Structural plasticity of helical nanotubes based on coiled-coil assemblies.

Numerous instances can be seen in evolution in which protein quaternary structures have diverged while the sequences of the building blocks have remained fairly conserved. However, the path through which such divergence has taken place is usually not known. We have designed two synthetic 29-residue α-helical peptides, based on the coiled-coil structural motif, that spontaneously self-assemble into helical nanotubes in vitro. Using electron cryomicroscopy with a newly available direct electron detection capability, we can achieve near-atomic resolution of these thin structures. We show how conservative changes of only one or two amino acids result in dramatic changes in quaternary structure, in which the assemblies can be switched between two very different forms. This system provides a framework for understanding how small sequence changes in evolution can translate into very large changes in supramolecular structure, a phenomenon that may have significant implications for the de novo design of synthetic peptide assemblies.

Cofilin cross-bridges adjacent actin protomers and replaces part of the longitudinal F-actin interface.

ADF/cofilins are abundant actin binding proteins critical to the survival of eukaryotic cells. Most ADF/cofilins bind both G and F-actin, sever the filaments and accelerate their treadmilling. These effects are linked to rearrangements of interprotomer contacts, changes in the mean twist, and filament destabilization by ADF/cofilin. Paradoxically, it was reported that under certain in vitro and in vivo conditions cofilin may stabilize actin filaments and nucleate their formation. Here, we show that yeast cofilin and human muscle cofilin (cofilin-2) accelerate the nucleation and elongation of ADP-F-actin and stabilize such filaments. Moreover, cofilin rescues the polymerization of the assembly incompetent tethramethyl rhodamine (TMR)-actin and T203C/C374S yeast mutant actin. Filaments of cofilin-decorated TMR-actin and unlabeled actin are indistinguishable, as revealed by electron microscopy and three-dimensional reconstruction. Our data suggest that ADF/cofilins play an active role in establishing new interprotomer interfaces in F-actin that substitute for disrupted (as in TMR-actin and mutant actin) or weakened (as in ADP-actin) longitudinal contacts in filaments.

What is the structure of the RecA-DNA filament?

The bacterial RecA protein has been a model system for understanding how a protein can catalyze homologous genetic recombination. RecA-like proteins have now been characterized from many organisms, from bacteriophage to humans. Some of the RecA-like proteins, including human RAD51, appear to function as helical filaments formed on DNA. However, we currently have high resolution structures of inactive forms of the protein, and low resolution structures of the active complexes formed by RecA-like proteins on DNA in the presence of ATP or ATP analogs. Within a crystal of the E. coli RecA protein, a helical polymer exists, and it has been widely assumed that this polymer is quite similar to the active helical filament formed on DNA. Recent developments have suggested that this may not be the case.

Molecular evolution: actin's long lost relative found.

The bacterial protein MreB has been identified as a prokaryotic homolog of the eukaryotic cytoskeletal protein actin. While we still know little about MreB's function, the structural similarities and differences between MreB and actin provide more insight into the remarkable properties of actin.

Archaeal RadA protein binds DNA as both helical filaments and octameric rings.

The Escherichia coli RecA protein has been a model for understanding homologous eukaryotic recombination proteins such as Rad51. The active form of both RecA and Rad51 appear to be helical filaments polymerized on DNA, in which an unusual helical structure is induced in the DNA. Surprisingly, the human meiosis-specific homolog of RecA, Dmc1, has thus far only been observed to bind DNA as an octameric ring. Sequence analysis and biochemical studies have shown that archaeal RadA proteins are more closely related to Rad51 and Dmc1 than the bacterial RecA proteins. We find that the Sulfolobus solfataricus RadA protein binds DNA in the absence of nucleotide cofactor as an octameric ring and in the presence of ATP as a helical filament. Since it is likely that RadA is closely related to a common ancestral protein of both Rad51 and Dmc1, the two DNA-binding forms of RadA may provide insight into the divergence that has taken place between Rad51 and Dmc1.

Comparison of bacteriophage T4 UvsX and human Rad51 filaments suggests that RecA-like polymers may have evolved independently.

The UvsX protein from bacteriophage T4 is a member of the RecA/Rad51/RadA family of recombinases active in homologous genetic recombination. Like RecA, Rad51 and RadA, UvsX forms helical filaments on DNA. We have used electron microscopy and a novel method for image analysis of helical filaments to show that UvsX-DNA filaments exist in two different conformations: an ADP state and an ATP state. As with RecA protein, these two states have a large difference in pitch. Remarkably, even though UvsX is only weakly homologous to RecA, both UvsX filament states are more similar to the RecA crystal structure than are RecA-DNA filaments. We use this similarity to fit the RecA crystal structure into the UvsX filament, and show that two of the three previously described blocks of similarity between UvsX and RecA are involved in the subunit-subunit interface in both the UvsX filament and the RecA crystal filament. Conversely, we show that human Rad51-DNA filaments have a different subunit-subunit interface than is present in the RecA crystal, and this interface involves two blocks of sequence similarity between Rad51 and RecA that do not overlap with those found between UvsX and RecA. This suggests that helical filaments in the RecA/Rad51/RadA family may have arisen from convergent evolution, with a conserved core structure that has assembled into multimeric filaments in a number of different ways.

The primase active site is on the outside of the hexameric bacteriophage T7 gene 4 helicase-primase ring.

Gene 4 of bacteriophage T7 encodes a protein (gp4) that can translocate along single-stranded DNA, couple the unwinding of duplex DNA with the hydrolysis of dTTP, and catalyze the synthesis of short RNA oligoribonucleotides for use as primers by T7 DNA polymerase. Electron microscopic studies have shown that gp4 forms hexameric rings, and X-ray crystal structures of the gp4 helicase domain and of the highly homologous RNA polymerase domain of Escherichia coli DnaG have been determined. Earlier biochemical studies have shown that when single-stranded DNA is bound to the hexameric ring, the primase domain remains accessible to free DNA. Given these results, a model was suggested in which the primase active site in the gp4 hexamer is located on the outside of the hexameric ring. We have used electron microscopy and single-particle image analysis to examine T7 gp4, and have determined that the primase active site is located on the outside of the hexameric ring, and therefore provide direct structural support for this model.

Probing the structure of F-actin: cross-links constrain atomic models and modify actin dynamics.

Cross-links between protomers in F-actin can be used as a very sensitive probe of both the dynamics and structure of F-actin. We have characterized filaments formed from a previously described yeast actin Q41C mutant, where disulfide bonds can be formed between the Cys41 that is introduced into subdomain-2 and Cys374 on an adjacent protomer. We find that the distribution of cross-linked n-mers shows no cooperativity and corresponds to a random probability cross-linking reaction. The random distribution suggests that disulfide formation does not cause a significant perturbation of the F-actin structure. Consistent with this lack of perturbation, three-dimensional reconstructions of extensively cross-linked filaments, using a new approach to helical image analysis, show very small structural changes with respect to uncross-linked filaments. This finding is in conflict with refined models but in agreement with the original Holmes et al. model for F-actin. Under conditions where 94 % of the protomers are linked by disulfide bonds, the distribution of filament twist becomes more heterogeneous with respect to control filaments. A molecular model suggests that strain, introduced by the disulfide, is relieved by increasing the twist of the long-pitch actin helices. Disulfide formation makes yeast actin filaments approximately three times less flexible in terms of bending and similar, in this respect, to vertebrate skeletal muscle F-actin. These observations support previous reports that the rigidity of F-actin can be controlled by the position of subdomain-2, and that this region is more flexible in yeast F-actin than in skeletal muscle F-actin.

Electron microscopic studies of the translin octameric ring.

Translin is thought to participate in a variety of cellular activities including chromosomal translocations, translational regulation of mRNA expression, and mRNA transport. It forms an octameric ring structure capable of sequence-specific binding of both DNA and RNA substrates. We have used electron microscopy and single-particle image analysis to generate a three-dimensional reconstruction of the Translin ring. The subunits appear to have two distinct domains that assemble to form an open channel with diameter of approximately 30 A at one end and approximately 50 A at the opposite end. In the presence of either DNA or RNA containing consensus binding sequences, the largest opening into the central cavity is filled with density. Strikingly, although Translin shows significant sequence homology to only one other protein, Translin-associated factor X, the quaternary organization and the dimerization of subunits in the ring are very similar to those observed for hexameric ring helicases. This suggests that many of the structures in DNA and RNA metabolism may have similar quaternary organization.

Domain structure and dynamics in the helical filaments formed by RecA and Rad51 on DNA.

Both the bacterial RecA protein and the eukaryotic Rad51 protein form helical nucleoprotein filaments on DNA that catalyze strand transfer between two homologous DNA molecules. However, only the ATP-binding cores of these proteins have been conserved, and this same core is also found within helicases and the F1-ATPase. The C-terminal domain of the RecA protein forms lobes within the helical RecA filament. However, the Rad51 proteins do not have the C-terminal domain found in RecA, but have an N-terminal extension that is absent in the RecA protein. Both the RecA C-terminal domain and the Rad51 N-terminal domain bind DNA. We have used electron microscopy to show that the lobes of the yeast and human Rad51 filaments appear to be formed by N-terminal domains. These lobes are conformationally flexible in both RecA and Rad51. Within RecA filaments, the change between the "active" and "inactive" states appears to mainly involve a large movement of the C-terminal lobe. The N-terminal domain of Rad51 and the C-terminal domain of RecA may have arisen from convergent evolution to play similar roles in the filaments.

Does a stretched DNA structure dictate the helical geometry of RecA-like filaments?

Proteins in the RecA/Rad51/RadA/UvsX family form helical filaments on DNA in which the DNA is stretched and untwisted. A comparison of the average helical parameters of these filaments from five different proteins, obtained from archaea, eubacteria and eukaryotes, suggests that an intrinsic state of DNA may be responsible for the conservation of these particular filament forms across evolution. In this view, these proteins stabilize this existing state of DNA, rather than induce a novel conformation.

Actin depolymerizing factor stabilizes an existing state of F-actin and can change the tilt of F-actin subunits.

Proteins in the actin depolymerizing factor (ADF)/cofilin family are essential for rapid F-actin turnover, and most depolymerize actin in a pH-dependent manner. Complexes of human and plant ADF with F-actin at different pH were examined using electron microscopy and a novel method of image analysis for helical filaments. Although ADF changes the mean twist of actin, we show that it does this by stabilizing a preexisting F-actin angular conformation. In addition, ADF induces a large ( approximately 12 degrees ) tilt of actin subunits at high pH where filaments are readily disrupted. A second ADF molecule binds to a site on the opposite side of F-actin from that of the previously described ADF binding site, and this second site is only largely occupied at high pH. All of these states display a high degree of cooperativity that appears to be an integral part of F-actin.

Bacterial conjugation: running rings around DNA.

Helicases are active in many aspects of DNA replication, recombination, repair and transcription. An integral membrane bacterial protein assembly involved in the transfer of DNA between cells has been shown to resemble a ring helicase, suggesting that it hydrolyzes ATP to pump DNA through a central channel.

Binding of dystrophin's tandem calponin homology domain to F-actin is modulated by actin's structure.

Dystrophin has been shown to be associated in cells with actin bundles. Dys-246, an N-terminal recombinant protein encoding the first 246 residues of dystrophin, includes two calponin-homology (CH) domains, and is similar to a large class of F-actin cross-linking proteins including alpha-actinin, fimbrin, and spectrin. It has been shown that expression or microinjection of amino-terminal fragments of dystrophin or the closely related utrophin resulted in the localization of these protein domains to actin bundles. However, in vitro studies have failed to detect any bundling of actin by either intact dystrophin or Dys-246. We show here that the structure of F-actin can be modulated so that there are two modes of Dys-246 binding, from bundling actin filaments to only binding to single filaments. The changes in F-actin structure that allow Dys-246 to bundle filaments are induced by covalent modification of Cys-374, proteolytic cleavage of F-actin's C-terminus, mutation of yeast actin's N-terminus, and different buffers. The present results suggest that F-actin's structural state can have a large influence on the nature of actin's interaction with other proteins, and these different states need to be considered when conducting in vitro assays.

Two conformations of G-actin related to two conformations of F-actin.

In summary, a number of different conformational states of F-actin have been described by several different laboratories. Crystal structures have revealed that an opening of the nucleotide-binding cleft, produced by a large rotation of subdomain 2, can occur in G-actin. We have shown that two crystal states of beta-actin, in an open and closed form, can provide a very good model for the conformational difference in F-actin between yeast the wild-type and a V159N mutant. This suggests that some of the dynamics associated with G-actin may provide insights into dynamic processes within the F-actin filament.

Three-dimensional reconstruction of transcription termination factor rho: orientation of the N-terminal domain and visualization of an RNA-binding site.

The Escherichia coli rho transcription termination protein is a hexameric helicase, and is believed to function by separating an RNA-DNA hybrid. Unlike hexameric DNA helicases, where a single strand of DNA passes through the central channel, it has been proposed that the RNA wraps around the outside of the ring. We have generated a three-dimensional reconstruction of rho, and localized a tRNA molecule bound to the primary RNA-binding site to the outside of the ring. An atomic structure of the N-terminal domain of rho fits into our reconstruction uniquely, with the residues involved in RNA-binding on the outside of the ring. Although rho shares a common structural core with the F1-ATPase and other hexameric helicases, there has been a divergence in function due to rho's N-terminal domain, which has no homology to other helicases.

The human Rad52 protein exists as a heptameric ring.

The RAD52 epistasis group was identified in yeast as a group of genes required to repair DNA damaged by ionizing radiation [1]. Genetic evidence indicates that Rad52 functions in Rad51-dependent and Rad51-independent recombination pathways [2] [3] [4]. Consistent with this, purified yeast and human Rad52 proteins have been shown to promote single-strand DNA annealing [5] [6] [7] and to stimulate Rad51-mediated homologous pairing [8] [9] [10] [11]. Electron microscopic examinations of the yeast [12] and human [13] Rad52 proteins have revealed their assembly into ring-like structures in vitro. Using both conventional transmission electron microscopy and scanning transmission electron microscopy (STEM), we found that the human Rad52 protein forms heptameric rings. A three-dimensional (3D) reconstruction revealed that the heptamer has a large central channel. Like the hexameric helicases such as Escherichia coli DnaB [14] [15], bacteriophage T7 gp4b [16] [17], simian virus 40 (SV40) large T antigen [18] and papilloma virus E1 [19], the Rad52 rings show a distinctly chiral arrangement of subunits. Thus, the structures formed by the hexameric helicases may be a more general property of other proteins involved in DNA metabolism, including those, such as Rad52, that do not bind and hydrolyze ATP.

F-actin retains a memory of angular order.

Modifications can be made to F-actin that do not interfere with the binding of myosin but inhibit force generation, suggesting that actin's internal dynamics are important for muscle contraction. Observations from electron microscopy and x-ray diffraction have shown that subunits in F-actin have a relatively fixed axial rise but a variable twist. One possible explanation for this is that the actin subunits randomly exist in different discrete states of "twist, " with a significant energy barrier separating these states. This would result in very slow torsional transitions. Paracrystals impose increased order on F-actin filaments by reducing the variability in twist. By looking at filaments that have recently been dissociated from paracrystals, we find that F-actin retains a "memory" of its previous environment that persists for many seconds. This would be consistent with slow torsional transitions between discrete states of twist.