Crystal structure analysis of the tetragonal crystal form are preliminary molecular model of pig-heart citrate synthase.

The crystal structure of pig heart citrate synthase was analyzed at 0.35-nm resolution. Chain tracing was possible and an initial molecular model constructed. The dimensions of the dimer molecule (located on a crystallographic diad) are 7.5 x 6.0 x 9.0 nm. The chain folding is characterized by the predominance of helices and the absence of sheet structure. The electron density accounts for 355 residues per monomer, so that about 80 residues must be disordered in the crystal. The disordered segment in probably N-terminal. The ordered part consists of two closely associated domains, a large domain with 300 residues and a C-terminal domain of 55 residues consisting of 3(anti)parallel helices. The large domain is built from 12 helical segments, some of which are buried in the interior of the molecule. Inhibitor binding studies with citrate and CoA revealed citrate binding sites but showed no electron density for CoA. It is suggested that CoA binds to the disordered, flexible N-terminal domain. Experiments of limited proteolysis with trypsin showed that under conditions a segment of Mr 9000 is cleaved off selectively. The remaining 35 000-Mr part is dimeric.     

The three-dimensional structure of bovine platelet factor 4 at 3.0-A resolution

Platelet factor 4 (PF4), which is released by platelets during coagulation, binds very tightly to negatively charged oligosaccharides such as heparin. To date, six other proteins are known that are homologous in sequence with PF4 but have quite different functions. The structure of a tetramer of bovine PF4 complexed with one Ni(CN)4(2-) molecule has been determined at 3.0 A resolution and refined to an R factor of 0.28. The current model contains residues 24-85, no solvent, and one overall temperature factor. Residues 1-13, which carried an oligosaccharide chain, were removed with elastase to induce crystallization; residues 14-23 and presumably 86-88 are disordered in the electron density map. Because no heavy atom derivative was isomorphous with the native crystals, the complex of PF4 with one Ni(CN)4(2-) molecule was solved using a single, highly isomorphous Pt(CN)4(2-) derivative and the iterative, single isomorphous replacement method. The secondary structure of the PF4 subunit, from amino- to carboxyl-terminal end, consists of an extended loop, three strands of antiparallel beta-sheet arranged in a Greek key, and one alpha-helix. The tetramer contains two extended, six-stranded beta-sheets, each formed by two subunits, which are arranged back-to-back to form a "beta-bilayer" structure with two buried salt bridges sandwiched in the middle. The carboxyl-terminal alpha-helices, which contain lysine residues that are thought to be intimately involved in binding heparin, are arranged as antiparallel pairs on the surface of each extended beta-sheet.     

The fusion activity of HIV-1 gp41 depends on interhelical interactions

Infection by human immunodeficiency virus type I requires the fusogenic activity of gp41, the transmembrane subunit of the viral envelope protein. Crystallographic studies have revealed that fusion-active gp41 is a "trimer-of-hairpins" in which three central N-terminal helices form a trimeric coiled coil surrounded by three antiparallel C-terminal helices. This structure is stabilized primarily by hydrophobic, interhelical interactions, and several critical contacts are made between residues that form a deep cavity in the N-terminal trimer and the C-helix residues that pack into this cavity. In addition, the trimer-of-hairpins structure has an extensive network of hydrogen bonds within a conserved glutamine-rich layer of poorly understood function. Formation of the trimer-of-hairpins structure is thought to directly force the viral and target membranes together, resulting in membrane fusion and viral entry. We test this hypothesis by constructing four series of gp41 mutants with disrupted interactions between the N- and C-helices. Notably, in the three series containing mutations within the cavity, gp41 activity correlates well with the stability of the N-C interhelical interaction. In contrast, a fourth series of mutants involving the glutamine layer residue Gln-653 show fusion defects even though the stability of the hairpin is close to wild-type. These results provide evidence that gp41 hairpin stability is critical for mediating fusion and suggest a novel role for the glutamine layer in gp41 function.     

The crystal structure of N(10)-formyltetrahydrofolate synthetase from Moorella thermoacetica

The structure was solved at 2.5 A resolution using multiwavelength anomalous dispersion (MAD) scattering by Se-Met residues. The subunit of N(10)-formyltetrahydrofolate synthetase is composed of three domains organized around three mixed beta-sheets. There are two cavities between adjacent domains. One of them was identified as the nucleotide binding site by homology modeling. The large domain contains a seven-stranded beta-sheet surrounded by helices on both sides. The second domain contains a five-stranded beta-sheet with two alpha-helices packed on one side while the other two are a wall of the active site cavity. The third domain contains a four-stranded beta-sheet forming a half-barrel. The concave side is covered by two helices while the convex side is another wall of the large cavity. Arg 97 is likely involved in formyl phosphate binding. The tetrameric molecule is relatively flat with the shape of the letter X, and the active sites are located at the end of the subunits far from the subunit interface.     

Structure of an insect virus at 3.0 A resolution

We report the first atomic resolution structure of an insect virus determined by single crystal X-ray diffraction. Black beetle virus has a bipartite RNA genome encapsulated in a single particle. The capsid contains 180 protomers arranged on a T = 3 surface lattice. The quaternary organization of the protomers is similar to that observed in the T = 3 plant virus structures. The protomers consist of a basic, crystallographically disordered amino terminus (64 residues), a beta-barrel as seen in other animal and plant virus subunits, an outer protrusion composed predominantly of beta-sheet and formed by three large insertions between strands of the barrel, and a carboxy terminal domain composed of two distorted helices lying inside the shell. The outer surfaces of quasi-threefold related protomers form trigonal pyramidyl protrusions. A cleavage site, located 44 residues from the carboxy terminus, lies within the central cavity of the protein shell. The structural motif observed in BBV (a shell composed of 180 eight-stranded antiparallel beta-barrels) is common to all nonsatellite spherical viruses whose structures have so far been solved. This highly conserved shell architecture suggests a common origin for the coat protein of spherical viruses, while the primitive genome structure of BBV suggests that this insect virus represents an early stage in the evolution of spherical viruses from cellular genes.     

Modeling the ion channel structure of cecropin.

Atomic-scale computer models were developed for how cecropin peptides may assemble in membranes to form two types of ion channels. The models are based on experimental data and physiochemical principles. Initially, cecropin peptides, in a helix-bend-helix motif, were arranged as antiparallel dimers to position conserved residues of adjacent monomers in contact. The dimers were postulated to bind to the membrane with the NH2-terminal helices sunken into the head-group layer and the COOH-terminal helices spanning the hydrophobic core. This causes a thinning of the top lipid layer of the membrane. A collection of the membrane bound dimers were then used to form the type I channel structure, with the pore formed by the transmembrane COOH-terminal helices. Type I channels were then assembled into a hexagonal lattice to explain the large number of peptides that bind to the bacterium. A concerted conformational change of a type I channel leads to the larger type II channel, in which the pore is formed by the NH2-terminal helices. By having the dimers move together, the NH2-terminal helices are inserted into the hydrophobic core without having to desolvate the charged residues. It is also shown how this could bring lipid head-groups into the pore lining.     

Three-dimensional structure determined for a subunit of human tRNA splicing endonuclease (Sen15) reveals a novel dimeric fold

Splicing of eukaryal intron-containing tRNAs requires the action of the heterotetrameric splicing endonuclease, which is composed of two catalytic subunits, Sen34 and Sen2, and two structural subunits, Sen15 and Sen54. Here we report the solution structure of the human tRNA splicing endonuclease subunit HsSen15. To facilitate the structure determination, we removed the disordered 35 N-terminal and 14 C-terminal residues of the full-length protein to produce HsSen15(36-157). The structure of HsSen15(36-157), the first for a subunit of a eukaryal splicing endonuclease, revealed that the protein possesses a novel homodimeric fold. Each monomer consists of three alpha-helices and a mixed antiparallel/parallel beta-sheet, arranged in a topology similar to that of the C-terminal domain of Methanocaldococcus jannaschii endonuclease. The dimeric interface is dominated by a beta-barrel structure, formed by face-to-face packing of two, three-stranded beta-sheets. Each of the beta-sheets results from reciprocal parallel pairing of one beta-strand from one subunit with two other beta-strands from the symmetric subunit. The structural model provides insights into the functional assembly of the human tRNA splicing endonuclease.     

Structure of human estrone sulfatase suggests functional roles of membrane association

Estrone sulfatase (ES; 562 amino acids), one of the key enzymes responsible for maintaining high levels of estrogens in breast tumor cells, is associated with the membrane of the endoplasmic reticulum (ER). The structure of ES, purified from the microsomal fraction of human placentas, has been determined at 2.60-A resolution by x-ray crystallography. This structure shows a domain consisting of two antiparallel alpha-helices that protrude from the roughly spherical molecule, thereby giving the molecule a "mushroom-like" shape. These highly hydrophobic helices, each about 40 A long, are capable of traversing the membrane, thus presumably anchoring the functional domain on the membrane surface facing the ER lumen. The location of the transmembrane domain is such that the opening to the active site, buried deep in a cavity of the "gill" of the "mushroom," rests near the membrane surface, thereby suggesting a role of the lipid bilayer in catalysis. This simple architecture could be a prototype utilized by the ER membrane in dictating the form and the function of ER-resident enzymes.     

Role of interchain interactions in the stabilization of the right-handed twist of beta-sheets

Conformational energy computations have been carried out for parallel and antiparallel beta-sheets composed of poly-L-Val and poly-L-Ile peptide chains, each consisting of four and of six residues, respectively, with CH3CO- and-NHCH3 end groups. The beta-sheets considered contained three and five equivalent chains, respectively. All computed minimum-energy beta-sheets were found to have a large right-handed twist of a magnitude that corresponds to the mean twist of beta-sheets observed in globular proteins. The twist has the same sign but is much larger than in beta-sheets of poly-L-Ala, because of intra- and interchain interactions between the bulky beta-branched side-chains. While the right-handed twist is a result of intrachain interactions between side-chains in the case of poly-L-Val, these interactions would favor a left-handed twist in poly-L-Ile, and the right-handed twist in the latter is a result of interchain interactions. Parallel beta-sheets are more stable than antiparallel sheets for both poly-L-Val and poly-L-Ile, in contrast to poly-L-Ala. This result agrees with observations on the preferred orientation of the chains in oligopeptides that form beta-structures. It also explains the observed high relative frequencies of occurrence of Val and Ile residues in parallel beta-sheets, as compared with antiparallel sheets, in globular proteins.     

Discovery of a significant,nontopological preference for antiparallel alignment of helices with parallel regions in sheets

To help elucidate the role of secondary structure packing preferences in protein folding, here we present an analysis of the packing geometry observed between alpha-helices and between alpha-helices and beta-sheets in 1316 diverse, nonredundant protein structures. Finite-length vectors were fit to the alpha-carbon atoms in each of the helices and strands, and the packing angle between the vectors, Omega, was determined at the closest point of approach within each helix-helix or helix-sheet pair. Helix-sheet interactions were found in 391 of the proteins, and the distributions of Omega values were calculated for all the helix-sheet and helix-helix interactions. The packing angle preferences for helix-helix interactions are similar to those previously observed. However, analysis of helix-strand packing preferences uncovered a remarkable tendency for helices to align antiparallel to parallel regions of beta-sheets, independent of the topological constraints or prevalence of beta-alpha-beta motifs in the proteins. This packing angle preference is significantly diminished in helix interactions involving mixed and antiparallel beta-sheets, suggesting a role for helix-sheet dipole alignment in guiding supersecondary structure formation in protein folding. This knowledge of preferred packing angles can be used to guide the engineering of stable protein modules.     

Beta-sheet capping: signals that initiate and terminate beta-sheet formation

In the present work, we address the question of whether different amino acids have different beta-sheet initiating and terminating characteristics. Using a large scale analysis of parallel and antiparallel beta-sheets in a non-redundant dataset of proteins, we observed that most of the amino acids show significant under- or over-representation in at least one of the positions at the two ends of beta-sheets, which are denoted as N-cap and C-cap. In addition, based on statistical data and structural comparison, we found that certain amino acids, especially Asp, Asn, Gly and Pro have strong tendencies to block beta-sheet continuation. Hence, we can consider these residues as beta-sheet terminators. It was also proposed that the dipole moments in parallel beta-sheets, whose direction is from C-terminal (partially negative) to N-terminal (partially positive), are much stronger than has previously been suggested. In fact, enhancement of dipole moments in parallel beta-sheets is a result of the positioning of positively charged residues at N-cap and negatively charged residues at C-cap. This enhancement in dipole moment magnitude leads to strengthened dipolar interactions between parallel beta-sheets dipoles and other partners especially alpha-helices dipoles. The results provide an explanation for the antiparallel alignment of parallel beta-sheets with alpha-helices.     

Twist and shear in beta-sheets and beta-ribbons

The structures of the beta-sheets and the beta-ribbons have been analysed using high-resolution protein structure data. Systematic asymmetries measured in both parallel and antiparallel beta-structures include the sheet twist and the strand shear. In order to determine the origin of these asymmetries, numerous interactions and correlations were examined. The strongest correlations are observed for residues in antiparallel beta-sheets and beta-ribbons that form non-H-bonded pairs. For these residues, the sheet twist is correlated to the backbone phi angle but not to the psi angle. Our analysis supports the existence of an inter-strand C(alpha)H(alpha)...O weak H-bond, which, together with the CO...HN H-bond, constitutes a bifurcated H-bond that links neighbouring beta-strands. Residues of beta-sheets and beta-ribbons in high-resolution protein structures form a distinct region of the Ramachandran plot, which is determined by the formation of the bifurcated H-bond, the formation of an intra-strand O...H(alpha) non-bonded polar interaction, and an intra-strand O...C(beta) steric clash. Using beta-strands parameterised by phi-psi values from the allowed beta-sheet region of the Ramachandran plot, the shear and the right-hand twist can be reproduced in a simple model of the antiparallel and parallel beta-ribbon that models the bifurcated H-bonds specifically. The conformations of interior residues of beta-sheets are shown to be subsets of the conformations of residues of beta-ribbons.     

Structure and orientation study of fusion peptide FP23 of gp41 from HIV-1 alone or inserted into various lipid membrane models (mono-, bi- and multibi-layers) by FT-IR spectroscopies and Brewster angle microscopy

In the present work, we study the structure and the orientation of the 23 N-terminal peptide of the HIV-1 gp 41 protein (AVGIGALFLGFLGAAGSTMGARS) called FP23. The behaviour of FP23 was investigated alone at the air/water interface and inserted into various lipid model systems: in monolayer or multibilayers of a DOPC/cholesterol/DOPE/DOPG (6/5/3/2) and in a DMPC bilayer. PMIRRAS and polarized ATR spectroscopy coupled with Brewster angle microscopy and spectral simulations were used to precisely determine the structure and the orientation of the peptide in its environment as well as the lipid perturbations induced by the FP23 insertion. The infra-red results show the structural polymorphism of the FP23 and its ability to transit quasi irreversibly from an alpha-helix to antiparallel beta-sheets. At the air/water interface, the transition is induced by compression of the peptide alone and is modulated by compression and lipid to peptide ratio (Ri) when FP23 is inserted into a lipid monolayer. In multibilayers and in a single bilayer, there is coexistence in quasi equal proportions of alpha-helix and antiparallel beta-sheets of FP23 at low peptide content (Ri=100, 200) while antiparallel beta-sheets are predominant at high FP23 concentration (Ri=50). In (multi)bilayer systems, evaluation of dichroic ratios and sprectral simulations show that both the alpha-helix and the antiparallel beta-sheets are tilted at diluted FP23 concentrations (tilt angle of alpha-helix with respect to the normal of the interface=36.5+/-3.0 degrees for FP23 in multibilayers of DOPC/Chol/DOPE/DOPG at Ri=200 and 39.0+/-5.0 degrees in a single bilayer of DMPC at Ri=100 and tilt angle of the beta-sheets=36.0+/-2.0 degrees for the beta-sheets in multibilayers and 30.0+/-2.0 degrees in the lipid bilayer). In parallel, the FP23 induces an increase of the lipid chain disorder which shows both by an increase of the methylene stretching frequencies and an increase of the average C-C-C angle of the acyl chains. At high FP23 content (Ri=50), the antiparallel beta-sheets induce a complete disorganization of the lipid chains in (multi)bilayers.     

Solid state NMR reveals a pH-dependent antiparallel beta-sheet registry in fibrils formed by a beta-amyloid peptide

We report solid state nuclear magnetic resonance (NMR) measurements that probe the supramolecular organization of beta-sheets in the cross-beta motif of amyloid fibrils formed by residues 11-25 of the beta-amyloid peptide associated with Alzheimer's disease (Abeta(11-25)). Fibrils were prepared at pH 7.4 and pH 2.4. The solid state NMR data indicate that the central hydrophobic segment of Abeta(11-25) (sequence LVFFA) adopts a beta-strand conformation and participates in antiparallel beta-sheets at both pH values, but that the registry of intermolecular hydrogen bonds is pH-dependent. Moreover, both registries determined for Abeta(11-25) fibrils are different from the hydrogen bond registry in the antiparallel beta-sheets of Abeta(16-22) fibrils at pH 7.4 determined in earlier solid state NMR studies. In all three cases, the hydrogen bond registry is highly ordered, with no detectable "registry-shift" defects. These results suggest that the supramolecular organization of beta-sheets in amyloid fibrils is determined by a sensitive balance of multiple side-chain-side-chain interactions. Recent structural models for Abeta(11-25) fibrils based on X-ray fiber diffraction data are inconsistent with the solid state NMR data at both pH values.     

Porcine pancreatic phospholipase A2: sequence-specific 1H and 15N NMR assignments and secondary structure

The solution structure of porcine pancreatic phospholipase A2 (124 residues, 14 kDa) has been studied by two-dimensional homonuclear 1H and two- and three-dimensional heteronuclear 15N-1H nuclear magnetic resonance spectroscopy. Backbone assignments were made for 117 of the 124 amino acids. Short-range nuclear Overhauser effect (NOE) data show three alpha-helices from residues 1-13, 40-58, and 90-109, an antiparallel beta-sheet for residues 74-85, and a small antiparallel beta-sheet between residues 25-26 and 115-116. A 15N-1H heteronuclear multiple-quantum correlation experiment was used to monitor amide proton exchange over a period of 22 h. In total, 61 amide protons showed slow or intermediate exchange, 46 of which are located in the three large helices. Helix 90-109 was found to be considerably more stable than the other helices. For the beta-sheets, four hydrogen bonds could be identified. The secondary structure of porcine PLA in solution, as deduced from NMR, is basically the same as the structure of porcine PLA in the crystalline state. Differences were found in the following regions, however. Residues 1-6 in the first alpha-helix are less structured in solution than in the crystal structure. Whereas in the crystal structure residues 24-29 are involved both in a beta-sheet with residues 115-117 and in a hairpin turn, the expected hydrogen bonds between residues 24-117 and 25-29 do not show slow exchange behavior. This and the absence of several expected NOEs imply that this region has a less well defined structure in solution. Finally, the hydrogen bond between residues 78-81, which is part of a beta-sheet, does not show slow exchange behavior.     

A multimeric model for murine anti-apoptotic protein Bcl-2 and structural insights for its regulation by post-translational modification.

A monomeric model for murine antiapoptotic protein Bcl-2 was constructed by comparative modeling with the software suite MPACK (EXDIS/DIAMOD/FANTOM) using human Bcl-xL as a template. The monomeric model shows that murine Bcl-2 is an all -helical protein with a central (helix 5) hydrophobic helix surrounded by amphipathic helices and an unstructured loop of 30 residues connecting helices 1 and 2. It has been previously shown that phosphorylation of Ser 70 located in this loop region regulates the anti-apoptotic activity of Bcl-2. Based on our current model, we propose that this phosphorylation may result in a conformational change that aids multimer formation. We constructed a model for the Bcl-2 homodimer based on the experimentally determined 3D structure of the Bcl-xL: Bad peptide complex. The model shows that it will require approximately a half turn in helix 2 to expose hydrophobic residues important for the formation of a multimer. Helices 5 and 6 of the monomeric subunit Bcl-2 have been proposed to form an ion-channel by associating with helices 5 and 6 of another monomeric subunit in the higher-order complex. In the multimeric model of Bcl-2, helices 5 and 6 of each subunit were placed distantly apart. From our model, we conclude that a global conformational change may be required to bring helices 5 and 6 together during ion-channel formation.Figure Hydrophobic interactions in the dimerization groove are shown. Helix 2' of monomer B interacts through residues V90, H91, L94, A97, G98, F101 and Y105 with the hydrophobic surface formed by residues in helices 3, 4, and 5 of the monomer A. Shown here is a lateral view of monomer A depicted in a surface model with hydrophobic regions in red. The backbone of the helix is shown using a neon representation in yellow and the interacting side chains are in blue.     

Aggregation of cateslytin beta-sheets on negatively charged lipids promotes rigid membrane domains. A new mode of action for antimicrobial peptides?

Cateslytin, a positively charged (5+) arginine-rich antimicrobial peptide (bCgA, RSMRLSFRARGYGFR), was chemically synthesized and studied against membranes that mimic bacterial or mammalian systems. Circular dichroism, polarized attenuated total reflection infrared spectroscopy, (1)H high-resolution MAS NMR, and (2)H and (31)P solid state NMR were used to follow the interaction from peptide and membrane points of view. Cateslytin, which is unstructured in solution, is converted into antiparallel beta-sheets that aggregate mainly flat at the surface of negatively charged bacterial mimetic membranes. Arginine residues are involved in the binding to negatively charged lipids. Following the interaction of the cateslytin peptide, rigid and thicker membrane domains enriched in negatively charged lipids are found. Much less interaction is detected with neutral mammalian model membranes, as reflected by only minor percentages of beta-sheets or helices in the peptide secondary structure. No membrane destruction was detected for both bacterial and mammalian model membranes. A molecular model is proposed in which zones of different rigidity and thickness bring about phase boundary defects that ultimately lead to permeability induction and peptide crossing through bacterial membranes.     

Structure and mechanism of activity of the cyclic phosphodiesterase of Appr>p, a product of the tRNA splicing reaction

The crystal structure of the cyclic phosphodiesterase (CPDase) from Arabidopsis thaliana, an enzyme involved in the tRNA splicing pathway, was determined at 2.5 A resolution. CPDase hydrolyzes ADP-ribose 1",2"-cyclic phosphate (Appr>p), a product of the tRNA splicing reaction, to the monoester ADP-ribose 1"-phosphate (Appr-1"p). The 181 amino acid protein shows a novel, bilobal arrangement of two alphabeta modules. Each lobe consists of two alpha-helices on the outer side of the molecule, framing a three- or four-stranded antiparallel beta-sheet in the core of the protein. The active site is formed at the interface of the two beta-sheets in a water-filled cavity involving residues from two H-X-T/S-X motifs. This previously noticed motif participates in coordination of a sulfate ion. A solvent-exposed surface loop (residues 100-115) is very likely to play a flap-like role, opening and closing the active site. Based on the crystal structure and on recent mutagenesis studies of a homologous CPDase from Saccharomyces cerevisiae, we propose an enzymatic mechanism that employs the nucleophilic attack of a water molecule activated by one of the active site histidines.     

Crystal structure of a deubiquitinating enzyme (human UCH-L3) at 1.8 A resolution.

Ubiquitin C-terminal hydrolases catalyze the removal of adducts from the C-terminus of ubiquitin. We have determined the crystal structure of the recombinant human Ubiquitin C-terminal Hydrolase (UCH-L3) by X-ray crystallography at 1.8 A resolution. The structure is comprised of a central antiparallel beta-sheet flanked on both sides by alpha-helices. The beta-sheet and one of the helices resemble the well-known papain-like cysteine proteases, with the greatest similarity to cathepsin B. This similarity includes the UCH-L3 active site catalytic triad of Cys95, His169 and Asp184, and the oxyanion hole residue Gln89. Papain and UCH-L3 differ, however, in strand and helix connectivity, which in the UCH-L3 structure includes a disordered 20 residue loop (residues 147-166) that is positioned over the active site and may function in the definition of substrate specificity. Based upon analogy with inhibitor complexes of the papain-like enzymes, we propose a model describing the binding of ubiquitin to UCH-L3. The UCH-L3 active site cleft appears to be masked in the unliganded structure by two different segments of the enzyme (residues 9-12 and 90-94), thus implying a conformational change upon substrate binding and suggesting a mechanism to limit non-specific hydrolysis.     

A hierarchic approach to the design of hexameric helical barrels

The design of large macromolecular assemblies is an endeavor with implications for protein engineering as well as nanotechnology. A hierarchic approach was used to design an antiparallel hexameric, tubular assembly of helices. In previous studies, a domain-swapped, dimeric three-helix bundle was designed from first principles. In the crystal lattice, three dimers associate around a 3-fold rotational axis to form a hexameric assembly. Although this hexameric assembly was not observed in solution, it was possible to stabilize its formation by changing three polar residues per monomer to hydrophobic (two Phe and one Trp) residues. Molecular models based on the crystallographic coordinates of DSD (PDB accession code 1G6U) show that these side-chains pack in the central cavity (the "supercore") of the hexameric bundle. Analytical ultracentrifugation, fluorescence spectroscopy, CD spectroscopy, and guanidine-HCl denaturation were used to determine the assembly of the hexamer. To probe the requirements for stabilizing the hexamer, we systematically varied the polarity and steric bulk of one of the Phe residues in the supercore of the hexamer. Depending on the nature of this side-chain, it is possible to modulate the stability of the hexamer in a predictable manner. This family of hexameric proteins may provide a useful framework for the construction of proteins that change their oligomeric states in response to binding of small molecules.