Plant-produced vaccines could be the way of the future for pandemic influenza vaccine production, ever since two different groups showed that production time for a new H1N1 pandemic vaccine could be as short as one month from the first announcement of the pandemic strain sequence, to having the end-product ready for testing in animals [12
]. This is significantly shorter than the 4 month minimum production time required for the seasonal vaccine made by egg-based technology. In addition, compared to plant-produced vaccines, egg-based vaccines also have the disadvantage that certain strains, such as H5N1, do not grow well in eggs and production is severely limited [3
To illustrate the advantage for vaccine production via transient plant expression, when a deadly H3N2 influenza strain emerged in Brisbane, Australia (A/Brisbane/10/2007), researchers at the Fraunhofer Institute (USA) took 36 days to produce a purified candidate vaccine via transient expression in tobacco plants (800 mg/kg fresh plant material) [8
]. In addition, a Canadian company (Medicago Inc.) were able to obtain purified H1-VLPs within 21 days after the initial pandemic H1N1 HA sequence became known [13
]. Medicago and the Fraunhofer Institute (USA) also announced at the ‘Influenza Vaccine for the World 2009’ conference that human clinical trials will be implemented for their plant-produced HA influenza vaccines. These are currently in progress [14
In our study, we aimed to establish a platform to produce influenza A virus subunit vaccines via tobacco plants in South Africa, in order to minimise the impact of a potential pandemic on our population. Pre-pandemic awareness and stockpiling has also been encouraged and is deemed necessary for H5N1 as it is still viewed as a potential global pandemic threat due to its high mortality rate [16
]. In this case, stable transgenic seeds can be very useful to have on standby in case of a H5N1 pandemic. Our H5 and H5tr stable transgenic plant lines expressed vaccine protein at high levels for two consecutive generations – which is a first for transgenic production of influenza HA protein. When needed, these stored transgenic seeds can be planted at large scale, and the resulting transgenic plants can be used to purify H5 vaccines within a few weeks.
By targeting the HA variants to different plant cell compartments, we were able to determine which localities in the plant cell had the highest potential for the accumulation of our recombinant proteins. Our findings from both the transient and the stable transgenic systems indicated that full-length H5 accumulated at the highest level in the apoplastic spaces. This may be due to the presence of the HA transmembrane domain which can allow budding from the cell membrane in the form of VLPs, as observed by D’Aoust et al. [10
]. In our study, the apoplast-targeted HA also provided us with a potential alternative to leaf homogenization which should reduce contaminating plant proteins and simplify the purification process. We were able to recover H5 protein from H5-expressing leaf tissue by infiltrating leaves with a buffer containing a detergent (Triton X-100), followed by low speed centrifugation of whole leaves. By using this method we were also able to extract H5 localised in the chloroplast or cytoplasm, but to a lesser extent (data not shown). This was likely due to the presence of Triton-X100 in the buffer, which allowed the elution of intracellular proteins by solubilising the membranes. This is an attractive alternative extraction method especially for apoplast-accumulating proteins, as it markedly reduces co-extraction of intracellular proteins and insoluble plant material.
H5tr accumulated at high levels in the ER. The ER is a popular target for expression of recombinant proteins, particularly glycoproteins, often resulting in high accumulation levels [17
]. The contrasting behaviour of H5tr and H5 with respect to accumulation – the former high in the ER and low in the apoplast, with the inverse for H5 – is not easily explained. It is possible that SEKDEL-tagged H5 is less stable than the H5tr equivalent.
In comparison to other studies investigating HA expression in plants [6
] we obtained similar transient expression levels (62–675 mg/kg) than previously reported (1–800 mg/kg). Our choice of expression vector, infiltration methods and codon optimised HA gene sequence can possibly account for our high HA accumulation. Although immobilized metal affinity chromatography (IMAC) purification (via His-tag) has previously been successful for this protein [8
], we could not utilize the same purification strategy for our full-length H5, as it did not contain a His-tag. We were able to purify the His-tagged truncated H5 by IMAC; however, our yields were too variable and too low for us to use the purified material for animal experiments. We therefore decided to generate our preliminary data using a crude concentrated and diafiltered product, in which the HA proteins were enriched. More important, though, is not the amount of protein produced, but the fact that the plant-produced H5 protein exhibited the correct conformation as shown by the success of HA and HI tests in comparisons with known standards.
The murine trial demonstrated the ability of plant-produced H5 and H5tr to induce HA-specific antibodies, which was verified by western blot. The candidate H5 and H5tr vaccines therefore elicited the correct antibody response in vivo
. However the HI titres were relatively low for all the sera. Results for the sera of mice vaccinated with H5 showed that only three had HI titres above 1:64 the minimum HI titre considered to be protective in humans is
] which indicates that the plant-produced H5 in this study would have elicited a protective immunity. The HI titres for mice vaccinated with H5tr were inconclusive. One of the reasons that the full length H5 did not elicit high HI titres might be due to the extraction method employed. The addition of Triton-X100 to the extraction buffer possibly also destroyed any VLPs that were formed.
Western blot analysis of H5-immunised chicken sera clearly demonstrated the presence of HA-specific antibodies. HA-specific antibodies were also present in the sera from H5tr-immunised chickens, but to a lesser degree. In general, the antibody concentration was too low to inhibit haemagglutination of red blood cells during HI testing, with the exception of two serum samples from the H5tr-immmunised chickens (1:16 and 1:32). Even though the HI titres were lower than 1:40 in both mice and chickens immunised with H5tr, it still indicated that our candidate H5tr vaccine warrants further investigation.
HI titres were relatively low for all chicken and mice sera: immune responses to our vaccine candidates may have been suboptimal because of dose and adjuvant choice (IFA). Other studies that have tested plant-produced HA subunit vaccine candidates in animals have used various influenza strains, regions of HA, purification protocols, dosages, adjuvants and test animals - all which may play a role in vaccine efficacy. In mice, Shoji et al
] tested three doses ranging from 5 μg to 30 μg together with 10 μg of Quil A adjuvant. D'Aoust [10
] administered two doses of 0.5 μg H5-VLPs and Spitsin et al.
] used two doses of 10 μg with alum-CpG adjuvant. When considering immunisation studies conducted on ferrets, the doses varied from a 100 μg with alum adjuvant [7
] to 45 and 95 μg HA in Quil A adjuvant [8
]. Ferrets, which are the preferred animal model for influenza [19
], were not available for our study; we decided to conduct our vaccine trial on chickens since these can be naturally infected by H5N1. We deduced that our vaccine regimen (three doses of 13 μg antigen) may have been too low to induce high antibody titres in chickens. A dose-ranging study will be necessary to determine the optimum antigen and adjuvant dosage. As mentioned above, the full-length H5 antigen may not have elicited high HI titres due to the presence of Triton-X100 in the extraction buffer, which may have destroyed any VLPs that were formed. D’Aoust [10
] showed that low doses of 0.5 μg VLPs protect against lethal challenge in mice and that H5 antigen which did not form VLPs induced up to six fold lower HI responses. It would be interesting to determine if VLPs were formed in our study and to develop extraction methods for them.
Based on our current results, it is not clear whether the plant-produced H5 or the H5tr showed the best potential as a candidate subunit vaccine. Overall, a higher concentration of H5-specific antibodies was detected in chicken than in mouse sera; however, only two of the H5tr chicken serum samples and three of the H5 mouse serum samples gave low positive HI results.