Expression of HPV-11 L1 protein in transgenic Arabidopsis thaliana and Nicotiana tabacum

Background We have investigated the possibility and feasibility of producing the HPV-11 L1 major capsid protein in transgenic Arabidopsis thaliana ecotype Columbia and Nicotiana tabacum cv. Xanthi as potential sources for an inexpensive subunit vaccine. Results Transformation of plants was only achieved with the HPV-11 L1 gene with the C-terminal nuclear localization signal (NLS-) encoding region removed, and not with the full-length gene. The HPV-11 L1 NLS- gene was stably integrated and inherited through several generations of transgenic plants. Plant-derived HPV-11 L1 protein was capable of assembling into virus-like particles (VLPs), although resulting particles displayed a pleomorphic phenotype. Neutralising monoclonal antibodies binding both surface-linear and conformation-specific epitopes bound the A. thaliana-derived particles and – to a lesser degree – the N. tabacum-derived particles, suggesting that plant-derived and insect cell-derived VLPs displayed similar antigenic properties. Yields of up to 12 μg/g of HPV-11 L1 NLS- protein were harvested from transgenic A. thaliana plants, and 2 μg/g from N. tabacum plants – a significant increase over previous efforts. Immunization of New Zealand white rabbits with ~50 μg of plant-derived HPV-11 L1 NLS- protein induced an antibody response that predominantly recognized insect cell-produced HPV-11 L1 NLS- and not NLS+ VLPs. Evaluation of the same sera concluded that none of them were able to neutralise pseudovirion in vitro. Conclusion We expressed the wild-type HPV-11 L1 NLS- gene in two different plant species and increased yields of HPV-11 L1 protein by between 500 and 1000-fold compared to previous reports. Inoculation of rabbits with extracts from both plant types resulted in a weak immune response, and antisera neither reacted with native HPV-11 L1 VLPs, nor did they neutralise HPV-11 pseudovirion infectivity. This has important and potentially negative implications for the production of HPV-11 vaccines in plants.


Background
Papillomaviruses are small species-and tissue specific double-stranded DNA tumour viruses, classified in the taxonomic family Papillomaviridae. High-risk genital HPVs types 16,18,33 and 58 are the leading cause of cervical cancer [1], and low-risk genital HPVs such as the related types 6 and 11 cause benign epithelial papillomas or warts. HPV-11 is recognised as one of the most prevalent anogenital papillomaviruses and is the main causal agent of benign genital warts (condyloma acuminata) and laryngeal condylomas. Furthermore, HPV-11 DNA has also been found to be associated with various other mucosal surfaces [2][3][4][5].
Given a HPV-11 prevalence rate of 5-12% in normal women [6][7][8], a serious recent concern is the impact of increasing human immunodeficiency virus (HIV) infection rates, and the associated immunosuppression of HIVpositive individuals, on HPV-6 or 11 coinfections. HPVassociated disease is the most common coinfection and comorbidity in immunosuppressed individuals [9]. HPV infections are more readily detected in HIV-seropositive women, are more persistent, more severe and more difficult to treat than HPV infections in HIV-seronegative women, and recur more frequently. Silverberg et al. [10] have found the prevalence of HPV-6 and 11 to be up to 5.6 times higher in HIV-seropositive women, thereby increasing the prevalence of genital warts by a factor of 3.2. The associated morbidity and negative effects on quality of life are a major problem among HIV-infected women (L Denny, pers comm). Thus, although HPVs 6 and 11 are not cancer-causing, infections can be disfiguring and cause severe discomfort. The complications of HPV-11 coinfection in HIV-seropositive individuals necessitate the urgent development of a safe, efficacious and inexpensive vaccine against HPV-11.
While efforts to develop HPV vaccines have largely concentrated on the cervical cancer-causing HPV-16, HPV-11 has also received widespread attention. Extremely high healthcare costs are associated with management of noncancerous HPV-6/11 disease, and inclusion of HPV-6 and 11 in a vaccine might be advantageous [11]. Phase I clinical trials have proven that HPV-11 L1 virus-like particle (VLP) vaccines are safe [12]. Merck has developed a quadrivalent HPV vaccine, Gardasil™, which includes HPV-6, 11, 16 and 18 and is produced in yeast: two large Phase III trials have been completed and it was licensed for use in the United States in June 2006 [13]. A modelling study [14] predicted significant improvements in the quality of life and prevention of cancer upon vaccination of young girls with an HPV vaccine that was only 75% efficacious. However, the Merck HPV vaccine will cost US$360 for 3 doses/person -an amount that is higher than the annual per capita health expenditure of many third-world nations. Therefore, other strategies for the production of stable, cheaper HPV vaccines are more immediately appropriate for these constituencies.
A strategy for the large-scale production of inexpensive HPV vaccines is production in plants: this could be between 10 and 50 times cheaper than its production in fermentation systems. Delivery of these by the oral or "needle-free" route is ideally suited to the background setting of vaccination campaigns in many developing countries [15]. We and others have shown that plant-expressed HPV-11 L1 and HPV-16 L1 proteins assemble into antigenically-appropriate capsomers and VLPs that are highly immunogenic upon parenteral and/or oral delivery to animals [16][17][18][19][20]. HPV-11 VLPs were expressed in potato; however, yields were very low and oral administration of tuber material was only weakly immunogenic [19]. We have also shown that a plant-produced cottontail rabbit papillomavirus (CRPV) capsomer vaccine protected rabbits against the formation of CRPV-induced papillomas [21], which so far is the only proof of concept for a plantproduced PV L1 protein vaccine.
This study aimed to develop a plant-derived L1 protein subunit vaccine against HPV-11 in a plant host from which it could be easily purified at reasonable yield, in order to formulate a potentially cheap vaccine that could help prevent HPV-11 infection in developing countries. Through Agrobacterium tumefaciens-mediated transformation, we genetically manipulated Arabidopsis thaliana and Nicotiana tabacum, to express the HPV-11 L1 gene. Comparative studies on the expression of the gene were performed on both expression systems. Products were characterised antigenically using monoclonal antibodies and used to immunise rabbits, in parallel with insect cellproduced antigens.

Genetic analysis of transgenic plants
Despite several attempts to transform A. thaliana and N. tabacum plants with the HPV-11 L1 and HPV-11 L1 NLSgenes, only plants containing the HPV-11 L1 NLSgene could be regenerated. Attempts to introduce the fulllength HPV-11 L1 gene failed in both plant systems. PCR screening of putative HPV-11 L1 NLStransgenic A. thaliana first generation (T 1 ) and N. tabacum regenerated generation (R 0 ) plants, using primer pair 4 as listed in Table  1, showed that 8/20 A. thaliana T 1 generation and 4/15 N. tabacum R 0 lines were positive for the gene.
Inheritance of the transgene was established by self-pollinating T 1 plants and screening seeds on selective media. Random PCR screening of plant lines up to the T 6 generation for A. thaliana and T 3 generation for N. tabacum showed that the gene was stably integrated and successfully inherited ( Figure 1A and 1B).
To determine transcription of the integrated HPV-11 L1 NLSgene, total RNA of successive generations of A. thaliana and N. tabacum plants was analysed by RT-PCR. Amplification of a 303 bp product indicated the presence of the HPV-11 L1 NLStranscript and we determined that the HPV-11 L1 NLS -mRNA was present in PCR-positive transgenic plants in all generations ( Figure 1C and 1D).

Analysis of plant-derived HPV-11 L1 NLSprotein
The final product resulting from homogenization, filtration and centrifugation of transgenic leaf material was analysed by standard techniques and showed that HPV-11 L1 NLSprotein was indeed present in the concentrated plant protein extracts (Figures 2 and 3).
Monoclonal antibodies (Mabs) to HPV-6 (H6:C6, H6:E51 and H6:I2); HPV-11 (H11:B2 and H11:H3) and to HPV-16 (H16:D9) displaying specific and cross-reactive properties (Table 2) were used in the ELISA to detect and characterise the antigenicity of HPV-11 L1 NLSprotein derived from both plant hosts (Figure 2A and 2B). For A. thaliana-derived HPV-11 L1 NLSprotein, T 3 generation plants from all 8 lines were harvested and pooled. Our initial interest was in determining whether or not the protein was expressed at all in A. thaliana, rather than in selecting high or low expressers. The limited biomass obtainable from these plants necessitated the pooling of leaf material from all transgenic lines. The resulting protein extract was characterised with all the above listed Mabs. The results in Figure 2A show that all Mabs bound to the A. thalianaderived L1-containing protein extract. Binding of HPV-11 neutralising conformation-specific Mabs H11:B2 and H11:H3 to antigen in the plant protein extract was essentially equivalent to the binding to the positive control, suggesting that A. thaliana-derived HPV-11 L1 NLSprotein could potentially elicit a neutralising antibody response once administered.
Further, Figure 2A shows evidence of binding of the HPV-6/11 cross-reactive surface linear Mabs H6:C6, H6:E51 and H6 I2 to the protein extract: these Mabs bind intact VLPs as well as denatured L1 protein. Although binding of these Mabs was not of the same magnitude as to the positive control (Figure 2A), together these results suggest that most of the A. thaliana-derived HPV-11 L1 NLSprotein is assembled similarly to insect cell-produced protein.
Analysis of the antigenic properties, of the N. tabacumderived HPV-11 L1 NLSprotein from all generations from 4 lines showed that the surface linear and HPV-11 neutralising antibody H6:I2 bound best. Two other cross-reactive Mabs (H6:C6 and H6:E51) also bound; however, binding efficiencies of these surface exposed linear epitope-recognizing and non-HPV-11-neutralising Mabs were significantly lower ( Figure 2B).
Conformation-specific and HPV-11 neutralising Mabs H11:B2 and H11:H3, which only bind capsomers and intact VLPs, did not react significantly with plant extracts, suggesting that most of the N. tabacum-derived HPV-11 L1 NLSprotein exists in an unassembled state. The rather weak detection of the HPV-11 L1 NLS -VLP spiked positive control by MAb H16:D9 was anticipated, as it is not suitable for detecting intact VLPs ( Figure 2B).
A similar assay on insect cell-derived HPV-11 VLPs diluted in PBS instead of non-transgenic plant extract gave qualitatively identical results to those in Figures 2A and 2B (result not shown), indicating that addition of plant sap does not change the VLP antigenic properties.

Quantitation of the L1 protein
The amount of L1 protein in transgenic plant extracts was measured by comparison to an ELISA-derived standard curve for known concentrations of insect cell-derived HPV-11 L1 NLSprotein. Non-transgenic plant protein extract was initially spiked with insect cell-derived HPV-11 L1 NLS -VLPs resulting in a known concentration of 0.4      Insect cell-produced VLPs in non-transgenic plant extract immunotrapped using Mab H11:H3 represent the positive control ( Figure 3C); negative controls were protein extracts derived from non-transgenic A. thaliana (not shown) and N. tabacum ( Figure 3D): these showed no HPV-like or any other particles upon examination.

Western blotting
Detection of HPV-11 L1 NLSprotein by western blot is shown in Figure 4. While characteristic polypeptide bands were detected in both plant extracts, the L1 protein extracted from A. thaliana-derived L1 protein showed no proteolysis: a ~55 kDa band in (A) matched that of the positive control. In contrast, however protein from N. tabacum (B) was extensively proteolysed (~36 kDa) indicating that protease activities differ markedly between the 2 plant systems. HPV-11 L1 NLSin insect cell lysate detected with H6:C6 (shown in C as Insect cell (IC)) was cleaved twice, resulting in 3 products: one of these is ~36 kDa in size. Non-infected insect cell lysate was used as negative control in C.

Animal immune response to plant-derived HPV-11 L1 protein
Eight New Zealand white rabbits were immunized with concentrated transgenic A. thaliana and N. tabacum protein extracts. Serum collected on days 1, 15, 28, 42, 56 and 70 was analysed for anti-HPV-11 L1 reactive antibodies by ELISA and results are shown in Figure 5. Rabbits #11 and #12 were immunised with T 3 generation transgenic A. thaliana-derived L1 protein extract calculated to contain between 21 μg and 43 μg.
A distinct difference was noted in the immune responses of these 2 rabbits: serum from rabbit #11 reacted well with insect cell-produced NLS -VLPs ( Figure 5), while rabbit #12 showed no significant antibody response and can hence be classified as a non-responder.
Rabbits #13 -#18 were immunised with N. tabacumderived protein extract: two rabbits per group were immunised with N. tabacum protein extract from T 1 generation transgenic lines X1, Xh and Xi respectively. According to the range of protein yield calculated per kg of fresh leaf material, inoculum from line X1 contained between 13-47 μg of L1 protein, whereas inocula from lines Xh and Xi contained between 10-114 μg and 9.5-42 μg of L1 protein respectively.
The most interesting result was obtained by evaluating serum reactivities in ELISA against insect cell-produced HPV-11 L1 NLS + VLPs. Serum from rabbits inoculated with A. thaliana-derived L1 protein extract failed to react, whereas sera from rabbits#16 and #17 immunised with N. tabacum-derived L1 protein appeared to weakly bind this antigen, as indicated by rising absorbances of successive bleeds about two fold higher than the prebleed level (results not shown). This lack of recognition of the HPV-11 L1 NLS + VLPs is of great concern because these VLPs most closely resemble the actual virus: the implications of this will be discussed later. To further investigate this, Day 56 antisera were tested against denatured and non-denatured insect cell-derived HPV-11 L1 NLS + and NLS -VLPs by ELISA. The results shown in Figure 6 essentially reconfirm those presented in Figure 5. With the exception of rabbit #11, the antisera from all remaining rabbits immunized with either plantderived protein extract reacted predominantly with nondenatured insect cell-derived HPV-11 L1 NLS -VLPs, and not with denatured protein from either type of VLPs. This was confirmed by western blots of the insect cell-derived NLS + and NLSprotein. While very strong reactions were seen with baculovirus and Sf21 cell proteins in both prebleed and last bleed sera, neither L1 protein was detected (data not shown). This suggested that the predominant antibody response was most probably not targeted against any linear epitopes.

Pseudovirion neutralisation assays
All rabbit sera taken on days 1 and 56 were tested for their capacity to neutralise HPV-11 pseudovirions. Serum sample dilutions ranging from 1:50 to 1:12150 were incubated together with HPV-11 L1 pseudovirions before Western blot analysis of plant extracts

88
being assayed. One of two internal controls was serum from rabbit #19 (not previously shown), which was inoculated with insect cell-derived HPV-11 L1 NLS + VLPs. Dilutions of the conformation-specific and neutralising MAb H11:H3 were the second positive control.
No specific neutralising capacity was observed for any prebleed serum. However, while rabbit #19 serum (internal positive control; anti-insect cell-produced L1) neutralised the pseudovirion at the highest dilution tested (1:12150), and the known conformation-specific and neutralising MAb H11:H3 worked to a dilution of 1:10 6 , none of the post-inoculation day 56 sera were neutralising, even at the lowest dilution of 1:50.
Taken together, these serological results indicate that plant-derived HPV-11 L1 NLS -VLPs are modified to some extent, yet still retain certain overall virion morphological features as seen by electron microscopy. Immunisation of animals, however, results in an antibody response that is not capable of recognizing or neutralising HPV-11 L1 NLS + VLPs, the closest counterpart to the actual virus.

Discussion
The focus of this study was on the expression of a native HPV-11 L1 gene, as full-length and NLSconstructs, in transgenic A. thaliana and N. tabacum, with the overall aim of gaining some insight as to the feasibility of developing a plant-derived vaccine against HPV type 11. This is against a background of the release in 2006 of the GARDASIL™ yeast-derived quadrivalent vaccine against HPV types 6, 11, 16 and 18 by Merck: this vaccine has already shown 100% effectiveness in protecting from infection. However, the high cost of one course of this vaccine (3 doses; US$360) means that it will not prevent the continuous spread of the virus-associated disease in the foreseeable future, because it will remain largely nonaffordable to people in disadvantaged regions of the globe.
Here we report on the successful transformation of both A. thaliana and N. tabacum with the HPV-11 L1 gene missing the C-terminal nuclear localisation signal (NLS). This increases the number of plant systems used for the successful expression of HPV L1 proteins to include not only Immunogenicity of plant extracts Figure 5 Immunogenicity of plant extracts. Analysis of serum from New Zealand white rabbits #11 and #12 immunized with A. thaliana-derived HPV-11 L1 NLS -VLPs and rabbits #13 to #18 immunized with N. tabacum-derived HPV-11 L1 NLS -VLPs. Rabbits were immunized on days 1, 28 and 56 and serum from days 1, 15, 28, 42, 56 and 70 post injection was tested by ELISA in wells coated with 0.4 μg insect cell-derived HPV-11 L1 NLS -VLPs. Data from rabbit #12 is incomplete as it had to be euthanized due to growth of an abscess on the neck. Error bars represent the standard deviation calculated from triplicate analysis of samples. Absorbance at 405nm tobacco and potato, but now also Arabidopsis. Overall, our tobacco lines yielded ~2 μg of L1 protein per g of fresh leaf material, whereas up to 12 μg/g of L1 protein could be extracted from whole Arabidopsis plants. Our group has previously expressed unmodified CRPV L1 and HPV-16 L1 genes in transgenic tobacco to levels of 1 μg/g and 4 ng/ g of plant tissue respectively [21,17], and HPV-16 L1 via a tobamovirus vector in N. benthamiana to ~40 ng/g [18].
Biemelt et al. [16] expressed a human codon-optimised HPV-16 L1 gene in tobacco at a level of ~0.5% of the total soluble protein content (TSP), and in potato to about 0.2% TSP or ~12 μg/g of tuber. Warzecha et al. [19] expressed a plant codon-optimised HPV-11 L1 NLSgene in potato and reported extracting up to ~23 ng/g of fresh tuber. The HPV-16 L1 levels of expression have since been far surpassed by Maclean et al. [20]; this work reports a 500 -1000 fold increase in yield compared to previous reported expression of HPV-11 L1 [19].
We found that in planta expression of the wild-type HPV-11 L1 gene presented great difficulties and that it was effectively impossible to transform either one of our host plants. This stands in direct contrast to results obtained with the HPV-16 L1 gene, where expression of full-length L1 in either native (Varsani et al., 2003d) or human codon-optimised forms [16,20] was readily achieved. The latter studies concluded that the human codon-optimised HPV-16 L1 gene expressed better in plants than the plant codonoptimised or wild-type genes. Although HPV-11 and 16 L1 genes differ significantly, their L1 proteins are functionally very similar. Thus, the reasons for the difference in transformability remain obscure. It is also interesting in light of the HPV-16 results that our native HPV-11 L1 gene expressed so much better than the plant codon-optimised candidate of Warzecha et al. [19]: this indicates that expression of HPV L1 genes is best assessed empirically, rather than by assuming that gene optimisation according to plant codon usage will automatically improve expression.
The NLS of the HPV-11 L1 gene is not required for correct VLP assembly [22]. Resulting particles remain morphologically and antigenically identical to native virions despite removal of 21 aa from the C-terminus [19]. Therefore the truncated HPV-11 L1 protein should still be usa-ble as a vaccine and given the difficult and unpredictable in planta expression levels achieved with a spectrum of HPV L1 genes, we had decided to focus on maximizing the expression of the wild-type HPV-11 L1 NLSgene in our two expression hosts.
We were able to show successful transgene integration as well as stable inheritance in consecutive generations (Figure 1A and 1B). Transcription of the gene was investigated using RT-PCR: active transcription was shown throughout successive generations of both plant systems ( Figure 1C and 1D). One possible shortcoming of this approach however, given that Biemelt et al. [16] and Warzecha et al. [19] found by northern blot that their respective L1 mRNAs were somewhat degraded, was that such degradation was not detectable in our system.
We further investigated if the L1 monomers formed VLPs. We found that expression of the HPV-11 L1 NLSgene via recombinant baculovirus in Sf21 cells resulted in a range of different sized particles that did not resemble the classical shape ( Figure 3C). Furthermore, particles seen in both A. thaliana and N. tabacum protein extracts had shown the same pleomorphy, ranging from 20 to 60 nm in diameter ( Figure 3A and 3B). Li et al. [23], previously found that removal of 11 carboxyl-terminal amino acids still allowed for VLP assembly, but that particles were less uniform than those resulting from expression of the fulllength L1 gene. Warzecha et al. [19] found particles in HPV-11 L1-transformed potato but did not report pleomorphy as seen here.
Given several observations of differences between plantand insect cell-produced HPV-11 L1 proteins, we determined whether proteolytic degradation occurred for insect cell-derived L1 protein. Crude insect cell lysate containing HPV-11 L1 NLSprotein was left standing at 4°C for several hours before being examined by western blot: 3 distinct L1 proteolytic products were seen ( Figure 4C), of ~53, ~45 and ~36 kDa relative to the purified 55 kDa positive control. Li et al. [23] had reported that the HPV-11 L1 protein is susceptible to trypsin digestion, resulting in two distinctive products of ~42 and ~48 kDa in size. Our western blot analyses of insect cell and plant-derived HPV-11 L1 NLSproteins showed that the N. tabacumexpressed HPV-11 L1 NLSprotein is subject to severe degradation: digestion by a presumably trypsin-like protease resulted in a product of ~36 kDa ( Figure 4B). Unlike proteolysis in the crude insect cell lysate ( Figure 4C), no intermediate species were observed in the N. tabacum protein extract, suggesting that despite the addition of complete protease inhibitor throughout the extraction process, complete proteolysis of the L1 protein occurred over time.
Little is known about the properties of L1 proteins that have undergone this degree of proteolysis. Assuming they are incapable of assembly into capsids or of exhibiting virus-specific conformational epitopes could partially explain the ELISA results for tobacco-derived proteins shown in Figure 2B. However, particles were observed by electron microscopy ( Figure 3B). It could be that VLP assembly occurred during the concentration process, rather than in planta; and then only of a subset of L1 monomers that had not yet been completely proteolysed. Although analysis of the A. thaliana protein extract was anticipated to yield similar results, surprisingly no degradation was observed in the extracted HPV-11 L1NLSpro-tein ( Figure 4A). Furthermore, no significant alterations to the N. tabacum-derived HPV-16, the potato-derived HPV-11 L1 protein or of plant-derived CRPV L1 had been reported in previous studies [16][17][18][19] [21]. However, ours was the only study with HPV-11 L1 expressed in tobacco -and given the already proven unpredictability of L1 expression in plants, perhaps this is just more proof that empiricism is the only way to approach this subject.
The phenomenon of the HPV-11 L1 NLSproteins being severely degraded in N. tabacum plants but not in A. thaliana is interesting, and would undoubtedly impact on their vaccine potential. Given the evidence of proteolysis Antigenic specificity of immune sera Figure 6 Antigenic specificity of immune sera. Analysis of Day 56 antiserum from individual rabbits immunized with HPV-11 L1 NLSplant protein extract. Sera were tested against equal concentrations of denatured and non-denatured insect cell-derived HPV-11 L1 NLS + and NLS -VLPs at a concentration of 0.4 μg/well. Error bars represent the standard deviation calculated from triplicate analysis of samples. Insect cell-derived VLPs were denatured at 100°C for 10 min. Rabbits #11 and #12 were immunized with A. thaliana-derived HPV-11 L1 NLSantigen. Rabbits #13 and #14, #15 and #16, and #17 and #18 were inoculated with N. tabacum-derived protein extract from T 1 generation transgenic plant lines X1, Xh and Xi respectively. Absorbance at 405nm Antigen Preparation and particle pleomorphism, we investigated whether the antigenic properties of the plant-derived VLPs had been retained. It has been reported that removal of 86 aa from the HPV-11 L1 C-terminal end resulted in the formation of pentameric capsomers that were no longer able to establish inter-capsomeric contacts: however, the MAb H11:H3 capsid-neutralising domain was entirely contained within pentameric L1 capsomers and interpentamer associations were not required for the induction of virus-neutralising antibodies [24]. However, tobacco-produced VLPs did not bind the conformation-specific MAb H11:H3 ( Figure 2B). Another conformation-specific and neutralising MAb (H11:B2) also failed to bind the tobacco plant-derived protein. This is similar to the case of Chen et al. [25], where the formation of VLPs from severely truncated COPV L1 protein did not necessarily preserve conformation-specific epitopes on the surface of the capsid. In contrast, all Mabs, including the conformation-specific and neutralising Mabs, were capable of binding to the non-proteolysed A. thaliana-produced HPV-11 L1 NLS - (Figure 2A).
It is of utmost importance to establish whether any potential viral vaccine is capable of eliciting an appropriate antibody response. Accordingly, we evaluated the ability of the plant extracts to induce an antibody response in rabbits. With perhaps the exception of serum collected from rabbit #17 all immune sera reacted with insect cellderived NLS -VLPs ( Figures 5 and 6), and not with NLS + VLPs. This was unexpected, given that especially the A. thaliana-derived L1 protein reacted with Mabs specific for conformation-dependent epitopes. Additionally, serum collected on day 56 reacted exclusively with non-denatured insect cell-derived NLS -VLPs ( Figure 6). Taken together, these results present a major problem -both for interpretation and potentially for the prospect of HPV-11 plant-produced vaccines. We note that Warzecha et al. [19] found that their plant-derived material reacted with appropriate antibodies, but that no reaction could be shown by sera from orally immunized mice with insect cell-derived VLPs -presumably NLS + -unless the mice had been orally boosted with those VLPs. In contrast, it has been shown that injection of plant-produced HPV-16 and CRPV L1 elicited antibodies reactive with native VLPs produced in insect cells [16,17,20,21]. Thus, this phenomenon is apparently limited to the HPV-11 L1 NLSprotein.
Of more concern to the prospects of a plant-produced HPV-11 vaccine was the fact that none of the sera collected on day 56 could neutralise HPV-11 pseudovirions. While we have previously shown that rabbits immunized with plant-derived CRPV L1 protein were protected against live CRPV challenge in the absence of detectable neutralising antibodies [21], this is the only such result to our knowl-edge, and it is generally taken as given that the primary correlate of protection against papillomavirus infection is the presence of neutralising antibodies. We note too that Maclean et al. [20] were able to show pseudovirion neutralisation after immunisation of mice with plant-derived HPV-16 L1 VLPs.
We suggest that an immune response to plant-derived HPV-11 L1 NLS-is elicited mainly against an immunodominant epitope that appears to be surface exposed in VLPs. The nature of the structural changes that occur within the capsid that allow for recognition of this epitope, which is otherwise not surface exposed on the actual HPV-11 virion, need further investigation. Nevertheless, the lack of recognition by immune sera of the HPV-11 L1 NLS + VLPs, the closest counterpart to the actual virus, implies that antibodies generated from injection of this particular plant-derived HPV-11 L1 NLSprotein will not be able to prevent infection and disease, which completely negates their potential use for prophylactic vaccination.
However, we feel it would be a mistake to neglect further work on the development of an HPV-11 vaccine in plants.
With the possibility of using a plant-derived HPV-11 vaccine as cost-effective vaccine boosters, the goal would be to make it work effectively. It might be possible to do this by engineering the protein in a different way, or making HPV-11:HPV-16 hybrid L1s, for example. It will also be interesting to investigate if with new second and third generation plant expression vectors, expression of full length native HPV-11 L1 protein can be achieved, and if alternative codon optimisation or expression strategies will have an impact on expression levels, conformation and immunogenicity of this product.

Conclusion
In summary, we expressed the HPV 11 L1 NLSgene in two different plants, upon which we found that in A. thaliana VLPs were formed. These particles were recognised by conformation-specific antibodies and when analysed by western blot, a 55 kDa protein was observed. In contrast, however, HPV 11 L1 NLS -VLPs from N. tabacum were not recognized by the same conformation-specific antibodies even though VLPs were observed by EM. The protein seemed to be partially degraded as shown by the presence of a 36 kDa band in western blots. This work also showed that yields of HPV-11 L1 protein in plants could be increased by between 500 and 1000-fold compared to the only previous report [19], despite the use of a native HPV L1 gene rather than a plant codon-optimised version. Inoculation of rabbits with plant extract from both types of plants resulted in a weak immune response, and antisera did not react with native HPV-11 L1 VLPs, nor did they neutralise HPV-11 pseudovirion infectivity. As HPV-11 L1 NLS + VLPs are the closest counterpart to the actual virus it follows that antibodies generated from injection of our plant-derived HPV-11 L1 NLSprotein will not be able to prevent infection and disease. This has serious implications for the further development of plant-derived HPV-11 vaccines.

Cloning of HPV-11 L1 into Agrobacterium tumefaciens
Full-length and truncated HPV-11 DNA fragments derived from the L1 ORF were amplified from a HPV-11 DNApositive biopsy sample clone (A-L Williamson, University of Cape Town) using primer pairs 1 and 2 respectively (Table 1), cloned into the pMOSBLUE vector and sequenced. Primer pair 2 was used to exclude the bipartite C-terminal nuclear localisation signal (NLS; KRpavskp-stapKRKRTKTKK) spanning aa 482 -502. PCR was performed using standard cycling conditions, with an annealing temperature of 55°C. A silent mutation at position 257 (C to T) was found, otherwise this HPV-11 L1 sequence does not differ from the prototype HPV-11 L1 sequence (GenBank accession number M 14119).

Synthesis of HPV-11 L1 protein in insect cells
The full-length and NLS -L1 genes described above were cloned, using standard techniques, into the pFastBac vector (Invitrogen™), transformed into Escherichia coli DH10BAC competent cells, and recombinant baculovirus was generated by transfection of the bacmid DNA into Sf21 cells (Manufacturer's protocol, Invitrogen™, Carlsbad, California). HPV-11 L1 protein was produced in Sf21 cells and virus-like particles (VLPs) were prepared as described [17].

Transformation of Arabidopsis thaliana
A. thaliana ecotype Columbia plants were transformed with A. tumefaciens constructs (HPV-11 L1/HPV-11 L1 NLS -) by a simplified transformation protocol [28]. Primary bolts were cut off to induce proliferation of secondary bolts, then after 4 days plants were dipped in A. tumefaciens cultures containing either pART27 L1 or pART27 L1 NLS -. A. tumefaciens cultures were harvested after 2 days and resuspended to OD 600 = 0.8 in phosphate buffered saline (PBS, 1.47 mM KH 2 PO 4 , 10 mM Na 2 HPO 4 , 2.7 mM KCl, 137 mM NaCl, pH 7.4) containing 5% (w/v) sucrose and surfactant Silwet-77 at a concentration of 0.03% (v/v). Dipped plants were grown at 16/8 hr day/night photoperiod until seeds were harvested. To determine the transformation efficiency and to screen for transformed plants, dry seeds were sterilised in 10% bleach, washed 3 times in water and plated on plant nutrient sucrose medium (PNS) containing kanamycin (250 μg/ml). Germination was allowed to take place under a 16/8 hr photoperiod, at 25°C and 80% relative humidity. Seedlings were left on PNS media until complete development of a mature 4 th leaf, before being transplanted to soil.

Transformation of Nicotiana tabacum
The transformation protocol of Horsch et al. [29] was used for the generation of transgenic N. tabacum plants. Sterilised leaf discs (±1 cm 2 ) were dipped into a 48 hr A. tumefaciens culture carrying either the HPV-11 L1 or HPV-11 L1 NLSconstruct. Squares were transferred to non-selective co-cultivation media [30] for 72 hrs, follwed by MS regeneration/shooting media with kanamycin (300 μg/ml) and cefotaxim (250 μg/ml) until small shoots appeared from the calli. Shoots with 1 to 2 leaves were transferred to MS root-inducing media with kanamycin (100 μg/ml). Once roots were formed, plants were transferred to soil and grown to maturity with a 16/8 hr photoperiod at 25°C and 80% relative humidity. Flowering plants were selfpollinated and dry seeds screened on media containing kanamycin (250 μg/ml) before putative transgenic seedlings were transferred to soil, once the 4 th leaves had grown to maturity.

Screening of plants for the HPV-11 L1 and L1 NLSgenes
Plant genomic DNA was extracted from putative transgenic and wild type A. thaliana and N. tabacum plants using the Dellaporta method [31] and screened by PCR for the HPV-11 L1 and HPV-11 L1 NLSgenes using primer pairs 3 and 4 ( Table 1).

Analysis of total RNA extracted from transgenic plants
Total RNA was extracted from fresh leaves using TRIzol™ reagent (Life Technologies). Detection of the L1 NLS -mRNA was achieved by RT-PCR using the Access RT-PCR system (Promega). Primer pair 5 (Table 1) was used to amplify an internal 300 base pair fragment according to the manufacture's instruction.

Processing and concentration of transgenic plant material
Entire A. thaliana plants and leaves from N. tabacum were harvested and homogenised in 1:2 (w/v) cold 0.1 m phos-