Protein Machines in the Molecular Arms Race

Birkbeck’s Science Week 2015 was held from Monday 23 to Thursday 26 March and included three evenings of public talks by senior researchers. The first two lectures, on the Tuesday, were given by two of the college’s most distinguished women scientists: Helen Saibil from the Department of Biological Sciences and Karen Hudson-Edwards from Earth and Planetary Sciences; they were billed together as a ‘Women in Science Evening’.

The lectures were all introduced by the Dean of the Faculty of Science, Nicholas Keep who described Saibil, a close colleague, as “our most eminent female scientist”. She came to Birkbeck from Canada via a PhD at King’s College London under the supervision of Nobel laureate Maurice Wilkins and post-doctoral work at Oxford.

Since arriving here in the 1980s she has built up an internationally renowned structural biology lab, focusing in particular on the technique of electron microscopy. She has been a Fellow of the Royal Society since 2006 and of the Academy of Medical Sciences since 2009.

Saibil began her lecture by explaining that proteins can act as little machines, performing mechanical tasks that are essential for the maintenance of life. Her group has been interested for some time in proteins that can punch holes in the walls of cells. This allows the cell contents to leak out in a damaging process known as lysis, and it also allows toxins to enter the cells. These proteins can therefore be thought of as powerful weapons, and they are deployed on both sides of a ‘molecular arms race’: by pathogens and by the immune systems of humans and other animals.

Most soluble proteins fold into a single stable structure that tries, as far as possible, to keep their hydrophobic (“water-hating”) parts – the side chains of certain amino acids – in the interior of the protein, with the hydrophilic (“water-loving”) side chains on the outside, in contact with the watery environment inside or outside cells.

Pore-forming proteins, however, have a ‘Jekyll and Hyde’ like identity: they can form two distinctly different shapes, one as individual, soluble molecules and the other when they associate with each other into membrane-bound rings to form cylindrical pores. These structures, and the conformational change between them, are remarkably similar in proteins from bacteria and from the immune system.

Pore-forming toxins have been found in types of bacteria that are responsible for some deadly human diseases, including meningitis and pneumonia. The structure of a monomeric form of one of these proteins in solution was first solved in 1998, using X-ray crystallography. However, large complexes of many protein molecules are more readily solved by electron microscopy, particularly when those complexes are embedded in membranes.

In 2005 Saibil and her group described structures of the pore-forming toxin pneumolysin from Streptococcus pneumoniae, in complex with a model cell membrane. They found that the proteins formed two distinctly different ring-shaped structures. Initially, they formed into a ring sitting on top of the membrane, which was termed the pre-pore; then they changed shape to burrow part of each protein deep into the membrane and form the pore itself. Each monomer in the pre-pore had a structure that was similar to that of the molecule in solution, but they underwent large structural changes to form the pore.

Schematic illustration of how suilysin, a bacterial cholesterol-dependent cytolysin, drills holes in cell membranes. Image © Adrian Hodel, London Centre for Nanotechnology

Most structures solved by electron microscopy are at lower resolution than those solved by X-ray crystallography, and it is not possible to trace the positions of individual atoms at lower resolutions (e.g. worse than 3 A). Saibil and her colleagues were able to interpret the structure of the proteins making up the pore by fitting pieces of the X-ray structure of the isolated molecule into their electron density.

They found a dramatic change in structure, with the tall, thin protein structure collapsing into an arch and a helical region stretching out to form a long, extended beta hairpin. It is these hairpins that join together to form the walls of the pore. The process of pore forming therefore has three stages: firstly the toxin molecules bind to the surface of their target cells, then they associate into the circular pre-pores and finally they change shape in a concerted manner, punching holes in the cell membranes by ejecting a disc of membrane, letting other toxins in and cell contents out.

Saibil then turned the focus of her talk from attack by bacteria to the human immune system’s defence. Natural killer (NK) cells are specialised lymphocytes (white blood cells) that kill virally-infected and cancerous cells in the bloodstream. They kill on contact with their target cells by releasing a toxic protein into those cells that stimulates those target cells to commit suicide in a process known as programmed cell death or apoptosis. We have only recently learned that the mechanism through which the NK cells work is very similar to the mechanism of the bacterial pore-forming toxins.

Natural killer cells express a protein called perforin that has a similar structure in solution to the bacterial pneumolysin. Although there is very little sequence similarity between these proteins – there is only one amino acid conserved throughout all the known bacterial and vertebrate proteins of this family, a glycine at a critical position for the conformational change – the structures are similar enough to suggest that the proteins all once had a common ancestor.

Saibil and her colleagues used electron microscopy to discover that this protein forms a pore through a similar mechanism to pneumolysin: the helical region that unfolds into the beta hairpin to form the pore forms the core of the molecular machine and is largely unchanged between the structures. There are some differences between the structures, however; in particular, there is no need for the perforin structure to ‘collapse’ as the molecule has ‘arms’ that are long enough to form the hairpin and punch the hole without bending into an arch.

The mechanism through which the NK cells kill their target cells is now quite well understood. When the two cells come into contact they form a temporary structure called an immune synapse that allows the pore to form and proteases called granzymes, which induce apoptosis, to enter the target cells. This YouTube video illustrates the natural killer cells’ mechanism of action, and this one shows a detailed view of the immune synapse. Other, similar proteins have been identified in oyster mushrooms; these form more rigid structures that are easier to work with. Saibil’s group and their collaborators have been able to solve the structure of this protein in intermediate stages of pore formation and are beginning to gain an understanding of exactly how it unfolds.

The pore of the oyster mushroom protein pleurotolysin, a member of the pneumolysin family. Image © Natalya Lukoyanova and Helen Saibil, from Lukoyanova et al., PLoS Biology 13:e1002049

Mutations in perforin that prevent it from functioning cause a rare disease called haemophagocytic lymphohistiocytosis, which is almost invariably fatal in childhood. Understanding the mechanism of action of this important family of protein ‘weapons’ in both attack and defence may help find a cure for this devastating condition, as well as for some commoner disorders of the immune system and important infectious diseases.

This post is cross-posted from the Birkbeck Events blog.

PPS students can learn much more about electron microscopy by taking the second-year module Techniques in Structural Molecular Biology to complete the MSc.

GroEL, giving misfolded polypeptides a second chance

Saibil, H.R., Fenton, W.A., Clare, D.K., Horwich, A.L. (2013) Structure and Allostery of the Chaperonin GroEL. J Mol Biol. 2013 May 13;425(9):1476-87

A recent paper, by Professor Helen Saibil’s team at Birkbeck, reviews the current understanding of chaperonin GroEL.  Chaperonins attract unfortunate proteins which are incompletely or incorrectly folded and provide them with an isolated chamber in which to bind and release until they achieve their native functional state.  GroEL and its partner GroES are profiled in PPS in Section 7 (symmetry) and Section 8 (action as a chaperone).  

GroEL is remarkable in its construction.  It consists of 14 identical protomers arranged in two back-to back rings, each of the two rings with seven subunits.  This forms a barrel with a 7-fold rotational symmetry axis through its centre and, perpendicular to this, seven 2-fold axes of symmetry, giving an overall symmetry of 72.

Each subunit comprises two main domains linked by an intermediate domain (see figure (c)).  The largest domain is equatorial at the centre of the barrel.  This contains the ATP binding site and is in contact with its two neighbours in the ring as well as the equatorial domains of its partner ring.  These domains form a stable platform from which the other two domains undergo large movements orchestrated by the cycle of ATP binding, hydrolysis and release.  

The apical domains are exposed at the outer ends of the GroEL barrel.  They are smaller and include a hydrophobic surface which is the binding site for many different nonnative polypeptides.  The intermediate domain has a hinge at the junction with each of the two main domains, such that it can mediate large movements of the domains as rigid bodies.  This can be seen by comparing figures (c) and (f).

 

X-ray crystal structures of GroEL and GroEL-GroES complexes.  (a) Longitudinal cross-section of GroEL (PDB 1OEL). (b) Top view of the GroEL barrel. (c) A protomer of GroEL, aligned approximately as the top left protomer in (a).  (d-f) Show the same set of views with GroES (d-e) and ATP (f) bound (PDB 1SVT).  Example helices have been coloured to demonstrate the extent of the rotation angles.  The red and orange helices of the apical domains can be seen to undergo a significant rotation.  Compare this with the relatively minor movements of the green helices in the intermediate domains and the violet helices of the equatorial domains.

The operations of the GroEL chaperone are initiated by rapid binding of ATP to the equatorial domain of one of the rings.  This is followed by the binding of the unstructured, partly folded or misfolded polypeptide.

Natively folded proteins have their hydrophobic residues buried in the stabilising core whilst those which have lost their way have exposed hydrophobic patches.  These patches bind to the hydrophobic surfaces of the apical domains.

The final ligand is GroES, a ring of seven homo-oligomers, which forms a lid for the GroEL barrel.  Each GroES monomer has a flexible hydrophobic loop which binds to the hydrophobic regions of the apical domain alongside the substrate polypeptide.  This loop is visible in figure (d). 

The apical domains undergo significant concerted rotations together as one movement, with the domains being held as rigid bodies (compare figures (a) and (d)).   These rotations replace the hydrophobic polypeptide binding surface with hydrophilic residues, so propelling the nonnative protein into the central lidded cavity, where it is isolated to refold.

As ATP binding stimulates positive cooperative movements within the cis ring, that is the ring binding the nucleotide, it is also responsible for negative cooperation between the rings.  This means that while the movements are coordinated to bind GroES and promote protein folding in the ATP bound cis ring, in the trans or partner ring the opposite rotation prompts the release of GroES, more than 100Å away, and the expulsion of the now native protein.  

Mutation studies have revealed that salt bridges, which are studied in PPS Section 9, hold the rings steady until full ATP occupancy is achieved and are probably involved in the positive cooperativity whilst the negative cooperativity is thought to be triggered by a pivoting of the equatorial domains.  This interferes with the staggered contacts between each equatorial domain and two of its partner equatorial domains on the opposite ring.

Recent work using single particle cryo-electron microscopy techniques, which is studied in the TSMB course, has captured images of the intermediate states between ATP binding and the active chaperone state where GroES is fully bound.  

Once ATP binds, the intermediate and apical domains tilt 35˚ sideways from the lower hinge.  This brings the intermediate domain towards the ATP binding pocket where the residue ASP398 forms several hydrogen bonds.  This action causes the breakage of salt bridges between the intermediate and apical domains of neighbouring protomers and between neighbouring apical domains with new salt bridges forming which support the new tilted architecture.

Following this, the apical domains lift and separate, to use an old advertising slogan, causing further breakage of salt bridges between apical domains.  This separation could help to unfold the misfolded polypeptide before it is released into the chaperone chamber and also positions the hydrophobic binding areas for docking of the GroES binding loops.

Once GroES is bound, the apical domains lift even further outwards and undergo a 100˚ twist to create the active folding chaperone with the GroES lid in a domed position and the polypeptide is released into the cavity to complete its folding.

The next stage is hydrolysis of the ATP, which triggers the release of the ligands on the cis ring, and the acceptance of ligands on the trans ring.   

The mechanism is believed to involve separation of a β sheet contact between equatorial domains of the trans ring.   The equatorial domains are primarily responsible for holding the rings together so that the ADP complex has reduced stability. Hydrolysis is followed by ATP binding to the equatorial domains of the trans ring.  This promotes the pivoting of the equatorial domains that defines negative cooperativity and the discharge of GroES, the native protein and ADP from the cis ring, although the exact movements which lead to the discharge are unknown.

It seems likely that the release is mediated through a reversal of the twist in the cis apical domains.  This is speculation, however, as this good Samaritan of nanomachines has not yet given up all of its trade secrets.  The progress made to date, however, in large part by Professor Saibil’s team, is a striking demonstration of the power of this recently developed method in structural biology.