β-dansyl
complex, the general trend is that the -dansyl-Phe binds more
tightly to the β micelle than the -dansyl-Phe, i.e. the complex
has a lower calculated binding energy. This trend is reproduced
for both the calculated lowest binding energy and the average
binding energy. Thus on an empirical basis the  form will
be retained longer on the micelle, since it has a larger (more
negative) binding energy, and hence will elute after the  form.
This empirical relationship is in agreement with experimental
data.
α-
α-
β-
β-
Min. binding energy/
kcal molϪ1
Ϫ114.83
Ϫ70.29
Ϫ75.75
Ϫ49.30
Ϫ90.98
Ϫ69.40
Ϫ143.80
Ϫ75.84
Average binding
energy/kcal molϪ1
The trend for the dansyl derivative and the β micelle can be
reversed for the chiral interactions with the α micelle. These
simulations predicted, prior to experiment, that the  form will
elute before the  form because for these systems the  form has
a larger binding energy than the  form. The experiment has
472
J. Chem. Soc., Perkin Trans. 2, 1998

View Article Online
Fig. 6 Transmission electron microscopy of solutions of surfactant in 5 m borate buffer: surfactant 1 at (a) 4 m and (b) 40 m; and surfactant 2
at (c) 4 m and (d) 40 m. The length of the scale bars is 100 nm.
since been performed and is in agreement with the prediction:
using the α micelle, the  enantiomer elutes before the  form.
ation was not evident in 2 where the strings appeared criss-
crossed between aggregates of spherical forms of varying size
and shape.
Electron microscopy
In the present study we have also explored the aggregation
behaviour of 1 and 2 using transmission electron microscopy.
It should be emphasized that as the samples were air dried
in the presence of an electron dense background stain, the
organization properties of 1 and 2 may be different from
those in free solution. The two surfactants were each studied at
4 and 40 m concentrations. Results are shown in Fig. 6(a)–
(d). Note that during the preparation of the negative stains
surfactant 1 spread the stain more evenly than 2. At 4 m con-
centration, the latter surfactant appeared to be less ordered
than 1; 5–10 nm spherical structures surrounded by a hap-
hazard array of fibrillar/string-like forms were observed for 2.
In contrast, 1 showed aggregates made up of spherical forms
that appeared in some instances to coalesce into ordered string-
like arrays.
Conclusions
We have shown that the chiral separation efficiency of
-glucopyranoside anionic surfactants depends strongly on
the orientation of the C14 hydrocarbon chain at the anomeric
carbon centre. The order of migration in the analysis of
racemic mixtures of dansyl derivatives using these α- and
β-surfactants at concentrations above their cmc could be
explained by molecular modelling studies. We have also shown
by transmission electron microscopy that inversion at the ano-
meric centre has a marked effect on the extent of organisation
of micro-structures formed by the two isomeric surfactants.
References
1 S. Fanali, J. Chromatogr., 1989, 474, 441.
2 M. J. Sepaniak, R. O. Cole and B. K. Clark, J. Liq. Chromatogr.,
1992, 15, 1023.
3 T. Ueda, F. Kitamura, R. Mitchell, T. Metcalf, T. Kuwara and
A. Nakamoto, Anal. Chem., 1991, 63, 2979.
4 S. Terabe, M. Shibita and Y. Miyashita, J. Chromatogr., 1989, 480,
403.
At 40 m the ordered nature of surfactant 1 was even more
apparent. The 5–10 nm spherical forms were reduced in number
with extensive developments of the string forms into parallel
arrays and whorls. Amongst these structures, foci of larger 25–
120 nm droplets were present that might represent a degree of
coalescence/nucleation between strings. This level of organiz-
J. Chem. Soc., Perkin Trans. 2, 1998
473

View Article Online
5 R. O. Cole, M. J. Sepaniak and W. L. Hinze, J. High Resolut.
Chromatogr., 1990, 13, 579.
6 G. E. Barker, P. Russo and R. A. Hartwick, Anal. Chem., 1992, 64,
3024.
12 F. Guarnieri and H. Weinstein, J. Am. Chem. Soc., 1996, 118,
5580.
13 R. Tejero, D. Bassolinoklimas, R. E. Bruccoleri and G. T.
Montelione, Protein Sci., 1996, 5, 578.
7 M. P. Gasper, A. Berthod, U. B. Nair and D. W. Armstrong, Anal.
Chem., 1996, 68, 2514.
8 D. C. Tickle, G. N. Okafo, P. Camilleri, R. F. D. Jones and A. J.
Kirby, Anal. Chem., 1994, 66, 4121.
9 V. de Biasi, J. Senior, J. A. Zukowski, R. C. Haltiwanger, D. E.
Eggleston and P. Camilleri, J. Chem. Soc., Chem. Commun., 1995,
1575.
10 M. Bouzige, G. Okafo, D. Dhanak and P. Camilleri, Chem.
Commun., 1996, 671.
14 BIOSYM, 9625 Scranton Road, San Diego, California, USA.
15 B. Honig, K. Charp and A. J. Yang, J. Phys. Chem., 1993, 97, 1101.
16 P. Camilleri and G. N. Okafo, J. Chem. Soc., Chem. Commun., 1992,
530.
17 M. F. Emmerson and A. Holtzer, J. Phys. Chem., 1965, 69, 3718.
Paper 7/08418H
Received 21st November 1997
Accepted 17th December 1997
11 N. J. Turro and A. Yekta, J. Am. Chem. Soc., 1978, 100, 5951.
474
J. Chem. Soc., Perkin Trans. 2, 1998