NMR in chiral polypeptide liquid crystals: the problem of amines

Article abstract of DOI:10.1016/j.tetasy.2007.06.018

It is shown that lyotropic liquid crystal mixtures made of poly-γ-benzyl-l-glutamate (PBLG) dissolved in N,N-dimethylformamide (DMF) are efficient anisotropic NMR solvents to distinguish the enantiomers of chiral amines through the effects of the differential ordering of enantiomers. This type of solvent overcomes problems often encountered when dissolving amines into the more conventional PBLG/CHCl₃ or PBLG/CH₂Cl₂ liquid crystals. Furthermore, it is shown that perdeuterobenzyl chloride is an excellent achiral deuterated derivatizing agent for enantiomeric excess measurements of chiral amines in conjunction with the PBLG/DMF solvent.

Full text of DOI:10.1016/j.tetasy.2007.06.018

Tetrahedron: Asymmetry 18 (2007) 1511–1516
NMR in chiral polypeptide liquid crystals: the problem of amines
a
b
b
b
Audrey Solgadi, Ludovic Jean, Marie-Claire Lasne, Jacques Rouden,
a
a,
*
Jacques Courtieu and Abdelkrim Meddour
a
Laboratoire de RMN en milieu orient e´ , Univ. Paris-Sud 11, ICMMO UMR CNRS 8182, Rue du Doyen Joseph P e´ r e` s,
Bat 410, 91405 Orsay Cedex, France
b
Laboratoire de Chimie Mol e´ culaire et Thio-organique, ENSICAEN, UMR CNRS 6507, 6, boulevard du Mar e´ chal Juin,
4050 Caen, France
1
Received 23 April 2007; accepted 13 June 2007
Abstract—It is shown that lyotropic liquid crystal mixtures made of poly-c-benzyl-L-glutamate (PBLG) dissolved in N,N-dimethylform-
amide (DMF) are efficient anisotropic NMR solvents to distinguish the enantiomers of chiral amines through the effects of the differential
ordering of enantiomers. This type of solvent overcomes problems often encountered when dissolving amines into the more conventional
PBLG/CHCl or PBLG/CH Cl liquid crystals. Furthermore, it is shown that perdeuterobenzyl chloride is an excellent achiral deuter-
3 2 2
ated derivatizing agent for enantiomeric excess measurements of chiral amines in conjunction with the PBLG/DMF solvent.
Ó 2007 Elsevier Ltd. All rights reserved.
1
. Introduction
for one reason or another, enantiomers cannot be observed
through the observation of a dipole–dipole coupling, one
can then try to observe a chemical shift anisotropic effect
or a deuterium quadrupolar splitting effect, even at a natu-
Over the last few years, we have been developing an origi-
nal NMR method to distinguish and quantify enantiomers.
This method is based on the use of chiral ordering media as
NMR solvents. The basic phenomenon is that two enan-
tiomers dissolved in chiral liquid crystals are not ordered
the same and consequently all the order sensitive NMR
interactions are different, namely, the chemical shift aniso-
tropies, the dipolar couplings and the quadrupolar split-
tings for spins larger than 1/2 such as deuterium. The
advantages of this technique are numerous.
1
ral abundance level. Indeed, there are so many opportuni-
1
,2
ties to observe the effects of the differential ordering of
enantiomers that, statistically over hundreds of examples
treated to date, the method failed to give the expected
answer in less than 1% of the cases.
Fourth, the price of the analysis is quite inexpensive as
quoted in Ref. 5.
First, the method is easy to implement on any spectrometer
and most of the time only requires a simple 1D NMR
experiment, although it may require some skill in the sam-
ple preparation (ꢀ1 h).
Currently, the best results have been obtained using lyotro-
pic chiral liquid crystals made up of organic solutions of
synthetic homo-polypeptides such as poly-(c-benzyl)-L-glu-
tamate (PBLG), poly-(e-carbobenzoxy)-L-lysin (PCBLL)
and poly-(c-ethyl)-L-glutamate (PELG). A large variety
of organic co-solvents can be used such as chloroform,
dichloromethane, 1,1,2,2-tetrachloroethane, tetrahydro-
furan, 1,4-dioxane or N,N-dimethylformamide (DMF).
The only condition is to use a helicogenic solvent in order
to preserve the a-helical conformation of the polypeptide
backbone, an essential condition in order to obtain a
homogeneous liquid crystalline solution.
1
Second, the method is very efficient as it works with all
types of compounds or chemical functions including
3
hydrocarbons.
Third, the chiral discrimination is not only seen through a
4
chemical shift process like conventional methods. It can be
observed through the measurement of any of the above-
mentioned anisotropic interactions. This means that if,
Amongst the collaborations with various chemical labora-
tories working on asymmetric synthesis, we have recently
been involved with a group working on chiral amines.
6
,7
*
0
Corresponding author. E-mail: ameddour@icmo.u-psud.fr
957-4166/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetasy.2007.06.018

1512
A. Solgadi et al. / Tetrahedron: Asymmetry 18 (2007) 1511–1516
On this occasion we encountered an uncommon problem
that made our technique inefficient. Owing to the impor-
tance of amines, noticeably in pharmaceutical industries,
herein we report a description of the problem encountered
and the solutions investigated to solve it.
2
. Results and discussion
2
.1. Problem
Lasne et al. are involved in the asymmetric synthesis of N-
arylaminopyrrolidines in the research of new ligands of
6
serotoninergic receptors and of new asymmetric reactions
7
using chiral palladium organometallic-amine complexes.
Over the course of arylation of N-phenyl-3-aminopyrrol-
idine, diamine 1 was obtained (Fig. 1). The absence of
any reaction of this amine with MTPA prevented the use
of the Mosher’s method to separate the enantiomers. Chi-
ral capillary electrophoresis or chiral HPLC were unsuc-
cessful neither. Consequently we decided to try the chiral
liquid crystal technique.
Owing to the high sensitivity of quadrupolar splitting to the
differential ordering of enantiomers, we decided to work
first with a deuterated isotopomer of 1, 1d , which can
5
be easily synthesized (Fig. 1). The proton decoupled deute-
2
1
rium ( H–{ H}) NMR spectrum of 1d dissolved in PBLG/
5
CHCl is reported in Figure 2a.
3
D
D
D
D
HN
HN
D
2
1
Figure 2. H–{ H} NMR spectra of racemic 1d
at T = 310 K. a: PBLG/CHCl (PBLG weight fraction X
PBLG/CHCl /DMF (100.0/407.3/30.5; X = 18.6%); c: PBLG/CHCl
DMF (100.0/404.6/61.5; X = 17.7%).
5
in PBLG liquid crystals
3
M
= 19.5%); b:
3
M
3
/
N
N
M
1
1d5
not enough to observe a clear chiral discrimination. We
then decided to add a small amount of DMF into the
Figure 1.
PBLG/CHCl liquid crystal as this aprotic polar solvent
3
is known to have a strong dissociative influence on electro-
static intermolecular associations. Furthermore, it is a
good solvent for PBLG. Spectra b and c in Figure 2 were
obtained by adding 7% and 13%, respectively, of DMF in
the chloroform co-solvent of PBLG. Clearly the lines
appear thinner as the DMF proportion in the co-solvent in-
creases. At the same time, the molecular order parameters
decrease with the amount of DMF as can be seen through
the general lowering of the quadrupolar splittings. This last
phenomenon supports the idea of an association to the
PBLG fibres that is lowered by the addition of DMF to
This spectrum exhibits anomalous wide lines, with line-
width Dm1/2 = 150–200 Hz, and no chiral discrimination
could be observed. In several other occasions working with
amines with the same solvent we came across this type of
problem of wide lines whose origin were still unclear. At
first we thought this problem was the result of some type
of exchange phenomenon. However, when heating or cool-
ing the sample, no coalescence process was observed. To
date we believe that the amines interact strongly with the
polypeptide fibres, maybe through hydrogen bonds. Conse-
quently, the molecular motion of the amine may be
the CHCl co-solvent.
3
strongly affected and whenever the motion becomes too
0
slow, T s are shortened to values where high resolution is
The decrease in the NMR linewidth induced by the DMF
led us to the conclusion that the latter solvent could be
the ideal PBLG co-solvent to study chiral amines with this
technique, despite the fact that PBLG/DMF is not the eas-
iest liquid crystal to work with, since PBLG/DMF forms a
gel at room temperature. Consequently, both the impor-
tant steps of PBLG/DMF/chiral solute sample preparation
2
lost due to the efficiency of the quadrupolar relaxation
process.
Believing that the problem could originate from the poly-
peptidic fibres we used a PCBLL/CHCl liquid crystal sol-
vent. The linewidth was narrower than in Figure 2a, but
3

A. Solgadi et al. / Tetrahedron: Asymmetry 18 (2007) 1511–1516
1513
nances allows evaluating the ee as 77%. Straightforward
calculations lead to a minimum value ee >97% in B sample.
In all these spectra, very small peaks are observed, which
are due to the natural abundance deuterium signals of
DMF. These may be used as an internal reference if needed.
2.2. Generalization of the method
The above result was first obtained owing to the convenient
properties of the PBLG/DMF liquid crystal towards the
amine orientation, but also to the simple introduction of
the deuterated moiety over the course of the synthesis of
1d5. The latter point cannot be made general for every
amine synthesis and consequently, other techniques must
be tested in order to demonstrate the efficiency and the gen-
erality of the PBLG/DMF anisotropic solvent in the enan-
tiomeric quantification of chiral amines. Amongst the
various possibilities derived from the above result we ex-
plored the idea of using perdeuterobenzyl chloride 2 as
an achiral deuterated derivatizing agent to introduce deute-
rium nuclei into the molecule to be studied.
The idea of achiral deuterated derivatizing agents (ADDA)
for alcohols acids, amines or amino acids is not new. The
advantages of this strategy are:
1
2
. There is no need to be concerned about the enantiomeric
purity of the derivatizing agent since this one is not
chiral.
. There is no need to be concerned about the completeness
of the reaction between the achiral derivatizing agent
and the enantiomers to be studied since no kinetic reso-
lution can take place.
2
1
3. The derivatizing agent is deuterated only to bring the
deuterium spy nuclei into the adduct in order to take
advantage of the deuterium quadrupolar splitting to
observe enantiomers using a chiral liquid crystalline
solvent such as PBLG/DMF.
5
Figure 3. H–{ H} NMR spectra of a: Racemic 1d in PBLG/DMF
(
(
(
PBLG weight fraction X
22.0 mg), in PBLG/DMF (X
22.0 mg) plus racemic 1d
M
= 36.4%) at T = 330 K; b: S-enriched 1d
5
M
= 33.9%) at T = 320 K; c: S-enriched 1d
5
5
M
(5.7 mg) in PBLG/DMF (X = 29.0%) at
T = 320 K.
We previously proposed various possible ADDA’s for
8
and NMR experiments must be carried out at a tempera-
ture above 310 K.
amines or amino acids such as acetyl chloride-d , trifluoro-
3
acetic anhydride (using it as a fluorinated achiral derivatiz-
ing agent for fluorine NMR) or pentadeuterated benzoyl
chloride. The main inconvenience of these derivatizing
9
2
1
10
In Figure 3a the H–{ H} NMR spectrum of 22 mg of
racemic 1d dissolved in PBLG/DMF chiral liquid crystal
agents is that they yield to amides for which a possible
chemical equilibrium between the s-cis and s-trans confor-
mations of the amide group often causes some complexity
in the spectral analysis along with a tremendous linewidth
broadening. The use of perdeuterobenzyl chloride 2 as an
ADDA can lead to amines and not amides, which therefore
will not cause such problems. The latter, although commer-
cially available, can be easily prepared from the much
cheaper perdeuterotoluene with a standard procedure
(Scheme 1a).
5
at 330 K is reported. This time, the spectrum is of a good
quality and the line separations between enantiomeric sig-
nals are large enough to allow the measurement of an even-
tual enantiomeric excess. In Figure 3b, the spectrum of the
(
S)-enriched enantiomeric mixture obtained through asym-
metric synthesis is presented. In this spectrum, no signal
from the (R)-enantiomer could be detected as only three
quadrupolar doublets are observed, one for each kind of
deuterium (ortho, meta and para). Nevertheless, the pres-
ence of a very small amount of (R) cannot be excluded,
the signal of which could be hidden in the foot of the
Monobenzylation of secondary amines and dibenzylation
of primary amines were achieved in refluxing acetonitrile
using one to three equivalent of benzyl chloride 2 (Scheme
1b).
(
S)-enantiomer lines. In order to determine an accurate
experimental value for the enantiomeric excess we added
a known quantity of racemic 1d (5.7 mg) to sample B.
5
Spectrum 3-C corresponds to this mixture, where the sig-
nals from the (R) enantiomer are now observable. Careful
integration and deconvolution of the para proton reso-
2
1
The H–{ H} NMR spectrum of a single enantiomer of
such perdeuterobenzylamines in PBLG-DMF liquid crys-

1514
A. Solgadi et al. / Tetrahedron: Asymmetry 18 (2007) 1511–1516
a
b
D
D
D
D
D
D
D
D
Benzoyl peroxide
Reflux, 30 min
D
CD3
+
SO2Cl2
D
CD2Cl
68%
2
D
R1
D D
NaHCO3
R1
D
N H
+
+
2 (2 3 eq.)
N
R2
R2
CH3CN
Reflux, 24 h
D
D
D
D
D
D D
D
NaHCO3
D
R
NH2
2 (1 eq.)
R
N
CH3CN
Reflux, 24 h
D
2
Scheme 1. a: Preparation of benzyl chloride 2 from perdeuterotoluene. b: Benzylation of amines.
tal was expected to contain ten peaks; that is, five quad-
rupolar doublets: three doublets with intensity one corre-
sponding to the para, and the a and a diastereotopic
phase are essentially the same as those measured from
the isotropic spectrum (7.35 and 7.32 ppm for ortho and
0
0
meta; 7.24 ppm for para; 4.05 and 3.29 ppm for a and a
deuterium of the methylene group and two others with
intensity two corresponding to the ortho and meta deute-
rium. For a mixture of enantiomers, the spectrum should
be doubled, meaning that five doublets for each enantio-
mer should be observed.
deuterium). Furthermore, as the mixture was not racemic,
it was possible to assign all the doublets of a given
enantiomer.
The signals of all the deuterium atoms are doubled, that is,
a proper signal to each enantiomer is observed, except for
the para. This is a good illustration of having several differ-
ent deuterium nuclei in an ADDA. Depending on the dif-
ference in orientation between the enantiomers, it can
happen that no discrimination is observable along a given
carbon–deuterium (C–D) direction, although this is almost
never the case for all the other C–D directions. It should be
The first perdeuterobenzylamine to be tested was derived
from a non-racemic mixture of 2-(hydroxymethyl)-pyrrol-
2
1
idine enantiomers (ee = 34.3% ± 0.1 (S)). The H–{ H}
NMR spectrum of this compound in PBLG/DMF at
3
50 K is presented in Figure 4. It exhibits the spectral fea-
tures depicted above.
S
noted in Figure 4 that the splitting of the a deuterium is
null at this temperature. This means that the average angle
between this C–D bond and the magnetic field is equal to
the magic angle (54.7°) for the (S)-enantiomer.
Table 1 presents the absolute values of the quadrupolar
splittings measured for each type of deuterium in each
R
S
enantiomer (Dm and Dm ) at several temperatures. It also
Q
Q
R
S
includes the value jDm ꢁ Dm j=2, which is the spectral line
Q
Q
separation of enantiomers for a given deuterium. Thus, for
this example, the best separations are obtained for the
ortho and meta deuterons. Increasing the temperature
resulted in a decrease of the quadrupolar splittings as well
as the NMR linewidth.
2
1
Figure 4. 61.4 MHz H–{ H} NMR spectrum the perdeuterobenzylamine
derived from (S)-2-(hydroxymethyl)-pyrrolidine 34.3% ee in PBLG/DMF
A simple integration of the different well separated signals
gave an NMR estimated enantiomeric excess of 33% ± 2,
which is in good agreement with the value of 34.4% ± 0.1
of the weighed preparation. A better precision on the
NMR measured ee may be obtained by carrying out several
independent acquisitions on the same sample, then taking
an average of the measured integrations. With a signal to
noise ratio larger than 100 and ten independent measure-
ments, the mean standard deviation (MSD) on the average
M
at T = 350 K. (PBLG weight fraction X = 26.2%).
The assignment of the quadrupolar doublets to the corre-
sponding deuterium and enantiomer was achieved using
chemical shifts and integrations. Chemical shift anisotropy
is negligible for deuterium NMR. Consequently, the chem-
ical shifts of the deuterium nuclei in PBLG/DMF meso-

A. Solgadi et al. / Tetrahedron: Asymmetry 18 (2007) 1511–1516
1515
R
S
R
S
Table 1. Absolute values of deuterium quadrupolar splittings ðDm , Dm Þ and spectral separation, jDm ꢁ Dm j/2, for a perdeuterobenzylamine derived
Q
Q
Q
Q
from (S)-2-(hydroxymethyl)-pyrrolidine 34.3% ee in PBLG/DMF (PBLG weight fraction X
M
= 26.2%) at different temperatures
R
S
R
Q
S
Amine
T (K)
10
Site
jDm j (Hz)
jDm j (Hz)
jDm ꢁ Dm j=2 (Hz)
Q
Q
Q
3
a
a
726
96
746
138
355
347
654
673
100
308
301
625
612
64
10
21
56.5
56.5
16
13
16.5
51
50.5
11
15
α'
o
HO
α
0
D
D
D
m
D
o,m
o,m
p
468
460
686
647
67
410
402
647
582
40
N
p
o
D
D
m
320
330
a
a
D
0
2-(Hydroxymethyl)-
o,m
o,m
p
pyrrolidine
ee = 34.3% (S)
PBLG/DMF X
M
= 26.2%
a
a
0
12
o,m
o,m
p
360
352
616
477
269
261
602
513
0
45.5
45.5
7
18
8
3
50
a
0
a
a
16
o,m
o,m
p
278
272
566
208
201
566
35
35.5
0
a
Temperature evolution indicates a change in the sign of this quadrupolar splitting around 345 K.
ee should reach values below 1%. Benzyl chloride 2 thus
appears to be an excellent ADDA for secondary amines
associated with the PBLG/DMF liquid crystalline solvent.
For primary amines, the problem is more complex. The
substitution conditions required to obtain mono-benzyl-
ated amines with a reasonable yield would necessitate the
reaction of an excess of amine against benzyl chloride 2.
This is not suitable because in the majority of cases, the
chemists will furnish a small amount of the amine to be
studied. As a result, we recommend an excess of ADDA
be used in order to obtain the di-benzylated amine with a
good yield. With such a di-benzylated amine derivative,
the expected spectrum is similar to that of a mono-benzyl-
ated one. Actually, due to rapid nitrogen atom inversion
the two benzyl groups are equivalent.
2
1
Figure 5. 61.4 MHz H–{ H} NMR spectrum the bis-(perdeuterobenz-
yl)amine derived from (R)-1-aminoindane 32.2% ee in PBLG/DMF
(
M
PBLG weight fraction X = 26.2%) at T = 320 K.
R
S
R
S
Table 2. Absolute values of deuterium quadrupolar splittings (Dm , Dm ) and spectral separation, jDm ꢁ Dm j/2, for a perdeuterobenzylated amine derived
Q
Q
Q
Q
M
from (R)-1-aminoindane 32.2% ee in PBLG/DMF (PBLG weight fraction X = 26.2%) at different temperatures
R
S
R
Q
S
Q
Amine
T (K)
Site
jDm j (Hz)
jDm j (Hz)
jDm ꢁ Dm j=2 (Hz)
Q
Q
1
-Aminoindane ee = 32.2% (R)
310
a
a
o,m
o,m
p
1270
649
675
660
206
1156
576
612
593
219
1053
508
554
535
234
1050
573
564
549
186
935
499
503
484
192
838
434
451
432
200
110
38
55.5
55.5
10
0
α'
D
320
330
a
a
o,m
o,m
p
110.5
α
N
0
D
o
38.5
54.5
54.5
13.5
107.5
37
51.5
51.5
17
D
o
D
m
2
m
D
D
p
D
a
a
o,m
o,m
p
0
M
PBLG/DMF X = 26.2%

1516
A. Solgadi et al. / Tetrahedron: Asymmetry 18 (2007) 1511–1516
An example of using a non-racemic mixture of 1-aminoin-
dane enantiomers (ee = 32.2% ± 0.1 (R)) as the chiral
amine is presented in Figure 5. Again, a total assignment
of the quadrupolar doublets was possible with the help of
both the known isotropic chemical shifts and enantiomeric
excess. The signal assignments are noted in Figure 5. The
quadrupolar splittings of the various deuterium nuclei are
reported in Table 2 for each enantiomer, together with
their evolution with temperature. Here, the largest spectral
seem to induce any racemization. We have shown two
demonstrative examples but the technique was also
successful with (± )-sec-butylamine, (± )-2-aminopentane,
(± )-3-aminoheptane, (± )-2-amino 1-butanol, (± )-a-
methyl-benzylamine and (± )-2 methylpiperidine.1
1
We think these results should be of interest to chemists in-
volved with the asymmetric synthesis of amines, noticeably
in pharmaceutical industries.
separation is observed on one of the benzylic deuterium
0
(
a ).
References
Integration led to an NMR estimate of the enantiomeric
excess ee = 33.3% ± 2, which correlate well with the
1. Sarfati, M.; Lesot, P.; Merlet, D.; Courtieu, J. Chem.
Commun. 2000, 2069–2081.
3
2.2% ± 0.1 ee of the weighed preparation and indicates
2
. Canet, I.; Courtieu, J.; Loewenstein, A.; Meddour, A.;
Pechin e´ , J. M. J. Am. Chem. Soc. 1995, 117, 6520–6526.
. Lesot, P.; Sarfati, M.; Courtieu, J. Chem. Eur. J. 2003, 9,
that no racemization occurred during the reaction
with 2.
3
1
724–1745.
4
5
. Parker, D. Chem. Rev. 1991, 91, 1441–1457.
. Science/Technology concentrate, Chemical Engineering News
3. Conclusions
1
997, 6, 24.
We have shown that a lyotropic liquid crystal made up of a
dimethylformamide solution of poly-c-benzyl-L-glutamate
with a PBLG weight fraction around 26% is efficient for
the NMR study of chiral amines. In this chiral anisotropic
solvent, the NMR linewidths are small enough to discrim-
inate enantiomers of amines in contrast with the more com-
mon PBLG/chloroform liquid crystal. It seems that DMF
is able to break some strong intermolecular interactions be-
tween the polypeptide fibres and amines, which causes
broad NMR lines in the chloroform liquid crystal solution.
6. Rouden, J.; Bernard, A.; Lasne, M.-C. Tetrahedron Lett.
1999, 40, 8109; Jean, L.; Rouden, J.; Maddaluno, J.; Lasne,
M.-C. J. Org. Chem. 2004, 69, 8893–8902.
7
. Corruble, A.; Davoust, D.; Desjardins, S.; Fressign e´ , C.;
Giessner-Prettre, C.; Harrison-Marchand, A.; Houte, H.;
Lasne, M.-C.; Maddaluno, J.; Oulyadi, H.; Valnot, J.-Y. J.
Am. Chem. Soc. 2002, 124, 15267–15279.
8
. Canet, J. L.; Canet, I.; Courtieu, J.; Da Silva, S.; Gelas, J.;
Troin, Y. J. Org. Chem. 1996, 61, 9035–9037.
9
. Jakubova, M.; Meddour, A.; Pechine, J. M.; Baklouti, A.;
Courtieu, J. J. Fluorine Chem. 1997, 86, 149–153.
1
0. Meddour, A.; Loewenstein, A.; Pechine, J. M.; Courtieu, J.
Furthermore, perdeuterobenzyl chloride appears to be a
good achiral deuterated derivatizing agent for primary
and secondary amines. The reaction is mild and does not
Tetrahedron: Asymmetry 1997, 8, 485–494.
11. Solgadi, A. PhD Thesis, Universit e´ Paris-Sud 11, No. 7804,
004.
2