Structural relationships in crystals accommodating different stereoisomers of 2-amino-3-methylpentanoic acid

Article abstract of DOI:10.1107/S0108768100002810

A reinvestigation of the crystal structure of the 1:1 mixture of the two racemates DL-isoleucine and DL-allo-isoleucine, with a detailed analysis of interatomic distances between alternative side-chain positions, reveals a systematic distribution of the four stereoisomers in this crystal. Two different molecular chains exist in the crystal and each such chain accommodates a single diastereomeric pair only (L-isoleucine:D-allo-isoleucine or D-isoleucine:L-allo-isoleucine). The crystal is built up by a stacking of such chains in two dimensions and three different packing modes for the two types of chains are discussed. Crystallization experiments of the two individual racemates in the 1:1 mixture of DL-isoleucine:DL-allo-isoleucine have been undertaken. The structure of the racemate DL-isoleucine is presented. The molecular arrangements in this racemate and the 1:1 DL-isoleucine:DL-allo-isoleucine mixture are closely related. Furthermore, the spontaneous resolution of enantiomers upon crystallization of the other racemate, DL-allo-isoleucine, is rationalized on the basis of the aforementioned analysis of interatomic distances in the 1:1 DL-isoleucine:DL-allo-isoleucine complex. Structural data for a new L-isoleucine: D-allo-isoleucine complex are also given.

Full text of DOI:10.1107/S0108768100002810

research papers
Acta Crystallographica Section B
Structural
Science
Structural relationships in crystals accommodating
different stereoisomers of 2-amino-3-methylpenta-
noic acid
ISSN 0108-7681
Received 29 November 1999
Bjùrn Dalhus* and Carl Henrik
G oÈ rbitz
A reinvestigation of the crystal structure of the 1:1 mixture of
the two racemates dl-isoleucine and dl-allo-isoleucine, with a
Accepted 18 February 2000
detailed analysis of interatomic distances between alternative
side-chain positions, reveals a systematic distribution of the
four stereoisomers in this crystal. Two different molecular
chains exist in the crystal and each such chain accommodates a
single diastereomeric pair only (l-isoleucine:d-allo-isoleucine
or d-isoleucine:l-allo-isoleucine). The crystal is built up by a
stacking of such chains in two dimensions and three different
packing modes for the two types of chains are discussed.
Crystallization experiments of the two individual racemates in
the 1:1 mixture of dl-isoleucine:dl-allo-isoleucine have been
undertaken. The structure of the racemate dl-isoleucine is
presented. The molecular arrangements in this racemate and
the 1:1 dl-isoleucine:dl-allo-isoleucine mixture are closely
related. Furthermore, the spontaneous resolution of enantio-
mers upon crystallization of the other racemate, dl-allo-
isoleucine, is rationalized on the basis of the aforementioned
analysis of interatomic distances in the 1:1 dl-isoleucine:dl-
allo-isoleucine complex. Structural data for a new l-isoleu-
cine: d-allo-isoleucine complex are also given.
Department of Chemistry, University of Oslo,
PO Box 1033, Blindern, N-0315 Oslo, Norway
Correspondence e-mail: bjornda@kjemi.uio.no
1
. Introduction
The ꢀ-amino acid 2-amino-3-methylpentanoic acid (isoleu-
ꢀ
ꢁ
cine) embodies two chiral C atoms, C and C (I), and
consequently four stereoisomers exist. The isomer with
ꢀ
ꢁ
absolute con®guration S at both C and C (denoted [S,S]) is
the naturally occurring ꢀ-amino acid l-isoleucine (l-Ile), while
the [R,R]-isomer represents its enantiomer, d-isoleucine (d-
Ile). The two stereoisomers [S,R] and [R,S] are l-allo-isoleu-
cine (l-allo-Ile) and d-allo-isoleucine (d-allo-Ile), respectively.
The four stereoisomers can form a total of four different 1:1
complexes; the racemates l-Ile:d-Ile [dl-Ile, (1)] and l-allo-
Ile:d-allo-Ile [dl-allo-Ile, (2)] as well as the two diaster-
eomeric complexes l-Ile:d-allo-Ile (3) and l-Ile:l-allo-Ile (4),
Table 1. As part of our program dealing with hydrogen-bond
geometries in the crystal structures of hydrophobic amino
acids (Dalhus & G oÈ rbitz, 1999, and references therein) crys-
tallization experiments with these complexes have been
#
2000 International Union of Crystallography
Printed in Great Britain ± all rights reserved
7
20 Dalhus and G oÈ rbitz ꢀ 2-Amino-3-methylpentanoic acid
Acta Cryst. (2000). B56, 720±727

research papers
carried out and in this paper we describe the crystal structures
of the racemate dl-Ile (1) and the 1:1 complex l-Ile:d-allo-Ile
Table 1
Various complexes of different stereoisomers of 2-amino-3-methyl-
pentanoic acid.
(
3).
Additionally, we have redetermined the crystal structures of
No.
Stereochemical con®guration
Complex
the 1:1:1:1 complex l-Ile:d-Ile:l-allo-Ile:d-allo-Ile (5), con-
taining all four stereoisomers, and d-allo-Ile (6). These two
structures have previously been reported by Benedetti et al.
(
(2)
1)
[S,S]:[R,R]
[S,R]:[R,S]
[S,S]:[R,S]
[S,S]:[S,R]
[S,S]:[R,R]:[S,R]:[R,S]
l-Ile:d-Ile
l-allo-Ile:d-allo-Ile
l-Ile:d-allo-Ile
l-Ile:l-allo-Ile
(3)
(4)
(5)
(
1973) and Varughese & Srinivasan (1975), respectively, but
l-Ile:d-Ile:l-allo-Ile:d-allo-Ile
the precision of the structural parameters is low by current
standards, with no information on the hydrogen-bond geo-
metries.
Crystal structures of seven 1:1 complexes of l-isoleucine
with selected hydrophobic d-amino acids have been presented
elsewhere (Dalhus & G oÈ rbitz, 1999).
mixed with tetramethoxysilane in the ratio 10:1 and 100±150 ml
of the resulting mixtures were dispensed in each of 10±12 30 Â
5
mm test-tubes, sealed with para®lm and left for some
minutes to polymerize. Applying the vapor diffusion tech-
nique, ethanol or 2-propanol subsequently diffused into the
gel at room temperature.
Thin and soft plates of racemate (1) were formed with
either alcohol as precipitating agents. Only a few crystals
extinguished and brightened satisfactorily when rotated in
plane-polarized light, and the specimen used for data collec-
tion (from the ethanol batch) was selected after testing a large
number of crystals. The diffraction patterns for some of the
other crystals are essentially two-dimensional with substantial
2
. Experimental
2
.1. Crystallization
Aqueous solutions of the four 1:1 mixtures (1)±(4) (Table 1)
as well as the equimolar mixture of all four stereoisomers (5)
Table 1) were prepared. Typically, 5±10 mg of each of the
amino acids in question were mixed and dissolved in 1000±
500 ml of water. The various amino acid solutions were then
(
*
1
streaking along c .
Figure 1
The asymmetric unit with atomic numbering for (a) dl-Ile:dl-allo-Ile (5), (b) l-Ile:d-allo-Ile (3) [molecule A is l-Ile and molecule B is d-allo-Ile], (c) dl-
Ile [l-isomer only] (1) and (d) d-allo-Ile (6). Displacement ellipsoids are drawn at the 50% probability level. H atoms are arbitrarily scaled. In (a) the
0
ꢂ
ꢃ2
0
disordered C methyl groups (C6 and C6 with bonded H atoms) are dotted and the disordered C methyl groups (C4 and C4 with bonded H atoms)
have been omitted for clarity.
Acta Cryst. (2000). B56, 720±727
Dalhus and G oÈ rbitz ꢀ 2-Amino-3-methylpentanoic acid 721

research papers
Table 2
Experimental details.
(1)
(3)
(5)
(6)
Crystal data
Chemical formula
Chemical formula weight
Cell setting
C H NO
13
C H NO
13
C H NO
13
C H NO
2
6
2
6
2
6
2
6
13
131.17
Triclinic
P1
131.17
Triclinic
P1
131.17
Triclinic
P1
131.17
Monoclinic
P21
9.6706 (1)
5.2583 (1)
14.1018 (2)
90
98.033 (1)
90
710.05 (2)
4
1.227
Space group
Ê
a (A)
Ê
5.2289 (1)
5.4102 (1)
13.1095 (3)
96.332 (1)
90.622 (1)
109.493 (1)
347.02 (1)
2
1.255
Mo Kꢀ
0.71073
4333
5.2438 (2)
5.3978 (2)
13.2562 (6)
93.042 (1)
92.811 (1)
109.897 (1)
351.42 (2)
2
1.240
Mo Kꢀ
0.71073
3483
5.2493 (1)
5.4006 (1)
13.2778 (2)
92.9433 (6)
92.8659 (5)
109.8571 (7)
352.66 (1)
2
b (A)
Ê
c (A)
ꢂ
ꢀ
ꢁ
ꢁ †
ꢂ
ꢁ †
ꢂ
ꢃ ꢁ †
Ê
V (A )
3
Z
D (Mg m
�
3
)
Radiation type
1.235
x
Mo Kꢀ
0.71073
6724
Mo Kꢀ
0.71073
7959
Ê
Wavelength (A)
No. of re¯ections for cell
parameters
�
1
ꢄ (mm
)
0.093
150 (2)
0.092
140 (2)
0.092
150 (2)
0.091
150 (2)
Temperature (K)
Crystal form
Crystal size (mm)
Crystal colour
Plate
0.50 Â 0.50 Â 0.10
Plate
0.45 Â 0.25 Â 0.10
Plate
0.75 Â 0.55 Â 0.30
Plate
0.90 Â 0.50 Â 0.05
Colourless
Colourless
Colourless
Colourless
Data collection
Diffractometer
Data collection method
Absorption correction
Siemens SMART CCD
! scans
Multi-scan (Sheldrick,
Siemens SMART CCD
! scans
Multi-scan (Sheldrick,
Siemens SMART CCD
! scans
Multi-scan (Sheldrick,
Siemens SMART CCD
! scans
Multi-scan (Sheldrick,
1
996)
996)
0.959
0.991
4788
1996)
0.933
0.973
1996)
0.921
0.995
Tmin
Tmax
No. of measured
re¯ections
0.955
0.991
8317
9079
16 583
No. of independent
re¯ections
No. of observed
re¯ections
5815
4439
4029
6187
11 372
8967
3746
5540
Criterion for observed re¯ec- I > 2ꢅ(I)
tions
I > 2ꢅ(I)
I > 2ꢅ(I)
I > 2ꢅ(I)
Rint
ꢆmax ( )
Range of h, k, l
0.0332
40
0.0096
32.5
� 8 ! h ! 5
� 8 ! k ! 9
� 22 ! l ! 22
0.0213
40.0
� 10 ! h ! 10
� 10 ! k ! 10
� 28 ! l ! 28
0.0286
40.0
� 16 ! h ! 20
� 10 ! k ! 10
� 30 ! l ! 25
ꢂ
� 10 ! h ! 9
�
�
9 ! k ! 11
26 ! l ! 21
Re®nement
Re®nement on
2
2
2
2
F
F
F
F
2
2
R‰F >2ꢅꢁF †Š
0.0576
0.1562
1.110
5815
0.0326
0.0934
1.104
4028
0.0605
0.1915
1.303
6187
0.0483
0.1080
1.074
2
wRꢁF †
S
No. of re¯ections used in
re®nement
11 371
No. of parameters used
Weighting scheme
101
w = 1/[ꢅ (Fo ) + (0.0888P)
0.0250P], where
201
w = 1/[ꢅ (Fo ) + (0.0675P)
+ 0.0025P], where
116
w = 1/[ꢅ (Fo ) + (0.0561P)
+ 0.1597P], where
196
w = 1/[ꢅ (Fo ) + (0.0567P)
+ 0.0061P], where
2
2
2
2
2
2
2
2
2
2
2
2
+
P = (Fo + 2F )/3
2
2
2
2
2
2
2
2
P = (Fo + 2F )/3
c
P = (Fo + 2F )/3
c
P = (Fo + 2F )/3
c
c
ꢁ
Á=ꢅ†
0.001
0.620
0.001
0.443
0.032
0.585
� 0.002
max
Ê
Ê
� 3
)
Áꢇ (e A
0.535
� 0.308
None
max
� 3
)
Áꢇmin (e A
Extinction method
� 0.463
� 0.221
� 0.344
None
None
SHELXTL
(
1.23 (7)
International Tables for
Crystallography (1992, Vol.
C, Tables 4.2.6.8 and
6.1.1.4)
Sheldrick, 1995)
Extinction coef®cient
Source of atomic scattering
factors
±
International Tables for
Crystallography (1992, Vol.
C, Tables 4.2.6.8 and
±
International Tables for
Crystallography (1992, Vol.
C, Tables 4.2.6.8 and
6.1.1.4)
±
International Tables for
Crystallography (1992, Vol.
C, Tables 4.2.6.8 and
6.1.1.4)
6
.1.1.4)
Computer programs
Structure solution
Structure re®nement
SHELXTL (Sheldrick, 1995) SHELXTL (Sheldrick, 1995) SHELXTL (Sheldrick, 1995) SHELXTL (Sheldrick, 1995)
SHELXTL (Sheldrick, 1995) SHELXTL (Sheldrick, 1995) SHELXTL (Sheldrick, 1995) SHELXTL (Sheldrick, 1995)
7
22 Dalhus and G oÈ rbitz ꢀ 2-Amino-3-methylpentanoic acid
Acta Cryst. (2000). B56, 720±727

research papers
Figure 2
Molecular packing diagram for (a) dl-Ile:dl-allo-Ile (5), (b) l-Ile:d-allo-Ile (3), (c) dl-Ile (1) and (d) d-allo-Ile (6). In (a) the disordered C methyl
ꢂ
0
ꢃ2
groups (C6 and C6 with bonded H atoms) are drawn with broken lines and the disordered C methyl groups (C4 and C4 with bonded H atoms) have
0
been omitted for clarity. Furthermore, the three molecules in the upper row in (a) are translated one unit along the a axis relative to the molecules in the
Ê
lower row. The short methylÁ Á Ámethyl contacts (in A, see text) are indicated by dotted lines.
The two enantiomers of racemate (2) were always sponta-
neously resolved upon crystallization, rendering a structure
determination of this racemate impossible.
small systematic errors (SHELXTL; Sheldrick, 1995).
Experimental conditions with information on the data
reduction and re®nement results are summarized in Table 2.
All structures were determined by direct methods using
SHELXTL (Sheldrick, 1995).
1
Well diffracting plate-shaped crystals of complex (3) were
obtained readily, with both ethanol and 2-propanol. However,
ethanol seems to have a positive effect on the thickness of the
crystals. Similarly, complex (5), with all four stereoisomers
present, formed relatively large plates with both alcohols. The
crystal used in the diffraction experiment was taken from the
ethanol batch. Complex (4), on the other hand, gave only
extremely thin needles (typically less than 0.01 mm) unsui-
table for conventional X-ray diffraction experiments.
A thin plate of d-allo-Ile (6), crystallized using 2-propanol
from a sample of known chirality, was used in the reinvesti-
gation of this structure.
All non-H atoms were re®ned anisotropically. Amino H
atoms were re®ned isotropically. H atoms bonded to C were
placed geometrically and re®ned with constraints to keep all
CÐH distances and all CÐCÐH angles on one C atom the
same. Free rotation about CÐC bonds was permitted for
methyl groups. Isotropic displacement parameters for the H
atoms were ®xed at 1.5U (for ÐCH ) or 1.2U (for ÐCH Ð
eq
3
eq
2
and ÐCHÐ) of the bonded C atom. In the model for (5)
0
0
atoms C4, C5 and C6 correspond to d/l-Ile, while C4 , C5 and
0
C6 correspond to d/l-allo-Ile. The occupancy of each
stereoisomer was con®ned by symmetry to be 0.5. [A re®ne-
ment of the occupancy factors gives 0.48 (1) and 0.52 (1) for
2
.2. Data collection, structure determination and refinement
0
0
C6 and C6 , respectively]. C6 and C6 give rise to two inde-
0
The data collections were performed with a Siemens
SMART CCD diffractometer (Siemens, 1995) and nominally
covered over a hemisphere of reciprocal space. The datasets
pendent peaks in the electron density map, while C4 and C5
1
Supplementary data for this paper are available from the IUCr electronic
archives (Reference: OS0046). Services for accessing these data are described
at the back of the journal.
ꢂ
ꢂ
are 99% complete to at least ꢆ = 32.5 [(1), (5) and (6)] and 40
3). Friedel pairs have not been merged, since this introduces
(
Acta Cryst. (2000). B56, 720±727
Dalhus and G oÈ rbitz ꢀ 2-Amino-3-methylpentanoic acid 723

research papers
Table 3
Fractional atomic coordinates and isotropic or equivalent isotropic
displacement parameters (A ) for dl-Ile (1), l-Ile:d-allo-Ile (3), dl-
Ile:dl-allo-Ile (5) and d-allo-Ile (6).
share the same set of parameters (constrained by the EXYZ
and EADP cards in SHELXTL; Sheldrick, 1995), as do also
C5 and C4 .
Ê
2
0
ij
i j
Ueq ˆ ꢁ1=3†Æ Æ U a a a :a :
i
j
i
j
3
. Results and discussion
Atomic coordinates for all four structures are listed in Table 3.
Selected geometric parameters are given in Table 4 and
hydrogen-bond geometries are listed in Table 5. Figs. 1(a)±(d)
show the molecular conformations, including atomic
numbering for the four structures.
x
y
z
Ueq
dl-Ile (1)
O1
O2
N1
C1
C2
C3
C4
C5
0.80323 (9)
0.36212 (9)
0.81846 (10)
0.58380 (11)
0.58889 (10)
0.61101 (12)
0.81181 (15)
0.33004 (15)
0.3293 (2)
0.75963 (8)
0.63846 (9)
0.26922 (9)
0.59662 (10) 0.61346 (4)
0.32193 (10) 0.63253 (4)
0.29197 (11) 0.74779 (5)
0.54061 (14) 0.80856 (6)
0.58960 (4)
0.62394 (4)
0.57840 (4)
0.01627 (9)
0.01616 (9)
0.01351 (8)
0.01216 (8)
0.01191 (8)
0.01447 (9)
0.02060 (12)
0.02330 (13)
0.0371 (2)
3.1. Molecular geometry
There are no unusual features in the overall geometry for
the molecules in these four structures, Table 4. In both dl-
0.2148 (2)
0.1342 (2)
0.79270 (6)
0.90085 (7)
C6
d-Ile:d-allo-Ile (3)
O1A
O2A
N1A
C1A
C2A
C3A
C4A
C5A
C6A
O1B
O2B
N1B
C1B
C2B
C3B
C4B
C5B
C6B
0.9524 (2)
0.5127 (2)
0.9645 (2)
0.7340 (2)
0.7406 (2)
0.7765 (2)
0.9848 (3)
0.5015 (3)
0.5155 (3)
0.3355 (2)
0.7748 (2)
0.3217 (2)
0.5541 (2)
0.5494 (2)
0.5233 (2)
0.8033 (3)
0.3228 (3)
0.2612 (3)
0.6557 (2)
0.5281 (2)
0.1650 (2)
0.4885 (2)
0.2121 (2)
0.1732 (2)
0.4177 (3)
0.0912 (3)
� 0.0009 (3)
0.57652 (6)
0.61114 (6)
0.56582 (7)
0.60067 (7)
0.62044 (7)
0.73401 (7)
0.79132 (9)
0.78135 (9)
0.88809 (9)
0.0149 (2)
0.01442 (15)
0.0118 (2)
0.0105 (2)
0.0105 (2)
0.0124 (2)
0.0184 (2)
0.0203 (2)
0.0303 (2)
0.01413 (15)
0.0144 (2)
0.0117 (2)
0.0108 (2)
0.0109 (2)
0.0141 (2)
0.15514 (15) 0.40100 (6)
0.2815 (2)
0.6441 (2)
0.3217 (2)
0.5993 (2)
0.6409 (2)
0.7279 (3)
0.3951 (3)
0.4403 (3)
0.36717 (6)
0.41060 (7)
0.37729 (7)
0.35807 (7)
0.24385 (8)
0.20184 (11) 0.0264 (3)
0.18385 (8)
0.07347 (8)
0.0187 (2)
0.0318 (3)
dl-Ile:dl-allo-Ile (5)
O1
O2
N1
C1
C2
0.30806 (12)
0.75015 (11) 0.58772 (5)
0.62306 (12) 0.62196 (5)
0.26020 (12) 0.57762 (5)
0.58294 (13) 0.61158 (5)
0.30626 (13) 0.63118 (5)
0.26586 (15) 0.74491 (6)
0.01528 (10)
0.01533 (10)
0.01253 (10)
0.01132 (10)
0.01136 (10)
0.01437 (11)
0.02080 (15)
0.0256 (2)
0.0337 (5)
0.0256 (2)
0.02080 (15)
0.0344 (5)
� 0.13094 (12)
0.32122 (13)
0.08990 (14)
0.09599 (13)
0.1269 (2)
C3
C4²
C5²
C6²
0.3310 (2)
0.5107 (2)
0.1817 (2)
0.0921 (6)
0.1817 (2)
0.5107 (2)
0.4649 (7)
0.80368 (7)
0.78999 (9)
0.8984 (2)
0.78999 (9)
0.80368 (7)
0.9153 (2)
� 0.1495 (2)
� 0.1310 (6)
� 0.1495 (2)
0.3310 (2)
0
C4 ²
C5 ²
0
0
C6 ²
0.3840 (7)
d-allo-Ile (6)
O1A
O2A
N1A
C1A
C2A
C3A
C4A
C5A
C6A
O1B
O2B
N1B
C1B
0.64390 (5)
0.73315 (5)
0.89125 (6)
0.74430 (6)
0.89164 (6)
0.94808 (6)
1.01413 (9)
0.83562 (8)
0.89085 (12)
1.16317 (5)
1.24741 (6)
1.43383 (6)
1.25852 (6)
1.39799 (6)
1.38807 (7)
1.52314 (9)
1.34993 (10)
1.31314 (10)
0.61873 (11) 0.58600 (4)
0.99335 (10) 0.63843 (4)
0.40288 (11) 0.57706 (4)
0.76242 (12) 0.61626 (4)
0.64770 (12) 0.63091 (5)
0.60705 (13) 0.73830 (5)
0.01513 (8)
0.01703 (9)
0.01317 (8)
0.01199 (9)
0.01219 (9)
0.01464 (10)
0.0249 (2)
0.8529 (2)
0.5049 (2)
0.4393 (2)
0.78089 (7)
0.79516 (5)
0.89933 (6)
0.01878 (11)
0.0304 (2)
1.22830 (11) 0.58923 (4)
1.62364 (11) 0.61395 (4)
1.04848 (12) 0.60816 (4)
1.38685 (12) 0.61819 (5)
1.27240 (13) 0.66580 (5)
1.18566 (14) 0.76921 (5)
0.01637 (9)
0.02034 (10)
0.01535 (9)
0.01262 (9)
0.01338 (9)
0.01600 (11)
0.0289 (2)
C2B
C3B
C4B
C5B
1.0579 (2)
1.4083 (2)
1.3257 (2)
0.81456 (7)
0.83059 (6)
0.92814 (7)
0.0320 (2)
0.0469 (3)
C6B
Figure 3
0
0
0
² C4, C5 and C6 apply to d/l-Ile, while C4 , C5 and C6 apply to d/l-allo-Ile. The
occupancy factors for all six atoms are ®xed at 0.5.
(a) Chain of l-Ile:d-allo-Ile molecules in (5). (b) Chain of d-Ile:l-allo-Ile
molecules in (5).
7
24 Dalhus and G oÈ rbitz ꢀ 2-Amino-3-methylpentanoic acid
Acta Cryst. (2000). B56, 720±727

research papers
Table 4
Selected distances and torsions (A, ) for dl-Ile (1), l-Ile:d-allo-Ile (3), dl-Ile:dl-allo-Ile (5) and d-allo-Ile
6).
3.2.1. DL-Ile:DL-allo-Ile (5).
This structure is a mixture of all
four stereoisomers of the title
compound. In the original
structure description (Benedetti
Ê
ꢂ
(
dl-Ile (1)
O1ÐC1
O2ÐC1
N1ÐC2
C1ÐC2
O1ÐC1ÐC2ÐN1
N1ÐC2ÐC3ÐC4
1.2592 (7)
1.2591 (7)
1.4927 (7)
1.5427 (7)
� 22.95 (7)
81.48 (6)
C2ÐC3
C3ÐC5
C3ÐC4
C5ÐC6
N1ÐC2ÐC3ÐC5
C2ÐC3ÐC5ÐC6
1.5445 (8)
1.5307 (9)
1.5328 (9)
1.5282 (12)
� 153.14 (5)
169.79 (7)
2
et al., 1973) the crystal is
described as a solid solution of
the two racemic pairs dl-Ile and
dl-allo-Ile in a ratio not signi®-
cantly different from 1:1.
l-Ile:d-allo-Ile (3)
O1AÐC1A
O2AÐC1A
N1AÐC2A
C1AÐC2A
C2AÐC3A
C3AÐC4A
C3AÐC5A
C5AÐC6A
1.2619 (12)
1.2622 (13)
1.4950 (15)
1.5397 (15)
1.5425 (13)
1.531 (2)
O1BÐC1B
O2BÐC1B
N1BÐC2B
C1BÐC2B
C2BÐC3B
C3BÐC4B
C3BÐC5B
C5BÐC6B
1.2589 (13)
1.2602 (13)
1.4932 (15)
1.542 (2)
1.5486 (14)
1.525 (2)
1.535 (2)
A close inspection of the
interatomic distances between
the alternative positions for the
disordered side-chain atoms in
the hydrophobic layer reveals
two relatively short CÁ Á ÁC
1.534 (2)
1.531 (2)
1.533 (2)
0
0
i
contacts; C6 Á Á ÁC6 [symmetry
O1AÐC1AÐC2AÐN1A
N1AÐC2AÐC3AÐC4A
N1AÐC2AÐC3AÐC5A
C2AÐC3AÐC5AÐC6A
� 23.81 (12)
81.08 (11)
O1BÐC1BÐC2BÐN1B
N1BÐC2BÐC3BÐC4B
N1BÐC2BÐC3BÐC5B
C2BÐC3BÐC5BÐC6B
22.43 (12)
152.52 (11)
� 82.05 (12)
172.06 (10)
code:
z ‡ 2]
(i)
� x ‡ 1; � y ‡ 1,
Ê
� 153.38 (10)
�
2.452 (5) A
and
169.10 (10)
ii
C6Á Á ÁC6 [symmetry code: (ii)
Ê
3.315 (6) A,
dl-Ile:dl-allo-Ile (5)²
O1ÐC1
O2ÐC1
N1ÐC2
C1ÐC2
� x; � y; � z ‡ 2]
1.2611 (9)
1.2623 (9)
1.4937 (9)
1.5404 (9)
1.5453 (10)
1.542 (2)
C3ÐC4
C3ÐC5
C5ÐC6
1.5338 (12)
1.5289 (13)
1.551 (3)
1.5289 (13)
1.5338 (12)
Fig. 2(a). The corresponding
shortest HÁ Á ÁH contacts are 1.84
0
Ê
and 2.45 A, respectively. The
C3ÐC4
C3ÐC5
0
C2ÐC3
0
molecules involved in these
contacts form chains parallel to
the ab diagonal. To avoid these
short methylÁ Á Ámethyl contacts,
one of the two con¯icting posi-
tions in each such contact has to
be vacant. This restriction gives
rise to two different molecular
chains, one with the diaster-
eomeric pair l-Ile:d-allo-Ile
0
C5 ÐC6
0
O1ÐC1ÐC2ÐN1
N1ÐC2ÐC3ÐC4
N1ÐC2ÐC3ÐC5
C2ÐC3ÐC5ÐC6
� 23.30 (9)
N1ÐC2ÐC3ÐC4
N1ÐC2ÐC3ÐC5
C2ÐC3ÐC5 ÐC6
� 152.95 (7)
81.64 (8)
� 174.73 (14)
0
0
81.64 (8)
� 152.95 (7)
169.89 (15)
0
d-allo-Ile (6)
O1AÐC1A
O2AÐC1A
N1AÐC2A
C1AÐC2A
C2AÐC3A
C3AÐC4A
C3AÐC5A
C5AÐC6A
O1AÐC1AÐC2AÐN1A
N1AÐC2AÐC3AÐC4A
N1AÐC2AÐC3AÐC5A
C2AÐC3AÐC5AÐC6A
1.2569 (8)
1.2623 (8)
1.4944 (8)
1.5344 (8)
1.5512 (9)
1.5271 (11)
1.5356 (10)
1.5306 (11)
17.68 (8)
O1BÐC1B
O2BÐC1B
N1BÐC2B
C1BÐC2B
C2BÐC3B
C3BÐC4B
C3BÐC5B
C5BÐC6B
O1BÐC1BÐC2BÐN1B
N1BÐC2BÐC3BÐC4B
N1BÐC2BÐC3BÐC5B
C2BÐC3BÐC5BÐC6B
1.2676 (8)
1.2504 (9)
1.4986 (9)
1.5424 (8)
1.5437 (9)
1.5275 (11)
1.5316 (11)
1.5312 (14)
43.22 (7)
(Fig. 3a) and one with d-Ile:l-
allo-Ile (Fig. 3b). Allthough the
structure is locally non-centro-
symmetric, the overall structure
is centrosymmetric. The 1:1:1:1
ratio between the four stereo-
isomers is a direct consequence
of the 1:1 ratio between the
152.47 (6)
� 81.92 (7)
174.68 (7)
56.35 (8)
� 179.37 (6)
169.71 (3)
0 0 0
C4, C5 and C6 apply to d/l-Ile, whilst C4 , C5 and C6 apply to d/l-allo-Ile.
²
Ile:dl-allo-Ile (5) and l-Ile:d-allo-Ile (3) the conformation of
d-allo-Ile corresponds to that of molecule A in the d-allo-Ile
diastereomers in each molecular chain combined with the
centrosymmetric space group.
(
6) structure. The molecular conformation of l-Ile is similar in
the three structures where this molecule is present, (1), (3) and
5).
A complete hydrophobic layer is generated by a stacking of
the molecular arrangements depicted in Figs. 3(a) and (b).
Three alternative models exist for the build-up of this
hydrophobic layer:
(
(
(
i) a stacking of one of the depicted arrangements only,
ii) a systematic alternating stacking and
3
.2. Crystal packing
The molecular packing arrangement in the four structures is
(iii) a random stacking of the two arrangements.
illustrated in Figs. 2(a)±(d). The layer-like build-up of crystals
is a consequence of the dual character of these molecules; the
charged ꢀ-amino and ꢀ-carboxylate groups form hydrogen
bonds with each other, while the hydrophobic side chains are
involved in van der Waals' interactions only. The hydrophilic
and hydrophobic layers are emphasized in Fig. 2(a).
In the ®rst model only a single diastereomeric pair (l-Ile:d-
allo-Ile or d-Ile:l-allo-Ile) is present within a speci®c hydro-
phobic layer. The molecular distribution in this model is illu-
2
Erroneously registered as dl-isoleucine (Refcode DLILEU) in the
Cambridge Structural Database (Allen & Kennard, 1993).
Acta Cryst. (2000). B56, 720±727
Dalhus and G oÈ rbitz ꢀ 2-Amino-3-methylpentanoic acid 725

research papers
strated in Fig. 4(a). Since only two of the four stereoisomers in
the crystal are included in each hydrophobic layer, the crystal
must be built up by a stacking of two types of layer, one with
the pair l-Ile:d-allo-Ile and one with d-Ile:l-allo-Ile. This
model puts no restriction on the ratio between the two
aforementioned diastereomeric pairs.
If, on the other hand, an alternating stacking of the
arrangements in Figs. 3(a) and (b) is assumed, all four
stereoisomers coexist in a single hydrophobic layer and,
automatically, in equimolar numbers. The Ile and allo-Ile
molecules within a single molecular layer alternate along both
the a and b axes, Fig. 4(b).
The last model, with a random stacking of the aforemen-
tioned alternative arrangements, also represents a model in
which all four stereoisomers are present within each hydro-
phobic layer. However, as with the ®rst model, the ratio
between the two diastereomeric pairs can, in principle, take
any value.
From the above analysis of the short methylÁ Á Ámethyl
contacts ± unveiling two different molecular arrangements
encompassing a single diastereomeric
pair ± it may be concluded that the
distribution of the four stereoisomers
is not random throughout the crystal.
However, it is not possible, from the
structural data for (5), to establish
which of the three proposed arrange-
ments most likely occurs in the crystals.
There are only small differences in the
short non-bonding methylÁ Á Ámethyl
distances in the three models; all such
contacts across the hydrophobic layer
in the three models have CÁ Á ÁC
Ê
distances in the range 3.91±3.93 A,
with corresponding HÁ Á ÁH distances
Ê
between 2.63 and 2.79 A.
3.2.2. L-Ile:D-allo-Ile (3). The two
stereoisomers in this complex aggre-
gate in double layers, one with l-Ile
and one with d-allo-Ile, Fig. 2(b). The
molecular packing in the hydrophobic
layer in this structure (Fig. 2b) and the
l-Ile:d-allo-Ile double layer (Fig. 3a) in
the ®rst model for (5) is identical. The
0
iii
C6Á Á ÁC6
[symmetry code: (iii)
Ê
3.935 (5) A
�
x; � y ‡ 1; � z ‡ 2]
contact in (5) is only slightly altered in
this complex; in (3) the corresponding
Ê
distance is 3.918 (2) A. The other
0
ii
Ê
C6 Á Á ÁC6 3.913 (5) A contact in (5) is
practically unchanged in (3),
.910 (2) A. It is noteworthy that the
Ê
3
molecular conformations in (3) and in
the l-Ile:d-allo-Ile pair in (5) are
almost identical; the largest difference
ꢂ
for corresponding torsion angles is 2.7
[
C2BÐC3BÐC5BÐC6B for d-allo-Ile
0
0
in (3) and C2ÐC3ÐC5 ÐC6 for d-
allo-Ile in (5)], Table 4.
3.2.3. DL-Ile (1). Analogous to (3),
the two enantiomers in this racemate
crystallize in double layers, with the d-
allo-Ile molecules in (3) replaced by d-
Ile (Fig. 2c). The structure of (1) can be
generated from the structure of (5) by
occupying all the C6 positions and
Figure 4
Stereoplot illustrating two of the three proposed molecular packing arrangements in a single
hydrophobic layer of dl-Ile:dl-allo-Ile (5). (a) Layer accommodating a single diastereomeric pair
(red = l-Ile, green = d-allo-Ile). (b) Layer with all four stereoisomers present (red = l-Ile, orange = l-
allo-Ile, green = d-allo-Ile, blue = d-Ile). The contacts illustrated in Figs. 3(a) and (b) are displayed
with yellow dots.
0
leaving all the C6 positions vacant.
7
26 Dalhus and G oÈ rbitz ꢀ 2-Amino-3-methylpentanoic acid
Acta Cryst. (2000). B56, 720±727

research papers
Table 5
Hydrogen-bond geometry (A, ).
3.2.5. D-allo-Ile (6). Each enantiomer in the racemate dl-
allo-Ile forms crystals with two crystallographically indepen-
dent molecules. Further, the hydrogen-bond arrangement is
different from that found in (1), (3) and (5), Table 5. The two
independent molecules in d-allo-Ile (6) (Figs. 1d and 2d),
Ê
ꢂ
a
DÐH, HÁ Á ÁA and DÐHÁ Á ÁA based on experimental H-atom positions;
b
Ê
HÁ Á ÁA for NÐH bonds normalized to 1.030 A (Taylor & Kennard, 1983).
a
b
DÐHÁ Á ÁA
DÐH
HÁ Á ÁA
HÁ Á ÁA
DÐHÁ Á ÁA DÁ Á ÁA
1,1
differ in the side-chain conformation; in molecule A ꢈ (N1Ð
dl-Ile (1)
�
i
C2ÐC3ÐC5) is gauche , while in molecule B the corre-
N1ÐH1Á Á ÁO2
N1ÐH2Á Á ÁO1
N1ÐH3Á Á ÁO2
0.95 (1) 1.97 (1) 1.886
0.85 (1) 1.91 (1) 1.725
0.91 (1) 2.10 (1) 1.988
164 (1)
175 (1)
157 (1)
2.888 (1)
ii
iii
2.752 (1) sponding torsion is trans. Furthermore, the carboxylate group
2.955 (1)
is approximately symmetric in molecule A, but clearly asym-
metric in molecule B, Table 4.
3.2.6. L-Ile:L-allo-Ile (4). Only extremely thin needles were
obtained upon crystallization of (4) and the crystal shape for
this complex is different from those of the individual amino
l-Ile:d-allo-Ile (3)
N1AÐH1AÁ Á ÁO2A
N1AÐH2AÁ Á ÁO1A
N1AÐH3AÁ Á ÁO2B
i
1.06 (4) 1.86 (3) 1.889
0.92 (3) 1.83 (3) 1.712
0.79 (2) 2.19 (2) 1.972
0.84 (2) 2.06 (2) 1.879
0.89 (2) 1.85 (2) 1.715
0.93 (2) 2.09 (2) 2.001
162 (2)
175 (2)
159 (2)
164 (1)
177 (2)
156 (2)
2.887 (1)
2.740 (1)
2.943 (1)
2.878 (1)
ii
iv
v
N1BÐH1BÁ Á ÁO2B
N1BÐH2BÁ Á ÁO1B
2.744 (1) acids l-Ile (G oÈ rbitz & Dalhus, 1996) and l-allo-Ile, which both
N1BÐH3BÁ Á ÁO2A
dl-Ile:dl-allo-Ile (5)
2.967 (1)
form plate-shaped crystals. This could very well indicate that
the molecular arrangement, and hence the hydrogen-bond
2.888 (1) pattern, in the 1:1 complex (4) is different from that of the
i
N1ÐH1Á Á ÁO2
0.91 (2) 2.00 (2) 1.889
0.89 (2) 1.85 (2) 1.715
0.82 (2) 2.16 (2) 1.956
164 (2)
175 (2)
165 (2)
ii
N1ÐH2Á Á ÁO1
2.742 (1)
2.958 (1)
individual amino acid structures. Unfortunately, the needles
were much too thin for diffraction experiments with a
conventional in-house diffractometer.
3.2.7. Hydrogen bonding. Experimental and normalized
(Taylor & Kennard, 1983) hydrogen-bond geometries are
vi
N1ÐH3Á Á ÁO2
d-allo-Ile (6)
ii
N1AÐH1AÁ Á ÁO2A
0.88 (2) 1.98 (2) 1.828
0.89 (1) 1.88 (1) 1.740
0.88 (2) 2.13 (2) 2.009
0.85 (2) 2.05 (2) 1.874
0.90 (1) 2.00 (1) 1.871
0.89 (1) 1.88 (1) 1.739
169 (1)
175 (1)
145 (1)
165 (1)
167 (1)
172 (2)
2.845 (1)
2.768 (1)
2.891 (1)
ii
N1AÐH2AÁ Á ÁO1B
vii
N1AÐH3AÁ Á ÁO1B
ii
N1BÐH1BÁ Á ÁO2B
2.879 (1) listed in Table 5. The hydrogen-bond patterns in (1), (3) and
i
N1BÐH2BÁ Á ÁO2A
2.881 (1)
2.761 (1)
(5) are isostructural, apart from the increased number of
viii
N1BÐH3BÁ Á ÁO1A
hydrogen bonds due to the lower symmetry in (3). This
hydrogen-bond arrangement recurs in other racemates of
hydrophobic amino acids with branched side chains; dl-
leucine (Di Blasio et al., 1975) and dl-valine (triclinic poly-
morph: Dalhus & G oÈ rbitz, 1996; monoclinic polymorph:
²
Symmetry codes: (i) x ‡ 1; y; z; (ii) x; y � 1; z; (iii) � x ‡ 1; � y ‡ 1; � z ‡ 1; (iv)
1
2
x � 1; y; z; (v) x; y ‡ 1; z; (vi) � x; � y ‡ 1; � z ‡ 1; (vii) � x ‡ 2; y � ; � z ‡ 1; (viii)
1
2
�
x ‡ 2; y ‡ ; � z ‡ 1.
ii
Ê
0
The short C6Á Á ÁC6 distance of 3.315 (6) A in (5) (Fig. 2a)
must then be increased to permit the presence of methyl
groups in all the C6 positions. Indeed, this is precisely what
Mallikarjunan & Thyagaraja Rao, 1969). The amino H1 (C Ð
ꢀ
C ÐNÐH = gauche ) and H2 (C ÐC ÐNÐH trans) atoms in
+
0
ꢀ
(1), (3) and (5) have normalized hydrogen-bond distances
within ranges of 0.010 A (1.879±1.889 A) and 0.013 A (1.712±
iv
happens in the crystal of dl-Ile; the C6Á Á ÁC6 [symmetry code:
Ê
Ê
Ê
Ê
(
3
iv) � x ‡ 1; � y; � z ‡ 2] distance is increased by 0.5 A to
Ê
.725 A), respectively.
1
Ê
.810 (3) A (Fig. 2c).
This repositioning of the terminal C6 methyl groups is
The purchase of a Siemens SMART diffractometer was
made possible through ®nancial support from The Research
Council of Norway (NFR).
feasible without gross alterations in the molecular arrange-
ment (Figs. 2a and c) and the hydrogen-bond arrangement in
the hydrophilic layer is left intact, Table 5.
3
structure of (5) is used as a model template for the generation
.2.4. Spontaneous resolution of DL-allo-Ile (2). If the
References
0
of the structure of dl-allo-Ile ± by occupying all C6 positions
and leaving all the C6 positions vacant ± a molecular packing
Allen, F. H. & Kennard, O. (1993). Chem. Des. Autom. News, 8, 1, 31±
3
7.
Benedetti, E., Pedone, C. & Sirigu, A. (1973). Acta Cryst. B29, 730±
33.
0
0
ii
Ê
with a short C6 Á Á ÁC6 contact of only 2.45 A is obtained (Fig.
7
iv
Ê
2
a). In dl-Ile a shift of 0.5 A in the C6Á Á ÁC6 interaction was
Dalhus, B. & G oÈ rbitz, C. H. (1996). Acta Cryst. C52, 1759±1761.
Dalhus, B. & G oÈ rbitz, C. H. (1999). Acta Cryst. C55, 1547±1555.
Di Blasio, B., Pedone, C. & Sirigu, A. (1975). Acta Cryst. B31, 601±
achieved by minor modi®cations in the molecular packing,
Ê
while in dl-allo-Ile the required shift is approximately 1.5 A.
6
02.
The observed separation of the enantiomers upon crystal-
lization suggests that it is not possible to rearrange the
molecules in this hypothetical dl-allo-Ile structure and elim-
G oÈ rbitz, C. H. & Dalhus, B. (1996). Acta Cryst. C52, 1464±1466.
Mallikarjunan, M. & Thyagaraja Rao, S. (1969). Acta Cryst. B25, 296±
303.
0
0
ii
inate the unfavorable C6 Á Á ÁC6
interaction without
Sheldrick, G. M. (1995). SHELXTL. Version 5.03. Siemens
Analytical X-ray Instruments Inc., Madison, Wisconsin, USA.
Sheldrick, G. M. (1996). Sadabs. University of G oÈ ttingen, Germany.
Siemens (1995). SMART/SAINT. Siemens Analytical X-ray Instru-
ments Inc., Madison, Wisconsin, USA.
Taylor, R. & Kennard, O. (1983). Acta Cryst. B39, 133±138.
Varughese, K. I. & Srinivasan, R. (1975). J. Cryst. Mol. Struct. 5, 317±
328.
disrupting the hydrogen-bond pattern. It is thus possible to
account for the spontaneous resolution of the enantiomers in
dl-allo-Ile at the molecular level. The molecular arrangement
in the enantiomeric structure (6), with an alternative
hydrogen-bond arrangement, is energetically favored (Fig.
2d).
Acta Cryst. (2000). B56, 720±727
Dalhus and G oÈ rbitz ꢀ 2-Amino-3-methylpentanoic acid 727