The use of vaginal microbicides containing human immunodeficiency virus (HIV)-neutralizing antibodies (nAbs) is a promising strategy to prevent HIV-1 infection. Although antibodies are predominantly manufactured using mammalian cells, elastin-like peptide (ELP) fusion technology improves the stability of recombinant, plant-produced proteins and facilitates their purification, making plants an alternative platform for antibody production. We generated transgenic tobacco plants accumulating four different formats of the anti-HIV-1 antibody 2G12 in the endoplasmic reticulum (ER), i.e. with ELP on either the light or heavy chain, on both, or on neither. Detailed analysis of affinity-purified antibodies by surface plasmon resonance spectroscopy showed that the kinetic binding parameters of all formats were identical to 2G12 lacking ELP produced in Chinese hamster ovary (CHO) cells. Importantly, protein purification from seeds by inverse transition cycling (ITC) did not affect the binding kinetics. Analysis of heavy chain N-glycans from leaf-derived antibodies showed that retrieval to the ER was efficient for all formats. In seeds, however, N-glycans on the naked antibody were extensively trimmed compared with those on the ELP fusion formats, and were localized to a different subcellular compartment. The in vitro HIV-neutralization properties of the tobacco-derived 2G12 were equivalent to or better than those of the CHO counterpart.
Transgenic plants are promising tools to address the limitations of current production systems for recombinant pharmaceutical proteins, which are both expensive and limited in scale. Accordingly, large-scale production in plants is feasible at a fraction of current costs and can be established in resource-poor areas (Teli and Timko, 2004; Ma et al., 2005; Twyman et al., 2005; Boehm, 2007). Functional antibodies have been expressed in transgenic plants and other plant-based systems (Twyman et al., 2005) including cereal seeds, which are particularly attractive because antibodies remain stable and functional even if mature seeds are stored at room temperature for several years (Fiedler and Conrad, 1995; Stoger et al., 2005).
However, it has proven difficult to achieve sufficient yields for some recombinant proteins in plants and the upstream cost savings brought about by using plants are often offset by the expense of downstream purification, which can account for up to 80% of the cost of goods (Twyman et al., 2005). The challenges of both low yields and high purification costs can be addressed through the use of elastin-like polypeptides (ELP) as fusion partners (Meyer and Chilkoti, 1999). ELP are artificial biopolymers derived from mammalian elastin and consist of the pentapeptide VPGXG, where X represents any amino acid except proline (Urry, 1992). ELP are structurally disordered and soluble below their inverse transition temperature (Tt) but at or above Tt intramolecular contacts between non-polar regions displace water molecules (Li et al., 2001; Li and Daggett, 2003) causing a phase transition that results in aggregation and hence precipitation (Urry, 1992, 1993). This reversible solubility is retained in ELP fusion proteins expressed in Escherichia coli, increasing the solubility and stability of many recombinant proteins and allowing them to be purified in a process termed inverse transition cycling (ITC) (Meyer and Chilkoti, 1999; Shimazu et al., 2003; Trabbic-Carlson et al., 2004; Chow et al., 2006).
Since cocktails of neutralizing antibodies (nAbs) appear to confer better protection than single antibodies (Trkola et al., 1995; Xu et al., 2001), we have applied the ELP fusion technology to the antibody 2G12, which is one of the few broadly HIV-nAbs discovered so far and a target of different plant expression strategies (Rademacher et al., 2008; Ramessar et al., 2008a). As already performed for the mAb 2F5, we investigated the effect of ELP fusion on 2G12 accumulation in transgenic tobacco leaves. However, we extended this study for seed-specific expression and further analysed the intracellular accumulation of the antibody molecule in cell compartments. We characterized the N-glycan chains on the antibody heavy chain and studied the antigen-binding activity of antibodies purified by traditional Protein A chromatography and ITC. A detailed investigation of the functionality of ITC purified antibodies by surface plasmon resonance (SPR) spectroscopy was done. Finally, we determined the HIV-1 neutralization activity of the antibody using a cell-based infection assay. To our knowledge, this is the first characterization of therapeutic plant-based, ITC purified antibody–ELP fusions including neutralization capacity and after the initial study of the antibody 2F5 an important step for the development of microbicide components derived from transgenic plants. The implications of our results in terms of using ELP to produce an inexpensive component for HIV microbicides are discussed.
Transgenic tobacco plants produce full-length 2G12 antibodies and ELP fusion proteins
Individual cDNAs encoding the 2G12 HC and LC, and their corresponding 100×ELP fusions, were inserted into plant expression cassettes containing either the constitutive cauliflower mosaic virus (CaMV) 35S promoter or the seed-specific unknown seed protein (USP) promoter (Baumlein et al., 1991; Fiedler et al., 1993) (Figure 1). Each construct also encoded a c-Myc tag for detection and a KDEL signal for retrieval to the endoplasmic reticulum (ER) (Scheller et al., 2004; Floss et al., 2008). Transgenic plants were generated by Agrobacterium-mediated leaf disk transformation (Horsch et al., 1985) and screened for accumulation of the individual antibody chains or ELP fusions. Fusion to 100×ELP resulted in a substantial increase in the accumulation of free antibody chains in tobacco leaves and seeds (data not shown).
Lines with a 3:1 transgene segregation pattern in the T1 generation, consistent with a single transgenic locus, were used as parental lines for the crosses to obtain offspring plants (hemizygous T2 plants) producing different antibody variants (Table 1). Four possible combinations of assembled antibodies were produced in plants showing constitutive expression, i.e. Nt2G12 (naked HC and LC), Nt2G12LELP (LC-ELP fusion, naked HC), Nt2G12HELP (HC-ELP fusion, naked LC) and Nt2G12ELP (ELP fusion on HC and LC). Individual double transgenic T2 plants were identified by Protein A/Protein L sandwich enzyme-linked immunosorbent assay (ELISA) on leaf extracts (Figure 2a) and Western blot analysis under non-reducing conditions with an anti-human Fc-specific antibody (data not shown). For seed-specific expressers, only the Nt2G12 and Nt2G12ELP crosses were carried out. Double transgenic hemizygous T2 plants were first identified by multiplex polymerase chain reaction (data not shown) and using the abovementioned sandwich ELISA on mature seeds (Figure 2b). The accumulation levels of HC/HC-ELP in leaves (constitutive expression) and seeds (seed-specific expression) were compared by Western blot under reducing conditions, allowing us to monitor the effects of ELP fusions on antibody accumulation (Figure 2a,b). The Nt2G12LELP and Nt2G12ELP variants accumulated to the highest levels in leaves (Figure 2a; Nt2G12LELP 1, 6, 10; Nt2G12ELP 1, 3, 4). The LC-ELP fusion increased the accumulation of the naked HC partner compared with the Nt2G12 variant (Figure 2a; Nt2G12LELP 1, 6, 10; Nt2G12 18, 19, 20). The antibody–ELP fusion proteins accumulated to higher levels both in leaves and seeds, except for the Nt2G12HELP variant (Table 2, Figure S1 presents the relevant Western blots). The plant-derived antibodies were compared with CHO-derived standards (CHO2G12), excluding the ELP component, so that expression levels could be compared directly.
Table 2. Accumulation levels of recombinant antibodies in tobacco leaves and seeds
% of TSP
TSP, total soluble protein.
Purification and characterization of 2G12 antibodies from tobacco plants
The leaf- and seed-derived Nt2G12ELP was purified by ITC at four different temperatures (37, 40, 45 and 50 °C). The supernatants (S) and the solubilized pellets (Psol) resulting from the ITC-precipitation were analysed by Western blot (Figure 3). Nt2G12ELP from leaves was detected in the solubilized precipitate (Psol; Figure 3a,b) and was completely absent in the supernatant (S) at all temperatures. This shows that ITC efficiently precipitates and recovers Nt2G12ELP from leaf extracts as already described for Nt2F5ELP (Floss et al., 2008). Analysis of the seed-derived Nt2G12ELP was complicated by the transgene segregation pattern in the T3 generation (seeds of hemizygous T2 plants were analysed), reflecting the significant proportion of seeds expressing LC-ELP or HC-ELP alone. The generation of homozygous transgenic tobacco lines accumulating full-length antibodies using the doubled haploid technology (Floss et al., 2009) can circumvent transgene segregation patterns. However, the ITC method was used successfully for the precipitation and recovery of seed-derived Nt2G12ELP from non-homogenous tobacco seeds (Figure 3c,d). Several notable differences among the seed extracts were identified. A Nt2G12ELP derivative was detected using the anti-human Fc-specific antibody (Figure 3c) and after Coomassie staining of the polyacrylamide gel (Figure 3d, arrow). This band was present even in crude extracts of the seed material and represents a HC-ELP dimer. Higher temperature precipitation (50 °C) abolished this fragment (Figure 3c,d). Another truncated form of Nt2G12ELP, lacking the ELP component, was detected in the crude seed extract but was successfully removed after ITC.
The 2G12 variants were also purified from tobacco leaves by Protein A affinity chromatography as previously described (Floss et al., 2008). After dialysis and ultrafiltration, the antibody preparations were analysed by Western blot using anti-human Fc-specific and anti-human kappa light chain antibodies. The variants migrated at their expected molecular weights (Figure 4a). Nt2G12 was detected as a double band which is probably a result of differences in the N-glycan composition on the antibody heavy chains (Rademacher et al., 2008). Notably, no bands were observed at positions corresponding to free HC-ELP fusions or HC-ELP dimers, demonstrating efficient antibody assembly. A band with a higher molecular weight was detected when analysing Nt2G12 with the anti-human Fc-specific antibody probably indicating the presence of a 2G12 dimer or aggregate (Figure 4a; Nt2G12). Weak signals were detected for LC-ELP with the anti-human kappa light chain antibody (Figure 4a; Nt2G12LELP and Nt2G12ELP) and this most likely reflects the molar excess of LC-ELP over HC and HC-ELP. Based on these results, we concluded that the ELP fusion to either or both antibody chains does not interfere with antibody assembly in tobacco leaves.
Protein A affinity chromatography was then used to purify Nt2G12 (Figure 4b) and Nt2G12ELP from tobacco seeds (lane 1; Figure 4b). Western blot analysis showed no differences between the molecular weight of Nt2G12 from seeds and that of CHO2G12 but as in case of the leaf-derived mAb, Nt2G12 was detected as a double band (see above; Figure 4a,b). However, the analysis of Nt2G12ELP from seeds revealed additional antibody fragments, one with a molecular weight identical to 2G12 without the ELP (indicating loss of the ELP parts, which is consistent with cleavage of the c-Myc tag). However, the dominant band in both the Protein A (lane 1; Figure 4b) and ITC preparations (lanes 2–5; Figure 4b) was exclusively detected by the anti-human Fc-specific antibody suggesting the formation of HC dimers (see above). Free LC-ELP fusions were detected in both preparations of Nt2G12ELP from seeds (lanes 1–5; Figure 4b, arrow) probably reflecting the molar excess of LC-ELP over HC-ELP and co-purification of free LC-ELP.
The comparison of Nt2G12ELP preparations from seeds obtained by affinity chromatography (lane 1; Figure 4b) and ITC (lane 2–5; Figure 4b), which was examined here for the first time, revealed that ITC facilitates the efficient removal of antibody fragments lacking ELP and prevents the concentration of HC-ELP dimers, but co-purifies LC-ELP.
Differential N-glycosylation of 2G12 heavy chains in tobacco leaves and seeds
To determine the N-glycan structures of the affinity-purified 2G12 antibody variants from tobacco leaves (Nt2G12, Nt2G12LELP, Nt2G12HELP, Nt2G12ELP) and seeds (Nt2G12, Nt2G12ELP), the antibodies were separated by non-reducing sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE), excised from the gel and digested with trypsin. Mass spectrometry showed that >90% of the antibody variants from tobacco leaves contained oligo-mannose type (OMT) N-glycans with Man7 being the most abundant form (Table 3). Only 2%–6% of the preparations comprised complex type (CT) N-glycans, showing that KDEL-mediated ER retrieval was efficient but not perfect.
Table 3. Relative abundance of complex type (CT) and oligo-mannose type (OMT) N-glycans on 2G12 heavy chains produced in tobacco leaves
Oligo-mannose type (%)
Complex type (%)
Surprisingly, although ∼45% of the N-glycans on the seed-derived Nt2G12 were single N-acetylglucosamine (GlcNAc) residues (Table 4) and the remainder were predominantly OMT (∼46%), with the remaining ∼8% represented by CT glycans, only 1% of the glycans on the HC-ELP of Nt2G12ELP were single GlcNAc residues, the remainder comprising OMT (94%) and CT (3%) glycans. This suggested that the naked and ELP variants of the antibody in tobacco seeds had different subcellular fates.
Table 4. N-Glycan profiles for Nt2G12 and Nt2G12ELP purified from tobacco seeds
Relative abundance (%)
Oligo-mannose type (Σ)
Complex type (Σ)
Subcellular localization of recombinant antibodies in tobacco seeds
Given the unexpected N-glycan profiles for the seed-derived 2G12 antibodies, immunolocalization studies were carried out to determine where each antibody accumulated. In seeds expressing Nt2G12, fluorescence microscopy showed that the antibody was deposited specifically in protein storage vacuoles (PSV), but was not found in the apoplast (Figure 5a-c). This contradicts the localization reported in an earlier study (Petruccelli et al., 2006). We confirmed our results by transmission electron microscopy, which showed even decoration of the PSV with gold particles (Figure 5b) but no labelling of the lipid bodies, the cell wall or the cytoplasm (Figure 5c). In contrast, Nt2G12ELP was localized in small, putative ER-derived protein bodies (PB) spread throughout the cytosol, as well as in the PSV (Figure 5d-g), confirming different subcellular trafficking for antibodies with and without the ELP fusion. The small PB were ∼400 nm in diameter and tended to cluster in small groups, resembling bunches of grapes (Figure 5g). Similarly, a scFv fused to ELP and expressed in seeds (Scheller et al., 2006) was found forming PB within the cytoplasm. In contrast to Nt2G12ELP, scFv-ELP bodies were lower in number (4–6 per cell) and bigger (∼3μm), already visible with a non-specific stain (Figure 6). The presence of such PB, of probable ER origin, in both fusion proteins indicates that the ELP retard the trafficking of the fusion proteins along the endomembrane system. No significant labelling was found within the apoplast (Figures 5d-f and 6a,b) or the lipid bodies (Figures 5g and 6b,c).
Antigen-binding properties of the recombinant antibodies
The antigen-binding properties of the different 2G12 variants were compared by SPR spectroscopy. For direct comparison, the antibodies were captured onto a Protein A surface, followed by the injection of soluble and monomeric gp120 to determine binding affinities compared with CHO2G12.
The apparent kinetic rate and equilibrium constants were determined by fitting the binding curves (Figure S2) to a simple monovalent interaction model without mass-transport limitation. The results showed that all antibody variants exhibited similar if not identical association and dissociation rate constants and the resulting equilibrium dissociation constants were identical within experimental limitations (>Tables 5 and S1). The only exception was Nt2G12LELP from leaves, with a higher dissociation rate constant. Since there was no similar effect in 2G12ELP, the ELP fusion is unlikely to be responsible and the slight deviation may be related to the purification. Overall, these results demonstrate that the ELP does not significantly interfere with the binding kinetics of 2G12, suggesting they have no negative impact on antibody folding and assembly in leaves or seeds.
Table 5. Comparative surface plasmon resonance analysis of the antigen-binding properties of 2G12 variants
Nt2G12ELP 37 °C ITC
Nt2G12ELP 40 °C ITC
Nt2G12ELP 45 °C ITC
Nt2G12ELP 50 °C ITC
The different antibodies displayed a range of binding activities (Table 5). The Nt2G12 purified from leaves bound to the antigen slightly more weakly than CHO2G12, and the relative activities of leaf-derived Nt2G12HELP and Nt2G12ELP were consistent with the absolute binding activities previously determined for the corresponding 2F5 variants (Floss et al., 2008). The ELP variants of 2G12 bound more strongly to gp120 than CHO2G12 because of the physical properties of the ELP. Nt2G12LELP should have an antigen-binding activity similar to Nt2G12HELP, so the lower activity observed for Nt2G12LELP indicates either a lower-quality protein preparation or the presence of partially-assembled antibodies or degradation products, which may also explain the faster dissociation.
The seed-derived Nt2G12 and Nt2G12ELP preparations showed significantly lower antigen-binding activities than the leaf-derived proteins (Table 5). This was not because of lower fidelity in the seeds but rather because of segregation of the two transgenes in the next generation, resulting in a significant fraction of the seeds expressing only 2G12HC or 2G12HC-ELP (Figures 3c and 4b). Interestingly, the antigen-binding activity increased significantly with the increasing temperatures used for purification by ITC. Analysis by non-reducing SDS-PAGE and Western blot convincingly showed that the increase in antigen-binding activity reflected the lower amount of degradation (Figures 3c and 4b) and unassembled or partially-assembled antibody chains, particularly because of elimination of 2G12 HC-ELP dimers (Figure 3d). The relative antigen-binding activities clearly demonstrated that ITC performed at higher temperatures facilitate to increase the purity of the preparation with no negative impact through thermal unfolding.
2G12 antibodies from tobacco plants neutralize HIV–1
The 2G12 and corresponding ELP derivatives from tobacco leaves were tested in a syncytium inhibition assay to compare HIV-1 neutralizing activity in vitro with the activity of CHO2G12. The 50% inhibitory concentration (IC50) of Nt2G12 was four times lower than that of CHO2G12 (Table 6), consistent with previous results obtained for KDEL-tagged 2G12 produced in corn (Rademacher et al., 2008). The presence of the ELP reduced the HIV-1 neutralization efficacy, i.e. the IC50 of Nt2G12HELP was 5.24 μg/mL, comparable with CHO2G12. ELP fusion to the heavy chain was tolerated better than to the light chain, as the latter resulted in a significantly higher IC50 of 18.7 μg/mL. Nt2G12ELP, in which all antibody chains carry ELP, had the lowest neutralization activity, so clearly the effect of the ELP on IC50 and neutralization efficacy is cumulative. The results further showed that both the amount of ELP and the antibody chain to which they were fused significantly affected HIV-1 neutralization efficacy.
Table 6. Virus neutralization activity of tobacco leaf-derived 2G12
IC50, 50% inhibitory concentration; ratio, IC50CHO2G12 divided by IC50 of mAb.
Microbicides containing nAbs are promising tools for the prevention of sexually transmitted HIV infections (Shattock and Moore, 2003; Klasse et al., 2008). However, the use of such prophylactic approaches requires large-scale production and cost-effective purification technologies, particularly in developing countries. Plants are attractive for this purpose because they are inexpensive, versatile, safe and amenable to rapid and economical scale-up, but the challenges of achieving sufficient yields and reducing the cost of downstream processing remain (Ramessar et al., 2008b).
Active immunization using HIV antigens has not been successful, so the production of nAbs in plants is one of several alternative approaches being explored to combat the spread of the virus (Chong, 2008). We expressed the heavy and light chains of the mAb 2G12 in transgenic tobacco plants, and consistent with previous studies (Floss et al., 2008), the accumulation of the individual light and heavy chains was clearly enhanced by ELP fusion in tobacco leaves and seeds. A two- to threefold increase in the levels of assembled 2G12 in leaves and seeds was achieved when ELP were fused to all four antibody chains, representing a total yield of up to 1% TSP (Table 2). This increase in yield was stronger than for the plant-derived 2F5 antibody formats (Floss et al., 2008), however, was not as striking as previously reported for seed-specific scFv expression (Scheller et al., 2006). This shows that ELP fusion have to be tested with different antibodies, even of identical format, to generally validate this technology. Under this aspect, we note that the antibody 2G12 exhibits a specific VH/VH′ interface between the variable heavy domains (Calarese et al., 2003). The elevated levels of recombinant 2G12 ELP fusions are particularly interesting in seeds, where recombinant proteins are deposited in specific storage compartments (Stoger et al., 2005).
In agreement with previously published data (Ko et al., 2003; Tekoah et al., 2004; Triguero et al., 2005), the KDEL-tagged antibodies predominantly comprised OMT glycans, whereas CT glycans were present only in minute amounts. The leaf-derived 2G12 variants followed this pattern rather than the plant-produced 2F5 (Floss et al., 2008), suggesting that neither N-glycan processing nor intracellular trafficking were significantly influenced by the presence of ELP on the LC, the HC or both. The quality of the leaf-derived antibodies was generally comparable with CHO2G12. Detailed Biacore analysis revealed similar or identical antigen-binding kinetics, and interestingly the relative activities determined for Nt2G12HELP and Nt2G12ELP were very similar to the absolute activities determined previously for the ELP fusions of the antibody 2F5 (Floss et al., 2008). This clearly demonstrates that the particular physical properties of the ELP result in a smaller change to the refractive index or an oriented capture within the dextran matrix of the sensorchip, as we reported for the tobacco-derived 2F5 ELP fusions (Floss et al., 2008), such that the antigen-binding activities based on the molar masses are overestimated. After verifying the antigen-binding properties of the plant-derived antibody preparations in vitro by biosensor analysis, their capacity for in vivo HIV-1 neutralization was investigated using a well-established cell-based neutralization assay. The KDEL-tagged plant-derived 2G12 lacking ELP was more efficient at neutralizing HIV-1 infection than CHO2G12 and had a fourfold lower IC50, as previously observed for 2G12 antibodies produced in maize (Rademacher et al., 2008; Ramessar et al., 2008b) and Nicotiana benthamiana (Strasser et al., 2008). The most likely explanation for this result is the presence of dimeric forms or aggregates of the KDEL-tagged tobacco-derived 2G12, which have a higher neutralization capacity, similar to polymeric forms of 2G12 (Wolbank et al., 2003). When analysing Nt2G12 by Western blot under non-reducing conditions a band with a higher molecular weight was observed probably indicating the presence antibody dimers or aggregates (Figure 4a). Fusion of the ELP to either or both antibody chains significantly reduced the neutralization efficacy, but ELP fusion to the HC was tolerated better than fusion to the LC, while fusion to both chains had a cumulative negative effect on neutralizing capacity. Therefore, both the position and the total number of ELP carried by the antibody are important, probably reflecting steric hindrance of molecular interactions with the virus. This was not observed in the Biacore experiments but the apparent contradiction can be explained as the measurements were performed with antibodies immobilized in the gel-like matrix of the sensorchip and a soluble monomeric form of the antigen. Clearly, efficient neutralization of HIV-1 requires careful fine-tuning of the ELP fusions, e.g. by choosing an appropriate length. Future studies will look at the influence of shorter ELP fusions on neutralization capacity.
The expression of recombinant proteins in seeds is advantageous because seeds offer a stable environment and relatively small volume in which proteins can accumulate (Stoger et al., 2005). Preliminary studies with ELP fusion proteins expressed in seeds indicated that the presence of an ELP moiety significantly affected the intracellular trafficking and deposition of the recombinant proteins and induced the formation of new PB that are not present in the absence of ELP. An increase in protein accumulation in seeds expressing ELP fusions in tobacco seeds expressing a scFv ELP fusion protein has been reported (Scheller et al., 2006). Immunohistochemical analysis of these seeds revealed the formation of PB filled with recombinant proteins (Figure 6). The naked Nt2G12 antibody produced in seeds was exposed to extensive endoglucanase trimming, such that only 46% contained OMT glycans, while 45% carried a single GlcNAc residue. In contrast, seed-derived Nt2G12ELP was not exposed to endoglucanase trimming and OMT glycans predominated, as seen when the 2G12 antibodies were expressed in leaves.
The different N-glycan profiles were concordant with different subcellular distributions of the Nt2G12 and Nt2G12ELP variants in tobacco seeds. Nt2G12 was localized in PSV, whereas Nt2G12ELP was also found in PB within the cytoplasm. Partial vacuolar delivery has been reported previously for a mAb expressed in tobacco seeds, but the glycan structures in that case suggested a Golgi-dependent transport pathway (Petruccelli et al., 2006). In contrast, Nt2G12 lacked significant amounts of vacuolar type N-glycans. The abundance of single GlcNAc residues suggested that OMT glycans were processed en route to the PSV or within it. Golgi-independent transport of proteins from the ER to the PSV has been suggested previously (Vitale and Hinz, 2005). The PB containing Nt2G12ELP budded directly from the ER, the absence of CT glycans confirming that the recombinant protein did not pass through the Golgi apparatus. In this respect, the glycan analysis gives more evidences for the ER origin of the Nt2G12ELP PB. Nt2G12ELP contained predominantly OMT glycans and only traces of single GlcNAc residues (Table 4). It is possible that the endoglucanase does not have access to the ELP fusion because of its different localization, or that the ELP somehow protect the glycan chain from trimming, e.g. by steric shielding. N-glycan trimming between the two core GlcNAc residues has been observed for both KDEL-tagged and untagged 2G12 expressed in maize endosperm (Rademacher et al., 2008; Ramessar et al., 2008a). This suggests that an endoglucanase activity is present in plant seeds, but only in post-ER compartments. Changes in N-glycan structures can influence the effector functions of antibodies, i.e. those with single GlcNAc residues behave more like aglycosylated antibodies, which tend to lack effector functions (Jefferis, 2005). Some seed-based systems are therefore unsuitable for the production of antibodies that require effector functions, unless genetic engineering or alternative targeting strategies (e.g. ELP fusion) can be used to protect the recombinant antibody from endoglucanase activity.
The 100×ELP fusion partner allows the efficient purification of 2G12 from leaf and seed tissue at 40–45 °C, although the higher temperature is preferable for ITC as this removes antibody fragments such as 2G12 without ELP, HC-ELP dimers and other degradation products. ITC has been already used to purify antibody 2F5 from tobacco leaves (Floss et al., 2008) and in this report we have firstly demonstrated that the same method can be used for seed-derived antibodies. The antigen-binding kinetics of seed-derived 2G12-ELP was not affected by ITC and the higher temperatures clearly increased the relative antigen-binding activity of the Protein A-captured antibodies, mainly because of the removal of HC-ELP dimers. The prolonged exposure of 2G12 to temperatures up to 50 °C had no negative impact, showing that ITC is a viable purification strategy.
Our results clearly show that HIV-nAbs can be purified from different plant tissues as ELP fusions, using simple and inexpensive methods, without affecting the functional properties of the antibody. The antibody–ELP fusion proteins retained the parent antibody’s ability to bind the antigen and neutralize HIV in vitro, albeit with slightly reduced potency. It therefore appears that antibody–ELP fusions could be included as microbicide components in the near future. The removal of the elastin-like polypeptide for an external application of the plant-based antibodies in microbicides into the vagina using conventional gel-based systems or so-called sustained-release and controlled-release systems (Klasse et al., 2008) is not necessary. Despite this fact the use of ELP as delivery vectors for chemotherapeutic drugs and therapeutic peptides (Rodriguez-Cabello et al., 2007) or elastin-like polymers with emphasis for medical uses (Raucher et al., 2008) is well described and a main focus of future research of human therapeutics. Furthermore, ELP are extraordinarily biocompatible (Urry et al., 1991; Rincon et al., 2006) and no immunomodulatory effect of ITC purified ELP (containing the c-Myc tag and the KDEL ER retention signal) from tobacco leaf material was detected in vitro (D.M. Floss, U. Conrad & L. Dedieu, unpublished data). Further improvements, such as shortening the ELP or fine-tuning of the ITC process (Conley et al., 2009), or using self-cleaving ELP-intein tags (Banki et al., 2005; Fong et al., 2009), may increase the potency of the antibody and the efficiency of purification and recovery.
Plant transformation vectors
Constitutive expression of 2G12 was achieved using a vector incorporating the CaMV 35S promoter, as previously described (Floss et al., 2008). The eight expression cassettes for the different versions of the heavy and light chains (Figure 1) were verified by sequencing. For seed-specific expression, the CaMV 35S promoter was replaced with the seed-specific USP promoter, excised from pRTRA7/3-USP-anti-oxa-scFv (Scheller et al., 2006). For stable transformation of tobacco, the expression cassettes were transferred individually into the binary vector pCB301-Kan (Gahrtz and Conrad, 2009), which is based on pCB301 (Xiang et al., 1999) and incorporates the T-DNA fragment from pBIN19 (Bevan, 1984).
Generation and selection of transgenic plants
The binary vectors were introduced into Agrobacterium tumefaciens C58C1 (pGV2260) (Deblaere et al., 1985) by electroporation, and transgenic tobacco (Nicotiana tabacum cv. SNN) plants were generated by leaf disk transformation (Horsch et al., 1985). T1 lines showing Mendelian segregation, consistent with a single transgenic locus, were crossed to obtain plants expressing all four possible combinations of the heavy and light chains with and without ELP in the T2 generation (Table 1).
Sodium dodecyl sulphate-polyacrylamide gel electrophoresis and Western blot
Leaf disks were homogenized in SDS sample buffer (72 mm Tris, 10% glycerine, 3% SDS, 5%β-mercaptoethanol, 0.25 μm bromophenol blue, pH 6.8), incubated for 10 min at 95 °C and centrifuged (15 000 g, 30 min, 4 °C). Seeds were ground under liquid nitrogen and homogenized in 50 mm Tris, 200 mm NaCl, 5 mm ethylenediaminetetraacetic acid (EDTA), 0.1% Tween 20, pH 8.0. The homogenates were centrifuged (15 000 g, 20 min, 4 °C) and TSP levels were determined using the Bradford assay (Bio-Rad, Munich, Germany).
Extracts were separated by reducing SDS-PAGE on 10% polyacrylamide (PAA) gels. Antibody assembly was analysed by non-reducing SDS-PAGE on 6% PAA gels as described (Floss et al., 2008). Individual antibody chains were detected with anti-c-Myc (9E10) supernatant followed by horseradish peroxidase (HRP)-conjugated sheep anti-mouse IgG (Amersham Biosciences, Piscataway, NJ, USA) as described (Conrad et al., 1997). The 2G12 heavy chain and HC-ELP fusions were also detected with HRP-conjugated anti-human Fc-specific IgG (Sigma-Aldrich, St Louis, MO, USA). The recombinant proteins were visualized using the ECL Western Blotting Analysis System (Amersham Biosciences). Alternatively, 2G12 antibody variants were detected with anti-human kappa light chain antibody conjugated to alkaline phosphatase (Sigma-Aldrich). After separation of the proteins by non-reducing SDS-PAGE and electroblotting, nitrocellulose membranes were blocked with 3% bovine serum albumin (BSA) in 10 mm Tris (pH 8.0), 150 mm NaCl, 0.1% Tween 20, and recombinant antibodies were visualized by Nitro blue tetrazolium chloride/5-bromo-4-chloro-3-indolyl phosphate staining. Coomassie® Brilliant Blue R-250 (SERVA Electrophoresis GmbH, Heidelberg, Germany) was used for the detection of recombinant antibody variants after SDS-PAGE.
Hemizygous T2 plants accumulating assembled antibodies were identified using Protein A/Protein L sandwich ELISA (Floss et al., 2008), in which assembled heavy and light chains were detected with HRP-conjugated Protein L (Affitech AS, Oslo, Norway) using tetramethylbenzidine and H2O2 as substrates for HRP. The extinction was measured at 450 nm (reference filter 650 nm). CHO2G12 (Polymun Scientific Immunbiologische Forschung GmbH, Vienna, Austria) was used as the standard, and extracts from wild-type tobacco (N. tabacum cv. SNN), and the BSA dilution buffer, were used as negative controls.
Protein A affinity purification
Protein A affinity purification of 2G12 antibody variants from tobacco leaves was carried out as described (Floss et al., 2008). Neutralized fractions containing the antibody were dialysed against 8 mm Na2HPO4, 2 mm KH2PO4, 150 mm NaCl, pH 7.6, concentrated by ultrafiltration (iCON™ Concentrator, molecular weight cut-off = 20 kDa; Pierce, Rockford, IL, USA) and the final retentate was centrifuged (15 000 g, 30 min, 4 °C) to remove any insoluble material.
Seed-derived antibodies were isolated by grinding 4 g seeds under liquid nitrogen and homogenizing the powder in 90 mL of seed extraction buffer. After centrifugation (3500 g, 30 min, 4 °C) and filtration (Corning® bottle-top vacuum filter, Corning®, St. Louis, MO, USA, polyethersulfone membrane, pore size 0.22 μm, Sigma-Aldrich), 4 mL of Protein A agarose (Roche Diagnostics GmbH, Hilden, Germany) was added to the clear supernatant and the mixture was incubated overnight at 4 °C with gentle continuous stirring. Protein A agarose was collected in a disposable 10-mL chromatography column (Qiagen GmbH, Hilden, Germany) and extensively washed with 100 mm Tris-HCl (pH 8.0) and 10 mm Tris-HCl (pH 8.0). The neutralized elution fractions were processed as described above.
Inverse transition cycling
Nt2G12ELP from tobacco leaves was purified by exploiting the salt- and temperature-dependent phase transition of ELP (Meyer and Chilkoti, 1999) as described (Floss et al., 2008). Seeds were ground under liquid nitrogen, homogenized in seed extraction buffer, cleared by centrifugation and filtration as described above. Sodium chloride was added to a final concentration of 2 m, and the solution was incubated 30 min at 37, 40, 45 and 50 °C to induce aggregation of the ELP. Nt2G12ELP was then pelleted by centrifugation at 37, 40, 45 and 50 °C (4500 g, 1 h) and re-dissolved overnight in 100 mm Tris-HCl (pH 8.0) with gentle stirring. Remaining insoluble material was removed (15 000 g, 30 min, 4 °C).
Surface plasmon resonance spectroscopy
Real-time biological interaction analysis was performed on a Biacore 2000 instrument (Biacore; GE Healthcare, Uppsala, Sweden) at 25 °C using 10 mm HEPES, 150 mm NaCl, 3.4 mm EDTA, 0.05% Tween 20, pH 7.4 as the running buffer. The amine coupling procedure was used at a flow rate of 5 μL/min to immobilize 3940 RU of Protein A (Sigma-Aldrich; 200 μg/mL in 10 mm sodium acetate, pH 4.2) on a CM5-rg sensorchip (Biacore; GE Healthcare). The contact times for activation and deactivation were increased to 10 min and for coupling to 15 min to achieve high immobilization levels. Flow cell 1 was activated and deactivated and used as reference. For kinetic analysis, purified 2G12 antibody variants were captured to yield antigen-binding capacities that were similar to the CHO-derived antibody using the predicted molecular masses of each antibody variant. Because of its limited availability, the binding of gp120 (HIV-1BaL gp120, NIH AIDS Research and Reference Reagent Program; http://www.aidsreagent.org) was analysed by a single 150 nm injection for 5 min and the dissociation was followed for 10 min at a flow rate of 30 μL/min. The referenced and blank subtracted sensorgrams were used for evaluation. The relative activity (relα) of the 2G12 variants in relation to their CHO-derived counterpart was calculated using eqn 1.
RCapture, the amount of antibody captured onto the Protein A surface; Rt = 310s, the gp120 binding response at 310 s after the start of the injection; and M, the molecular mass of the antibody variant.
The validity of eqn 1 for determining the relative activity was verified for different capture levels of the CHO-derived antibody (data not shown). As a result of the interchanged/cross-armed structure of the 2G12 antibody, interaction with the monomeric recombinant gp120 is monovalent. As observed previously (Zeder-Lutz et al., 2001) the binding curves were fitted with and well described by a simple 1:1 model. Thus, eqn 1 is generally valid, provided that the influence of mass transport is negligible.
Coomassie-stained bands representing the assembled, full-size antibodies were excised from non-reducing SDS-polyacrylamide gels, destained, carbamidomethylated and digested with trypsin. The peptides were extracted from the gel and analysed as described previously (Kolarich and Altmann, 2000; Kolarich et al., 2006) using a capillary liquid chromatography-electrospray ionization mass spectrometry (LC-ESI-MS) system. Data analysis was carried out using masslynx 4.0 SP4 (Waters Micromass, Milford, MA, USA).
The MS data from the tryptic peptides were compared with data sets generated by in silico tryptic digestion of 2G12, using the PeptideMass program (http://www.expasy.org/tolls/peptide-mass.html). Based on the tryptic peptide data sets, tryptic glycopeptides data sets were generated by the addition of N-glycan masses of the two identified glycopeptides (EEQYN297STYR and TKPREEQYN297STYR).
The HIV-1 neutralization was assessed using a syncytium inhibition assay. Starting at a concentration of 100 μg/mL, 10 twofold serial dilutions of each antibody variant, CHO2G12 and a non-neutralizing control, were pre-incubated with HIV-1 strain RF (102–103 TCID50/mL) for 1 h at 37 °C. CD4+ human AA-2 cells were added at a density of 4 × 105 cells/mL and incubated for 5 days. Experiments were performed with eight replicates per antibody dilution step. The presence of one or more syncytium per well after 5 days was scored as a positive infection. The IC50 values were calculated according to the method of Reed and Muench (Reed and Muench, 1938) using the concentrations present during the antibody virus pre-incubation step.
Seeds expressing Nt2G12, Nt2G12ELP or scFv antibody–ELP fusions (Scheller et al., 2006) were fixed, dehydrated and embedded as described (Arcalis et al., 2004). Thin sections (1.2 μm) were mounted on glass slides and either stained with toluidine blue or immunolabelled. Sections showing silver interference for electron microscopy were mounted on glass slides or gold grids respectively. After blocking in 5% BSA in 0.1 m phosphate buffer (pH 7.4), sections were probed with monoclonal mouse anti-c-Myc, polyclonal goat anti-gamma chain or polyclonal anti-kappa chain antibodies at room temperature for 2 h. After several washes with 0.1 m phosphate buffer (pH 7.4) containing 0.05% Tween 20, the antigen-antibody reaction was visualized with a polyclonal goat anti-mouse serum, conjugated to Alexa Fluor® 488 (Invitrogen, Paisley, UK), or donkey anti-goat conjugated to Alexa Fluor® 546 (Molecular Probes®; Invitrogen) for fluorescence microscopy. For electron microscopy, the reaction was visualized with goat anti-mouse or donkey anti-goat antisera conjugated to 15-nm or 10-nm gold particles respectively (British Biocell International, Cardiff, UK).
This work was funded by the EU FP6/Pharma-Planta. The authors thank Dr Richard Twyman for critical reading of the manuscript and Professor Hermann Katinger (Polymun Scientific, Vienna, Austria) for providing 2G12 hybridomas and CHO-derived reference material. We are grateful to C. Helmold and A. Winger for excellent technical assistance.