• Open Access

Immunogenic assessment of plant-produced human papillomavirus type 16 L1/L2 chimaeras



Cervical cancer is caused by infection with human papillomaviruses (HPV) and is a global concern, particularly in developing countries, which have ~80% of the burden. HPV L1 virus-like particle (VLP) type–restricted vaccines prevent new infections and associated disease. However, their high cost has limited their application, and cytological screening programmes are still required to detect malignant lesions associated with the nonvaccine types. Thus, there is an urgent need for cheap second-generation HPV vaccines that protect against multiple types. The objective of this study was to express novel HPV-16 L1-based chimaeras, containing cross-protective epitopes from the L2 minor capsid protein, in tobacco plants. These L1/L2 chimaeras contained epitope sequences derived from HPV-16 L2 amino acid 108–120, 56–81 or 17–36 substituted into the C-terminal helix 4 (h4) region of L1 from amino acid 414. All chimaeras were expressed in Nicotiana benthamiana via an Agrobacterium-mediated transient system and targeted to chloroplasts. The chimaeras were highly expressed with yields of ~1.2 g/kg plant tissue; however, they assembled differently, indicating that the length and nature of the L2 epitope affect VLP assembly. The chimaera containing L2 amino acids 108–120 was the most successful candidate vaccine. It assembled into small VLPs and elicited anti-L1 and anti-L2 responses in mice, and antisera neutralized homologous HPV-16 and heterologous HPV-52 pseudovirions. The other chimaeras predominantly assembled into capsomeres and other aggregates and elicited weaker humoral immune responses, demonstrating the importance of VLP assembly for the immunogenicity of candidate vaccines.


Cervical cancer is the third most common cancer among women worldwide (Ferlay et al., 2010), and the causal association between human papillomavirus (HPV) infection and cervical cancer has been well described (zur Hausen et al., 1981). HPV has also been established as a cause of cancer of the penis, vulva, anus, vagina and oropharynx (Münger et al., 2004); therefore, HPV vaccine development is a priority for preventative cancer research.

Human papillomaviruses are small, nonenveloped, double-stranded DNA viruses that infect human squamous and cutaneous epithelial cells (zur Hausen, 2000; Lin et al., 2010). The high-risk types HPV-16 and HPV-18 are the most prevalent genotypes in invasive cervical cancers, accounting for approximately 70% of diagnosed cervical cancers worldwide (Smith et al., 2007). Although their hierarchy may differ geographically, the six next most common oncogenic HPV types are HPV-31, HPV-33, HPV-35, HPV-45, HPV-52 and HPV-58, as consistently reported by the global IARC pooled analysis (Muñoz et al., 2004), meta-analyses (Clifford et al., 2003; Li et al., 2011) and a recent cross-sectional worldwide study (de Sanjosé et al., 2010).

The viral capsid is arranged in a T = 7 icosahedral lattice and contains two structural proteins: L1 (major capsid protein) and L2 (minor capsid protein). The capsid comprises of 72 pentameric capsomeres, assembled from 360 copies of L1 and associated with 36–72 copies of L2 (Buck et al., 2008). L1 spontaneously self-assembles into virus-like particles (VLPs), similar morphologically and immunogenically to native virions (Kirnbauer et al., 1992). Immunization with VLPs in animal models elicits high titres of neutralizing antibodies (NAb), which are protective and are predominantly directed against L1 type-specific conformational epitopes (Christensen and Kreider, 1990; Kirnbauer et al., 1992; Rose et al., 1994). As a result, the L1 major capsid protein is the antigen of choice in the development of prophylactic vaccines.

Other higher-order structures are also immunogenic. Immunization with ‘small’ T = 1 VLPs (30–40 nm) elicits antibody titres comparable to full-sized T = 7 VLPs (Schädlich et al., 2009). Furthermore, L1 capsomere-only vaccines elicit both NAb and T-cell responses (Ohlschläger et al., 2003; Thönes et al., 2008) and protect against viral challenge in animal models (Rose et al., 1998; Yuan et al., 2001). Although capsomeres elicit 20- to 40-fold lower humoral immune responses in comparison with VLPs, this can be addressed by the use of an adjuvant (Jagu et al., 2010).

Two multivalent HPV L1 VLP-based prophylactic vaccines have been licensed and are highly effective in the prevention of vaccine-type infections and associated disease (Future, II Study Group, 2007). Gardasil® (Merck & Co., Inc.) contains L1 VLPs of low-risk genital wart types 6 and 11 and high-cancer-risk types 16 and 18, produced in Saccharomyces cerevisiae. Cervarix™ (GlaxoSmithKline Biologicals) contains L1 VLPs from types 16 and 18, produced via recombinant baculoviruses in insect cells (Schiller et al., 2008). However, their type restriction and high cost of vaccines have limited their widespread application, particularly in developing countries (Muñoz et al., 2004; Parkin and Bray, 2006). Therefore, there is an urgent need for affordable second-generation HPV vaccines that broaden protection to include multiple oncogenic HPV types.

Plant expression systems present a cost-effective alternative for vaccine production. Although it is now generally accepted that plant-derived vaccines will also need to be processed and reproducibly formulated, estimates suggest that generic production costs could be reduced by at least 31% (Rybicki, 2009). Plants are easier to cultivate than mammalian cell cultures, allow expression of recombinant proteins free of contamination by bacterial toxins or human pathogens and allow appropriate subcellular targeting, proper folding and most post-translational modifications (Fischer et al., 2004).

Several groups have expressed HPV-16 L1 and L1-based VLPs in plants using transient and transgenic plant expression systems (reviewed by Giorgi et al., 2010). Plant-derived L1s self-assemble into higher-order structures and are immunogenic and protective in animal models (Kohl et al., 2006), thus demonstrating proof of efficacy. Transient expression systems are particularly useful for the rapid production of antigens (Rybicki, 2010), and significantly higher protein levels have been obtained in comparison with stable nuclear transformation (Varsani et al., 2003b). Furthermore, low yields of recombinant protein, potentially a result of protein instability or low-level expression, have been effectively addressed by several different strategies: human codon optimization of the gene—possibly due to the disruption of transcription and RNA processing inhibitory sequences in the HPV-16 L1 gene (Collier et al., 2002; Maclean et al., 2007), targeting the proteins to intracellular organelles (Maclean et al., 2007) and the use of a bean yellow dwarf geminivirus (BeYDV)-derived DNA replicon system (Regnard et al., 2010).

The L2 minor capsid protein has emerged as a candidate for the development of prophylactic HPV vaccines (Alphs et al., 2008; Gambhira et al., 2006, 2007a): the N-terminus of L2 is highly conserved and contains several epitopes, which are cross-neutralizing (Kawana et al., 1999, 2001, 2003; Kondo et al., 2007, 2008; Schellenbacher et al., 2009) and which are protective in vivo (Embers et al., 2002). Although L2 is immunogenically subdominant to L1 and lacks exposure on the surface of mature capsids (Buck et al., 2008; Roden et al., 2000), epitope display on the surface of VLPs or capsomeres may improve the immunogenicity of L2 regions and broaden the protection of L1-based vaccines.

Large regions of the HPV L1 C-terminal are surface-exposed and antigenic, particularly the loop containing residues 420–429 (Modis et al., 2002; Murata et al., 2009). Several groups have successfully inserted L2 epitopes into the core sequences of the HPV-16 L1 (McGrath et al., 2013; Slupetzky et al., 2001; Varsani et al., 2003a), and chimaeras containing HPV-16 L2 epitopes amino acids 56–81 and amino acids 17–36 (McGrath et al., 2013), and amino acids 108–120 (Varsani et al., 2003a), replacing the L1 helix 4 (h4) region, have elicited both anti-L1 and anti-L2 immune responses, suggesting that this insertion site has potential for epitope display. Furthermore, chimaeric HPV-16 L1 containing M2e influenza epitopes in a similar helix 4 region was successfully expressed in plants and self-assembled into cVLPs and capsomeres (Matić et al., 2011).

We report here the Agrobacterium-mediated transient expression of three L1/L2 chimaeras in Nicotiana benthamiana. The human codon-optimized chimaeras contained the cross-neutralizing HPV-16 L2 epitope amino acids 108–120 (Kawana et al., 1999; similar to ChiΔF-L2 in Varsani et al., 2003a), amino acids 56–81 (Kawana et al., 1998) or amino acids 17–36 (Gambhira et al., 2007b), in the helix 4 sequence of HPV-16 L1. Expression was optimized either by targeting the protein to the chloroplast or by utilizing a self-replicative BeYDV-based vector, strategies that have previously produced high yields of HPV-16 L1 in plants but have not been comparatively analysed. Mice were immunized with three L1/L2 chimaeras, and the elicited anti-L1 and anti-L2 antibodies were analysed with respect to cross-neutralizing ability. The degree of chimaera assembly was also correlated with immunogenicity.


Optimization of L1/L2 chimaera expression

The schematic design of the three HPV-16 L1/L2 chimaeras is shown in Figure 1. The chimaera backbone consists of South African HPV-16 L1 (GenBank: DQ067889) with an L2 epitope replacing the L1 C-terminal h4 helix at amino acid 414 (Varsani et al., 2003a). The three HPV-16 L1 chimaeras contained HPV-16 L2 sequences from amino acids 108–120, 56–81 or 17–36, replacing 13, 26 and 20 residues, respectively. The chimaera genes were human codon-optimized and synthesized by GENEART AG (Regensburg, Germany).

Figure 1.

The amino acid location of the L2 epitopes in the HPV-16 L1/L2 chimaeras. The L2 epitopes replacing the sequence from amino acid 414 are indicated as white bars.

To optimize yields, L1/L2 chimaera expression in N. benthamiana was initially examined in 1–9 days postinfiltration (dpi) time trials, for all the plant expression vectors, and analysed by Western blotting. Chimaeras were expressed either with or without NSs, a Tomato spotted wilt virus (TSWV) RNA silencing suppressor protein, which inhibits the onset of post-transcriptional gene silencing (PTGS)—a plant defence mechanism against viral infections (Takeda et al., 2002). Expression of the L1/L2 chimaeras peaked at 5 days postinfiltration (dpi), and expression with pTRAkc-rbcs1-cTP was both improved and prolonged when co-expressed with NSs (data not shown), suggesting NSs was effective in preventing post-transcriptional gene silencing and enhancing protein accumulation in plants, and as a result, all chimaeras were co-expressed with NSs in further comparative and quantitative experiments.

Comparative vector expression of L1/L2 chimaeras

Expression using the three expression vectors (pTRAc, pTRAkc-rbcs1-cTP and pRIC3) was directly compared by Western blotting (Figure 2a) and capture ELISA (Figure 2b) at 5 dpi. The pTRA vectors are nonreplicative plant expression vectors, which target the protein to the cytoplasm (pTRAc) or chloroplast (pTRAkc-rbsc-cTP), while pRIC3 is a self-replicative vector derived from bean yellow dwarf (BeYDV) mastrevirus, which targets protein to the cytoplasm. Plant-expressed HPV-16 L1 produced via pTRAc and pTRAkc-rbcs1-cTP was used as a positive control, and NSs-infiltrated plants were used as the negative control.

Figure 2.

Comparative expression of L1/L2 chimaeras using the plant expression vectors pTRAc, pTRAkcrbcs1-cTP and pRIC3. Chimaeras were co-expressed with NSs, extracted 5 days postinfiltration (dpi) and detected with CamVir1. HPV-16 L1 was expressed as a positive expression control for pTRAc and pTRAkc-rbcs1-cTP, and the negative expression control was NSs-infiltrated plants. (a) Western blot detection of the L1/L2 chimaeras. M, molecular weight marker. The black arrows indicate the L1/L2 chimaeras or HPV-16 L1 (~56 kDa). Equal volumes of sample were loaded. (b) The average HPV protein yield was determined by capture ELISA (n = 3). Yields that are significantly different at P = 0.01 are denoted with a star and were normalized according to TSP. The error bars indicate the standard deviation.

Expression with pTRAkc-rbcs1-cTP consistently gave the highest chimaera yields—with chimaeras presumed to be localized in the chloroplast due to the signal sequence—followed by expression with pRIC3 and then pTRAc (represented in Figure 2a). Chloroplast-targeted chimaeras produced yields of 1040–1310 mg/kg (2–3% total soluble protein, TSP), improving chimaera expression by up to 28-fold in comparison with the cytoplasm-targeting vector pTRAc (50–260 mg/kg; <1% TSP) and up to sevenfold in comparison with the self-replicative pRIC3 vector (190–660 mg/kg; <1% TSP). The expression yields obtained using pTRAkc-rbcs1-cTP were significantly higher than using pTRAc for all the L1/L2 chimaeras and HPV-16 L1 (Figure 2a) and significantly higher than using pRIC3 for L1/L2 (56–81). Although chloroplast-targeted HPV-16 L1 demonstrated the highest average yields (1710 mg/kg, 4% TSP), the differences in pTRAkc-rbcs1-cTP expression between the L1/L2 chimaeras and native L1 were not statistically significant (> 0.01).

Antigen enrichment using heparin chromatography

Human papillomavirus L1 and L1/L2 proteins were enriched from the crude plant supernatant by heparin chromatography. Heparin reversibly bound L1 and L1/L2 chimaeras in a similar manner, and the elution gradient was optimized using an initial linear 0–100% 1.5 m NaCl gradient (Figure S1). All HPV antigens eluted between 0.45 and 0.75 m NaCl (comparable to Bazan et al., 2009; Kim et al., 2010 and Baek et al., 2011); thus, a 50% (0.75 m NaCl) step gradient was used to enrich the antigens for the mouse immunogenicity study (Figure S1). Detection of the HPV antigens in the eluate fractions was by CamVir1 dot blots, and fractions containing the partially purified HPV antigens were pooled and desalted using ultrafiltration spin columns.

To determine the effect of chromatography on the enrichment of HPV antigens, the ELISA-detected L1 or L1/L2 yield was directly compared with the TSP yield for both the crude and purified samples (Figure 3) and confirmed by Coomassie-staining and Western blot analysis (Figure S2). Chromatography reduced both the TSP and total HPV protein, as expected. However, relative to TSP, there is up to a fivefold enrichment of HPV antigen in purified samples, suggesting that heparin chromatography is effective in removing a large proportion of contaminating protein. Full-length L1 and L1/L2 antigens were then quantified by Western blot densitometry using the commercial vaccine Cervarix as a standard (see Figure S3).

Figure 3.

Purification and antigen enrichment by heparin chromatography. The antigen yields relative to TSP in both the crude and the partially purified samples were used to determine antigen enrichment. TSP was determined using the Lowry assay, and HPV protein was detected with H16.J4 (linear epitope-specific MAb). The error bars indicate the standard deviation.

Electron microscopy of vaccine antigens

The assembly of the partially purified L1 and L1/L2 chimaeras after heparin affinity chromatography was analysed by immunocapture transmission electron microscopy (Figure 4). In comparison with the negative control (NSs-infiltrated plant extract), all the HPV samples appeared to contain tertiary structures: these were either capsomeres (~10 nm), capsomere aggregates, small VLPs (~25 to 40 nm) or full-sized VLPs (~50 nm). Enriched L1/L2 (108–120) assembled into small ‘chimaeric VLPs’ (cVLPs), which were regular in shape but varied in size, while L1/L2 (56–81) samples appeared to contain only capsomeres and some aggregates. L1/L2 (17–36) contained a mixed population of amorphous protein aggregates. HPV-L1 assembled into distinct VLPs (~50 nm), as described in our previous study (Maclean et al., 2007).

Figure 4.

Transmission electron micrographs of enriched vaccine antigens following heparin affinity chromatography. L1/L2 (108–120), L1/L2 (56–81), L1/L2 (17–36), HPV-16 L1 (positive control) and NSs plant extract (negative control) were immunotrapped with CamVir1 (1 : 1000) onto carbon-coated copper grids and viewed on a Zeiss 912 Omega Cryo EFTEM. Left scale bar = 200 nm, right scale bar = 50 nm. White arrows indicate HPV-16 capsomeres (~10 nm), grey arrows represent capsomere aggregates, and black arrows indicate small VLPs (25–40 nm) and full-sized VLPs (50–55 nm).

Antibody responses elicited by L1/L2 chimaeras

To analyse the humoral immunogenicity of the L1/L2 chimaeras in comparison with type-restricted HPV-16 L1, mice were subcutaneously injected with 10 μg of plant-derived antigen in Freund's incomplete adjuvant and boosted on four occasions. Indirect ELISA was used to detect antibodies elicited against HPV-16 L1, using insect cell-expressed HPV-16 L1 as the coating antigen (Figure 5a). Responses were regarded as positive if ELISA values were significantly different from their corresponding prebleeds and the NSs-infiltrated plant extract (= 0.01). No anti-L1 response was detected for L1/L2 (56–81) and the NSs-infiltrated negative control. In comparison, the plant-derived L1 positive control showed a good response, and both the plant-derived L1/L2 (17–36) and L1/L2 (108–120) chimaeras elicited anti-L1 titres of 200 and 12 800, respectively. Although HPV-16 L1 elicited the highest anti-L1 titres (12 800–51 200), L1/L2 (108–120) showed a similar response, suggesting the insertion of the L2 amino acids 108–120 epitope had less of an effect on L1 immunogenicity in comparison with the other chimaeras.

Figure 5.

Anti-L1 and anti-L2 responses in mouse sera. (a) Direct ELISA detection and titration of anti-L1 in pooled mouse sera, for vaccine L1/L2 (108–120), L1/L2 (56–81), L1/L2 (17–36), HPV L1 (+ve vaccine control) and NSs plant extract (−ve vaccine control). Insect cell-produced HPV-16 L1 was used as the coating antigen. All prebleed absorbance values were <0.05 at a 1 : 50 dilution. Markers represent the mean value of triplicate samples, and error bars indicate the standard deviation. (b) Western blot detection of anti-L2 responses in pooled mouse sera (1 : 100), using E. coli-expressed His-tagged HPV-16 L2 as the antigen. M, molecular weight marker. PB, prebleed sera. FB, final bleed sera. For the Western blot controls: +ve, mouse anti-His (1 : 2000; Serotec); −ve, no primary antibody. The black arrow indicates L2 (~80 kDa).

The anti-L2 response was determined using Western blotting, with the E. coli-produced His-tagged HPV-16 L2 protein as the target antigen (Figure 5b). A nonspecific reaction at a position similar to the ~80-kDa L2 band was detected in both the antisera from the negative vaccine control (NSs-infiltrated plant extract) and HPV-16 L1, which serves as an additional negative L2 control. All chimaera vaccines appeared to give an anti-L2 response; however, only the L1/L2 (108–120) and L1/L2 (17–36) chimaeras gave a definite response, with bands >2X intensity of those in the HPV-16 L1 control.

Cross-neutralizing responses elicited by L1/L2 chimaeras

The ability of the chimaeras to elicit cross-neutralizing antibodies was analysed using PsV neutralization assays. Pooled gradient fractions containing high concentrations of PsVs were examined by TEM (Figure S4) and subsequently used in the neutralization assays. Sera from mice immunized with plant-produced HPV-16 L1 and L1/L2 chimaeras were tested for homologous neutralization of HPV-16 PsVs and heterologous cross-protection against HPV-18, HPV-45 and HPV-52 PsVs. The neutralization titre was defined as the highest dilution of serum, which reduced SEAP activity by >50% in comparison with the control untreated sample.

Anti-HPV-16 L1 sera strongly neutralized HPV-16 PsVs, and anti-L1/anti-L2 (108–120) displayed a similar neutralization curve (Figure 6). Both L1/L2 (56–81) and L1/L2 (17–36) did not appear to elicit HPV-16 NAb, as antisera showed similar neutralization curves to the negative control. None of the L1/L2 chimaera vaccines elicited significant HPV-18 or HPV-45 NAb titres, with L1/L2 (56–81) and L1/L2 (17–36) antisera displaying similar neutralizing curves to HPV-16 L1 and the NSs-infiltrated extract sera (data not shown). However, L1/L2 (108–120) sera had HPV-52 PsV neutralizing activity at low reciprocal dilutions (50–200), reducing SEAP levels by >50% in comparison with the unneutralized HPV-52 PsV control (Figure S5).

Figure 6.

HPV-16 PsV neutralization assay for (a) L1/L2 (108–120), L1/L2 (56–81), L1/L2 (17–36), HPV-16 L1 (+ve vaccine control) and NSs-infiltrated plant extract (−ve vaccine control) using pooled sera from vaccinated mice. Infection and SEAP expression controls include the cell control (−ve control represents 100% neutralization) and the PsV control (+ve control represents 0% neutralization), as well as (b) the MAb H16.V5 as the +ve neutralization control.

Table 1 summarizes the HPV-16, HPV-18, HPV-45 and HPV-52 PsV neutralization antibody titres elicited by the plant-derived vaccines. L1/L2 (108–120) elicited homologous HPV-16 NAb, and the antisera cross-neutralized heterologous HPV-52 PsV, suggesting this vaccine has the most potential for protection. L1/L2 (17–36) chimaeras elicited anti-L1 titres, but homologous HPV-16 NAb were not detected, suggesting the immunogenicity against HPV-16 L1 may be compromised. L1/L2 (56–81) did not elicit anti-L1, anti-L2 or NAb and thus does not appear to have potential as a prophylactic vaccine. None of the HPV vaccines elicited detectable cross-neutralizing antibodies against phylogenitically related HPV types 18 and 45.

Table 1. Summary of the neutralization titres for plant-derived L1 and the L1/L2 chimaera candidate vaccines
VaccineChimaeraPsV neutralization assay titresa
  1. a

    The reciprocal of the highest dilution which reduced SEAP activity by at least 50%.

V1L1/L2 (108–120)50–5000–500–5050–200
V2L1/L2 (36–58)0–500–500–500–50
V3L1/L2 (17–36)0–500–500–5050–200
V4HPV-16 L1500–50000–500–500–50
V5Plant extract0–500–500–500–50
+ControlH16.V52 × 105–2 × 106   
α-CamVir1 12 800–51 200  
H45.N5  3200–12 800 
H52.C1   2 × 104–2 × 105
H52.D11   2 × 105–2 × 106


There is a strong financial and practical incentive to develop cheaper second-generation HPV vaccines, which are broadly protective against multiple oncogenic types. This study examined the plant-based transient expression of four HPV-16 L1-based chimaeras containing cross-neutralizing L2 epitopes and their immunogenicity in mice.

Expression of L1/L2 chimaeras in plants

The highest HPV-16 L1 yields obtained thus far in an Agrobacterium-mediated transient expression system were achieved by human codon optimization of the L1 gene and by either targeting the expressed protein to the chloroplast (Maclean et al., 2007) or using an agroinfiltration-delivered self-replicative BeYDV-based vector (Regnard et al., 2010). This is the first study that directly compares these strategies using human codon-optimized L1/L2 chimaeras, with results suggesting that chloroplast targeting is the most effective strategy for high-level accumulation. Although the self-replicative vector pRIC3 increased the accumulation of L1/L2 chimaeras in the cytoplasm, chimaera expression using the chloroplast-targeting pTRAkc-rbcs1-cTP vector significantly improved yields when compared to the cytoplasm-targeting pTRAc vector.

The three chloroplast-targeted L1/L2 chimaeras consistently produced high expression yields (1000–1300 mg/kg plant tissue), which were twofold higher than the maximum HPV-16 L1 yields reported in similar plant expression studies (530–550 mg/kg; Maclean et al., 2007; Regnard et al., 2010). Although HPV-16 L1 (positive control) expression gave higher yields, differences were not statistically significant, suggesting that the L2 epitope substitutions do not affect the expression and accumulation of HPV L1 chimaeras.

Immunogenicity of the L1/L2 chimaeras

Several groups have demonstrated that plant-derived HPV-16 L1 and L1-based chimaeras are immunogenic and induce the production of NAb (Fernández-San Millan et al., 2008; Maclean et al., 2007; Paz De la Rosa et al., 2009), which is widely considered the gold standard for demonstrating the potential of prophylactic HPV candidate vaccines (Rybicki, 2010). Insect cell-expressed L1/L2 (108–120) (Varsani et al., 2003a), or L2 amino acids 17–36, 18–31, 35–75, 69–81, 75–112, 108–120 and 115–154 inserted into the BPV-1 L1 DE surface loop (amino acids 133–134), elicited anti-L1 and anti-L2 responses in mice (Schellenbacher et al., 2009; Slupetzky et al., 2007), suggesting L1/L2 chimaeras may be an effective strategy to broaden the protection of HPV prophylactic vaccines. As a result, the humoral immunogenicity of plant-derived HPV-16 L1 and three L1/L2 chimaeras was investigated in mice.

Plant-derived HPV-16 L1 elicited the highest anti-L1 titres in mice, followed by L1/L2 (108–120) and L1/L2 (17–36). Furthermore, L1/L2 (108–120) and L1/L2 (17–36) demonstrated anti-L2 responses, suggesting the L2 peptides were effectively displayed on the surface of the chimaeric protein. L1/L2 (56–81) did not elicit a detectable anti-L1 response, and the anti-L2 response was inconclusive, with responses similar to the negative controls. As a result, we believe L1/L2 (56–81) does not have potential for further development as a HPV vaccine.

The L1/L2 chimaeras were examined for their ability to elicit neutralizing antibodies to HPV-16, HPV-18, HPV-45 and HPV-52 PsVs. All the L2 epitopes analysed in this study have previously been shown by other groups to elicit HPV-16 neutralizing and HPV-52 cross-neutralizing antibodies. Furthermore, L2 amino acids 56–81 sera cross-neutralize HPV-18, and L2 amino acids 17–36 sera cross-neutralize both HPV-18 and HPV-45 (Alphs et al., 2008; Gambhira et al., 2007b; Kawana et al., 2003; Kondo et al., 2007, 2008; Rubio et al., 2009; Schellenbacher et al., 2009; Slupetzky et al., 2007).

Both L1 and L1/L2 (108–120) elicited NAb against the homologous HPV-16, the most prevalent type detected in invasive cervical cancers worldwide (de Sanjosé et al., 2010). The titres were comparable to other L1/L2 vaccines with similar epitopes from other expression systems (Schellenbacher et al., 2009; Slupetzky et al., 2007) and to other immunogenicity studies using plant-derived HPV L1 antigens (Fernández-San Millán et al., 2008; Maclean et al., 2007; Paz De la Rosa et al., 2009).

Although neutralizing activity against phylogenetically related HPV-18 and HPV-45 PsV was not detected for any of our vaccines, L1/L2 (108–120) also elicited low titres of HPV-52 NAb, as described by other studies (Kawana et al., 2003; Kondo et al., 2008; Schellenbacher et al., 2009). This result is particularly significant in a local context, as HPV-52 is highly prevalent in low- and high-grade cervical lesions in South African women (Allan et al., 2008). Overall, L1/L2 (108–120) appears to be the best candidate vaccine as it elicited high anti-L1 titres, elicited an anti-L2 response and was the only chimaera to elicit both HPV-16 NAb and sera that cross-neutralized heterologous HPV-52, unlike the type-specific L1 vaccine.

Purification of L1/L2 chimaeras

The use of simple and efficient purification processes is a high priority in any commercial protein production system. Although size-dependent purification methods have traditionally been used to purify plant-expressed HPV L1 (Biemelt et al., 2003; Fernández-San Millán et al., 2008; Maclean et al., 2007), these methods were inefficient and nonreproducible for the L1/L2 chimaeras, as the L1/L2 chimaeras assembled into a variety of heterologous higher-ordered structures (Figure 4).

In this study, heparin chromatography was used to enrich the plant-expressed L1 and L1/L2 chimaeras. Heparin selectively purifies assembled L1 (Giroglou et al., 2001; Rommel et al., 2005) by binding to a conformational motif, which is not present on the C-terminal of L1 (Fleury et al., 2009) and thus is not directly affected by the L2 sequence replacements. Recent studies have shown the recovery, yield and purity of HPV L1 protein are improved using heparin chromatography (Kim et al., 2010), and to our knowledge, this is the first report of plant-expressed HPV particles purified using this strategy.

Heparin similarly bound and eluted the L1/L2 chimaeras (Figure S1), regardless of their differential assembly patterns and successfully enriched the antigens (Figure 3). However, the antigens were only partially purified (Figure S2) and contained contaminating proteins, as also described by Kim et al. (2010). As a result, a single-step method using heparin chromatography is not sufficient to obtain highly purified HPV L1 and L1/L2 chimaeras produced in plants.

Assembly of L1/L2 chimaeras

Plant-expressed HPV-16 L1 assembled into VLPs (~50 nm), which were similar in size and morphology to those expressed in other studies. Plant-expressed HPV-16 L1 VLPs are typically 30–65 nm in size (Biemelt et al., 2003; Liu et al., 2005; Maclean et al., 2007; Varsani et al., 2003b, 2006a), with the majority of VLPs reported as 50–60 nm in diameter when localized to or produced in tobacco chloroplasts (Fernández-San Millán et al., 2008; Lenzi et al., 2008; Maclean et al., 2007).

The chimaeras L1/L2 (108–120), L1/L2 (56–81) and L1/L2 (17–36), containing L2 epitope sequence replacements of 13, 26 and 20 residues, respectively, assembled into a variety of heterologous higher-ordered structures prior to heparin affinity purification (data not shown) and maintained their structure after purification (Figure 4). Chimaera L1/L2 (56–81), with the longest L2 epitope, predominantly assembled into capsomeres, with no VLP-like structures observed. In contrast, L1/L2 (17–36) predominantly formed protein aggregates, although the presence of larger amorphous VLP-like structures suggests there may be partial assembly of VLPs. L1/L2 (108–120), with the shortest L2 epitope, successfully assembled into regular chimaeric VLPs (cVLPs), in contrast to a similar chimaera expressed in insect cells (Varsani et al., 2003a)—possibly due to subtle differences in viral assembly chaperones such as heat shock protein 70 (Hsp70) and karyopherins (Sullivan and Pipas, 2001).

A recent study of plant-expressed HPV-16 L1 chimaeras containing influenza virus type A M2e epitopes supports the notion that epitope length and sequence influence VLP assembly (Matić et al., 2011). Although insertion of epitopes into h4 did not hinder capsomere formation, the chimaera containing a longer 23-residue epitope assembled into cVLPs and was more effective at displaying the epitope than a similar chimaera containing a smaller 8-residue epitope. These results provide further evidence that the h4 region plays a role in VLP assembly and suggest that epitopes located at amino acid 414 have a significant impact on assembly.

Differences in the structural assembly of the L1/L2 chimaeras could be explained by the presence of a disulphide bond between the highly conserved cysteine residues 175 and 428 in VLPs. Mutations of these L1 cysteines result in the formation of capsomeres rather than VLPs (Fligge et al., 2001; Li et al., 1998; McCarthy et al., 1998; Sapp et al., 1998; Varsani et al., 2006b), and thus, L1/L2 (108–120), with the epitope located at amino acids 414–426, does not replace Cys 428, allowing VLP assembly. The deletion of L1 residues 428–465 eliminates assembly of capsomeres into VLPs (Varsani et al., 2006b); thus, L1/L2 (17–36) and L1/L2 (56–81), which contain L2 sequence replacements overlapping this region (amino acids 414–433 and 414–439, respectively), do not assemble into VLPs. Another consideration is the epitope sequence and, in particular, the addition of cysteine residues to the L1 backbone via the inserted L2 amino acids 17–36 epitope. This L2 epitope contains two cysteine residues (Cys 22 and Cys 28), which may have formed disulphide bonds with Cys 175 in the L1 backbone, and may account for the partial assembly of L1/L2 (17–36) cVLPs (Figure 4). Further work is required to determine whether the insertion of cysteine residues affects (and potentially restores) cVLP assembly.

Link between assembly of L1/L2 and immunogenicity

This study provides further evidence of the correlation between VLP assembly and L1 immunogenicity (Thönes et al., 2008). L1 VLPs and L1/L2 (108–120) cVLPs elicited higher anti-L1 responses and HPV-16 NAb compared with the other chimaeras, which assembled into capsomeres and aggregates (Figure 4). Furthermore, only L1/L2 (108–120) cVLPs elicited detectable levels of HPV-16 and HPV-52 NAb. Although antisera to the L2 amino acids 17–36 and L2 amino acids 56–81 epitopes have been shown to neutralize HPV-16, HPV-18, HPV-45 and HPV-52 PsVs (Alphs et al., 2008; Gambhira et al., 2007b; Kondo et al., 2007, 2008; Rubio et al., 2009; Schellenbacher et al., 2009), the partial assembly of the plant-derived L1/L2 chimaeras may have affected L2 immunogenicity.

Other studies have shown that L2 peptides fused to keyhole limpet haemocyanin (KLH) elicit lower HPV-16 and HPV-18 NAb titres compared with those displayed on the surface of HPV-16 L1 cVLPs (Kondo et al., 2007, 2008), and denaturation of L1/L2 chimaeras containing L2 amino acids 17–36 reduces both HPV-16 and HPV-18, HPV-45 and HPV-52 NAb titres to the extent that HPV-45 and HPV-52 NAb were not detected in rabbit antisera (Schellenbacher et al., 2009). As a result, there is strong evidence that assembly of L1/L2 chimaeras into cVLPs elicits stronger immune responses than capsomeres. Assembly into cVLPs may enhance both the L1 and L2 immunity, and stronger L2 responses may improve the cross-protection potential of the L1/L2 chimaera vaccines.

In summary, three HPV-16 L1/L2 chimaeras were expressed in N. benthamiana using Agrobacterium-mediated transient expression systems. Although the self-replicative BeYDV-derived vector improved yields in the cytoplasm, maximum yields were obtained by targeting the protein to the chloroplast. We anticipate that the expression of L1-based proteins could be further optimized by combining the two high-yielding strategies: that is, fusing chloroplast signal sequences to transgenes expressed using a self-replicative vector. In addition, expression yields and downstream purification may be enhanced using elastin-like polypeptide (ELP) fusions (Duvenage et al., 2013). This preliminary study offers crucial practical advice for future chimaera design; the insertion and the length of the L2 sequence replacement should be reconsidered as both may affect assembly, efficient epitope display and thus immunogenicity of the vaccine antigen. Overall, the L1/L2 (108–120) chimaera is the best candidate vaccine: it was produced at very high yield, assembled into cVLPs and elicited anti-L1 and anti-L2 antibody responses that neutralized HPV-16 and HPV-52.

Experimental procedures

Plant expression vectors

Three binary Agrobacterium plant expression vectors were used to optimize L1/L2 chimaera expression: pTRAc and pTRAkc-rbcs1-cTP (provided by Prof. Rainer Fischer; Fraunhofer Institute for Molecular Biology and Applied Ecology, Germany) and the BeYDV vector pRIC3 (Figure S6). The pTRA vectors are nonreplicative plasmids targeting the expressed protein to either the cytoplasm (pTRAc) or chloroplasts (pTRAkc-rbcs1-cTP) via the chloroplast transit peptide sequence of the potato rbcS1 gene (Maclean et al., 2007), while pRIC3 is a self-replicating cytoplasm-targeting vector derived from BeYDV (Regnard et al., 2010).

Subcloning of the L1/L2 genes

The L1/L2 chimaera genes were excised from pGA4 vectors (using 3′ XhoI and 5′ BspHI, MluI or HindIII sites) and directionally subcloned into the expression vectors. AflIII/XhoI was used for pTRAc; MluI/XhoI, for pTRAkc-rbcs1-cTP; and HindIII/XhoI, for pRIC3. DH5-α chemically competent E. coli cells (E. cloni™, Lucigen) were transformed with the plasmid constructs and recombinants selected using ampicillin (100 μg/mL). Recombinant clones were screened by colony PCR, using pTRAc vector-specific primers and chimaera-specific primers binding to different L2 epitopes (Table 2). Recombinants were verified by sequencing.

Table 2. Primers used in PCR and sequencing of the L1/L2 chimaeras
Primer targetDetectionPrimer namePrimer sequenceProduct (kb)
pTRAc vectorAll chimaeraspTRAcFwd5′-CATTTCATTTGGAGAGGACACG-3′~1.7
L1/L2 chimaerasAll L1/L2 chimaerasModNewFwd5′-CGACGACCTGTACATCAAGG-3′
L1/L2 (108–120)VEET Rvs5′-GATGAAGCTGGTCTCCTCC-3′0.41
L1/L2 (56–81)SAF2 Rvs5′-GGATGTAGCCGGTCCTGC-3′0.44

Agrobacterium-mediated transient expression

Agrobacterium-mediated protein expression was studied as described by Maclean et al. (2007). Successful transformation was confirmed by colony PCR and restriction enzyme digestion. Infiltration methods used in this study were direct injection (used for small-scale optimization studies) and vacuum infiltration (used for mass production of antigen). For injection, expression time trials were conducted 1–9 days post-infiltration (dpi), and chimaeras were expressed with or without the NSs silencing suppressor. For the comparative vector expression study using pTRAkc-rbcs1-cTP, pTRAc and pRIC3, leaves on three different plants were injection-infiltrated and analysed (providing three biological repeats). For vacuum infiltration, whole N. benthamiana plants were co-infiltrated with NSs and the L1/L2 chimaeras by submerging the plant in the bacterial suspension and subjecting it to a vacuum of 80 kPa for ~1 min and then releasing the vacuum rapidly. The plants were grown for 5 days in 16-h light, 8 dark, at 22 °C.

Extraction of plant-expressed L1/L2 chimaeras

Crude extract from the injection-infiltrated leaves was prepared as described by Maclean et al. (2007), with the inclusion of complete EDTA-free protease inhibitor as per manufacturer's instructions (Roche, Germany). The vacuum-infiltrated leaves were harvested, weighed and ground in liquid nitrogen using a mortar and pestle; low-salt PBS (LS-PBS: 10 mm NaCl PBS with protease inhibitor) was added at a ratio of 1 : 4 (w/v); and samples were homogenized in a Waring blender for 10 min on ice. The homogenate was sonicated on ice for 6x 20-s intervals of sonication and rest (Macrotip sonication; Level 8; Heat Systems—Ultrasonics, Inc. Sonicator Cell Disruptor Model W-225 R), and the lysate was filtered through a double layer of Miracloth (CALBIOCHEM). The crude extract was clarified twice by centrifugation at 26 000 g for 10 min.

ELISA quantification of L1/L2 chimaera yields and detection of anti-L1 responses

Plant-expressed L1/L2 chimaeras were quantified by capture ELISA, using a modified polyvinyl alcohol (PVA)-blocking ELISA method (Studentsov et al., 2002). Briefly, a 96-well Maxisorp microtitre plate was coated with 1 : 2000 mouse anti-HPV-16 L1 MAb (either CamVir1 or H16.J4) overnight at 4 °C, washed and blocked with PVA. Diluted plant extract was added to the wells and incubated for 1 h at 37 °C, followed by a washing step and the addition of rabbit anti-HPV-16 polyclonal serum (1 : 1000) overnight at 4 °C. After washing, swine anti-rabbit horseradish peroxidase (HRP) conjugate (1 : 5000; DAKO) was added to wells, plates were incubated for 30 min at 37 °C, and the proteins were detected with OPD substrate (DAKO; Denmark). Plates were developed in the dark, the reaction was stopped with 0.5 m H2SO4, and the absorbance was detected at 490 nm.

To normalize the ELISA data, TSP was determined using a Lowry protein assay (Bio-Rad) as per the manufacturer's instructions, with a bovine plasma IgG standard (Bio-Rad). Statistical differences in chimaera expression were determined using ANOVA and the Fisher LSD post hoc test. Differences were reported at statistically significant at < 0.01.

ELISA was used to detect L1 antibodies in mouse sera. A 96-well plate was coated with insect cell-produced HPV-16 L1 (30 ng/100 μL/well). After blocking and washing, pooled sera were diluted fourfold (1 : 50 to 1 : 51 200) and incubated in triplicate for 2 h at room temperature. Positive control wells contained 1 : 50 dilution of the anti-L1 antibodies CamVir1 (Abcam®) and H16.V5 MAb. After 4x PBS washing, goat anti-mouse horseradish peroxidase conjugate (1 : 2000; Sigma) was added and incubated for 1 h at 37 °C. Plates were washed 4x with PBS and OPD substrate added. Anti-L1 binding titres were expressed as a reciprocal of the maximum serum dilution. A two-tailed, nonpaired t-test was used to calculate statistical significance of the anti-L1 response, compared with the negative control vaccine (= 0.01). One-way analysis of variance (ANOVA) was used to compare the vaccines, and the Fisher LSD, Turkey HSD and Bonferroni tests were used to determine the significance (= 0.01).

Western blot quantification of L1/L2 chimaera expression and detection of anti-L2 responses

Plant extracts were incubated at 95 °C for 5 min in loading buffer (Sambrook et al., 1989), similar volumes of sample were loaded into the wells, separated on a 10% SDS-PA gel and transferred onto a nitrocellulose membrane by semidry electroblotting. L1 was detected either with CamVir1 (1 : 10 000; Abcam, UK), which binds to the L1 linear epitope GFGAMDF located at amino acids 230–236 (McLean et al., 1990), or with H16.J4 mAb (1 : 2500), which binds a linear epitope located at amino acids 261–280 within the FG loop of the L1 protein (Christensen et al., 1996). MAbs were detected with secondary goat anti-mouse-alkaline phosphatase conjugate (1 : 10 000; Sigma) and blots developed with NBT/BCIP substrate (Roche). The commercial vaccine Cervarix, containing 40 μg/mL of insect cell-produced HPV-16 L1, was used as a known standard for the plant-produced HPV antigens, using equal volumes of antigens and Cervarix dilutions.

Escherichia coli-expressed His-tagged HPV-16 L2 antigen was used for Western blot detection of anti-L2 antibodies. His-tagged L2 (2.5 mg) was loaded into a single wide well, separated and transferred as described above. The membrane region between 55 and 130 kDa, containing the L2 protein at ~80 kDa, was divided into 12 similar-sized strips, blocked and then probed individually with either pooled sera (1 : 100) or a mouse anti-His antibody (1 : 2000, Serotech) as a positive control. Blots were probed with secondary antibody (1 : 5000, Sigma), washed and developed.

Densitometry (GeneTools, Syngene, Synoptics Ltd.) was used to quantify the L1/L2 chimaeric antigens using standard curves generated by the Cervarix dilution series and also to measure the absorbance intensity of the L2 epitopes detected using anti-L2 Abs. Values were normalized for nonspecific background absorbance, using the value associated with the negative control vaccine. Sera were assumed to show an anti-L2 response if L2 bands had absorbance values >2x, the value observed for the HPV-16 L1 final bleeds.

Enrichment using heparin chromatography

The crude plant extracts for HPV-16 L1 and the L1/L2 chimaeras were double-clarified and filtered through a 0.22-μm Millipore filter. Chromatography was performed on an ÄKTA explorer system 10, using a prepacked 1 mL HiTrap Heparin cation-exchange column (GE Healthcare, Amersham Biosciences AB, Sweden), equilibrated with 10 column volumes (cv) of LS-PBS wash buffer at a flow rate of 0.5 mL/min. The crude extract was loaded onto the column and washed with 10 cv of wash buffer. A step elution gradient was applied to enrich the antigens for mouse immunogenicity studies: 10 cv of 50% HS-PBS (0.75 m NaCl), followed by 10 cv of 100% HS-PBS. Fractions (1 mL) were collected and analysed by dot blot, using CamVir1 (1 : 10 000) to detect L1 and Cervarix as a positive control. Eluted fractions containing a high concentration of enriched antigen were pooled, concentrated and desalted using ultrafiltration spin tubes with 10-kDa MWCO filters as per the manufacturer's instructions (Macrosep® Centrifugal Devices, 10K Omega, PALL Life Sciences). The NSs-infiltrated plant extract (negative control) was similarly enriched, and the fractions that corresponded with the eluted protein peak were pooled.

Antigen purity was analysed by comparing the crude plant extract to the enriched antigen, using Coomassie-stained SDS-PA gels and Western blot analysis with CamVir1 (1 : 10 000). Capture ELISA using the linear epitope-specific monoclonal antibody (MAb) H16.J4 was used to determine the HPV antigen yields. The antigen yields relative to TSP in both the crude and the partially purified samples were used to determine antigen enrichment.

Structural analysis using transmission electron microscopy

The crude and heparin-enriched antigens were diluted 10x in PBS, immunotrapped using CamVir1 (1 : 1000) and captured on glow-discharged carbon-coated copper grids. The proteins were negatively stained with 2% uranyl acetate and viewed on a Zeiss 912 Omega Cryo EFTEM.

Immunization of mice and serology

Ethical permission for the study was obtained from the University of Cape Town Research Ethics Committee (AEC 008/037). Female C57/BL6 mice (10 mice per group) were immunized subcutaneously with 10 μg of antigen (either the plant-derived L1/L2 chimaeras or L1 as a positive control) in 100 μL Dulbecco's PBS (DPBS, Sigma). Control mice were immunized with the same volume of NSs-infiltrated plant extract. The TSP was similar for all doses (0.1–0.3 mg/mL), and the immunogen was prepared by homogenization in Freund's incomplete adjuvant (FIA) (1 : 1 v/v) using the syringe extrusion technique (Koh et al., 2006). Prebleeds were taken 12 days prior to vaccination (day 0), and mice were boosted on days 13, 27, 41 and 48 before obtaining the final bleeds at day 62.

HPV pseudovirion neutralization assays

Human papillomavirus pseudovirion (PsV) production, purification and neutralizing assays were performed according to Buck et al. (2005), with alterations described below. The plasmids and HEK293TT cells required for the assay were provided by Dr John Schiller (Laboratory of Cellular Oncology, National Cancer Institute, Bethesda, MD), and the protocols are available in the Technical Files at http://home.ccr.cancer.gov/Lco/. The ‘Production of Papillomaviral Vectors (Pseudoviruses)’ protocol was used, with the inclusion of 250 μg/mL hygromycin B (Roche). For transfection, 175 μL FuGene6 (Roche) was added to 5.7 mL DMEM with GlutaMAX (serum-free media) and incubated for 5 min at room temperature. A total of 40 μg of endotoxin-free DNA was added (20 μg of each plasmid) and incubated for a further 30 min at room temperature before addition to the cells. Pseudovirions were harvested 40–48 h post-transfection and resuspended at a cell density of >100 × 106 cells/mL. Using the ‘Improved Maturation of HPV and Polyomavirus’ protocol, sterile ammonium sulphate (1 m, pH 9.0) was added to a final concentration of 25 mm. The mixture was incubated at 37 °C for 15 min to allow lysis and then transferred to the preferred temperature for pseudovirion maturation overnight (25 °C for HPV-16 and HPV-18 and 37 °C for HPV-45 and HPV-52).

Human papillomavirus PsVs were purified from the clarified cell supernatant by density gradient ultracentrifugation on a 27–39% OptiPrep gradient. Dot blots were performed on each fraction. Cervarix (HPV-16 L1) or the clarified HPV-16, HPV-18, HPV-45 or HPV-52 supernatants initially loaded onto the gradient were used as positive controls. CamVir1 (1 : 5000; Abcam) and H16.I23, H45.N5, H52.C1 and H52.D11 MAbs (1 : 2000; kindly provided by Dr Neil Christensen) were used to detect HPV-16, HPV-18, HPV-45 and HPV-52, respectively. Peak fractions containing a high concentration of L1 were pooled and stored at −70 °C. Purified HPV PsVs were also analysed using TEM: PsVs (1 : 1000) were trapped on glow-discharged carbon-coated copper grids, stained with 2% uranyl acetate and viewed using a Zeiss EM 912 CRYO EFTEM.

The PsV titrations and neutralization assays were based on the ‘Papillomavirus Neutralisation Assay’ protocol, with the exception that no NAb were included in the titration. Serial dilutions of PsVs were prepared in neutralization media (twofold dilutions from 1 : 250 to 1 : 64 000) in nontreated sterile 96-well polystyrene plates (Nunc) and tested in triplicate. SEAP activity was detected using a microplate luminometer (Digene DML 2000) and the Great EscAPe™SEAP Chemiluminescence Kit 2.0 (Clontech Laboratories, Inc.) according to manual instructions, except volumes were adjusted to 0.6 v of those given in the manufacturer's protocol. An in vitro neutralization assay was used to detect HPV-specific antibody responses in mouse sera and to determine endpoint neutralization titres. Pooled sera were diluted fourfold (1 : 50 to 1 : 12 800) and tested in triplicate for the neutralization of HPV-16, HPV-18, HPV-45 and HPV-52 PsVs. The HPV-16, HPV-45 and HPV-52 positive controls were H16.V5, H45.N5, H52.C1 and H52.D11 MAb. The HPV-18 control was rabbit anti-Cervarix sera from our laboratory. Neutralization titres were defined as the reciprocal of the highest dilution that reduced SEAP activity by at least 50%.


The authors would like to thank Gillian de Villers, Dr Brandon Weber and Mohammed Jaffer (EM Unit, UCT) for their technical assistance; Prof Anna-Lise Williamson for the use of her group's facilities for the immunogenicity studies, as well as Prof Enid Shephard and Bruce Allan (IIDMM, UCT) for advice and support; Dr Neil Christensen (Dept Pathology, Milton S. Hershey Medical Center, PA) for providing monoclonal antibodies; Dr Marcel Prins (Wageningen Agricultural University, The Netherlands) for the A. tumefaciens LBA4404 (pBIN-NSs) strain; and Prof Rainer Fischer (Fraunhofer Institute, Aachen, Germany) for the pTRA vectors. We thank Dr John Schiller (Laboratory of Cellular Oncology, National Cancer Institute, Bethesda, MD) for supplying the HEK293TT cells and the plasmids used for the HPV pseudovirion neutralization assays. We thank the South African Department of Science and Technology (DST), the National Research Foundation (NRF) and the South African Medical Research Council (MRC) for funding this work. CP was supported by the Poliomyelitis Research Foundation (PRF) and UCT (PPI fund).