High-performance liquid chromatographic enantiomer separation and determination of absolute configurations of phosphinic acid analogues of dipeptides and their α-aminophosphinic acid precursors

Article abstract of DOI:10.1016/S0957-4166(03)00537-8

The enantiomers of N-benzyloxycarbonyl-phosphinic pseudodipeptides and their N-benzyloxycarbonyl-α-aminophosphinic acid precursors as well as various other structural analogues were separated on a set of cinchona alkaloid-derived chiral anion-exchangers b

Full text of DOI:10.1016/S0957-4166(03)00537-8

TETRAHEDRON:
ASYMMETRY
Pergamon
Tetrahedron: Asymmetry 14 (2003) 2557–2565
High-performance liquid chromatographic enantiomer separation
and determination of absolute configurations of phosphinic acid
analogues of dipeptides and their a-aminophosphinic acid
precursors
Michael La¨mmerhofer,a,* Dieter Hebenstreit,a Elena Gavioli,a Wolfgang Lindner,a Artur Mucha,b
Paweł Kafarskib and Piotr Wieczorekc
aChristian Doppler Laboratory for Molecular Recognition Materials, Institute of Analytical Chemistry, University of Vienna,
Wa¨hringerstrasse 38, A-1090 Vienna, Austria
bInstitute of Organic Chemistry, Biochemistry, and Biotechnology, Wrocław University of Technology,
Wybrzez; e Wyspian´skiego 27, 50-370 Wrocław, Poland
cInstitute of Chemistry, University of Opole, ul. Oleska 48, PL-45 052 Opole, Poland
Received 9 May 2003; accepted 4 July 2003
Abstract—The enantiomers of N-benzyloxycarbonyl-phosphinic pseudodipeptides and their N-benzyloxycarbonyl-a-aminophos-
phinic acid precursors as well as various other structural analogues were separated on a set of cinchona alkaloid-derived chiral
anion-exchangers by HPLC in the reversed-phase mode. Semi-preparative scale chromatography provided single enantiomers in
100 mg quantities. The configurations of the enantiomers were assigned indirectly by enantioselective chromatography on the basis
of the elution order and was confirmed by enantiomeric reference compounds.
© 2003 Elsevier Ltd. All rights reserved.
1. Introduction
phinic acid moiety. Many of the members in this com-
pound class are effective enzyme inhibitors, e.g. of
Zn–metalloproteases, aspartic acid proteinases, serine
proteinases and many others.1 It is to be expected that
the activity is not equal for the individual stereoiso-
Phosphinic acid peptides, also called phosphinic pseu-
dopeptides, are a class of compounds, which have a
peptide bond substituted by a non-hydrolysable phos-
Figure 1. Structures of (a) N-benzyloxycarbonyl-protected phosphinic acid pseudodipeptides 1–8 and (b) their precursors 9–12 as
well as some structurally related reference compounds 13–15.
* Corresponding author. Tel.: +43-1-4277-52323; fax: +43-1-4277-9523; e-mail: michael.laemmerhofer@univie.ac.at
0957-4166/$ - see front matter © 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0957-4166(03)00537-8

2558
M. La¨mmerhofer et al. / Tetrahedron: Asymmetry 14 (2003) 2557–2565
mers, but resides preferentially in one of them. The
study of the stereospecificity of action of those com-
pounds depends on methods that allow the preparative
production of the individual stereoisomers in sufficient
amount and stereochemical purity. In addition,
stereoselective analytical assays are required for the
determination of their enantiomeric purity as well as for
stereoselective bioanalytical investigations.
peptides N- and C-terminal amino acids with an (S)-
configuration and two other configurations either (R)
or (S)-have been separated by conventional CE without
any chiral selector, for the purpose of determining the
dissociation constants.5,6
2. Results and discussion
Herein we report a stereoselective liquid chromato-
graphic method for phosphinic pseudodipeptides
employing cinchona-alkaloid derived chiral stationary
phases. Since the target compounds leave the synthesis
protocol as N-benzyloxycarbonyl-derivatives (Fig. 1a),
these derivatives can be used directly without deprotec-
tion of either the N-terminal or C-terminal protecting
group, although other N-derivatives, e.g. N-9-fluorenyl-
methoxycarbonyl (FMOC), N-3,5-dinitrobenzyloxycar-
bonyl or 2,4-dinitrophenyl derivatives are supposed to
provide much better enantioselectivities than the Z-pro-
tecting group. The developed enantioselective chro-
matographic methods allowed us to separate the
enantiomers of the target N-benzyloxycarbonyl-deriva-
tives of phosphinic pseudopeptide esters 1–4 as well as
the corresponding hydrolyzed compounds with free C-
terminal carboxylic groups 5–8, along with their
aminophosphinic acid precursors 9–12 (Fig. 1). Fur-
thermore, some other structurally related analogues
13–15 have been investigated as reference compounds
(Fig. 1). For biological activity tests, the enantiomers of
Z-Leuc(PO2HCH2)GlyOMe 1 and Z-Phgc(PO2HCH2)-
GlyOMe 2 were prepared in 100 mg quantities by
semi-preparative HPLC in the batch mode. The same
chromatographic separation system was used for deter-
mining the enantiomeric purity of the prepared Z-phos-
phinic pseudodipeptide ester enantiomers. This also
allowed us to indirectly assign the absolute configura-
tions of the individual enantiomers, which was then
verified with a sample obtained by a stereospecific
synthesis. Thus, the proposed methods provide also a
means for the control of the quality and effectiveness of
stereoselective synthesis concepts of such chiral entities.
2.1. Separation of phosphinic pseudodipeptide ester
derivatives (monovalent acids)
The phosphinic acid group of the pseudo-dipeptides has
a pKa of ꢀ1.8.5,6 It is dissociated at the mobile phase
pH 5.6 and is thus negatively charged, whilst the quinu-
clidine of the cinchonan carbamate selectors (Fig. 2) is
protonated and positively charged. Accordingly, an
ion-exchange process can be established in which the
quinuclidinium ion of the chiral selectors represents the
fixed positive charge of the anion-exchange type CSP
that interacts with the oppositely ionized solute by
strong ionic interactions. This primary electrostatic
interaction, which is active over a long distance and
attracts the solute towards the ion-exchange binding
site, is located in the center of a cleft-like hydrophobic
binding pocket that is pre-formed by a number of
lipophilic and bulky moieties, viz. planar quinoline
ring, bulky tert-butyl residue at the carbamate group,
and the rigid quinuclidine ring itself (Fig. 3a).7 The
polar interaction sites in the center of the binding site
are thus embedded in a hydrophobic surrounding
which can strengthen the electrostatics of the selector-
analyte interactions.
N-Derived amino acids, e.g. 3,5-dinitrobenzoylated
leucine, have been shown to fit favorably into this cleft
of the binding pocket featuring besides H-bond-medi-
ated ionic interaction, hydrogen bonding between the
solute’s amide and the selector’s carbamate, as well as
face-to-face p-p-interaction (Fig. 3a). The herein inves-
tigated pseudodipeptide derivatives may bind to the
cinchonan carbamate by essentially the same binding
mechanism (Fig. 3b) and can therefore also be sepa-
rated into the enantiomers on the investigated cincho-
nan carbamate CSPs (see Table 1).
The presented methodologies should be of interest,
since reports about enantiomeric separation of chiral
aminophosphinic acids with free acidic functionality are
scarce. Only a few separations of phosphinic acids and
their derivatives have been published, e.g. CE with
cyclodextrins as chiral additives to the background
electrolyte.2–4 Diastereomers of phosphinic pseudotetra-
The additional P-alkyl group on the phosphorus atom
of the phosphinic acid, as compared to carboxylic,
phosphonic, or sulfonic acid analogues, however,
Figure 2. Structures of the investigated chiral anion-exchange type cinchonan-based chiral stationary phases.

M. La¨mmerhofer et al. / Tetrahedron: Asymmetry 14 (2003) 2557–2565
2559
Figure 3. X-Ray crystal structure (a) and tentative binding model (b) of the selector-analyte ion-pair with the favorable
stereochemistry for tight binding. (a) O-9-(b-Chloro-tert-butylcarbamoyl)quinidine in complex with (R)-3,5-dinitrobenzoyl
leucine,7 and (b) O-9-(tert-butylcarbamoyl)quinidine in complex with (S)-Z-Phgc(PO2HCH2)GlyOMe, (S)-2.
Table 1. Chromatographic data of phosphinic pseudopeptides obtained with different anion exchange type cinchonan-based
chiral stationary phasesa
Compound
k1
h
RS
e.o.b
k1
h
RS
e.o.b
CSP 1, O-9-(tert-butylcarbamoyl) quinidine
CSP 2, O-9-(tert-butylcarbamoyl) quinine
1
2
3
4
5
6
7
8
3.22
7.07
4.11
4.79
4.07
17.02
7.69
11.74
1.15
1.17
1.00
1.12
1.00
1.08
1.11
1.00
1.76
2.18
–
1.48
–
0.95
1.14
–
(R)-(−)
(R)-(+)
–
(R)-(−)
–
(R)-(+)
(S)-(+)
–
1.91
5.62
3.21
4.76
8.70
18.53
11.32
13.57
1.08
1.18
1.00
1.03
1.00
1.13
1.11
1.00
1.14
2.53
–
0.41
–
1.63
1.22
–
(S)-(+)
(S)-(−)
–
(S)-(+)
–
(S)-(−)
(R)-(−)
–
CSP 3, O-9-(2,6-diisopropylphenylcarbamoyl) quinidine
CSP 4, O-9-(2,6-diisopropylphenylcarbamoyl) quinine
1
2
3
4
5
6
7
8
1.59
3.34
1.98
2.60
4.82
11.70
5.81
8.56
1.08
1.05
1.32
1.06
1.09
1.09
1.36
1.05
<0.5
<0.5
2.40
<0.5
0.84
<0.5
2.69
<0.5
(R)-(−)
(R)-(+)
(R)-(−)
(R)-(−)
(R)-(−)
(R)-(+)
(R)-(−)
(R)-(−)
3.42
6.47
4.15
5.56
5.39
11.81
5.79
9.10
1.06
1.14
1.20
1.00
1.00
1.15
1.47
1.00
0.62
1.49
1.89
–
(S)-(+)
(S)-(−)
(S)-(+)
–
–
–
1.95
5.62
–
(S)-(−)
(S)-(+)
–
a Column dimensions, 150×4 mm I.D.; mobile phase, methanol–50 mM sodium phosphate buffer (80:20; v/v) (pHa 5.6); temperature, 40°C; flow
rate, 1 mL/min; UV detection, 250 nm.
b e.o., elution order, i.e. assigned configuration of the first eluted enantiomer.
appears to put larger steric demands on the size of the
binding pocket which is to a great extent defined in the
relatively rigid selector molecule. The additional volu-
minous residue fills the empty space in the binding
pocket of the complex meaning it can more easily
collide with other bulky groups with the result being
detrimental for effective molecular recognition and chi-
ral separation. Indeed, both retention factors of the
second eluted enantiomer, i.e. the high-affinity enan-
tiomer, as well as the enantioselectivity constantly
decreased in the order H>CH3>C2H5 as the X-sub-
stituent within the series of comparable Z-PhgPO2H-X
homologues (see Fig. 1b; with R1 being a phenyl
group). For example, on the tert-butylcarbamoyl-
quinidine CSP (Chiral AX QD-1, CSP 1) the reten-
tion factors k2 diminished from 8.51 for 10 over 7.34
for 13 to 7.10 for 14, while enantioselectivity values h
declined in the same order from 1.28 over 1.22 to 1.11
(Table 2). The corresponding target phosphinic pseu-
dodipeptide analogue
2
with X=2-(methoxycar-
bonyl)ethyl was a little better separated with h=1.17
(k1=7.07, k2=8.29, RS=2.18) than the ethyl group-
bearing analogue 14. On the other hand, the replace-
ment of H by OH as the X-substituent (Fig. 1b) led to
the bis-protic phosphonic acid congener 15, which in
accordance with the ion-exchange mechanism showed a
stronger retention (k1=12.43) than the phosphinic acid
10 (k1=6.65) with minor effect on enantioselectivity

2560
M. La¨mmerhofer et al. / Tetrahedron: Asymmetry 14 (2003) 2557–2565
h=1.30 (RS=2.63) when compared to 10 (h=1.28,
RS=3.23) (Table 2). Similar trends were also observed
on the other investigated CSPs.
ing and enantioselectivity. Besides different solvation
effects, the aminophosphinic acid residue can come into
contact with the bulky carbamate residues and interact
with this moiety by steric and/or Van der Waals type
interactions. Alternatively, in the hydro-organic envi-
ronment this binding contribution can also be inter-
preted as a hydrophobic interaction with significant
effect on observed enantioselectivities. For example, for
both the tert-butylcarbamoylated CSPs, the h values
Although in the tentative binding models (Fig. 3b) it
appears that the alkyl, arylalkyl and phenyl groups,
respectively, at the stereogenic center of the Z-protected
phosphinic pseudodipeptide esters point towards free
space, they make a significant contribution to the bind-
Table 2. Chromatographic data of phosphinic acid precursors and structurally related compounds obtained with different
anion-exchange type cinchonan-based chiral stationary phasesa
Compound
k1
h
RS
e.o.b
k1
h
RS
e.o.b
CSP 1, O-9-(tert-butylcarbamoyl) quinidine
CSP 2, O-9-(tert-butylcarbamoyl) quinine
9
3.64
6.65
4.96
5.56
6.02
6.40
12.43
3.16
7.00
5.82
1.19
1.28
1.04
1.18
1.22
1.11
1.30
1.25
1.14
1.24
2.15
3.23
0.5
2.21
2.10
1.15
2.63
2.62
1.46
2.83
(R)-(−)
(R)-(+)
(R)-(−)
(R)-(−)
(R)-(+)
(R)-(+)
(R)-(+)
(S)-(−)
(S)-(+)
(S)-(+)
1.86
6.26
4.57
5.37
5.63
6.17
12.37
4.28
7.92
7.17
1.14
1.23
1.00
1.10
1.21
1.14
1.23
1.19
1.13
1.15
1.36
3.40
1.00
1.33
2.81
1.90
2.77
2.86
1.91
2.41
(S)-(+)
(S)-(−)
–
(S)-(+)
(S)-(−)
(S)-(−)
(S)-(−)
(R)-(+)
(R)-(−)
(R)-(−)
10
11
12
13
14
15
16c
17c
18c
CSP 3, O-9-(2,6-diisopropylphenylcarbamoyl) quinidine
CSP 4, O-9-(2,6-diisopropylphenylcarbamoyl) quinine
9
2.60
4.69
3.51
4.30
4.40
4.90
7.03
3.63
7.43
6.28
1.21
1.25
1.46
1.23
1.13
1.04
1.27
1.26
1.03
1.45
2.51
3.22
4.75
2.55
1.76
<0.5
3.27
3.43
<0.5
4.62
(R)-(−)
(R)-(+)
(R)-(−)
(R)-(−)
(R)-(+)
(R)-(+)
(R)-(+)
(S)-(−)
(S)-(+)
(S)-(+)
3.53
6.31
5.00
6.43
6.37
7.21
10.35
3.58
8.71
6.81
1.18
1.27
1.38
1.16
1.18
1.14
1.26
1.22
1.08
1.39
2.19
3.04
4.76
2.02
2.06
1.58
3.12
2.63
0.97
4.69
(S)-(+)
(S)-(−)
(S)-(+)
(S)-(+)
(S)-(−)
(S)-(−)
(S)-(−)
(R)-(+)
(R)-(−)
(R)-(−)
10
11
12
13
14
15
16c
17c
18c
a Column dimensions, 150×4 mm I.D.; mobile phase, methanol–50 mM sodium phosphate buffer (80:20; v/v) (pHa 5.6); temperature, 40°C; flow
rate, 1 mL/min; UV detection, 250 nm.
b e.o., elution order; assigned configuration of the first eluted enantiomer.
c 16, N-benzyloxycarbonyl-leucine; 17, N-benzyloxycarbonyl-phenylglycine; 18, N-benzyloxycarbonyl-phenylalanine.
Figure 4. Effect of the side chain residue R1 on enantioselectivity h of various phosphinic acid derivatives and carboxylic acid
analogues on CSP 1 (a) and CSP 3 (b). (1) Y=-PO2H(CH2)2COOCH3, (2) Y=-PO2H2, (3) Y=-PO2H(CH2)2COOH, and (4)
Y=-COOH. R1: iBu=isobutyl, Ph=phenyl, Bz=benzyl, Phet=2-phenethyl. The separation factor h is here defined as k(S)/k(R)
yielding a value below 1 for compound 7 with Bz residue and Y=-PO2H(CH2)2COOH on CSP 1, which indicates reversed elution
order compared to the other compounds of the series (vide supra). For the experimental conditions see Tables 1 and 2.

M. La¨mmerhofer et al. / Tetrahedron: Asymmetry 14 (2003) 2557–2565
2561
decrease in the order phenyl>isobutyl>2-phenylethyl>
benzyl (see also Fig. 4a, curve 1). Most remarkably the
latter, i.e. the benzyl congener, which had not been
separated under the given conditions on either of the
both tert-butyl carbamate phases, is the solute yielding
the highest enantioselectivity value on both the diiso-
propylphenyl carbamate CSPs (Fig. 4b, curve 1). This is
a typical observation for bis-aromatic compounds such
as N-benzyloxycarbonyl and N-benzoyl a-amino acids
with a benzyl side chain on the diisopropylphenyl cin-
chonan carbamate CSPs also featuring two aromatic
binding sites. While the h values of the amino acids
with aliphatic residues are substantially reduced on the
diisopropylphenyl phase when compared to the tert-
butyl CSP analogue, enantioselectivity, e.g. bis-aro-
matic Z-Phe or Bz-Phe, is enhanced. This can be
explained by a spatial fit of the aromatic moieties for a
simultaneous p-p-interaction between the quinoline ring
and the Z-group, as well as the diisopropylphenyl
residue and benzyl side chain. Improper spatial require-
ments for simultaneous bi-valent p-p-interaction of the
aromatic moieties in the corresponding phenyl and
2-phenethyl-congeners in contrast led to a substantial
loss in the chiral recognition ability of the diisopropy-
lphenyl cinchonan carbamate selectors for these two
solutes. Here it is noteworthy that on both the tert-
butyl and 2,6-diisopropylphenyl carbamate CSPs, the
entire series of the phosphinic acid precursors (Fig. 4a
and b, curve 2) yielded higher enantioselectivities than
the corresponding phosphinic pseudodipeptide ester
congeners (curve 1).
considerably better resolved on the 2,6-diisopropyl-
phenyl carbamoyl quinidine CSP (3, h=1.32; RS=2.4)
(Table 1). Remarkably, only the 2,6-diisopropylphenyl
carbamoyl quinidine CSP revealed an enantiomer sepa-
ration capability for all 4 ester derivatives 1–4, although
with exception of compound 3 only at low levels (RS
typically <0.5). This clearly agrees with our earlier
findings that the cinchonan carbamate CSPs with the
2,6-diisopropylphenyl residue show broader enantiose-
lectivity than the corresponding tert-butyl analogues
which however afford higher enantioselectivity for
many classes of compounds. Representative chro-
matograms of the phosphinic pseudodipeptide ester
derivatives are depicted in Figure 5 and of the corre-
sponding aminophosphinic acid precursors in Figure 6.
The chromatographic results for the phosphonic pseu-
dopeptides in Table 1 as well as those for the precursors
and analogues in Table 2 moreover largely reflect the
expected pseudo-enantiomeric behavior of quinidine
and quinine carbamate selectors as indicated by the
reversal of the elution order on the corresponding
Chiral AX QD (CSP 1 and 3) and QN CSPs (CSP 2
and 4). However, it became quite evident that the
corresponding quinidine and quinine selectors are not
true enantiomers, but actually diastereomers (opposite
configurations at C8 and C9, but identical configura-
tions at the stereogenic centers at N1, C3, and C4). The
a values of the quinidine and quinine CSP counter-
parts, though exhibiting mostly similar levels of enan-
tioselectivity, are not equal as would be expected for
real enantiomers on the CSPs with similar selector
loadings. The selectivity differences between pseudo-
enantiomeric counterparts once more underlines the
stringent demands for the steric and functional fit of the
binding geometries of the selectors.
Compounds 1 and 4 showed the highest degree of
resolution on the tert-butyl carbamoyl quinidine CSP
(1, h=1.15; RS=1.8 and 4, h=1.12; RS=1.5), com-
pound 2 the best enantiomer separation on the tert-
butyl carbamoyl quinine CSP (2, h=1.18; RS=2.5 but
also sufficient resolution on the corresponding pseudo-
enantiomeric quinidine analogue CSP 1; 2, h=1.17;
RS=2.2), while compound 3 on the other hand was
On the basis of our binding model for N-carbonylated
amino acids shown in Figure 3, which is not only
backed by X-ray crystal structures of selector-analyte
Figure 5. Chromatograms of the direct HPLC enantiomer separation of compounds 1 (a), 2 (b), and 4 (c) on O-9-(tert-butylcar-
bamoyl) quinidine based CSP 1, and (d) compound 3 on O-9-(2,6-diisopropylphenylcarbamoyl) quinine based CSP 4. For the
experimental conditions, see Table 1.

2562
M. La¨mmerhofer et al. / Tetrahedron: Asymmetry 14 (2003) 2557–2565
co-crystallisates.7 but also by solution-phase NMR as
well as molecular modeling experiments (molecular
dynamic simulations) on selector–analyte complexes,8
the absolute configuration of the first eluted enantiomer
of the Z-aminophosphinic acid derivatives was chro-
matographically assigned to be (R) on the quinidine
phase, taking into account the change in CIP priorities
caused by the phosphor atom. As a tacit assumption, a
homology principle between the N-carbonylated amino
acids like dinitrobenzoyl or benzyloxycarbonyl amino
acids and herein investigated N-benzyloxycarbonylated
aminophosphinic acids was thereby assumed. More-
over, it was supposed that the overall primary binding
and chiral recognition mechanism does not change
significantly when compared to the N-carbonylated
amino acids derivatives so that the tentative binding
model shown in Figure 3b still holds. Thus, it can be
seen that the (S)-enantiomer of 2 fits very well with the
favorable binding requirements and is therefore able to
exhibit a higher affinity towards the quinidine carba-
mate selector. It should hence be eluted second. To
2.2. Separation of phosphinic pseudodipeptides with free
carboxylic acid at the C-terminus (bis-acidic
compounds)
By hydrolysis of the esters 1–4 the bis-acidic phosphinic
pseudodipeptides 5–8 with free C-terminal carboxylic
group are obtained. A second acidic group in the
analyte molecules has implications for the separation
mechanism, as it can compete with the phosphinic acid
group for binding at the fixed ionic site of the CSPs,
and thus scramble the above discussed chiral recogni-
tion mechanism. However, the phosphinic acid group is
more acidic and so is preferred as the negatively ionized
binding site. Also the existence of topologically favor-
able additional interactions such as the H-bond
between analytes’ carbamate NꢀH and the selector’s
carbamate carbonyl can be used as strong arguments in
favor of adherence to the discussed binding mechanism
even in the hydrolyzed analytes. Accordingly, the phos-
phinic acid group is supposed to dominate the chiral
recognition mechanism, while the additional carboxylic
group merely adds a significant retention increment.
verify
this
the
(R)-enantiomer
of
Z-
Phgc(PO2HCH2)GlyOMe 2 was prepared by stereose-
lective synthesis. As predicted the (R)-enantiomer
eluted first on the quinidine CSP, whilst the (S)-enan-
tiomer showed a higher affinity to the selector confirm-
ing the validity of the chromatographic assignment of
the absolute configurations of aminophosphinic acids,
as previously demonstrated for amino phosphonic
acids.9 The change of the specific rotation within the
series for the Z-phosphinic pseudodipeptide ester with a
phenyl group as the R1 residue (compound 2) origi-
nated from structure (chromophore) induced reversal of
the sign of rotation and does not reflect a reversal in the
elution order, as could have been argued before verifi-
cation with the enantiomerically pure 2. Moreover, the
results with the enantiomeric precursors, viz. (R)-(−)-9,
(R)-(+)-10, and (R)-(−)-11 are fully in line with the
above assignment (see Table 2) and also confirm the
correctness of the chromatographic determination of
the absolute configurations for the investigated
aminophosphinic acid derivatives.
The chromatographic data presented in Table 1 under-
pins this hypothesis. There is only one exception to this
general rule within the series: Compound 7, which
features a benzyl side chain residue, shows a reversed
elution order on both of the tert-butyl carbamoylated
CSPs (CSP 1 and CSP 2), but not on the 2,6-diiso-
propylphenyl carbamoylated analogues (CSP 3 and
CSP 4). This can be explained by a more favorable
arrangement of the solute upon primary ionic interac-
tion with the carboxylic group, supported by additional
other interactions such as p-p-interaction with the aro-
matic group of the side chain instead of the phenyl ring
of the protection group. The fact that both pairs of
interaction sites, i.e. the phosphinic acid group and the
phenyl of the Z-protection group as well as the C-termi-
nal carboxylic group and phenyl of the side chain, are
separated by five atoms and thus topologically very
similar, can be invoked to indicate a competitive bind-
ing mode. Since single enantiomers of 7 or 3 were not
Figure 6. Chromatograms of the direct HPLC enantiomer separation of compounds 9 (a) and 10 (b) on O-9-(tert-butylcarbamoyl)
quinidine based CSP 1, and of compounds 11 (c) and 12 on O-9-(2,6-diisopropylphenylcarbamoyl) quinidine based CSP 3. For
the experimental conditions, see Table 1.

M. La¨mmerhofer et al. / Tetrahedron: Asymmetry 14 (2003) 2557–2565
2563
available, the confirmation of the reversed elution order
was carried out chromatographically. Compound 3 was
separated chromatographically on a semi-preparative
scale on CSP 4 providing single enantiomers of 3. The
first eluted enantiomer was assigned to be the (S)-(+)-
enantiomer and was after hydrolysis to (S)-(+)-7 with 1
M LiOH (ca. 3 h at 60°C) re-injected on CSP 2, with a
racemic sample as a reference. Thus it was clearly seen
that the (S)-(+)-enantiomer eluted after the (R)-(−)-
enantiomer of 7 on CSP 2, which means that the
elution order was reversed on both of the tert-butyl
carbamate phases when compared to all the other
resolved phosphinic pseudodipeptide esters and di-acid
solutes of the series due to different binding and chiral
recognition mechanisms.
ring as the side chain also gave the highest resolution
on CSP 4, but was separated with a significantly lower
separation factor (h=1.15) and resolution (RS=1.95),
while compounds 5 and 8 with the isobutyl and 2-
phenylethyl residue, respectively, were only poorly sep-
arated under the given conditions and solely on CSP 3
(e.g. 5, h=1.09 and RS=0.84; 8, h=1.05 and RS <0.5).
In this context it must be noted, however, that the
conditions have not been specifically optimized for the
present compounds. On the contrary, standard condi-
tions as established previously for aminophosphonic
acid derivatives have been utilized throughout.9 Hence,
better separations could be obtained, in particular for
the bis-acidic compounds after a systematic optimiza-
tion of the major influential experimental variables such
as pH, eluent modifier and type, temperature, and flow
rate. However this would have been beyond the focus
of the present investigation.
In any case, the retention factors of the bis-acidic
compounds 5–8 were increased on all CSPs compared
to the corresponding ester analogues (Table 1), which is
very much in accordance with the primary anion-
exchange retention mechanism. With the exception of 7,
all the compounds of the series with the hydrolyzed
ester group (Fig. 4, curve 3) are separated with lower h
values than the corresponding esters (Fig. 4, curve 1) on
CSP 1 and CSP 2. In contrast, on CSP 3 the difference
in enantioselectivities for the ester and carboxylic acid
pseudodipeptides are more or less negligible (Fig. 4b,
curves 1 and 3). Noticeably, on all four CSPs, com-
pound 7, with the benzyl side chain at the stereogenic
center, was resolved with the highest selectivity, e.g.
1.47 (RS=5.62) on the 2,6-diisopropylphenylcarbamoyl
quinine-based CSP 4. Compound 6 with the phenyl
2.3. Semi-preparative enantiomer separation and appli-
cation for enantiomeric excess determination
Compounds 1 and 2 were separated on a semi-prepara-
tive scale on the tert-butylcarbamoylquinidine CSP 1.
Despite high loading (ca. 40 mg) the resolution was still
satisfactory (see Fig. 7). No advanced fraction collec-
tion and peak shaving technology was utilized, but only
two fractions corresponding to the first and second
eluted enantiomer were collected with a cut in the peak
valley. The direct chromatographic enantiomeric purity
control afforded an enantiomeric excess ee of 98% for
the first eluted (R)-enantiomer and 82% for the second
Figure 7. Semi-preparative separation of compound 2 (a) and compound 1 (b) on O-9-(tert-butylcarbamoyl) quinidine based CSP
1 (column dimension, 250×16 mm I.D.) and analytical control of the enantiomeric purity of the individual fractions. For other
conditions see Section 4.

2564
M. La¨mmerhofer et al. / Tetrahedron: Asymmetry 14 (2003) 2557–2565
eluted (S)-enantiomer of Z-Phgc(PO2HCH2)Gly-OMe
2 (see Fig. 7a) as well as an ee of 74% for the first
eluted (R)-enantiomer and 86% for the second eluted
(S)-enantiomer of Z-Leuc(PO2HCH2)Gly-OMe 1 (Fig.
7b). The modest ee values for the second eluted enan-
tiomer of 2 and the first eluted enantiomer of 1 has to
be sought in the tailing and fronting, respectively, of
the peaks during semi-preparative chromatography. If a
higher enantiomeric purity is required, recycling of a
middle fraction is strongly recommended and certainly
allows the production of enantiomers in ee exceeding
98%.
well as their a-aminophosphinic acid precursors. Obvi-
ously, these test compounds largely adhered to a selec-
tor-analyte binding and chiral recognition mechanism
which was previously proposed for N-carbonylated
amino acid derivatives and was thoroughly validated
spectroscopically. Hence, the separation method could
be exploited for the chromatographic determination of
the absolute configurations of the N-derivatized
aminophosphinic acids and phosphinic pseudodipep-
tides on the basis of the tentative binding model previ-
ously proposed for N-carbonylated amino carboxylic
and phosphonic acids. These results have been con-
firmed with enantiomeric samples that were obtained
by stereoselective synthesis. The presented enantioselec-
tive chromatographic method thus proved its usefulness
in analytical and preparative enantiomer separation of
phosphinic acid derivatives and their stereochemical
description as well as enantiomeric purity control.
Figure 8 on the other hand shows the chromatograms
monitored in the course of the enantiomeric purity
determination of the enantiomeric reference compound
(R)-(+)-2 that was prepared by stereoselective synthesis.
It can be seen that the compound has a high ee (ca.
98%). The fact that the impurity is eluted in front of the
main component still allows its correct quantitation.
Also the N-benzyloxycarbonyl-a-aminophosphinic
acids that were used as enantiomeric reference com-
pounds and precursors showed high enantiomeric
excess, namely 98%, >98%, and >99% ee for (R)-(−)-9,
(R)-(−)-11, and (R)-(+)-10, respectively.
4. Experimental
4.1. Materials
The N-benzyloxycarbonyl-a-aminophosphinic acid
pseudodipeptide ester derivatives 1–4 depicted in Figure
1 were synthesized by a Michael addition reaction of
methyl acrylate to the corresponding Z-protected 1-
aminoalkylphosphinic acid preactivated into its triva-
lent silyl ester according to the procedure described.10
This synthesis yielded compounds 1–4 as racemates.
Compound 2 was also synthesized in its enantiomeric
form by the same reaction protocol but starting from
3. Conclusion
The concept of enantioselective anion-exchange
employing quinidine and quinine carbamate CSPs
turned out to be highly suitable for the enantiomer
separation of the target phosphinic pseudopeptides as
Figure 8. Enantiomeric purity determination of (R)-(+)-2 synthesized by a stereoselective method on O-9-(tert-butylcarbamoyl)
quinine based CSP 2, on which the enantiomeric impurity is eluted in front of the main enantiomer peak. For the experimental
conditions, see Table 1.

M. La¨mmerhofer et al. / Tetrahedron: Asymmetry 14 (2003) 2557–2565
2565
the (R)-(+)-enantiomer of N-benzyloxycarbonyl-1-
amino-1-phenylmethylphosphinic acid 10, which was
obtained by fractional crystallization of racemic 10 as
diastereomeric salts with (R)-(+)-1-phenylethylamine as
previously published.11,12 Similarly, the (R)-(−)-enan-
tiomers of 9 and 11 were obtained from the correspond-
ing racemates and (R)-(+)-1-phenylethylamine, and the
racemic N-benzyloxycarbonyl-1-aminoalkylphosphinic
acids 9–12 by a procedure described previously.11 Com-
pounds 13–15 were from a previous study.2 The bis-
acidic phosphinic pseudodipeptide compounds 5–8 were
obtained by hydrolysis of the corresponding Z-phos-
phinic acid pseudodipeptide esters 1–4 with lithium
hydroxide. Thus, 10 mg of the respective compound 1
to 4 were dissolved in 1 mL 1 M lithium hydroxide and
kept at 60°C for 3 h. The HPLC control of the hydrol-
ysis showed the reaction had gone to completion after
2.5 h.
Chiral AX QD-1 column (Bischoff Chromatography,
Leonberg Germany) containing CSP 1, 15 mm particles
(column dimension 250×16 mm I.D.) using the same
mobile phase as for analytical scale separations at a
flow rate of 4.5 mL/min and ambient temperature. The
amount injected onto the column per run was 35 mg of
1 and 45 mg of 2. Both the enantiomers were collected
separately and the pooled product fractions evaporated
to dryness. The residue was dispersed in 1 M hydro-
chloric acid had been dried with the phosphinic dipep-
tide ester extracted with ethyl acetate. After the organic
phase had been dried with sodium sulfate and the
solvent evaporated, a white crystalline product was
obtained after stirring with petroleum ether.
Acknowledgements
The chiral stationary phases, Chiral AX QD-1, QN-1,
QD-2, and QN-2 (Fig. 2) were from Bischoff Chro-
matography (Leonberg, Germany) and were also com-
mercially available from Iris Technologies (Lawrence,
KS, USA). Methanol was of HPLC grade and supplied
by Merck (Darmstadt, Germany). Sodium dihydrogen
phosphate of analytical grade was employed to prepare
the buffer and was also from Merck. The pH of the
mobile phase was adjusted with phosphoric acid thus
representing the apparent pH (pHa) of the hydro-
organic mixture. The mobile phase was filtered through
a 0.2 mm Nylon membrane filter and degassed by
sonication prior to use.
The financial support of the Austrian Christian
Doppler Research Society and the industrial partners
AstraZeneca (Mo¨lndal, Sweden) and Merck (Darm-
stadt, Germany) is gratefully acknowledged.
References
1. Collinsova´, M.; Jira´cek, J. Curr. Med. Chem. 2000, 7,
629–647.
2. Dzygiel, P.; Rudzinska, E.; Wieczorek, P.; Kafarski, P. J.
Chromatogr. A 2000, 895, 301–307.
3. Fischer, C.; Schmidt, U.; Dwars, T.; Oehme, G. J. Chro-
matogr. A 1999, 845, 273–283.
4.2. Instrumentation and HPLC method
4. Rudzinska, E.; Dzygiel, P.; Wieczorek, P.; Kafarski, P. J.
Chromatogr. A 2002, 979, 115–122.
5. Koval, D.; Kasicka, V.; Jiracek, J.; Collinsova, M. Elec-
trophoresis 2003, 24, 774–781.
HPLC experiments were performed with a Hitachi–
Merck HPLC system which consisted of a L-7100
intelligent pump, L-7400 UV–vis detector, L-7200
autosampler, D-7000 interface, HSM 7000 chromatog-
raphy data station software from Merck (Darmstadt,
Germany) and a Jasco OR-990 optical rotation detector
(Jasco, Groß-Umstadt, Germany). The HPLC column
was thermostated at 40°C with a column thermostat
from W.O. Electronics (Langenzersdorf, Austria).
Column dimensions were 150×4 mm I.D. The mobile
phase consisted of methanol–50 mM sodium dihydro-
gen phosphate buffer (80:20; v/v) (pHa 5.6). The flow
rate for the analytical runs was 1 mL/min. The phos-
phinic acid derivatives were dissolved in a mobile phase
at a concentration of ca. 1 mg/mL where 20 mL of the
sample solution were injected. The detection wave-
length was set at 250 nm.
6. Koval, D.; Kasicka, V.; Jira´cek, J.; Collinsova´, M.; Gar-
row, T. A. J. Chromatogr. A 2002, 770, 145–154.
7. Czerwenka, C.; La¨mmerhofer, M.; Maier, N. M.; Ris-
sanen, K.; Lindner, W. Anal. Chem. 2002, 74, 5658–5666.
8. Maier, N. M.; Schefzick, S.; Lombardo, G. M.; Feliz, M.;
Rissanen, K.; Lindner, W.; Lipkowitz, K. B. J. Am.
Chem. Soc. 2002, 124, 8611–8629.
9. Zarbl, E.; La¨mmerhofer, M.; Hammerschmidt, F.; Wug-
genig, F.; Hanbauer, M.; Maier, N. M.; Sajovic, L.;
Lindner, W. Anal. Chim. Acta 2000, 404, 169–177.
10. Vassiliou, S.; Mucha, A.; Cuniasse, P.; Georgiadis, D.;
Lucet-Levannier, K.; Beau, F.; Kannan, R.; Murphy, G.;
Kna¨uper, V.; Rio, M.-C.; Basset, P.; Yiotakis, A.; Dive,
V. J. Med. Chem. 1999, 42, 2610–2620.
11. Baylis, E. K.; Campbell, C. D.; Dingwall, J. G. J. Chem.
Soc., Perkin 1 1984, 2845–2853.
4.3. Semi-preparative chromatography
12. Mucha, A.; Tyka, R.; Sawka-Dobrowolska, W.;
Glowiak, T. Phosphorus, Sulfur and Silicon 1994, 92,
129–138.
Compounds 1 and 2 (see Fig. 1) were separated into the
enantiomers on a semi-preparative scale employing a