Atoms and Empty Space: the Structural Biology of Specificity

The eminent structural biologist Laurence Pearl, Professor of Structural Biology in the Genome Damage and Stability Centre at the University of Sussex, has strong links with Birkbeck. He studied for his PhD under Professor (now Sir) Tom Blundell at Birkbeck in what was then the Department of Crystallography (and the only full university department so named in the UK). The latest of the many honours and awards that he has received is the Biochemical Society's prestigious Novartis Medal for 2018. This is an annual award, made "in recognition of contributions to the development of any branch of biochemistry" and the winner is invited to present a lecture either at the Society's London headquarters or at one of its many conferences. Pearl gave his Novartis Medal Lecture, entitled Atoms and Empty Space: the Structural Biology of Specificity at the London office of the Biochemical Society on 13 November 2018.

This lecture was live-streamed on Facebook. It is linked from the Biochemical Society Facebook page, but you don't need to have a Facebook account to view it; just click on the video link below. (It is about an hour long.)

Video © Biochemical Society 2018

To summarise the lecture briefly, Pearl began with a well-known but still controversial quote from the Greek philosopher Democritus: "Nothing exists except atoms and empty space. All else is opinion." He then summarised the beginning of his scientific career as a postgraduate student at Birkbeck. During his PhD he learned computer programming and wrote - or contributed to - some of the first structural bioinformatics programs, as well as studying the structure of the aspartic protease, endothiapepsin (PDB 4APE). This was in the early 1980s, when the world was becoming aware of the scourge of AIDS and the necessity of targeting its viral cause, HIV. Sequencing the tiny HIV genome revealed a protease with similar sequence motifs to aspartic proteases, but less than half as long. It was Pearl, by then a postdoc at the Institute of Cancer Research in London, and Willie Taylor from Birkbeck who predicted, years before the structure was solved, that HIV's protease would be active as a dimer. (There is more about this in section 7 of PPS, on Quaternary Structure.) The HIV protease structure - exactly as Pearl and Taylor had predicted it - became an important tool for the discovery of protease inhibitors as drugs against AIDS. The first protease inhibitor to enter the clinic was named saquinavir after the amino acid motif SQNI that led them to predict the enzyme's dimeric structure.

Not surprisingly, this discovery also proved to be a gateway to Pearl's career as an independent researcher. He set up his lab in 1989 in the biochemistry department at UCL, with research projects in three areas: DNA repair, signal transduction, and a group of unrelated proteins that help other proteins to fold and that have collectively been named chaperones. He spent the rest of his lecture describing just a few of the many proteins that his group has studied during the last 30 years: first at UCL, then back at the Institute of Cancer Research and since 2009 at Sussex. These included the enzyme uracil DNA glycosylase, which is involved in the specific repair of cytosine residues in DNA that have been deaminated to form uracil. His co-worker on this project was Renos Savva, who is now a senior lecturer at Birkbeck and director of our MSc course in Biobusiness. The structure of this enzyme (PDB 1UDI) explains the exquisite specificity of this enzyme. Another, much smaller DNA repair enzyme, mismatch uracil glycosyase, was found to have the same fold. His more recent work, which uses electron microscopy as well as X-ray crystallography, includes the structure and mechanism of glycogen synthase kinase, and the structure and dynamics of further DNA damage and repair systems. Listen to the lecture if you want to find out more!

HIV gp41: conformational plasticity in response to cholesterol

Lai, A.L., Eswara Moorthy, A., Li, Y., Tamm, L.K. (2012) Fusion activity of HIV gp41 fusion domain is related to its secondary structure and depth of membrane insertion in a cholesterol dependent fashion.  Journal of Molecular Biology 418 (1-2): 3-15

Human immunodeficiency virus, HIV, attacks the human immune system by targeting CD4⁺ T- cells, which will be covered in PPS section 12.  The virus is coated in a lipid bilayer which supports trimeric glycoprotein complex gp120/gp41.  The gp41 trimer is transmembrane whilst the gp120 glycoprotein projects from the viral surface, recognising and binding the CD4 receptors and chemokine coreceptors expressed on the T-cell surface.  This binding provokes change in the interaction between the glycoproteins, causing gp41 to extend its N terminus towards the T-cell, leaving its C terminus firmly embedded in the viral lipid membrane.  The N terminus, which includes conserved hydrophobic residues, inserts into the lipid bilayer of the T-cell and then the elongated section between the two membranes refolds into a helical hairpin, so bringing the membranes close enough to fuse and allowing the virus to enter the cell.  Several therapies, for example Fuzeon®, target gp41 by inhibiting the formation of the six helix bundle (one hairpin per each of the three monomers).  This process is covered in section 7 of PPS under The Life Cycle of HIV.  

The mechanism of insertion of the gp41 fusion domain into the T-cell’s lipid bilayer is clearly of paramount importance in understanding the process of membrane fusion but the structure of the fusion domain has been controversial with several studies directly contradicting each other.  NMR studies of the gp41 fusion domain in solution in lipid micelles revealed an α helical structure but NMR studies of the fusion domain in solid state in lipid bilayers indicate a structure which is primarily composed of β sheets.  (NMR is a widely used spectroscopic technique which utilises the oscillations of nuclei in a strong magnetic field.  This technique is covered in the TSMB course but current PPS students will not be able to follow the link.)  Neither of the sets of conditions is close enough to physiological conditions to be conclusive and so a recent study, (Lai, A.L. et al., (2012)), was designed to examine the structure taken by the domain under a range of possible physiological conditions.

The cell membranes of both the virus and the T-cell contain approximately 30 mol% cholesterol.  (Mol% is mole percentage, the molar mass of a constituent as a percentage of the average molar mass of the sample.)  This cholesterol is not distributed evenly, however, but forms lipid rafts of high cholesterol concentration surrounded by cholesterol poor membrane.  This paper used circular dichroism (again, this link is to the TSMB course) to study the structure adopted by the gp41 fusion domain bound to a lipid bilayer in conditions of zero cholesterol, 20 mol% cholesterol and 30 mol% cholesterol.  Circular dichroism exploits the fact that different secondary structures absorb alternating circularly polarised light at characteristic wavelengths and so the percentage of α helical and β sheet secondary structure within a protein can be calculated.

The surprising results are that where there is an absence of cholesterol, the structure is clearly α helical.  As the cholesterol content increased, β sheet structures appear and α helical structures dissipate until at 30 mol% cholesterol the β sheets are seen to predominate.  Since the length of the domain in question is quite short it seems unlikely that it is a mixed structure but rather that the domain changes conformation as the lipid composition alters.  

To increase understanding of this, electron paramagnetic resonance, or EPR, was used.  The fusion domain was labelled at four points with a nitroxide which would give a saturation signal as it reacted with the oxygen that is more prevalent in the hydrophobic centre of the lipid bilayer.  This could be used to show that the α helical conformation penetrated the bilayer more deeply at approximately 8 Å while the β sheet was held slightly closer to the headgroup region.  This data could also be used to perform an innovative docking exercise.  Several structures of the fusion domain of gp41 in α helical conformation had been previously solved and the lowest energy one, PDB 2PJV, was selected and the four nitroxides were added to it computationally.  This structure was then introduced to a computational model of a lipid bilayer and rotations and translations were applied until the derived saturation signals gave the best fit with the experimentally measured ones.  The resulting model shows the α helix embedded beneath the phosphate headgroups with the hydrophobic sidechains extending into the hydrocarbon centre of the lipid bilayer.

Model of the fusion domain of HIV gp41 docking in the lipid bilayer

Lai, A.L., Eswara Moorthy, A., Li, Y., Tamm, L.K. (2012) Fusion activity of HIV gp41 fusion domain is related to its secondary structure and depth of membrane insertion in a cholesterol dependent fashion.  Journal of Molecular Biology 418 (1-2): 3-15

Surprisingly, it has been shown that membrane fusion between the virus and the T-cell is initiated whichever of the two conformations is adopted.   Given this, the question is raised as to the mechanistic implication of this extraordinary ability to switch conformations depending on local lipid conditions without altering functionality.  Does the fusion domain enter the target membrane at a cholesterol rich region in β sheet conformation or a cholesterol poor region in α helical conformation?  There is currently no answer but the study puts forward some appealing suggestions.

Possibly the insertion occurs across the boundary of the cholesterol rich raft.  In this case there may be a mix of conformations dependent on the exact location of each fusion domain.  Alternatively, the insertion may initiate in the lipid raft but then switch into the α helical conformation in order to pass through the less ordered cholesterol poor membrane.  This may be because this region of membrane is more conducive to fusion or because of the unproven hypothesis that the fusion domain interacts with the gp41 transmembrane domain to engender membrane fusion.  There are indications of this interaction in the fusion mechanism of the influenza virus and it is considered that a more deeply embedded α helix would provide a more attractive binding platform than a β sheet.  Having elucidated the structural effects of cholesterol on the fusion of the HIV virus with a human T-cell, the next step will be to understand the mechanistic implications as a route to possible new therapies.