Seeing the Wood for the Trees in Structural Biology

The British Crystallographic Association (BCA) was set up in 1982 to support UK scientists working in crystallography and other structure-based sciences. It has five specialist groups (four discipline-based, and one for young crystallographers): the Biological Structures Group for structural biologists holds its main annual conference each December, generally just before the Christmas break. Several of these one-day Winter Meetings have been previously described in this blog. The 2016 meeting, however, was particularly relevant for anyone connected to Birkbeck: not only was it held in the college, but it celebrated the work of one of the college’s most distinguished structural biologists, Steve Wood. The meeting title was, of course, a pun on his name.

Wood worked with Professor Sir Tom Blundell at Birkbeck in the 1990s to solve the structure of an important small human protein, serum amyloid P component (SAP or pentraxin; PDB 1SAC). This protein forms pentamers that bind to amyloid fibres and it is thought to be involved in the protection of those fibres from breakdown by proteases. Pentraxin-binding compounds that interfere with this process might be useful as treatments for amyloidosis and other diseases associated with protein aggregation, perhaps including Alzheimer’s disease.

Blundell, a former head of Birkbeck’s Crystallography Department and now emeritus professor of Biochemistry at the University of Cambridge, kicked off the meeting in fine style. He had known Wood since they were, respectively, a young lecturer and a PhD student at the University of Sussex in the 1970s, and they have published over 60 papers together. His talk surveyed the structural biology of multi-protein signalling systems over the last 40 years. The earliest such system to be discovered involved the control of blood sugar levels through insulin and glucagon binding to their receptors. The general principles developed through structural studies of this relatively simple system have been applied to other, more complex ones including the interaction between the breast cancer susceptibility protein BRCA2 and a recombinase enzyme that controls one type of DNA repair. Mutations that interfere with this binding lead to greatly enhanced susceptibility to some cancer types. Blundell’s group at Cambridge set up a database, CREDO, to catalogue the interactions involved in all macromolecular complexes in the PDB. Many protein-protein interactions are now actual or potential drug targets. Some promising drugs for solid tumours act by inhibiting the interactions between cyclins and cyclin dependent kinases (CDKs) that drive cells through the cell cycle. Astex Pharmaceuticals, the drug discovery company set up by Blundell and some of his Cambridge colleagues in 1999, has one such CDK inhibitor – ribociclib – that has completed Phase III clinical trials for advanced breast cancer.

Garry Taylor, who gave the next talk, joined Blundell’s group as a postdoc soon after its move to Birkbeck in the mid-70s, where he established a long, productive collaboration with Wood and with Jim Pitts, who now directs the PPS course. Taylor is now a professor at the University of St Andrews in Scotland where he studies the structure and mechanism of sialidases. These enzymes hydrolyse (break) the bond between a terminal sialic acid residue and the remainder of a polysaccharide or glycoprotein; both bacterial and viral sialidases are involved in the pathology of infectious disease. All sialidases share a catalytic domain with a characteristic beta propeller fold, but the bacterial enzymes have a separate carbohydrate-binding domain (CBD). This binds tightly to the sialic acid substrate of all sialidases, including that of influenza virus neuraminidase (which will be covered in detail in section 10 of the PPS course). Taylor and his group were awarded a grant to explore the idea that this domain, alone, might bind tightly enough to sialic acids on the surface of influenza virus host cells to prevent both virus entry and the release of progeny virions. They have now developed multi-valent CBDs that can protect mice from challenge with a lethal dose of influenza virus. Taylor suggested that, if these molecules are as successful in protecting against influenza in human trials, they might also be useful prophylactics for other respiratory pathogens that bind to cells via sialic acid receptors.

Jonas Emsley, one of Wood’s many PhD students at Birkbeck, is now at the University of Nottingham where his group studies the structures and mechanisms of proteins involved in blood coagulation. His talk focused on the activation and assembly of proteases in the contact system, in which the presence of ‘foreign’ surfaces such as bacteria triggers several physiological processes including blood clotting. Inappropriate activation of this system has been linked to heart disease and stroke, and mice that lack either of the coagulation factors Factor XI and Factor XII are protected to some extent from thrombosis. Factor XI, which is activated by Factor XII, contains four repeats of a domain with six conserved cysteine residues that can be drawn in the shape of an apple, hence its name of ‘apple domain’. The protein circulates as a dimer with the monomer-monomer interactions mediated by one apple domain and the catalytic domains sitting on top of the eight apple domains like a cup on a saucer. There is a pocket on the surface of each apple domain, and the pocket on the second such domain binds a conserved tripeptide, DFP, that is found in many of its substrates. Small-molecule inhibitors of this interaction might be useful anticoagulants.

Structure of factor XI apple domain with bound peptide substrate showing the conserved DFP motif. Image (c) Jonas Emsley

Other speakers included Birkbeck’s Helen Saibil, whose ground-breaking high resolution electron microscopy of protein complexes has been covered many times in this blog (see e.g. posts from April 2015 and July 2013) and Neil McDonald, now based at the Francis Crick Institute in London, who described some largely unpublished work on the structure and mechanism of RET receptor tyrosine kinases. Appropriately, however, the final talk was devoted to Wood’s structure: SAP. It was given by Simon Kolstoe who joined the Wood group in Southampton as a PhD student in 1999, moved with him back to UCL and is now at the University of Portsmouth. He first presented a ‘potted history’ of structural studies of this protein, describing how a competitive inhibitor of SAP-amyloid binding was developed as a potential treatment for amyloidosis at the turn of the millennium. This compound, CPHPC, was found to deplete SAP levels in serum but, unfortunately, clinical amyloid levels were unchanged. A high-resolution structure of this compound binding to SAP was published in 2014 (PDB 4AVV). Kolstoe and his co-workers have now turned their attention to SAP binding to DNA, which might also be clinically relevant.

The meeting ended with the usual votes of thanks, with the award of a poster prize to Jingxu Guo from University College London, and with a gift to Wood: a molecular model of a SAP-drug complex, presented by Tony Savill of Molecular Dimensions Ltd.

Image of two molecules of SAP coordinated with five molecules of CPHPC. Image (c) Simon Kolstoe, PDB 4AVV

E coli BirA protein: the multi-tasking interaction surface

Adikaram, P.R., Beckett, D. (2012) Functional Versatility of a Single Protein Surface in Two Protein:Protein Interactions. Journal of Molecular Biology 419 (3-4): 223-233

 

A recently published study (Adikaram, P.R., Beckett, D. (2012)) brings together a number of topics examined in the Principles of Protein Structure course.  Section 10, Protein Interactions and Function, discusses dynamic protein-protein interactions including those of various enzymes and Section 8, The Protein Lifecycle, covers protein-DNA bonding.  This paper investigates a protein, BirA from E. coli, which uses a single interaction surface to interact as either a metabolic enzyme or a transcription repressor depending on the cellular requirement at any time and it considers the evolution of this multipurpose surface.

Both of the interactions start with the binding of the protein BirA to biotin and ATP to form an intermediate enzyme complex.  Biotin is a critical B complex vitamin which is synthesized by bacteria in the gut in humans and which has crucial roles in metabolism and the Krebs cycle.  If the organism is in growth mode the intermediate complex forms a heterodimer with acetyl-coenzyme A (CoA) carboxylase and transfers the biotin to a receptor subunit, so constructing the enzyme which catalyses the initiation of fatty acid synthesis.  Only when the CoA carboxylase is depleted is it energetically favourable for the BirA-biotin-ATP intermediate to fulfil its other function and form a homodimer.  The homodimer binds sequence specifically to DNA at the biotin biosynthetic operon and acts to repress the initiation of transcription.

Image adapted from (Adikaram, P.R., Beckett, D. (2012)).  This model was created using PDB 2EWN for the homodimer, PDB 1BDO for the biotin carboxyl carrier protein subunit and PDB 2EJG as a template for the heterodimer.

 

The interaction surface at the centre of both these complexes is a β sheet surrounded by five loops and both the heterodimer and the homodimer form by extension of this β sheet.  The differences occur in the loops.  Two of the loops have sequences which are found to be conserved in biotin ligases in organisms ranging from humans to bacteria.  This implies that these loops have preserved critical functionality in either the formation of the BirA-biotin-ATP intermediate or in the formation of the heterodimer or in both.  The other three loops have variable sequences and this was taken to indicate that degenerate evolution across species has produced different methods of the homodimerization required to form a transcription repressor.

In this study, 18 residues were selected across the constant and variable loops and were individually substituted for alanine to elucidate the effect of each one on the energetics of both reactions.  It was found that the transfer of biotin to its receptor protein was significantly affected by seven of the residues, most of which were part of the constant loops.  This was consistent with expectation given the conserved nature of these residues.  More surprisingly, 11 residues were found to impact the homodimerization and these were distributed across both the variable and the constant loops with some of the constant loop active residues being key to both reactions.  

So how do these results on the active residues fit with what is known about the surface interactions of the two dimers?  The heterodimer is maintained by two separate interaction sites; the constant loops provide the primary bonding region which is also the active site of the enzyme whilst a second bonding surface is provided by one of the variable loops.  This is consistent with the finding that not all of the seven key residues in heterodimerization were located on the constant loops.  The homodimer interface also comprises two interaction sites but these are symmetrical and each consists of a variable loop of one of the monomers in continuous contact with a constant loop from the other monomer.  This explains why the critical residues for homodimerization were found to include some of the constant loop residues which are also critical for heterodimerization.  This discovery has led to some interesting conclusions on the evolution of this multi-functional interaction surface.

The conserved nature of the constant loops demonstrates that the interaction with biotin and subsequent heterodimerization, which is critical to the formation of the metabolic enzyme, was the primary function of BirA, evolved in an ancestor common to bacteria and humans.  Substitution of these key residues leads to a serious depletion in the energetics of this reaction, showing that further evolution to incorporate a second function would have needed to accommodate these original sequences.  The variable loops have therefore evolved subsequently both to play a supporting role in the stabilisation of the heterodimer and, when cellular regulation allows it, to be complementary to the constant loop residues in formation of a homodimer for transcription repression.

The findings illustrate that a single surface can be used to perform two distinct functions, necessitating two distinct protein-protein interactions, where the structure required for one function has evolved to be complementary to that required for the other and where a regulatory switch is present to activate the appropriate mode.

The topic of protein-protein interactions is explored in much more detail in section 11 of the TSMB course.  This link takes you to the overview of TSMB but clicking the syllabus tag on the left hand side will give you the topics covered in each section.