• Open Access

Fusion of an Fc chain to a VHH boosts the accumulation levels in Arabidopsis seeds

Authors

  • Sylvie De Buck,

    1. Department of Plant Systems Biology, VIB, Gent, Belgium
    2. Department of Plant Biotechnology and Bioinformatics, Ghent University, Gent, Belgium
    Search for more papers by this author
  • Jonah Nolf,

    1. Department of Plant Systems Biology, VIB, Gent, Belgium
    2. Department of Plant Biotechnology and Bioinformatics, Ghent University, Gent, Belgium
    Search for more papers by this author
  • Thomas De Meyer,

    1. Department of Plant Systems Biology, VIB, Gent, Belgium
    2. Department of Plant Biotechnology and Bioinformatics, Ghent University, Gent, Belgium
    Search for more papers by this author
  • Vikram Virdi,

    1. Department of Plant Systems Biology, VIB, Gent, Belgium
    2. Department of Plant Biotechnology and Bioinformatics, Ghent University, Gent, Belgium
    Search for more papers by this author
  • Kirsten De Wilde,

    1. Department of Plant Systems Biology, VIB, Gent, Belgium
    2. Department of Plant Biotechnology and Bioinformatics, Ghent University, Gent, Belgium
    Search for more papers by this author
  • Els Van Lerberge,

    1. Department of Plant Systems Biology, VIB, Gent, Belgium
    2. Department of Plant Biotechnology and Bioinformatics, Ghent University, Gent, Belgium
    Search for more papers by this author
  • Bart Van Droogenbroeck,

    1. Department of Plant Systems Biology, VIB, Gent, Belgium
    2. Department of Plant Biotechnology and Bioinformatics, Ghent University, Gent, Belgium
    Current affiliation:
    1. Institute for Agricultural and Fisheries Research (ILVO), Merelbeke, Belgium
    Search for more papers by this author
  • Ann Depicker

    Corresponding author
    1. Department of Plant Systems Biology, VIB, Gent, Belgium
    2. Department of Plant Biotechnology and Bioinformatics, Ghent University, Gent, Belgium
    Search for more papers by this author

Summary

Nanobodies® (VHHs) provide powerful tools in therapeutic and biotechnological applications. Nevertheless, for some applications, bivalent antibodies perform much better, and for this, an Fc chain can be fused to the VHH domain, resulting in a bivalent homodimeric VHH-Fc complex. However, the production of bivalent antibodies in Escherichia coli is rather inefficient. Therefore, we compared the production of VHH7 and VHH7-Fc as antibodies of interest in Arabidopsis seeds for detecting prostate-specific antigen (PSA), a well-known biomarker for prostate cancer in the early stages of tumour development. The influence of the signal sequence (camel versus plant) and that of the Fc chain origin (human, mouse or pig) were evaluated. The accumulation levels of VHHs were very low, with a maximum of 0.13% VHH of total soluble protein (TSP) in homozygous T3 seeds, while VHH-Fc accumulation levels were at least 10- to 100-fold higher, with a maximum of 16.25% VHH-Fc of TSP. Both the camel and plant signal peptides were efficiently cleaved off and did not affect the accumulation levels. However, the Fc chain origin strongly affected the degree of proteolysis, but only had a slight influence on the accumulation level. Analysis of the mRNA levels suggested that the low amount of VHHs produced in Arabidopsis seeds was not due to a failure in transcription, but rather to translation inefficiency, protein instability and/or degradation. Most importantly, the plant-produced VHH7 and VHH7-Fc antibodies were functional in detecting PSA and could thus be used for diagnostic applications.

Introduction

During the last 20 years, plant-derived systems have become an interesting platform for the production of important heterologous proteins and form a promising alternative for the conventional production systems based on bacteria, yeast or mammalian cell cultures (Twyman et al., 2003). Indeed, plants have many practical, economical and safety advantages (De Muynck et al., 2010; Karg and Kallio, 2009; Ma et al., 2005; Nagels et al., 2012; Paul and Ma, 2011; Peters and Stoger, 2011). Plants can correctly assemble complex proteins, incorporate post-translational modifications and show a high potential for scaled-up production. Furthermore, plants show a low risk of animal and human pathogen contaminations. Already several plant tissues, including leaves, seeds, roots and cell cultures, are successfully used for the production of these recombinant proteins, but especially seeds are very attractive bioreactors. Seeds provide the possibility of prolonged and cheap storage of the recombinant proteins in a relatively small volume and in a stable environment at room temperature (Lau and Sun, 2009).

Our research group developed a plant transformation vector, based on the regulatory sequences of the seed storage proteins arcelin-5I and β-phaseolin of Phaseolus vulgaris (common bean; Goossens et al., 1999), for the production of recombinant antibodies in Arabidopsis thaliana seeds (De Jaeger et al., 2002). The combination of these regulatory sequences with the 2S2 signal peptide to target the recombinant protein to the secretory pathway, and with a KDEL sequence to retain the proteins in the endoplasmic reticulum (ER), resulted in very high levels of single-chain variable antibody fragments (scFv) and scFv-Fc antibodies, up to 36% and 14% of total soluble protein (TSP) in T3 homozygous seeds, respectively (De Jaeger et al., 2002; Loos et al., 2011b; Van Droogenbroeck et al., 2007). Recently, this seed-specific expression cassette was also used for the production of full-length antibodies (up to 4.9% of TSP) and antigens (up to 7.7% of TSP; Loos et al., 2011a; Morandini et al., 2011).

Nowadays, there is an increasing interest in the production and use of Nanobodies® (VHHs; Vanlandschoot et al., 2011), which are the small antigen-binding domains of the heavy chain antibodies, produced by camelids (Hamers-Casterman et al., 1993; Muyldermans, 2001). VHHs have a molecular weight of 14–15 kDa, and their monomeric behaviour, high stability and solubility and ability to bind epitopes that are not accessible to conventional antibodies make them highly suitable for many therapeutic and biotechnological applications (Harmsen and De Haard, 2007; Saerens et al., 2008). VHH production is especially successful in Escherichia coli (Arbabi Ghahroudi et al., 1997), but production in Saccharomyces cerevisiae, Pichia pastoris, filamentous fungi and mammalian cell cultures has also been documented (Frenken et al., 2000; Joosten et al., 2005; van der Linden et al., 2000; Rahbarizadeh et al., 2006; Zhou et al., 2008). Until now, only a few studies have reported the accumulation of VHHs in transgenic plants. In transgenic tobacco, potato and Arabidopsis plants, the accumulation levels of VHHs varied between 0.005% and 3.4% of TSP (Ismaili et al., 2007; Jobling et al., 2003; Korouzhdehy et al., 2011; Magee et al., 2004; Winichayakul et al., 2009), while upon transient Agrobacterium infiltration of Nicotiana benthamiana leaves, VHHs accumulated to an exceptionally high level of up to 30% of TSP (Teh and Kavanagh, 2010).

Prostate-specific antigen is a well-known biomarker for prostate cancer in the early stages of tumour development (Bok and Small, 2002). Several monoclonal or polyclonal antibodies and recombinant scFv fragments targeted against PSA have been generated (Oesterling et al., 1995; Whitlow et al., 1993). An anti-PSA VHH (cAbPSA-N7, here termed VHH7) was found to outperform conventional antibodies in terms of sensitivity (Huang et al., 2005) and enabled the detection of total PSA levels at clinically relevant concentrations. Moreover, this VHH showed a higher conformational stability (Saerens et al., 2005). Fusion of a ‘fragment crystallizable’ chain (Fc chain) of an antibody to the VHH domain results in a VHH-Fc fusion protein, and due to the Fc auto-oligomerization based on the disulphide bridges in the hinge region, a stable bivalent homodimeric complex is formed. This bivalence can result in a strengthened antigen–antibody interaction, because avidity superimposes on the affinity (Giersberg et al., 2010), and a prolonged serum half-life (Harmsen and De Haard, 2007). In several antibody-based assays, the Fc chain also allows the use of secondary polyclonal antibodies, resulting in an amplification of the signal and an increased sensitivity. The production of VHH antibodies in E. coli is very efficient, whereas the production of bivalent VHH-Fc fusions is not. The aim of this research was therefore to evaluate the production of VHH7 and VHH7-derived antibodies in the Arabidopsis seed platform. We analysed whether the type of signal sequence (plant versus camel) and the origin of the Fc chain (human, mouse or pig) have an influence on the accumulation level and stability of the antibody. Also, we determined the functionality of the in planta produced VHH7 and VHH7-Fc proteins in a PSA ELISA at concentrations needed for clinical diagnostics assays.

Results

Generation of transgenic lines accumulating VHH7 and VHH7-Fc proteins

In order to evaluate the accumulation of VHH and VHH-Fc proteins in Arabidopsis seeds, different T-DNA vectors were constructed, all containing the neomycin phosphotransferase II (NPTII) gene, conferring resistance to kanamycin, and an expression cassette under the control of the seed-specific β-phaseolin promoter (De Jaeger et al., 2002) to obtain seed-specific expression of the VHH and VHH-Fc fusions (Figure 1a). As gene of interest, the coding sequence of the anti-PSA nanobody® (VHH7) was chosen (Saerens et al., 2004). To target VHH7 to the ER, the plant 2S2 signal peptide (p) and the original signal peptide of the camel heavy chain antibody (c) were evaluated (pPphas-pVHH7 and pPphas-cVHH7, respectively). For the production of VHH7-Fc fusions in seeds, the Fc fragment of a human IgG1 (hGFc) antibody was linked to these two versions of the VHH7 coding sequence, giving rise to the pPphas-pVHH7-hGFc and pPphas-cVHH7-hGFc T-DNA vectors. All coding sequences contained a KDEL signal for retention of the recombinant protein in the ER (Figure 1b). Transgenic Arabidopsis plants were obtained by Agrobacterium-mediated floral dip transformation, and 18 pVHH7, 10 cVHH7, 10 pVHH7-hGFc and nine cVHH7-hGFc T2 single-locus transformants were retained for further analysis.

Figure 1.

Schematic outline of the different T-DNAs used. (a) pPhas GW-derived T-DNA (Morandini et al., 2011). (b) Detailed overview of the gene of interest in the pVHH7, cVHH7, pVHH7-hGFc, cVHH7-hGFc, pVHH7-pAFc, pVHH7-pGFc and pVVH7-mGFc T-DNAs (not to scale). In all constructs, the VHH7 nanobody (Saerens et al., 2004) was used and fused to the His-tag or different Fc regions, in combination with a KDEL ER-retention signal. PNOS, nopaline synthase gene promoter; NPTII, coding sequence of the neomycine phosphotransferase gene; 3′OCS, 3′ regulatory region of the octopine synthase gene; CcdB, negative selection marker, a lethal gene that targets Escherichia coli DNA gyrase; CmR, chloramphenicol resistance gene; PPHAS, β-phaseolin gene promoter (−1 to −1470; GenBank accession no. Z5020); 3′ARC5I, 3′ regulatory region of the arcelin-5I gene from Phaseolus vulgaris; 2S2, signal peptide of the Arabidopsis 2S2 seed storage protein gene; CS, camel leader sequence; Omega leader, the 5′ untranslated region of the tobacco mosaic virus (31 base pairs); VHH7, nanobody against the prostate-specific antigen; hinge, hinge region of the corresponding Fc region; hG-Fc, crystallizing fragment (CH2 and CH3 domain) of the human IgG1 immunoglobulin; pAFc, crystallizing fragment of the porcine IgAb immunoglobulin (derived from U12594); pGFc, crystallizing fragment of the porcine IgG3 immunoglobulin (accession no. EU372658); mGFc, crystallizing fragment of the mouse IgG3 immunoglobulin (NCBI accession no. X00915; UniProt accession no. PO3987); KDEL, endoplasmic reticulum retention signal; LB, left border; RB, right border; attR1-attR2-attB1-attB2, gateway recombination sequences.

VHH7-hGFc accumulated 10- to 100-fold higher as compared to VHH7 in Arabidopsis seeds

To determine the VHH7 and VHH7-hGFc accumulation levels in the transgenic Arabidopsis seed stocks, a direct ELISA was performed by coating seed extracts, followed by detection with an anti-histidine or an anti-human IgG1 antibody for the VHH7 and VHH7-hGFc proteins, respectively. For pVHH7, the accumulation levels ranged from 0.01% to 0.13% of TSP, with an average of 0.06% of TSP for the T2 seed stocks, and between 0.01% and 0.11% of TSP, with an average of 0.07% of TSP for the T3 seed stocks (Figure 2, Table S1). Similarly for cVHH7, the average accumulation levels were 0.03% and 0.06% of TSP, in the T2 and T3 seed stocks, respectively (Figure 2, Table S1). Thus, the VHH7 accumulation levels in transgenic seeds were very low. In the pVHH7-hGFc and cVHH7-hGFc seed stocks, on the other hand, much higher accumulation levels of the VHH7-Fc fusion were observed. In the T2 seed stocks, the accumulation levels of pVHH7-hGFc ranged from 1.8% to 10.7% of TSP, with an average of 5.0% of TSP, and those of cVHH7-hGFc from 0.6% to 5.9% of TSP, with an average of 3.1% of TSP (Figure 2, Table S1). Similarly, in the pVHH7-hGFc and cVHH7-hGFc homozygous T3 seed stocks, the accumulation levels showed an average of 8.6% of TSP (varying between 3.0% and 16.3% of TSP) and of 7.4% of TSP (varying between 0.60% and 12.9% of TSP), respectively (Figure 2, Table S1). For the majority of the transformants, the accumulation level in the T3 seed stock was two times higher than in the corresponding T2 seed stock, which is indicative for a positive gene dosage effect.

Figure 2.

Accumulation levels of VHH7 and VHH7-hGFc proteins in T2 and T3 Arabidopsis seed stocks as determined by ELISA. Every dot represents the antibody accumulation in a different transgenic seed stock. The average accumulation level per construct is indicated. The number (n) of analysed segregating T2 and homozygous T3 seed stocks is mentioned below. The accumulation levels are expressed as % VHH7 or VHH7-hGFc protein per microgram of total soluble protein (TSP).

The plant-produced VHH7 and VHH7-hGFc antibodies detect sub-nanogram amounts of PSA

To analyse whether the plant-produced VHH7 and VHH7-hGFc antibodies are able to recognize PSA, a sandwich ELISA was performed. After coating the wells with a fixed amount of polyclonal anti-PSA antibodies, a PSA dilution series was loaded. The PSA was then detected with VHH7 and VHH7-hGFc, purified from pPphas-pVHH7 and pPphas-pVHH7-hGFc/pPphas-cVHH7-hGFc seed stocks, respectively. In this ELISA set-up, VHH7 detected up to 200 pg of PSA in concentrations of 2 ng/mL, while the two purified VHH7-hGFc antibody preparations both detected PSA down to 100 pg in a concentration of 1 ng/mL (Figure 3). The performance of VHH and VHH-Fc can, however, not be compared as they are detected by different secondary antibodies.

Figure 3.

Functional analysis of the plant-produced VHH7 and VHH7-hGFc antibodies in a sandwich ELISA. The pVHH7 detected up to 200 pg of prostate-specific antigen (PSA) per well, whereas the pVHH7-hGFc and cVHH7-hGFc antibodies both detected 100 pg of PSA. In the blank wells, no PSA was added. The background, indicated as a horizontal line, was determined as twice the OD640 value of the blank.

The N-terminal signal sequence has no influence on the accumulation levels or the proteolysis pattern in Arabidopsis seeds

In order to obtain more data on the integrity of the in planta produced VHH7 and VHH7-hGFc proteins, the seed extracts were analysed via Western blot analysis (Figure 4). The pVHH7 and cVHH7 bands were revealed with anti-His antibodies at approximately 15 kDa (Figure 4a). The intensity of the signal corresponds to 5–20 ng in a 20-μg seed protein extract when compared to a dilution series of VHH7 purified out of a pVHH7 Arabidopsis seed stock, corresponding to 0.02%–0.1% of TSP and confirming the low accumulation found with ELISA. For the pVHH7-hGFc and cVHH7-hGFc seed extracts, the main signal obtained with anti-human IgG1 antibodies has a molecular weight of approximately 50 kDa, indicated with an arrow in Figure 4b,c, as expected for the pVHH7-hGFc and cVHH7-hGFc antibodies. The intensity of this signal varies between 20 and 200 ng in 2 μg of seed protein extract, as derived by comparison with the dilution series of the purified standard, corresponding to 1%–10% of TSP (Figure 4b,c). In the blots, also a smaller fragment of approximately 30 kDa was detected (Figure 4b,c, arrowhead) and one could deduce that this fragment contains at least part of the Fc chain. This was confirmed by N-terminal sequencing. Analysis of the gel-purified pVHH7-hGFc 30-kDa protein band revealed that the degradation product has its cleavage between valine and serine in the C-terminal part of VHH7, just before the hinge region (Figure 4d). Both the 50- and 30-kDa fragments bear glycosylation, since PNGase F treatment results in protein bands with lower molecular weights (Figure S1). N-terminal sequencing revealed that the plant or camel signal peptides were correctly cleaved off from the pVHH7, cVHH7-hGFc and pVHH7-hGFc proteins (B. De Vreese, pers. commun.; data not shown).

Figure 4.

(a–c) Western blot analysis and (d) proteolysis cleavage analysis of the VHH7-hGFc protein. Immunodetection of (a) pVHH7 and cVHH7 via antihistidine antibodies, (b) pVHH7-hGFc and (c) cVHH7-hGFc proteins, via anti-human IgG1, HRP-linked whole antibodies. The names of the seed stocks are indicated below the bar. Twenty microgram of total soluble protein (TSP) of the pVHH7 and cVHH7 seed extracts and 2 μg of TSP of the pVHH7-hGFc and cVHH7-hGFc seed extracts were loaded. As standards, purified VHH7 protein produced in Escherichia coli and purified VHH7-hGFc protein produced in Arabidopsis thaliana seeds were used. The arrow and arrowhead indicate the expected full-length and degradation product, respectively. (d) Amino acid sequence of the C-terminal part of the VHH7 nanobody (capital letters), the hinge region (capital and bold letters) and the N-terminal part of the human Fc region (capital and italic letters). The 30-kDa degradation pattern, detected on Western blot analysis, arose due to the cleavage between valine (V) and serine (S) in the carboxy terminal end of the nanobody (arrowhead). HRP, horseradish peroxidase.

In conclusion, no difference in accumulation was observed when either the plant or the camel signal peptide was used.

Evaluation of the transgene RNA and protein levels in single-copy T-DNA transformants

In order to evaluate whether the observed differences in accumulation of VHH7 versus VHH7-hGFc proteins in seeds were the result of transcriptional or translational changes, mRNA levels were determined in developing seeds, harvested 13 days postanthesis (Figure 5). To exclude possible dosage effects, seven transgenic lines harbouring only one T-DNA insertion were selected by Southern blot analysis. The accumulation levels of VHH7 in these single-copy homozygous T3 seed stocks varied between 0.04% and 0.07% of TSP for both the pVHH7 and cVHH7 seed stocks, and the VHH7-hGFc levels varied between 2.2% and 3.0% of TSP in the pVHH7-hGFc and cVHH7-hGFc seed stocks (Figure 5a). Thus, the difference in VHH versus VHH-Fc accumulation level in single-copy T-DNA transformants was about 50-fold. Total RNA was isolated from micro-dissected developing T4 seeds and a qRT-PCR was performed with VHH7-specific primers. Both VHH7 and VHH7-hGFc mRNA levels were comparable in all samples (Figure 5b), indicating that analogous amounts of recombinant mRNA were produced in the seed stocks of single-copy pVHH7, cVHH7, pVHH7-hGFc and cVHH7-hGFc transformants. This implies that the differences in VHH7 versus VHH7-hGFc antibody accumulation levels are due to post-transcriptional effects.

Figure 5.

Comparison of the antibody protein accumulation levels and the mRNA expression levels in single-copy T-DNA transgenic lines. (a) Antibody accumulation levels in homozygous T3 seed stocks. The amount of antibody is expressed as percentage of TSP. (b) mRNA expression analysis of VHH7 and VHH7-hGFc in homozygous T4 developing seeds, 13 days after anthesis. The fold difference of expression in the analysed lines compared to untransformed control seeds is shown. Error bars are standard deviation of three biological repeats.

Fusion of pVHH7 to different Fc regions results in different levels of accumulation and stability

To investigate whether the origin of the Fc chain has an influence on the accumulation level and the stability, three other pVHH7-Fc fusions were cloned. pVHH7 was fused to the Fc chain of the mouse IgG3 immunoglobulin (pPphas-pVHH7-mGFc), the porcine IgG3 immunoglobulin (pPphas-pVHH7-pGFc) and the porcine IgAb immunoglobulin (pPphas-pVHH7-pAFc; Figure 1b). The T2 seed stocks of 27 pVHH7-mGFc, 27 pVHH7-pGFc and 25 pVHH7-pAFc were analysed by Western blot analysis and compared with a purified standard of another VHH fused to the corresponding Fc fragment (Virdi et al., 2013; Figure 6). In both the pVHH7-mGFc and pVHH7-pGFc seed stocks, high accumulation of VHH7-Fc antibodies could be demonstrated, along with some minor degradation fragments (Figure 6a,b). A higher degradation was seen in the pVHH7-pAFc seed stocks: the full-length VHH7-pAFc antibody of 50 kDa was visualized together with a conspicuous fragment of approximately 30 kDa, corresponding to the IgAb Fc fragment (Figure 6c). A similar extensive degradation pattern was also observed in another VHH fusion to the pig Fc fragment of IgA antibodies (V2-pAFc; Figure 6c).

Figure 6.

Influence of different Fc fusions on accumulation and stability of the VHH7-Fc fusion. Western blot analysis (a) with anti-mouse IgG3 (γ3 chain-specific), horseradish peroxidase (HRP) linked (from goat) to detect VHH7-mGFc proteins, (b) with anti-pig IgG (whole molecule), HRP linked (from rabbit) to detect VHH7-pGFc proteins, (c) and with anti-pig IgA, HRP linked (from goat) to detect VHH7-pAFc proteins. The arrows and arrowheads indicate the full protein and the degradation products, respectively. As standards, fusions were used between a particular VHH and the respective Fc fragment purified from Arabidopsis seed stocks. Because the VHHs in these fusions were different, the fusions did have a different molecular weight. The standard used in (a) was GBP-mGFc (De Meyer T., Eeckhout D., De Rycke R., De Buck S., Muyldermans S. and Depicker A. in prep.), in (b) was V2-pGFc (Virdi et al.,2013.) and in (c) was V2-pAFc (Virdi et al.,2013.). (d) Graphical representation of the accumulation levels of the VHH7-hGFc, VHH7-mGFc, VHH7-pGFc and VHH7-pAFc antibodies in different seed stocks, determined by Western blot analysis. The number of analysed seed stocks (n) together with the average accumulation level is indicated.

Based on these Western blots, the full-length VHH7-Fc accumulation levels for the different Fc fusions were determined and compared to those in the pVHH7-hGFc transformants (Figure 6d). The average accumulation level of the pVHH7-hGFc antibody fusion measured from immunoblots (Figure 4b) was about 2% of TSP. This is approximately 2.5 times lower than estimated by ELISA (5% of TSP; Figure 2 and Table S1). This can be explained by the fact that in the ELISA, the anti-human IgG1-HRP conjugate detects both full-length antibodies and degradation products, while the quantification in the Western blots was carried out based on the signal of the full-length 50-kDa band. The VHH7-Fc accumulation levels were comparable in the pVHH7-hGFc, pVHH7-mGFc and pVHH7-pGFc T2 seed stocks, with an average accumulation level of 2.0%, 2.7% and 1.1%, respectively. In the pVHH7-pAFc seed stocks, lower accumulation levels were obtained, with an average of 0.4% of TSP, and the maximum level being 1.0% of TSP. Therefore, we can conclude that fusion of VHH7 to a different Fc chain leads to a drastic increase in accumulation levels and that the Fc chain determines to a large extent the amount of cleavage in the C-terminal part of the nanobody.

Discussion

Here, we report the accumulation levels of VHH7 and VHH7-Fc antibodies in A. thaliana seeds. As determined by ELISA and Western blot, the accumulation levels of VHH7 in seeds were rather low (<0.13% of TSP in ELISA) compared to those of other antibody fragments, such as scFv and scFv-Fc (De Jaeger et al., 2002; Van Droogenbroeck et al., 2007). Except for one study (Ismaili et al., 2007), the reported accumulation levels of VHHs in stably transformed plants have been generally very low (<0.1% of TSP; Jobling et al., 2003; Korouzhdehy et al., 2011; Winichayakul et al., 2009). The reason for this is not yet known, especially because VHHs are known to have a very high stability (Harmsen and De Haard, 2007).

Fusion of an Fc chain, irrespective of its origin (mouse, human or pig), to the VHH7 resulted in 10- to 100-fold higher accumulation levels (1%–16% of TSP) and these levels were in the same range as those observed for other seed-produced scFv, scFv-Fc and full-length antibodies (De Jaeger et al., 2002; Loos et al., 2011a; Van Droogenbroeck et al., 2007). Fusion of an Fc chain to the VHH7 domain thus drastically increases the accumulation levels of these VHH-Fc proteins. Similar observations were described before. A bivalent head-to-tail fusion of camelid antibodies in Arabidopsis showed a higher accumulation level than that of a single VHH (Winichayakul et al., 2009), and also, the fusion of a nanobody to the β-glucuronidase enzyme or the elastin-like polypeptides (ELPs) increased the accumulation levels 100- to 1000-fold (Conrad et al., 2011; Lentz et al., 2012). Fusion of the Fc fragment of a human immunoglobulin A to the HIV p13 antigen resulted in a 13-fold increase in expression in tobacco leaves (Obregon et al., 2006), and also, fusion of the HA1 antigenic domain of haemagglutinin from H5N1 avian influenza virus to a mouse or human Fc antibody fragment improved the overall yield of the construct in tobacco leaves and allowed a single-step purification (Spitsin et al., 2009). Also in other systems, such as P. pastoris, mammalian cells and different plant production platforms, fusion of heterologous proteins to an Fc chain is known to improve yield, stability and purification (Andrianov et al., 2010; Asano et al., 2010; Bitonti et al., 2004; Logan et al., 2009; Shapiro et al., 2003; Vitale et al., 2006; Wycoff et al., 2011). Also, other protein-stabilizing fusion partners have been identified (Benchabane et al., 2008 and references therein; Fischer et al., 2012 and references therein). Fusion to ELPs resulted in an enhanced accumulation of spider silk proteins (Patel et al., 2007; Scheller et al., 2004), two different avian flu H5N1 antigens (Phan and Conrad, 2011), human interleukin-10 and murine interleukin-4 (Patel et al., 2007) and scFvs (Scheller et al., 2006) in plants. Similarly, fusion of recombinant proteins to ubiquitin, zein, hydrophobin, cholera toxin B subunit and viral coat proteins resulted in increased recombinant protein levels in plants (Conley et al., 2011 and references therein; Hondred et al., 1999; Joensuu et al., 2010; Mainieri et al., 2004).

It is not clear why fusion to such ‘carrier’ proteins increases the accumulation levels of the ‘cargo’. The Fc fusion proteins may be more stable than the VHH domains due to the assembly into a dimeric structure (Hiatt et al., 1989; Obregon et al., 2006). The Fc chain might also result in improved packing of the complex in protein storage vacuoles and/or reduce aggregation and/or enhance proteolytic stability. Fusion to ELPs and hydrophobins induces the formation of novel protein bodies, which might exclude the heterologous proteins from normal physiological turnover (Conley et al., 2011 and references therein).

The use of the native immunoglobulin camel signal peptide or the plant 2S2 signal peptide had no influence on the accumulation levels of both the VHH7 and VHH7-hGFc proteins. Similarly, several studies demonstrated that the native signal peptide could be replaced by a plant or a yeast signal peptide without affecting antibody secretion in plants (Düring et al., 1990; Hein et al., 1991). However, others showed that protein production could be improved by replacing the heterologous signal peptide by the host signal peptide and that the use of native signal sequences might result in unexpected targeting (Arakawa et al., 1997; Schaaf et al., 2005; Streatfield, 2007 and references therein; Yoshida et al., 2004).

By analysing different single-copy T-DNA transformants, and thus excluding the gene dosage effect, we could demonstrate that differences in transcription/transcript levels were not the reason for the 100-fold increase in accumulation of the VHH7-hGFc proteins compared to the VHH7 proteins, suggesting that the low amount of VHH7s in Arabidopsis seeds could be the result of protein misfolding, instability or degradation. Gething (1999) postulated that binding protein (BiP) might stabilize partially folded intermediates during folding and assist in the assembly of protein oligomers, and Nuttall et al. (2002) demonstrated that BiP takes part in the immunoglobulin folding and assembly in transgenic plants. In the described VHH7-hGFc fusions, the VHH7 domain was fused via the hinge to the CH2 and CH3 domains of the Fc, and interaction with the BiP protein may result in a decreased sensitivity to proteolysis through secondary protein structure, as postulated earlier by Obregon et al. (2006). This hypothesis is indeed strengthened by the fact that the BiP1, BiP2 and BiP3 genes are up-regulated in pVHH7-hGFc accumulating seed stocks (De Wilde et al., 2013).

Production of VHH7 fused to the human IgG1 Fc chain in all transformants was accompanied with the presence of a degradation fragment of 30 kDa, corresponding to the Fc domain. Production of Fc fusion antibodies or full-length antibodies in plants indeed often results in a significant amount of specific degradation products (Castilho et al., 2011; De Muynck et al., 2009; Loos et al., 2011a,b; Nagels et al., 2012; Spitsin et al., 2009; Villani et al., 2009), although in some instances the extent can be minimal (Andrianov et al., 2010; Obregon et al., 2006; Van Droogenbroeck et al., 2007; Wycoff et al., 2011). What determines the degree of degradation is not yet completely understood, but most probably it is a combination of several parameters. First, the antibody hinge and closely related regions are the most sensitive to proteolytic activity (De Muynck et al., 2009). N-terminal sequencing of the 30-kDa VHH7-Fc degradation product revealed that cleavage occurred within the C-terminal end of the VHH7 sequence and not in the hinge. Indeed, when the same human IgG1 Fc region was fused to an anti-MBP single-chain variable fragment (scFv), no degradation was observed upon accumulation in Arabidopsis seeds (Van Droogenbroeck et al., 2007), while fusion to the 2G12 scFv and HepA78 scFv did result in degradation products (Loos et al., 2011b). Another parameter is the plant species and/or plant tissue determining to a large extent the degree of degradation. Different proteolysis patterns have been observed in different plant species, such as A. thaliana, Nicotiana tabacum and N. benthamiana, and also within different tissues of the same plant, indicating that a different set of proteases are active in different tissues at different time points during development (Loos et al., 2011b). Furthermore, although the anti-MBP scFv-Fc did not show any degradation in Arabidopsis seeds, degradation was observed when produced in N. tabacum or Petunia seeds (B. Van Droogenbroeck and A. Ann Depicker, unpubl. results). Also, the nature of the Fc chain seems to determine the degree of degradation and the pattern of observed degradation bands. In combination with VHH7, the human IgG1 Fc chain and the porcine IgAb Fc chain displayed much degradation, while the Fc chains of the mouse IgG3 and porcine IgG3 antibodies showed no or a very limited degree of proteolysis. We conclude that particular hinge–Fc region combinations influence the 3D structure and surface exposure of protease-sensitive cleavage sites of the upstream VHH domain, leading to a unique proteolysis pattern in each VHH-Fc combination. Such influence on protein degradation is also noticed in green fluorescent protein fusion experiments (Yewdell et al., 2011). Thus, it is tempting to postulate that the functionality in antigen binding can also be modulated.

The PNGase F endoglycosidase–based mobility shift assay showed that the heterologous proteins bear N-linked glycans. Moreover, as PNGase F is incapable of cleaving N-linked oligosaccharides with core α-1,3 fucosylation, a glycosylation that is typically seen in later Golgi apparatus, one can presume that the heterologous proteins are primarily ER retained, as expected. However, analysing the VHH7-hGFc subcellular localization will help in further confirmation.

In conclusion, we demonstrated that the accumulation level of a nanobody in Arabidopsis seeds is very low, but that this accumulation can be 100-fold increased by the fusion of the nanobody to an Fc chain. The Fc-fused VHH can be easily purified from seed extracts and is still functional. We therefore claim that the Arabidopsis seed production platform is well suited to produce VHH-derived Fc-fused monoclonal antibodies with all the advantages of bivalent antibodies and the opportunities offered by the Fc for easy purification and signal amplification.

Experimental procedures

Construction of the VHH and VHH-Fc expression vectors

Vector pcDNAcAbPSAN7c harbours the original leader sequence of a camel heavy chain antibody, and the coding sequence of the VHH7 nanobody, directed against the PSA (Saerens et al., 2005), fused to the human Fc chain (=hinge, CH2 and CH3 domains) of the human IgG1 antibody (hGFc; Van Droogenbroeck et al., 2007). Initially, four different T-DNA vectors were produced. The pPphas-pVHH7 and pPhas-pVHH7-hGFc encode the VHH7 nanobody and the VHH7-hGFc antibody, respectively, with the 2S2 plant signal peptide (p). The pPphas-cVHH7 and pPphas-cVHH7-hGFc code for the same VHH7 nanobody and VHH7-hGFc antibody, respectively, but with the camel leader sequence (c; Nguyen et al., 1998). VHH7 in both the pPphas-pVHH7 and pPphas-cVHH7 vector contains a His-tag for purification and detection. The VHH7 and VHH7-hGFc coding sequences all contained a C-terminal KDEL sequence and they were inserted into vector pPphasGW by initially transferring it to the Gateway® vector pDonor221 (Life Technologies Europe B.V., Gent, Belgium) in a two-step PCR process with the final amplicon carrying an attB1 site and the 2S2 signal sequence or the camel leader sequence at the 5′ end and KDEL and attB2 sequences at the 3′ end. The coding sequences of VHH7 or VHH7-hGFc were amplified using forward primer 1 and reverse primer 1, specific for each construct (Table S2), and the resulting amplicon was then amplified using construct-specific forward primer 2 and reverse primer 2 (Table S2). The final amplicon was used in a BP recombination reaction with the pDonor221 vector, resulting in the entry clones pEnpVHH7, pEnpVHH7-hGFc, pEncVHH7 and pEncVHH7-hGFc. These entry clones were independently used with the Gateway® destination vector pPphasGW (Morandini et al., 2011) for LR recombination generating the four expression vectors pPphas-pVHH7, pPhas-pVHH7-hGFc, pPhas-cVHH7 and pPphas-cVHH7-hGFc.

The expression vectors pPphas-pVHH7-pGFc, pPphas-pVHH7-pAFc and pPphas-pVHH7-mGFc were created by exchanging the human Fc region with the Fc chain from porcine IgG3 (accession no: EU372658; pGFc), porcine IgAb (derived from U12594; pAFc) and mouse IgG3 (NCBI accession no. X00915; UniProt accession no. PO3987; mGFc), respectively. Briefly, the porcine IgG3, porcine IgAb and mouse IgG3 Fc DNA fragments were chemically synthesized with flanking BstEII-BamHI sites, and via these restriction sites, the Fc fragments were interchanged with the human Fc in the entry vector pEnpVHH7-hGFc. The resulting entry vectors were then recombined in a Gateway® LR reaction with the destination vector pPphasGW (Morandini et al., 2011). The resulting expression vectors were transferred to the Agrobacterium strain C58C1RifR (pMP90; Koncz and Schell, 1986) by electroporation.

Plant transformation and screening for single-locus and homozygous seed stocks

Arabidopsis thaliana (L.) Heyhn, ecotype Columbia 0 were transformed by floral dip (De Buck et al., 2012). Transgenic T1 plants were selected on Murashige and Skoog agar medium supplemented with kanamycin (50 mg/L), nystatin (50 mg/L) and vancomycin (750 mg/L). Single-locus transgenic lines were identified (De Buck et al., 2012), and homozygous T3 seed stocks were distinguished by analysing T-DNA segregation.

Protein extraction, protein concentration determination, SDS-PAGE and Western blot analysis

For each seed stock, 5 mg of Arabidopsis seeds was crushed together with two metal balls (5 mm) in a Retch MM200 device for 2 min at a mill frequency of 20 oscillations/s and proteins were extracted in 900 μL of 50 mm Tris(hydroxymethyl)aminomethane (Tris)–HCl, pH 8.0, 200 mm NaCl, 5 mm ethylenediaminetetraacetic acid (EDTA), 0.1% (v/v) of polyoxyethylene sorbitan monolaurate (Tween 20) as described by De Buck et al. (2012). The total concentration of proteins in the seed extracts was determined by the Lowry method (De Buck et al., 2012).

Seed protein extracts were separated by SDS–polyacrylamide gel electrophoresis (SDS-PAGE) using a 10% polyacrylamide separating gel under reducing conditions. The proteins were transferred electrophoretically from the gel onto Immobilon-P polyvinylidene fluoride membrane (Millipore, Billerica, MA). Blots were blocked in 4% (w/v) skimmed milk and 0.1% (v/v) Tween 20 in phosphate-buffered saline (PBS). VHH7 proteins were detected with mouse antihistidine monoclonal antibody (Serotec, Düsseldorf, Germany), followed by anti-mouse IgG (from sheep) coupled to horseradish peroxidase (HRP; GE Healthcare, Diegem, Belgium). VHH7-hGFc proteins were detected with anti-human IgG1, HRP-linked whole antibody (from sheep; GE Healthcare), VHH7-pGFc proteins with anti-pig IgG (whole molecule), HRP linked (from rabbit; Sigma-Aldrich, Diegem, Belgium), VHH7-pAFc proteins with anti-pig IgA, HRP linked [from goat; Serotec), and VHH7-mGFc proteins with anti-mouse IgG3 (γ3 chain-specific), HRP linked [from goat; Southern Biotech (distibuted by ImTec Diagnostic NV), Antwerpen, Belgium]. Accumulation levels of VHH7 and VHH7-Fc proteins were estimated via the Bio-Rad ChemiDoc™ Imaging System (Bio-Rad, Hercules, CA).

The mobility shift assay after glycosidase treatment of pVHH7-hGFc proteins was performed with PNGase F enzyme [New England Biolabs (distributed by BIOKE), Leiden, The Netherlands].

Purification of VHH and VHH-Fc antibodies from Arabidopsis seeds

To purify pVHH7-hGFc and cVHH7-hGFc fusions from Arabidopsis seeds, seed extracts were prepared in binding buffer (50 mm phosphate buffer, 300 mm NaCl and 0.1% CHAPS; pH 7.4) and purified using Protein A affinity purification (MabSelect SuRe 1 mL; GE Healthcare) on an Äkta Explorer® purification system (GE Healthcare). After washing the column with 50 mm phosphate buffer (pH 7.4), the antibodies were eluted with 0.1 m glycine buffer (pH 3.0) and immediately neutralized with 1 m Tris (pH 9.0). The same Protein A purification protocol was followed to purify the reference antibodies GBP-mGFc and V2-pGFc with IgG Fcs, while the resin SSL7/Agarose [InvivoGen (distributed by Cayla), Toulouse, France] was used for the purification of V2-pAFc. The pVHH7 nanobody was purified on a His Trap® Fast Flow column (1 mL; GE Healthcare), equilibrated with binding buffer (50 mm phosphate buffer, 300 mm NaCl, 20 mm imidazole and 0.1% CHAPS; pH 7.4). The VHH7 protein was eluted with 50 mm phosphate buffer, 300 mm NaCl, 400 mm imidazole and 0.1% CHAPS. To remove the imidazole from eluate, the purified protein was washed with PBS on a Vivaspin column (Sartorius, Vilvoorde, Belgium).

ELISA to determine the total amount of VHH7 antibodies in the seed extracts

Maxisorp 96-well plates (NUNC; Sigma-Aldrich) were coated overnight at 4 °C with 1/100 and 1/250 dilutions of the seed extract in 0.1 m NaHCO3 coating buffer, pH 9.6 in duplicate. As standards, a twofold dilution series of 20, 10, 5, 2.5. 1.25 and 0.625 ng purified VHH7 produced in E. coli (Saerens et al., 2004) were spiked into the seed extract. Free protein-binding sites in the wells were blocked for 1 h at room temperature with blocking solution [1% bovine serum albumin (BSA) and 0.1% Tween 20 in PBS]. After washing the plate three times with 0.1% Tween 20 in PBS, wells were incubated with a mouse anti-His IgG (Serotec), 1/1000 diluted in blocking buffer for 1 h and 30 min. After washing three times, detection was performed with an alkaline phosphatase sheep anti-mouse IgG (whole molecule)–F(ab’)2 fragment conjugate (Sigma-Aldrich), 1/1000 diluted in blocking buffer, for 1 h and 30 min. The absorption at 405 nm was measured 30 min after the addition of the substrate p-nitrophenyl phosphate. For the calculation of the amount of VHH7 antibodies in the samples, the standard curves in 1/100 and 1/250 Col-0 diluted seed extract were used for the 1/100 and 1/250 diluted samples, respectively. The detection limit of this assay was 0.01% VHH7 of TSP, or 10 ng in 100 μg seed protein per well.

ELISA to determine the total amount of VHH7-hGFc antibodies in the seed extracts

Maxisorp 96-well plates (NUNC; Sigma-Aldrich) were coated overnight at 4 °C with 1/2000 and 1/5000 dilutions of the seed extract in 0.1 m NaHCO3 coating buffer, pH 9.6 in duplicate. As standard curve, 0.625–20 ng purified VHH7-hGFc from Arabidopsis seeds was added to 1/2000 or 1/5000 diluted Col-0 extract in 0.1 m NaHCO3 coating buffer, pH 9.6. Wells were incubated with a sheep anti-human IgG–HRP conjugate (Sigma-Aldrich), 1/5000 diluted in blocking buffer (1% BSA and 0.1% Tween 20 in PBS). The absorption at 640 nm was measured 30 min after the addition of 3,3′,5,5′-tetramethylbenzidine liquid substrate (Sigma-Aldrich).

ELISA to analyse the functionality of the plant-produced VHH7 and VHH7-hGFc antibodies

Maxisorp 96-well plates (NUNC; Sigma-Aldrich) were coated overnight at 4 °C with a 1/1000 dilution of chicken anti-PSA antibodies [Abcam (distributed by VWR), Leuven, Belgium] in 0.1 m NaHCO3 coating buffer, pH 9.6. Free protein-binding sites in the wells were blocked for 1 h at room temperature with blocking buffer (2% skimmed milk and 0.1% Tween 20 in PBS). After washing the plate three times with 0.1% Tween 20 in PBS, the wells were incubated with a twofold dilution series of PSA (Scipac, Kent, UK) from 1 to 0.78 pg per well in blocking buffer for 1 h and 30 min. After washing three times, the PSA was detected with 200 ng of purified pVHH7, pVHH7-hGFc or cVHH7-hGFc per well, diluted in blocking buffer. The pVHH7 nanobody was detected with 1/1000 diluted mouse anti-His IgG (Serotec), followed by 1/5000 diluted sheep anti-mouse IgG fused with HRP (GE Healthcare). The pVHH7-hGFc or cVHH7-hGFc antibodies were detected in one-step with a 1/5000 diluted sheep anti-human IgG–HRP conjugate (Sigma-Aldrich). The absorption at 640 nm was measured 30 min after the addition of the substrate 3,3′,5,5′-tetramethylbenzidine (Sigma-Aldrich).

RNA extraction and quantitative real-time PCR analysis

qRT-PCR analyses were performed on 13 dpa (days postanthesis) developing seeds, by collecting seeds from 20 to 25 pooled siliques. For this, flowers were labelled when anthers had just reached the stigma (0 dpa), allowing self-fertilization to take place. RNA was extracted with the Spectrum Plant Total RNA Kit (Sigma-Aldrich), and a DNase treatment was performed with the On Column DNaseI Digestion Set (Sigma-Aldrich). Poly d(T) cDNA was synthesized from 1 μg of total RNA with the iScript cDNA Synthesis Kit (Bio-Rad, Nazareth Eke, Belgium). Nineteen microlitre of 1/50 diluted cDNA was used to amplify the VHH7 gene and the housekeeping genes cyclophilin and elongation factor 1α (EF1A; Baud et al., 2003; Fallahi et al., 2008). Sample loading in 384-well plates was performed using the Janus robotic platform (JANUS Mini Format; Perkin Elmer, Zaventem, Belgium), and the analysis was performed with an iCycler (Bio-Rad) using SYBR Green (Eurogentec, Seraing, Belgium) for detection in a 5-μL volume. PCRs were performed as described by De Wilde et al. (2013). Relative expression levels were first normalized to two endogenous genes, EF1A and cyclophilin, and then to the respective untransformed controls with the math formula method (Livak and Schmittgen, 2001; Schmittgen and Livak, 2008). For VHH7, forward primer was 5′ TACGCCGACTCCACGAAGG3′ and reverse primer was 5′CGTCTGCCGCACAGTAATAGG3′. For EF1A, forward primer was 5′CTGGAGGTTTTGAGGCTGGTAT3′ and reverse primer was 5′CCAAGGGTGAAAGCAAGAAGA3′ (Baud et al., 2003), and for cyclophilin (At2 g29960), forward primer was 5′TGGACCAGGTGTACTTTCAATGG3′ and reverse primer was 5′CCACTGTCTGCAATTACGACTTTG3′ (Fallahi et al., 2008).

Acknowledgements

The authors thank Dirk Saerens, Katia Conrath and Serge Muyldermans for providing the plasmid pcDNAcAbPSA7Fc and helpful discussion; Deborah Ongenaert for practical assistance; the research groups of Professor De Jaeger (VIB-UGent), Professor Witters (UAntwerpen) and Professor Devreese (UGent) for help with the N-terminal sequencing; and Annick Bleys for help in preparing the manuscript. This work was supported by the European Commission 6th Framework (Pharma Planta, LSHB-CT-2003-503565) and the ‘Bijzonder Onderzoeksfonds’ of the Ghent University. We also thank the colleagues involved in the European Cooperation in Science and Technology (COST) action FA0804 for helpful discussions. T.D.M and K.D.W are indebted to the Agency for Innovation by Science and Technology (IWT) for predoctoral fellowships.

Conflict of interest

No conflicts of interest are to be declared.

Ancillary