Crystal structure of the putative carbohydrate recognition domain of human galectin‐related protein

Galectins are β-galactoside binding proteins sharing a conserved carbohydrate recognition domain (CRD).1 Galectin-encoding genes are present in the genomes of a wide variety of eukaryotic organisms ranging from marine sponges up to humans.2 They are however restricted to the animal kingdom and to the subclass of agaricomycetidae in the fungal kingdom. Members of the galectin family can be assigned to one of three subgroups: The prototype subgroup having only one CRD per monomer, the tandem-repeat type having two CRDs and the chimera type which possesses a long peptide extension in addition to its CRD.3 The highly conserved residues responsible for β-galactoside association have been defined based on the X-ray structures of galectin family members or respective CRDs in complex with their carbohydrate ligands.4-9 Residues His49, Asn51, Arg53, Asn62, and Glu72 recognize theβ-galactoside ligand via specific hydrogen bonds, while Trp69 confers binding affinity through hydrophobic stacking with the galactose moiety [numbering of human galectin-7 (hGal-7)]. However, within the galectin family, there are a growing number of proteins containing a galectin CRD based on sequence similarity, but which lack some of the strictly conserved amino acids responsible for β-galactoside binding and accordingly do not display significant affinity for β-galactosides. Members of this so-called galectin-related protein (GRP) family are galectin-related interfiber protein (GRIFIN), the mannose-binding Charcot-Leyden crystal protein, Coprinopsis cinerea GRP (CGL3) and human GRP, to name a few. Human GRP contains a short N-terminal segment of unknown function and a putative CRD, which shares consensus amino acids at 51 out of the 64 most highly conserved residues in other galectins2 (see Fig. 1). No ligand has so far been identified. To get insight into the function of this GRP highly conserved among animals, we have determined the three-dimensional structure of the GRP CRD fragment at 2.0 Å resolution. Sequence alignment of H. sapiens GRP with orthologs from different species (in italics) and hGal-7. Strictly conserved and similar residues are boxed in red and represented by red letters, respectively. Amino acid residues directly involved in carbohydrate coordination by hGal-7 are highlighted with asterisks. Ppyg, Pongo pygmaeus (CAH92960); Mmus, Mus musculus (NP_776113); Btau, Bos taurus (XP_594786); Rnor, Rattus norvegicus (EDL97949); Oana, Ornithorhynchus anatinus (XP_001511457); Mdom, Monodelphis domestica (XP_001374836); Cfam, Canis lupus familiaris (XP_538509); Ptro, Pan troglodytes (XP_525768); Ecab, Equus caballus (XP_001494811); Mmul, Macaca mulatta (XP_001085876); Ggal, Gallus gallus (NP_001008456); Xtro, Xenopus tropicalis (NP_001001900); hGal-7 (NP_002298). The N-terminus of the crystallized protein is indicated with a vertical arrow. Secondary structure elements derived from the crystal structure of GRP-C using DSSP10 are shown on top of the sequences. The sequences were aligned using Multalin11 and the figure was prepared with ESPript 2.2.12 The HSPC159 gene, coding for human GRP, was amplified by PCR using cDNA generated from mRNA of HEK293 cells as a template. The PCR product was cloned into pBAD-TOPO (Invitrogen), which served as a template for the generation of a plasmid for bacterial expression of the N-terminally truncated and his-tagged GRP (GRP-C; residues 38–172, accession NP_054900). The site of truncation was chosen according to a GRP expression and crystallization study.13 The respective insert was generated by PCR using the forward oligonucleotide primer NdeI-His8-GRP (5′-CAT ATG GGC CAT CAT CAT CAT CAT CAT CAT CAC AGC GGC GTT CCA TTT TGT GGG CAC ATT-3′) and the reverse primer BamHI-GRP (5′-GGG GGG GGA TCC TCA GCC AAG CTT GGT GAT CTG GAG-3′). The PCR product was TOPO-cloned, digested with NdeI/BamHI and ligated into pET24a (Novagen). The plasmid was transformed into Escherichia coli strain BL21 (DE3) and protein production was performed in LB medium at 310 K. When OD600 of the culture reached ∼1.0, protein expression was induced with a final concentration of 0.5 mM isopropyl-β-D-thiogalactoside. The bacterial culture was incubated at the same temperature for an additional 6 h and cells were harvested by centrifugation. Cells were then resuspended in TBS (10 mM Tris-HCl pH 7.5, 150 mM NaCl) containing 1 mM phenylmethylsulfonyl fluoride on ice and lysed by three cycles of french pressing. All subsequent steps were carried out at 277 K. Cell debris was removed in two consecutive steps of centrifugation at 4300g for 10 min and 27,000g for 30 min, respectively. The resulting raw extract was incubated with talon metal affinity resin (BD Biosciences) on a turning wheel for an hour, washed with TBS supplemented with 5 mM imidazole and recombinant protein was finally eluted at room temperature in TBS containing 200 mM imidazole. After size-exclusion chromatography on a HiLoad 16/60 Superdex 75 column (GE Healthcare) equilibrated in TBS at 279 K, fractions containing the protein were pooled and concentrated using Amicon Ultra-4 centrifugal filter device (Millipore) with a molecular weight cutoff of 10 kDa. Protein concentration was calculated by measuring the absorbance at 280 nm using an extinction coefficient of 2560 M−1 cm−1 and the purified protein was immediately used for crystallization. Preliminary crystallization conditions were screened using Crystal Screen I, II, Grid Screen PEG 6000 and Grid Screen Ammonium Sulfate from Hampton Research with the hanging-drop vapor-diffusion method at 291 K. Drops containing equal volumes (2.5 μL) of protein (40 mg/mL) and reservoir solution were equilibrated against 500 μL reservoir solution. The best crystal grew in Tris/HCl pH 8.0 with 14% PEG 6000. Crystals reached a maximum size of ∼0.1 × 0.1 mm and were stabilized by adding 30% PEG 6000 directly to the drop. Crystals were then flash frozen in liquid nitrogen. Data were collected at the X06SA beam line at the Swiss Light Source in Villigen. The diffraction data were indexed with the HKL2000 package14 and belonged to the space group P3(1)21 or the enantiomeric P3(2)21. Data collection statistics are summarized in Table I. From the solvent content value, two to three monomers could be present in the asymmetric unit. The structure was successfully solved by molecular replacement using the hGal-7 monomer (PDB ID: 1BKZ) as a search model in PHASER.15 Search attempts were made in both P3(2)21 and P3(1)21 space group. However, a solution could only be found for the space group P3(2)21. After the location of the first monomer, the second molecule was located using resolution ranging from 30.0 to 3.65 Å and no other monomer could be added. Model rebuilding was done with the program O16 and annealing,B-group and finally B-individual refinement was carried out in CNS.17 After final rebuilding, water molecules were added automatically and checked individually. Residues B1–B2, B19–B28, B39–B40, B50–B56, and B80–B85 are disordered but were included in the model. Their conformation was built similarly to chain A despite very weak electron density for the side chains and the Cα chain. The final Rcryst and Rfree are 21.7 and 25.9%, respectively. The model displays satisfactory stereochemical qualities as determined by PROCHECK18 and SFCHECK.19 Refinement statistics are indicated in Table I. Atomic coordinates and structure factors have been deposited in the Protein Data Bank under the PDB code 2jj6. The GRP-C monomer shows a typical galectin fold composed of two antiparallel β-sheets (F1–F5 and S1–S6) which associate in a β-sandwich arrangement [Fig. 2(A)]. Like in the structure of hGal-7, a short 310 helix (H1) is located between strands F5 and S2. Overall, the CRD of GRP displays a nearly identical topology and a very similar three-dimensional structure compared to the CRDs of human galectins-1, −2, −3, −4, −7, and Charcot-Leyden crystal protein. GRP-C is showing the highest structural similarity despite only 29% overall amino acid sequence identity with hGal-7 in the region where they perfectly align (GRP amino acids 38–172). A structural comparison performed with DaliLite using the GRP-C and the hGal-7 protein structures gives a root mean square deviations ranging from 1.0 to 1.4 Å (Cα atoms) [Fig. 2(B)]. The GRP-C and hGal-7 structures only deviate slightly in three positions where the GRP protein contains additional residues: 1. Asp75, Ser76 (green) 2. Ile113 (yellow) 3. Thr153 (pink) [Figs. 1 and 2(B)]. Crystal structure of the putative GRP CRD (A) Cartoon representation of one monomer of GRP-C. Strands of the two β-sheets and the short helix are labeled as in Figure 1. (B) The Cα trace of GRP-C is overlayed with the one of hGal-7. GRP-C and hGal-7 are colored in blue and gray, respectively. Insertions in the structure of GRP-C are colored as highlighted in Figure 1. N-acetyllactosamine bound by hGal-7 is shown in orange. (C) Comparison between sugar binding pockets of hGal-7 and GRP-C. Residues directly involved in LacNAc (yellow) binding are shown in gray. Amino acids located at the equivalent positions of GRP-C are depicted in green. Hydrogen bonds are shown as dashed lines. (D) Surface representation of the GRP-C dimer. Protein electrostatics have been mapped onto the Connolly surface (vacuum electrostatics: red negatively, blue positively charged). (E) Cartoon representations of GRP-C (top) and human galectin-2 (bottom, PDB ID: 1HLC) dimers are shown. S1 and F1 strands are highlighted in red and green, respectively. Bound lactose (orange) indicates the location of the sugar binding pocket in the case of hGal-2. From the seven most critical amino acid residues found in the CRDs of galectins (marked with asterisks in Fig. 1), only two are conserved in the putative CRD of GRP. It is therefore rather unlikely that the protein retains some lactose-binding capacity. In agreement with these predictions, binding tests performed with recombinant GRP-C and lactosyl-sepharose did not show any association (data not shown). In Figure 2(C), the coordination of LacNAc in the binding pocket of hGal-7 is depicted.8 The amino acid sidechains of GRP-C located at the respective positions are shown in the same orientation to illustrate the apparent differences. Trp69 (hGal-7) and Arg53 (hGal-7) are known to abolish carbohydrate binding completely when mutated.20, 21 It is possible that the amino acid residues differing between GRP and the galectin family generate a novel sugar-binding specificity and that GRP represents an actual lectin. Such a shift in carbohydrate-binding specificity is observed in the fungal GRP (CGL3) from Coprinopsis cinerea which coordinates chitooligosaccharides and LacDiNac instead of lactose (Wälti et al., in press). Results obtained using size exclusion chromatography indicate that the protein behaves as a monomer in solution (data not shown). Similarly, hGal-7 had initially been reported to be monomeric in solution22 until Morris et al.23 showed its dimeric nature in solution, which was in accordance with the published dimeric quarternary structure in the crystalline state.8 In the case of GRP-C, one of the crystal contacts observed between the protein subunits leads to the formation of one continuous slightly twisted plane of β-sheets encompassing two monomers. This apparent dimer uses a mode of interaction previously observed in the hGal-2 dimer with the significant difference that one monomer is rotated around the longitudinal axis by 180° [Fig. 2(E)]. β-strand S1 of one subunit is in contact with β-strand F1 of the other subunit and vice versa. These interactions result in the two putative sugar-binding pockets pointing into opposite directions unlike in the hGal-2 dimer where the two binding sites are found on the same side of the dimer [Fig. 2(E)]. This GRP-C dimer is clearly different from any previously observed dimer in X-ray crystal structures of prototype galectins, but its existence in solution will have to be proven. Since the removal of N-terminal 37 amino acids is quite an intervention, it remains to be elucidated whether full-length GRP can interact in an identical way. The authors would like to thank M. Imperiali for supplying RNA from HEK cells. They are grateful to N. Ban for advice on crystallization strategy and they deeply acknowledge S. Gutman and C. Schulze-Briese for assistance during data collection at the Swiss Light Source (Villigen, Switzerland). S.T. was recipient of an EMBO long term fellowship.