Solution structure of the antiapoptotic protein bcl-2.

The structures of two isoforms of Bcl-2 that differ by two amino acids have been determined by NMR spectroscopy. Because wild-type Bcl-2 behaved poorly in solution, the structures were determined by using Bcl-2/Bcl-x(L) chimeras in which part of the putative unstructured loop of Bcl-2 was replaced with a shortened loop from Bcl-x(L). These chimeric proteins have a low pI compared with the wild-type protein and are soluble. The structures of the two Bcl-2 isoforms consist of 6 alpha-helices with a hydrophobic groove on the surface similar to that observed for the homologous protein, Bcl-x(L). Comparison of the Bcl-2 structures to that of Bcl-x(L) shows that although the overall fold is the same, there are differences in the structural topology and electrostatic potential of the binding groove. Although the structures of the two isoforms of Bcl-2 are virtually identical, differences were observed in the ability of the proteins to bind to a 25-residue peptide from the proapoptotic Bad protein and a 16-residue peptide from the proapoptotic Bak protein. These results suggest that there are subtle differences in the hydrophobic binding groove in Bcl-2 that may translate into differences in antiapoptotic activity for the two isoforms.

Rationale for Bcl-xL/Bad peptide complex formation from structure, mutagenesis, and biophysical studies.

The three-dimensional structure of the anti-apoptotic protein Bcl-xL complexed to a 25-residue peptide from the death promoting region of Bad was determined using NMR spectroscopy. Although the overall structure is similar to Bcl-xL bound to a 16-residue peptide from the Bak protein (Sattler et al., 1997), the Bad peptide forms additional interactions with Bcl-xL. However, based upon site-directed mutagenesis experiments, these additional contacts do not account for the increased affinity of the Bad 25-mer for Bcl-xL compared to the Bad 16-mer. Rather, the increased helix propensity of the Bad 25-mer is primarily responsible for its greater affinity for Bcl-xL. Based on this observation, a pair of 16-residue peptides were designed and synthesized that were predicted to have a high helix propensity while maintaining the interactions important for complexation with Bcl-xL. Both peptides showed an increase in helix propensity compared to the wild-type and exhibited an enhanced affinity for Bcl-xL.

Retention of immunosuppressant activity in an ascomycin analogue lacking a hydrogen-bonding interaction with FKBP12.

C24-Deoxyascomycin was prepared in a two-step process from ascomycin and evaluated for its immunosuppressant activity relative to ascomycin and FK506. An intermediate in the synthetic pathway, Delta(23,24)-dehydroascomycin, was likewise evaluated. Despite lacking the hydrogen-bonding interactions associated with the C24-hydroxyl moiety of ascomycin, C24-deoxyascomycin was found to be equipotent to the parent compound both in its immunosuppressive potency and in its interaction with the immunophilin, FKBP12. Conversely, Delta(23,24)-dehydroascomycin which also lacks the same hydrogen-bonding interactions did not exhibit this potency. NMR studies were conducted on the FKBP12/C24-deoxyascomycin complex in an attempt to understand this phenomenon at the molecular level. The NMR structures of the complexes formed between FKBP12 and ascomcyin or C24-deoxyascomcyin were very similar, suggesting that hydrogen-bonding interactions with the C24 hydroxyl moiety are not important for complex formation.

Solution structure of an rRNA methyltransferase (ErmAM) that confers macrolide-lincosamide-streptogramin antibiotic resistance.

The Erm family of methyltransferases is responsible for the development of resistance to the macrolide-lincosamide-streptogramin type B (MLS) antibiotics. These enzymes methylate an adenine of 23S ribosomal RNA that prevents the MLS antibiotics from binding to the ribosome and exhibiting their antibacterial activity. Here we describe the three-dimensional structure of an Erm family member, ErmAM, as determined by NMR spectroscopy. The catalytic domain of ErmAM is structurally similar to that found in other methyltransferases and consists of a seven-stranded beta-sheet flanked by alpha-helices and a small two-stranded beta-sheet. In contrast to the catalytic domain, the substrate binding domain is different from other methyltransferases and adopts a novel fold that consists of four alpha-helices.

Solution structure of the DNA-binding domain of a human papillomavirus E2 protein: evidence for flexible DNA-binding regions.

The three-dimensional structure of the DNA-binding domain of the E2 protein from human papillomavirus-31 was determined by using multidimensional heteronuclear nuclear magnetic resonance (NMR) spectroscopy. A total of 1429 NMR-derived distance and dihedral angle restraints were obtained for each of the 83-residue subunits of this symmetric dimer. The average root mean square deviations of 20 structures calculated using a distance geometry-simulated annealing protocol are 0.59 and 0.90 angstroms for the backbone and all heavy atoms, respectively, for residues 2-83. The structure of the human virus protein free in solution consists of an eight-stranded beta-barrel and two pairs of alpha-helices. Although the overall fold of the protein is similar to the crystal structure of the bovine papillomavirus-1 E2 protein when complexed to DNA, several small but interesting differences were observed between these two structures at the subunit interface. In addition, a beta-hairpin that contacts DNA in the crystal structure of the protein-DNA complex is disordered in the NMR structures, and steady-state 1H-15N heteronuclear NOE measurements indicate that this region is highly mobile in the absence of DNA. The recognition helix also appears to be flexible, as evidenced by fast amide exchange rates. This phenomenon has also been observed for a number of other DNA-binding proteins and may constitute a common theme in protein/DNA recognition.

Structure and ligand recognition of the phosphotyrosine binding domain of Shc.

The nuclear magnetic resonance structure of the phosphotyrosine binding (PTB) domain of Shc complexed to a phosphopeptide reveals an alternative means of recognizing tyrosine-phosphorylated proteins. Unlike in SH2 domains, the phosphopeptide forms an antiparallel beta-strand with a beta-sheet of the protein, interacts with a hydrophobic pocket through the (pY-5) residue, and adopts a beta-turn. The PTB domain is structurally similar to pleckstrin homology domains (a beta-sandwich capped by an alpha-helix) and binds to acidic phospholipids, suggesting a possible role in membrane localization.

Backbone dynamics of a two-domain protein: 15N relaxation studies of the amino-terminal fragment of urokinase-type plasminogen activator.

The amino-terminal fragment (ATF) of urokinase-type plasminogen activator (u-PA) is a two-domain protein which consists of a kringle and a growth factor domain (GFD). The dynamics of uniformly 15N-labeled ATF was examined by measuring the longitudinal (T1) and transverse (T2) 15N relaxation times and heteronuclear NOEs. The data were interpreted in terms of the model-independent spectral density function. The GFD was found to exhibit a high degree of anisotropy, whereas the kringle domain of ATF undergoes isotropic reorientation. This difference in anisotropy is best explained by the two domains moving independently such as differently shaped beads on a string. With the exception of the N- and C-terminal regions of the protein, the most flexible region of ATF was the seven-residue omega loop (N22-I28) of the GFD which has been implicated in the binding of u-PA to its receptor. The amides of the linker region between the domains displayed high values of the order parameter, indicating restricted motion on the picosecond time scale. This is in contrast to the flexible linker of calmodulin [Barbato et al. (1992) Biochemistry 31, 5269-5278], which displayed low values of S2 and unrestricted motion in the linker region.

Solution structure of a pleckstrin-homology domain.

Pleckstrin, the major protein kinase C substrate of platelets, contains domains of about 100 amino acids at the amino and carboxy termini that have been found in a number of proteins, including serine/threonine kinases, GTPase-activating proteins, phospholipases and cytoskeletal proteins. These conserved sequences, termed pleckstrin-homology (PH) domains, are thought to be involved in signal transduction. But the details of the function and binding partners of the PH domains have not been characterized. Here we report the solution structure of the N-terminal pleckstrin-homology domain of pleckstrin determined using heteronuclear three-dimensional nuclear magnetic resonance spectroscopy. The structure consists of an up-and-down beta-barrel of seven antiparallel beta-strands and a C-terminal amphiphilic alpha-helix that caps one end of the barrel. The overall topology of the domain is similar to that of the retinol-binding protein family of structures.

Solution structure of the amino-terminal fragment of urokinase-type plasminogen activator.

The amino-terminal fragment (ATF) of urokinase-type plasminogen activator is a two domain protein which consists of a growth factor and a kringle domain. The 1H, 13C, and 15N chemical shifts of this protein have been assigned using heteronuclear two- and three-dimensional NMR experiments on selective and uniformly 15N- and 15N/13C-labeled protein isolated from mammalian cells that overexpress the protein. The chemical shift assignments were used to interpret the NOE data which resulted in a total of 1299 NOE restraints. The NOE restraints were used along with 27 phi angle restraints and 21 hydrogen-bonding restraints to produce 15 low energy structures. The individual domains in the structures are highly converged, but the two domains are structurally independent. The root mean square deviations (rmsd) between residues 11-46 in the growth factor domain and the mean atomic coordinates were 0.99 +/- 0.2 for backbone heavy atoms and 1.65 +/- 0.2 for all non-hydrogen atoms. For residues 55-130 in the kringle domain, the rmsd was 0.84 +/- 0.2 for backbone heavy atoms and 1.42 +/- 0.2 for all non-hydrogen atoms. The overall structures of the individual domains are very similar to the structures of homologous proteins. However, important structural differences between the growth factor and other homologous proteins were observed in the region which has been implicated in binding the urokinase receptor which may explain, in part, why other growth factors show no appreciable affinity for the urokinase receptor.

Three-dimensional structure of the FK506 binding protein/ascomycin complex in solution by heteronuclear three- and four-dimensional NMR.

A high-resolution three-dimensional solution structure of the FKBP/ascomycin complex has been determined using heteronuclear multidimensional nuclear magnetic resonance spectroscopy (NMR) and a distance geometry/simulated annealing protocol. A total of 43 structures of the complex, including 3 tightly bound water molecules, were obtained using 1958 experimental restraints consisting of 1724 nuclear Overhauser effect (NOE) derived distances, 66 chi 1 and 46 phi angular restraints, and 122 hydrogen bond restraints. The root mean square (rms) deviations between the 43 FKBP/ascomycin solution structures and the mean atomic coordinates were 0.43 +/- 0.08 A for the backbone heavy atoms and 0.80 +/- 0.08 A for all non-hydrogen atoms. Angular order parameters for the family of 43 conformations were calculated to determine dihedral convergence. Order parameters for phi, psi, and chi 1 angles exhibited mean values of 0.98, 0.97, and 0.95, respectively, while the mean of the chi 2 order parameter was 0.63. Comparisons were made between the FKBP/ascomycin complex and two NMR-derived solution structures of unbound FKBP and the X-ray crystal structure of an FKBP/FK506 complex. Differences were observed between the FKBP/ascomycin complex and uncomplexed FKBP for residues 33-45 and 78-92. In contrast, the NMR-derived solution structure of the FKBP/ascomycin complex and the X-ray crystal structure of the FKBP/FK506 complex were very similar. Differences between the two complexes were mainly observed in the conformations of some highly solvent exposed side chains.

A practical method for uniform isotopic labeling of recombinant proteins in mammalian cells.

A method to obtain uniformly isotopically labeled (15N and 15N/13C) protein from mammalian cells is described. The method involves preparation of isotopically labeled media consisting of amino acids isolated from bacterial and algal extracts supplemented with cysteine and enzymatically synthesized glutamine. The approach is demonstrated by producing 15N-labeled and 15N/13C-labeled urokinase from Sp2/0 cells and successfully growing Chinese hamster ovary (CHO) cells on the labeled media. Thus, using the procedures described, isotopically labeled proteins that have been expressed in mammalian cells can be prepared, allowing them to be studied by heteronuclear multidimensional NMR techniques.

Conformation of two non-immunosuppressive FK506 analogs when bound to FKBP by isotope-filtered NMR.

The 3D structure of two unlabeled FK506 analogs, (R)- and (S)-[18-OH]ascomycin, when bound to [U-13C,15N]FKBP were determined by isotope-filtered 2D NMR experiments. The structures for the R and S isomers that bind tightly to FKBP but lack immunosuppressive activity are compared to each other and to the conformation of the potent immunosuppressant, ascomycin, when bound to FKBP. The results are interpreted in terms of calcineurin binding to the FKBP/ascomycin complex.

NMR studies of an FK-506 analog, [U-13C]ascomycin, bound to FK-506-binding protein.

Multidimensional, heteronuclear NMR methods were used to determine the complete 1H and 13C resonance assignments for [U-13C]ascomycin bound to recombinant FKBP, including stereospecific assignment of all 22 methylene protons. The conformation of ascomycin was then determined from an analysis of NOEs observed in a 13C-edited 3D HMQC-NOESY spectrum of the [U-13C]ascomycin/FKBP. This structure is found to be quite different from the solution structure of the two forms of uncomplexed FK-506. However, it is very similar to the X-ray crystal structure of FK-506 bound to FKBP, rms deviation = 0.56 A. The methods used for resonance assignment and structure calculation are presented in detail. Furthermore, FKBP/ascomycin NOEs are reported which help define the structure of the ascomycin binding pocket. This structural information obtained in solution was compared to the recently described X-ray crystal structure of the FKBP/FK-506 complex.

High-level expression of recombinant human FK-binding protein from a fusion precursor.

The human peptidyl-prolyl isomerase FK-binding protein (FKBP) was cloned as a fusion partner with CMP-KDO synthetase (CKS), and the resultant construct was characterized as an improved high-expression source for FKBP. The CKS-FKBP fusion was expressed as a soluble protein at levels approaching 1 gm/L in Escherichia coli fermentations. The fusion protein was purified to near homogeneity by a one-step ammonium sulfate fractionation of whole cell lysate. After selective cleavage, the fusion precursor produced yields approaching 300 mg of purified FKBP per liter of harvested culture, a approximately 30 to 60-fold increase over that observed for a nonfusion construct. Selective cleavage of the fusion partners was accomplished using either hydroxylamine or specific, limited proteolysis. Once separated from the CKS fusion partner, the FKBP was isolated in a single step by either reversed-phase HPLC or chromatography on Q-Sepharose. For comparison of physical and chemical properties, a nonfusion construct of recombinant human FKBP was expressed in E. coli and isolated. The purified FKBPs exhibited expected SDS-PAGE molecular weights and N-terminal sequences. The proteins had similar proton NMR spectra and binding to [3H]FK-506. The fusion construct, CKS-FKBP, was also found to bind [3H]FK-506. These data indicate that FKBP fused to the C-terminus of CKS folds independently of the fusion partner and suggests the fused FKBP adopts a conformation resembling that of the native protein.

Identification of solvent-exposed regions of an FK-506 analog, ascomycin, bound to FKBP using a paramagnetic probe.

The solvent-exposed regions of (U-13C)ascomycin when bound to its putative target protein, FKBP, have been identified based on the different proton longitudinal relaxation rates (R1 = 1/T1) measured in the absence and presence of the paramagnetic relaxation reagent, 4-hydroxy-2,2,6,6-tetramethyl-piperidinyl-1-oxy (HyTEMPO). The proton T1S of bound ascomycin were determined using a pulse sequence (T1-HMQC) which consists of a 180 degree proton pulse and a variable delay (tau) followed by a heteronuclear multiple quantum correlation (HMQC) experiment. The solvent-exposed regions of ascomycin determined by these experiments are compared to NOE data in which ascomycin/FKBP contacts were identified and to the X-ray structure of the FK-506/FKBP complex.

NMR identification of protein surfaces using paramagnetic probes.

Paramagnetic agents produce line broadening and thus cancellation of anti phase cross-peak components in two-dimensional correlated nuclear magnetic resonance spectra. The specificity of this effect was examined to determine its utility for identifying surface residues of proteins. Ubiquitin and hen egg white lysozyme, for which X-ray crystal structures and proton NMR assignments are available, served as test cases. Two relaxation reagents were employed, 4-hydroxy-2,2,6,6-tetramethylpiperidinyl-1-oxy and the gadolinium (III) diethylenetriaminepentaacetate complex ion. Correlations were sought between reagent-produced decreases of side-chain cross-peak volumes in double-quantum-filtered proton correlation (DQF-COSY) spectra and the solvent-exposed side-chain surface area of the corresponding residues. The lanthanide complex produced strong effects ascribable to association with carboxylate groups but was not otherwise useful in delineating surface residues. The nitroxyl, on the other hand, produced clear distinctions among the Val, Leu, and Ile residues that generally paralleled side-chain exposure in the crystal, although consistent correlations were not observed with residues of other types. Although an instance of possible specific protein-nitroxyl association was noted, the nitroxyl appears to be a tool for identifying hydrophobic surface residues.

Ligand-binding effects on the kringle 4 domain from human plasminogen: a study by laser photo-CIDNP 1H-NMR spectroscopy.

Photo-chemically induced dynamic nuclear polarization (photo-CIDNP) one-dimensional and two-dimensional (2D) 1H-NMR techniques have been applied to the study of the kringle 4 domain of human plasminogen both ligand-free and complexed to the antifibrinolytic drugs epsilon-aminocaproic acid and p-benzylaminesulfonic acid (BASA). A number of aromatic side-chains (His3, Trp72, Tyr41, Tyr50 and Tyr74) appear to be exposed and accessible to 3-N-carboxymethyl-lumiflavin, the photopolarizing flavin dye, both in the presence and in the absence of ligands. A lesser exposure is observed for the Trp25 and Trp62 indole groups in the presence of BASA. The spin-spin (J-coupling) and dipolar (Overhauser) connectivities in the 2D experiments afford absolute assignment of aromatic resonances for the above residues, as well as of those stemming from the Trp72 ring in the presence of BASA. Moreover, a number of H beta resonances can be identified and sorted according to specific types of amino acid residues.