Membrane pores: from structure and assembly, to medicine and technology

Pathogens have evolved weapons to invade and damage our cells, and our immune system has evolved defences against these attacks. Among the weaponry used by both sides in this continual war are proteins that punch holes in cell membranes. Membrane perforation enables pathogens to take over host cells and resources for their own replication, and also enables host immune systems to kill invading pathogens. The membrane attack complex-perforin (MACPF)/ cholesterol dependent cytolysin (CDC) superfamily of membrane pore-forming proteins is used by a wide range of pathogens as well as by host immune systems. This talk focuses on the mechanisms by which MACPF and CDC proteins convert from their soluble, monomeric forms into large arcs and rings that insert into membranes and perforate them. Studies of pore assembly on liposomes in vitro and examples of their action in vivo will be presented.

The a-pore-forming toxin Cytolysin A (ClyA) is responsible for the hemolytic phenotype of several Escherichia coli and Salmonella enterica strains and is cytotoxic towards cultured mammalian cells. ClyA is a soluble, monomeric 34 kDa protein that spontaneously forms ring-shaped, homododecameric pore complexes upon contact with target membranes or detergent. The pore complex forms a hollow cylinder with an inner diameter of 70 Å that is reduced to 30 Å in the membrane-inserted part of the complex. The structural comparison between the soluble momoner and the protomer in the pore complex revealed one of the largest conformational transitions observed in a protein so far, in which more than half of the ClyA residues are reorganized and 16% of the residues are located in different secondary structure elements after protomer formation. Analysis of detergent-induced ClyA assembly in vitro showed that the rate-limiting step in pore formation at ClyA concentrations above 100 nM is the unimolecular transition of the soluble monomer to the assembly-competent protomer, which is followed by rapid protomer assembly. Single-molecule Förster resonance energy (FRET) experiments at picomolar  ClyA concentrations preventing pore assembly showed that a molten globule-like off-pathway intermediate is formed during the monomer-to-protomer transition. The global kinetic analysis of detergent-induced pore formation from picomolar to micromolar ClyA concentrations supports a model in which all sterically compatible, linear oligomeric ClyA intermediates participate in pore assembly.

This talk will describe how we have developed an understanding of natural coiled-coil sequence motifs found in proteins, which lead to specific protein oligomerization; and then how we can apply this knowledge in rational peptide and protein design.  For the latter, it is now clear that nature has almost certainly only shown us a glimpse of what is possible structurally and functionally with polypeptide chains.  To help explore past the confines of natural coiled-coil structures, and into what has been termed the dark matter of protein space, a toolkit of de novo peptides has been developed. These can be used as building blocks for the rapid construction of new protein structures and assemblies. The talk will demonstrate the utility of this approach to make water-soluble protein-like barrels and pores, which can be engineered to bind biological molecules and catalyse simple reactions.  More recently membrane-spanning variants of these α-helical barrels have been engineered, and in collaboration with the Bayley lab, it has shown that these insert into lipid bilayers and conduct ions in voltage-dependent manners.  The two systems have been explored.  In both cases, the insertion and orientation of peptides into the membranes can be directed, and the channel activities can be controlled through mutagenesis and blocking experiments.

The mechanism of proton transport through membrane proteins is of general interest to multiple areas of biology.  Using a variety of spectroscopic, crystallographic, and computational methods, the mechanism by which protons are conducted through the M2 proton channel from influenza A virus was investigated, and this information was used to design new anti-influenza drugs that target highly drug-resistant forms of the virus.  A second topic of the talk will focus on the use of de novo protein design to test the mechanism by which a class of transporters uses proton gradients to drive the conduction of molecules into or out of cells. Transporters have been hypothesized to arise by physical association or gene duplication of primordial units, leading to an assembly with “frustrated symmetry” that rocks between two states with the substrate-binding site alternately accessing each side of the membrane. Rocker, a minimalist Zn2+/proton antiporter was designed to test these principles, although it bears no sequence similarity to any known natural protein. Structural, dynamic and functional studies indicate that Rocker is a primordial transporter, which recapitulates many of the properties of this class of proteins.  These studies also demonstrate the feasibility of using de novo design to design membrane proteins with complex dynamic and functional properties.

The membrane attack complex (MAC) is a fundamental component of immune defence that drills holes in bacterial membranes and kills pathogens. MAC lesions were first identified in 1964, yet half a century later details of its structure and assembly mechanism remain undiscovered. Here electron cryo-microscopy is used to visualize the human pore complex at subnanometer resolution. The protein composition of the MAC is determined and interaction interfaces that hold the assembly together are identified. Unlike closely related pore-forming proteins, the MAC’s asymmetric pore and "split-washer" shape suggest a killing mechanism that involves not only membrane rupture, but also distortion.

Pore-forming toxins (PFT) constitute a fascinating group of proteins belonging to the molecular offensive and defensive machinery of virtually all kingdoms of life. This class of water-soluble proteins shares the remarkable ability to metamorphose in the presence of cell membranes, generating lytic pores and causing cell-damage. Actinoporins are a family of potent hemolytic toxins from sea anemone forming alpha-helical pores on cellular and model membranes.

In general, two requirements are sufficient to trigger pore-formation by actinoporins:

(i) the presence of the lipid sphingomyelin, and (ii) the segregation of the membrane on domains or lipid-rafts. Until recently, the molecular basis of pore-formation by actinoporins, and specially the specific requirement for sphingomyelin were unclear.

However, a number of recent studies have shed light into critical steps of their mechanism of action, such as binding of the toxins to the membrane, self-assembly via protein-protein interactions, and assembly of the transmembrane pore.

Collectively, the data suggests that sphingomyelin facilitates pore-formation at the binding and assembly stages, and reveal the first example of a hybrid lipid/protein pore by a PFT. The structural and thermodynamic basis of this novel architecture will be explained in detail during this presentation. Surprisingly, the entire process can be made reversible under mild experimental conditions by the careful selection of detergents, challenging current perceptions in the field of membrane-protein interactions.

Bax is a key player in apoptosis that mediates of the permeabilization of the outer mitochondrial membrane. Despite intense research, the underlying process remains poorly understood. By combining biophysical approaches at different scales, new insight into the molecular mechanism of Bax is provided. Electron paramagnetic resonance data shows a key conformational change in the central hairpin of Bax that is involved in pore formation. In this configuration, Bax is present as a mixture of oligomers based on dimer units, as revealed by single molecule imaging. Moreover, the nanoscale organization of Bax at mitochondria of apoptotic cells is provided by superresolution microscopy. Based on this, a new model for the molecular mechanism of Bax is proposed.

Two members of the Bcl-2 family, Bak and Bax, drive apoptotic cell death by changing conformation and forming oligomers that permeabilise the mitochondrial outer membrane. The two proteins are activated by BH3-only family members binding to the α2-α5 hydrophobic surface groove. Newly exposed hydrophobic regions then either “collapse” onto the membrane surface to lie in-plane, or interact to generate BH3:groove symmetric dimers. We found that in each Bak dimer the N-termini are fully solvent-exposed and mobile, allowing disulphide bonding between certain residues (e.g. V61C:V61C') to specifically interrogate how dimers associate into high order complexes. These data informed mathematical simulations that support a model in which Bak dimers interact in a random manner to form compact clusters that generate lipidic pores.       

It was also found that antibodies can trigger activation of Bak and mitochondrial Bax, and do so by binding to a new activation site, the α1-α2 loop. The mechanism of antibody-mediated Bak activation involves α1 dissociation, revealed by biochemical studies and a structural model of Fab bound to Bak. Intriguingly, antibodies to the α1-α2 loop in cytosolic Bax could block its translocation to mitochondria. These data thus identify the α1-α2 loop as a new target for regulating Bak and Bax apoptotic function.

The mechanism by which amphipathic peptides permeabilize biological membranes is not well understood. Do the peptides form well-defined pores or simply dissolve the membranes in a detergent-like fashion? What is the lifetime of these pores? How do the sequence and structure of these peptides determine their permeabilizing function? Both experiment and theory face formidable challenges in obtaining detailed information on such labile structures.  We have used two theoretical approaches to study peptide-induced pore formation in lipid bilayers. The first treats water and lipids implicitly and the peptide in atomistic detail. This simplified representation allows one to obtain useful insights into protein-membrane interactions. Using this approach we showed that antimicrobial peptides bind more strongly to membrane pores than to the flat membrane, consistent with the idea that they stabilize them. The second approach is fully atomistic molecular dynamics simulations, some on the 10-microsecond timescale, starting from inserted peptide aggregates. Such simulations of melittin, magainin, PGLa, alamethicin, and protegrin have revealed interesting differences between these peptides and possible explanations of the observed synergy between some of them.