Vaccine design at the atomic level

Abhay Kotecha from the Division of Structural Biology of the Nuffield Department of Medicine at the University of Oxford explains how scientists are creating new safer and more stable empty particle vaccines for foot-and-mouth disease virus

Understanding the three-dimensional structure of biological macromolecules such as proteins provides a wealth of information about the organisation of individual atoms and their chemical makeup. This atomic detail information offers enormous potential to rationally engineer proteins to enhance their properties. In this case, the coat proteins of viruses that form a shell (capsid) surrounding and protecting the genetic material can be optimised to improve the quality of vaccines. Along with collaborators from The Pirbright Institute and the University of Reading, this is essentially what I and colleagues here at the Division of Structural Biology at the University of Oxford have done. We have rationally designed new, safe and more stable, recombinant empty particle vaccines against foot-and-mouth disease virus (FMDV) that can be produced at commercially viable levels.

Foot-and-mouth disease (FMD) is highly contagious and globally it is the most economically important disease of livestock, affecting cattle, pigs, sheep, goats and other cloven-hoofed animals. The disease is characterised by fever and lesions resembling blisters inside the mouth and on the feet of infected animals. The disease is not fatal in adult animals; however mortality is high in new-borns. Although most animals recover within four to six weeks, it leaves them debilitated resulting in severe loss in milk and meat production. FMD is endemic in large parts of Africa, the Middle East, Asia and South America.  Other parts of the world, such as North and Central America, New Zealand, Australia and Chile, as well as some European countries including the United Kingdom, are currently FMD free (Figure 1). In endemic countries, FMD affects both national and international trade and has a major economic impact on the livestock industries and damaging consequences for local farmers leading to an invariable loss of income and increase in poverty. Although, the disease is mainly restricted to the developing world, developed countries are not completely immune to the threat posed by sporadic outbreaks. The UK’s 2001 FMD outbreak had a massive impact on the economy. It lasted for almost seven months spreading all across the country, resulting in the slaughter of over 6 million animals and cost in excess of £8 billion.

[caption id="attachment_33711" align="aligncenter" width="352" caption="Figure 1: Global status of FMDV. FMD is currently endemic in many parts of the world including the Middle East, Asia, Africa and some parts of South America."][/caption]

There are seven different serotypes of foot-and-mouth disease virus (FMDV): O, A, Asia1, C and three South African Territory viruses (SAT-1, SAT-2 and SAT-3). Each serotype consists of a number of antigenic variants or sub-types contributing to the dynamic pool of antigenic variation. In addition there is limited antigenic cross-reactivity within serotypes and sub-types of FMDV; therefore a vaccine against one serotype does not protect animals against infection from other serotypes or necessarily even sub-types of the same serotype, necessitating the continual development of new vaccine strains. Routine vaccination programmes are employed in the endemic regions to control the spread of the disease. Current vaccines are made by chemically inactivating the live virus particles which no longer remain infectious after inactivation; however, the production of these vaccines requires strict bio-containment manufacturing facilities which are very expensive to maintain.

A major factor affecting the potency of the FMDV vaccine is the requirement of intact inactivated virus particles. Individual proteins or peptides have proven to be insufficient at producing immunogenic response. However, FMD vaccines are unstable in various environmental conditions, including mild acidic conditions (pH<7.0) or warmer temperatures (>30?C) where inactivated capsids readily fall apart into immunogenically useless pentameric subunits. As a result, vaccines have a very limited shelf-life of about six months and require expensive cold chains for storage and transport which can be very difficult to maintain, especially in poor countries. Because of the inherent instability of FMDV capsid, the vaccine efficacy is also severely reduced and the immune response is short lived, requiring frequent immunisation of animals. Improvement of vaccine stability is thus of the utmost priority to control the disease and maintain a country’s FMDV free status. In addition to stability, there is a potential for the escape of live virus from the production facilities involved in current vaccine production, or as a consequence of incomplete inactivation of vaccines.

In order to design better FMD vaccines, it is essential to understand the structure of the capsid in atomic detail. The first structure of an FMDV capsid was solved in our lab back in 1989¹ but it is only recently that the technologies have become available to re-engineer these capsids to produce stable vaccines. The capsid initially assembles with three proteins VP0, VP1 and VP3, which are derived from a single polyprotein by cleavage with the viral 3C protease. Then as the virus matures VP0 is cut into VP2 and VP4 to form the final structure consisting of 60 copies each of four structural proteins, VP1-4, arranged in an icosahedral lattice of 12 pentameric building blocks (Figure 2). These pentameric subunits are major structural intermediates for capsid assembly and disassembly. Five protomers self-assemble into pentameric subunits and twelve pentamers are subsequently assembled around the viral genome to form the capsid. In the course of natural infection, empty virus particles, which lack the viral RNA genome but resemble the mature virus in structure and antigenicity, are also produced but these empty capsids are very unstable.

[caption id="attachment_33712" align="aligncenter" width="300" caption="Figure 2: The design for the production of recombinant FMDV empty capsids and the assembly of protomers into the capsid showing the icosahedral lattice arrangement of the viral proteins; different viral proteins are coloured as VP1 (blue), VP2 (green) and VP3 (red). The pentameric building blocks are highlighted and expanded and the histidine residue located on the ?-helix at the two-fold symmetry axis between pentamers, shown in atomic detail in the bottom panel."][/caption]

A number of different strategies have been employed by several research groups around the world to produce novel FMD vaccines as an alternative approach to conventional vaccines. These include synthetic peptide vaccines, empty capsid vaccines and virus-like-particle vaccines. However, all these approaches have so far proved to only confer partial protection against infection for livestock. The most successful approach was demonstrated by scientists at the Plum Island Animal Disease Centre in the USA which used in vivo production of empty FMDV particles generated with the live replication defective recombinant adenovirus vector containing a minimal FMDV genome encoding VP0, VP1, VP3 structural proteins and 3C protease². However, there are a number of limitations with this approach such as the need for large doses of virus, presumably due to severe instability of the FMDV empty capsids, pre-existing and induced immunity to adenovirus which can restrict re-immunisation and moreover the use of live replication defective adenovirus which can cause regulatory issues in meat production industries in some countries. More recently FMDV empty capsids were expressed using baculovirus and silkworm larvae to express FMDV structural proteins. The empty capsid vaccines prepared in this manner for two FMDV serotypes, Asia-1 and O, produced immunogenic responses in mice and conferred protection in cattle³. Similarly, O serotype empty capsids were also assembled using the baculovirus insect cell expression system and were shown to confer protection against challenge in cattle experiments⁴. However, recombinant expression yielded very limited amounts of intact capsids not suitable for large scale commercial vaccine manufacture.

The major barrier to producing recombinant empty capsid, in addition to the capsid instability, is the effect of the 3C protease. It is toxic to the insect cells used to express the empty capsids and kills the cells, terminating the capsid production. With a careful balancing of the expression of structural proteins and 3C protease we demonstrated the expression of recombinant FMDV empty capsids at commercially viable levels⁵. In addition, to solve the problem of capsid instability, the three-dimensional structure of FMDV was computationally analysed. This highlighted specific atomic interactions along the inter-pentameric interfaces and, in turn, allowed us to model a potential energetically favourable disulfide bond at the 2-fold axis between adjacent pentamers which would essentially lock the pentamers together. Thus we have re-engineered the capsid interface at an atomic level. Disulfide bonds are more robust covalent cross-links than the non-covalent interactions that generally hold proteins together. Because of the high symmetry of the virus shell and the disulfide bond at a 2-fold axis, there are thirty such bonds in the re-engineered capsid increasing the overall stability of the empty capsids by many folds. Both original and stabilised capsids were successfully produced in insect cells.

Following the rational design and production, the crystal structures of the original and the re-engineered capsids were analysed at atomic level using the microfocus macromolecular crystallography beamline (I24) at Diamond Light Source, the UK’s synchrotron science facility. I24 is the best beamline of its kind in the world and indeed, exactly as predicted, the structure showed the presence of the disulfide bonds locking the pentamers together (Figure 3). When tested for stability, the re-engineered capsids were found to be significantly more stable, surviving for at least two hours at temperatures of up to 56?C and pH of down to 5.5. These properties will make transporting and storing the eventual vaccine much easier in warmer climates.

[caption id="attachment_33713" align="aligncenter" width="300" caption="Figure 3: Microcrystals of FMDV capsids (top panel) and X-ray crystallographic analysis of the original (bottom left) and re-engineered FMDV capsids (bottom right). The original capsid shows two histidine residues (93His) whereas the re-engineered capsid shows the presence of the 93 Cys residues forming a disulfide bond between pentamers, as predicted."][/caption]

When the stabilised re-engineered empty capsids were used in cattle vaccination trials, they were found to induce equivalent neutralising antibody titres to that of standard inactivated FMDV vaccines and, nine months after vaccination, the animals were still protected from infection. Detailed findings of this research are published in journal PLOS Pathogens⁶.

This new vaccine is safe to produce and does not require extensive bio-containment regulations during manufacturing. In addition, the enhanced stability of re-engineered empty capsids will reduce losses during production, storage and transport of the vaccines.  Finally, the absence of non-structural proteins in recombinant empty capsid vaccines will allow the development of diagnostic tests to discriminate between infected and vaccinated animals. Such empty capsid vaccines are likely to form the basis of the next generation of stable, safe and infection risk-free vaccines against the disease. The rational structure-based approach that we have demonstrated here may also be applicable across a wide range of human and animal diseases including other picornaviruses that affect humans such as polio and coxsackieviruses.