Alpha-helical coiled coils and bundles: how to design an alpha-helical protein

Nuclear matrix proteins as structural and functional components of the mitotic apparatus

The eukaryotic nucleus is a membrane-enclosed compartment containing the genome and associated organelles supported by a complex matrix of nonhistone proteins. Identified as the nuclear matrix, this component maintains spatial order and provides the structural framework needed for DNA replication, RNA synthesis and processing, nuclear transport, and steroid hormone action. During mitosis, the nucleoskeleton and associated chromatin is efficiently dismantled, packaged, partitioned, and subsequently reassembled into daughter nuclei. The dramatic dissolution of the nucleus is accompanied by the assembly of a mitotic apparatus required to facilitate the complex events associated with nuclear division. Until recently, little was known about the fate or disposition of nuclear matrix proteins during mitosis. The availability of specific molecular probes and imaging techniques, including confocal microscopy and improved immunoelectron microscopy using resinless sections and related procedures, has enabled investigators to identify and map the distribution of nuclear matrix proteins throughout the cell cycle. This chapter will review the structure, function, and distribution of the protein NuMA (nuclear matrix mitotic apparatus) and other nuclear matrix proteins that depart the nucleus during the interphase/mitosis transition to become structural and functional components within specific domains of the mitotic apparatus.

Boehringer Mannheim award lecture 1995. La conference Boehringer Mannheim 1995. De novo design of alpha-helical proteins: basic research to medical applications

The two-stranded alpha-helical coiled-coil is a universal dimerization domain used by nature in a diverse group of proteins. The simplicity of the coiled-coil structure makes it an ideal model system to use in understanding the fundamentals of protein folding and stability and in testing the principles of de novo design. The issues that must be addressed in the de novo design of coiled-coils for use in research and medical applications are (i) controlling parallel versus antiparallel orientation of the polypeptide chains, (ii) controlling the number of helical strands in the assembly (iii) maximizing stability of homodimers or heterodimers in the shortest possible chain length that may require the engineering of covalent constraints, and (iv) the ability to have selective heterodimerization without homodimerization, which requires a balancing of selectivity versus affinity of the dimerization strands. Examples of our initial inroads in using this de novo design motif in various applications include: heterodimer technology for the detection and purification of recombinant peptides and proteins; a universal dimerization domain for biosensors; a two-stage targeting and delivery system; and coiled-coils as templates for combinatorial helical libraries for basic research and drug discovery and as synthetic carrier molecules. The universality of this dimerization motif in nature suggests an endless number of possibilities for its use in de novo design, limited only by the creativity of peptide-protein engineers.

The C-terminal region of the UvrB protein of Escherichia coli contains an important determinant for UvrC binding to the preincision complex but not the catalytic site for 3'-incision

The UvrABC endonuclease from Escherichia coli repairs damage in the DNA by dual incision of the damaged strand on both sides of the lesion. The incisions are in an ordered fashion, first on the 3'-side and next on the 5'-side of the damage, and they are the result of binding of UvrC to the UvrB-DNA preincision complex. In this paper, we show that at least the C-terminal 24 amino acids of UvrB are involved in interaction with UvrC and that this binding is important for the 3'-incision. The C-terminal region of UvrB, which shows homology with a domain of the UvrC protein, is part of a region that is predicted to be able to form a coiled-coil. We therefore propose that UvrB and UvrC interact through the formation of such a structure. The C-terminal region of UvrB only interacts with UvrC when present in the preincision complex, indicating that the conformational change in UvrB accompanying the formation of this complex exposes the UvrC binding domain. Binding of UvrC to the C-terminal region of UvrB is not important for the 5'-incision, suggesting that for this incision a different interaction of UvrC with the UvrB-DNA complex is required. Truncated UvrB mutants that lack up to 99 amino acids from the C terminus still give rise to significant incision (1-2%), indicating that this C-terminal region of UvrB does not participate in the formation of the active site for 3'-incision. This region, however, contains the residue (Glu-640) that was proposed to be involved in 3'-catalysis, since a mutation of the residue (E640A) fails to promote 3'-incision (Lin, J.J., Phillips, A.M., Hearst, J.E., and Sancar, A. (1992) J. Biol. Chem. 267, 17693-17700). We have isolated a mutant UvrB with the same E640A substitution, but this protein shows normal UvrC binding and incision in vitro and also results in normal survival after UV irradiation in vivo. As a consequence of these results, it is still an open question as to whether the catalytic site for 3'-incision is located in UvrB, in UvrC, or is formed by both proteins.

The effect of pH on the folding and stability of the myosin rod

The far-ultraviolet circular dichroism and fluorescence emission intensities of the myosin rod were studied at pH 2-11, in the absence and presence of guanidine hydrochloride. The protein kept its helicity in this pH range. Its stability in the denaturant was higher at acidic pH than at pH 7. This may be due to favorable interactions involving protonated glutamic acid residues at the interface of the two alpha-helical chains of the molecule. At alkaline pH, the fluorescence of the myosin rod was quenched, and the tryptophan region of the protein became less stable in the presence of guanidine hydrochloride, due to ionization of tyrosine residues or other amino acids close to tryptophans in the double-helix arrangement.

Genetic strategy for analyzing specificity of dimer formation: Escherichia coli cyclic AMP receptor protein mutant altered in its dimerization specificity

Many transcriptional regulators function in homo- or heterodimeric combinations. The same protein can carry out distinct regulatory functions depending on the partner with which it associates. Here, we describe a mutant of the Escherichia coli cAMP receptor protein (CRP) that has an altered dimerization specificity; that is, mutant/mutant homodimers form preferentially over wild-type/mutant heterodimers. CRP dimerization involves the formation of a parallel coiled-coil structure, and our CRP mutant bears an amino acid substitution affecting the first "d" position residue within the alpha-helix that mediates CRP dimerization. The genetic strategy we used to isolate this CRP altered dimerization specificity (ADS) mutant is generalizable and could be utilized to isolate ADS mutants of other dimeric transcriptional regulators.

Identification of a three-amino-acid region in G protein gamma 1 as a determinant of selective beta gamma heterodimerization

Guanine-nucleotide-binding protein beta and gamma subunits belong to large protein families encompassing at least five and ten members, respectively, from mammalian cells. The formation of stable beta gamma heterodimers is a selective process determined by the primary sequences of both the beta and gamma subunit. For example, gamma 2 dimerizes with both beta 1 and beta 2, gamma 1 with beta 1, but not with beta 2. To identify the structural elements of gamma subunits relevant to the selectivity of beta gamma dimerization, we have used the baculovirus-insect cell-expression system to produce chimeric beta and gamma subunits and have studied their dimerization using an assay based on the ability of isoprenylation-resistant gamma subunit mutants to draw beta subunits into the cytosol and including sucrose density gradient analysis of soluble recombinant beta gamma dimers. The results show that replacement of three consecutive residues of gamma 1, Cys36-Cys37-Glu38, by the corresponding residues of gamma 2, Ala33-Ala34-Ala35, suffices to render the mutant gamma 1 subunit capable of forming heterodimers with beta 2. The ability of mutant gamma 1 subunits to dimerize with beta 2 does not correlate with the probability of the mutated region to participate in coiled-coil structures. The tripeptide region identified here as a critical determinant of the selectivity of beta gamma dimer formation is distinct from, but partially overlaps with, the region reported by Lee et al. [Lee, C., Murakami, T. & Simonds, W. F. (1995) J. Biol. Chem. 270, 8779-8784]. The results of this study, therefore, not only extend the region of gamma 1 selecting between beta 1 and beta 2 to the five-residue sequence between Cys36 and Phe40, but also argue against the notion that the hydrophobic terminal residue of this motif represents the key determinant of selective beta gamma interaction.

The effects of interhelical electrostatic repulsions between glutamic acid residues in controlling the dimerization and stability of two-stranded alpha-helical coiled-coils

The effects of interhelical electrostatic repulsions in controlling the dimerization and stability of two-stranded alpha-helical coiled-coils have been studied using de novo designed synthetic coiled-coils. A native coiled-coil was snythesized, which consisted of two identical 35-residue polypeptide chains with a heptad repeat QgVaGbAcLdQeKf and a Cys residue at position 2 to allow formation of an interchain 2-2' disulfide bridge. This peptide, designed to contain no intrachain or interchain electrostatic interactions, forms a stable coiled-coil structure at 20 degrees C in benign medium (50 mM KCl, 25 mM PO4, pH 7) with a [urea]1/2 value of 6.1 M. Five mutant coiled-coils were designed in which Gln residues at the e and g positions of the heptad repeat were substituted with Glu systematically from the N terminus toward the C terminus, resulting in each polypeptide chain having 2, 4, 6, 8, or 10 Glu residues. These substituted Glu residues are able to form interchain i to i' +5 electrostatic repulsions across the dimer interface. As the number of interchain repulsions increases, a steady loss of helical content is observed by circular dichroism spectroscopy. The effects of the interchain Glu-Glu repulsions on the coiled-coil structure are partly overcome by the presence of an interchain disulfide bridge; the peptide with six Glu substitutions is only 15% helical in the reduced form but 85% helical in the oxidized form. The stabilities of the coiled-coils were determined by urea and guanidine hydrochloride (GdnHCl) denaturation studies at 20 degrees C. The stabilities of the coiled-coils determined by urea denaturation indicate a decrease in stability, which correlates with an increasing number of interchain repulsions ([urea]1/2 values ranging from 8.4 to 3.7 M in the presence of M KCl). In contrast, all coiled-coils had similar stabilities when determined by GdnHCl denaturation (approximately 2.9 M). KCl could not effectively screen the effects of interchain repulsions on coiled-coil stability as compared to GdnHCl.

Alternative mechanisms of CAK assembly require an assembly factor or an activating kinase

We have cloned a mouse cDNA that encodes p36, a novel subunit of the CDK-activating kinase (CAK). p36 contains a C3HC4 zinc-binding domain or RING factor and is associated both with a TFIIH-bound form of CAK and with a free trimeric form. p36 promotes the assembly of CDK7 and cyclin H in vitro, stabilizing the transient CDK7-cyclin H complex. Stabilization and activation of CAK by p36 is independent of the phosphorylation state of T170, the conserved activating residue of CDK7. Assembly of active CDK7-cyclin H dimers can also occur through an alternative p36-independent pathway that requires phosphorylation of T170 by a CAK-activating kinase, or CAKAK. Thus, CDK7-cyclin H complex formation can be achieved by multiple mechanisms.

Hagfish biopolymer: a type I/type II homologue of epidermal keratin intermediate filaments

In contrast to most intermediate filaments (IF) which function intracellularly or constitute epidermal appendages, the single massive (approximately 60 cm length, approximately 3 microns width) IF-rich 'thread' biopolymer synthesized by the specialized hagfish gland thread cell is released extracellularly via holocrine secretion to interact with mucins and seawater, thereby modifying the viscoelastic properties of the copious mucous exudate. Recently, using the Pacific hagfish (Eptatretus stouti, class Agnatha), a jawless scaleless marine vertebrate of ancient lineage, we determined that the deduced amino acid sequence of one thread IF chain (alpha, 66.6 kDa, native pI 7.5) contained an atypical, threonine-rich central rod domain of low identity (< 30%) with other vertebrate IF types, but that the N- and C-terminal domains exhibited several keratin-like features. From these and other unexpected characteristics, it was concluded that hagfish alpha is best categorized as a type II homologue of an epidermal keratin. We now report the deduced sequence of a second thread IF subunit (gamma, 62.7 kDa, native pI 5.3) which is co-expressed and co-assembles in vitro with alpha in a 1:1 ratio. As was found for alpha, the N- and C-terminal domains of gamma have keratin-like parameters, but the central rod has low identity to IFs of types I-V (< 31%), a cephalochordate IF (< 29%) and invertebrate IFs (< 20%) and no particular homology to type I or type II keratins. Central rod identity between gamma and alpha is also low (approximately 23%), as is typical of comparisons between different rod types but atypical of similar rod types (> 50%). The central rods of both gamma and alpha lack the 42-residue insert of helix 1B present in lamins and invertebrate IFs, have unusually high threonine contents (gamma, 10%; alpha, 13%) compared to other IF types (2-5%), contain a number of unexpected residues in consensus conserved sites, and employ a L12 segment of 21 residues rather than the 16 or 17 residues found in keratins. Theoretical analyses indicate that the hagfish molecules exist as coiled coil heterodimers (alpha/gamma) in which the chains are parallel, in axial register, and stabilized by significant numbers of ionic interactions. Fast Fourier-transform analyses revealed that the linear distribution period of approximately 9.55 for basic and acidic residues in other IF chains is not completely maintained, partly due to the high threonine content. The threonine residues occupy mainly outer sites b, c, f in the heptad substructure, possibly abetting parallel alignment of thousands of IFs within the thread, interactions with mucins at the thread periphery, and hierarchical IF chain assembly. It is suggested that the gamma and alpha chains from this most primitive extant vertebrate are type I and type II homologues of epidermal keratin chains, possibly related to early specialized keratins.

Molecular modeling indicates that homodimers form the basis for intermediate filament assembly from human and mouse epidermal keratins

Mammalian epidermal keratin molecules adopt rod-shaped conformations that aggregate to form cytoplasmic intermediate filaments. To investigate these keratin conformations and the basis for their patterns of molecular association, graphical methods were developed to relate known amino acid sequences to probable spacial configurations. The results support the predominantly alpha-helical conformation of keratin chains, interrupted by short non-alpha-helical linkages. However, it was found that many of the linkages have amino acid sequences typical of beta-strand conformations. Space-filling atomic models revealed that the beta-strand sequences would permit the formation of 2-chain and 4-chain cylindrical beta-helices, fully shielding the hydrophobic amino acid chains that alternate with hydrophilic residues in these sequences. Because of the locations of the beta-helical regions in human and mouse stratum corneum keratin chains, only homodimers of the keratins could interact efficiently to form 2-chain and 4-chain beta-helices. Tetramers having the directions and degrees of overlap of constituent dimers that have been identified by previous investigators are also predicted from the interactions of beta-helical motifs. Heterotetramers formed from dissimilar homodimers could combine, through additional beta-helical structures, to form higher oligomers having the dimensions seen in electron microscopic studies. Previous results from chemical crosslinking studies can be interpreted to support the concept of homodimers rather than heterodimers as the basis for keratin filament assembly.

Design, expression, and initial characterization of MB1, a de novo protein enriched in essential amino acids

Using recently emerging protein folding principles we have designed a protein enriched in the essential amino acids methionine, threonine, lysine and leucine. Our preliminary study of consensus residues (based on charge, hydrophobicity and volume) of natural alpha-helical bundle proteins indicated that the residues M, T, K, and L could be inserted in an alpha-helical bundle structure. We therefore attempted to create a stable de novo protein, highly enriched in these essential amino acids, that would adopt the alpha-helical bundle fold. The design process was an iterative one. The consensus residues (based on the properties profile) for bundle helices were found considering the four helices taken together, helices I to IV individually, or only their N- and C-termini. Using these data, the helices in our de novo protein were designed by inserting the residues M, T, K and L as often as possible at positions where their volume, hydrophobicity and charge match the consensus found in natural bundle helices. Short sequences of strong turn formers were used to join the helices and adjust the predicted p1 to 7.7, while a number of local and global factors were used to refine our design. Further, the sequence was checked to eliminate various known protease targets in E. coli. The sequence of our de novo protein, MB1, is: MAT-EDMTDMMTTLFKTMQLLTK-SEPTA-MDEATKTATTMKNHLQNLMQK-TKNKE DMTDMATTYFKTMQLLTK-TEPSA-MDEATKTATTMKNHLQNLMQK-GVA+ ++ , where dashes separate long helices from short, turn forming linkers. A gene coding for this protein was assembled from synthetic oligonucleotides, then fused to the maltose binding protein gene under the control of a tac promoter. The fusion protein was expressed in E. coli, purified and cleaved to yield maltose binding protein and our de novo protein, MB1. MB1 was found to be helical, to have the expected molecular weight (11 kDa) and the expected content (57%) of the essential amino acids M, T, K and L.

Predicting coiled coils by use of pairwise residue correlations

A method is presented that predicts coiled-coil domains in protein sequences by using pairwise residue correlations obtained from a (two-stranded) coiled-coil database of 58,217 amino acid residues. A program called PAIRCOIL implements this method and is significantly better than existing methods at distinguishing coiled coils from alpha-helices that are not coiled coils. The database of pairwise residue correlations suggests structural features that stabilize or destabilize coiled coils.

Characterization of a new four-chain coiled-coil: influence of chain length on stability

Limited information is available on inherent stabilities of four-chain-coils. We have developed a model system to study this folding motif using synthetic peptides derived from sequences contained in the tetramerization domain of Lac repressor. These peptides are tetrameric as judged by both gel filtration and sedimentation equilibrium and the tetramers are fully helical as determined by CD. The four-chain coiled-coils are well folded as judged by the cooperativity of thermal unfolding and by the extent of dispersion in aliphatic chemical shifts seen in NMR spectra. In addition, we measured the chain length dependence of this four-chain coiled-coil. To this end, we developed a general procedure for nonlinear curve fitting of denaturation data in oligomeric systems. The dissociation constants for bundles that contain alpha-helical chains 21, 28, and 35 amino acids in length are 3.1 x 10(-12), 6.7 x 10(-23), and 1.0 x 10(-38) M3, respectively. This corresponds to tetramer stabilities (in terms of the peptide monomer concentration) of 180 microM, 51 nM, and 280 fM, respectively. Finally, we discuss the rules governing coiled-coil formation in light of the work presented here.

Predicting oligomerization states of coiled coils

An algorithm based on the profile method was developed that faithfully distinguishes between the amino acid sequences of dimeric and trimeric coiled coils. Normalized sequence profiles derived from nonhomologous, two- and three-stranded, coiled-coil sequences with unambiguous registers were used to assign dimer and trimer propensities to test sequences. The difference between the dimer and trimer profile scores accurately reflected the preferred oligomerization state. The method relied on two strategies that may be generally applicable to profile calculations--profile values of solvent-exposed residues and of amino acids that were underrepresented in the data-base were given zero weight. Differences between the dimer and trimer profiles revealed sequence patterns that match and extend experimental studies of oligomer specification.

Identification of two highly homologous presynaptic proteins distinctly localized at the dendritic and somatic synapses

Through screening of a murine brain cDNA library, we have isolated two brain specific cDNAs encoding highly homologous proteins, named 921-L and 921-S, comprised of 134 amino acids with 80% identity. Immunohistological study with the mAbs raised against the bacterially expressed 921 proteins showed that 921-L protein is distributed at the dendritic region and 921-S at the neuronal somatic surface. Immuno-electron microscopic study revealed that both 921 proteins are localized at the presynaptic terminal, indicating that the 921 proteins are differentially expressed at the dendritic and somatic presynapses.

A peptide fragment of human DNA topoisomerase II alpha forms a stable coiled-coil structure in solution

Results are presented on a peptide fragment (1013-1056) from human DNA topoisomerase II alpha. This was selected using the procedure of Lupas et al. (Lupas, A., Van Dyke, M., and Stock, J. (1991) Science 252, 1162-1164) for its potential to adopt a stable coiled-coil structure. The same theoretical treatment rejected the segment 994-1021 proposed by Zwelling and Perry (Zwelling, L. A., and Perry, W. M. (1989) Mol. Endocrinol. 3, 603-604) as a possible core for leucine-zipper formation. Our experimental studies combine cross-linking and CD analysis. Cross-linking establishes that the 1013-1056 fragment forms a stable homodimer in solution. Effects of increasing peptide concentration on CD spectra confirm that only the 1013-1056 fragment can undergo a coiled-coil stabilization from an isolated alpha-helix. Unfolding experiments further show that the coiled-coil is more stable in guanidium chloride than in urea. Values of -6.8 and -7.4 kcal/mol for the dimerization free energy are determined by thermal and urea unfolding, respectively. These are strikingly similar to the value recently found for the dissociation/reassociation of the entire yeast topoisomerase II from sedimentation equilibrium experiments (Lamhasni, S., Larsen, A. K., Barray, M., Monnot, M., Delain, E., and Fermandjian, S. (1995) Biochemistry 34, 3632-3639), although their significance relatively to topoisomerase II undoubtedly requires further analysis.

Molecular characterization of NDP52, a novel protein of the nuclear domain 10, which is redistributed upon virus infection and interferon treatment

The nuclear domain (ND)10 also described as POD or Kr bodies is involved in the development of acute promyelocytic leukemia and virus-host interactions. Immunofluorescence analysis using a variety of human autoimmune sera and monoclonal antibodies showed a typical dot like nuclear staining for ND10, suggesting that this structure consists of several proteins. Two of the ND10 proteins, Sp100 and PML are genetically characterized and show homology with several transcription factors. Here we describe NDP52, an additional novel protein of the ND10. We raised a new mAb C8A2, that specifically recognizes NDP52. Immunofluorescence analysis using this mAb showed a typical nuclear dot staining as it was described for ND10. Isolation and sequencing of the corresponding cDNA revealed that NDP52 has a predicted molecular mass of 52 kD. The deduced amino acid sequence exhibits an extended central coiled coil domain containing a leucine zipper motif. The COOH terminus of NDP52 shows homology with LIM domains, that have recently been described to mediate protein interactions, which let NDP52 appear as a suitable candidate for mediating interactions between ND10 proteins. In vivo, NDP52 is transcribed in all human tissues analyzed. Furthermore, we show that NDP52 colocalizes with the ND10 protein PML and can be redistributed upon viral infection and interferon treatment. These data suggest that ND10 proteins play an important role in the viral life cycle.

Quantitative approaches to utilizing mutational analysis and disulfide crosslinking for modeling a transmembrane domain

The transmembrane domain of chemoreceptor Trg from Escherichia coli contains four transmembrane segments in its native homodimer, two from each subunit. We had previously used mutational analysis and sulfhydryl cross-linking between introduced cysteines to obtain data relevant to the three-dimensional organization of this domain. In the current study we used Fourier analysis to assess these data quantitatively for periodicity along the sequences of the segments. The analyses provided a strong indication of alpha-helical periodicity in the first transmembrane segment and a substantial indication of that periodicity for the second segment. On this basis, we considered both segments as idealized alpha-helices and proceeded to model the transmembrane domain as a unit of four helices. For this modeling, we calculated helical crosslinking moments, parameters analogous to helical hydrophobic moments, as a quantitative way of condensing and utilizing a large body of crosslinking data. Crosslinking moments were used to define the relative separation and orientation of helical pairs, thus creating a quantitatively derived model for the transmembrane domain of Trg. Utilization of Fourier transforms to provide a quantitative indication of periodicity in data from analyses of transmembrane segments, in combination with helical crosslinking moments to position helical pairs should be useful in modeling other transmembrane domains.

Design of two-stranded and three-stranded coiled-coil peptides

The structural features required for the formation of two- versus three-stranded coiled coils have been explored using de novo protein design. Peptides with leucine at the 'a' and 'd' positions of a coiled-coil (general sequence: Leua Xaab Xaac Leud Glue Xaaf Lysg) exist in a non-cooperative equilibrium between unstructured monomers and helical dimers and helical trimers. Substituting valine into each 'a' position produces peptides which still form trimers at high concentrations, whereas substitution of a single asparagine at the 'a' position of the third heptad yields a dimer. During the course of this work, we also re-investigated a helical propensity scale derived using a series of coiled-coil peptides previously believed to exist in a monomer-dimer equilibrium (O'Neil & DeGrado 1990). Detailed analysis of the concentration dependence of ellipticity at 222 nm reveals that they exist in a non-cooperative monomer-dimer-trimer equilibrium. However, the concentration of trimer near the midpoint of the concentration-dependent transition is small, so the previously determined values of delta delta G alpha using the approximate monomer-dimer scheme are indistinguishable from the values obtained employing the complete monomer-dimer-trimer equilibrium.

Isolation, characterization, and primary structure of a calcium-binding 63-kDa bone protein

A novel noncollagenous protein of the mineralized matrix of bovine bone was isolated by ion exchange and gel permeation chromatography. The apparent M(r) of the protein is 63,000 as determined by SDS-polyacrylamide gel electrophoresis. The protein is a rather minor constituent in bone and could not be detected in other connective tissues by enzyme-linked immunosorbent assay of guanidine HCl extracts. The 63-kDa protein was detected in the osteoid and around the osteocytes upon immuno-histochemical staining of bovine compact bone. The sequence of the 63-kDa protein was deduced from cDNA clones isolated from a rat calvaria lambda gt11 expression library. The protein contains two centrally located EF-hand Ca(2+)-binding domains. Seven heptad repeats are present indicating the ability of the protein for coiled-coil interactions. Ability to bind calcium was confirmed by 45Ca2+ binding to protein blotted onto nitrocellulose membrane. The protein was synthesized in calvaria explants as detected by immunoprecipitation of radiolabeled protein from the culture medium. Although the protein can be detected in biochemical amounts in bone only, varying amounts of mRNA for this protein were detected in several rat tissues by RNase protection assay with highest levels in rat calvaria. This extracellular protein corresponds to a mouse protein called nucleobindin.

Some properties of acid-reassembled tropomyosin

The preference for parallelism of the two chains in tropomyosin coiled coils is thought to result from interchain salt bridges. To examine this idea, studies are presented of tropomyosin molecules reassembled from chaotropic solvents in acid solution, where cross-links cannot exist. The acid-reassembled molecules are appreciably less disulfide cross-linkable in acid than native molecules, a result explainable if some antiparallel dimers indeed form at low pH. Physical studies (backbone- and tyrosine-region CD and intrinsic viscosity) indicate that refolding in acid yields a molecular population demonstrably different in tyrosine-region CD from native, but having comparable (but not identical) helix content, thermal stability, and dimensions. Moreover, the refolding in acid after either thermal or chaotropic-solvent denaturation yields the same final state, arguing that it is an equilibrium state. All these results are consistent with, but do not prove, that the acid-reassembled population includes an appreciable fraction (2/3) of antiparallel coiled-coil dimers.

Strategies and rationales for the de novo design of a helical hairpin peptide

The de novo design of alpha t alpha, a helical hairpin peptide, is described, alpha t alpha (alpha-helix/turn/alpha-helix) was developed to provide a model system for protein folding at the level of secondary structure association and stabilization. According to the prevailing models of protein folding, the second step in the folding process is the association and stabilization of secondary structural elements or microdomains. A brief description of the design, along with CD and NMR evidence confirming the conformation of the peptide in solution, has been published (Fezoui Y, Weaver DL, Osterhout JJ, 1994, Proc Natl Acad Sci USA 91:3675-3679). The present work includes a full description of the design process, including the trade-offs that were made during the development of the peptide, a discussion of recent experimental results that were not available at the time of the original design, indications of areas where, in retrospect, the design might have been done differently, and a discussion of how the present work fits into the field of de novo protein design.

Dissecting titin into its structural motifs: identification of an alpha-helix motif near the titin N-terminus

Titin, also known as connectin, is a giant modular protein specifically found in vertebrate striated muscle. Since the huge size of titin does not allow a direct structure determination, we have started a long-term project to characterize the protein by cutting it into smaller domains or structural units. The major part of the titin sequence is assembled by modules approximately 100 amino acids long that belong to two major protein superfamilies. Most of these modules are linked together by stretches of variable length with unique sequence. No direct structural characterization has been achieved so far for any of these linkers. We present here a study of a stretch located in the titin N-terminus and part of a linker between two modules. Our attention was drawn toward this region because it shows 100% probability to form a coiled coil when analyzed by a prediction program. A synthetic 38 amino acid peptide spanning such a sequence was studied in aqueous solution by circular dichroism, nuclear magnetic resonance, and analytical ultracentrifugation at various pH, salt, and peptide concentrations. Under all conditions, it shows a strong tendency to form alpha-helical structures. In the presence of salt, this conformation is associated with the formation of helical bundles below pH 5. Above pH 5, any aggregate breaks, and the titin peptide is a monomeric helix in equilibrium with its random coil conformation. We discuss the factors which stabilize the helical conformation and the possible role of this stretch in vivo.

Designed polyanionic coiled-coil proteins: acceleration of heparin cofactor II inhibition of thrombin

Novel polyanionic proteins were designed to increase the rate of heparin cofactor II (HC) inhibition of alpha-thrombin, an essential protease in the coagulation cascade. Two alpha-helical coiled-coil proteins, a 62-residue dimer containing 8 Glu residues (E8C) and a 104-residue dimer containing 14 Glu residues (E14C), plus two 31-residue control peptides containing 8 Glu residues each (E8A and E8B), were chemically synthesized, structurally characterized and enzymatically assayed. Circular dichroic spectrophotometry indicated that both E8C and E14C formed stable two-chain alpha-helical coiled coils at pH 7 and 25 degrees C. The control peptides were only partially alpha-helical. E14C remained folded at 90 degrees C but E8C was half unfolded at 49 degrees C. Coiled-coil proteins E8C and E14C maximally accelerated by 35- and 33-fold, respectively, the rate of HC inhibition of alpha-thrombin. None of these compounds accelerated antithrombin inhibition of alpha-thrombin, and neither control peptide accelerated HC inhibition of alpha-thrombin. Acceleration of the HC inhibition of alpha-thrombin showed bimodal dependence on the concentration of the polyanionic protein, which is consistent with formation of a HC-coiled-coil-thrombin ternary complex. The results suggest that antithrombotic polyanionic alpha-helical coiled-coil proteins can be designed and synthesized and that the occurrence of secondary structure can be correlated with biological activity.

Structural stability of short subsequences of the tropomyosin chain

The native tropomyosin molecule is a parallel, registered, alpha-helical coiled coil made from two 284-residue chains. Long excised subsequences (> or = 95 residues) form the same structure with comparable thermal stability. Here, we investigate local stability using shorter subsequences (20-50 residues) that are chemically synthesized or excised from various regions along the protein chain. Thermal unfolding studies of such shorter peptides by CD in the same solvent medium used in extant studies of the parent protein indicate very low helix content, almost no coiled-coil formation, and high thermal lability of such secondary structure as does form. This behavior is in stark contrast to extant data on leucine-zipper peptides and short "designed" synthetic peptides, many of which have high alpha-helix content and form highly stable coiled coils. The existence of short coiled coils calls into question the older idea that short subsequences of a protein have little structure. The present study supports the older view, at least in its application to tropomyosin. The intrinsic local alpha-helical propensity and helix-helix interaction in this prototypical alpha-helical protein is sufficiently weak as to require not only dimerization, but macro-molecular amplification in order to attain its native conformation in common benign media near neutral pH.

The alpha-5 segment of Bacillus thuringiensis delta-endotoxin: in vitro activity, ion channel formation and molecular modelling

A peptide with a sequence corresponding to the highly conserved alpha-5 segment of the Cry delta-endotoxin family (amino acids 193-215 of Bacillus thuringiensis CryIIIA [Gazit and Shai (1993) Biochemistry 32, 3429-3436]), was investigated with respect to its interaction with insect membranes, cytotoxicity in vitro towards Spodoptera frugiperda (Sf-9) cells, and its propensity to form ion channels in planar lipid membranes (PLMs). Selectively labelled analogues of alpha-5 at either the N-terminal amino acid or the epsilon-amine of its lysine, were used to monitor the interaction of the peptides with insect membranes. The fluorescent emission spectra of the 7-nitrobenz-2-oxa-1,3-diazole-4-yl (NBD)-labelled alpha-5 peptides displayed a blue shift upon binding to insect (Spodoptera littoralis) mid-gut membranes, reflecting the relocation of the fluorescent probes to an environment of increased apolarity, i.e. within the lipidic constituent of the membrane. Moreover, midgut membrane-bound NBD-labelled alpha-5 peptides were protected from enzymic proteolysis. Functional characterization of alpha-5 has revealed that it is cytotoxic to Sf-9 insect cells, and that it forms ion channels in PLMs with conductances ranging from 30 to 1000 pS. A proline-substituted analogue of alpha-5 is less cytolytic and slightly more exposed to enzymic digestion. Molecular modelling utilizing simulated annealing via molecular dynamics suggests that a transbilayer pore may be formed by alpha-5 monomers that assemble to form a left-handed coiled coil of approximately parallel helices. These findings further support a role for alpha-5 in the toxic mechanism of delta-endotoxins, and assign alpha-5 as one of the transmembrane helices which form the toxic pore. The suggested role is consistent with the recent finding that cleavage of CryIVB delta-endotoxin in a loop between alpha-5 and alpha-6 is highly important for its larvicidal activity [Angsuthanasombat, Crickmore and Ellar (1993) FEMS Microbiol. Lett. 111, 255-262].

Thermal unfolding equilibria in homodimeric chicken gizzard tropomyosin coiled coils

CD studies are presented on thermal unfolding of coiled-coil homodimers of two genetic variant chains of chicken gizzard tropomyosin (CG-Tm). The experiments include the effects of cross-linking both isoforms and the dependence on protein concentration of unfolding in both reduced isoforms, variables not examined in extant work. The general shapes of the unfolding curves for singly cross-linked species depend on whether the cross-link is at C190 (its site on one isoform) or at C36 (its site on the other). These curves are compared with extant ones for various cross-linked species of rabbit tropomyosin. The comparison supports the view that the unfolding behavior of cross-linked species results from a complex interaction of strain at the cross-link, local variations in structural stability, and loop entropy. The observed concentration dependence of the transition temperature for the uncross-linked (reduced) species of CG-Tm is very small (2.9 degrees C) for one variant homodimer and unobservably small (< 2 degrees C) for the other in the 100-fold concentration range (approximately 0.01-1.0 mg/mL) accessible here. These experimental values of delta Tm are much smaller than are predicted from extant values of the van't Hoff transition enthalpies, calling the latter into question.

Synaptic vesicle membrane fusion complex: action of clostridial neurotoxins on assembly

Clostridial neurotoxins inhibit neurotransmitter release by selective and specific intracellular proteolysis of synaptobrevin/VAMP, synaptosomal-associated protein of 25 kDa (SNAP-25) or syntaxin. Here we show that in binary reactions synaptobrevin binds weakly to both SNAP-25 and syntaxin, and SNAP-25 binds to syntaxin. In the presence of all three components, a dramatic increase in the interaction strengths occurs and a stable sodium dodecyl sulfate-resistant complex forms. Mapping of the interacting sequences reveals that complex formation correlates with the presence of predicted alpha-helical structures, suggesting that membrane fusion involves intermolecular interactions via coiled-coil structures. Most toxins only attack the free, and not the complexed, proteins, and proteolysis of the proteins by different clostridial neurotoxins has distinct inhibitory effects on the formation of synaptobrevin-syntaxin-SNAP-25 complexes. Our data suggest that synaptobrevin, syntaxin and SNAP-25 associate into a unique stable complex that functions in synaptic vesicle exocytosis.

An unusual intermediate filament subunit from the cytoskeletal biopolymer released extracellularly into seawater by the primitive hagfish (Eptatretus stouti)

Each slime gland thread cell from the primitive Pacific hagfish (Eptatretus stouti) contains a massive, conical, intermediate filament (IF)-rich biopolymer (‘thread,’ approximately 60 cm length, approximately 3 microns width). In view of the unusual ultrastructure of the thread, its extracellular role in modulation of the viscoelastic properties of mucus, and the ancient lineage of this primitive vertebrate, we report the nucleotide and deduced amino acid sequences of one major thread IF subunit, alpha (pI 7.5), which is coexpressed with a second polypeptide, gamma (pI 5.3). These two polypeptides coassemble in vitro into approximately 10 nm filaments. The alpha-thread chain, a 66.6 kDa polypeptide, has an unusual central rod domain containing 318 residues flanked by N- and C-terminal domains of 192 and 133 residues, respectively. Each peripheral region exhibits some epidermal keratin-like features including peptide repeats and a high total content of glycine and serine residues. The terminal domains, however, lack the H1 and H2 subdomains characteristic of known keratins. Moreover, when the central rod is aligned either in relation to established homology profiles (J. F. Conway and D. A. D. Parry (1988) Int. J. Biol. Macromol. 10, 79–98) of other IF subunits (types I-V, nestin, non-neuronal invertebrate), or by computer-based homology searches of the GenBank/EMBL Data Bank, a low identity (&lt; 30%) is evident, with no preferred identity to keratins or other known IF types. Although the central rod of 318 residues consists of the canonical apolar heptad repeats interspersed with three linker regions, a discontinuity in phasing of the heptad substructure in rod 2B, and conserved sequences at either end of the rod domain, other collective characteristics are atypical: overall high threonine content (13.2% vs 2.3-5.4% for other IFs), high threonine content in rod 1B (18.8% vs 1–6%), five Thr-Thr repeats in coiled coil segments, L12 of length greater than in keratins, substitution of phenylalanine for a highly conserved glutamate in the sixth position of L2, and a glycine-proline sequence in segment 2B. Possibly as a result of the high threonine content, the percentage of both acidic and basic residues in most helical subdomains is reduced relative to type I and II chains. Fast Fourier transform analyses show that only the acidic residues in segment 1B and basic residues in segment 2 have near typical IF periods.(ABSTRACT TRUNCATED AT 400 WORDS)

Parallel helix bundles and ion channels: molecular modeling via simulated annealing and restrained molecular dynamics

A parallel bundle of transmembrane (TM) alpha-helices surrounding a central pore is present in several classes of ion channel, including the nicotinic acetylcholine receptor (nAChR). We have modeled bundles of hydrophobic and of amphipathic helices using simulated annealing via restrained molecular dynamics. Bundles of Ala20 helices, with N = 4, 5, or 6 helices/bundle were generated. For all three N values the helices formed left-handed coiled coils, with pitches ranging from 160 A (N = 4) to 240 A (N = 6). Pore radius profiles revealed constrictions at residues 3, 6, 10, 13, and 17. A left-handed coiled coil and a similar pattern of pore constrictions were observed for N = 5 bundles of Leu20. In contrast, N = 5 bundles of Ile20 formed right-handed coiled coils, reflecting loosened packing of helices containing beta-branched side chains. Bundles formed by each of two classes of amphipathic helices were examined: (a) M2a, M2b, and M2c derived from sequences of M2 helices of nAChR; and (b) (LSSLLSL)3, a synthetic channel-forming peptide. Both classes of amphipathic helix formed left-handed coiled coils. For (LSSLLSL)3 the pitch of the coil increased as N increased from 4 to 6. The M2c N = 5 helix bundle is discussed in the context of possible models of the pore domain of nAChR.

Coiled-coil stutter and link segments in keratin and other intermediate filament molecules: a computer modeling study

Structural discontinuities have previously been identified in four regions of the coiled-coil rod domain structure present in intermediate filament (IF) protein molecules. These include a point at which a phase shift occurs in the heptad periodicity characteristic of the sequence of polar and apolar residues in alpha-helical coiled-coils, and three links that lack a heptad substructure. We have studied these regions by computer-based molecular modeling and comparative sequence analysis and conclude that the phasing discontinuity can be accommodated without significant distortion of the overall double-helical chain conformation; the L2 link has a similar conformation in all different types of IF molecules, a favorable conformation being one in which the two strands wrap tightly around each other; the L12 links vary in length between different IF types but contain important sequence similarities suggestive of a partial beta structure; the L1 links show larger variations in length, a lower degree of similarity, and probably diverse structures. Variations in the overall charges of the different links suggest that ionic interactions may play a significant role in filament assembly. The results also have general significance for other alpha-fibrous proteins in which either the characteristic heptad phasing undergoes a discontinuity or where a short non-coiled-coil sequence occurs within a coiled-coil rod domain structure.

Restored heptad pattern continuity does not alter the folding of a four-alpha-helix bundle

The sequences of alpha-helical coiled-coils and bundles are characterized by a specific pattern of hydrophobic and hydrophilic residues which is repeated every seven residues. Highly conserved breaks in this pattern frequently occur in segments of otherwise continuous heptad substructures. The hairpin bend of the ROP protein coincides with such a break and provides a model system for the study of the structural effects induced by heptad discontinuities. The structure of a ROP mutant which re-establishes a continuous heptad pattern, shows insignificant changes relative to the wild-type protein, as is also reflected in its conformational stability, spectroscopic properties and unfolding behaviour. Thus, formation of alpha-alpha-hairpin bends may occur both in the presence and absence of heptad breaks.

Non-globular domains in protein sequences: automated segmentation using complexity measures

Computational methods based on mathematically-defined measures of compositional complexity have been developed to distinguish globular and non-globular regions of protein sequences. Compact globular structures in protein molecules are shown to be determined by amino acid sequences of high informational complexity. Sequences of known crystal structure in the Brookhaven Protein Data Bank differ only slightly from randomly shuffled sequences in the distribution of statistical properties such as local compositional complexity. In contrast, in the much larger body of deduced sequences in the SWISS-PROT database, approximately one quarter of the residues occur in segments of non-randomly low complexity and approximately half of the entries contain at least one such segment. Sequences of proteins with known, physicochemically-defined non-globular regions have been analyzed, including collagens, different classes of coiled-coil proteins, elastins, histones, non-histone proteins, mucins, proteoglycan core proteins and proteins containing long single solvent-exposed alpha-helices. The SEG algorithm provides an effective general method for partitioning the globular and non-globular regions of these sequences fully automatically. This method is also facilitating the discovery of new classes of long, non-globular sequence segments, as illustrated by the example of the human CAN gene product involved in tumor induction.

NuMA/centrophilin: sequence analysis of the coiled-coil rod domain

Nuclear mitotic apparatus protein (NuMA), also known as centrophilin, has been shown in previous work to contain a centrally located sequence of length 1485 residues that has both a heptad substructure and a high propensity for alpha-helix formation. Further analysis of this sequence here has revealed that NuMA will form a two-stranded coiled-coil structure with multiple (18) points at which the conformation is interrupted either by proline-containing segments or by discontinuities in the phasing of the heptad substructure. It has also been shown that the two chains will be parallel (rather than antiparallel), that they will lie in axial register, and that this arrangement will be stabilized by a large number of interchain ionic interactions. Interestingly the coiled-coil rod domain is also shown to lack any significant long-range periodicity in the linear distribution of either its acidic or its basic residues. Hence there is no direct evidence from the sequence data that NuMA molecules will aggregate to form closely packed filaments within nuclear space.

Crystal structure of an isoleucine-zipper trimer

Subunit oligomerization in many proteins is mediated by short coiled-coil motifs. These motifs share a characteristic seven-amino-acid repeat containing hydrophobic residues at the first (a) and fourth (d) positions. Despite this common pattern, different sequences form two-, three- and four-stranded helical ropes. We have investigated the basis for oligomer choice by characterizing variants of the GCN4 leucine-zipper dimerization domain that adopt trimeric or tetrameric structures in response to mutations at the a and d positions. We now report the high-resolution X-ray crystal structure of an isoleucine-containing mutant that folds into a parallel three-stranded, alpha-helical coiled coil. In contrast to the dimer and tetramer structures, the interior packing of the trimer can accommodate beta-branched residues in the most preferred rotamer at both hydrophobic positions. Compatibility of the shape of the core amino acids with the distinct packing spaces in the two-, three- and four-stranded conformations appears to determine the oligomerization state of the GCN4 leucine-zipper variants.

Identification of the DNA-binding domains of the switch-activating-protein Sap1 from S.pombe by random point mutations screening in E.coli

Mating type switching in fission yeast, Schizosaccharomyces pombe, is initiated by a site-specific double-strand break (DSB) at the mat1 locus. The DSB is controlled from a distance by cis- and trans-acting elements. The switch-activating protein, Sap1 binds to the SAS1 cis-acting element which controls the frequency of the DSB at the mat1 locus and, consequently the efficiency of mating type switching. We developed a general method for screening randomly mutagenized expression libraries of DNA-binding protein in E.coli. Sap1 gene was mutagenized by PCR under conditions of reduced Taq polymerase fidelity. The mutated DNA was expressed in E.coli and screened for SAS1-recognition. This method was used to isolated 16 point mutations that abolished SAS1 interaction together with 18 mutations that did not affect binding. The position of these point mutations allowed the identification of three protein domains located in the N-terminal part of Sap1 that are essential for DNA-binding. Deletions and biochemical analysis showed that Sap1 is a dimer both in solution and when bound to SAS1 sequence. The dimerization domain was localized C-terminally to the three domains described above and when used in exess it inhibited DNA binding.

Molecular and genetic characterization of GABP beta

This report outlines three observations relating to GABP beta, a polypeptide constituent of the heterotetrameric transcription factor GABP. Evidence is presented showing that the mouse genome encodes two highly related GABP beta polypeptides, designated GABP beta 1-1 and GABP beta 2-1. Genomic and cDNA copies of the newly defined Gabpb2 gene were cloned and characterized, providing the conceptually translated amino acid sequence of GABP beta 2-1. The genes encoding these two proteins, as well as GABP alpha, were mapped to three unlinked chromosomal loci. Although physically unlinked, the patterns of expression of the three genes were strikingly concordant. Finally, the molecular basis of GABP beta dimerization was resolved. Carboxy-terminal regions of the two GABP beta polypeptides, which mediate dimerization, bear highly related primary amino acid sequences. Both sequences are free of alpha-helix destabilizing residues and, when displayed on idealized alpha-helical projections, reveal marked amphipathy. Two observations indicate that these regions adopt an alpha-helical conformation and intertwine as coiled-coils. First, the dimer-forming region of GABP beta 2-1 can functionally replace the leucine zipper of a bZIP transcription factor. Second, a synthetic peptide corresponding to this region shows distinctive helical properties when examined by circular dichroism spectroscopy. Finally, evidence is presented showing that GABP beta 1-1 and GABP beta 2-1 can heterodimerize through this carboxy-terminal domain, but neither protein can heterodimerize via the dimer-forming region of the bZIP protein C/EBP beta.

A thermodynamic model for the helix-coil transition coupled to dimerization of short coiled-coil peptides

A simple thermodynamic formalism is presented to model the conformational transition between a random-coil monomeric peptide and a coiled-coil helical dimer. The coiled-coil helical dimer is the structure of a class of proteins also called leucine zipper, which has been studied intensively in recent years. Our model, which is appropriate particularly for short peptides, is an alternative to the theory developed by Skolnick and Holtzer. Using the present formalism, we discuss the multi-equilibriatory nature of this transition and provide an explanation for the apparent two-state behavior of coiled-coil formation when the helix-coil transition is coupled to dimerization. It is found that such coupling between multi-equilibria and a true two-state transition can simplify the data analysis, but care must be taken in using the overall association constant to determine helix propensities (w) of single residues. Successful use of the two-state model does not imply that the helix-coil transition is all-or-none. The all-or-none assumption can provide good numerical estimates when w is around unity (0.35 < or = w < or = 1.35), but when w is small (w < 0.01), similar estimations can lead to large errors. The theory of the helix-coil transition in denaturation experiments is also discussed.

Specificity and promiscuity in membrane helix interactions

Transmembrane alpha-helices can associate with one another in lipid bilayers. This association is important in the folding and oligomerization of many integral membrane proteins, and may also play a role in their function. The interactions between helices may be highly specific or relatively non-specific, and their roles may differ accordingly. These two cases are discussed.

Nuf2, a spindle pole body-associated protein required for nuclear division in yeast

The NUF2 gene of the yeast Saccharomyces cerevisiae encodes an essential 53-kd protein with a high content of potential coiled-coil structure similar to myosin. Nuf2 is associated with the spindle pole body (SPB) as determined by coimmunofluorescence with known SPB proteins. Nuf2 appears to be localized to the intranuclear region and is a candidate for a protein involved in SPB separation. The nuclear association of Nuf2 can be disrupted, in part, by 1 M salt but not by the detergent Triton X-100. All Nuf2 can be removed from nuclei by 8 M urea extraction. In this regard, Nuf2 is similar to other SPB-associated proteins including Nufl/SPC110, also a coiled-coil protein. Temperature-sensitive alleles of NUF2 were generated within the coiled-coil region of Nuf2 and such NUF2 mutant cells rapidly arrest after temperature shift with a single undivided or partially divided nucleus in the bud neck, a shortened mitotic spindle and their DNA fully replicated. In sum, Nuf2 is a protein associated with the SPB that is critical for nuclear division. Anti-Nuf2 antibodies also recognize a mammalian 73-kd protein and display centrosome staining of mammalian tissue culture cells suggesting the presence of a protein with similar function.

Specificity and promiscuity in membrane helix interactions

The membrane-spanning portions of many integral membrane proteins consist of one or a number of transmembrane α-helices, which are expected to be independently stable on thermodynamic grounds. Side-by-side interactions between these transmembrane α-helices are important in the folding and assembly of such integral membrane proteins and their complexes. In considering the contribution of these helix–helix interactions to membrane protein folding and oligomerization, a distinction between the energetics and specificity should be recognized. A number of contributions to the energetics of transmembrane helix association within the lipid bilayer will be relatively non-specific, including those resulting from charge–charge interactions and lipid–packing effects. Specificity (and part of the energy) in transmembrane α-helix association, however, appears to rely mainly upon a detailed stereochemical fit between sets of dynamically accessible states of particular helices. In some cases, these interactions are mediated in part by prosthetic groups.