Conserved amino acid motifs from the novel Piv/MooV family of transposases and site-specific recombinases are required for catalysis of DNA inversion by Piv.

Piv, a site-specific invertase from Moraxella lacunata, exhibits amino acid homology with the transposases of the IS110/IS492 family of insertion elements. The functions of conserved amino acid motifs that define this novel family of both transposases and site-specific recombinases (Piv/MooV family) were examined by mutagenesis of fully conserved amino acids within each motif in Piv. All Piv mutants altered in conserved residues were defective for in vivo inversion of the M. lacunata invertible DNA segment, but competent for in vivo binding to Piv DNA recognition sequences. Although the primary amino acid sequences of the Piv/MooV recombinases do not contain a conserved DDE motif, which defines the retroviral integrase/transposase (IN/Tnps) family, the predicted secondary structural elements of Piv align well with those of the IN/Tnps for which crystal structures have been determined. Molecular modelling of Piv based on these alignments predicts that E59, conserved as either E or D in the Piv/MooV family, forms a catalytic pocket with the conserved D9 and D101 residues. Analysis of Piv E59G confirms a role for E59 in catalysis of inversion. These results suggest that Piv and the related IS110/IS492 transposases mediate DNA recombination by a common mechanism involving a catalytic DED or DDD motif.

The C-terminal tail of UNC-60B (actin depolymerizing factor/cofilin) is critical for maintaining its stable association with F-actin and is implicated in the second actin-binding site.

Actin depolymerizing factor (ADF)/cofilin changes the twist of actin filaments by binding two longitudinally associated actin subunits. In the absence of an atomic model of the ADF/cofilin-F-actin complex, we have identified residues in ADF/cofilin that are essential for filament binding. Here, we have characterized the C-terminal tail of UNC-60B (a nematode ADF/cofilin isoform) as a novel determinant for its association with F-actin. Removal of the C-terminal isoleucine (Ile152) by carboxypeptidase A or truncation by mutagenesis eliminated F-actin binding activity but strongly enhanced actin depolymerizing activity. Replacement of Ile152 by Ala had a similar but less marked effect; F-actin binding was weakened and depolymerizing activity slightly enhanced. Truncation of both Arg151 and Ile152 or replacement of Arg151 with Ala also abolished F-actin binding and enhanced depolymerizing activity. Loss of F-actin binding in these mutants was accompanied by loss or greatly decreased severing activity. All of the variants of UNC-60B interacted with G-actin in an indistinguishable manner from wild type. Cryoelectron microscopy showed that UNC-60B changed the twist of F-actin to a similar extent to vertebrate ADF/cofilins. Helical reconstruction and structural modeling of UNC-60B-F-actin complex reveal how the C terminus of UNC-60B might be involved in one of the two actin-binding sites.

Blood flow regulation by S-nitrosohemoglobin in the physiological oxygen gradient.

The binding of oxygen to heme irons in hemoglobin promotes the binding of nitric oxide (NO) to cysteinebeta93, forming S-nitrosohemoglobin. Deoxygenation is accompanied by an allosteric transition in S-nitrosohemoglobin [from the R (oxygenated) to the T (deoxygenated) structure] that releases the NO group. S-nitrosohemoglobin contracts blood vessels and decreases cerebral perfusion in the R structure and relaxes vessels to improve blood flow in the T structure. By thus sensing the physiological oxygen gradient in tissues, hemoglobin exploits conformation-associated changes in the position of cysteinebeta93 SNO to bring local blood flow into line with oxygen requirements.

Puzzle pieces defined: locating common packing units in tertiary protein contacts.

Puzzle pieces are defined as small packing units which make up the unique tertiary interactions in proteins. Anti-parallel and perpendicular helix-helix contacts were broken down into basic puzzle-piece pairs in order to study the traits of such contacts: their limited geometry, preferred residue involvement, residue conformation and other common constraints. These traits can then be used for continued comparison of other protein structures, improving models of and designing proteins de novo and, in time, predicting 3D structure from primary sequence. Results from a small (100 proteins) database of anti-parallel helix-helix contacts and from preliminary work on a large database (600 proteins) of perpendicular helix-helix contacts are presented.

The Alacoil: a very tight, antiparallel coiled-coil of helices.

The Alacoil is an antiparallel (rather than the usual parallel) coiled-coil of alpha-helices with Ala or another small residue in every seventh position, allowing a very close spacing of the helices (7.5-8.5 A between local helix axes), often over four or five helical turns. It occurs in two distinct types that differ by which position of the heptad repeat is occupied by Ala and by whether the closest points on the backbone of the two helices are aligned or are offset by half a turn. The aligned, or ROP, type has Ala in position "d" of the heptad repeat, which occupies the "tip-to-tip" side of the helix contact where the C alpha-C beta bonds point toward each other. The more common offset, or ferritin, type of Alacoli has Ala in position "a" of the heptad repeat (where the C alpha-C beta bonds lie back-to-back, on the "knuckle-touch" side of the helix contact), and the backbones of the two helices are offset vertically by half a turn. In both forms, successive layers of contact have the Ala first on one and then on the other helix. The Alacoil structure has much in common with the coiled-coils of fibrous proteins or leucine zippers: both are alpha-helical coiled-coils, with a critical amino acid repeated every seven residues (the Leu or the Ala) and a secondary contact position in between. However, Leu zippers are between aligned, parallel helices (often identical, in dimers), whereas Alacoils are between antiparallel helices, usually offset, and much closer together. The Alacoil, then, could be considered as an "Ala anti-zipper." Leu zippers have a classic "knobs-into-holes" packing of the Leu side chain into a diamond of four residues on the opposite helix; for Alacoils, the helices are so close together that the Ala methyl group must choose one side of the diamond and pack inside a triangle of residues on the other helix. We have used the ferritin-type Alacoil as the basis for the de novo design of a 66-residue, coiled helix hairpin called "Alacoilin." Its sequence is: cmSPDQWDKE AAQYDAHAQE FEKKSHRNng TPEADQYRHM ASQY QAMAQK LKAIANQLKK Gsetcr (with "a" heptad positions underlined and nonhelical parts in lowercase), which we will produce and test for both stability and uniqueness of structure.

The tyrosine corner: a feature of most Greek key beta-barrel proteins.

The Tyr corner is a conformation in which a tyrosine (residue "Y") near the beginning or end of an antiparallel beta-strand makes an H bond from its side-chain OH group to the backbone NH and/or CO of residue Y - 3, Y - 4, or Y - 5 in the nearby connection. The most common "classic" case is a delta 4 Tyr corner (more than 40 examples listed), in which the H bond is to residue Y - 4 and the Tyr chi 1 is near -60 degrees. Y - 2 is almost always a glycine, whose left-handed beta or very extended beta conformation helps the backbone curve around the Tyr ring. Residue Y - 3 is in polyproline II conformation (often Pro), and residue Y - 5 is usually a hydrophobic (often Leu) that packs next to the Tyr ring. The consensus sequence, then, is LxPGxY, where the first x (the H-bonding position) is hydrophilic. Residues Y and Y - 2 both form narrow pairs of beta-sheet H-bonds with the neighboring strand. delta 5 Tyr corners have a 1-residue insertion between the Gly and Tyr, forming a beta-bulge. One protein family has a delta 4 corner formed by a His rather than a Tyr, and several examples use Trp in place of Tyr. For almost all these cases, the protein or domain is a Greek key beta-barrel structure, the Tyr corner ends a Greek key connection, and it is well-conserved in related proteins. Most low-twist Greek key beta-barrels have 1 Tyr corner. "Reverse" delta 4 Tyr corners (H bonded to Y + 4) and other variants are described, all less common and less conserved. It seems likely that the more classic Tyr corners (delta 4, delta 5, and delta 3 Tyr, Trp, or His) contribute to the stability of a Greek key connection over a hairpin connection, and also that they may aid in the process of folding up Greek key structures.

Looking at proteins: representations, folding, packing, and design. Biophysical Society National Lecture, 1992.

Looking at proteins is an active process of interpretation and selection, emphasizing some features and deleting others. Multiple representations are needed, for such purposes as showing motions or conveying both the chain connectivity and the three-dimensional shape simultaneously. In studying and comparing protein structures, ideas are suggested about the determinants of tertiary structure and of folding (e.g., that Greek key beta barrels may fold up two strands at a time). The design and synthesis of new proteins "from scratch" provides a route toward the experimental testing of such ideas. It has also been a fruitful new perspective from which to look at structures, requiring such things as statistics on very narrowly defined structural categories and explicit attention to "negative design" criteria that actively block unwanted alternatives (e.g., reverse topology of a helix bundle, or edge-to-edge aggregation of beta sheets). Recently, the field of protein design has produced a rather unexpected general result: apparently we do indeed know enough to successfully design proteins that fold into approximately correct structures, but not enough to design unique, native-like structures. The degree of order varies considerably, but even the best designed material shows multiple conformations by NMR, more similar to a "molten globule" folding intermediate than to a well ordered native tertiary structure. In response to this conclusion, we are now working on systems that test useful questions with approximate structures (such as determining which factors most influence the choice of helix-bundle topology) and also analyzing how natural proteins achieve unique core conformations (e.g., for side chains on the interior side of a beta sheet, illustrated in the kinemages).

Three-dimensional structure of human serum albumin.

The three-dimensional structure of human serum albumin has been solved at 6.0 angstrom (A) resolution by the method of multiple isomorphous replacement. Crystals were grown from solutions of polyethylene glycol in the infrequently observed space group P42(1)2 (unit cell constants a = b = 186.5 +/- 0.5 A and c = 81.0 +/- 0.5 A) and diffracted x-rays to lattice d-spacings of less than 2.9 A. The electron density maps are of high quality and revealed the structure as a predominantly alpha-helical globin protein in which the course of the polypeptide can be traced. The binding loci of several organic compounds have been determined.

A simple apparatus for controlling nucleation and size in protein crystal growth.

A simple device is described for controlling vapor equilibrium in macromolecular crystallization as applied to the protein crystal growth technique commonly referred to as the "hanging drop" method. Crystal growth experiments with hen egg white lysozyme have demonstrated control of the nucleation rate. Nucleation rate and final crystal size have been found to be highly dependent upon the rate at which critical supersaturation is approached. Slower approaches show a marked decrease in the nucleation rate and an increase in crystal size.