Fungal galectins, sequence and specificity of two isolectins from Coprinus cinereus.

Galectins are members of a genetically related family of β-galactoside-binding lectins. At least eight distinct mammalian galectins have been identified. More distantly related, but still conserving amino acid residues critical for carbohydrate-binding, are galectins in chicken, eel, frog, nematode, and sponge. Here we report that galectins are also expressed in a species of fungus, the inky cap mushroom, Coprinus cinereus. Two dimeric galectins are expressed during fruiting body formation which are 83% identical to each other in amino acid sequence and conserve all key residues shared by members of the galectin family. Unlike most galectins, these have no N-terminal post-translational modification and no cysteine residues. We expressed one of these as a recombinant protein and studied its carbohydrate-binding specificity using a novel nonradioactive assay. Binding specificity has been well studied for a number of other galectins, and like many of these, the recombinant C. cinereus galectin shows particular affinity for blood group A structures. These results demonstrate not only that the galectin gene family is evolutionarily much older than previously realized but also that fine specificity for complex saccharide structures has been conserved. Such conservation implies that galectins evolved to perform very basic cellular functions, presumably by interaction with glycoconjugates bearing complex lactoside carbohydrates resembling blood group A.

Galectins are animal lectins that are related in amino acid sequence and specifically bind to ␤-galactoside carbohydrates such as lactose (1). Members of this gene family all include a conserved carbohydrate-binding domain but vary in inclusion of other domains and in tissue expression patterns. More distantly related, but conserving critical amino acid residues involved in carbohydrate-binding, are galectins in chicken, eel, frog, nematode, and sponge (2).
Although galectins have been studied for 20 years now, physiological functions for these proteins have not yet been clearly established. Their affinity for oligosaccharides found on glycoconjugates on cell surfaces or in extracellular matrix has suggested that galectins function extracellularly by binding to such ligands. Indeed, certain galectins have particular affinity for specific glycoprotein ligands, such as polylactosamine chains on laminin (1,3). When added to cells or overexpressed after transfection, galectins can have major effects on cell adhesion, proliferation, apoptosis, metastasis, and immune function (1)(2)(3)(4)(5)(6). However, evidence has also been presented for intracellular functions of galectins, for instance in message splicing (7) or as nuclear proteins (8). Therefore, efforts are being directed at exploring the evolutionary origin of galectins in the hope that their functions will be easier to define in simple model organisms.
Here we report that a species of fungus, the inky cap mushroom, Coprinus cinereus, expresses two lectins related in sequence and carbohydrate-binding specificity to other galectins. This discovery means that the galectin gene family is even older than previously realized and must have evolved at least a billion years ago. Like many other galectins, recombinant C. cinereus galectin shows particular affinity for blood group A structures, suggesting that fine specificity for complex saccharide structures has also been conserved. Such conservation implies that galectins evolved to perform very basic cellular functions, presumably by interaction with glycoconjugates bearing complex lactoside carbohydrates resembling blood group A.

EXPERIMENTAL PROCEDURES
Lactose Affinity Chromatography-Fruiting stage C. cinereus proteins were first extracted and partially purified using ammonium sulfate precipitation and ion exchange chromatography, as described previously (9). Lactose-binding proteins were then isolated by affinity chromatography from approximately 0.5 mg of protein in 1 ml of 50 mM Tris-HCl (pH 7.2), 150 mM NaCl (TBS) plus 10 mM mercaptoethanol. This partially purified fruiting body extract was passed through a 1-ml (bed volume) column of lactosyl-Sepharose, prepared as described previously (10). The column was washed with 10 ml of TBS, 10 mM mercaptoethanol, and eluted with another 5 ml by inclusion of 0.1 M lactose, while collecting 1-ml fractions. Protein concentrations were determined using a bicinchoninic acid protein assay (BCA Protein Assay, Pierce). Fractions were then analyzed by SDS-PAGE 1 (17% gel) and silver staining. Protein molecular weights were estimated by comparison with protein standards (SDS-PAGE Molecular Weight Standards Ϫ Low Range, Bio-Rad). It was subsequently found that the reducing agent, mercaptoethanol, was not required for purification of the lactosebinding proteins.
Endonuclease Assay-Endonuclease activity was measured basically * This work was supported by Grant AR41459 from the NIAMSD from the National Institutes of Health (to D. N. W. C.) and Grant A4693 from the Natural Sciences and Engineering Research Council of Canada (to B. C. L.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  as described previously (9) by incubating 23 l of serially diluted (in 100 mM Tris-Cl (pH 7.4), 10 mM MgCl 2 ) test fractions with 2 l (0.5 g) of supercoiled plasmid DNA for 30 min at 37°C and then monitoring any DNA nicking by migration in a 1% agarose gel, stained with ethidium bromide.
cDNA Sequencing-DNA was purified from overnight cultures of additional phage clones isolated during the original antibody screen of a C. cinereus gt11 cDNA library (11), and the insert cDNA was removed by restriction digestion and subcloned into pGEM4Z plasmids (Promega Corp., Madison, WI). Both strands of the cDNA were then fully sequenced by standard techniques using Sequenase (U.S. Biochemical Corp.), T7 and T3 oligonucleotide primers flanking the cloning site, some sequence-determined internal primers, and incorporation of ␣-35 S-dCTP and dideoxynucleotides for chain termination according to protocols suggested by the enzyme manufacturer. Restriction digests, electrophoresis, ligation, bacterial transformation, and other basic molecular genetic manipulations all followed standard methods.
Peptide Sequencing-Partial amino acid sequences were determined by established protocols (12) from the N terminus of proteins separated by SDS-PAGE and electroblotted onto a polyvinylidene difluoride membrane. The membrane with the blotted proteins was stained with Coomassie Blue R-250, and individual bands were excised and analyzed by Edman degradation at the Core Facility for Protein/DNA Chemistry at Queen's University, Kingston, Ontario, Canada.
Production and Purification of Recombinant Galectin-To produce recombinant lectin in Escherichia coli, an NdeI site was engineered at the translation start site in the C. cinereus galectin-II clone in pGEM4Z by PCR between a mutagenic primer (5Ј CCAGTCTAACATATGCTC-TACCACC 3Ј) and a primer for the T7 promoter in pGEM4Z. The PCR product was ligated into the HphI site of pCR1000 (Stratagene, La Jolla, CA) and transformed into E. coli INV␣F (Stratagene). This plasmid was then purified, digested with NdeI and BamHI, and ligated into NdeI and BamHI digested pET-11 (Novagen Inc., Madison, WI). Cultures of E. coli strain BL21(DE3) transformed with this plasmid were induced with 0.1 M isopropyl-␤-thiodigalactoside and lysed by sonication, and the recombinant lectin was purified by affinity chromatography on lactosyl-Sepharose, all as described previously (13).
Gel Filtration Chromatography-To estimate approximate native molecular masses, proteins were analyzed by gel filtration chromatography using a Superdex 75 HR 10/30 molecular sieve column (bed volume approximately 24 ml) (Pharmacia Biotech Inc.) and a Perkin-Elmer Series 4 HPLC system with detection at 214 nm. Approximately 10 g of purified recombinant protein or 0.5 mg of the partially purified fruiting body extract at 100 g/ml in TBS, 4 mM mercaptoethanol, 0.1 M lactose was injected into the HPLC, equilibrated in the same buffer, and chromatographed at a flow rate of 0.5 ml/min. In some cases, the eluent was collected in 0.33-ml fractions for SDS-PAGE analysis. Approximate molecular masses of the test proteins were calculated by comparison with molecular weight standards (Sigma) and recombinant rat galectin-1 and human galectin-3 (13).
Electrospray Ionization Mass Spectrometry (ESI-MS)-Mass spectrometry was used to determine precise subunit molecular masses for the native galectins purified by lactose-affinity chromatography from fruiting body extracts. ESI-MS analysis was performed by the Biological Mass Spectrometry Laboratory, Department of Chemistry, University of Waterloo, Ontario. Prior to analysis the purified lectins were extensively dialyzed against 2 mM ammonium acetate (pH 7.0). 400 pmol of the protein was introduced into the ion source at a cone voltage of 22 V. The molecular mass spectrum was reconstructed from multiply charged ions in the m/z (mass to charge ratio) spectrum.
Novel Carbohydrate-binding Assay-Carbohydrate-binding specificity of recombinant lectin was determined using a novel assay and a panel of standard oligosaccharides (as numbered in Table I, numbers 3 and 18 kindly donated by Hakon Leffler, UCSF, numbers 8, 10, 11, and 20 from Oxford GlycoSystems, numbers 9 and 19 from Accurate Chemical and Scientific Corp., and the rest from Sigma). In brief, each saccharide was tested for inhibition of the binding of biotin-labeled asialofetuin (biotin-ASF) to lectin-Sepharose beads, detected by subsequent binding of streptavidin-peroxidase, and colorimetric quantification of bound peroxidase activity using soluble tetramethylbenzidine substrate.
Purified recombinant C. cinereus galectin-II or rat galectin-1 were first conjugated to Sepharose by standard techniques (14). In brief, each protein was dialyzed into 75 mM NaCl, 75 mM Na 2 HPO 4 /KH 2 PO 4 (pH 7.2) (PBS). The pH was raised by addition of 1 ⁄5 volume of 0.5 M NaHCO 3 (pH 8.3), 1 M NaCl, 5 mM EDTA, and lactose was added to 0.1 M. Then 15 ml of this solution at approximately 1 mg/ml galectin was added to 15 ml (packed volume) of cyanogen bromide-activated Sepharose 4B beads (Pharmacia Biotech Inc.), which had been hydrated and washed with dilute HCl according to the manufacturer's instructions. Coupling was allowed to proceed overnight at 4°C with gentle rocking to keep the beads suspended. The beads were then pelleted by centrifugation; the supernatant was aspirated to waste, and unreacted sites were blocked by resuspension and incubation for 2 h at 4°C in 20 ml of 0.2 M ethanolamine in the above coupling buffer. The beads were then washed three times with 20 ml of TBS and stored at 4°C. Fetuin (Sigma) was desialylated by mild acid hydrolysis (15), dialyzed against PBS, and biotinylated by addition of 0.5 ml at 20 mg/ml to an equal volume of 0.2 M NaCO 3 (pH 8.8), and addition of 300 l of 12 mg/ml N-hydroxysuccinimide-biotin (Calbiochem) in dimethylformamide. After reaction in the dark for 1 h at room temperature, the reaction was quenched by addition of 100 l of 1 M NH 4 Cl. The solution was then dialyzed twice against 1 liter of TBS and stored in frozen aliquots.
A competitive binding assay was developed in which 5 l of 50 g/ml biotin-ASF was added to 35 l of 1:7 suspension of lectin-Sepharose beads in TBS, 1% bovine serum albumin (Sigma, RIA grade), 0.2% Triton X-100, and 10 l of a given test saccharide. After incubating overnight at 4°C with gentle rocking to maintain suspension, the beads with any bound biotin-ASF were pelleted by microcentrifuge, washed three times with 0.5 ml of the above buffer, and then incubated rocking for 1 h at 4°C in 0.5 ml of a 1:500 dilution of streptavidin biotinylated horseradish peroxidase complex (RPN 1051, Amersham Life Sciences Inc.) in the same buffer. Again, the beads carrying any streptavidinperoxidase bound to biotin-ASF were pelleted and washed three times. The beads were then resuspended by vortexing in 100 l of H 2 O, and 25 l of suspension was placed in a microtiter plate well. 100 l of tetramethylbenzidine reagent (TMB-Soluble, Intergen) was then added to each well, and the absorption at 620 nm of each well was recorded every minute for 30 min with mechanical agitation before each reading. From these readings the reaction velocity while in the linear range, reflecting the amount of bound peroxidase, was determined using Kineticalc software (Bio-Tek Instruments, Inc., Winooski, VT). Values obtained in the presence of saccharide (averaged for triplicate experiments) were then calculated as a fraction (F) of that obtained in the absence of any inhibitor and plotted against the saccharide concentration (C) using Sigmaplot software (Jandel Scientific Software, San Rafael, CA). The approximate K i for each saccharide, which should be approximately the concentration giving 50% inhibition of biotin-ASF binding to the lectin beads, was calculated by fitting the data to a simple competitive binding function for a trace ligand (F ϭ 1/(1 ϩ C/K i )).

Identification of a Candidate Galectin in C. cinereus-A can-
didate galectin in C. cinereus was initially identified by using the "Findpatterns" program (GCG Sequence Analysis Software Package, version 8, Genetics Computer Group, Inc., Madison, WI) to screen GenBank (National Center for Biotechnology Information, National Library of Medicine) for sequences which include a motif chosen as conserved in the carbohydratebinding site of all known galectins (L, I, V, F)(L, I, V, F)XN(S, T)(5X-7X)(W, Y)XXEX(R, K, H). A partial cDNA encoding such a motif (GenBank Accession No. L03301) had been cloned from a basidiomycete fungus, C. cinereus, by screening an gt11 library with an antibody raised to a partially purified endonuclease (11). Although the cDNA was presumed to encode an endonuclease, this had not been directly demonstrated.
Identification of Lactose-binding Proteins in the Partially Purified Fruiting Body Extract-Lactose affinity chromatography was used to test for the presence of galectins in a partially purified endonuclease preparation. Fruiting body extract was first treated by ammonium sulfate precipitation and ion exchange chromatography, as described previously (9). Analysis by SDS-PAGE and silver staining of this preparation revealed three bands, one with an apparent molecular mass of 17 kDa and a barely resolved doublet at approximately 15.5 kDa (Fig.  1, lane 2). When passed over a lactosyl-Sepharose affinity matrix, all endonuclease activity flowed through the column, as did the lower band in the 15.5-kDa doublet (Fig. 1, lane 3), whereas the upper band of the doublet and the band at 17 kDa were almost completely retained and specifically eluted with lactose ( Fig. 1, lanes 4 -8). This indicates that the retained proteins possess lactose-binding activity. The endonuclease activity might be due to the lower band of the 15.5-kDa doublet or to some other minor component of the preparation. Regardless, all endonuclease activity is clearly separable from lactose-binding activity. Therefore, given its sequence similarity to galectins, the above cDNA sequence seemed quite likely to encode at least one of the lactose-binding proteins at 15.5 and 17 kDa and not an endonuclease.
Isolation of cDNAs for Two Closely Related C. cinereus Galectins-To confirm that the above C. cinereus cDNA encodes at least one of the lactose-binding bands identified here, we searched for a corresponding full coding length cDNA in order to produce and test the recombinant protein for lectin activity. Sequencing of additional clones isolated during the original antibody screen of a C. cinereus gt11 library (11) yielded an apparently full coding length cDNA (Coprinus galectin-II, Gen-Bank Accession No. U64676) 635 base pairs in length with a sequence closely related to the original cDNA. We also discovered that a portion of the original cDNA (Coprinus galectin-I) had been overlooked due to an internal EcoRI restriction site. The full 509-base pair length of this cDNA (corrected GenBank Accession No. L03301) was subcloned by PCR amplification from the phage vector using primers flanking the gt11 EcoRI sites.
These two cDNAs are very closely related, sharing 84% identity in nucleic acid sequence and 83% identity in deduced amino acid sequence (Fig. 2), and both cDNAs share extensive amino acid sequence similarity with other members of the galectin family (Fig. 3) (16 -37), including all critical amino acid residues found by crystallography to be directly involved in carbohydrate binding (marked with an asterisk in Fig. 3) (38 -40). Of the 26-amino acid residue differences between the two Coprinus lectins (nonidentical amino acid residues are marked at bottom of Fig. 3), 8 are clearly conservative substitutions, and only 3 nonconservative substitutions occur at positions generally conserved in other galectins. While certain galectins, such as mammalian galectin-1 (41,42), include cysteine residues that must remain reduced for the protein to retain carbohydrate-binding activity, neither of the deduced Coprinus galectin sequences includes any cysteine residues, and we found that these lectins remain active in the absence of reducing agents. Molecular masses calculated from the deduced amino acid sequences of the cDNAs are 16,408 and 16,671 for Coprinus galectin-I (Cgl-I) and galectin-II (Cgl-II), respectively.
Based on N-terminal peptide sequence obtained from the SDS-PAGE bands described above, the two cDNAs correspond to the two lactose-binding proteins at 15.5 and 17 kDa. The peptide sequence MLYHLFVNNQVKLQNDFKPE from the band migrating at approximately 17 kDa exactly matches the predicted N-terminal sequence of the Cgl-II cDNA. The peptide sequence MLYRLFVNNQIK?QDDFKAE from the band migrating at approximately 15.5-kDa protein matches the predicted N-terminal sequence of the Cgl-I cDNA with the exception of histidine encoded by the cDNA as the fourth residue, instead of the arginine indicated by amino acid sequencing.
Biochemical Characterization of Recombinant Coprinus Galectin-II-Because N-terminal sequence matching the cDNA was obtained for both proteins by Edman degradation, it appeared that neither protein has a modified N terminus, retaining their initial methionines. This was confirmed by mass spectrometry of the Coprinus lectins purified by lactose-affinity chromatography from fruiting body extracts. ESI-MS revealed two major peaks with apparent masses of 16,408 (Ϯ1.2 Da) and 16,672 (Ϯ1.2 Da) (Fig. 4), corresponding exactly to the masses predicted from the cDNA sequences for Cgl-I and Cgl-II, respectively. Also apparent are minor peaks 96 Da larger than each of the major peaks and even smaller peaks 96 Da larger than those, but these appear to be ionization artifacts, because their presence and size varied depending on buffer conditions. These results confirm the deduced protein sequences and demonstrate that neither protein is post-translationally modified in any way.
To confirm the lectin activity indicated by similarity to galectins, the coding sequence for Cgl-II was subcloned into a prokaryotic expression vector and recombinant protein was produced in E. coli. This protein was purified from the lysed bacteria by affinity chromatography using a lactosyl-Sepharose column and elution with lactose, confirming its lectin activity. SDS-PAGE revealed a single band in the purified recombinant protein matching in size the 17-kDa band in the partially purified extract (Fig. 1, lane 1). The purified recombinant lectin had no detectable endonuclease activity even at 2.0 mg/ml (Fig.  5, lane 6), whereas plasmid degrading activity was easily detectable in the partially purified fruiting body extract at 25 g/ml (Fig. 5, lane 3).
Recombinant Cgl-II elutes as two distinct peaks on HPLC gel filtration chromatography under nondenaturing conditions  4 -8, 1/100 each). Proteins were resolved by SDS-PAGE followed by silver staining. Approximate sizes, determined from the migration of molecular mass standards, are shown on the left.

FIG. 2. Nucleotide and deduced amino acid sequences from cDNAs for Coprinus galectin-I and galectin-II.
Nucleotide and amino acids residues differing between Cgl-I and Cgl-II are in bold for Cgl-I. (Fig. 6A, profile A), and based on comparison to molecular mass standards these peaks appear to represent a dimer (about 2 ⁄3 of the total) with an approximate size of 41.5 kDa and a monomer (about 1 ⁄3 of the total) with an approximate size of 18.5 kDa (Fig. 6B). To ask whether Cgl-I also dimerizes, fungal proteins in the partially purified fruiting body extract were analyzed. These proteins elute with a more complex profile (Fig. 6A,  profile B), but major peaks appear at the same positions as for recombinant Cgl-II. SDS-PAGE of fractions from the apparent dimer peak of the partially purified extract (Fig. 6C, lane 4) revealed bands matching in size both Cgl-I (the upper band of the 15.5-kDa doublet bands) and Cgl-II (the 17-kDa band). These bands plus the lower band of the 15.5-kDa doublet appeared in the apparent monomer fractions (Fig. 6C, lane 8 -11). Whether the lectins can also form heterodimers or only homodimers can not be determined from these data.
Characterization of Coprinus Galectin-II Carbohydrate-binding Specificity Using a Novel Assay-To compare the carbohydrate-binding specificity of Coprinus galectin with other galectins, a novel assay was designed based on saccharide inhibition of biotinylated asialofetuin binding to galectin protein immobilized on Sepharose beads. The advantage of this assay is that it is nonradioactive, whereas characterization of specificity for most other galectins has been based on saccharide inhibition of the binding of radioiodinated lectin to immobilized saccharide, such as lactosyl-Sepharose. To be sure that results from this assay are comparable with results reported for other assays, well characterized rat galectin-1 was tested simultaneously. For each saccharide tested, inhibition curves were plotted using a log scale for the saccharide concentration (as shown for lactose in Fig. 7). All curves for Cgl-II and rat galectin-1 were parallel, apparently reflecting simple noncooperative binding in each case. Based on these curves, the approximate K i or the concentration giving 50% inhibition of biotin-ASF binding to the lectin beads was calculated for each saccharide. Results are shown in Table I for recombinant Cgl-II and rat galectin-1 in comparison with published results for inhibition of rat galectin-1 (43) binding to lactosyl-Sepharose. Clearly, the novel assay developed here, in which soluble labeled ligand binds to immobilized lectin, yields values for rat galectin-1 specificity remarkably close to values obtained from an assay in which soluble labeled lectin binds to immobilized ligand. This novel nonradioactive assay should be more practical, given the greater stability of its components.
The carbohydrate-binding specificity of recombinant Cgl-II is very similar to the specificities reported for other galectins (1). As with all other galectins, only galactoside, and especially lactoside, sugars competed for the lectin binding site. Also as with all other galectins, binding seems to require a free hydroxyl at the glucose 3 carbon of lactose (Gal␤1-4Glc), whereas binding seems only moderately affected by substitution at the galactose 2Ј-or 3Ј-hydroxyl. Thus, substitution of lactose with fucose at position 3 (Table I, (31), chicken (32)(33)(34), Caenorhabditis elegans (35,36), and Geodia cydonium (37), respectively. In some cases, only partial sequences covering the carbohydrate-binding domains are shown, as indicated by three periods preceding the sequence. For those galectins that include two carbohydrate-binding domains, the N-terminal domain is labeled A, the C-terminal domain is labeled B, and dashes indicate where these domains join. Underlined gaps have been introduced to maximize alignment of conserved residues. Residue numbering above the sequences is based on the mammalian galectin-1 sequence. The most highly conserved residues are capitalized and shaded. Asterisks above the sequences designate residues found by x-ray crystallography to be directly involved in carbohydrate binding (38 -40). Residues that are not identical in Cgl-I and Cgl-II are marked (#) under the Cgl-II sequence.
(  (Table I, number 5) is a significantly more potent inhibitor than lactose. Like many other galectins, Cgl-II shows particular affinity for lactose substituted with GalNAc at the 3Ј position on galactose (Table I, number 9). Blood group A tetrasaccharide (GalNAc␣1-3[Fuc␣1-2]Gal␤1-4Glc), which has this structure, was the most potent competitive inhibitor tested (apparent K i ϭ 0.09 mM). DISCUSSION Identification of galectin expression in a fungus, C. cinereus, extends the known antiquity of this gene family. The Coprinus isolectins are the first galectins identified outside the animal kingdom, which diverged from the fungi approximately 1 billion years ago (44,45). Such evolutionary conservation implies that galectins evolved to perform very basic biological functions. Physiological processes are considerably more limited in mushrooms than in most other species in which galectins have been studied, restricting the possible functions played by galectins as they originally evolved. Furthermore, because fungi (and sponges) diverged from the main phylogenetic tree before insects and animals diverged, it seems likely that galectins are also expressed in Drosophila and that, if these could be identified, the experimental advantages of that organism could be brought to bear on the question of galectin function.
Carbohydrate-binding specificity was only studied for one of the two Coprinus galectins, Cgl-II. Because they are so similar in amino acid sequence (83% identity), it seems unlikely that there are major differences in binding specificity between Cgl-I and Cgl-II. However, it is clear that these are not just alleles of each other, because preliminary studies have revealed tandem arrangement of the two corresponding genes. 2 One possibility is that gene duplication has served to facilitate high level expression.
Some aspects of Coprinus galectin binding specificity can be interpreted from the crystal structures determined for mammalian galectin-1 (38,39) and galectin-2 (40) complexed with saccharide. Those crystals reveal a carbohydrate-binding site composed of four adjacent ␤ strands (amino acids 31-83 of mammalian galectin-1), and all of the galectins, including both Coprinus galectins, conserve certain critical amino acid residues which directly interact with lactose, His-44 (except in the 32-kDa nematode galectin), Arg-48, Val-59 (except in the 32-kDa nematode galectin C-terminal domain), Asn-61, Trp-68, and Glu-71 (amino acid residues numbered by position in mammalian galectin-1 as shown in Fig. 3). In the crystallized galectins, Asn-46 also contributes to lactose binding through watermediated interaction with the 3-OH of lactose. However, this residue is not conserved in the Coprinus galectins, electric eel galectin (30), or in the N-terminal carbohydrate-binding domain of the 32-kDa nematode galectins (36). Most galectins, including the Coprinus galectins, also share amino acid residue Arg-73, which interacts through a water molecule with the N-acetyl group of GlcNAc and is at least partially responsible for preferential binding to Gal␤1-4GlcNAc. Arg-73 has also been suggested to restrict the equatorial 3-OH of GlcNAc and 2 R. P. Boulianne and B. C. Lu, unpublished observations.

FIG. 4.
Electrospray ionization mass spectra of native Coprinus lectins. ESI-MS analysis of the Coprinus galectins purified by lactose affinity chromatography from fruiting body extracts revealed two major peaks with apparent masses of 16,408 (Ϯ1.2 Da) and 16,672 (Ϯ1.2 Da), corresponding exactly to the masses predicted from the cDNA sequences for Cgl-I and Cgl-II, respectively. thereby block binding of Gal␤1-3GalNAc (46). Many galectins show enhanced binding to larger oligosaccharides, but the amino acid residues responsible have not yet been clearly established.
In general, the K i values measured for Cgl-II are higher than those reported for other galectins. K i is a measure of the competitive potency of a test saccharide for competitive inhibition of lectin binding to a standard ligand (asialofetuin in this case). However, these values are not necessarily directly comparable across studies, because they can be assay-dependent. K i values depend partly on the lectin's affinity for the given standard ligand (unless the standard ligand concentration is much smaller than its dissociation constant), but values reported for various galectins have derived from assays using different standard ligands (in some cases with very different valencies). Even as measured here in the same assay, the fact that the K i values for Cgl-II are higher than those for rat galectin-1 could mean that Cgl-II has a higher affinity than rat galectin-1 for asialofetuin or that Cgl-II has lower affinity than rat galectin-1 for the tested saccharides. What should be comparable across different assays (assuming similar interaction valency) is the relative potency of test saccharides compared with a standard, such as lactose. Analyzed in this way, the 30-fold lower K i measured for blood group A tetrasaccharide compared with lactose for inhibition of Cgl-II is similar to the 25-fold difference found for rat galectin-3.
Preferential binding to blood group A structures has now been observed for a number of galectins, including mammalian galectins Ϫ3 (43,47), Ϫ4 (24), and Ϫ5 (43) and a sponge galectin (48). The observation that blood group A tetrasaccharide is also the best of the tested competitors for the Cgl-II binding site marks such fine specificity as an evolutionarily old characteristic of galectins and suggests that it might be functionally important. At least some fungi incorporate GalNAc into glycoproteins and polysaccharides (49), but there has not yet been enough structural characterization of fungal glycoconjugates to use this information to identify any candidate ligands. It is also possible that relevant ligands are not fungal, but perhaps animal, insect, or nematode glycoconjugates mediating, for instance, spore adhesion to those organisms (50).
While evidence has been presented for both intra-and extracytoplasmic functions of galectins, the complex carbohydrate structures preferred by galectins are extracytoplasmic, and in many cases galectins have been shown to be secreted and accumulated extracellularly (1,2). However, none of the galectins yet described, including the Coprinus galectins, include a secretion signal sequence. Nevertheless, mammalian galectin-1 and galectin-3 have been shown to be exported from mammalian cells by nonclassical mechanisms (51)(52)(53)(54)(55)(56), and galectin-1 can even be exported when expressed as a recombinant protein in Saccharomyces cerevesiae (57). Thus, it is likely that the Coprinus galectins, too, are secreted proteins.
The Coprinus galectins share a number of additional characteristics with other galectin family members. Like most other galectins, the Coprinus galectins form non-disulfide bonded dimers and are thus functionally divalent with the potential to cross-link glycoconjugates on cell surfaces or in extracellular matrix. Like many other galectins, the Coprinus galectins are expressed at very high levels (11), a characteristic of proteins with more structural roles, as opposed to catalytic or signaling roles. Also like many other galectins, the Coprinus galectins show marked developmental regulation, in this case during fruiting body formation (11). These characteristics reinforce previous suggestions that galectin functions may be particularly important in regulating developmental changes in cellcell or cell-matrix interactions.
Unlike other galectins, the Coprinus galectins have unmodified N termini. It is not clear why the mobilities of Cgl-I and Cgl-II diverge on SDS-PAGE, because the molecular masses, as predicted from deduced sequence and confirmed by mass spectrometry, are very close, 16,408 and 16,671, respectively. Therefore, these proteins have no post-translational modifications. In contrast, all other galectins for which this has been studied have acetylated N termini (after cleavage of the initiator methionine in most cases) (2). Mammalian galectin-3 also undergoes regulated phosphorylation (58).
At one time galectins were referred to as S-type lectins (59), because they were thought to require reduced thiols to retain carbohydrate-binding activity. While this is true for some galectins, such as mammalian galectin-1 (41,42), it is not true for many other galectins. The Coprinus galectins fall into the latter category, and like a galectin in nematodes (36) and one in electric eel (30), Cgl-I and Cgl-II have no cysteine residues and retain carbohydrate-binding activity in the absence of reducing conditions. If, as has been proposed (41), oxidative inactivation serves to limit the extracellular lifespan of some galectins, this is not the case for the Coprinus galectins.
Small, saline-soluble lectins have been found in many other fungus species (50,60), and almost all are developmentally regulated with high expression in fruiting bodies and little or no expression in vegetative mycelia. Most show binding specificity for GalNAc, Gal␤1-3GalNAc, or asialomucin. Some of these could be galectins, such as the ␤-galactosyl-specific lectin isolated from fruiting bodies of another basidiomycete mushroom, Ischnoderma resinosus (61). However, the few fungal lectins that have been sequenced are clearly not members of the galectin family. For instance, the galactosamine-binding lectins from the nematode-trapping fungus Arthrobotrys oligospora and the common edible mushroom Agaricus bisporus are related to each other in sequence and form a novel lectin gene family (62,63), whereas the fucose-binding lectin from the mushroom Aleuria aurantia has an unrelated sequence (64).
It is notable that galactoside-binding lectins are also specifically expressed during fruiting body formation in other organisms, such as the cellular slime mold, Dictyostelium discoideum (65), and the bacterium, Myxococcus xanthus (66). These lectins are clearly unrelated in sequence to the galectins but like the galectins have been shown to accumulate extracellularly despite the lack of classical secretion signal sequences (67,68). The fact that in each of these organisms small, soluble galactoside-binding lectins are synthesized at high levels at a time when the organisms are starving and that such lectins seem to have evolved independently several times might indicate that galactoside-binding lectins play some important role in fruiting. Indeed, suppression of lectin expression in D. discoideum (69) or in M. xanthus (70) resulted in impaired fruiting body formation, apparently due to a defect in the cells' ability to migrate into an aggregate, the initial step in fruiting body formation for these species. Fruiting in basidiomycetes, including C. cinereus, involves a similar aggregation of cells (from the vegetative mycelium) to form a more complex tissue, the mushroom. However, this is believed to involve only changes in cell adhesion and elongation, not cell migration (71). We hope to better define the function of the C. cinereus galectins by using homologous recombination to eliminate expression of both C. cinereus galectin genes.