Refinement of the structure of human basic fibroblast growth factor at 1.6 A resolution and analysis of presumed heparin binding sites by selenate substitution

The three-dimensional structure of human basic fibroblast growth factor has been refined to a crystallographic residual of 16.1% at 1.6 A resolution. The structure has a Kunitz-type fold and is composed of 12 antiparallel beta-strands, 6 of which form a beta-barrel. One bound sulfate ion has been identified in the model, hydrogen bonded to the side chains of Asn 27, Arg 120, and Lys 125. The side chain of Arg 120 has two conformations, both of which permit hydrogen bonds to the sulfate. This sulfate binding site has been suggested as the binding site for heparin (Eriksson, A.E., Cousens, L.S., Weaver, L.H., & Matthews, B.W., 1991, Proc. Natl. Acad. Sci. USA 88, 3441-3445). Two beta-mercaptoethanol (BME) molecules are also included in the model, each forming a disulfide bond to the S gamma atoms of Cys 69 and Cys 92, respectively. The side chain of Cys 92 has two conformations of which only one can bind BME. Therefore the BME molecule is half occupied at this site. The locations of possible sulfate binding sites on the protein were examined by replacing the ammonium sulfate in the crystallization medium with ammonium selenate. Diffraction data were measured to 2.2 A resolution and the structure refined to an R-factor of 13.8%. The binding of the more electron-dense selenate ion was identified at two positions. One position was identical to the sulfate binding site identified previously. The second selenate binding site, which is of lower occupancy, is situated 5.6 A from the first. This ion is hydrogen bonded by the side chain of Lys 135 and Arg 120. Thus the side chain of Arg 120 binds two selenate ions simultaneously. It is suggested that the observed second selenate binding site should also be considered as a possible binding site for heparin, or that both selenate binding sites might simultaneously contribute to the binding of heparin.

Refined structure of bovine carbonic anhydrase III at 2.0 A resolution

The three-dimensional structure of bovine carbonic anhydrase III (BCA III) from red skeletal muscle cells has been determined by molecular replacement methods. The structure has been refined at 2.0 A resolution by both constrained and restrained structure-factor least squares refinement. The current crystallographic R-value is 19.2% and 121 solvent molecules have so far been found associated with the protein. The structure is highly similar to the refined structure of human carbonic anhydrase II. Some differences in amino acid sequence and structure between the two isoenzymes are discussed. In BCA III, Lys 64 and Arg 91 (His 64 and Ile 91 in HCA II) are both pointing out from the active site cavity forming salt bridges with Glu 4 and Asp 72 (His 4 and Asp 72 in HCA II), respectively. However, Arg 67 and Phe 198 (Asn 67 and Leu 198 in HCA II) are oriented towards the zinc ion and significantly reduce the volume of the active site cavity. Phe 198 particularly reduces the size of the substrate binding region at the "deep water" position at the bottom of the cavity and we suggest that this is one of the major reasons for the differences in catalytic properties of isoenzyme III as compared to isozyme II.

Similar hydrophobic replacements of Leu99 and Phe153 within the core of T4 lysozyme have different structural and thermodynamic consequences

Two bulky amino acids within the core of phage T4 lysozyme have each been replaced in turn with a series of hydrophobic amino acids. In one set of experiments, Leu99 was replaced with Phe, Met, Ile, Val and Ala. In the second series, Phe153 was replaced with Leu, Met, Ile, Val and Ala. The double mutant in which both Leu99 and Phe153 were replaced with alanine was also constructed. The change in stability of the protein associated with each substitution and the crystal structure of each variant have been determined. In the case of replacements at position 99 the protein behaves in a relatively rigid manner, and changes very little in response to substitutions. In contrast, the protein is more flexible and adjusts much more in response to substitutions of Phe153. In both cases there is a roughly linear dependence between the stability of the mutant protein relative to wild-type (delta delta G) and the difference in the hydrophobic strength of the amino acids involved in the substitution based on solvent transfer measurements (delta delta Gtr). The change in delta delta G is, however, much greater than delta delta Gtr. For the Phe153 replacements the discrepancy is about 1.9-fold, while for the Leu99 series it is about 2.6-fold. Mutants such as Leu99-->Ala, for which the protein remains essentially rigid, tend to create larger cavities and so incur a larger energy of destabilization. Mutants such as Phe153-->Ala, for which the protein structure tends to relax, result in smaller cavities and so are less destabilized. Mutants L99I and L99V are less stable than expected from considerations of transfer free energy and cavity formation due to introduced strain caused by the replacement of Leu99 with a residue of different shape. Mutant F153L is more stable than the reference wild-type, even though the transfer free energy of Leu is less than that of Phe. The increase in stability is apparently due to torsional strain in the side-chain of Phe153 that is present in wild-type lysozyme, but is relieved in the mutant structure.

A cavity-containing mutant of T4 lysozyme is stabilized by buried benzene

The hydrophobic cores of proteins are generally well packed, with few cavities. Mutations in which a bulky buried residue such as leucine or phenylalanine is replaced with a small residue such as alanine can create cavities in the core of a protein (our unpublished results). The sizes and shapes of such cavities can vary substantially depending on factors such as local geometry, whether or not a cavity already exists at the site of substitution, and the degree to which the protein structure relaxes to occupy the space vacated by the substituted residue. We show by crystallographic and thermodynamic analysis that the cavity created by the replacement Leu 99----Ala in T4 lysozyme is large enough to bind benzene and that ligand binding increases the melting temperature of the protein by 6.0 degrees C at pH 3.0. Benzene does not, however, bind to the cavity created by the Phe 153----Ala replacement. The results show that cavities can be engineered in proteins and suggest that such cavities might be tailored to bind specific ligands. The binding of benzene at an internal site 7 A from the molecular surface also illustrates the dynamic nature of proteins, even in crystals.

Response of a protein structure to cavity-creating mutations and its relation to the hydrophobic effect

Six "cavity-creating" mutants, Leu46----Ala (L46A), L99A, L118A, L121A, L133A, and Phe153----Ala (F153A), were constructed within the hydrophobic core of phage T4 lysozyme. The substitutions decreased the stability of the protein at pH 3.0 by different amounts, ranging from 2.7 kilocalories per mole (kcal mol-1) for L46A and L121A to 5.0 kcal mol-1 for L99A. The double mutant L99A/F153A was also constructed and decreased in stability by 8.3 kcal mol-1. The x-ray structures of all of the variants were determined at high resolution. In every case, removal of the wild-type side chain allowed some of the surrounding atoms to move toward the vacated space but a cavity always remained, which ranged in volume from 24 cubic angstroms (A3) for L46A to 150 A3 for L99A. No solvent molecules were observed in any of these cavities. The destabilization of the mutant Leu----Ala proteins relative to wild type can be approximated by a constant term (approximately 2.0 kcal mol-1) plus a term that increases in proportion to the size of the cavity. The constant term is approximately equal to the transfer free energy of leucine relative to alanine as determined from partitioning between aqueous and organic solvents. The energy term that increases with the size of the cavity can be expressed either in terms of the cavity volume (24 to 33 cal mol-1 A-3) or in terms of the cavity surface area (20 cal mol-1 A-2). The results suggest how to reconcile a number of conflicting reports concerning the strength of the hydrophobic effect in proteins.

Dissection of protein structure and folding by directed mutagenesis

The lysozyme from bacteriophage T4 is being used as a model system to determine the roles of individual amino acids in the folding and stability of a typical globular protein. One general finding is that the protein is very adaptable, being able to accommodate many potentially destabilizing replacements. In order to determine the importance of 'alpha-helix propensity' in protein stability, different replacements have been made within alpha-helical segments of T4 lysozyme. Several such substitutions of the form Xaa-->Ala increase the stability of the protein, supporting the idea that alanine is a strongly helix-favouring amino acid. It is possible to engineer a protein that has up to ten alanines in succession, yet still folds and has normal activity. This illustrates the redundancy that is present in the amino acid sequence. A number of 'cavity-creating' mutants of the form Leu-->Ala have been constructed to understand better the nature of hydrophobic stabilization. The structural consequences of these mutations differ from site to site. In some cases the protein structure hardly changes at all; in other cases removal of the wild-type side-chain allows surrounding atoms to move in and occupy the vacated space, although a cavity always remains. The destabilization of the protein associated with these cavity-creating mutations also varies from case to case. The results suggest how to reconcile recent conflicting reports concerning the strength of the hydrophobic effect in proteins.

Three-dimensional structure of human basic fibroblast growth factor

The three-dimensional structure of human basic fibroblast growth factor (bFGF) has been determined by x-ray crystallography and refined to a crystallographic residual of 17.4% at 2.2-A resolution. The structure was initially solved at a nominal resolution of 2.8 A by multiple isomorphous replacement using three heavy-atom derivatives. Although the map clearly showed the overall fold of the molecule, electron density was not observed for the first 19 amino-terminal and the last 3 carboxyl-terminal amino acids, suggesting that they are disordered. The bFGF crystals were grown from 2.0 M ammonium sulfate at pH 8.1 in space group P1 with cell dimensions a = 30.9 A, b = 33.4 A, c = 35.9 A, alpha = 59.5 degrees, beta = 72.0 degrees, and gamma = 75.6 degrees. There is one molecule per unit cell and the crystals diffract to spacings beyond 1.9 A. The overall structure of bFGF can be described as a trigonal pyramid with a fold very similar to that reported for interleukin 1 beta, interleukin 1 alpha, and soybean trypsin inhibitor. An apparent sulfate ion is bound within a basic region on the surface of the molecule and has a ligands the main-chain amide of Arg-120 and the side chains of Asn-27, Arg-120, and Lys-125. This is suggested as the presumed binding site for heparin. Residues 106-115, which are presumed to bind to the bFGF receptor [Baird, A., Schubert, D., Ling, N. & Guillemin, R. (1988) Proc. Natl. Acad. Sci. USA 85, 2324-2328], include an irregular loop that extends somewhat from the surface of the protein and is about 25 A from the presumed heparin binding site. The backbone structure of this putative receptor-binding loop is very similar, although not identical, to the corresponding region of interleukin 1 beta.

Refined structure of human carbonic anhydrase II at 2.0 A resolution

The structure of human erythrocytic carbonic anhydrase II has been refined by constrained and restrained structure-factor least-squares refinement at 2.0 A resolution. The conventional crystallographic R value is 17.3%. Of 167 solvent molecules associated with the protein, four are buried and stabilize secondary structure elements. The zinc ion is ligated to three histidyl residues and one water molecule in a nearly tetrahedral geometry. In addition to the zinc-bound water, seven more water molecules are identified in the active site. Assuming that Glu-106 is deprotonated at pH 8.5, some of the hydrogen bond donor-acceptor relations in the active site can be assigned and are described here in detail. The O gamma 1 atom of Thr-199 donates its proton to the O epsilon 1 atom of Glu-106 and can function as a hydrogen bond acceptor only in additional hydrogen bonds.

Crystallographic studies of inhibitor binding sites in human carbonic anhydrase II: a pentacoordinated binding of the SCN- ion to the zinc at high pH

The binding of four inhibitors--mercuric ion, 3-acetoxymercuri-4-aminobenzenesulfonamide (AMS), acetazolamide (Diamox), and thiocyanate ion--to human carbonic anhydrase II (HCA II) has been studied with X-ray crystallography. The binding of mercury to HCA II at pH 7.0 has been investigated at 3.1 A resolution. Mercuric ions are observed at both nitrogens in the His-64 ring. One of these sites is pointing toward the zinc ion. The only other binding site for mercury is at Cys-206. The binding of the two sulfonamide inhibitors AMS and Diamox, has been reinvestigated at 2.0 and 3.0 A, respectively. Only the nitrogen of the sulfonamide group binds to the zinc ion replacing the hydroxyl ion. The sulfonamide oxygen closest to the zinc ion is 3.1 A away. Thus the tetrahedral geometry of the zinc is retained, refuting earlier models of a pentacoordinated zinc. The structure of the thiocyanate complex has been investigated at pH 8.5 and the structure has been refined at 1.9 A resolution using the least-squares refinement program PROLSQ. The crystallographic R factor is 17.6%. The zinc ion is pentacoordinated with the anion as well as a water molecule bound in addition to the three histidine residues. The nitrogen atom of the SCN- ion is 1.9 A from the zinc ion but shifted 1.3 A with respect to the hydroxyl ion in the native structure and at van der Waals' distance from the O gamma l atom of Thr-199. This is due to the inability of the O gamma l atom of Thr-199 to serve as a hydrogen bond donor, thus repelling the nonprotonated nitrogen. The SCN- molecule reaches into the deep end of the active site cavity where the sulfur atom has displaced the so-called "deep" water molecule of the native enzyme. The zinc-bound water molecule is 2.2 A from the zinc ion and 2.4 A from the SCN- nitrogen. In addition, this water is hydrogen bonded to the O gamma l atom of Thr-199 and to another water molecule. We have observed that solvent and inhibitor molecules have three possible binding sites on the zinc ion and their significance for the catalysis and inhibition of HCA II will be discussed. All available crystallographic data are consistent with a proposed catalytic mechanism in which both the OH moiety and one oxygen of the substrate HCO3- ion are ligated to the zinc ion.