research papers

Acta Cryst. (2012). B68, 549–557 doi:10.1107/S0108768112033484 549

Acta Crystallographica Section B

StructuralScience

ISSN 0108-7681

Single-crystal investigation of L-tryptophan with Z000 =16

Carl Henrik Gorbitz,a*

Karl Wilhelm Tornroosb and

Graeme M. Dayc

aDepartment of Chemistry, University of Oslo,

1033 Blindern, Oslo N-0315, Norway,bDepartment of Chemistry, University of Bergen,

Allegt. 41, Bergen N-5007, Norway, andcDepartment of Chemistry, University of

Cambridge, Lensfield Road, Cambridge CB2

1EW, England

Correspondence e-mail:

c.h.gorbitz@kjemi.uio.no

# 2012 International Union of Crystallography

Printed in Singapore – all rights reserved

A complex, disorder-free structure in the space group P1 has

been established for l-tryptophan, for which no crystal

structure has previously been available. The 16 molecules in

the asymmetric unit can be divided into two groups of eight;

one where the side chains have gauche orientations and one

with trans orientations. Molecules within each group have

almost identical molecular geometries. The unit-cell para-

meters mimic a hexagonal cell, but deviations from 90� for the

cell angles � = 84.421 (4) and � = 87.694 (4)� give a small tilt

that rules out hexagonal symmetry. The hydrogen-bonding

pattern resembles that found in the crystal structure of the

racemic structure of dl-tryptophan, but a lower density

combined with longer hydrogen bonds and inter-aromatic

interactions show that the enantiomeric structure is less

efficiently packed.

Received 13 June 2012

Accepted 24 July 2012

1. Introduction

Due to their obvious interest as the building units of proteins

and their role in many metabolic processes, amino acids were

among the first chiral compounds to be investigated with X-

ray diffraction methods (the structure of achiral glycine was

determined by Albrecht & Corey from an incomplete data set

in 1939). In 1950, when Shoemaker et al. published their work

on l-threonine, just 14 other detailed structures of organic

molecules crystallizing in Sohnke space groups were known

[according to the Cambridge Structural Database (CSD);

version 5.33 of November 2011; Allen, 2002], of which only

half dealt with chiral substances (i.e. not achiral or meso

forms). More amino acid structures appeared in the following

years (as hydrates for l-asparagine and l-arginine), the

sequence to some extent reflecting the ease with which high

quality crystals could be obtained. The initial rush thus ended,

in 1976, with the 17th amino acid, l-leucine, notorious for its

thin, flaky crystals. After a 14 year break (or 10 if the deter-

mination of the � form of l-glutamic acid is considered), a

crude structure (R factor 0.147) was revealed for phenylala-

nine (as the d enantiomer), for which crystals are plagued by

various, as yet undetermined, types of stacking disorders.

Recently, anhydrous structures were finally obtained for both

l-asparagine (Yamada et al., 2007) and l-arginine (Courvoisier

et al., 2012), meaning that prior to the present investigation

pure crystal structures were known for 18 out of the 20 stan-

dard amino acids, with l-tryptophan (l-Trp) and l-lysine still

missing (Table 1).

l-Trp has previously been investigated in complexes with

acetic acid (as a hydrate, GOMDAO: Li et al., 2009), formic

acid (MUGKAA: Hubschle et al., 2002; MUGKAA01: Scheins

et al., 2004), pyridine-2,4-dicarboxylic acid (as an ethanol

solvate, NUQHIR: Di, 2010) and d-(R)-mandelate (as a 1.5

hydrate, UGITAG: Fujii, 2009). Unravelling the solid-state

structure for pure l-Trp has, however, proved very difficult.

Khawas & Murti (1969) reported a unit cell with the para-

meters a = 16.81, b = 17.90 and c = 6.90 A in the orthorhombic

space group Pmmm (which in retrospect seems a bit strange

for a chiral compound). We have now obtained a single-crystal

structure of l-Trp (I), which is described here and compared

with the racemate dl-Trp (II) (space group P21/c with a =

18.90, b = 5.74, c = 9.31 A, � = 101.8�, CSD refcode

QQQBTP01: Bakke & Mostad, 1980; QQQBTP02: Hubschle

et al., 2004).

l-Trp is an essential amino acid for humans, meaning that it

must be part of our diet. It is the least common of the standard

amino acids in proteins with an average occurrence of about

1.3%, but Trp residues play important roles in protein stability

and recognition. Additionally, Trp is a biochemical precursor

for serotonin (a neurotransmitter), melatonin (a neuro-

hormone) and niacin (vitamin B3 or nicotinic acid).

2. Experimental

2.1. Crystal preparation

From a saturated solution of (I) in water (approximately

10 mg ml�1), 30 ml was deposited into a series of 30 � 6 mm

test tubes that were subsequently sealed with Parafilm1. A

needle was then used to prick a single small hole in the

Parafilm1 of each tube, after which it was placed inside a

larger test tube filled with 1 ml of acetonitrile. The systems

were ultimately capped and left for 3 d at 293 K. l-Trp regu-

larly precipitated as thin flakes, but one tube produced some

crystals of a different habit, as rhombohedral prisms, and the

largest specimen was used for data collection.

2.2. Data collection and structure refinement

Data collection was carried out on a Bruker AXS APEX II

ULTRA rotating anode Pt135 CCD diffractometer, using

graphite-monochromated Mo K� radiation (� = 0.71073 A)

for 0.3� scans over 182� in ! in four orthogonal ’ positions.

Structure solution was carried out by direct methods

combined with iterative difference-Fourier synthesis (Shel-

drick, 2008).

The structure, with 16 molecules labelled A–P in the

asymmetric unit, was refined without constraints or restraints

on C, N or O positions. No positional disorder was indicated.

H atoms were introduced in theoretical positions with fixed

N—H = 0.88 (aromatic) or 0.91 A (amino) and C—H = 0.95

(aromatic), 0.99 (methylene) or 1.00 A (methine), while

permitting free rotation for the amino groups. Uiso values were

set to 1.2Ueq of the carrier atom, or 1.5Ueq for amino groups.

In the absence of significant anomalous scattering effects,

23 163 Friedel pairs were merged. Experimental details are

given in Table 2.

Alternative refinements utilized a significantly reduced

number of refinement parameters through the employment of

SHELX EADP commands, forcing the use of the same

anisotropic displacement parameters for atoms related by

pseudotranslational symmetry. One such model (CIF available

as supplementary material1) yielded an R value of 0.088 for

1338 parameters. The estimated standard uncertainties for

calculated geometrical parameters were unchanged compared

with the unconstrained refinement, or even increased

slightly.

research papers

550 Carl Henrik Gorbitz et al. � Single-crystal investigation Acta Cryst. (2012). B68, 549–557

Table 1First complete single-crystal structure determinations (including atomcoordinates) of the 20 standard amino acids according to the CambridgeStructural Database (CSD; Version 5.33 of November 2011; Allen, 2002).

Amino acid† CSD refcode Reference

l-Alanine LALNIN Simpson & Marsh (1966)l-Arginine‡ – Courvoisier et al. (2012)l-Aparagine§ VIKKEG Yamada et al. (2007)l-Aspartic acid} LASPRT Derissen et al. (1968)l-Cysteine (mon) LCYSTN Harding & Long (1968)l-Cysteine (ort) LCYSTN21 Kerr & Ashmore (1973)l-Glutamine GLUTAM Cochran & Penfold (1952)l-Glutamic acid (�) LGLUAC03 Lehmann & Nunes (1980)l-Glutamic acid (�) LGLUAC Hirokawa (1955)Glycine (�)†† GLYCIN02 Marsh (1958)Glycine (�) GLYCIN Iitaka (1960)Glycine (�) GLYCIN01 Iitaka (1961)l-Histidine (ort) LHISTD10 Madden, McGandy &

Seeman (1972)l-Histidine (mon) LHISTD01 Madden, McGandy,

Seeman, Harding &Hoy (1972)

l-Isoleucine LISLEU Torii & Iitaka (1971)l-Leucine LEUCIN Harding & Howieson

(1976)l-Lysine – –l-Methionine LMETON10 Torii & Iitaka (1973)d-Phenylalanine SIMPEJ Weissbuch et al. (1990)l-Proline‡‡ PROLIN Kayushina & Vainshtein

(1965)l-Serine§§ LSERIN10 Benedetti et al. (1973)l-Threonine LTHREO Shoemaker et al. (1950)l-Tryptophan - This workl-Tyrosine LTYROS10 Mostad et al. (1972)l-Valine LVALIN Torii & Iitaka (1970)

† Polymorphism is indicated by either (mon) and (ort) for monoclinic and orthorhombicforms, respectively, or by greek letters. ‡ From powder data, not yet in the CSD,dihydrate (ARGIND): Karle & Karle (1964). § Monohydrate (ASPARM): Kartha &de Vries (1961). } Monohydrate (IJEQET): Umadevi et al. (2003). †† Solved andrefined from an incomplete data set by Albrecht & Corey (1939). ‡‡ Monohydrate(RUWGEV): Janczak & Luger (1997). §§ Monohydrate (LSERMH): Frey et al.(1973).

1 Supplementary data for this paper are available from the IUCr electronicarchives (Reference: EB5018). Services for accessing these data are describedat the back of the journal.

3. Results and discussion

3.1. Molecular structure, asymmetric unit and unit cell

The asymmetric unit and the unit cell are shown in Fig. 1.

The molecular conformation of l-Trp is essentially described

by the two torsion angles N1—C2—C3—C5 (�1), which may

be gauche+, trans or gauche�, and C2—C3—C5—C4 (�2),

which may be positive or negative, giving rise to a total of six

basic conformations (combinations of �1 and �2). The torsion

angles listed in Table 3 show that the 16 l-Trp molecules in the

structure of (I) populate only two of these conformations;

molecules A, B, E, F, I, J, M and N have �1 = trans and �2’

�114� (called the T family), while molecules C, D, G, H, K, L,

O and P have �1 = gauche and �2’ 112� (G family). The

neighbouring molecules B and C, shown in Fig. 2, are repre-

sentative examples of both conformations.

Each conformation found in the structure of (I) has been

observed for l-Trp in two previous structures, while the formic

acid solvate MUGKAA01 (Scheins et al., 2004) with �1 =

gauche+ and negative �2 provides the only example of a third

l-Trp conformation.

The very narrow ranges for torsion angles given in Table 3

and a molecular overlay diagram (supplementary material)

illustrate that molecular geometries within each of the

conformational families are remarkably similar. Side-chain

geometries and orientations are hardly distinguishable, while a

limited rotation is seen for the amino group, which is deter-

mined by a single refinement parameter (no s.u.’s available

owing to constrained refinement). The calculated r.m.s.

deviation for the best fit between O, N and C atoms in pairs of

molecules with the same conformation is typically around

0.03 A.

Fig. 1 shows that the structure of (I) has a layered

construction. Hydrogen bonding occurs in two separate

hydrophilic layers, including the polar heads of molecules A–

G and I–P, respectively. Between the hydrophilic layers are

hydrophobic bilayers of the side chains, one composed of the

side chains from molecules E–L and one from the side chains

of molecules A–D and M–P.

research papers

Acta Cryst. (2012). B68, 549–557 Carl Henrik Gorbitz et al. � Single-crystal investigation 551

Figure 1The asymmetric unit, chosen so that all 16 molecules are within the unitcell, viewed approximately along the ab diagonal. H atoms bonded to Chave been omitted for clarity. Molecules of the T family (trans side-chainorientation, see text) are shown with dark grey C atoms; those of the Gfamily have light grey C atoms. Hydrogen bonds appear as dotted lines.The side chain of molecule G has been depicted in wireframerepresentation to relieve overlap with molecule F; an extra copy ofmolecule G with all atoms in white, called G0, at (x; 1þ y; z) is included.The blue, dashed line represents the interface at the centre of thehydrophobic bilayer, while the open arrow shows a pseudo twofold screwaxis. Various terms used in describing the structure are given at thebottom.

Table 3Torsion angles (�) for l-Trp in crystal structures.

Refcode† �1‡ �2‡ Ref

(I)§ �173.4, �172.9 (4),�174.1 (4)

�113.8, �112.2 (6),�115.2 (6)

–

(I) �79.0, �77.8 (5), �80.0 (5) 111.6, 109.6 (6), 113.6 (6) –(II) �166.8 �107.9 (a)GOMDAO �62.1 103.3 (b)MUGKAA01 53.5 �104.1 (c)NUQHIR �58.5 114.4 (d)UGITAG �164.4 �102.3 (e)

References: (a) Hubschle et al. (2004); (b) Li et al. (2009); (c) Scheins et al. (2004); (d) Di(2010); (e) Fujii (2009). † CSD (Allen, 2002). ‡ For (I) �1 and �2 are N1—C2—C3—C5 and C2—C3—C5—C4, respectively (see Fig. 2). § First line is for the T family,second line is for the G family. For each torsion angle the average for eight molecules isgiven, followed by the extreme values (in italic type face with calculated s.u.s).

Table 2Experimental details.

Crystal dataChemical formula C11H12N2O2

Mr 204.23Crystal system, space group Triclinic, P1Temperature (K) 123a, b, c (A) 11.430 (3), 11.464 (4), 35.606 (9)�, �, � (�) 84.421 (4), 87.694 (4), 60.102 (2)V (A3) 4025.6 (19)Z, Z0 16Radiation type Mo K�� (mm�1) 0.095Crystal size 0.62 � 0.28 � 0.14

Data collectionDiffractometer Bruker APEXII ULTRA CCDAbsorption correction Multi-scanTmin, Tmax 0.943, 0.987No. of measured, independent and

observed [I > 2�(I)] reflections66 471, 24 736, 19 659

Rint 0.043(sin /�)max (A�1) 0.721

RefinementR[F2 > 2�(F2)], wR(F2), S 0.085, 0.255, 1.13No. of reflections 24 736No. of parameters 2178No. of restraints 3H-atom treatment H-atom parameters constrained�max, �min (e A�3) 0.53, �0.47

Computer programs used: APEX2 (Bruker, 2007a), SAINT-Plus (Bruker, 2007b),SHELXTL (Sheldrick, 2008), SADABS (Sheldrick, 1996).

3.2. Hydrogen bonding

Fig. 3 shows the hydrogen-bonding pattern in the crystal

structure of (I). Amino acids with bulky side chains are often

unable to form good hydrogen-bonding patterns on their own,

presumably as this would introduce unacceptable steric

conflict, and rather depend on incorporating water molecules

as integral parts of the network. Examples include non-

aromatic compounds like tert-butylglycine monohydrate and

neopentylglycine monohydrate (Weissbuch et al., 1990)

[neopentyl is —CH2—C—(CH3)3] as well as amino acids with

aromatic side chains like 3-fluoro- and 2,4,5-trifluor-

ophenylalanine monohydrates (In et al., 2003), 4-nitropheny-

lalanine monohydrate (Dai & Fu, 2008) and �-

methyltryptophan monohydrate (Cyr et al., 1999). l-Trp can

evidently do without water, but forms a pattern that is distinct

from those found for other layered amino acid structures

(Gorbitz et al., 2009). Three types of hydrogen

bonds are present in Fig. 3(a) (as well as in the

qualitatively identical pattern in Fig. 3b), forming

what can be described as a series of tapes with a

repeating unit of eight molecules, e.g.

E:A:H:D:F:B:G:C, utilizing hydrogen-bond type 1

with C1—C2—N1—H1 = gauche� and type 2

with C2—C1—N1—H3 = gauche+. Notably,

molecules with similar conformations occur in

pairs along the tapes, which are subsequently

interconnected by type 3 interactions with C1—

C2—N1—H2 = trans into perpendicular head-to-

tail chains after an off-set of four amino acids in

the tape direction. This generates E:F:E:F and

other equivalent sequences in Fig. 3(a), with

research papers

552 Carl Henrik Gorbitz et al. � Single-crystal investigation Acta Cryst. (2012). B68, 549–557

Figure 3Hydrogen bonding within the two distinct hydrophilic layers in the crystalstructure of (I) with reference to Fig. 1. Side chains are not shown beyondC� (C3 in Fig. 2). The colour coding for C atoms is similar to Fig. 1, whiletwo different colours have been used in an arbitrary fashion for N and Oatoms to give each molecule in each layer a unique appearance. Views areperpendicular to the ab plane, i.e. along c*. The three numbers in (a)identify three types of hydrogen bonds, while the rectangles highlight ahydrogen-bonded tape (red) and a head-to-tail chain (blue).

Figure 2Molecule B with a trans orientation for �1 (N1—C2—C3—C5) andmolecule C with a gauche orientation. Displacement ellipsoids have beendrawn at the 50% probability level; H atoms are shown as spheres ofarbitrary size.

Table 4Hydrogen-bond distances (A) and angles (�) in the structures of (I) and (II) (Hubschleet al., 2004).

Type† H� � �O N� � �O N—H� � �O

l-Trp (I)N1—H1� � �O1 1 1.94, 1.90, 1.98‡ 2.804, 2.784 (6), 2.833 (5) 159, 152, 165N1—H3� � �O1 2 2.00, 1.97, 2.04 2.902, 2.870 (6), 2.928 (6) 171, 161, 179N1—H2� � �O2 3 1.82, 1.79, 1.86 2.711, 2.692 (6), 2.735 (6) 167, 161, 177

dl-Trp (II)§N15—H15A� � �O13 1 1.91} 2.814 172}N15—H15B� � �O13 2 1.94 2.826 165N15—H15C� � �O14 3 1.80 2.707 175

† As identified in Figs. 3 and 4. ‡ For each parameter the average for 16 interactions is given, followedby the extreme values (in italic type face with calculated s.u.s). § Atomic numbers retained from theoriginal contribution. } Values listed are after normalization of the N—H bond lengths to 0.91 A as usedin the refinement of (I).

adjacent head-to-tail chains having opposite directions.

Interestingly the hydrogen-bonding pattern of the racemate

(II), depicted in Fig. 4(a), resembles the pattern of (I) despite

the fact that the characteristic tapes are missing. The dimer in

Fig. 4(a) is centrosymmetric and thus could not occur for l-

Trp, but head-to-tail chains are nevertheless very similar in

both structures, the main difference being that the orientation

of chains occurs in pairs for (II), i.e. ""##"" rather than "#"

for (I). The hydrogen bond lengths of (I) and (II) are

compared in Table 4 [a complete description of hydrogen-

bond geometry for (I) is available as supplementary material].

The hydrogen bond of type 2, with H�—C�—N—H = trans, is

clearly shorter for (II) than for (I). For the other two inter-

actions there are no significant differences. The type 1 inter-

action is more linear for (II) than for (I), but not shorter.

The structure of the achiral compound 8-amino-1,4-diox-

aspiro[4.5]decane-8-carboxylic acid (JAPJUF: Vela et al.,

1989), in space group Pna21, is also related to (I) with the

hydrogen-bonding pattern shown in Fig. 4(b). The only

essential modification compared with (I) is that the carboxy-

late O atom involved in the head-to-tail chains also accepts a

second H atom, while for (I) this O atom is not involved in the

tape motif.

3.3. Aromatic interactions and pseudosymmetry structure,asymmetric unit and unit cell

If all l-Trp molecules had the same side-chain orientation,

the screw axis along the ab diagonal involving molecules I–P

readily visible in Fig. 1, would have been proper rather than

pseudo (a similar pseudo-twofold screw along the b axis

operates on the polar heads of the molecules A–H). Evidently,

such a simple arrangement is not compatible with an efficient

stacking of the aromatic side chains, as the approximate top

view in Fig. 5 shows that side chains instead are related by

pseudo-glide planes. The presence of such pseudo-glide planes

is rendered possible by the occurrence of two different side

chain conformations in the structure of (I), and results in a

research papers

Acta Cryst. (2012). B68, 549–557 Carl Henrik Gorbitz et al. � Single-crystal investigation 553

Figure 5A pseudo twofold screw axis for molecules I–P operating on the polarheads of the amino acid molecules (inside circle). The hydrophobic sidechains are instead related by horizontal pseudo-glide planes (dashedlines). In this illustration molecules belonging to different conformationalfamilies are not only distinguished by the colour of the C atoms, butmembers of the T family also have O and N atoms in darker colour.

Figure 6(a) Representative section of a hydrophobic region in the structure of (I)showing interactions between the indole groups of the Trp side chains ofmolecules A, B, C and D. C atoms are coloured according to side-chainconformation as in Fig. 1, while two different colours have been used forN atoms to distinguish individual molecules. (b) Interactions betweenside chains in the structure of (II) (Hubschle et al., 2004) with l or d givingamino acid chirality. The H� � �C(�) distances (in A) were measured afternormalization of the N—H bond to 0.88 A and are directly comparablewith the corresponding distances in (a). The dashed horizontal lines showa true crystallographic glide plane in (b) and a pseudo glide plane in (a).

Figure 4Hydrogen-bonding patterns in the crystal structures of (a) dl-Trp (II)(Hubschle et al., 2004) and (b) an aminocyclohexanecarboxylic acid (Velaet al., 1989), both curtailed beyond C�. Labels and coloured rectanglesappear as in Fig. 3: (a) with an additional hydrogen-bonded dimerhighlighted in (a).

stacking arrangement of aromatic groups that is amazingly

similar to that found for the racemate (II), Fig. 6.

The A and B molecules in Fig. 6(a) belong to the T family

and have the same function as the conformationally identical

l-enantiomers in Fig. 6(b), while the C and D molecules

belong to the G family and play the role of the D enantiomers

in (II). The representative lengths of the two shortest N—

H� � �C(�) interactions indicate, however, that hydrophobic

groups are slightly less efficiently packed in the crystal struc-

ture of the pure enantiomer. Together with the longer type 2

hydrogen bond in (I), this contributes to a slightly lower

density for (I) than for (II), calculated values being 1.348 (T =

123 K) and 1.371 g cm�3 (T = 173 K), respectively.

3.4. Unit cell revisited: crystal construction principles

With only eight entries in the CSD (Allen, 2002), crystal

structures of organic molecules with Z0 � 16 are exceedingly

rare. Accordingly, whenever a crystal structure with a Z0 value

of this magnitude is presented, suspicion will immediately be

raised concerning potential failure to include missing

symmetry. It is therefore imperative to obtain a fundamental

understanding of the crystal build-up of (I) to verify that the

P1 space group with Z0 = 16 is not an artefact.

The stack of molecules I–P in Figs. 1 and 6 is essentially

obtained from two similar repeating units M:I:O:K and

N:J:P:L with conformations T:T:G:G. Thus, the operational

structural unit of (I) consists of four independent molecules.

Potentially, such units could be used to build triclinic, mono-

clinic or even orthorhombic crystal structures with Z0 = 4, all

retaining the hydrogen-bonding pattern and aromatic inter-

actions (within hydrophobic monolayers) of (I). These hypo-

thetical packing arrangements furthermore have in common

that they would include just a single type of hydrophobic

bilayer or interface, which is the default for enantiomeric and

racemic layered amino acid structures. This includes the

structure of (II), compared in Fig. 7 with the structure of

(I).

In Fig. 8 the focus is on the dashed hydrophobic interfaces

in Fig. 7. The surfaces of (II) shown in Figs. 8(a) and (b) are

related by inversion symmetry, which reverses the indicated

directions of the C�—C� bond vectors (for enantiomeric

structures a reversal of monolayer directions within a double

layer is usually accomplished by a twofold screw operation).

Putting the two surfaces together in (c) makes the ‘knobs-in-

holes’ fit easily visible. In contrast, the surface generated by

molecules E, F, G and H in Fig. 8(e) is qualitatively obtained

from the IJKL surface in Fig. 8(d) by first a 180� rotation

around the horizontal x axis (C�—C� bond vector) and then

an in-plane anti-clockwise 60� rotation around the z axis. If the

resulting interface shown in detail in Fig. 8(f) and labelled 1 in

Fig. 7(b) had been the only way to fit the two surfaces, the

structure of (I) would have belonged to the hexagonal space

group P65 with Z0 = 8 (two units of four molecules) and

approximate cell parameters a0 = b0 = 11.44 A [= (a + b)/2] and

c0 = 106.22 A (= c�sin ��sin ��3). There appears to be no

structures in the CSD with hexagonal symmetry and such a

�60� twist between distinct layers, but the dimeric amino acid

l-cystine (Dahaoui et al., 1999), in space group P6122 with cell

parameters a = b = 5.412 (1) and c = 55.956 (1) A, gives an

impression of such an arrangement, a conceptual difference to

(I) being that hydrophobic bilayers are covalently linked by

S—S bonds.

What makes the structure of (I) particularly intriguing,

however, is that there is not just a single way to fit two

hydrophobic surfaces, but also a second alternative that is

operational at interface 2 in Fig. 7(b). An understanding of the

subtle difference between 1 and 2 can be gained by comparing

Figs. 8(d)–(f) with Figs. 8(g)–(i). The IJKL surface in Fig. 8(d)

and the ABCD surface in Fig. 8(g) are indistinguishable. The

same is also true for the EFGH and MNOP surfaces in Figs.

9(e) and (h), but while the final in-plane rotation in Fig. 8(e) to

give the fit shown in Fig. 8(f) is anti-clockwise, it is clockwise in

Fig. 8(h), resulting in a new fit shown in Fig. 8(i). In this way all

16 independent molecules (four units of four molecules)

obtain a unique set of neighbours with respect to inter-

molecular interactions. Note for instance in Fig. 8 how mole-

cules A, B, I and J, which all belong to the T family, have

dissimilar contacts with overlapping molecules O (G), N (T),

research papers

554 Carl Henrik Gorbitz et al. � Single-crystal investigation Acta Cryst. (2012). B68, 549–557

Figure 7(a) Hydrophobic bilayers in the crystal structure of (II), viewed along theb axis. For each bilayer C atoms in the top and bottom monolayer arecoloured in light grey and dark grey, respectively. The pink, dashed lines,separating structurally identical surfaces, show the central interface ineach unit cell. Hydrophilic layers appear as shaded rectangles (amino andcarboxylate groups are not shown), while C� atoms are small spheres andC�—C� bonds are yellow. The arrow shows the viewing direction in Fig. 8.(b) Similar for (I) with a view along the a axis. (Note: in Figs. 7 and 8 side-chain colour does not reflect conformation.) For each monolayer themolecules contributing a side chain are identified, e.g. A, B, C, D for thelayer at the top. Two distinct interfaces, labelled 1 and 2, separatestructurally different surfaces.

E (T) and H (G), respectively. The � 60� rotation of layers

leaves the triclinic unit cell of (I) some hexagonal-like traits,

like a and b axes of about the same length and a � angle close

to 60� (Table 2), but as the alternating +60 and �60� rotations

along the c axis operate with centres of rotation that are

slightly translocated in the other two directions, � and � are

shifted significantly away from 90� (Table 2).

Two other structures have crystallized in the space group P1

with Z0 = 16 (2,2-aziridine-dicarboxamide, BIPCOS01:

Bruckner, 1982; cholesterol at 310 K, CHOEST21, Hsu et al.,

2002), but none are divided into hydrophobic and hydrophilic

layers with this kind of relationship between groups of

molecules.

The mystery of (I) is of course why and how hardly

distinguishable surfaces should interact in two different ways

in a crystal structure, and indeed doing so in a fully ordered

manner (neither twinning nor disorder was indicated in the

refinement).

3.5. Systematic absences in the diffraction pattern

The presence of pseudo-screw axes and glide planes

described above leads to certain groups of reflections being

systematically weak or absent, but much more important for

the observed diffraction pattern is translational pseudosym-

metry. It is useful in this connection to regard first each of the

two blocks in Fig. 1 independently, and then turn to the

complete structure.

From Fig. 3(a) it is clear that the A–H block has close to

perfect translational symmetry along the vertical b axis,

effectively cutting the repeat unit into b/2. Reflections with k

odd accordingly do not have a contribution from this block,Fig. 9(a). In a similar manner, the

pattern shown in Fig. 4(b) is face-

centred, meaning that it gives no

contributions to reflections with h +

k odd, Fig. 9(b). When combining

information from Figs. 9(a) and (b)

in Fig. 9(c), we find that reflections

that do not fulfil any of the condi-

tions hkl: k = 2n or hkl: h + k = 2n,

e.g. (0,1,0) or (2,1,1), do not have

significant contributions from any of

the two blocks and thus are missing.

This explains the unusual

‘systematic absences’ for the

triclinic space group P1 observed in

the diffraction pattern of (I), Fig.

9(d).

3.6. Search for polymorphs

To establish whether the single

crystal used for data collection was

the result of a freak occurrence of

another polymorph than usually

obtained for l-Trp, additional crys-

tallizations were performed from

various solvents, uniformly giving

the familiar, very thin flakes. Cell

parameters determined form two

such specimens matched those

listed in Table 2.

Furthermore, X-ray powder

diffraction data were collected for a

typical sample, and the observed

pattern fitted the simulated pattern

of the single-crystal structure [illus-

tration available as supplementary

material, room temperature cell

parameters are a = 11.5223 (13), b =

11.5156 (12), c = 35.877 (6) A, � =

84.393 (9), � = 87.681 (7), � =

research papers

Acta Cryst. (2012). B68, 549–557 Carl Henrik Gorbitz et al. � Single-crystal investigation 555

Figure 8(a) and (b) Monolayers in the structure of (II). The arrows indicate the directions of the C�—C� bondvectors; l and d give amino acid chirality. (c) Fit between the two monolayers to give a full bilayer, atomsin the bottom monolayer (a) are displayed here as a space-fill representation. (d)–(f) Similar drawingsfor the two independent bilayers in the structure of (I). See text for details.

59.987 (8)�, V = 4102.1 (10) A3]. Although other polymorphs

certainly may exist, we thus see no indication of such in our

experiments.

3.7. A modulated structure?

Structures of organic compounds with high Z0 values may

sometimes be described (and potentially better understood) as

commensurately (or incommensurately) modulated structures

while applying the higher-dimensional superspace approach

(Schonleber, 2011). Examples include 4,40-dimethyl-2-

hydroxy-6-oxocyclohexene-1-carboxylic acid with Z0 = 2.5

(five half-molecules; Duncan et al., 2002) as well as various

crown complexes such as [Cu(H2O)2(15-crown-5)](NO3)2 with

Z0 = 10 (Schonleber & Chapuis, 2004; Hao et al., 2005). It was

thus appropriate to check whether this could also be the case

for the structure of (I).

Several tests were made using JANA2006 (Petrıcek et al.,

2006), with the conclusion that there is in fact no reasonable

way to apply the superspace theory and handle the structure

as modulated. There are 16 independent molecules in the

space group P1, in two different conformations (see above).

We would then be looking for some unified pseudo-translation

between A, B, E, F, I, J, M, N (T family) and similarly for C, D,

G, H, K, L, O, P (G family). Both sets have similar relation-

ships to the leading molecule. In the molecular refinement the

A molecule was used as a motive with which the remaining

family members were related by rotations and translations

A 0; 0; 0; xA; yA; zA

B 0; 0; 0; xA; yAþ 1=2; zA

E �60; 0; 180; xE; yE; zE

F �60; 0; 180; xE; yEþ 1=2; zE

I �60; 0; 0; xI; yI; zI

J �60; 0; 0; xI þ 1=2; yI þ 1=2; zI

M 0; 0; 180; xM; yM; zM

N 0; 0; 180; xM þ 1=2; yM þ 1=2; zM

So for A + B and E + F we have the pseudo-translation

(0,1/2,0), which really would lead to supercell effects and the

possibility of a description as being modulated, but for I + J

and M + N the pseudo translation is different (1/2,1/2,0).

Accordingly, there is no chance to use the formalism for

modulated structures. A similar scheme is valid for molecules

of the G family.

4. Conclusion

The elusive solid-state structure of

l-tryptophan (l-Trp) has been

determined from a single-crystal X-

ray analysis based on diffraction

data collected from a specimen of

unusually high quality. The packing

arrangement, incorporating a

hydrogen-bonding pattern not

previously observed for amino

acids, mimics the structure of dl-

Trp, where fully ordered l-Trp

molecules belonging to two

different conformations called T

and G play the roles of the l- and

d-enantiomers in the crystal struc-

ture of the racemate. This gives rise

to extended pseudo-symmetry and

together with the occurrence of two

independent types of hydrophobic

interfaces in the crystal bring the

number of molecules in the asym-

metric unit up to 16, a very rare

phenomenon indeed. There is no

indication that this is in fact a

modulated structure. From a theo-

retical point of view, the packing is

unique in that it, for the triclinic

space group P1, gives apparent

systematic absences based on the

condition hkl: h + k = 2n or k = 2n.

To investigate the energy gained

by l-Trp in increasing Z0 from 1 or

research papers

556 Carl Henrik Gorbitz et al. � Single-crystal investigation Acta Cryst. (2012). B68, 549–557

Figure 9(a)–(c) Simulated diffraction patterns of (I) for a hk layer. Atoms in the A–H block give contributionsonly to the reflections coloured in blue in (a), while the I–P block (b) give rise to the red flections in (b).In the combined pattern in (c) reflections with contributions from both blocks are coloured in black. (d)A projection of the experimentally observed hk0 layer.

2, as seen for the other enantiomerically pure amino acids, to 8

or 16, and to study the alternative low Z0 structures that are

available to the molecule, we are in the process of performing

extended structure prediction calculations and energy calcu-

lations on the observed structure described here. Crystal

structures with multiple, flexible molecules in the asymmetric

unit pose a particular challenge for theoretical structure

prediction methods, usually based on global lattice energy

minimization (Day, 2011), due to the high dimensionality of

the energy surface that must be explored (van Eijck & Kroon,

2000). These studies will be discussed in a future paper, now in

preparation.

The authors thank Dr David Wragg, Department of

Chemistry, University of Oslo, Norway, for collecting and

refining the powder XRD data, and Dr Vaclav Petrıcek,

Institute of Physics, Academy of Sciences of the Czech

Republic, Czech Republic, for testing the experimental data

set with respect to potential modulation.

References

Albrecht, G. & Corey, R. B. (1939). J. Am. Chem. Soc. 61, 1087–1103.Allen, F. H. (2002). Acta Cryst. B58, 380–388.Bakke, Ø. & Mostad, A. (1980). Acta Chem. Scand. B, 34, 559–570.Benedetti, E., Pedone, C. & Sirigu, A. (1973). Gazz. Chim. Ital. 103,

555–561.Bruckner, S. (1982). Acta Cryst. B38, 2405–2408.Bruker (2007a). APEX2. Bruker AXS, Inc., Madison, Wisconsin,

USA.Bruker (2007b). SAINT Bruker AXS, Inc., Madison, Wisconsin,

USA.Cochran, W. & Penfold, B. R. (1952). Acta Cryst. 5, 644–653.Courvoisier, E., Williams, P. A., Lim, G. K., Hughes, C. E. & Harris,

K. D. M. (2012). Chem. Commun. 48, 2761–2763.Cyr, L. V., Newton, M. G. & Phillips, R. S. (1999). Bioorg. Med. Chem.

7, 1497–1503.Dahaoui, S., Pichon-Pesme, V., Howard, J. A. K. & Lecomte, C.

(1999). J. Phys. Chem. A, 103, 6240–6250.Dai, W. & Fu, D.-W. (2008). Acta Cryst. E64, o1446.Day, G. M. (2011). Crystallogr. Rev. 17, 3–52.Derissen, J. L., Endeman, H. J. & Peerdeman, A. F. (1968). Acta Cryst.

B24, 1349–1354.Di, K. (2010). Acta Cryst. E66, o1125–o1126.Duncan, L. L., Patrick, B. O. & Brock, C. P. (2002). Acta Cryst. B58,

502–511.Eijck, B. P. van & Kroon, J. (2000). Acta Cryst. B56, 535–542.Frey, M. N., Lehmann, M. S., Koetzle, T. F. & Hamilton, W. C. (1973).

Acta Cryst. B29, 876–884.Fujii, I. (2009). Anal. Sci. 25, 35–36.Gorbitz, C. H., Vestli, K. & Orlando, R. (2009). Acta Cryst. B65, 393–

400.

Hao, X., Siegler, M. A., Parkin, S. & Brock, C. P. (2005). Cryst.Growth Des. 5, 2225–2232.

Harding, M. M. & Howieson, R. M. (1976). Acta Cryst. B32, 633–634.Harding, M. M. & Long, H. A. (1968). Acta Cryst. B24, 1096–1102.Hirokawa, S. (1955). Acta Cryst. 8, 637–641.Hsu, L.-Y., Kampf, J. W. & Nordman, C. E. (2002). Acta Cryst. B58,

260–264.Hubschle, C. B., Dittrich, B. & Luger, P. (2002). Acta Cryst. C58,

o540–o542.Hubschle, C. B., Messerschmidt, M. & Luger, P. (2004). Cryst. Res.

Technol. 39, 274–278.Iitaka, Y. (1960). Acta Cryst. 13, 35–45.Iitaka, Y. (1961). Acta Cryst. 14, 1–10.In, Y., Kishima, S., Minoura, K., Nose, T., Shimohigashi, Y. & Ishida,

T. (2003). Chem. Pharm. Bull. 51, 1258–1263.Janczak, J. & Luger, P. (1997). Acta Cryst. C53, 1954–1956.Karle, I. L. & Karle, J. (1964). Acta Cryst. 17, 835–841.Kartha, G. & de Vries, A. (1961). Nature, 192, 862–863.Kayushina, R. L. & Vainshtein, B. K. (1965). Kristallografiya, 10, 833–

844.Kerr, K. A. & Ashmore, J. P. (1973). Acta Cryst. B29, 2124–2127.Khawas, B. & Krishna Murti, G. S. R. (1969). Acta Cryst. B25, 1006–

1009.Lehmann, M. S. & Nunes, A. C. (1980). Acta Cryst. B36, 1621–

1625.Li, J., Liang, Z.-P. & Tai, X. S. (2009). Personal communication.Madden, J. J., McGandy, E. L. & Seeman, N. C. (1972). Acta Cryst.

B28, 2377–2382.Madden, J. J., McGandy, E. L., Seeman, N. C., Harding, M. M. & Hoy,

A. (1972). Acta Cryst. B28, 2382–2389.Marsh, R. E. (1958). Acta Cryst. 11, 654–663.Mostad, A., Nissen, H. M. & Romming, C. (1972). Acta Chem. Scand.

26, 3819–3833.Petrıcek, V., Dusek, M. & Palatinus, L. (2006). JANA2006. Institute

of Physics, Praha, Czech Republic.Scheins, S., Dittrich, B., Messerschmidt, M., Paulmann, C. & Luger, P.

(2004). Acta Cryst. B60, 184–190.Schonleber, A. (2011). Z. Kristallogr. 226, 499–517.Schonleber, A. & Chapuis, G. (2004). Ferroelectrics, 305, 99–102.Sheldrick, G. M. (1996). SADABS. University of Gottingen,

Germany.Sheldrick, G. M. (2008). Acta Cryst. A64, 112–122.Shoemaker, D. P., Donohue, J., Schomaker, V. & Corey, R. B. (1950).

J. Am. Chem. Soc. 72, 2328–2349.Simpson, H. J. & Marsh, R. E. (1966). Acta Cryst. 20, 550–555.Torii, K. & Iitaka, Y. (1970). Acta Cryst. B26, 1317–1326.Torii, K. & Iitaka, Y. (1971). Acta Cryst. B27, 2237–2246.Torii, K. & Iitaka, Y. (1973). Acta Cryst. B29, 2799–2807.Umadevi, K., Anitha, K., Sridhar, B., Srinivasan, N. & Rajaram, R. K.

(2003). Acta Cryst. E59, o1073–o1075.Vela, M. A., McLaughlin, M. L. & Fronczek, F. R. (1989). Acta Cryst.

C45, 1091–1093.Weissbuch, I., Frolow, F., Addadi, L., Lahav, M. & Leiserowitz, L.

(1990). J. Am. Chem. Soc. 112, 7718–7724.Yamada, K., Hashizume, D., Shimizu, T. & Yokoyama, S. (2007). Acta

Cryst. E63, o3802–o3803.

research papers

Acta Cryst. (2012). B68, 549–557 Carl Henrik Gorbitz et al. � Single-crystal investigation 557