Role of the carbohydrate moieties in chiral recognition on teicoplanin- based LC stationary phases

Article abstract of DOI:10.1021/ac991004t

For this study, we used the macrocyclic antibiotic teicoplanin, a molecule consisting of an aglycone peptide 'basket' with three attached carbohydrate (sugar) moieties. The sugar units were removed and the aglycone was purified. Two chiral stationary phases (CSPs) were prepared in a similar way, one with the native teicoplanin molecule and the other with the aglycone. Twenty-six compounds were evaluated on the two CSPs with seven RPLC mobile phases and two polar organic mobile phases. The compounds were 13 amino acids or structurally related compounds (including DOPA, folinic acid, etc.) and 13 other compounds (such as carnitine, bromacil, etc.). The chromatographic results are given as the retention, selectivity, and resolution factors along with the peak efficiency and the enantioselective free energy difference corresponding to the separation of the two enantiomers. The polarities of the two CSPs are similar. It is clearly established that the aglycone is responsible for the enantioseparation of amino acids. The difference in enantioselective free energy between the aglycone CSP and the teicoplanin CSP was between 0.3 and 1 kcal/mol for amino acid enantioseparations. This produced resolution factors 2-5 times higher with the aglycone CSP. Four non amino acid compounds were separated only on the teicoplanin CSP. Six and five compounds were better separated on the teicoplanin and aglycone CSPs, respectively. Although the sugar units decrease the resolution of α-amino acid enantiomers, they can contribute significantly to the resolution of a number of non amino acid enantiomeric pairs.

Full text of DOI:10.1021/ac991004t

Anal. Chem. 2000, 72, 1767-1780
Ro le o f t h e Ca rb o h yd ra t e Mo ie t ie s in Ch ira l
Re c o g n it io n o n Te ic o p la n in -Ba s e d LC S t a t io n a ry
P h a s e s
†
Ala in Be rthod, Xia nghong Che n, J ohn P. Kullm a n, a nd Da nie l W. Arm s trong*
Department of Chemistry, University of MissourisRolla, Rolla, Missouri 65409
Fra nc e s c o Ga s pa rrini, Ila ria D’Ac qua ric a , a nd Cla udio Villa ni
Dipartimento di Studi di Chimica e Tecnologia delle Sostanze Biologicamente Attive, Universit a` “La Sapienza”,
00185 Roma, Italy
Ange lo Ca rotti
Dipartimento Farmaco-Chimico, Universita di Bari, 70125 Bari, Italy
For this study, we used the macrocyclic antibiotic teico-
planin, a molecule consisting of an aglycone peptide
appearance of penicillin-resistant bacteria prompted the search
for different molecules such as methicillin, the first manmade
antibiotic. Strains of methicillin-resistant Staphylococcus aureus
(MRSA) also appeared, and a neverending struggle against
infectious agents began. Today, strains of MSRA and Enterococcus
faecalis are resistant to almost all antibiotics and have become
“
basket” with three attached carbohydrate (sugar) moi-
eties. The sugar units were removed and the aglycone was
purified. Two chiral stationary phases (CSP s) were pre-
pared in a similar way, one with the native teicoplanin
molecule and the other with the aglycone. Twenty-six
compounds were evaluated on the two CSP s with seven
RP LC mobile phases and two polar organic mobile
phases. The compounds were 13 amino acids or structur-
ally related compounds (including DOP A, folinic acid,
etc.) and 1 3 other compounds (such as carnitine, bro-
macil, etc.). The chromatographic results are given as the
retention, selectivity, and resolution factors along with the
peak efficiency and the enantioselective free energy dif-
ference corresponding to the separation of the two enan-
tiomers. The polarities of the two CSP s are similar. It is
clearly established that the aglycone is responsible for the
enantioseparation of amino acids. The difference in enan-
tioselective free energy between the aglycone CSP and the
teicoplanin CSP was between 0 .3 and 1 kcal/ mol for
amino acid enantioseparations. This produced resolution
factors 2-5 times higher with the aglycone CSP . Four non
amino acid compounds were separated only on the
teicoplanin CSP . Six and five compounds were better
separated on the teicoplanin and aglycone CSP s, respec-
tively. Although the sugar units decrease the resolution
of r-amino acid enantiomers, they can contribute signifi-
cantly to the resolution of a number of non amino acid
enantiomeric pairs.
1
killers throughout the world. Currently, the antibiotics of last
resort are glycopeptides of the vancomycin family. It was dem-
onstrated that the mode of action of these antibiotics was different
2
from that of earlier compounds. The vancomycin-related antibiot-
ics bind to the bacterial cell wall D-alanyl-D-alanine terminal group,
blocking the process of wall building. Several members of the
group, vancomycin, avoparcin, ristocetin, and teicoplanin, are used
more wisely today to contain the “superbug” MSRA proliferation.
Since the target of these antibiotics is the
D-alanyl-D-alanine
group, we thought early on that they could be used in the
separation of amino acid enantiomers.3,4 It turned out that chiral
stationary phases (CSPs) based on these macrocyclic antibiotics
were very successful, not only in amino acid enantioresolution5,6
but also in the resolution of a wide variety of different enanti-
omers,7,8 using both HPLC and capillary electrophoresis (CE)
techniques.3
-11
(
(
(
1) Williams, D. H.; Westwell, M. S. CHEMTECH 1 9 9 6 (March), 17.
2) Harris, C. M.; Harris, T. M. J. Am. Chem. Soc. 1 9 9 2 , 104, 363.
3) Armstrong, D. W.; Tang, Y.; Chen, S.; Zhou, Y.; Bagwill, C.; Chen, J. R.
Anal. Chem. 1 9 9 4 , 66, 1473.
(
(
4) Armstrong, D. W.; Liu, Y.; Ekborg-Ott, K. H. Chirality 1 9 9 5 , 7, 474.
5) Berthod, A.; Liu, Y.; Bagwill, C.; Armstrong, D. W. J. Chromatogr. 1 9 9 6 ,
7
31, 123.
(6) Peter, A.; Torok, G.; Armstrong, D. W. J. Chromatogr., A 1 9 9 8 , 793, 283.
(
(
7) Armstrong, D. W. LC-GC 1 9 9 7 (May), S20.
8) Armstrong, D. W.; Nair, U. B. Electrophoresis 1 9 9 7 , 18, 2331-2342.
After the 1928 discovery of penicillin, it was thought that life-
threatening infections could be completely controlled. Soon, the
(9) D’Acquarica, I.; Gasparrini, F.; Misiti, D.; Villani, C.; Carotti, A.; Cellamare,
S.; Muck, S. J. Chromatogr., A. 1 9 9 9 , 857, 145; originally presented at the
9th International Symposium on Chiral Descrimination, Nagoya, Japan, Oct
2
7-30, 1997.
†
On leave from the Laboratoire des Sciences Analytiques, University of Lyon
(10) Canceliere, G.; D’Acquarica, I.; Gasparrini, F.; Misiti, D.; Villani, C. Pharm.
1
, Bat. 308-D, UA CNRS 5619, 69622 Villeurbanne cedex, France.
Sci. Technol. Today 1 9 9 9 , 2, 484.
1
0.1021/ac991004t CCC: $19.00 © 2000 American Chemical Society
Analytical Chemistry, Vol. 72, No. 8, April 15, 2000 1767
Published on Web 03/09/2000

In some antibiotic studies, it was found that vancomycin
derivatives with altered carbohydrate moieties killed bacteria by
a mechanism differing from that of native vancomycin.12 In some
cases, carbohydrate units alone had antibiotic activity.12 Transpos-
ing these findings to the chiral recognition process is a logical
step. The question arises as to the exact role of the sugar moieties
in the highly successful teicoplanin CSP. To answer this question
and investigate the role of the teicoplanin sugar units, the three
carbohydrate units of the teicoplanin molecule were removed and
the teicoplanin aglycone “basket” was isolated and purified. Two
CSPs were prepared in a similar way: one with the complete
teicoplanin molecule and the other with only the aglycone
obtained was cooled to 5 °C and left overnight. Then the
precipitated solid was recovered by filtration, washed with water
(5 mL), and dried under vacuum (0.1 mbar) at 60 °C for 15 h,
giving 2.60 g of crude light brown aglycone (molar yield 80%).
The molecular weight of each product was checked by high-
resolution MS after each step as indicated in Figure 1. The crude
aglycone (2.60 g) was further purified as follows. It was dissolved,
with stirring, in 15 mL of DMSO, plus 15 mL of demineralized
water and 0.6 mL of 20% HCl(aq). When the dissolution was
complete, another 9 mL of water was added and the pH of the
solution was brought to 7 by adding 20% NaOH(aq) solution. The
suspension was then cooled to 5 °C for 12 h, and the solid
aglycone was filtered, washed with water (15 mL) and acetone,
and dried under vacuum (0.1 mbar) at 50 °C, giving 1.82 g of
pure aglycone (molar yield 70%).
“basket.” The enantiorecognition capability of the two CSPs was
evaluated using an identical set of racemic compounds containing
a variety of functionalities.
P reparation of the Chiral Stationary P hases. The full
EXPERIMENTAL SECTION
9
procedure was recently reported. A 5 g sample of LiChrospher
Chemicals. The stationary-phase base was LiChrospher Si-
Si(100) was dried at 150 °C for 1 h under vacuum (0.1 mbar) in
a round-bottom flask. A 120 mL portion of dry toluene was added,
and the mixture was heated to reflux to azeotropically remove
any residual water. Then, 2.5 mL of (3-aminopropyl)triethoxysilane
2
(
100) silica gel (5 µm particle size, 10 nm pore diameter, 400 m / g
specific surface area (Merck, Darmstadt, Germany)). The reagents
3-aminopropyl)triethoxysilane and 1,6-diisocyanatohexane were
(
obtained from Fluka (Buchs, Switzerland). Table 1 lists the
racemic compounds that were tested on the CSPs, each with its
structure and a reference number used in the other tables. They
were all obtained from Sigma-Aldrich (St. Louis, MO). The
chemical solvents dry toluene and pyridine were from J. T. Baker
(
11 mmol) was added dropwise and the mixture was heated to
reflux for 4 h. After cooling, the modified silica was filtered off,
and washed with toluene, methanol, and dichloromethane, and
dried at 90 °C (0.1 mbar, 1 h). The elemental analysis gave 4.18%
C and 1.14% N, corresponding to 1045 µmol/ g of aminopropyl
(
Phillipsburg, NJ). The chromatographic solvents methanol and
2
groups or 2.61 µmol/ m based on the N percentage.
acetonitrile and the buffer additives (acetic acid, ammonium
acetate, and triethylamine) were from Fisher Scientific (Fair Lawn,
NJ). Teicoplanin was a gift of the LePetit Research Center
A 2.5 mL portion of 1,6-diisocyanatohexane (15 mmol) was
added to an ice-bath-cooled slurry of 2.5 g of 3-aminopropyl-
LiChrospher in 50 mL of dry toluene. Then, the mixture was
heated at 70 °C for 2 h. After cooling, the supernatant toluene
phase was removed under an argon atmosphere. The excess
reactant was removed by dry toluene washing. A suspension of 1
g of teicoplanin (0.53 mmol) in 100 mL of dry pyridine was added
dropwise to the wet activated silica. Next, the mixture was heated
at 70 °C for 12 h with stirring under an argon atmosphere.
Disappearance of the main teicoplanin peak from the reaction
(
Gerenzano, Italy).
P reparation of the Teicoplanin Aglycone. A 5 g sample of
the teicoplanin complex (2.7 mmol) was dissolved with magnetic
stirring at room temperature in 30 mL of DMSO and 1 mL of 80%
H
2
SO
linkage of the nonylglucosamine unit was then hydrolyzed (Figure
); disappearance of the main teicoplanin peak from the reaction
4
. The mixture was heated to 65 °C for 1.5 h. The phenoxy
1
liquid phase was checked by HPLC (column 250 × 4 mm i.d. ODS
9
liquid phase was checked by HPLC, as previously described. After
Hypersil; mobile phases (A) 0.1 M ammonium acetate and (B)
cooling, the teicoplanin-bonded silica was washed with 50 mL
sequential portions of pyridine, water, methanol, acetonitrile, and
dichloromethane. It was then dried under vacuum (70 °C, 0.1
mbar, 2 h). The elemental analysis gave 15.6% C and 3.61% N,
0
1
.1 M ammonium acetate/ acetonitrile, 20/ 80; linear gradient from
0% B to 50% B in 20 min, to 75% B in 15 min, to 100% B in 5 min;
flow rate 1.00 mL/ min; T ) 20 °C; UV detection at 254 nm). A 1
mL aliquot of 80% H SO was added, and the reaction mixture
was kept at 65 °C for 3 h. The phenoxy linkage of the mannose
unit was then hydrolyzed. To remove the N-acetyl-â- -glucosamine
2
4
2
corresponding to 142 µmol/ g of teicoplanin or 0.36 µmol/ m based
on the C percentage.
D
The very same procedure was followed to prepare the aglycone
CSP. The 1,6-diisocyanatohexane-activated phase was prepared
and reacted in situ with a suspension of 0.70 g of the aglycone
unit, it was necessary to raise the temperature to 80 °C for 24 h.
Appearance of the aglycone peak from the reaction liquid phase
was checked by HPLC, as described above (same conditions
except isocratic elution; A/ B ) 75/ 25, by volume). After cooling
to 30 °C, the DMSO solution was diluted with 63 mL of
demineralized water, and the resulting pH was 1.5. To this cloudy
solution was added 750 mg of activated carbon (Darco G 60,
Aldrich Catalog No. 24227-6), and the stirring was continued for
(
0.58 mmol) in 70 mL of dry pyridine. The final elemental analysis
gave 13.64% C and 3.37% N, corresponding to 151 µmol/ g of
aglycone or 0.38 µmol/ m based on the C percentage.
2
Column P reparation. A classical packing procedure was
used: A 3.3 g sample of the bonded LiChrospher was suspended
in 60 mL of a 50/ 50 acetone/ chloroform mixture with 15%
acetic acid. After 5 min of ultrasonication, the slurry was
packed in a 250 × 4.6 mm stainless steel column at 700 bar
with a Haskel DSTV-122 pump using methanol as the pres-
surizing agent. The column efficiencies were in the 40 000
plate/ m of column range, checked with a 90/ 10 hexane/
1
h. The mixture was filtered on Celite, and the filtrate was
neutralized with 10% NaOH aqueous solution. The suspension
(
(
11) Beesley, T. E.; Scott, R. P. W. Chiral Chromatography; J. Wiley & Sons: New
York, 1998; pp 275-277.
12) Ge, M.; Zhong, C.; Onoshi, H. R.; Kohler, J.; Silver, L. L.; Kerns, R.;
Fukusawa, S.; Thompson, C.; Kahne, D. Science 1 9 9 9 , 284, 508.
1768 Analytical Chemistry, Vol. 72, No. 8, April 15, 2000

Ta b le 1 . S o lu t e s Us e d in t h e S t u d y
Analytical Chemistry, Vol. 72, No. 8, April 15, 2000 1769

Ta b le 1 . (Co n t in u e d )
1770 Analytical Chemistry, Vol. 72, No. 8, April 15, 2000

Ta b le 1 . (Co n t in u e d )
chloroform mobile phase and acetophenone (k′ > 11).
Chromatographic System. The 26 compounds were eluted
on two 25 cm columns with nine different mobile phases at 22 °C
as listed in Table 2. The results are listed in Tables 2 and 3. The
chromatographic system consisted of a Shimadzu SCL-10A System
Controller and an SIL-10A Auto Injector, a Shimadzu LC-10AT
pump, and a Shimadzu SPD-10A UV detector with a Chromatopac
CR 501 integrator (Shimadzu, Kyoto, Japan; obtained from Delta
Instruments, Mountain View, MO).
chromatographic results obtained when an enantiomeric separa-
tion was achieved on at least one CSP.
Stationary-P hase P olarity and Solute Retention. Consider-
ing the nature of the bonded chiral selectors on the silica surface,
it is apparent that different functional groups are present in the
two CSPs. The linkage between the silica surface and the antibiotic
has nine apolar methylene units and either a ureido or carbamate
9
linkage. The polar groups of the free teicoplanin antibiotic are
1
4 hydroxyl groups, of which 4 are phenolic groups, 1 free amino
group, and 1 free carboxylic acid group. Its apolar groups are the
nine methylene units of its sugar alkyl chain and the row of six
amide linkages in the macrocyclic portion of the molecule (see
Figure 1) and the seven benzene rings attached to it. For the
aglycone molecule, the basket itself has only seven polar hydroxyl
groups, of which six are phenols, and it lacks the apolar alkyl
chain connected to a sugar. It has the free amino and carboxylic
acid groups, the apolar row of six amide linkages, and the seven
benzene rings (Figure 1). However, it is difficult to evaluate the
polarities of these two stationary phases by considering just their
molecular structure. The retention factors of the different test
molecules should give a better idea of the relative polarity of these
stationary phases.
RESULTS AND DISCUSSION
Analyte Selection. Table 1 shows that the selected solutes
can be arranged in four classes. Class A contains R-amino acids
and structurally related compounds. All have the general structure
R-C*H(COOH)-NH-R′, with R and R′ differing by atoms or
groups. Two non amino acid herbicide molecules (i.e., A1 2 and
A1 3 ) with the general structure R-C*H(COOH)-O-R′ were
added as a subsection of class A because of their similar
stereogenicity. Class B consists of non amino acid enantiomeric
compounds where the stereogenic center is part of a ring structure
that introduces some rigidity into the molecule. Class C com-
pounds are non amino acid molecules with one or more aromatic
moieties directly connected to the stereogenic center. An aromatic
structure is not necessarily a benzene ring. It can be any other
π-electron-rich ring such as uracil or a pyrimidinedione ring
The retention factors of the first eluting enantiomers of the
intermediate-polarity compounds of the four classes are somewhat
lower with the same RPLC mobile phase on the aglycone phase
than on the teicoplanin phase (Tables 2 and 3). However, the
retention factors of the whole set of compounds are relatively
similar on the two CSPs with the same mobile phase. For example,
the k′ values of the first enantiomer of pindolol (D4 ) are 10.0 and
(
bromacil) or a pyranone ring (coumachlor and warfarin). Class
D compounds are non amino acid solutes that contain an alcohol
moiety connected to the stereogenic center (i.e., a general
structure R-CH
2
-C*HOH-CH -R′). Acetylcarnitine also was
2
placed in class D even though its hydroxyl group was acetylated.
These compounds were selected as a result of previous studies
1
0.7 with the 85% methanol mobile phase on the teicoplanin and
with the teicoplanin CSP.3
-5,8-10
Table 2 lists the results for the
aglycone CSPs, respectively. These k′ values both drop to 4.03
class A compounds (amino acids). Table 3 lists the results
obtained for the three other compound classes. The retention
factor, selectivity, resolution factor, peak efficiency, and enanti-
oselective energy differences are given for each compound, in
addition to the mobile phase used.
Mobile P hases and Results P resentation. All compounds
in Table 1 were evaluated with seven different RPLC mobile
phases plus a 100% methanol mobile phase. Also, a polar organic
mobile phase (acetonitrile/ methanol, 95%/ 5% v/ v, with 0.2%
triethylamine (TEA) and 0.3% acetic acid) was evaluated due to
its low hydrogen-bonding capability. A total of 234 chromatograms
were obtained on each of the two CSPs (26 compounds × 9 mobile
phases). To simplify the presentation, Tables 2 and 3 list only the
and 3.45, respectively, with the 100% methanol mobile phase
(
(
Table 3). For the more polar 5-methyl-5-phenylhydantoin solute
B2), the k′ values are 4.00 and 3.96 with the 20% methanol mobile
phase at pH 4.1. They drop to 0.30 and 0.26 when the 85%
methanol mobile phase is used. These retention factors are almost
identical between the two CSPs (Table 3). Since the stationary-
phase coverages of the two CSPs are fairly similar, it can be stated
that the overall polarity of the aglycone stationary phase used in
this study is somewhat similar to the polarity of the corresponding
teicoplanin stationary phase. The retention factors of the second
eluting isomers differ much more widely. This will be discussed
in the enantioselectivity section of this article.
Analytical Chemistry, Vol. 72, No. 8, April 15, 2000 1771

1772 Analytical Chemistry, Vol. 72, No. 8, April 15, 2000

Ta b le 2 . Ch ro m a t o g ra p h ic Re s u lt s Ob t a in e d a t 2 2 °C o n t h e Tw o CS P s w it h Nin e Diffe re n t Mo b ile P h a s e s fo r t h e
Cla s s A Co m p o u n d s (Am in o Ac id s a n d Re la t e d Co m p o u n d s )
CSPa
mobile phaseb
k′
k′
R
R
s
N
(plates)
N
(plates)
∆G (kcal/ mol)
c
compd
1
2
1
2
Amino Acid Compounds
A1
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
60-3.8
100
0.56
nd
2.20
nd
0.82
nd
4.44
nd
0.96
4.52
0.66
3.9
1.08
4.55
1.09
5.76
0.9
1.46
0.7
350
100
300
90
0.22
0.41
d
2.02
1.25
A2
20
0.54
0.51
0.40
0.40
0.66
0.48
0.53
0.57
0.45
0.38
0.50
0.45
1.10
1.26
9.43
2.39
6.31
3.84
1.93
2.29
2.72
2.42
1.79
2.45
0.53
1.13
2.25
2.68
0.67
0.41
2.70
1.83
2.32
8.45
0.32
0.92
1.56
0.28
0.36
0.30
0.26
0.16
0.51
0.60
0.72
0.46
0.63
0.48
0.80
0.57
0.98
0.64
0.64
0.46
0.51
0.49
1.02
0.98
1.78
1.68
2.00
2.82
1.55
0.47
1.21
0.98
2.25
2.16
0.81
0.70
0.05
0.01
1.78
8.86
1.65
9.75
1.64
9.48
2.06
10.1
2.00
11.3
2.32
11.8
2.76
11.4
1.5
5.3
1.8
6.0
2.0
7.0
2.4
8.2
2.3
8.2
3.2
8.3
4.2
7.7
1.5
0
3.4
0
2.6
0
3.6
0
3.5
0
1.2
0
2.8
0
0
1.2
0
2.1
2.2
2.3
1.2
4.8
2.2
4.2
2.8
4.6
2.6
5.3
3.6
6.5
0
700
1500
2900
2200
1600
1400
1300
1500
1300
2000
2100
1700
1800
700
600
200
1300
350
1100
600
900
600
1100
600
900
600
500
400
180
0.33
1.29
0.29
1.33
0.29
1.32
0.42
1.35
0.41
1.42
0.49
1.44
0.59
1.43
0.27
0
0.38
0
0.35
0
0.38
0
0.44
0
0.50
0
0.35
0
0
0.28
0
0.24
0.34
0.39
0.27
0.67
0.18
0.91
0.47
0.87
0.28
0.52
0.57
1.11
0
20-4.1
40
60
60-3.8
60-4.1
85
4.29
1.16
5.31
3.04
14.4
15.1
2.39
12.1
3.84
3.53
2.29
5.17
2.42
3.79
2.45
1.25
1.13
4.07
2.68
0.67
0.66
2.70
2.76
4.13
16.4
0.51
2.88
2.14
1.32
0.80
1.33
0.63
1.30
1.34
4.02
0.72
1.03
0.74
1.02
0.80
1.43
1.30
1.75
0.75
1.36
0.66
1.42
1.55
3.11
2.75
5.10
11.2
14.7
4.34
6.07
4.11
4.49
9.68
10.28
6.11
5.85
2.00
1.65
A3
20-4.1
40
1.60
1.00
1.92
1.00
1.83
1.00
1.90
1.00
2.12
1.00
2.36
1.00
1.81
1.00
1.00
1.61
1.00
1.51
1.78
1.94
1.59
3.13
1.37
4.71
2.22
4.43
2.42
8.12
2.63
6.70
1.00
2.24
1.17
2.13
1.00
2.51
1.33
2.73
1.17
2.96
1.29
2.90
1.52
3.17
1.54
3.04
5.60
5.21
2.80
210
750
650
500
600
700
500
120
650
60
60-3.8
60-4.1
85
1200
800
200
100
650
A4
A5
60-3.8
60-4.1
85
900
800
1200
500
300
700
400
250
900
800
1400
600
1500
900
1400
1000
900
20
2400
2000
2900
2000
1900
1400
2000
1100
1600
1700
40
60
60-3.8
85
500
A6
20
3.0
0.5
2.6
0
2200
2200
2000
1000
1300
800
0.47
0.09
0.44
0
20-4.1
40
4.8
1.25
5.1
0.6
5.0
0.9
4.9
1.6
5.2
1.5
4.2
8.0
3.6
1.6
2000
1500
2600
1900
2600
1600
2100
800
1000
700
1000
1000
400
80
500
90
80
800
600
1000
550
1300
1300
1500
800
1300
1100
1500
1200
1500
700
800
280
360
650
120
70
350
50
70
550
450
700
350
450
550
0.54
0.17
0.59
0.09
0.63
0.15
0.62
0.25
0.68
0.26
0.65
1.01
0.97
0.60
1.50
0.72
0.90
0.85
0.91
1.18
1.25
2.16
2.75
60
60-3.8
60-4.1
85
100
A7
200.1
40
12.9
11
60
3.40
4.58
4.30
4.76
7.54
8.36
1.7
2.0
6.5
6.0
9
5.5
5.7
5.8
60-3.8
60-4.1
85
40
165
Analytical Chemistry, Vol. 72, No. 8, April 15, 2000 1773

Ta b le 2 . Co n t in u e d
CSPa
mobile phaseb
k′1
k′2
Amino Acid Compounds
R
Rs
N1 (plates)
N2 (plates)
∆G (kcal/ mol)
c
compd
A8
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
20
0.27
0.43
0.40
0.77
0.58
0.98
1.56
2.02
0.15
0.17
1.13
1.00
0.06
0.07
0.12
0.03
0.39
0.07
0.71
0.46
2.20
0.93
1.74
1.50
1.38
1.04
1.59
1.40
1.28
1.43
0.90
0.77
0.76
0.85
1.21
1.37
2.23
2.05
0.79
1.88
0.40
3.35
1.68
4.34
1.89
2.61
0.34
0.41
1.44
1.62
0.14
0.23
0.23
0.15
0.39
0.30
1.14
1.51
2.20
2.76
1.93
3.33
1.53
2.89
1.99
3.21
1.66
3.26
1.19
2.04
1.13
2.29
1.95
4.08
3.64
14.2
2.93
4.37
1.00
4.35
2.90
4.43
1.21
1.29
2.27
2.41
1.27
1.62
2.33
3.28
1.92
5.00
1.00
4.28
1.60
3.28
1.00
2.97
1.11
2.22
1.11
2.78
1.25
2.29
1.30
2.28
1.32
2.65
1.49
2.69
1.61
2.98
1.63
6.93
3.1
6.2
0
1300
2300
1200
1100
0.63
0.86
0
40
6.2
5.0
6.0
0.7
1.0
1.5
1.8
0.7
2.0
0.7
1.5
1.2
1.4
0
2.0
1.7
4.0
0
3.7
0.5
2.6
0.7
4.2
1.4
4
2200
2400
1200
600
600
1200
800
500
500
1100
1300
400
0.86
0.62
0.87
0.11
0.15
0.49
0.52
0.14
0.28
0.48
0.70
0.38
0.95
0
60-3.8
60-3.8
60-4.1
100
40
A9
500
1800
1500
500
1000
1700
1900
4400
2700
700
A1 0
1500
1700
1900
2300
60
60-3.8
85
1700
1000
1100
1600
900
800
0.85
0.28
0.70
0
100
20
1100
1800
1000
3800
1700
2600
2000
1300
1700
2000
1700
1100
2100
1000
800
400
600
250
1100
500
1200
600
1000
500
1300
900
900
900
400
600
230
300
0.64
0.06
0.47
0.06
0.60
0.13
0.49
0.15
0.48
0.16
0.57
0.23
0.58
0.28
0.64
0.29
1.13
A1 1
20-4.1
40
60
1.3
3.8
1.4
4.4
1.5
4.8
1.7
4.7
1.5
6.5
60-3.8
60-4.1
85
100
400
650
Non Amino Acid Compounds
A1 2
T
A
T
A
T
A
T
A
T
A
T
A
T
A
20-4.1
60-4.1
100
12.4
2.6
1.42
0.97
0.27
0.4
1.61
0.06
10.2
2.12
2.25
0.3
14.1
2.6
1.69
0.97
0.38
0.4
2.09
0.06
11.3
2.12
4.07
0.3
1.14
1.00
1.19
1.00
1.41
1.00
1.30
1.00
1.11
1.00
1.81
1.00
1.23
1.00
0.7
0
0.7
0
0.8
0
1.3
0
0.6
0
2.4
0
1.0
0
1300
900
400
600
0.07
0
0.10
0
0.20
0.49
0.15
0
0.06
0
1400
1800
1900
1300
1500
600
95ACN
20-4.1
100
A1 3
800
500
500
1800
1600
500
1500
600
0.35
0.43
0.12
0
95ACN
1.65
0.03
2.03
0.03
a
T ) teicoplanin chiral stationary phase (CSP); A ) aglycone CSP. b Mobile phase code: 20 ) 20% methanol/ 80% water v/ v; 20-4.1 ) 20%
methanol/ 80% water buffered at pH 4.1 by 1% TEAA; 40 ) 40% methanol/ 60% water v/ v; 60 ) 60% methanol/ 40% water v/ v; 60-3.8 ) 60% methanol/
4
2
0% water buffered at pH 3.8 by acetic acid; 60-4.1 ) 60% methanol/ 40% water buffered at pH 4.1 by 1% TEAA; 85 ) 85% methanol/ 15% water with
-2
.5 × 10 M ammonium acetate; 100 ) 100% methanol with 0.1% TEA and 0.1% v/ v acetic acid; 95ACN ) 95% acetonitrile/ 5% methanol with 0.2%
TEA and 0.3% acetic acid. Enantioselective energy (-RT ln R). d nd ) not detected, weak absorbance, and/ or k′ > 40.
c
Figure 2 shows the retention factors of three compounds
versus the mobile-phase composition. For the amino acid pheny-
lalanine (A6 ), an increase in the retention factor is observed when
the mobile-phase methanol content increases. This behavior was
80 min) were listed in the tables because this retention time was
considered as a maximum acceptable value. When the injected
solute did not elute after 90 min, the column was rinsed with pure
methanol and reequilibrated with the mobile phase for another
solute.
7
observed on other stationary phases. It is due to the reduced
solubility of amino acids in methanol-rich mobile phases. For most
compounds, a linear decrease of the log k′ value is observed with
an increase in the methanol content of the mobile phase (bromacil,
C1 , Figure 2). No retention factors higher than 26 (retention time
Enantioselectivity. Tables 2 and 3 list the enantioselectivity
factors (R ) k′
factors (R ). The classical equation for the resolution factor
calculation was not used because the peak efficiency observed
2 1
/ k′ ), the peak efficiencies (N), and the resolution
s
1774 Analytical Chemistry, Vol. 72, No. 8, April 15, 2000

Ta b le 3 . Ch ro m a t o g ra p h ic Re s u lt s Ob t a in e d a t 2 2 °C o n t h e Tw o CS P s w it h Nin e Diffe re n t Mo b ile P h a s e s fo r t h e
Cla s s B-D Co m p o u n d s
CSPa
mobile phaseb
k′1
k′2
R
Rs
N1 (plates)
N2 (plates)
∆G (kcal/ mol)
c
compd
Class B Compounds: Structures Where the Stereogenic Center is Part of a Ring System
B1
B2
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
85
1.03
1.42
4.37
3.55
4.00
3.96
1.90
1.64
0.75
0.63
0.86
0.61
0.55
0.56
0.30
0.26
0.15
0.29
1.31
1.16
4.63
3.32
4.20
3.80
1.77
1.47
0.65
0.50
0.77
0.49
0.57
0.46
4.55
4.25
4.07
3.55
1.86
1.41
1.11
1.42
8.39
6.92
7.92
7.56
3.70
3.85
1.82
1.68
1.96
1.60
1.41
1.48
0.75
0.91
0.53
0.93
1.70
1.68
5.05
3.65
4.51
4.14
1.88
1.68
0.71
0.61
0.86
0.69
0.57
0.55
4.91
4.25
4.27
3.55
2.01
1.41
1.08
1.00
1.92
1.95
1.98
1.91
1.95
2.35
2.43
2.67
2.28
2.62
2.56
2.64
2.50
3.50
3.53
3.21
1.30
1.45
1.09
1.10
1.07
1.09
1.06
1.14
1.09
1.22
1.12
1.41
1.00
1.20
1.08
1.00
1.05
1.00
1.08
1.00
1.04
1.09
1.00
1.09
0.4
0
2300
1200
0.05
0
20
3.7
3.5
5.4
3.7
4.2
5.2
5.0
5.9
4.6
5.5
5.0
5.5
2.5
3.5
3.6
4.8
1.3
1.4
0.8
0.8
0.7
0.8
0.5
1.1
0.3
0.9
0.4
1.1
0
1300
1300
2300
1300
1400
2000
1500
2900
2500
2800
1700
3000
2400
2400
2700
2000
1200
1300
2300
1600
800
600
500
1200
600
0.38
0.39
0.40
0.38
0.39
0.50
0.52
0.57
0.48
0.56
0.55
0.57
0.54
0.73
0.74
0.68
0.15
0.22
0.05
0.06
0.04
0.05
0.03
0.08
0.05
0.12
0.06
0.20
0
20-4.1
40
1100
1000
2200
2100
1300
1900
2500
2000
2400
2400
2500
2600
800
1200
1800
2200
800
1500
1700
2900
1100
2100
1200
1200
60
60-3.8
60-4.1
85
100
95ACN
20
B3
20-4.1
40
2500
3300
3200
1300
2100
1100
1400
60
60-3.8
60-4.1
20
0.6
0.8
0
0.4
0
0.5
0
1.4
1.4
0
1800
2600
1800
2500
0.11
0.05
0
0.03
0
0.05
0
0.02
0.05
0
B4
20-4.1
40
1800
1600
1500
1500
B5
C1
85
14.7
23.1
6.80
8.60
15.3
25.2
6.80
9.38
2600
4200
1000
4300
100
1.3
4500
4500
0.05
Class C Compounds: Non-AA Compounds with at Least One Aromatic Ring
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
20
8.71
6.51
8.00
7.01
2.62
2.30
1.06
0.68
1.13
0.71
0.89
0.61
3.07
3.19
6.53
4.20
0.83
0.93
1.76
0.53
1.21
0.16
2.08
2.32
0.38
0.31
0.25
0.42
0.78
0.98
4.16
8.00
11.7
6.90
10.5
7.49
1.34
1.06
1.31
1.07
1.22
1.00
1.13
1.00
1.17
1.00
1.13
1.00
1.18
1.00
1.24
1.10
3.57
2.80
1.78
1.00
1.82
1.00
2.42
1.88
7.29
4.29
8.76
4.07
5.40
2.65
4.59
1.77
3.1
0.6
2.9
0.7
1.6
0
0.6
0
0.8
0
0.6
0
0.8
0
1.8
0.7
3.8
3.6
1.2
0
2200
3100
2800
2200
1800
2000
1700
2000
1700
2000
0.17
0.03
0.16
0.04
0.12
0
0.07
0
0.09
0
0.07
0
0.10
0
0.13
0.06
0.74
0.60
0.34
0
20-4.1
40
3.20
2.30
1.20
0.68
1.32
0.71
1.01
0.61
3.62
3.19
8.10
4.62
2.96
2.60
3.13
0.53
2.20
0.16
5.03
4.36
2.77
1.33
2.19
1.71
4.21
2.60
60
1500
1800
1600
700
1300
1100
1400
600
60-3.8
60-4.1
60
C2
C3
60-3.8
20-4.1
40
1400
1300
700
700
300
1500
1300
350
500
100
60
0.8
0
60
70
0.35
0
60-3.8
60-4.1
85
2.6
2.0
5.4
3.9
5.6
3.1
4.5
2.2
5.0
2.0
500
200
340
700
400
700
350
150
300
260
200
360
650
800
250
700
330
300
140
240
0.52
0.37
1.16
0.85
1.28
0.83
0.99
0.57
0.89
0.33
100
95ACN
19.1
14.16
Analytical Chemistry, Vol. 72, No. 8, April 15, 2000 1775

Ta b le 3 . Co n t in u e d
CSPa
mobile phaseb
k′1
k′2
R
Rs
N1 (plates)
N2 (plates)
∆G (kcal/ mol)
c
compd
Class C Compounds: Non-AA Compounds with at Least One Aromatic Ring
C4
T
A
T
A
T
A
T
A
40
12.4
3.84
16.1
3.84
1.30
1.00
1.33
1.00
1.28
1.13
1.20
1.08
1.8
0
1.5
0
1.9
1.0
1.3
0.7
700
800
1000
900
0.15
0
0.17
0
0.14
0.07
0.11
0.04
60
2.00
1.84
3.50
2.06
2.55
1.56
2.66
1.84
4.48
2.32
3.06
1.68
60-3.8
60-4.1
1400
2000
1600
1400
1600
2100
1400
1200
Class D Compounds: R-CH2-CHOH-CH2-R′
D1
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
T
A
40
0.37
1.47
4.50
1.63
1.08
1.50
0.63
2.07
4.50
2.45
2.19
4.50
1.69
1.41
1.00
1.50
2.03
3.00
1.09
1.00
1.19
1.08
1.19
1.43
1.10
1.33
1.06
1.06
1.19
1.00
1.7
2.5
0
4.0
4.6
7.0
1.1
0
1100
2000
2300
2200
0.31
0.20
0
0.24
0.41
0.64
0.05
0
0.10
0.05
0.10
0.21
0.06
0.17
0.03
0.02
0.10
0
60
3600
1800
1200
4900
3500
2000
1500
2200
85
D2
D3
D4
85
10.3
11.2
5.00
5.93
5.30
0.80
1.21
3.06
2.68
10.0
10.7
4.03
3.45
5.00
7.06
5.72
0.95
1.73
3.37
3.56
10.6
11.3
4.78
3.45
100
85
1.5
0.9
0.9
1.8
0.5
1.6
1.3
0.7
1.7
0
2100
3400
2100
1000
1500
450
9500
5500
2600
1300
2800
2000
1300
500
100
85
500
8500
5000
2500
100
a
T ) teicoplanin chiral stationary phase (CSP); A ) aglycone CSP. b The mobile phase code is the same as that given in Table 2. c Enantioselective
energy (-RT ln R).
with the first eluting enantiomer often differed widely from the
efficiency observed with the second enantiomer. The resolution
factor was calculated using
factors can be as high as 8. These stationary phases are sometimes
so selective in resolving enantiomers that it may take several tens
of minutes after the first enantiomer appears to see the peak
representing the second enantiomer. Figure 3 shows the chro-
matograms of DOPA (A2 ) on the two CSPs with the 60%/ 40%
1
R ) / (t - t )/ (t / ( N ) + t / ( N ))
s
2
R2
R
R
x
2
R
1
x
1
1
2
MeOH/ H
2
O unbuffered mobile phase. A 2 min span separates
-DOPA enantiomer from the -enantiomer on
the first eluting
L
D
R
where t is the retention time of an enantiomer and N, the
the teicoplanin CSP. This time difference increases to 16 min for
these enantiomers on the aglycone CSP. As already described in
corresponding peak efficiency. The difference in enantioselective
free energy was estimated using13
our previous work, the
L-forms of all the common amino acids
-5,7
elute first and the
D
-forms elute second.3
-∆∆G ) RT ln R
For the amino acid compounds, the change in mobile-phase
composition induced relatively small variations in enantioselectivity
factors and enantioselective free energies (Table 2). The difference
in enantioresolution free energy of nonionizable compounds is
almost independent of the methanol content in the mobile phase
on both CSPs (Table 3).
Role of the Carbohydrate Moieties. The aglycone of teico-
planin is not water soluble. The teicoplanin solubility in water is
in the gram per liter range.14 Thus, at least one of nature’s roles
for the three sugar units is to render the teicoplanin molecule
water soluble. The nonyl chain of the glucosamine unit is
responsible for its surfactant properties. This would tend to
enhance the concentration of teicoplanin at various interfaces,
including the bacterial cell wall. These features, water solubility
and surfactant properties, are not essential for a chiral stationary
phase. However, the three carbohydrate units are themselves
The highest enantioselectivity factors obtained on the aglycone
CSP were R ) 110 and 11.8 for N-acetylfluorophenylalanine (A7 )
and DOPA (A2 ), respectively. The highest enantioselectivity
factors obtained on the teicoplanin CSP were R ) 40 and 8.8 for
N-acetylfluorophenylalanine (A7 ) and 4-hydroxymandelic acid
(
C3 ), respectively. The very high values for compound A7 may
have high errors because the first enantiomer of this phenylalanine
derivative is very poorly retained on both CSPs. The near-zero
1
k′ values for this first enantiomer were difficult to estimate
accurately. However, several R values higher than 10 were
obtained for both CSPs. This corresponds to a difference in
enantioselective free energy in the 1.4 kcal/ mol range, which is
indicative of the very high enantiorecognition capability of these
two stationary-phase chiral selectors. Tables 2 and 3 show that
the resolution factors associated with these high enantioselectivity
(
14) Budavari, S. The Merck Index, 11th ed.; Merck & Co.: Rahway, NJ, 1989; p
(
13) Berthod, A.; Li, W.; Armstrong, D. W. Anal. Chem. 1 9 9 2 , 64, 873.
776 Analytical Chemistry, Vol. 72, No. 8, April 15, 2000
1438.
1

Fig u re 2 . Retention factors versus methanol mobile-phase content: (a) phenylalanine (A6); (b) 5-methyl-5-phenylhydantion (B2); (c) bromacil
C1); (1) k′1 for teicoplanin; (9) k′2 for teicoplanin; (3) k′1 for aglycone; (0) k′2 for aglycone.
(
chiral. Comparing the results obtained on the two CSPs may help
to determine the role of the pendant sugar moieties in enantio-
recognition.
To quantify the effect of the sugar units, the differences in
enantioselective free energies, ∆G, listed in Tables 2 and 3,
between the two CSPs were used. The values obtained for a given
compound were summed for every mobile phase, and the rounded
average values for the compounds were plotted as shown in Figure
1.25 - 2.59 ) -1.34 kcal/ mol for three different mobile phases
or an average value of -0.447 kcal/ mol that is rounded to -0.4
kcal/ mol for Figure 4. A negative number means that the
compound is better separated on the aglycone CSP. A positive
number means that the compound is better separated on the
native teicoplanin CSP, which contains the carbohydrate moieties.
Note that most of the negative numbers are for amino acids (the
class A compounds) while most of the positive numbers are for
compounds that do not contain carboxylate functionalities.
Cases Where the Carbohydrate Units Decrease Enantiorecogni-
tion. Clearly, the R-amino acids are much better resolved by the
aglycone CSP than by the native teicoplanin CSP. Figure 3 shows
the DOPA (A2 ) enantioseparations on the two CSPs. The 1 kcal/
mol energy difference corresponds to a 4 times higher R value
4
. For example, for
phases were shown to produce successful enantioseparation. The
G energies were 0.63, 0.00, and 0.62 kcal/ mol on the teicoplanin
D,L-proline (compound A8), three mobile
∆
CSP, that is, a total of 1.25 kcal/ mol. The corresponding values
were 0.86, 0.86, and 0.87 kcal/ mol on the aglycon CSP, a total of
2
.59 kcal/ mol. The energy difference between the two CSPs is
Analytical Chemistry, Vol. 72, No. 8, April 15, 2000 1777

The high free energy difference between the two related CSPs
is mainly observed with amino acids (Table 2). This may be due
to the effect of steric hindrance, but the two other possibilities
should be considered as well. The 1 kcal/ mol enantiorecognition
energy difference in the case of DOPA (A2 ) compared to the 0.5
kcal/ mol value for phenylalanine (A6 ) clearly indicates that the
hydroxyl groups of the benzene ring of DOPA interact with a
hydroxyl group of the aglycone that was formerly occupied by a
sugar unit in the teicoplanin molecule. This interaction enhances
the enantiorecognition of DOPA.
In general, it seems that the steric hindrance effect of the sugar
moieties is predominant for R-amino acids, which are thought to
“
(
dock” and bind inside the cleft of the aglycone near its amine
or ureidosif attached to a linkage chain) functional group.5,9,16
This constitutes a highly specific binding site for R- -amino acids.
It appears that most -amino acids can associate more strongly
D
D
and easily with this active binding site of the aglycone than they
can on the native teicoplanin molecule. This closer approach
produces stronger enantiointeraction and better enantioselectivity.
This can be seen in Figure 2, which shows that the L-enantiomer
of phenylalanine (A6 ) is slightly less retained on the aglycone
CSP than on the teicoplanin CSP. However, the opposite trend is
seen for the
D
-enantiomer. The addition of these two effects
-retention and increased -retention on the
(
slightly reduced
L
D
aglycone CSP) produces the large difference in the observed
enantioselectivities of the two CSPs. Besides steric hindrance, the
two phenols and the hydroxyl group created on the aglycone seem
to further enhance the interaction with the amino acids.
Fig u re 3 . Chromatograms of D,L-DOPA on the native teicoplanin
CSP (top) and the aglycone CSP (bottom). Conditions: mobile phase
The cases of the carnitine and acetylcarnitine compounds (D1
and D3 ) are different: they are partially resolved on the teico-
planin CSP, and the enantioresolution is enhanced on the aglycone
CSP (Table 3). Once again, steric hindrance may be invoked to
explain the better enantioresolution of carnitine by the aglycone
CSP. However, carnitine is a γ-amino acid whose stereogenic
center bears a hydroxyl group (Table 1). Since the acetylation of
this hydroxyl group (acetylcarnitine, A1 ) produces better enan-
tiorecognition on both CSPs (Table 3), it is likely that the primary
stereoselective interactions take place at or near the active binding
site for R-amino acids15 and the carboxylate moiety of the carnitine
molecules. Reducing the capability of the alcohol group to
hydrogen-bond to the carboxylic group helps the enantiorecog-
nition in this case.
Cases Where the Carbohydrate Units Are Necessary for Enantio-
recognition. Of the class A compounds, only folinic acid (A3 ) is
better resolved by the native teicoplanin CSP. However, compound
A3 (which is a metabolite of folic acid) is quite different from the
common protein amino acids in Table 1. Its nitrogen atom bears
a large amido substitutent (Table 1). Compounds A1 2 and A1 3
are structurally related to amino acids, but the amino group is
replaced by oxygen. This substitution significantly decreases the
binding energy of the amino acid binding site. Thus, their
enantioresolution factors are lower than those of the amino acids.
Consequently, the mobile phase must be optimized to obtain
baseline resolution (i.e., >99% methanol, Table 2). There are no
clear differences between the two CSPs, suggesting that the
stereoselective binding site, in this case, may not be the amino
60% methanol/40% water, pH 4.1 with TEA + acetic acid; 1 mL/min;
UVdetection at 254 nm; T ) 22 °C.
on the aglycone CSP compared to the teicoplanin CSP. For the
R-amino acids, the difference in enantioselective free energy is
about 0.5 kcal/ mol higher on the aglycone CSP than on the
teicoplanin CSP. The R values of amino acids are 2-3 times higher
on the aglycone CSP than on the native teicoplanin CSP. This
significant energy difference means that the sugar units decrease
the amino acid enantiorecognition. This also indicates that it is
the aglycone basket of the teicoplanin molecule that is solely
responsible for enantiorecognition of the common R-amino acids.
The sugar units may be present to decrease the affinity for amino
acids other than
D
-ala, since the teicoplanin target is the
D
2
-ala-
D
-
ala peptide termination in the Gram(+) bacterial cell wall. It was
shown for vancomycin, ristocetin, and eremomycin that the sugar
units promoted the dimerization of the antibiotic molecules,
-alanine.15
enhancing the binding to the target L-lysine-D-alanine-D
From a chiral separations point of view, the sugar units of the
native teicoplanin molecule may intervene in the chiral recognition
process in at least three ways: (i) steric hindrance, where the
sugar units occupy room inside the “basket”, which limits the
access of other molecules to binding sites; (ii) blocking possible
interaction sites on the aglycone, where two sugars are linked
through phenol hydroxyl groups and the third sugar is linked
through an alcohol moiety (Figure 1); (iii) offering competing
interaction sites, where the three sugars are themselves chiral
and have hydroxyl, ether, and amido functional groups.
(
15) Gerhard, U.; Mackay, J. P.; Maplestone, R. A.; Williams, D. H. J. Am. Chem.
Soc. 1 9 9 3 , 115, 232.
(16) Nair, U. B.; Chang, S. C.; Armstrong, D. W.; Rawjee, Y. Y.; Eggleston, D.
S.; McArdle, J. V. Chirality 1 9 9 6 , 8, 590.
1778 Analytical Chemistry, Vol. 72, No. 8, April 15, 2000

Fig u re 4 . Average enantioselectivity energy differences between the aglycone and the native teicoplanin CSPs. A positive value corresponds
to a better enantioresolution by the teicoplanin column. A negative value corresponds to a better enantioresolution by the aglycone column. For
compound codes, see Table 1.
acid binding site. Excluding the two carnitines, of the 13 non
amino acid compounds, 8 are better enantioseparated by the
teicoplanin CSP, that is, ∼70%. Four compounds are better
resolved by the aglycone CSP. Of the eight racemates better
resolved by the teicoplanin CSP, two are not separated by the
aglycone CSP. This means that 2-aminocyclohexane carboxylic
acid (B1 ) and styrene oxide (B4 ) need the carbohydrate moieties
to be enantioresolved. The six other compounds are separated
by both CSPs. For example, 4-hydroxymandelic acid (C3 ) shows
values of 3.56, 7.30, and even 18 on the teicoplanin CSP. The
corresponding R values are 2-3 times lower on the aglycone
stationary phase. This demonstrates that the sugar units often
help the enantioresolution process in the case of non amino acid
enantiomers.
can be made with non amino acid compounds such as methylphe-
nylhydantoin (B2 ) and methylphenylsuccinimide (B3 ), which
differ by a methylene group in B3 and an NH group in B2 . This
simple molecular change destroys the enantiorecognition of
hydantoin. The corresponding energy is reduced by 0.4-0.5 kcal/
mol for the succinimide derivative. But the resolution factors that
were in the 3-5 range on the two CSPs for the hydantoin
derivative drop to 0.8 down to 0, indicating no separation for the
succinimide compound. These energy changes are independent
of the mobile-phase composition. The positive point about teico-
planin-based CSPs is that if these stationary phases have a clear
receptor accepting amino acid substrates, they also have several
other enantioselective binding sites that accommodate, more
loosely, a variety of other enantiomers (Table 3).
Enantioselective Association. It is well-known that a small
change in a molecule can cause a large change in its enantiorec-
ognition by a CSP. This effect is observed among phenylalanine
Kinetic Considerations. Tables 2 and 3 list the chromato-
graphic efficiencies, N, roughly estimated using the classical
equation
7
(
A6 ), DOPA (A2 ), and N-acetylfluorophenylalanine (A7 ). The
2
addition of two hydroxy groups to the benzene ring of phenyla-
lanine produces DOPA (Table 1). The enantiorecognition free
energy increased by about 0.3 kcal/ mol on the teicoplanin CSP
and by more than 0.8 kcal/ mol on the aglycone phase. A fluorine
atom on the benzene ring and acetylation of the free amino group
produce compound A7 , whose enantiorecognition free energy
increased by almost 1 kcal/ mol on the two CSPs (Table 2), proving
that additional interactions take place with the aglycone basket.
These could be π-π interactions between the chlorine-substituted
benzene ring of the basket and the fluorine-substituted benzene
ring of A7 or hydrogen-bond interactions between the acetylated
amine and an amido group in the basket. Similar observations
N ) 5.54(t / W0.5h)
r
r
in which t is the peak retention time and W0.5h is the peak width
at half-height. This equation is exact only if the peak is perfectly
symmetric. The obtained efficiencies are in the thousand plate
range, which is low for a modern 25 cm HPLC column. These
values are 10-40 times lower than the efficiencies of these
columns when new and tested with a n-hexane/ chloroform normal
phase (∼40 000 plates/ m). Since the chromatographic system was
not optimized for band-broadening minimization, our goal is not
to give the exact values of the peak efficiencies but to show that
Analytical Chemistry, Vol. 72, No. 8, April 15, 2000 1779

to a mere 350 plates for the second, a 7-fold lower value. The
mass transfer of the more retained enantiomer is much slower
than that of the first eluting enantiomer. The interaction between
the D-amino acids is thermodynamically strong, producing high
retention times and slow adsorption-desorption kinetics, which
result in poor mass transfer and kinetics. This trend is observed
to a much lesser extent for the non amino acid solutes (Table 3
and Figure 5). This phenomenon was described previously for
5
the teicoplanin CSP. Indeed it can even be used to distinguish
between
D- and L-amino acids when no standards are available.
CONCLUSION
It is clear that the carbohydrate units on teicoplanin are not
needed for the enantioresolution of the common R-amino acids.
The cleft near the amine end (or ureido groupsif attached to a
linkage chain) of the aglycone basket is an important part of the
receptor site for amino acid recognition. Amino acids appear to
have an easier access to this site on the aglycone CSP, which
produces much higher enantioselectivities and resolution factors
for these compounds compared to those obtained on the native
teicoplanin CSP. Non amino acid compounds that have carboxylate
groups also can associate with the aglycone amino acid binding
site. Sometimes, these analytes are better resolved by the aglycone
CSP. However, many non amino acid compounds were enantio-
resolved by a combination of interactions involving both the
aglycone basket and its attached sugar units. These compounds
are better resolved, or resolved only, on the native teicoplanin
CSP. They are often neutral compounds or amine-containing
compounds lacking an anionic group that could enhance their
ability to bind at the amino acid site of the teicoplanin “basket”.
Fig u re 5 . Chromatograms of 4-hydroxymandelic acid on the native
teicoplanin CSP (top) and the aglycone CSP (bottom). Conditions:
mobile phase 85% methanol/15% water, pH 4.1 with TEA + acetic
acid; 1 mL/min; UV detection at 254 nm; T ) 22 °C.
ACKNOWLEDGMENT
Support from the National Institutes of Health (Grant NIH R01
GM53825-04), the Ministero dell’Universit a` e della Ricerca Sci-
entifica e Tecnologia, Italy, and the Consiglio Nazionale delle
Ricerche, Italy, is gratefully acknowledged. The authors are
indebted to the Lepetit Reserch Center (Gerenzano, Italy) for the
generous gift of teicoplanin.
there is often a significant difference in efficiency between the
first eluting enantiomer and the second (Figures 3 and 5). For
D
,L-DOPA and the 60% methanol mobile phase (A2 and Figure
3
), the efficiencies on the teicoplanin CSP were 1300 plates for
the first eluting enantiomer and 800 plates for the second eluting
enantiomer, a 38% drop. This trend was even more pronounced
for the aglycone CSP, with 1500 and 600 plates for the first and
second enantiomers, respectively, a 60% drop. It was even lower
with the more polar 20% methanol mobile phase, where the
efficiencies were 2200 plates for the first eluting peak and dropped
Received for review August 31, 1999. Accepted December
7, 1999.
AC991004T
1780 Analytical Chemistry, Vol. 72, No. 8, April 15, 2000