Vaccine farming in Cape Town

PlanT-Derived vaccines: REVIEW

review

Human Vaccines 7:3, 339-348; March 2011; © 2011 Landes Bioscience

Vaccine farming in Cape Town Edward P. Rybicki,1,2,* Anna-Lise Williamson,1 Ann Meyers2 and Inga I. Hitzeroth2 Institute of Infectious Disease and Molecular Medicine and 2Department of Molecular & Cell Biology; University of Cape Town; Cape Town, South Africa

1

Key words: biopharming, vaccines, plants, HIV, HPV, papillomavirus, geminivirus The review details the development of the Subunit Vaccine Group at the University of Cape Town, from its beginnings as a plant virology laboratory in the 1980s. The investigation and development of Human papillomavirus (HPV) and Human immunodeficiency vaccine candidates are covered in detail, with an emphasis on how this work allowed the evolution of a systematic approach to the optimization of expression of these and other proteins especially in plants, but also in insect cell culture. We discuss various insights gained during our work, such as approaches to codon optimization, use of different vector systems and plant hosts, intracellular targetting and gene modification. The future prospects for both our work and for the field of plant-made vaccines in general, are discussed.

new avenue also was the molecular characterisation by cDNA cloning of the ssRNA potyviruses; 8,9 this led to more cloning and sequencing work on geminiviruses,10-12 which in turn led inexorably to the application of DNA and cDNA amplification by PCR to the characterisation of poty- and especially of geminiviruses,13,14 but also of Human papillomaviruses.15-17 A major parallel interest that developed at this time was the use of sequence data obtained by whole- or partial-genome sequencing for phylogenetic analysis of virus evolutionary relationships: this resulted in a number of reviews and papers, on phylogeny of potyviruses,18,19 geminiviruses,20-22 papillomaviruses,17 HIV,23 Beak and feather parrot circovirus (BFDV),24 and viruses in general.25 The first dabbling that led to the biotechnology that we now practice—expression of foreign proteins in plants—was a collaboration with Jennifer Thomson of this Department on the Agrobacterium tumefaciens-mediated transformation of tobacco—Nicotiana tabacum—with constructs derived from South African strains of Tobacco necrosis (TNV) and Cucumber mosaic (CMV) viruses, aimed at conferring resistance to these viruses.26 This line of work has continued to the present, with our recent success—twenty years after initiation of the project—in engineering transgenic resistance into maize (Zea mays) against Maize streak virus.27,28 Two early and abortive projects on expressing vaccine-related proteins were initiated (and ended) in the mid-1990s: these were the expression of HPV-16 L1 major capsid protein in transgenic tobacco (see later), and the development of CMV as a plant expression vector (Gehringer M, Williamson AL, Varsani A, Rybicki EP, unpublished). The first successful effort at expressing foreign proteins in plants for purposes not linked to resistance to infection, was the high-level expression of the enzyme phosphinothricin acetyl transferase (PAT) in Black Mexican sweet maize cell suspensions, via a replicating MSV-derived vector,29 and an attempt to make MSV into an infectious gene vector.30 This was part of a focussed effort at the time concentrating on MSV molecular biology;31 however, we very much had in mind adapting geminiviruses to be useful expression vectors (reviewed in ref. 32). It is interesting that we and others have come full circle on this a decade later33 (see below). Another important development in our lab in the early 2000s was the establishment of a centre of expertise in the use of recombinant baculoviruses for expression of proteins in insect cell cultures. This was first done in order to express individual MSV proteins (van der Walt E and Rybicki EP, unpublished), then found use in the expression of BFDV proteins;34 however, it rapidly found use as a “gold standard” system for making recombinant vaccine-related proteins (see below), and in particular,

©201 1L andesBi os c i enc e. Donotdi s t r i but e. Introduction and History

The Subunit Vaccine Group at the University of Cape Town has its origins in a plant virology laboratory first established by Marc van Regenmortel and Barbara von Wechmar in the 1970s and continued since 1985 by Ed Rybicki. The lab used serology as a means of plant virus detection and characterisation, which led to a familiarity with the then-novel techniques such ELISA and western blotting and immunoaffinity purification, which has been very useful in present work.1-4 Work on plant viruses moved on from serology to characterisation of complexes of viruses infecting or found in or associated with crop plants—and especially Brome mosaic bromovirus and Barley yellow dwarf luteoviruses (BYDV), and the insect viruses Rhopalosiphum padi and Aphid lethal paralysis dicistroviruses (RhPV and ALPV) which were found in wheat and barley, and then in the aphids infesting these crops.5,6 This led to further skill development, largely to do with purification of vanishingly small amounts of viruses (1–10 μg/kg) from large amounts of plant material, as well as working with nucleic acids for the first time. The next phase of work—in the mid-1980s—took in molecular biological techniques, and in particular nucleic acid-based technologies, for the characterisation and differentiation of viruses. The first step in this direction was the novel application of restriction mapping as a means of characterizing and differentiating the dsDNA replicative intermediates of the genomes of Maize streak mastreviruses (MSV; Geminiviridae), the causative agents of the most serious virus disease of maize in Africa.7 A *Correspondence to: Edward P. Rybicki; Email: [email protected] Submitted: 06/28/10; Accepted: 11/21/10 DOI: 10.4161/hv.7.3.14263

www.landesbioscience.com

Human Vaccines

339

vaccines. The HIV vaccines were DNAand poxvirus-expressed versions of a multigene HIV construct40,41 as well as Salmonella-expressed Gag protein,42 and a major effort on developing Pr55Gag-based subunit vaccines via recombinant baculovirus and plant expression of protein. The HPV vaccines included plant- and insect cell-made HPV-11 and HPV-16 L1 protein, and a Bacillus Calmette-Guerin (BCG)-expressed HPV-16 L1.43 These projects were followed by awards for BFDV vaccine development in insect cells and plants, an award for development of plant and insect cell expression of novel chimaeric HPV L1 proteins and rotavirus capsid proteins, and a grant for novel HPV vaccines as well as local funding for emerging and pandemic influenza virus vaccines. This continuity of good funding, in the unlikely environment of a developing Figure 1. Schematic of the approach that should be taken to produce vaccines or other “farmacountry, led to the establishment of an ceuticals” in plants. excellent infrastructure and solid expertise in a variety of vaccine protein expression HPV-16 and other PV L1 major capsid proteins and derivatives systems—and has enabled us to recently source funding from the European Union (FP6 CompuVac and FP7 PlaProVa consortia), of them. As a result of this varied experience, our laboratory was partic- as well as from the international biotech industry (Era Biotech, ularly well placed in South Africa in the late 1990s and the begin- Spain). The following review will detail particularly the “bioning of the 2000s to exploit a surge of interest from local funding farming” or plant expression-related aspects of our work, as well agencies in molecular and plant biotechnology in general, and as on where we think the field should go in future. vaccine biotechnology in particular. We had established collaboPlant-Based Vaccine Research rations inside and outside South Africa on human viruses of vaccine interest; we had expertise in molecular as well as in basic virology; we had the infrastructure and the expertise to handle It is worth pointing out here that, unlike many other groups who extraction of proteins from large amounts of plant material; we started in the “pharming field” in the early years, we have always had experience in making transgenic plants and plant cells; we taken the view that plant-made vaccines should prove themselves had established “conventional” animal cell culture systems—and as biosimilars to established vaccines, rather than as totally new the premier basic science funding agency in South Africa had just edible or oral vaccines. This approach is outlined in Figure 1, effectively dropped plant virology as an interest area, meaning we a graphic that Ed Rybicki has been using for some ten years in talks describing our work: thus, what we believe is the way to were badly in need of research funding.35 We started with small grants for making HPV L1 protein in go for this kind of work is to produce proteins in a non-crop plants and insect cells, and incidentally became involved in an plant—transgenically or transiently—and extract these from the investigation with Robert C Rose of the University of Rochester plants for formulation as possibly orally delivered or (preferably) into the feasibility of using insect cell-made HPV L1 virus-like injectable preparations. We will describe our work on human particles (VLPs) as oral vaccines: this was highly successful, in papillomavirus (HPV) vaccines in particular, but also our investhat we showed that orally-dosed mice developed humoral and tigations into the feasibility of making vaccines in plants against cellular immunity to HPV-11, -16 and -18 as well as neutralizing HIV and influenza viruses, and draw together lessons from all of antibodies.36,37 This sparked an interest in our lab in being able to these examples on how to improve vaccine-related protein promake L1 VLPs more cheaply—which led to ten years of work on duction in plants. plant and insect cell expression of HPV proteins. In 1999, we became part of two major local vaccine develHPV Vaccines opment consortia: these were a South African AIDS Vaccine Initiative (SAAVI) project on development of vaccines against In the mid-1990s, our laboratory first attempted to express HIV-1 subtype C,38 and a project on novel vaccines against the L1 capsid protein of a South African isolate of HPV-16 in HPV.39 Both projects involved conventional as well as plant-based transgenic tobacco. However, subsequent testing with HPV-16

©201 1L andesBi os c i enc e. Donotdi s t r i but e.

340

Human Vaccines

Volume 7 Issue 3

pseudovirions,44 when these became available, showed that no neutralising antibodies had been elicited (Varsani A, Stewart D, Williamson AL, Rybicki EP; unpublished results): this was unsurprising as the protein was subsequently shown not to assemble correctly.45 This notwithstanding, this was the first time anyone had done this with a HPV L1 gene, and paved the way for subsequent successful work. Successful expression of a native Figure 2. HPV-16 L1 VLPs produced in plants and in insect cells. Electron micrographs of uranyl acetateHPV-16 L1 protein gene in our lab stained preparations: (A) rTMV-produced VLPs made in N. benthamiana, trapped with Mab H16:V5; (B) CsCl-purified VLPs made via baculovirus (bar = 70 nm).49 was achieved early in the 2000s: 46 the gene detailed above was corrected by PCR-mediated in vitro mutagenesis (1,234 A→T) (Varsani A, Williamson AL and Rybicki protein and plant system could not be extrapolated to another. EP, unpublished; GenBank accession AY177679), and used to In the first place, it was found that, unlike the case with HPVmake constructs which would express full-length (L1) and a 16 L1, transgenic N. tabacum cv. Xanthi plants expressed native version with the C-terminal 22 aa double nuclear localization CRPV L1 at levels higher than were achieved in N. benthamisignal (NLS; 484-KRKatpttsststtaKRKKRKL-505) deleted ana via rTMV—1 mg/kg vs. 0.4 mg/kg, due to instability of the (L1ΔC483). These were cloned in pART27, which was used to rTMV—and at much higher levels than were obtained using the transform Nicotiana tabacum cv. Xanthi. Transgenic plants pro- HPV-16 L1 native gene, whether transgenically or transiently duced both HPV-L1 capsomers and VLPs (see Fig. 2); however, expressed.50 This protein did not appear to assemble into VLPs; yields were no higher than 4 μg/kg wet weight of leaves. The however, it was protective in animal experiments (see below). reason for leaving off the NLS was that similarly truncated ver- Second, transgenic N. tabacum cv. Xanthi or Arabidopsis thalisions are known to be expressed better than the native protein ana plants—transformed using pART27 binary vector-base in both the yeast and insect cell systems used for commercial constructs, exactly as for HPV-16 and CRPV L1 genes—could vaccine manufacture; however, in this case it did not appear not be regenerated if they expressed a full-length native HPVto affect protein expression and was possibly negatively corre- 11 L1 gene, but only if they expressed a C-terminal truncated lated with particle formation. As previously, concentrated plant (22 aa) NLS-version:51 this was in contrast to what had been extracts were only weakly immunogenic if injected into rabbits; seen for HPV-16 L1 by us and by others,46,47 but similar to the however, partially purified preparations containing VLPs bound only other successful expression of HPV-11 L1 protein, which conformation-specific MAbs which are used as a surrogate for was done in potato using a plant codon optimized (“plantized”) predicting neutralization, and in this were effectively identical gene also lacking the NLS. However, this was only expressed at to insect cell-made VLPs. Our study, together with two oth- levels around 20 μg/kg,48 compared to the levels we achieved of ers published almost simultaneously, on HPV-16 and HPV-11 2 mg/kg in tobacco and 11 mg/kg in Arabidopsis. Interestingly, L1 VLPs respectively,47,48 clearly showed that it was possible to we achieved better expression levels of HPV-11 L1 NLS-via produce HPV vaccine candidates in plants—but that the yields rTMV in N. benthamiana than in transgenic N. tabacum (10 needed to be significantly increased. mg/kg vs. 2 mg/kg)—and the protein did assemble into VLPs, We therefore investigated the potential of a Tobacco mosaic and the rTMV was stable51 (Kohl T, Hitzeroth II, Stewart D, virus (TMV)-based expression system—the so-called pBSG Williamson AL, Rybicki EP; unpublished results). A finding Geneware from the now-defunct Large Scale Biology Corp—for that has important implications for HPV vaccines produced in increasing yields of HPV-16 L1 protein, in Nicotiana benthami- plants was that the N. tabacum HPV-11 L1 product was signifiana plants infected using in vitro-generated viral ssRNA.49 This cantly proteolytically degraded compared to the Arabidopsis or system was better for L1 expression—however, while yields went N. benthamiana-produced L1, and antibodies elicited by injecup to 40 μg/kg wet weight, and particles were made that were tion of rabbits or guinea pigs with adjuvanted concentrated plant immunogenic and antigenically identical to insect cell-made extracts from the latter containing NLS-L1 cross-reacted only VLPs, the yield was still far short of a workable figure. weakly with full-length insect cell-produced L1, and were not Work being done in parallel in the lab on Cottontail rabbit neutralising in pseudovirion infection assays. Thus, while the papillomavirus (CRPV) L1 and HPV-11 L1 protein expression work showed that other vaccine-relevant PV L1s could be propointed up an interesting phenomenon to do with the effect of duced at reasonable levels in plants, it further showed that results the nature of the HPV L1 gene on expression levels: this was obtained using HPV-16 could not necessarily be extrapolated to that even related HPV L1 native genes expressed at very differ- other PV L1 proteins—which may have important implications ent levels in identical expression systems, and that results for one for making mixed-VLP type vaccines in plants.



Human Vaccines

341

Figure 3. Papilloma growth on the backs of rabbits following challenge with infectious CRPV. Papilloma sizes were measured weekly beginning at day 14 and the gross mean diameters (GMDs) calculated. (A) The mean GMDs and SEM of papillomas were plotted against time for the sites challenged with the 10-2 dilution of infectious CRPV. Control group, immunized with BCG expressing irrelevant rotavirus antigen; VLP, rabbits immunized with purified CRPV L1 VLPs produced via baculovirus in insect cells. (B) Papilloma development on negative control rabbit, and (C) plant-derived vaccine treated rabbit. Figure adapted from reference 50.

native genes—our use of a native HPV-11 L1 gene gave far better yields (~500-fold higher) than a plantized gene used in reference 48, so there was no simple rule to follow. Initially, we repeated our previous tactic of using pART7/27 for transgenic expression in N. tabacum cv. Xanthi, this time with plant codon optimized and human codon-optimized versions: we note that the latter two variants were independently formulated to those previously used in Germany.47 By selection from a large number of lines, we were able to show maximum production levels of about 1 mg/kg for the plantized, and 4 mg/kg for the humanized—compared to our previous 4 μg/kg for the native gene (Maclean J, Williamson AL, Rybicki EP, unpublished results). While this was a significant improvement, it essentially reiterated the results claimed by Biemelt et al. for a humanized gene—and was still not an optimal yield. We then did a systematic investigation of the effects of codon optimization and intracellular localization on accumulation of the protein, transiently expressed in N. benthamiana via agroinfiltration of a new set of optimized expression vectors obtained from Rainer Fischer’s laboratory (pTra series), using the native gene in parallel with the humanized and plantized genes.53 The results were as successful as they were unexpected: while the humanized gene expressed better than any other, the native gene was better than the plantized version—and localizing the protein into plastids via a Rubisco small subunit-derived chloroplast import signal (pTraCTP) further increased protein accumulation. It was the scale of the improvement in yield, however, that was the interesting feature: the humanized (hL1) gene producing a plastidlocalized protein allowed yields of up to 550 mg/kg or over 17% of total soluble protein, two orders of magnitude better than anything we had seen for this protein. Interestingly, in light of results with transgenic HPV-16 and HPV-11 L1 expression mentioned earlier, a NLS-version of the L1 accumulated in plastids to far lower levels (~100 mg/kg) than the full-length version of L1 (550 mg/kg). Thus, the strategy for increasing yield that worked in insect cells and in yeast does not work for HPV-16 L1 in plants. Injection of mice once with 11 μg of centrifugally-concentrated L1 extract—with or without Freund’s incomplete adjuvant—elicited sera with endpoint titres in excess of 1/40,000 and HPV-16 pseudovirion neutralization titres of greater than 1/6,400, as good or better than two doses of 10 μg of insect cell-produced VLPs. These values are comparable with titres elicited in humans by the commercial vaccines,54,55 meaning we have proof of principle that a plant-made HPV vaccine would probably work. We extended the optimization strategy that worked for HPV16 L1 protein to the L2 protein, as this has significant vaccine potential in its own right—given that it elicits a wider range of cross-protective antibodies than L1,56 —and also stabilizes VLPs if co-expressed with L1. As for HPV-16 L1, it was necessary to use a human codon-optimized gene rather than plant-optimized or the native version, but the protein accumulated equally well via pTra-mediated transient expression in chloroplasts, cytoplasm and the endoplasmic reticulum, to a level of ~30 mg/kg (Pereira R, Hitzeroth II, Rybicki EP, unpublished). This is potentially very useful, as the protein occurs in virions at a ratio of between 1–6:30 (i.e., a maximum of 1 L2 molecule per L1 capsomere), and as we get yields of L1 that are ~10–20-fold higher, co-expression of the


The most important outcome of this phase of work was the investigation of the efficacy of the CRPV L1 produced in either transgenic N. tabacum or transiently via rTMV in N. benthamiana as vaccines in an animal model.50 This protein apparently did not assemble past the pentameric capsomer stage; however, it reacted well with conformation-specific MAbs. Rabbits injected three times with concentrated plant extracts with Freund’s incomplete adjuvant—a total of 80 μg for transgenic extracts, and 12 for rTMV extracts—developed immune sera that reacted well with baculovirus-made CRPV L1 VLPs, but did not detectably neutralize CRPV pseudovirions in an in vitro assay. While the lack of complete assembly and lack of neutralization indicated that the vaccine might not work, to our surprise the plant-derived vaccines both protected rabbits against challenge with live CRPV better than the “gold standard” baculovirus-made product (Fig. 3). This was an important first proof of efficacy of a plant-made papillomavirus L1 protein vaccine, and further contradicted the now-weakening established wisdom that immunity to PV infections requires neutralizing antibodies (reviewed in ref. 43 and 52). Given this success, the next phase of work with HPV involved attempting to increase the very poor yields obtained with HPV16 L1. While codon optimization had led to yield increases in others’ hands previously—Biemelt et al. had found that a “humanized” HPV-16 L1 gene was far better than plantized or

342

Human Vaccines

Volume 7 Issue 3

proteins could result in efficient L1+L2 VLP formation. Given these results, one would be forgiven for thinking that there was nothing left to do in making a successful plant-produced HPV vaccine—especially when generation of transgenic N. tabacum cv. Petit Havana localizing L1 protein to plastids also gave spectacular yields, with one line in particular accumulating over 250 mg/kg.53 However, it proved impossible to maintain these yields in transgenics: seed produced from these transformed lines or even plantlets produced from them by micropropagation, either yielded no L1 protein at all or very little—a phenomenon almost certainly due to post-transcriptional gene silencing following meiosis (Maclean J, Koekemoer M, Stewart D, Williamson AL, Rybicki EP, unpublished results). However, agroinfiltration can be used successfully on a commercial scale to produce pharmaceutically-important proteins,57 so the prospect is not too diminished. We have also recently managed to increase the yield of transiently expressed HPV-16 L1 in N. benthamiana by 50% by use of a replicating geminivirus-derived vector,33 improving the prospect still further. Another group were successful in getting up to 3 g/kg of HPV-16 L1 produced in transplastomic tobacco,58 means there is also another route to a plant-produced HPV vaccine—either of which could hopefully be exploited in the not too distant future. Current work in the laboratory on HPV vaccines is focussed on making chimaeric second-generation L1-based vaccines—based on earlier successful work done in insect cells with Figure 4. HIV-1 Pr55 Gag virus-like particles made in insect cells and in plants. Transmission HPV-16 L1 protein containing the L2 protein electron micrographs of HIV-1 VLPs. (A) baculovirus-produced Gag VLPs and (B) VLPs immunotrapped using HIV-1 p17 antiserum from crude sap of N. benthamiana infected with rTMV cross-type protective neutralizing epitope-conexpressing humanized gag (Meyers A, Williamson AL, Rybicki EP, unpublished results). 59 taing sequence LVEETSFIDAGAP —as well ® as in harnessing Era Biotech’s Zera peptide in order to make easily-purifiable “protein bodies” of an other- HIV virions; they are also potent immunogens with adjuvantwise soluble HPV-16 E7 oncoprotein-derived immunogen as a ing activity which can elicit both humoral and cellular immune potential therapeutic vaccine for cervical cancer.60,61 Both inves- responses,62 and Gag contains probably the greatest density of tigations are being done in insect cells as well as in plants via T-cell epitopes in the HIV proteome,63 making it a good target agroinfiltration: interesting parallels and differences have been for vaccine-stimulated immune responses. found in expression, which will be reported soon (Burger M, We were possibly the first to show that a South African Whitehead M, Hitzeroth II and Rybicki EP, unpublished). HIV-1 subtype C isolate gag gene inserted into a recombinant baculovirus genome could express VLPs in insect cells; imporHIV Vaccines tantly, we also showed that a low dose of the particles provided a potent boost to a gag DNA vaccine prime in mice.64 In fact, Our involvement with HIV vaccine research was as part of a vaccination with DNA plus VLPs resulted in 29% of Ifnγ+/CD8 + large multidisciplinary team, with a number of approaches being T cells assayed by flow cytometry being Gag peptide-specific, investigated in parallel. One of these—the focus of the Subunit compared to 6% for gag alone, 9% for VLPs alone and 9% for Vaccine Group—was the use of Pr55Gag virus-like particles 2x gag vaccinations. These results were followed by studies in as part of a mixed vaccination approach. Gag VLPs are truly baboons which showed that parenteral vaccination with insect virus-like in being enveloped particles of about the same size as cell-made Gag VLPs was very effective at boosting Gag-specific



Human Vaccines

343

Table 1. Types of gene construct and modes of expression of papillomavirus L1 genes in plants L1 Gene

Host

Vector

Presumed Localization

Yield

Ref

Native HPV-16

N. tabacum cv. Xanthi

pART7/27

cytoplasm/nucleus

~4 μg/kg

46

Native HPV-16 ΔC483 (NLS-)

N. tabacum cv. Xanthi

pART7/27 (transgenic)

cytoplasm