Advancing Vaccine Technology to Combat Global Pandemic Threats

ABSTRACT

Current egg-based manufacturing methodologies for the production of influenza vaccines are slow and have an inherent lag period from variant identification to vial. In addition, seasonal influenza variants are predicted up to 12 months in advance by the World Health Organization but may only be confirmed five to six months before a vaccine reaches the market. Vaccine manufacturers have little time to produce and stockpile the selected candidates, and often have to manufacture one or more vaccine strains in advance at risk of the preferred strains altering. There is a need for a more flexible and rapid production methodology to produce cheap and effective influenza vaccines with minimal notice for pandemic variants. A new approach to combat this threat is required. There are several technologies in early development that may offer a more viable solution to the pandemic threat, such as the use of microbial-derived production processes and platform virus-like particle manufacturing strategies to alleviate some of these constraints and lead to a more rapid response time.

As of June 2009, the World Health Organization updated the status of the current outbreak of influenza A H1N1 swine flu (Figure 1) to a global pandemic, highlighting the requirement for a cheap, effective prophylactic vaccine and manufacturing strategy to raise the current state of preparedness for such situations. In 1918, one such influenza pandemic swept the globe and an estimated 50–100 million people died as a direct consequence. This historical data suggests that a similar flu pandemic in the modern era would affect 30% of the global population (approximately 2 billion) and an estimated 60 million individuals (1%) would die.¹ Combating the virus poses some significant and difficult challenges. Because the virus is an RNA virus, the lack of proof reading capability when the virus replicates can cause the virus to mutate and combine genetic material with another virus serotype resulting in variant forms, which may cross the species barrier wwAny vaccine must be specific to the current common strains of influenza virus to be effective; most vaccines are trivalent meaning they contain three inactivated virus strains to further increase the vaccine's efficiency and to overcome minor drifts in the virus morphology. Seasonal variants are predicted up to six months in advance with scientists selecting the virus strain and subtype that is most likely to protect against the virus in the first quarter of the year. Production of the vaccine by infecting live hen eggs, or more recently cells in culture, then begins and may take up to six months. The current production rate globally is approximately 300 million doses for each 6-month cycle in ideal conditions, providing vaccinations for only 5% of the global population.²

(EDEN BIODESIGN LTD)

There are several anti-viral drugs on the market, such as Roche's Tamiflu (oseltamivir phosphate), which can be used as an effective treatment for H1N1 influenza patients but have a different mode of action than a viral antigen vaccine. Tamiflu's mode of action is to prevent the release of the viral particles from infected cells, reducing the severity and duration of the infection; this is invaluable in reducing infection and controlling the spread of the disease but it has no prophylactic properties whatsoever. It has been estimated that Tamiflu production will be at 110 courses for 2009 alone.³

Figure 1. Images of H1N1 influenza virus. Image taken in the Centers for Disease Control and Prevention Influenza Laboratory2

Vaccination—Supply and Demand

The original licensed influenza vaccines were inactivated versions of a mixture of three influenza strains produced by an egg-based manufacturing process. To give an idea of scale, on average, between one and two eggs are needed to produce one dose of vaccine. During the production process itself, fertile hen eggs are infected with the candidate influenza strain and incubated for several days. The vaccine is purified from the allantoic fluid of virus-infected chick embryos by a combination of tangential flow filtration (TFF), density centrifugation, and, more recently, by anion exchange column chromatography. The virus is also chemically inactivated through formaldehyde or β-propiolactone treatment at some point during processing.

More recently, licensed inactivated influenza vaccines are made in mammalian cell culture and purified by more modern techniques. The major hurdle facing the manufacturers for the production of a pandemic vaccine is the time of production with the entire production process for a season's influenza vaccine taking an average of six months. The challenge to the vaccine producers is to develop a production process with a rapid turnaround time with minimal notice to combat a pandemic threat.

There are many approaches that may be taken for the development and production of a vaccine. An attractive prospect is the use of platform production systems which, in turn, would offer significant advantages for the production of pandemic vaccines, significantly reducing the lead in and production times and ultimately time to patient once the virus serotype has been identified. This approach will also allow for strategic manufacturing sites to be at a state of operational readiness for the production of the vaccine candidate with minimal notice. One such approach is the use of a microbial system for the production of virus-like particle (VLP) based vaccines.

An Emerging Solution—Virus-Like Particles (VLPs)

There are a number of expression technologies currently in use for the production of VLPs, including mammalian, insect and microbial-based systems. A microbial-based platform approach is ideally suited for this application. They are simple, cheap, and have a rapid turnaround time required for pandemic vaccine production. A microbial production system may also be favourable because, unlike insect and mammalian systems, they do not require incorporation of specific viral inactivation or reduction steps into the production process, which may be problematic when processing such large molecules as VLPs.

VLPs for pharmaceutical applications typically exploit the inherent ability of many viral structural proteins to self assemble. For instance, VLPs are formed when the 16 kDa hepatitis B viral core proteins spontaneously self-assemble to form icosahedral VLP structures of approximately 30–50 nm in a host cell as shown in Figure 2. They can be engineered to display a range of surface antigens, such as the hemagglutinin (HA) surface antigen providing a useful carrier for vaccine platforms. By presenting antigens in a VLP form, where typically 240 copies of the same protein can be presented in a single particle, it may be possible to stimulate the immune system in a more appropriate way than can be achieved through presentation of dissociated antigens adjuvanted with other macromolecules. One such example is iQur Ltd's (London, UK) tandem core VLP expression technology.⁴

Figure 2. Scanning electron microscope image of a preparation of virus-like tandem core particles displaying (a) Hepatitis A virus complete P1 polypeptide or (b) Hepatitis B virus surface antigen from E. coli (picture courtesy of iQur Ltd.)4

Manufacturing Strategies

Vaccines are complex and diverse biomolecules ranging from recombinant subunit antigens to live attenuated organisms. Therefore, a range of production technologies and process formats are required to manufacture sufficient quantities of these varied products. However, this wide range of production methods can be problematic because each one requires specific capital expenditure and specialized development approaches coupled with operator expertise and concomitant high cost, which limits the availability of vaccines to developing nations. With current comparisons suggesting productivity in the range of 10-g/L for prokaryotic systems versus a realistic short- or medium-term 3-g/L goal from eukaryotic processes, there is a growing trend and focus toward a lower number of production platforms by manufacturing in similar expression systems including microbial backgrounds.

These microbial platforms in particular offer an attractive solution to the often numerous challenges associated with the development of vaccine products. The rapid establishment of strains expressing the target recombinants is especially attractive to product developers, often facilitating early efficacy studies from multiple product variants for pandemic threats while paralleling process development and manufacturing activities.

In addition to a reduction in process costs of goods and the process durations associated with microbial systems, the production scales required for low-dose, high-potency products have an associated reduction in capital expenditure for manufacturing facilities; thus an all round attractive proposition for reactive development of vaccine products.

Fermentation Process

The final decision on the expression system of choice will be determined by the effect of post-translational modification (PTM) on product efficacy. However, where there are no specific PTM requirements, microbial-based production platforms gain an advantage. Previous resistance to using such systems, for instance in the production of Fab-based therapies have centred around the limitation for manipulation of vH to vL expression levels, which can affect both productivity and product efficacy. However, advances in transcriptional regulatory elements mean that coordinated control of these and other multicomponent products can be optimized to significantly increase process titers while assisting in the decrease of normally associated product impurities.

In the VLP vaccine arena, various companies have focussed their attention on decreasing development timelines while increasing product titers through the use of alternative cell-culture–based systems. Novavax, Inc. (Rockville, MD), has focussed on a rapid development strategy for a VLP product targeted toward emerging influenza viruses, which has resulted in a concomitant increase in both product potency and significant relative dose yields compared to the traditional egg-based manufacturing systems. An effective vaccine can be manufactured in approximately 10–12 weeks from the point of strain identification while clearly the requirement for scale in response to required doses has been significantly reduced. However, typical process (raw material) cost of goods may still remain high.

More recently, the development of effective vaccine treatments through yeast and E. coli-based VLP production platforms have further decreased development and manufacturing timelines and scale requirements while further increasing patient safety. With the development of platform downstream processing strategies and estimated recoveries in the region of 50–60%, even small-scale bioreactors remain a firm target for manufacturing campaigns using these systems. Indeed, estimated dose yields post purification are in the region of 500–1,000/L while development and manufacturing program timelines are expected to halve those associated with historical tissue-culture–based systems.

Downstream Processing

A platform approach to the purification process can also be adopted. The size of the particles can be advantageous during the purification process allowing generic steps to be developed. One such purification strategy is outlined in Figure 3.

Figure 3. Platform purification strategy for the production of virus like particles

The process broadly consists of cell harvest to separate the cells from the fermentation media, resuspension into a controlled, buffered environment, and cell lysis through high-pressure cell disruption. The bulk of the cellular debris is then cleared from the solution by microfiltration and the process solution concentrated and buffer exchanged using a high molecular weight cut-off TFF system. Further purification and polishing can then be achieved using anion exchange (AEX) and size exclusion chromatography.

Harvest and Cell Disruption

VLPs are produced as an intracellular product from a microbial production system. They must be released into the process stream to allow for subsequent purification. This can be achieved by several means, the most common being the use of high-pressure lysis. This is a process whereby the process solution is forced through a small fixed orifice at high pressure. The rapid transfer of the sample from a region of high pressure to one of low pressure causes cell disruption.

This type of cell disruption is by far the most efficient and reproducible but does have a number of drawbacks in processing of VLPs. The extremely high shear not only lyses the cells but also micronizes the cellular debris; this coupled with the large size of the VLPs themselves can make the subsequent clarification step problematic. Careful attention must be paid during the development of such a step to strike a balance between cellular disruption and the release of the VLPs into the process medium and the micronization of cellular debris to avoid fouling issues further downstream in the production process.

Clarification

A number of options are available for clarification including centrifugation, depth filtration, or micro-filtration to name a few. Generally, microfiltration is the preferred option as it is well established in the biopharmaceutical industry as a scalable and robust clarification technique. Whichever technique is used, this clarification step may be problematic post high-pressure lysis.

Concentration and Buffer Exchange

Because of the relatively large size of the VLP molecules, a high molecular weight cut off membrane such as 500 or 1,000 kDa may be used for this application. The large molecular weight cut off has the advantage of reducing the host cell protein (HCP) content while recovering the VLPs in the retentate fraction.

Anion Exchange Chromatography

An effective approach for the reduction of DNA and endotoxin levels is the use of AEX. Owing to the shear size of the particles, there are inherent mass transfer limitations that play a significant role during any chromatographic step. This reduces mass diffusion during column loading, reducing the effectiveness of the step and limiting of the column's overall dynamic binding capacity. Scale may then become an issue. There are several technologies available to negate this problem such as membrane adsorbtion technology, available from a number of suppliers such as Millipore, Pall, and Natrix Separations. Eden Biodesign has found the Poros HQ diffusion chromatography media (Applied Biosystems) to be effective. A typical chromatogram obtained during an anion exchange purification of a VLP is shown in Figure 4.

Figure 4. Anion exchange chromatogram for the purification of virus like particles (VLPs)

Size Exclusion Chromatography

Taking the platform approach, the size of the particles should be in the range of 30–50 nm. This is above the molecular exclusion limit for a number of size exclusion resins, and therefore, they may be used in a group separations mode whereby the large VLPs are above the upper exclusion limit for the resin to pass through in the void volume. The smaller HCPs and misformed VLPs pass through into the pores of the resin and are retained on the column for longer, eluting with a greater residence time.

A typical group separations chromatogram is shown in Figure 5, peak one in the void volume being the VLPs and peak two being gross impurities such as residual HCPs and misformed particles.

Figure 5. Size exclusion chromatogram for the purification of virus-like particles (VLPs)

There are a number of problems associated with the production of therapeutic products and vaccines from microbial systems, specifically from E. coli. These are centred mainly on the primary separations and reduction of lipopolysaccharides (LPS) from the process stream.

Processing Challenges

A major challenge in the production of any therapeutic products from E. coli is the requirement for the removal of LPS from the process stream, a specific processing step for this application may be required. LPS or endotoxins are a major component of gram negative bacteria such as E. coli. Endotoxins, even at relatively low doses can initiate an innate immune response resulting in a fever and in severe cases, septic shock and even death. Regulatory specifications exist for many parenteral products and are typically 2 IU per dose or less. There are many technologies specifically for this application such as LPS affinity chromatography and membrane adsorbtion technology. Membranes are a more attractive proposition because they are cheap, simple to operate, and require little or no cleaning validation. There are several options available for this application such as Pall's Mustang E and Sartorius's Q membrane. Membranes such as the Sartobind Q have a slightly different chemistry to the Mustang E but work on a similar charge effect, with the added advantage of a 3 μm pore size, so that few or no losses are seen because of the size exclusion effect that may be observed when processing such large molecules as VLPs.

However, this is not the complete picture; mechanically removing free endotoxins from the process stream is only half the challenge. Given the typically hydrophobic nature of VLPs, there is the additional problem of endotoxins adhering to the particles themselves, offering additional and significant further challenges to reduction. Bound endotoxins may be released by treatment with solvents, detergents, or a combination of both, further complicating the manufacturing process and adding expense. A combination of endotoxin reduction techniques is the more favourable approach. A production strategy in yeast such as Pichia pastoris may also be the favoured approach to avoid this problem.

Conclusions

The prospect of using platform production strategies, both upstream and downstream for the rapid production of pandemic vaccines is becoming a reality. The major bottlenecks and time constraints imposed by current egg-based or mammalian cell culture manufacturing strategies could be alleviated by switching to a platform microbial production process. There are a number of such systems currently in development that in real terms will significantly reduce the vaccine's time-to-market and ultimately save lives.

ANDREW CLUTTERBUCK is the downstream purification team leader, NITIN JAIN is a downstream purification scientist, and DAVID SIMPSON is the process development manager, all at Eden Biodesign, Ltd, Liverpool, UK, +44 151 728 1750, andrew.clutterbuck@edenbiodesign.com

References

1. Jeffery K. Taubenberger JK, Morens DM. 1918 Influenza: the Mother of All Pandemics. Emerg Infect Dis. 2006 Jan;12(1)15–22 Available from: http://www.cdc.gov/eid.

2. Centers for Disease Control and Prevention. Available from: http://www.cdc.gov/h1n1flu/images.htm.

3. Roche provides additional donation of 5.65 million packs of Tamiflu to World Health Organization. Roche Media News. 12 May 2009. Available from: http://www.roche.com/med-cor-2009-05-12-e.pdf.

4. Technical data reproduced with kind permission from iQur, Ltd. (London, UK) Available from: http://www.iqur.com.