doi:10.1016/S0003-9861(02)00088-7 | How to Cite or Link Using DOI
Copyright © 2002 Elsevier Science (USA). All rights reserved.
|Permissions & Reprints|
|Referred to by:||Erratum to “Purification, characterization, and crystallization of alliinase from garlic” [Arch. Biochem. Biophys. 402 (2002) 192–200]|
Archives of Biochemistry and Biophysics, Volume 404, Issue 2, 15 August 2002, Page 339,
E. Bartholomeus Kuettner, Rolf Hilgenfeld, Manfred S. Weiss
Glycosylated dimeric alliinase (EC 18.104.22.168) was purified to homogeneity from its natural source, garlic. With 660 units/mg, the specific enzymatic activity of the pure enzyme is the highest reported to date. Based on both CD spectroscopy data and sequence-derived secondary structure prediction, the α-helix content of alliinase was estimated to be about 30%. Comparisons of all available amino acid sequences of alliinases revealed a common cysteine pattern of the type C–x18–19–C–x–C–x2–C–x5–C–x6–C in the N-terminal part of the sequences. This pattern is conserved in alliinases but absent in other pyridoxal 5′-phosphate-dependent enzymes. It suggests the presence of an epidermal growth factor-like domain in the three-dimensional structures of alliinases, making them unique among the various families of pyridoxal 5′-phosphate-dependent enzymes. Well-ordered three-dimensional crystals of garlic alliinase were obtained in four different forms. The best diffraction was observed with crystal form IV (space group P212121, a=68.4, b=101.1, c=155.7 Å) grown from an ammonium sulfate solution. These crystals diffract to at least 1.5 Å resolution at a synchrotron source and are suitable for structure determination.
Author Keywords: Garlic; Allium sativum; C–S lyase; Alliin lyase; Pyridoxal 5′-phosphate; Protein crystallization; Epidermal growth factor
- • Results and discussion
- • Purity of the protein
- • Enzyme activity
- • UV/Vis, fluorescence, and CD spectroscopy; secondary structure prediction
- • The conserved N-terminal segment—a novel epidermal growth factor (EGF)-like domain
- • Protein crystallization
- • Diffraction data collection and processing
- • Relationship between crystal forms
Alliinase is a homodimeric glycoprotein found in many plants of the genus Allium such as garlic (A. sativum) , onion (A. cepa) , leek (A. porrum) , shallot (A. ascalonicum) , Welsh onion (A. fistulosum) , Chinese chives (A. tuberosum) , rakkyo (A. chinense) , and ramson (A. ursinum) . Furthermore, alliinases were isolated from wild garlic (Tulbaghia violacea) , ornamental plants of genus Leucocoryne , Brassica species , Bacillus subtilis , or Penicillium corymbiferum . Alliinase-like enzymes were also reported to occur in cassie (Acacia farnesia) , broccoli (Brassica oleralea var. botrytis) , or shiitake (Lentinus edodes) .
In the parenchymatous bundle sheaths of garlic, alliinase comprises up to 10–12% of the soluble clove protein material . It is found in vacuoles and is thereby physically separated from its natural substrate alliin ((+)-S-allyl- -cysteine sulfoxide), which occurs in the cytosol . Only upon injuring the garlic cloves does it come into contact with alliin, suggesting that the alliinase/alliin system may be a primitive defense system of the plant [17 and 19].
In garlic, each alliinase subunit consists of 448 amino acids accounting for a molecular weight of 51,500, four putative N-glycosylation sites , and one PLP1 molecule as cofactor [21 and 22]. With its C–S lyase activity , alliinase is able to cleave the Cβ–Sγ bond of sulfoxide derivatives of the amino acid cysteine [23 and 24] ( Fig. 1), thus giving rise to all the garlic sulfur compounds which are responsible for most of the properties of garlic, such as the specific smell and flavor as well as the health benefits like blood lipid or blood pressure lowering  and anticancer, antimicrobial ( ), and antiviral  effects.
Based on the comparison of amino acids around the cofactor binding site, alliinases appear to be similar to other C–S lyases, among which are enzymes involved in the metabolism of cysteine, homocysteine, and methionine or their derivatives . The C–S lyases belong to the fold type I group of the PLP-dependent enzymes , a group which contains aspartate aminotransferase as its most prominent member. However, the overall similarity to this class of enzymes is low (sequence identities are between 20% for 273 aligned amino acid residues and 25% for 153 aligned residues), so that structural investigations of alliinase are necessary in order to reveal the determinants of its activity and selectivity.
We have purified the enzyme from its natural source to homogeneity, characterized it, crystallized it in its native, i.e., glycosylated form, and analyzed the obtained crystal forms by X-ray diffraction.
Materials and methods
Protein source. Fresh German garlic was grown in our own garden in Weimar (Thuringia, Germany); fresh French, Spanish, or Egyptian garlic was purchased from local supermarkets.
Protein purification. The enzyme was purified based on protocols previously described [29 and 30]. The whole purification procedure except for the peeling of the cloves was performed at 4 °C. About 120 g of freshly peeled garlic cloves in 180 ml of buffer A (0.05 M Hepes, pH 7.2, 1 mM Pefabloc, 10% (v/v) glycerol, 20 μM PLP, 1 mM CaCl2, 1 mM MgCl2, 1 mM MnCl2, 3 mM NaN3) was homogenized in a blender (Waring Model 38BL40) for 1 min at maximum speed. The garlic mush was then filtered through two layers of cotton and centrifuged at 20,000g for 30 min. PEG-8000 was added to the crude extract to a final concentration of 25% (w/v) and the solution stirred slowly for 20 min. Precipitated proteins were sedimented at 20,000g for another 15 min. The yellowish pellet was then resuspended in 240 ml of buffer A and centrifuged again at 20,000g for 15 min. The supernatant was filtered through a 0.45-μm filter syringe. The clear protein solution was then applied to an affinity column for glycoproteins (75-ml bed volume of ConA–Sepharose; Pharmacia) which was preequilibrated with buffer B (buffer A without Pefabloc and with 0.5 M NaCl). Glycoproteins were eluted with 200 ml buffer C (buffer B plus 0.15 M methyl-α- -mannoside). The fractions with the strongest light absorption at 430 nm were pooled and concentrated to a volume of less than 10 ml. The final purification step was a gel-filtration column (HiLoad 26/60 Superdex 200 prep grade; Pharmacia) equilibrated with buffer D (0.1 M NaH2PO4, pH 7.4, 0.15 M NaCl, 10% (v/v) glycerol, 20 μM PLP, 3 mM NaN3). Fractions containing alliinase were pooled again and concentrated to 10–30 mg/ml, at which the solution appears lemon to golden yellow.
Chemicals and reagents used in this work were obtained from Sigma, Merck, or Fluka BioChemika unless mentioned otherwise.
Analytical methods. SDS–PAGE on gels containing 12% (w/v) polyacrylamide was carried out according to the method of Laemmli  and stained with Coomassie blue (Pharmacia). Proteins used as molecular mass standards were obtained from the low-molecular-weight kit (Pharmacia). Protein concentrations were determined according to Bradford  using a protein assay kit (Bio-Rad) with known concentrations of BSA (Serva) and γ-globulin (Bio-Rad) as standard.
Enzyme activity. The enzymatic activity of alliinase was assayed using racemic alliin (ICN Biomedicals) as substrate following a procedure described previously  which, in turn, was based on earlier work by Schwimmer and Mazelis . In brief, the standard reaction volume of 300 μl consisted of 0.1 M NaH2PO4 buffer at pH 6.5, 25 μM PLP, 0.2 mM NADH, 3 units of lactate dehydrogenase (Fluka BioChemika), 6 mM alliin, and an alliinase sample of 10–50 ng. Km and Vmax values were determined from enzymatic reactions containing 50 ng of pure alliinase and 0.1–12 mM alliin. Enzymatic activity was monitored at room temperature by a decrease in UV light absorption at 340 nm over a period of 60 s using quartz cuvettes of 1 cm in pathlength and a Zeiss UV/Vis spectrometer (Model Spekol). One unit of enzyme activity was defined as releasing 1 μmol pyruvate per minute.
UV/Vis, fluorescence, and CD spectroscopy; secondary structure prediction. UV/Vis spectra were recorderd using a Varian UV/Vis spectrophotometer (Model Cary 4E), fluorescence spectra using a Perkin–Elmer fluorescence spectrophotometer (Model LS 50), and CD spectra using a Jasco CD spectrophotometer (Model J-710). All spectra were recorded at room temperature. Prior to the recording of the spectra, the alliinase samples were dialyzed against PLP-free buffer (buffer D without PLP) for the UV/Vis and fluorescence spectra or CD-compatible Tes buffer (10 mM Tes, pH 7.4, 50 mM Na2SO4) for the CD spectra, and then diluted to a protein content of 1 mg/ml or 0.1–0.45 mg/ml, respectively. Based on the CD spectra, the secondary structure contents were estimated using software supplied by Jasco Corp. Sequence-derived secondary structure prediction was carried out using the computer program JPred , available on the Internet (www.compbio.dundee.ac.uk/Software/JPred/jpred.html). α-Helix contents of the known structures of PLP-dependent enzymes (type I) were extracted from the Protein Data Bank (PDB) . For AATs, the entries 9AAT , 1AJR , 1AMQ , 1BJW , 2CST , and 1YAA  were chosen and for CBLs, the entries 1C7N , 1CL1 , 1D2F , and 1IBJ .
Sequence alignment. Amino acid sequences of alliinases from different plants (A. sativum, A. cepa, A. ascalonicum, A. ceparoots, A. tuberosum, Arabidopsis thaliana) were taken from the SWISS-PROT database  (entries Q01594, P31757, P31756, Q9M7L9, O04927, and Q9FE98, respectively) and aligned using the University of Wisconsin computer program package UW-GCG v9.0  with the alignment option PILEUP in default mode.
Crystallization. Purified alliinase solution was dialyzed against crystallization buffer (5 mM Hepes, pH 7.4, 10% (v/v) glycerol, 20 μM PLP, 3 mM NaN3). The alliinase concentration was adjusted to 5 mg/ml and a 10-fold excess of S-ethyl- -cysteine, an alliinase inhibitor , was added. Crystallization experiments were performed using hanging drops in standard crystallization trays (Hampton Research) at both 4 and 20°C. The reservoirs contained 1 ml of precipitant solution, the drops 4 μl of protein/inhibitor solution and 4 μl of the respective reservoir solution.
Data collection and processing. For diffraction data collection, the alliinase crystals were mounted in cryoloops (Hampton Research) and slid through dried paraffin oil  at a temperature just above the melting point of the oil. Very quickly thereafter, the crystals were flash-cooled in a nitrogen stream at 100 K.
Diffraction data collection was carried out at beamline ID14.2 of the ESRF (Grenoble, France) and at beamlines X11 and BW7B of the EMBL Hamburg outstation (DESY, Germany) as well as on our in-house rotating anode (FR591; Nonius, Delft, The Netherlands). Diffraction data were processed and scaled using the HKL program suite . The redundancy-independent merging R factor Rrim and the precision-indicating merging R factor Rpim [51 and 52] were calculated using our own program RMERGE (available under http://www.imb-jena.de/www_sbx/projects/sbx_qual.html or from M.S.W. upon request). Further data reduction was achieved using computer programs from the CCP4 suite .
Purity of the protein
The elution profile of the gel-filtration column showed two major peaks which could be separated (Fig. 2). The higher molecular weight peak corresponds to higher order alliinase oligomers and/or to alliinase–lectin complexes. The other peak belongs to pure dimeric alliinase and clearly exhibits the higher alliinase activity (Fig. 2). An SDS–PAGE showed only one band at an apparent molecular weight of 53,000 ( Fig. 3). Only when the gel was overloaded did other faint bands appear (lane C). The overall purity was therefore estimated to be >95%. In an IEF–PAGE, alliinase exhibited multiple bands (data not shown), which may be attributed to different isoforms or glycosylation patterns. No attempt was made to separate the bands before crystallization. NaCl had to be present throughout the purification procedure owing to its ability to preserve the enzymatic activity of alliinase . This property of NaCl seems to be related to its ability to stabilize the alliinase dimer, an effect which could also be observed in analytical gel-filtration experiments of NaCl-containing and NaCl-free alliinase samples (data not shown). These observations suggest a structural role for either Na+ or Cl− in the alliinase dimer, but further clarification of this has to await the determination of the three-dimensional structure of the protein.
|Full-size image (6K)|
Fig. 2. Elution diagram from a gel-filtration column of alliinase from garlic. The line shows the absorbance of the eluate at 280 nm, the bars show the enzymatic activity of the fractions. The peak on the right corresponds to pure dimeric, the one on the left to higher order alliinase oligomers and/or alliinase–lectin complexes. The activity was measured using 5 μl of 40-fold diluted fractions.
|Full-size image (29K)|
Fig. 3. SDS–PAGE of alliinase from garlic. Lane A, molecular weight standard proteins (beginning from top: 94,000, 67,000, 43,000, 30,000, 20,000, 14,400); lane B, alliinase after gel-filtration—50 μg protein; lane C, alliinase after gel-filtration—50 μg protein; lane D, crude garlic extract—50 μg protein.
The specific activity values of purified alliinase were observed to be between 228 and 660 units/mg, depending on the source of the garlic. The highest specific activity was observed with fresh Egyptian garlic. With 660 units/mg this was 69% higher than the previously reported highest alliinase-specific activity of 390 units/mg . However, the latter value was determined for nonglycosylated alliinase from Spanish garlic. The highest value reported for natural, glycosylated alliinase purified using a procedure similar to the one reported here was 332 units/mg . The source in this case was Chinese garlic. The Michaelis–Menten constant Km and the maximum reaction speed Vmax of a freshly prepared French garlic sample were determined to be 2.9 mM and 27.8 milliunits/min respectively. The Km value is close to the 2.2 mM, which was reported by Jansen et al.  for racemic alliin. Nevertheless, their Vmax of 20 milliunits/min is much smaller than our observed value. These observations suggest that the enzymatic activity of alliinase depends not only on the purity and the purification procedure, but also on the brand and the source of garlic.
UV/Vis, fluorescence, and CD spectroscopy; secondary structure prediction
As had been reported previously, alliinase exhibits an absorption maximum at 430 nm (Fig. 4a) due to the absorption of the cofactor PLP . UV/Vis spectra of alliinase stored for about 2 weeks show a small shoulder at around 325 nm (data not shown), a feature which had been observed in another study as well . It has been ascribed to a mixture of protonated internal PLP–Lys aldimin (maximum) and nonprotonated PLP–Lys aldimin (shoulder) .
|Full-size image (15K)|
Fig. 4. Spectroscopic analysis of alliinase from garlic: (a) UV/Vis spectrum. (b) Fluorescence spectrum at three excitation wavelengths (solid line, 430 nm; dotted line, 390 nm; and dashed line, 330 nm). (c) CD spectrum.
A fluorescence scan at 430 nm (or longer wavelengths) excitation (Fig. 4b) shows a fluorescence signal with a maximum at 519 nm. Shorter excitation wavelengths yield fluorescence spectra which appear different (Fig. 4b). For example, excitation at 390 nm results in a weakened and blue-shifted fluorescence maximum at 511 nm and an additional shoulder at 460 nm, and excitation at 330 nm yielded a strong fluorescence at 391 nm. These observations indicate that there is a significant fluorescence signal originating from the PLP cofactor, which may exist as a mixture of different species, with respect to protonation of the internal PLP–Lys aldimin complex, as has been observed by UV/Vis spectroscopy as well.
In the CD spectrum, the enzyme shows a positive circular dichroism in the range around 190 nm (Fig. 4c), which is common for α-helix-containing proteins. Based on the spectrum, the α-helix content was calculated to be about 27–30%, which is in good agreement with an α-helix content of 29% calculated by sequence-derived prediction. The predicted α-helix content is in the lower range of PLP-dependent type I enzymes, e.g., CBLs (26–40%) or AATs (42–46%).
The conserved N-terminal segment—a novel epidermal growth factor (EGF)-like domain
PLP-dependent enzymes of type I from different families share similar domains such as the large central cofactor-binding domain and a small C-terminal domain. Their N-terminal segments do not show a common fold but often contribute to the C-terminal domain . The SWISS-PROT database entries of a few alliinases refer to a PROSITE  domain pattern consistent with an EGF-like domain consensus sequence. A sequence alignment of all available, nonredundant, and complete alliinase sequences from different sources reveals that this EGF-like domain pattern is strictly conserved ( Fig. 5). The derived Cys pattern of C–x18–19–C–x–C–x2–C–x5–C–x6–C differs from the common EGF-like pattern with respect to its first, fourth, and last cysteine residue. This observation points to the existence of a novel EGF-like domain pattern in alliinases, which might also exhibit a different disulfide bridge arrangement. The functional role of this domain in alliinases in unknown. However, it may be related to the vacuolar localization of the enzyme. EGF domains are frequently found in the extracellular portion of membrane-bound or secreted proteins  and alliinase represents an example of an intracellular, secreted protein. Since EGF-like domains from plants are not as common as their animal counterparts, the structure of alliinase would be the first example of a plant enzyme fused to an EGF-like domain.
|Full-size image (22K)|
Fig. 5. Sequence alignment of the N-terminal segments of alliinases from different sources. ALLI_ALSA. A. sativum: ALLI_ALCE, A. cepa; ALLI_ALAS, A. ascalonicum; ALLI_ALCEr, A. ceparoots; ALLI_ALTU, A. tuberosum; ALLI_ARTH, Ar. thaliana. White letters on dark gray background, conserved residues compared to garlic alliinase; white letters on black background, residues of the EGF-like domain PROSITE pattern; black letters on light gray background, additionally conserved cysteine residues which were not part of the PROSITE pattern; black triangles, conserved cysteine residues in EGF-like domain.
Crystals of alliinase were grown under three conditions yielding four different crystal forms (Table 1; Figs. 6A–D). In all cases, it was observed that the best crystals could be grown using freshly purified protein. Once the protein solution was frozen, stored, and rethawed, crystallization was more difficult, and usually microseeding was required. All crystals exhibited a yellowish color indicative of an intact protein–PLP complex.
|Full-size image (89K)|
Fig. 6. The four crystal forms of alliinase from garlic. (A) Stereo image (parallel eyes) of a crystal of form I which was recorded using a digital optical 3D microscope . (B–D) Mono images of crystals of forms II–IV, respectively. The yellow color of the crystals is due to the presence of the cofactor. The bars represent a length of about 100 μm.
The monoclinic crystal form I (space group P2) and the orthorhombic crystal form II (space group C222 or C2221) grew under exactly the same conditions, sometimes even in the same drop. The addition of NaCl led to a new crystal form (form III), which exhibits a crystal habit similar to form I and which also belongs to the same monoclinic space group P2, but with approximately half the unit cell volume (Table 2). This suggests a relationship in the packing of the molecules in these two crystal forms. An interesting observation is also that form III could be grown only using microseeding with a freshly prepared seeding suspension. Another orthorhombic crystal form (form IV, space group P212121) was observed under very different conditions. The pH range was very broad for these form IV crystals and the ammonium sulfate precipitant concentration very high. Nonetheless, these crystals took about 1 month to appear in the drops and grew to their maximum size within another 2–3 months. They exhibited a far better internal order as indicated by their diffraction limit (Table 3). Presumably it was the slow growth that led to such well-ordered crystals.
Note. n.d., not determined.
Diffraction data collection and processing
The unit cell parameters of all four crystal forms are given in Table 2. From visual inspection of the diffraction images, forms I and II diffracted to slightly better than 3.0 Å resolution and forms III and IV to far beyond 2.0 Å (Fig. 7). However, the diffraction data sets of forms I and II extend to only 4.0 and 3.2 Å, respectively, due to experimental difficulties such as ice formation and cryostat failure. Since the data set of crystal form II was rather incomplete, an unambiguous space group assignment was not possible. It remains to be determined whether the correct space group is C222 or C2221. In principle, all four crystal forms would be suitable for structure analysis but due to the superior quality of the form IV crystals our initial focus was laid on that form.
|Full-size image (18K)|
Fig. 7. Diffraction image of one crystal of form IV of alliinase from garlic. The image was recorded using a MAR345 detector on the wiggler beamline BW7B of the EMBL outstation (DESY, Hamburg). The dark area has been enhanced to make the spots reaching out to 1.5 Å resolution visible.
Relationship between crystal forms
From the unit cell parameters given in Table 2, it can be inferred that some relationship exists between the different forms. Form II seems to be a shrunken version of form I, and form III appears to be related to form II by approximately halving the dimension of the c axis. Form IV has only two of its three dimensions similar to the other forms. The a axis seems to be completely unrelated. The reason for this polymorphism is not clear at present. In an IEF–PAGE, dissolved crystals of types I, III, and IV exhibited almost the same multiple band pattern as the protein solution used for crystallization. Therefore, differently charged isoforms of the protein can be ruled out as the cause for the polymorphism. It may be, however, that the protein in solution exists in various glycosylation states. Under the assumption that the different sugar chains occupy the solvent channels in order to not disrupt crystal packing, such a mixture could give rise to different but related crystal forms. This will be an interesting aspect to analyze once the structures in these different crystal forms have been determined.
The glycoprotein alliinase from its natural source was purified to homogeneity. The observed enzymatic activity was the highest reported so far. Even though alliinases belong to the type I group of PLP-dependent enzymes, they are unique in that they are likely to carry a novel EGF-like domain of unknown function. The protein was crystallized in four different but seemingly related crystal forms. At least two of the forms are of sufficient quality to allow the structure to be determined. Structure elucidation by isomorphous replacement is currently in progress.
We thank Dr. Peter Schellenberg, Dr. Hans-Martin Striebel, and Dr. Steffen Kirschstein of the Department of Single Cell and Single Molecule Techniques (IMB Jena) for their assistance with UV/Vis and CD spectroscopy; Dr. Eckard Birckner at the Institute of Physical Chemistry (Friedrich-Schiller-Universität Jena) for the collection of the fluorescence data and helpful discussions; Dr. Santosh Panjikar and Dr. Gottfried J. Palm for their help during the collection of the 1.53-Å diffraction data set of crystal form IV; and the beamline staff of the EMBL outstation in Hamburg and the ESRF in Grenoble for their data collection facilities and assistance.
5. M. Fujita, M. Endo and M. Sano Agric. Biol. Chem. 54 (1990), pp. 1077–1079.
7. S. Tsuno Bitamin (Kyoto) 14 (1958), pp. 659–664.
9. J. Jacobson, Y. Yamaguchi, L.K. Mann and F.D. Howard Phytochemistry 7 (1968), pp. 1099–1108.
12. F. Murakami Bitamin (Kyoto) 20 (1960), pp. 131–135.
16. K. Iwami and K. Yasumoto Agric. Biol. Chem. 44 (1980), pp. 3003–3004.
20. A. Rabinkov, X.-Z. Zhu, G. Grafi, G. Galini and D. Mirelman Appl. Biochem. Biotechnol. 48 (1994), pp. 149–171.
21. E.V. Goryachenkova Dokl. Akad. Nauk SSSR 87 (1952), pp. 457–460.
35. H.M. Berman, J. Westbrook, Z. Feng, G. Gilliland, T.N. Bhat, H. Weissig, I.N. Shindyalov and P.E. Bourne Nucl. Acids Res. 28 (2000), pp. 235–242. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (9114)
37. S. Rhee, M.M. Silva, C.C. Hyde, P.H. Rogers, C.M. Metzler, D.E. Metzler and A. Arnone J. Biol. Chem. 272 (1997), pp. 17293–17302. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (66)
39. T. Nakai, K. Okada, S. Akutsu, I. Miyahara, S. Kawaguchi, R. Kato, S. Kuramitsu and K. Hirotsu Biochemistry 38 (1999), pp. 2413–2424. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (54)
40. V.N. Malashkevich, B.V. Strokopytov, V.V. Borisov, Z. Dauter, K.S. Wilson and Y.M. Torchinsky J. Mol. Biol. 247 (1995), pp. 111–124. Abstract | PDF (1551 K) | View Record in Scopus | Cited By in Scopus (63)
44. T. Clausen, A. Schlegel, R. Peist, E. Schneider, C. Steegborn, Y.-S. Chang, A. Haase, G.P. Bourenkov, H.D. Bartunik and W. Boos EMBO J. 19 (2000), pp. 831–842. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (35)
45. U. Breitinger, T. Clausen, S. Ehlert, R. Huber, B. Laber, F. Schmidt, E. Pohl and A. Messerschmidt Plant Physiol. 126 (2001), pp. 631–642. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (19)
53. Collaborative Computational Project No. 4, 1994. Acta Crystallogr. D 50, 760–763.
59. V. Schubert GIT Laborfachzeitschrift 44 (2000), pp. 942–943.
61. G.S. French and K.S. Wilson Acta Crystallogr. A 34 (1978), pp. 517–525.
*1 This work was supported by the Deutsche Forschungsgemeinschaft, Grant WE2520/1-1.
Corresponding author. Fax: +49-40-89902-149; email: email@example.com
1 Abbreviations used: ConA, concanavalin A; BSA, bovine serum albumin; Vis, visible light; AAT, aspartate aminotransferase; CBL, cystathionine β-lyase; EGF, epidermal growth factor; IEF, isoelectric focusing; PEG-8000, polyethylene glycol with average molecular weight of 8000; PLP, pyridoxal 5′-phosphate; Tes, N-tris(hydroxymethyl)methyl-2-aminoethanesulfonic acid.