Ingrid Kjøller Larsen

Structural Basis for AMPA Receptor Activation and Ligand Selectivity: Crystal Structures of Five Agonist Complexes with the GluR2 Ligand-binding Core

Glutamate is the principal excitatory neurotransmitter within the mammalian CNS, playing an important role in many different functions in the brain such as learning and memory. In this study, a combination of molecular biology, X-ray structure determinations, as well as electrophysiology and binding experiments, has been used to increase our knowledge concerning the ionotropic glutamate receptor GluR2 at the molecular level. Five high-resolution X-ray structures of the ligand-binding domain of GluR2 (S1S2J) complexed with the three agonists (S)-2-amino-3-[3-hydroxy-5-(2-methyl-2H-tetrazol-5-yl)isoxazol-4-yl]propionic acid (2-Me-Tet-AMPA), (S)-2-amino-3-(3-carboxy-5-methylisoxazol-4-yl)propionic acid (ACPA), and (S)-2-amino-3-(4-bromo-3-hydroxy-isoxazol-5-yl)propionic acid (Br-HIBO), as well as of a mutant thereof (S1S2J-Y702F) in complex with ACPA and Br-HIBO, have been determined. The structures reveal that AMPA agonists with an isoxazole moiety adopt different binding modes in the receptor, dependent on the substituents of the isoxazole. Br-HIBO displays selectivity among different AMPA receptor subunits, and the design and structure determination of the S1S2J-Y702F mutant in complex with Br-HIBO and ACPA have allowed us to explain the molecular mechanism behind this selectivity and to identify key residues for ligand recognition. The agonists induce the same degree of domain closure as AMPA, except for Br-HIBO, which shows a slightly lower degree of domain closure. An excellent correlation between domain closure and efficacy has been obtained from electrophysiology experiments undertaken on non-desensitising GluR2i(Q)-L483Y receptors expressed in oocytes, providing strong evidence that receptor activation occurs as a result of domain closure. The structural results, combined with the functional studies on the full-length receptor, form a powerful platform for the design of new selective agonists.

Structural basis of cell-cell adhesion by NCAM.

The neural cell adhesion molecule NCAM, a member of the immunoglobulin superfamily, mediates cell–cell recognition and adhesion via a homophilic interaction. NCAM plays a key role during development and regeneration of the nervous system and is involved in synaptic plasticity associated with memory and learning. The 1.85 Å crystal structure of the two N-terminal extracellular domains of NCAM reported here provides a structural basis for the homophilic interaction. The molecular packing of the two-domain structure reveals a cross shaped antiparallel dimer, and provides fundamental insight into trans-cellular recognition mediated by NCAM.

Crystal structure of tetranectin, a trimeric plasminogen-binding protein with an α-helical coiled coil

Tetranectin is a plasminogen kringle 4‐binding protein. The crystal structure has been determined at 2.8 Å resolution using molecular replacement. Human tetranectin is a homotrimer forming a triple α‐helical coiled coil. Each monomer consists of a carbohydrate recognition domain (CRD) connected to a long α‐helix. Tetranectin has been classified in a distinct group of the C‐type lectin superfamily but has structural similarity to the proteins in the group of collectins. Tetranectin has three intramolecular disulfide bridges. Two of these are conserved in the C‐type lectin superfamily, whereas the third is present only in long‐form CRDs. Tetranectin represents the first structure of a long‐form CRD with intact calcium‐binding sites. In tetranectin, the third disulfide bridge tethers the CRD to the long helix in the coiled coil. The trimerization of tetranectin as well as the fixation of the CRDs relative to the helices in the coiled coil indicate a demand for high specificity in the recognition and binding of ligands.

STRUCTURE OF THE C-TYPE LECTIN CARBOHYDRATE RECOGNITION DOMAIN OF HUMAN TETRANECTIN

Tetranectin (TN) is a C-type lectin involved in fibrinolysis, being the only endogenous ligand known to bind specifically to the kringle 4 domain of plasminogen. TN was originally isolated from plasma, but shows a wide tissue distribution. Furthermore, TN has been found in the extracellular matrix of certain human carcinomas, whereas none or little is present in the corresponding normal tissue. The crystal structure of full-length trimeric TN (2.8 A resolution) has recently been published [Nielsen et al. (1997). FEBS Lett. 412, 388-396]. The crystal structure of the carbohydrate recognition domain (CRD) of human TN (TN3) has been determined separately at 2.0 A resolution in order to obtain detailed information on the two calcium binding sites. This information is essential for the elucidation of the specificity of TN towards oligosaccharides. TN3 crystallizes as a dimer, whereas it appears as a monomer in solution. The overall fold of TN3 is similar to other known CRDs. Each monomer is built of two distinct regions, one region consisting of six beta-strands and two alpha-helices, and the other region is composed of four loops harboring two calcium ions. The calcium ion at site 1 forms an eightfold coordinated complex and has Asp116, Glu120, Gly147, Glu150, Asn151, and one water molecule as ligands. The calcium ion at site 2, which is believed to be involved in recognition and binding of oligosaccharides, is sevenfold coordinated with ligands Gln143, Asp145, Glu150, Asp165, and two water molecules. One sulfate ion has been located at the surface of TN3, forming contacts to Glu120, Lys148, Asn106 of a symmetry-related molecule, and to an ethanol molecule.

Atomic resolution structure of human HBP/CAP37/azurocidin.

Crystals of human heparin binding protein (HBP) diffract to 1.1 A when flash-frozen at 120 K. The atomic resolution structure has been refined anisotropically using SHELXL96. The final model of HBP consists of 221 amino-acid residues of 225 possible, three glycosylation units, one chloride ion, 15 precipitant ethanol molecules and 323 water molecules. The structure is refined to a final crystallographic R factor of 15.9% and Rfree(5%) of 18.9% using all data. A putative protein kinase C activation site has been identified, involving residues 113-120. The structure is compared to the previously determined 2.3 A resolution structure of HBP.

Crystallization and crystallographic investigations of the small subunit of mouse ribonucleotide reductase

The R2 protein component of mouse ribonucleotide reductase has been obtained from overproducing Escherichia coli bacteria. It has been crystallized using NaCl as precipitant. The crystals are orthorhombic, space group C2221 with cell dimensions and diffract to at least 2.5 Å. The asymmetric unit of the crystals contains one monomer. Rotation and translation function searches using a model based on the weakly homologous E. coli R2 gave one significant peak. Rotation about a crystallographic 2‐fold axis parallel to the a‐axis produces an R2 dimer with dimer interactions very similar to those found for E. coli R2.

Two mutants of human heparin binding protein (CAP37): Toward the understanding of the nature of lipid A/LPS and BPTI binding†

Heparin binding protein (HBP) is an inactive serine protease homologue with important implications in host defense during infections and inflammations. Two mutants of human HBP, [R23S,F25E]HBP and [G175Q]HBP, have been produced to investigate structure‐function relationships of residues in the putative lipid A/lipopolysaccharide (LPS) binding site and BPTI (bovine pancreatic trypsin inhibitor) binding site. The X‐ray structures have been determined at 1.9 Å resolution for [G175Q]HBP and at 2.5 Å resolution for the [R23S,F25E]HBP mutant, and the structures have been fully refined to R‐factors of 18.2% and 20.7%, respectively. The G175Q mutation does not alter the overall structure of the protein, but the ability to bind BPTI has been eliminated, and the mutant mediates only a limited stimulation of the LPS‐induced cytokine release from human monocytes. The lipid A/LPS binding property of [G175Q]HBP is comparable with that of native HBP. The R23S,F25E mutations do not affect the binding of lipid A/LPS and BPTI or the LPS‐induced cytokine release from human monocytes. This shows that two diverse ligands, lipid A/LPS and BPTI, do not share binding sites. Previously, there was convincing evidence for the proposed lipid A/LPS binding site of HBP. Unexpectedly, the extensive structural changes introduced by mutation of Arg23 and Phe25 do not affect the binding of lipid A/LPS, indicating that another not yet identified site on HBP is involved in the binding of lipid A/LPS. Proteins 2001;42:442–451.

Crystallization and molecular-replacement solution of a truncated form of human recombinant tetranectin

The two C-terminal domains, TN23 (residues 17-181), of human recombinant tetranectin, a plasminogen kringle 4 binding C-type lectin, have been crystallized in two different space groups. Using PEG 8000 as precipitant and at a pH of 8.5, crystals belonging to the monoclinic space group C2 are obtained, with unit-cell parameters a = 160.4, b = 44.7, c = 107.5 A, beta = 127.6 degrees. Using sodium formate as precipitant and at a pH of 5.0, TN23 crystallizes in a rhombohedral space group, with unit-cell parameters a = b = c = 107.4 A, alpha = beta = gamma = 78.3 degrees. A full data set to 4.5 A has been collected from the monoclinic crystals. Using the structure of full-length tetranectin, a molecular-replacement solution has been obtained. The crystal packing shows that TN23 crystallizes as a trimer, with one trimer in the asymmetric unit.

Expression, crystallization and preliminary X-ray analysis of the two amino-terminal Ig domains of the neural cell adhesion molecule (NCAM)

The two amino-terminal Ig domains of the rat neural cell adhesion molecule, NCAM, have been expressed in the yeast strain Pichia pastoris. The double domain consisting of 191 amino acids was overexpressed, purified and crystallized. The crystals belong to the monoclinic space group P2(1), with unit-cell parameters a = 47.1, b = 122.5, c = 72.9 A and beta = 98.3 degrees. Assuming there are four double domains per asymmetric unit, V(m) is estimated to be 2.41 A(3) Da(-1) and the solvent content to be 42.3%. A native data set has been collected from a single flash-frozen crystal to 1.85 A resolution.

Human plasminogen binding protein tetranectin: crystallization and preliminary X-ray analysis of the C-type lectin CRD and the full-length protein.

The recombinant human plasminogen binding protein tetranectin (TN) and the C-type lectin CRD of this protein (TN3) have been crystallized. TN3 crystallizes in the tetragonal space group P4(2)2(1)2 with cell dimensions a = b = 64.0, c = 75.7 A and with one molecule per asymmetric unit. The crystals diffract X-rays to at least 2.0 A resolution. A complete diffraction data set has been collected to 2.7 A resolution. The crystals of TN, obtained by the vapour-diffusion reverse salting-in method at 280 K, are rhombohedral, space group R3, with the hexagonal axes a = b = 89.1, c = 75.8 A, and diffract to at least 2.5 A. A full data set has been collected to 3.0 A. The asymmetric unit contains one monomer of TN. Molecular replacement solutions for TN3 and TN have been obtained using the structure of the C-type lectin CRD of rat mannose-binding protein as search model. The rhombohedral space group indicates that trimers of TN are formed in accordance with the observation of trimerization in solution.