Search

Type IV pili (T4P) are bacterial virulence factors responsible for attachment to surfaces and for twitching motility, a motion that involves a succession of pilus extension and retraction cycles. In the opportunistic pathogen Pseudomonas aeruginosa, the PilM/N/O/P proteins are essential for T4P biogenesis, and genetic and biochemical analyses strongly suggest that they form an inner-membrane complex. Here, we show through co-expression and biochemical analysis that the periplasmic domains of PilN and PilO interact to form a heterodimer. The structure of residues 69-201 of the periplasmic domain of PilO was determined to 2.2 A resolution and reveals the presence of a homodimer in the asymmetric unit. Each monomer consists of two N-terminal coiled coils and a C-terminal ferredoxin-like domain. This structure was used to generate homology models of PilN and the PilN/O heterodimer. Our structural analysis suggests that in vivo PilN/O heterodimerization would require changes in the orientation of the first N-terminal coiled coil, which leads to two alternative models for the role of the transmembrane domains in the PilN/O interaction. Analysis of PilN/O orthologues in the type II secretion system EpsL/M revealed significant similarities in their secondary structures and the tertiary structures of PilO and EpsM, although the way these proteins interact to form inner-membrane complexes appears to be different in T4P and type II secretion. Our analysis suggests that PilN interacts directly, via its N-terminal tail, with the cytoplasmic protein PilM. This work shows a direct interaction between the periplasmic domains of PilN and PilO, with PilO playing a key role in the proper folding of PilN. Our results suggest that PilN/O heterodimers form the foundation of the inner-membrane PilM/N/O/P complex, which is critical for the assembly of a functional T4P complex.

Adenine phosphoribosyltransferase (APRTase) is a widely distributed enzyme, and its deficiency in humans causes the accumulation of 2,8-dihydroxyadenine. It is the sole catalyst for adenine recycling in most eukaryotes. The most commonly expressed APRTase has subunits of approximately 187 amino acids, but the only crystal structure is from Leishmania donovani, which expresses a long form of the enzyme with 237 residues. Saccharomyces cerevisiae APRTase was selected as a representative of the short APRTases, and the structure of the apo-enzyme and sulfate bound forms were solved to 1.5 and 1.75 A, respectively. Yeast APRTase is a dimeric molecule, and each subunit is composed of a central five-stranded beta-sheet surrounded by five alpha-helices, a structural theme found in all known purine phosphoribosyltransferases. The structures reveal several important features of APRTase function: (i) sulfate ions bound at the 5'-phosphate and pyrophosphate binding sites; (ii) a nonproline cis peptide bond (Glu67-Ser68) at the pyrophosphate binding site in both apo-enzyme and sulfate-bound forms; and (iii) a catalytic loop that is open and ordered in the apo-enzyme but open and disordered in the sulfate-bound form. Alignment of conserved amino acids in short-APRTases from 33 species reveals 13 invariant and 15 highly conserved residues present in hinges, catalytic site loops, and the catalytic pocket. Mutagenesis of conserved residues in the catalytic loop, subunit interface, and phosphoribosylpyrophosphate binding site indicates critical roles for the tip of the catalytic loop (Glu106) and a catalytic site residue Arg69, respectively. Mutation of one loop residue (Tyr103Phe) increases k(cat) by 4-fold, implicating altered dynamics for the catalytic site loop.

A structural genomics comparison of purine nucleoside phosphorylases (PNPs) indicated that the enzyme encoded by Mycobacterium tuberculosis (TB-PNP) resembles the mammalian trimeric structure rather than the bacterial hexameric PNPs. The crystal structure of M. tuberculosis PNP in complex with the transition-state analogue immucillin-H (ImmH) and inorganic phosphate was solved at 1.75 A resolution and confirms the trimeric structure. Binding of the inhibitor occurs independently at the three catalytic sites, unlike mammalian PNPs which demonstrate negative cooperativity in ImmH binding. Reduced subunit interface contacts for TB-PNP, compared to the mammalian enzymes, correlate with the loss of the cooperative inhibitor binding. Mammalian and TB-PNPs both exhibit slow-onset inhibition and picomolar dissociation constants for ImmH. The structure supports a catalytic mechanism of reactant destabilization by neighboring group electrostatic interactions, transition-state stabilization, and leaving group activation. Despite an overall amino acid sequence identity of 33% between bovine and TB-PNPs and almost complete conservation in active site residues, one catalytic site difference suggests a strategy for the design of transition-state analogues with specificity for TB-PNP. The structure of TB-PNP was also solved to 2.0 A with 9-deazahypoxanthine (9dHX), iminoribitol (IR), and PO(4) to reconstruct the ImmH complex with its separate components. One subunit of the trimer has 9dHX, IR, and PO(4) bound, while the remaining two subunits contain only 9dHX. In the filled subunit, 9dHX retains the contacts found in the ImmH complex. However, the region of IR that corresponds to the oxocarbenium ion is translocated in the direction of the reaction coordinate, and the nucleophilic phosphate rotates away from the IR group. Loose packing of the pieces of ImmH in the catalytic site establishes that covalent connectivity in ImmH is required to achieve the tightly bound complex.

Fascin is an actin crosslinking protein that organizes actin filaments into tightly packed bundles believed to mediate the formation of cellular protrusions and to provide mechanical support to stress fibers. Using quantitative rheological methods, we studied the evolution of the mechanical behavior of filamentous actin (F-actin) networks assembled in the presence of human fascin. The mechanical properties of F-actin/fascin networks were directly compared with those formed by alpha-actinin, a prototypical actin filament crosslinking/bundling protein. Gelation of F-actin networks in the presence of fascin (fascin to actin molar ratio >1:50) exhibits a non-monotonic behavior characterized by a burst of elasticity followed by a slow decline over time. Moreover, the rate of gelation shows a non-monotonic dependence on fascin concentration. In contrast, alpha-actinin increased the F-actin network elasticity and the rate of gelation monotonically. Time-resolved multiple-angle light scattering and confocal and electron microscopies suggest that this unique behavior is due to competition between fascin-mediated crosslinking and side-branching of actin filaments and bundles, on the one hand, and delayed actin assembly and enhanced network micro-heterogeneity, on the other hand. The behavior of F-actin/fascin solutions under oscillatory shear of different frequencies, which mimics the cell's response to forces applied at different rates, supports a key role for fascin-mediated F-actin side-branching. F-actin side-branching promotes the formation of interconnected networks, which completely inhibits the motion of actin filaments and bundles. Our results therefore show that despite sharing seemingly similar F-actin crosslinking/bundling activity, alpha-actinin and fascin display completely different mechanical behavior. When viewed in the context of recent microrheological measurements in living cells, these results provide the basis for understanding the synergy between multiple crosslinking proteins, and in particular the complementary mechanical roles of fascin and alpha-actinin in vivo., (Copyright 2001 Academic Press.)

CTLA-4 is a dimeric T-cell surface receptor responsible for transducing signals that down-regulate activated T cells upon binding B7 ligands. The disulfide-linked homodimer of the extracellular segment of human CTLA-4 and the receptor-binding domain of human B7-2 were purified and cocrystallized. Diffraction from these crystals is consistent with the monoclinic space group P2(1) (unit-cell parameters a = 47.85, b = 54.56, c = 103.09 A, beta = 91.63); native data have been collected to 3.2 A resolution.

Regulation of T-cell activity is dependent on antigen-independent co-stimulatory signals provided by the disulphide-linked homodimeric T-cell surface receptors, CD28 and CTLA-4 (ref. 1). Engagement of CD28 with B7-1 and B7-2 ligands on antigen-presenting cells (APCs) provides a stimulatory signal for T-cell activation, whereas subsequent engagement of CTLA-4 with these same ligands results in attenuation of the response. Given their central function in immune modulation, CTLA-4- and CD28-associated signalling pathways are primary therapeutic targets for preventing autoimmune disease, graft versus host disease, graft rejection and promoting tumour immunity. However, little is known about the cell-surface organization of these receptor/ligand complexes and the structural basis for signal transduction. Here we report the 3.2-A resolution structure of the complex between the disulphide-linked homodimer of human CTLA-4 and the receptor-binding domain of human B7-2. The unusual dimerization properties of both CTLA-4 and B7-2 place their respective ligand-binding sites distal to the dimer interface in each molecule and promote the formation of an alternating arrangement of bivalent CTLA-4 and B7-2 dimers that extends throughout the crystal. Direct observation of this CTLA-4/B7-2 network provides a model for the periodic organization of these molecules within the immunological synapse and suggests a distinct mechanism for signalling by dimeric cell-surface receptors.

Immucillin-H [ImmH; (1S)-1-(9-deazahypoxanthin-9-yl)-1,4-dideoxy-1,4-imino-D-ribitol] is a 23 pM inhibitor of bovine purine nucleoside phosphorylase (PNP) specifically designed as a transition state mimic [Miles, R. W., Tyler, P. C., Furneaux, R. H., Bagdassarian, C. K., and Schramm, V. L. (1998) Biochemistry 37, 8615-8621]. Cocrystals of PNP and the inhibitor are used to provide structural information for each step through the reaction coordinate of PNP. The X-ray crystal structure of free ImmH was solved at 0.9 A resolution, and a complex of PNP.ImmH.PO(4) was solved at 1.5 A resolution. These structures are compared to previously reported complexes of PNP with substrate and product analogues in the catalytic sites and with the experimentally determined transition state structure. Upon binding, ImmH is distorted to a conformation favoring ribosyl oxocarbenium ion formation. Ribosyl destabilization and transition state stabilization of the ribosyl oxocarbenium ion occur from neighboring group interactions with the phosphate anion and the 5'-hydroxyl of the ribosyl group. Leaving group activation of hypoxanthine involves hydrogen bonds to O6, N1, and N7 of the purine ring. Ordered water molecules provide a proton transfer bridge to O6 and N7 and permit reversible formation of these hydrogen bonds. Contacts between PNP and catalytic site ligands are shorter in the transition state analogue complex of PNP.ImmH.PO(4) than in the Michaelis complexes of PNP.inosine.SO(4) or PNP.hypoxanthine.ribose 1-PO(4). Reaction coordinate motion is dominated by translation of the carbon 1' of ribose between relatively fixed phosphate and purine groups. Purine and pyrimidine phosphoribosyltransferases and nucleoside N-ribosyl hydrolases appear to operate by a similar mechanism.

The effective regulation of T cell responses is dependent on opposing signals transmitted through two related cell-surface receptors, CD28 and cytotoxic T lymphocyte-associated antigen 4 (CTLA-4). Dimerization of CTLA-4 is required for the formation of high-avidity complexes with B7 ligands and for transmission of signals that attenuate T cell activation. We determined the crystal structure of the extracellular portion of CTLA-4 to 2.0 angstrom resolution. CTLA-4 belongs to the immunoglobulin superfamily and displays a strand topology similar to Valpha domains, with an unusual mode of dimerization that places the B7 binding sites distal to the dimerization interface. This organization allows each CTLA-4 dimer to bind two bivalent B7 molecules and suggests that a periodic arrangement of these components within the immunological synapse may contribute to the regulation of T cell responsiveness.

The actin monomer-binding protein, profilin, influences the dynamics of actin filaments in vitro by suppressing nucleation, enhancing nucleotide exchange on actin, and promoting barbed-end assembly. Profilin may also link signaling pathways to actin cytoskeleton organization by binding to the phosphoinositide PIP(2) and to polyproline stretches on several proteins. Although activities of profilin have been studied extensively in vitro, the significance of each of these activities in vivo needs to be tested. To study profilin function, we extensively mutagenized the Saccharomyces cerevisiae profilin gene (PFY1) and examined the consequences of specific point mutations on growth and actin organization. The actin-binding region of profilin was shown to be critical in vivo. act1-157, an actin mutant with an increased intrinsic rate of nucleotide exchange, suppressed defects in actin organization, cell growth, and fluid-phase endocytosis of pfy1-4, a profilin mutant defective in actin binding. In reactions containing actin, profilin, and cofilin, profilin was required for fast rates of actin filament turnover. However, Act1-157p circumvented the requirement for profilin. Based on the results of these studies, we conclude that in living cells profilin promotes rapid actin dynamics by regenerating ATP actin from ADP actin-cofilin generated during filament disassembly.

One of the central functions of actin cytoskeleton is to provide the mechanical support required for the establishment and maintenance of cell morphology. The mechanical properties of actin filament assemblies are a consequence of both the available polymer concentration and the actin regulatory proteins that direct the formation of higher order structures. By monitoring the displacement of well-dispersed microspheres via fluorescence microscopy, we probe the degree of spatial heterogeneity of F-actin gels and networks in vitro. We compare the distribution of the time-dependent mean-square displacement (MSD) of polystyrene microspheres imbedded in low- and high-concentration F-actin solutions, in the presence and absence of the F-actin-bundling protein fascin. The MSD distribution of a 2. 6-microM F-actin solution is symmetric and its standard deviation is similar to that of a homogeneous solution of glycerol of similar zero-shear viscosity. However, increasing actin concentration renders the MSD distribution wide and asymmetric, an effect enhanced by fascin. Quantitative changes in the shape of the MSD distribution correlate qualitatively with the presence of large heterogeneities in F-actin solutions produced by increased filament concentration and the presence of actin bundles, as detected by confocal microscopy. Multiple-particle tracking offers a new, quantitative method to characterize the organization of biopolymers in solution.

Protein tyrosine phosphatase 1B (PTP1B) displays a preference for peptides containing acidic as well as aromatic/aliphatic residues immediately NH(2)-terminal to phosphotyrosine. The structure of PTP1B bound with DADEpYL-NH(2) (EGFR(988)(-)(993)) offers a structural explanation for PTP1B's preference for acidic residues [Jia, Z., Barford, D., Flint, A. J., and Tonks, N. K. (1995) Science 268, 1754-1758]. We report here the crystal structures of PTP1B in complex with Ac-ELEFpYMDYE-NH(2) (PTP1B.Con) and Ac-DAD(Bpa)pYLIPQQG (PTP1B.Bpa) determined to 1.8 and 1.9 A resolution, respectively. A structural analysis of PTP1B.Con and PTP1B.Bpa shows how aromatic/aliphatic residues at the -1 and -3 positions of peptide substrates are accommodated by PTP1B. A comparison of the structures of PTP1B.Con and PTP1B.Bpa with that of PTP1B.EGFR(988)(-)(993) reveals the structural basis for the plasticity of PTP1B substrate recognition. PTP1B is able to bind phosphopeptides by utilizing common interactions involving the aromatic ring and phosphate moiety of phosphotyrosine itself, two conserved hydrogen bonds between the Asp48 carboxylate side chain and the main chain nitrogens of the pTyr and residue 1, and a third between the main chain nitrogen of Arg47 and the main chain carbonyl of residue -2. The ability of PTP1B to accommodate both acidic and hydrophobic residues immediately NH(2)-terminal to pTyr appears to be conferred upon PTP1B by a single residue, Arg47. Depending on the nature of the NH(2)-terminal amino acids, the side chain of Arg47 can adopt one of two different conformations, generating two sets of distinct peptide binding surfaces. When an acidic residue is positioned at position -1, a preference for a second acidic residue is also observed at position -2. However, when a large hydrophobic group occupies position -1, Arg47 adopts a new conformation so that it can participate in hydrophobic interactions with both positions -1 and -3.

Giardia lamblia, the protozoan parasite responsible for giardiasis, requires purine salvage from its host for RNA and DNA synthesis. G. lamblia expresses an unusual purine phosphoribosyltransferase with a high specificity for guanine (GPRTase). The enzyme's sequence significantly diverges from those of related enzymes in other organisms. The transition state analogue immucillinGP is a powerful inhibitor of HGXPRTase from malaria [Li, C. M., et al. (1999) Nat. Struct. Biol. 6, 582-587] and is also a 10 nM inhibitor of G. lamblia GPRTase. Cocrystallization of GPRTase with immucillinGP led unexpectedly to a GPRTase.immucillinG binary complex with an open catalytic site loop. Diffusion of ligands into preformed crystals gave a GPRTase.immucillinGP.Mg(2+).pyrophosphate complex in which the open loop is stabilized by crystal contacts. G. lamblia GPRTase exhibits substantial structural differences from known purine phosphoribosyltransferases at positions remote from the catalytic site, but conserves most contacts to the bound inhibitor. The filled catalytic site with an open catalytic loop provides insight into ligand binding. One active site Mg(2+) ion is chelated to pyrophosphate, but the other is chelated to two conserved catalytic site carboxylates, suggesting a role for these amino acids. This arrangement of Mg(2+) and pyrophosphate has not been reported in purine phosphoribosyltransferases. ImmucillinG in the binary complex is anchored by its 9-deazaguanine group, and the iminoribitol is disordered. No Mg(2+) or pyrophosphate is detected; thus, the 5'-phosphoryl group is needed to immobilize the iminoribitol prior to magnesium pyrophosphate binding. Filling the catalytic site involves (1) binding the purine ring, (2) anchoring the 5'-phosphate to fix the ribosyl group, (3) binding the first Mg(2+) to Asp125 and Glu126 carboxyl groups and binding Mg(2+).pyrophosphate, and (4) closing the catalytic site loop and formation of bound (Mg(2+))(2). pyrophosphate prior to catalysis. Guanine specificity is provided by two peptide carbonyl oxygens hydrogen-bonded to the exocyclic amino group and a weak interaction to O6. Transition state formation involves N7 protonation by Asp129 acting as the general acid.

The crystal structure of the F-actin binding domain 2 of severin, the gelsolin homologue from Dictyostelium discoideum, has been determined by multiple isomorphous replacement and refined to 1.75 A resolution. The structure reveals an alpha-helix-beta-sheet sandwich similar to the domains of gelsolin and villin, and contains two cation-binding sites, as observed in other domain 1 and domain 2 homologues. Comparison of the structures of several gelsolin family domains has identified residues that may mediate F-actin binding in gelsolin domain 2 homologues. To assess the involvement of these residues in F-actin binding, three mutants of human gelsolin domain 2 were assayed for F-actin binding activity and thermodynamic stability. Two of the mutants, RRV168AAA and RLK210AAA, demonstrated a lowered affinity for F-actin, indicating a role for those residues in filament binding. Using both structural and biochemical data, we have constructed a model of the gelsolin domain 1-domain 2-F-actin complex. This model highlights a number of interactions that may serve as positive and negative determinants of filament end- and side-binding.

Transcription of human respiratory syncytial virus (RSV) genome RNA exhibited an obligatory need for the host cytoskeletal protein actin. Optimal transcription, however, required the participation of another cellular protein that was characterized as profilin by a number of criteria. The amino acid sequence of the protein, purified on the basis of its transcription-optimizing activity in vitro, exactly matched that of profilin. RSV transcription was inhibited 60 to 80% by antiprofilin antibody or poly-L-proline, molecules that specifically bind profilin. Native profilin, purified from extracts of lung epithelial cells by affinity binding to a poly-L-proline matrix, stimulated the actin-saturated RSV transcription by 2.5- to 3-fold. Recombinant profilin, expressed in bacteria, stimulated viral transcription as effectively as the native protein and was also inhibited by poly-L-proline. Profilin alone, in the absence of actin, did not activate viral transcription. It is estimated that at optimal levels of transcription, every molecule of viral genomic RNA associates with approximately the following number of protein molecules: 30 molecules of L, 120 molecules of phosphoprotein P, and 60 molecules each of actin and profilin. Together, these results demonstrated for the first time a cardinal role for profilin, an actin-modulatory protein, in the transcription of a paramyxovirus RNA genome.

With access to whole genome sequences for various organisms and imminent completion of the Human Genome Project, the entire process of discovery in molecular and cellular biology is poised to change. Massively parallel measurement strategies promise to revolutionize how we study and ultimately understand the complex biochemical circuitry responsible for controlling normal development, physiologic homeostasis and disease processes. This information explosion is also providing the foundation for an important new initiative in structural biology. We are about to embark on a program of high-throughput X-ray crystallography aimed at developing a comprehensive mechanistic understanding of normal and abnormal human and microbial physiology at the molecular level. We present the rationale for creation of a structural genomics initiative, recount the efforts of ongoing structural genomics pilot studies, and detail the lofty goals, technical challenges and pitfalls facing structural biologists.

Three-dimensional structural models of three members of the phosphoglucomutase (PGM) superfamily, parafusin, phosphoglucomutase-related protein and sarcoplasmic reticulum phosphoglucomutase, were constructed by homology modeling based on the known crystal structure of rabbit muscle phosphoglucomutase. Parafusin, phosphoglucomutase-related protein and sarcoplasmic reticulum phosphoglucomutase each have 50% or more identity with rabbit muscle phosphoglucomutase at the amino acid level and all are reported to exhibit no or minor phosphoglucomutase activity. There are four major insertions and two deletions in the parafusin sequence relative to PGM, all of which are located in surface-exposed loops connecting secondary structural elements. The remaining amino acid substitutions are distributed throughout the sequence and are not predicted to alter the polypeptide fold. Parafusin contains a putative protein kinase C site located on a surface loop in domain II that is not present in the homologs. Although the general domain structure and the active site of rabbit muscle phosphoglucomutase are preserved in the model of phosphoglucomutase-related protein, a major structural difference is likely to occur in domain 1 due to the absence of 55 amino acid residues in PGM-RP. This deletion predicts the loss of three alpha-helices and one beta-strand from an anti-parallel beta-sheet in this domain as compared with the rabbit muscle phosphoglucomutase.

Malaria is a leading cause of worldwide mortality from infectious disease. Plasmodium falciparum proliferation in human erythrocytes requires purine salvage by hypoxanthine-guanine-xanthine phosphoribosyltransferase (HGXPRTase). The enzyme is a target for the development of novel antimalarials. Design and synthesis of transition-state analogue inhibitors permitted cocrystallization with the malarial enzyme and refinement of the complex to 2.0 A resolution. Catalytic site contacts in the malarial enzyme are similar to those of human hypoxanthine-guanine phosphoribosyltransferase (HGPRTase) despite distinct substrate specificity. The crystal structure of malarial HGXPRTase with bound inhibitor, pyrophosphate, and two Mg(2+) ions reveals features unique to the transition-state analogue complex. Substrate-assisted catalysis occurs by ribooxocarbenium stabilization from the O5' lone pair and a pyrophosphate oxygen. A dissociative reaction coordinate path is implicated in which the primary reaction coordinate motion is the ribosyl C1' in motion between relatively immobile purine base and (Mg)(2)-pyrophosphate. Several short hydrogen bonds form in the complex of the enzyme and inhibitor. The proton NMR spectrum of the transition-state analogue complex of malarial HGXPRTase contains two downfield signals at 14.3 and 15.3 ppm. Despite the structural similarity to the human enzyme, the NMR spectra of the complexes reveal differences in hydrogen bonding between the transition-state analogue complexes of the human and malarial HG(X)PRTases. The X-ray crystal structures and NMR spectra reveal chemical and structural features that suggest a strategy for the design of malaria-specific transition-state inhibitors.

The actin regulatory protein profilin is targeted to specific cellular regions through interactions with highly proline-rich motifs embedded within its binding partners. New X-ray crystallographic results demonstrate that profilin, like SH3 domains, can bind proline-rich ligands in two distinct amide backbone orientations. By further analogy with SH3 domains, these data suggest that non-proline residues in profilin ligands may dictate the polarity and register of binding, and the detailed organization of the assemblies involving profilin. This degeneracy may be a general feature of modules that bind proline-rich ligands, including WW and EVH1 domains, and has implications for the assembly and activity of macromolecular complexes involved in signaling and the regulation of the actin cytoskeleton.

The Ena-VASP homology (EVH1) domain is a protein interaction module found in several proteins that are involved in transducing migratory and morphological signals into cytoskeletal reorganization. EVH1 specifically recognizes proline-rich sequences in its binding partners and directs the localization and formation of multicomponent assemblies involved in actin-based motile processes and neural development. The structure of the complex between an EVH1 domain and the target peptide sequence EFPPPPT identifies the interactions responsible for recognition and distinguishes it from other proline-rich binding modules, including SH3 and WW domains. Surprisingly, the EVH1 domain has structural similarity to pleckstrin homology (PH), phosphotyrosine-binding (PTB) and ran-binding (RanBD) domains.

The structure of human HGPRT bound to the transition-state analog immucillinGP and Mg2+-pyrophosphate has been determined to 2.0 A resolution. ImmucillinGP was designed as a stable analog with the stereoelectronic features of the transition state. Bound inhibitor at the catalytic site indicates that the oxocarbenium ion of the transition state is stabilized by neighboring-group participation from MgPPi and O5'. A short hydrogen bond forms between Asp 137 and the purine ring analog. Two Mg2+ ions sandwich the pyrophosphate and contact both hydroxyls of the ribosyl analog. The transition-state analog is shielded from bulk solvent by a catalytic loop that moves approximately 25 A to cover the active site and becomes an ordered antiparallel beta-sheet.

Profilin-I from Acanthamoeba castellanii is a 13-kDa protein that binds actin and poly-l-proline. The native protein has been crystallized in two different but closely related forms. The second form proved more amenable to three-dimensional structural determination using heavy-atom isomorphous methods to obtain crystallographic phase information. We used the second crystal structure as a test molecule in the molecular replacement procedure to determine the structure of the first crystal form of profilin-I. More residues participate in crystal lattice contacts in the first crystal form than in the second. The two crystal forms differ significantly in the C-terminal helix that interacts with actin and in the loop preceding this helix. Coordinates of some main chain atoms here differ by about 1.0 A, and side chain atoms differ by more than 2.0 A., (Copyright 1998 Academic Press.)

Somatostatin (SRIF) is a 14-residue peptide hormone synthesized in the hypothalamus and pancreatic islets. SRIF-14 and an N-terminally extended form, SRIF-28, are generated by the proteolytic processing of an approx. 102-residue precursor prosomatostatin (proSRIF) at a single set of paired basic residues (Arg-Lys) and at a monobasic (Arg) site respectively. Previous work in our laboratory demonstrated that the propeptide of SRIF mediates intracellular sorting; we suggested that this information resides in the prohormone structure. To identify putative sorting domains we have investigated structural features of recombinant anglerfish proSRIF-II purified from Escherichia coli. Two species of proSRIF-II were obtained: a monomeric form and a disulphide-linked dimer. CD analyses revealed that monomeric proSRIF-II lacks appreciable periodic secondary structure; however, on slow heating (2 degrees C/min) and cooling, it assumed a predominantly alpha-helical conformation. When subjected to a second heating-and-cooling cycle, the alpha-helical conformation was maintained. In contrast, the dimeric form of proSRIF-II was predominantly alpha-helical and its helicity did not increase in response to heating and recooling. Our results suggest that proSRIF-II might exist in several different folding intermediate states.

The structure of profilin from the budding yeast Saccharomyces cerevisiae has been determined by X-ray crystallography at 2.3 A resolution. The overall fold of yeast profilin is similar to the fold observed for other profilin structures. The interactions of yeast and human platelet profilins with rabbit skeletal muscle actin were characterized by titration microcalorimetry, fluorescence titrations, and nucleotide exchange kinetics. The affinity of yeast profilin for rabbit actin (2.9 microM) is approximately 30-fold weaker than the affinity of human platelet profilin for rabbit actin (0.1 microM), and the relative contributions of entropic and enthalpic terms to the overall free energy of binding are different for the two profilins. The titration of pyrene-labeled rabbit skeletal actin with human profilin yielded a Kd of 2.8 microM, similar to the Kd of 2.0 microM for the interaction between yeast profilin and pyrene-labeled yeast actin. The binding data are discussed in the context of the known crystal structures of profilin and actin, and the residues present at the actin-profilin interface. The affinity of yeast profilin for poly-L-proline was determined from fluorescence measurements and is similar to the reported affinity of Acanthamoeba profilin for poly-L-proline. Yeast profilin was shown to catalyze adenine nucleotide exchange from yeast actin almost 2 orders of magnitude less efficiently than human profilin and rabbit skeletal muscle actin. The in vivo and in vitro properties of yeast profilin mutants with altered poly-L-proline and actin binding sites are discussed in the context of the crystal structure.

One of the most promising approaches for the development of a synthetic blood substitute has been the engineering of novel mutants of human hemoglobin (Hb) A which maintain cooperativity, but possess lowered oxygen affinity. We describe here two crystal structures of one such potential blood substitute, recombinant (r) Hb(alpha 96Val-->Trp), refined to 1.9 A resolution in an alpha-aquomet, beta-deoxy T-state, and to 2.5 A resolution in a carbonmonoxy R-state. On the basis of molecular dynamics simulations, a particular conformation had been predicted for the engineered Trp residue, and the lowered oxygen affinity had been attributed to a stabilization of the deoxy T-state interface by alpha 96Trp-beta 99Asp hydrogen bonds. Difference Fourier maps of the T-state structure clearly show that alpha 96Trp is in a conformation different from that predicted by the simulation, with its indole side chain directed away from the interface and into the central cavity. In this conformation, the indole nitrogen makes novel water-mediated hydrogen bonds across the T-state interface with beta 101Glu. We propose that these water-mediated hydrogen bonds are the structural basis for the lowered oxygen affinity of rHb(alpha 96Val-->Trp), and discuss the implications of these findings for future molecular dynamics studies and the design of Hb mutants.

Purine nucleoside phosphorylase (PNP) is a key enzyme in the purine salvage pathway, which provides an alternative to the de novo pathway for the biosynthesis of purine nucleotides. PNP catalyzes the reversible phosphorolysis of 2'-deoxypurine ribonucleosides to the free bases and 2-deoxyribose 1-phosphate. Absence of PNP activity in humans is associated with specific T-cell immune suppression. Its key role in these two processes has made PNP an important drug design target. We have investigated the structural details of the PNP-catalyzed reaction by determining the structures of bovine PNP complexes with various substrates and substrate analogues. The preparation of phosphate-free crystals of PNP has allowed us to analyze several novel complexes, including the ternary complex of PNP, purine base, and ribose 1-phosphate and of the completely unbound PNP. These results provide an atomic view for the catalytic mechanism for PNP proposed by M. D. Erion et al. [(1997) Biochemistry 36, 11735-11748], in which an oxocarbenium intermediate is stabilized by phosphate and the negative charge on the purine base is stabilized by active site residues. The bovine PNP structure reveals several new details of substrate and inhibitor binding, including two phosphate-induced conformational changes involving residues 33-36 and 56-69 and a previously undetected role for His64 in phosphate binding. In addition, a well-ordered water molecule is found in the PNP active site when purine base or nucleoside is also present. In contrast to human PNP, only one phosphate binding site was observed. Although binary complexes were observed for nucleoside, purine base, or phosphate, ribose 1-phosphate binding occurs only in the presence of purine base.

Nucleoside N-ribohydrolases are targets for disruption of purine salvage in the protozoan parasites. The structure of a trypanosomal N-ribohydrolase in complex with a transition-state inhibitor is reported at 2.3 A resolution. The nonspecific nucleoside hydrolase from Crithidia fasciculata cocrystallized with p-aminophenyliminoribitol reveals tightly bound Ca2+ as a catalytic site ligand. The complex with the transition-state inhibitor is characterized by (1) large protein conformational changes to create a hydrophobic leaving group site (2) C3'-exo geometry for the inhibitor, typical of a ribooxocarbenium ion (3) stabilization of the ribooxocarbenium analogue between the neighboring group 5'-hydroxyl and bidentate hydrogen bonds to Asn168; and (4) octacoordinate Ca2+ orients a catalytic site water and is liganded to two hydroxyls of the inhibitor. The mechanism is ribooxocarbenium stabilization with weak leaving group activation and is a departure from glucohydrolases which use paired carboxylates to achieve the transition state.

Human glutathione-S-transferase M2-2 (hGSTM2-2) was expressed in Escherichia coli and purified by GSH-affinity chromatography. The recombinant enzyme and the protein isolated from human tissue were indistinguishable based on physicochemical, enzymatic and immunological criteria. The catalytically active dimeric hGSTM2-2 was crystallized without GSH or other active-site ligands in two crystal forms. Diffraction from form A crystals extends to 2.5 A and is consistent with the space group P21 (a = 53.9, b = 81.5, c = 55.6 A, beta = 109.26 A) with two monomers in the asymmetric unit. Diffraction from form B crystals extends to 3 A and is consistent with a space group P212121 (a = 57.2, b = 80.7, c = 225.9 A) with two dimers in the asymmetric unit. This is the first report of ligand-free mu-class GST crystals, and a comparison with liganded complexes will provide insight into the structural consequences of substrate binding which are thought to be important for catalysis.

Filamentous actin structures possess unique biophysical and biochemical properties and are required for cell locomotion, cell division, compartmentalization and morphological processes. The site-specific assembly and disassembly of these structures are directed by actin-regulatory proteins. This article reviews how structural studies are now defining the atomic details of small modular domains present in actin-regulatory proteins responsible for crosslinking, severing and capping of actin filaments, as well as for localization of actin filament assembly. These studies have identified three modular strategies for the design of proteins that regulate the actin cytoskeleton.