• Open Access

Utility of the P19 suppressor of gene-silencing protein for production of therapeutic antibodies in Nicotiana expression hosts


Correspondence (fax 1-519-837-0442;

email jchall@uoguelph.ca)


To study how the P19 suppressor of gene-silencing protein can be used effectively for the production of therapeutic glycoproteins, the following factors were examined: the genetic elements used for expressing recombinant proteins; the effect of different P19 concentrations; compatibility of P19 with various Nicotiana tabacum cultivars for transgenic expression; the glycan profile of a recombinant therapeutic glycoprotein co-expressed with P19 in an RNAi-based glycomodified Nicotiana benthamiana expression host. The coding sequences for the heavy and light chains of trastuzumab were cloned into five plant expression vectors (102–106) containing different 5′ and 3′ UTRs, designated as vector sets 102–106 mAb. The P19 protein of Tomato bushy stunt virus (TBSV) was also cloned into vector 103, which contained the Cauliflower mosaic virus (CaMV) 35S promoter and 5′UTR together with the terminator region of the nopaline synthase gene of Agrobacterium. Transient expression of the antibody vectors resulted in different levels of trastuzumab accumulation, the highest being 105 and 106 mAb at about 1% of TSP. P19 increased the concentration of trastuzumab approximately 15-fold (to about 2.3% of TSP) when co-expressed with 103 mAb but did not affect antibody levels with vectors 102 and 106 mAb. When 103 mAb was expressed together with P19 in different N. tabacum cultivars, all except Little Crittenden showed a marked discolouring of the infiltrated areas of the leaf and decreased antibody expression. Co-expression of P19 also abolished antibody accumulation in crosses between N. tabacum cv. I-64 and Little Crittenden, indicating a dominant mode of inheritance for the observed P19-induced responses.


The gene-silencing machinery of plants is involved in regulating expression of endogenous gene transcripts as well as reducing or eliminating the effects of invading pathogens such as viruses (Baulcombe, 2004; Reinhart et al., 2002). As a countermeasure to this defence mechanism, viruses encode for proteins that act as suppressors of gene silencing (SGS). P19 from the Tomato bushy stunt virus (TBSV) is an example of proteins known to function as a potent suppressor of gene silencing in plants as well as in animals (Scholthof, 2006; Voinnet et al., 1999). Plants also react to most transfer DNA (T-DNA) transgenes that invade their genomes by initiating a post-transcriptional gene-silencing response (Baulcombe, 2004; Brodersen and Voinnet, 2006). The inhibitory effect of P19 on the gene-silencing pathway has been exploited to enhance expression levels of recombinant proteins in plants (Voinnet et al., 2003), but its use has been limited to transient expression only, mainly owing to the deleterious effects of this protein when expressed constitutively at high levels in a transgenic setting (Siddiqui et al., 2008).

There are two main cellular gene-silencing mechanisms in plants, the small interfering RNA (siRNA) and the micro-RNA (miRNA)-silencing pathways (Carthew and Sontheimer, 2009), which are collectively referred to as interfering RNA (RNAi). The two systems show a great deal of similarity in their mechanisms of action, as they share some key enzymes (Brodersen and Voinnet, 2006). Both systems identify their target nucleic acid (viral RNA, viral DNA, transgene mRNA and endogenous mRNA) by a complex known as RNA-induced silencing complex (RISC). RISC carries a complementary single-stranded RNA probe for its target, which upon binding, is either blocked or degraded.

Recently, our understanding of the mechanism of action of P19 in suppressing gene silencing at the molecular level has increased a great deal (Burgyan et al., 2004), but certain aspects still remain unclear. P19 is a multifunctional protein that is active as a dimer and found in both the cytosol and the nucleus (Park et al., 2004). It is capable of binding siRNA and miRNA molecules in a nonspecific fashion (Dunoyer et al., 2004). As there is a rise in virus-derived siRNA levels in plants in response to infection, P19 acts to reduce the amount of free siRNA duplexes through nonspecific binding and represses the silencing response by interfering with siRNA loading of RISC (Hsieh et al., 2009). Studies on TBSV mutants with lowered levels of P19 have shown that a high titre of the protein is critical for exerting its biological activity (Qiu et al., 2002; Scholthof et al., 1999). Despite P19's nonspecific siRNA binding, the effects brought about by this protein show host-specificity (Ahn et al., 2011; Angel et al., 2011; Siddiqui et al., 2008).

Nicotiana species display a variation in induction of the hypersensitive response (HR) to the P19 protein of TBSV, which is indicated by an initial leaf discoloration that leads to a marked necrosis of the infiltrated area (Angel et al., 2011). A putative R gene product, suggested to originate in Nicotiana sylvestris (a progenitor of Nicotiana tabacum), is thought to be responsible in triggering this response (Angel et al., 2011). In N. tabacum cv. Samsun, discoloration from HR develops 2–3 days after infiltration of leaves with P19, leading to fully dehydrated spots on day 7, while the same treatment yields no necrosis or discoloration in Nicotiana benthamiana. Stable transgenic expression of P19, however, does not elicit HR in either N. tabacum cv. Xanthi or N. benthamiana, indicating that high titres of P19 are required for triggering this response (Siddiqui et al., 2008).

The extent to which P19 increases expression seems to vary for different recombinant proteins. Several reports indicate that the expression of green fluorescent protein (GFP), a commonly used reporter, is boosted approximately 50-fold when co-expressed with P19 (Voinnet et al., 2003; Zheng et al., 2009). The levels of a Yersinia pestis antigen expressed in transgenic tomato was enhanced up to 2% of TSP from near undetectable levels when P19 was transiently expressed in the leaves of this plant (Alvarez et al., 2008). Expression of antibodies (Saxena et al., 2011) and other recombinant proteins (Zheng et al., 2009), on the other hand, have only been enhanced by fivefold.

In the context of plant-derived therapeutic proteins, another very important consideration is their glycan profile, as this post-translational modification impacts the efficacy of therapeutic proteins and can be a major factor in batch-to-batch variability of recombinant therapeutic proteins (Gomord et al., 2010; Schiestl et al., 2011). Plant-specific sugar residues on the N-glycan core, namely core α1,3-fucose and β1,2-xylose, are immunogenic in mammals (Bardor et al., 2003; Jin et al., 2008). As a result, a great deal of effort has been directed towards creating plants with modified humanized glycosylation patterns (Cox et al., 2006; Sourrouille et al., 2008; Strasser et al., 2008). For the most part, glycomodified plants have been created through RNAi gene-silencing technology, mainly owing to the existence of multiple endogenous fucosyltransferase and xylosyltransferase genes in most plants (Cox et al., 2006; Sourrouille et al., 2008; Strasser et al., 2008). Although it is generally believed that P19 cannot reverse established gene-silencing pathways (Scholthof, 2006), interference in the siRNA pathway by P19 becomes a concern when RNAi-generated genetic backgrounds are to be used as expression hosts for producing therapeutic proteins, primarily because there are not reports in the literature on using P19 with an RNAi-based glycomodified host, but also since in an unrelated case, P19 was shown to repress the knockdown of a previously established RNAi transgenic line (Ahn et al., 2011). With the above-discussed results in mind, we present a series of experiments that demonstrate how P19 can be used effectively to enhance the expression of a recombinant therapeutic glycoprotein, such as a biosimilar therapeutic antibody, in plant-based expression systems.


The untranslated regions used in recombinant expression of trastuzumab significantly impact antibody accumulation

Trastuzumab is a therapeutic antibody used in the treatment of HER2+ breast cancer (Baselga et al., 1998; Lewis et al., 1993). To produce this antibody in Nicotiana hosts, we cloned its heavy (HC) and light (LC) chains into plant expression cassettes and placed them either on a single binary vector (102 mAb), or on separate binary vectors (103–106HC and 103–106LC), in which case they were co-expressed (referred to as vector sets 103–106 mAb) to produce the fully assembled antibody. The different expression cassettes were designed to carry different combinations of promoters, 5′UTRs and 3′ UTR/terminators (Figure 1), but in all cases carried the same signal peptide from the Arabidopsis basic chitinase protein for apoplastic targeting. We transiently expressed trastuzumab in N. benthamiana using vector sets 103–106 mAb to compare the levels of recombinant antibody production. A 7-day expression time-course with whole-plant vacuum infiltration showed a considerable difference in the dynamics and maximal antibody expression among the four vector sets (Figure 2a). Vectors 105 and 106 mAb resulted in higher maximal antibody accumulation compared with 103 and 104 mAb. Antibody expression peaked at 3–4 days postinfection (d.p.i.) for 103 and 104 mAb, and at 4–5 d.p.i. for 105 mAb, whereas 106 mAb showed a steady increase in expression up to 7 days postinfection. Maximal antibody levels were achieved with the 105 and 106 mAb vector sets, estimated by ELISA at approximately 1% of TSP. The protein content of tobacco leaves has been estimated at approximately 2% of the fresh weight (Stevens et al., 2000); therefore, antibody expression at 1% of TSP translates into approximately 200 mg of antibody per kg of fresh weight (FW). A 20-fold range in antibody accumulation was observed when the different vectors were compared. We conclude that in transient expression, different UTR elements fused to the coding sequences for the heavy and light chains of trastuzumab significantly affect the accumulation of the fully assembled antibody. In addition, N-glycosylation profiles of mAbs were determined by LC–ESI–MS (liquid–chromatography–electrospray ionization–mass spectrometry). Typically, mAbs exhibited a largely homogeneous GnGnXF3 oligosaccharide pattern with plant-specific β1,2-xylose and core α1,3-fucose residues (see last section of 'Results').

Figure 1.

Diagram of the expression cassettes used in this study. The expression cassettes shown here were situated on the T-DNA region of binary vectors. Vector 102 mAb was the only vector carrying the heavy (HC) and light chains (LC) of trastuzumab on the same T-DNA. When expressing trastuzumab with vectors 103–106 mAb, the HC and LC were co-expressed to produce IgG molecules.

Figure 2.

Western blot analysis of trastuzumab expressed transiently with different plant expression vectors in Nicotiana benthamiana. Blots were probed with a 1 : 10 000 mix of alkaline phosphatase conjugated to anti-human γ- and κ-chain antibodies. Expression of the 103–106 mAb vectors was analysed over 6 days. Plants were treated by vacuum infiltration. Each lane represents a pooled sample, created by mixing three leaf samples. The vectors were either expressed alone (a), or together with P19 (b). All vector sets carried the same codes for the HC and LC of trastuzumab coupled with different UTRs. Different expression dynamics were observed when vectors were expressed alone or together with P19, as determined by ELISA (c). Semi-quantitative RT-PCR (d) shows the levels of HC, LC, P19 and Actin at 5 dpi. Messenger RNA levels for HC and LC directly correlate with protein levels.

P19 does not have a similar boosting effect on the expression of trastuzumab from different vector sets

The effect of co-expressing P19 with vectors 103–106 mAb was analysed over 7 days in N. benthamiana (Figure 2b,c) to determine whether the different UTR combinations had any effect on the ability of P19 to boost antibody expression, as previously reported (Saxena et al., 2011; Vezina et al., 2009). All Agrobacterium cultures used in this experiment were adjusted to an OD600 of 0.2. Not all vector sets were positively affected by P19 (Figure 2c, Table 1). For 103 and 104 mAb, P19 resulted in a 15-fold increase in the concentration of trastuzumab, although maximal expression was almost threefold greater with 103 mAb compared with 104 mAb, both without and with P19. For 105 mAb, P19 only resulted in a approximately twofold increase in the concentration of trastuzumab, from 1% to approximately 2.1% of TSP. Antibody accumulation with vector set 106 mAb, which contained only plant-derived UTRs, was unaffected by P19. Antibody expression peaked for 103 and 105 mAb when co-expressed with P19 at just over 2% of TSP. It was also noted that P19 changed the peak time of expression for the vectors 103, 104 and 105 mAb (Figure 2c). Similarly, semi-quantitative RT-PCR data showed that P19 boosted the levels of HC and LC messages expressed with 103–105 mAb, but not with 106 mAb.

Table 1. Maximal trastuzumab level produced with different expression vectors
Vector set (mAb)Max expression (% TSP)Max expression with P19 (% TSP)Expression increase with P19
1030.15 ± 0.012.35 ± 0.1315.6x
1040.05 ± 0.010.75 ± 0.0415.0x
1051.02 ± 0.072.17 ± 0.312.1x
1060.98 ± 0.040.85 ± 0.081.0x

We also tested the ability of P19 to boost expression of trastuzumab with vector 102 mAb, which is similar to 106 mAb, that is it only contains plant-derived UTRs (Figure 1). Similar to 106 mAb, P19 did not affect the antibody expression level of 102 mAb (Figure 3a). P19 was also found to have no boosting effect on trastuzumab expressed with Tobacco mosaic virus- (TMV) and Potato virus X-based (PVX) deconstructed vectors (Figure 3b), which are virus-based plant expression vectors developed by ICON Genetics (Halle, Germany) for transient expression (see Grohs et al., 2010, for details regarding expression of trastuzumab with ICON vectors).

Figure 3.

Western blot analysis of trastuzumab expressed transiently with (a) 102 mAb and with (b) TMV/PVX (virus-based) expression vectors in Nicotiana benthamiana. Blots were probed with a 1 : 10 000 mix of alkaline phosphatase conjugated to anti-human γ- and κ-chain antibodies. Plants were treated by spot infiltration. Pooled samples were generated by combining three infiltrated spots. Two pooled samples (harvested 5 d.p.i.) are shown for each treatment. Similar to 106 mAb, co-expression of P19 did not affect the level of trastuzumab expressed with either vector.

Boosting effect of P19 on recombinant antibody expression is concentration dependent

It is believed that in order for P19 to exert its biological function for a successful TBSV infection, it has to accumulate beyond a certain threshold concentration in plant cells (Qiu et al., 2002; Scholthof et al., 1999). We showed that in transient expression, co-expression of P19 with 103 mAb at an Agrobacterium concentration of OD600 = 0.2 significantly boosted antibody production. As the expression level of recombinant proteins are, for the most part, lower for a transgenic versus transient expression, we looked at co-infiltration of 103 mAb with P19 at lower Agrobacterium concentrations to simulate transgenic-like levels of P19 accumulation in N. benthamiana. Infiltration of 103 mAb vectors alone at OD600 values of 0.2, 0.02 and 0.002 resulted in different levels of antibody production, with a direct correlation between the applied Agrobacterium concentration and antibody expression level (Figure 4a). The ability of P19 to boost antibody expression with 103 mAb progressively diminished when P19 was applied at lower concentrations, while co-expressing P19 at OD600 = 0.2 resulted in a significant increase in antibody expression irrespective of 103 mAb concentration (Figure 4a). Semi-quantitative RT-PCR data showed that P19 message levels were much greater when Agro-infiltration was performed at OD600 = 0.2, compared with applications at OD600 = 0.02 and OD600 = 0.002 (Figure 4b). We conclude that unless P19 can be expressed at very high levels, its utility in transgenic production of recombinant proteins is limited because of its concentration-dependent mode of action.

Figure 4.

(a) Western blot analysis showing the dose-dependent effect of P19 on enhancing recombinant antibody expression in Nicotiana benthamiana. Blots were probed with a 1 : 10 000 mix of alkaline phosphatase conjugated to anti-human γ- and κ-chain antibodies. 103 mAb vectors were co-expressed with P19 at three different concentrations of Agrobacterium, OD600 = 0.2, 0.02 and 0.002. Plants were treated by spot infiltration. Pooled sampled were generated by combining three infiltrated spots. Each lane represents a pooled sample. The boosting effect of P19 was most prominent when applied at the higher concentration, regardless of the concentration of 103 mAb. P19 had no boosting effect on antibody expression when applied at OD600 = 0.002. (b) Semi-quantitative RT-PCR data show that application of the P19 expression vector at OD600 = 0.2 results in greater P19 mRNA compared with OD600 = 0.02 and OD600 = 0.002.

Tobacco varieties respond differently to P19

To utilize P19 for transgenic production of recombinant proteins in plants, it is imperative that P19 does not adversely affect plant growth, and more importantly, gene expression. We showed that when expressed transiently at high levels in N. benthamiana, P19 effectively boosts recombinant protein levels. However, owing to having a much greater biomass, N. tabacum is often selected over N. benthamiana for transgenic production. The downside of using N. tabacum is the development of necroses at the site of infection by 7 days after the introduction of recombinant P19 to leaf cells via Agroinfiltration (Angel et al., 2011). The reaction that is triggered by P19 in N. tabacum, also known as the hypersensitive response, has been well documented but only tested in two tobacco cultivars (Angel et al., 2011; Siddiqui et al., 2008). To identify a tobacco that could be used with P19 in a transgenic expression system, we screened five cultivars for the development of the hypersensitive response. We transiently expressed trastuzumab with 103 mAb, alone or together with P19 by spot infiltration of 5-week-old plants. Four of five tobacco cultivars, namely I-64, TI-95, Xanthi and Petite Havana H4, showed a marked decrease in antibody expression at day 6 when co-expressed with P19, compared to antibody expression without P19 (Figure 5a). Conversely, the co-expression of P19 with 103 mAb in N. tabacum cv. Little Crittenden (LCR) resulted in a significant boost in trastuzumab expression (Figure 5a). Furthermore, the tobacco cultivars that showed a decrease in antibody expression in the presence of P19 also showed a marked discoloration of the treated areas after 3 days, while the infiltrated areas of N. benthamiana and N. tabacum cv. LCR were unaffected up to 10-day postinfection (shown in Figure 5c at 5 d.p.i). N. benthamiana shows no necrosis after Agroinfiltration of P19 even at OD600 = 0.8 (data not shown). In reports where necrosis is observed in N. benthamiana with P19 (Hsieh et al., 2009; Voinnet et al., 1999), other viral proteins were also present in the treatments. We conclude that N. tabacum cv. Little Crittenden and N. benthamiana can be used as a host for transgenic production of recombinant proteins using P19.

Figure 5.

Differential response of Nicotiana tabacum and Nicotiana benthamiana to P19. Western blot analysis showing transient expression of trastuzumab alone or together with P19 in N. benthamiana and five different N. tabacum cultivars (a) and N. tabacum crosses (b). Blots were probed with a 1 : 10 000 mix of alkaline phosphatase conjugated to anti-human γ- and κ-chain antibodies. Plants were treated by spot infiltration. Samples were pooled by combining three infiltrated spots. Each lane represents a pooled sample. Co-expression of 103 mAb with P19 resulted in a significant reduction in antibody expression in all tobacco cultivars except in LCR. The drop in antibody expression reflects tissue death cause by the induction of the hypersensitive response. Crosses between N. tabacum I-64 and LCR, and N. benthamiana and N. tabacum I-64 (NBT) showed a similar drop in antibody expression when P19 was co-expressed with 103 mAb (b). All Nicotiana species that showed a drop in antibody expression in the presence of P19 also displayed discoloration at the site of infection, which lead to necrosis about day 3 days postinfection (c). Images here show infiltrated spots at 5-day postinfection. TI-95, N. tabacum cv. TI-95; XAN, N. tabacum cv. Xanthi; PH, N. tabacum cv. Petite Havana H4; LCR, N. tabacum cv. Little Crittenden; BEN, N. benthamiana; NBT, N. benthamian × N. tabacum cv. I-64.

Reciprocal crosses were made between tobacco cultivars I-64 and LCR to look at the manner in which induction of the hypersensitive response by P19 is inherited in N. tabacum. We also tested the induction of this response in a sterile cross between N. benthamiana and N. tabacum, named NBT (F. Garabagi, unpublished). Five-week-old seedlings were infiltrated with 103 mAb, with or without P19. The level of trastuzumab was reduced in all the crosses at 5 d.p.i. when 103 mAb was co-expressed with P19 (Figure 5b). This reduction in antibody expression correlated with discoloration of the treated leaves, followed by necrosis (Figure 5c). These results indicate that the putative R gene responsible for triggering the HR is nuclear and that LCR is homozygous recessive for that gene.

P19 does not affect the silencing of fucosyltransferase and xylosyltransferase activity in RNAi-based glycomodified Nicotiana benthamiana

RNAi-based silencing is a commonly used method to modify the N-glycosylation pattern towards human-like structures in plants (Cox et al., 2006; Sourrouille et al., 2008; Strasser et al., 2008). This strategy was also applied to eliminate plant-specific N-glycan residues (i.e. xylose and core α1,3 fucoce) in N. benthamiana (ΔXTFT) by the down-regulation of the respective enzymes fucosyltransferase and xylosyltransferase (Cox et al., 2006; Sourrouille et al., 2008; Strasser et al., 2008). Here, we set out to determine possible adverse effects of P19 in such plants. Thus, ΔXTFT mutants were vacuum infiltrated with the mAb103 vectors with or without P19. Trastuzumab was purified 5-day postinfiltration and analysed by LC–ESI–MS to determine the N-glycosylation pattern (Figure 6b,c). Interestingly, both 103 mAb versions carried an identical N-glycosylation profile, lacking plant-specific oligosaccharides. This suggests that P19 does not affect the RNAi-silencing pathway responsible for the silencing of FT and XT in ΔXTFT at this time point, while it interferes with the gene-silencing pathway that reduces transgene expression. In another experiment, trastuzumab was purified from a 21-day time-course expressed with 103 mAb + P19 and subjected to Western blot analysis with an anti-HRP probe to look at the presence of plant-specific sugars, α1,3-fucose and β1,2-xylose (Strasser et al., 2008). We found that P19 brings about a minor reversal of the silencing of fucosyltransferase and xylosyltransferase in ΔXTFT N. benthamiana expression host at around 11 dpi (Figure S1). Thus, we conclude that P19 can be used for boosting recombinant protein levels expressed in an RNAi-based expression host without altering the protein's glycan profile up to 9 days postinfiltration.

Figure 6.

N-glycan profiles of 103 mAb expressed in Nicotiana benthamiana WT (a) and in ∆XTFT without (b) and with P19 (c). N-glycan analyses were carried out by liquid–chromatography–electrospray ionization–mass spectrometry (LC–ESI–MS) of tryptic glycopeptides as described previously (Stadlmann et al., 2008; Strasser et al., 2008). Note, that incomplete tryptic digest results in the generation of two glycopeptides that differ by 482 Da. Glycopeptide 1 is indicated with asterisks (*). See http://www.proglycan.com for N-glycan abbreviations.


In this study, we approached three important issues for the application of P19 to improve recombinant protein production in plants: P19's ability to sufficiently boost recombinant protein expression; its compatibility with the expression host; interference with the state of gene silencing in RNAi-based glycomodified plants. Our results indicate that all three requirements are met and that P19 can be effectively utilized for transient expression of recombinant glycoproteins in RNAi-based expression hosts.

The highest antibody expression reports using suppressors of gene silencing to date have been on transient expression of mAb 2G12 with P19 at 400 mg/kg FW (Saxena et al., 2011) and mAb C5-1 with HcPro at 757 mg/kg FW (Vezina et al., 2009). Both of these mAbs were retained in the ER by the addition of auxiliary C-terminal tags. In the context of biosimilar therapeutic antibodies, however, ER retention signals are problematic because they add extra amino acids to the primary sequence of the innovator protein. In addition, ER typical oligomannisidic N-glycsoylation is uncommon for most therapeutic proteins and thus unwanted.

Here, we report P19-enhanced expression of the model therapeutic mAb trastuzumab that is targeted to the apoplast using classical binary vectors at about 2.3% of TSP, or approximately 460 mg/kg FW. We found that regulatory genetic elements used in recombinant constructs determine whether or not P19 can boost expression. This was demonstrated by co-expressing P19 together with the heavy and light chains of trastuzumab cloned in five different expression vectors containing different 5′ and 3′ UTRs (Figure 1). Our results indicate that transcripts with at least one virus-derived UTR, such as in 103, 104 and 105 mAb, were boosted by P19. This suggests that the transcripts of these expression cassettes are subjected to RNAi silencing, albeit to different extents and therefore are boosted in the presence of a suppressor of a gene silencing. Semi-quantitative RT-PCR analysis of the HC and LC of trastuzumab corroborated this notion. The extent of gene silencing in vector 105 mAb, which contained a dicot 3′UTR was less than that of vectors 103 and 104 mAb, with nondicot 3′UTRs. In contrast, transcripts that only contained plant-derived UTRs, such as 106 and 102 mAb, were unaffected by P19, suggesting they were not subjected to any significant RNAi silencing during the observation period (Figures 2d and 3a). Deconstructed TMV vectors contain a potent suppressor of gene silencing that binds short interfering RNAs much like P19 (Csorba et al., 2007). This is most likely why the co-infiltration of suppressors of silencing does not have an effect on expression levels from these vectors (Figure 3b). The fact that the same signal peptide was used in all expression cassettes suggests that this element did not affect P19 functionality. These findings may have direct implications in the design of expression vectors.

It is apparent from several reports on constitutive transgenic expression of P19 that the onset of adverse effects occurs only when the protein is used in high titres, especially because high-level constitutive expression of P19 is known to be lethal in Arabidopsis (Dunoyer et al., 2004). This dose-dependent functionality of P19 is supported by the fact that transgenic tobacco cv. Xanthi is tolerant to P19 at low levels (Siddiqui et al., 2008), while the protein generates necrosis in tobacco cvs. Samsun and NC95 leaves when transiently expressed at high titres (Angel et al., 2011). Inducible expression systems that are capable of producing high-levels of recombinant protein have been employed to circumvent the lethality that is caused by producing high titres of P19 in transgenic systems. Nonetheless, unfavourable effects such as malformed leaves and flowers appear when P19 is expressed with such systems at high-levels, such as with the pOp/LhG4 transactivation systems described by Stav et al. (2009). Saxena et al. (2011) present data on a modified P19 that does not cause adverse physiological side effects, but displays a mediocre boosting capability because of the overall lower level of expression that is associated with transgenic systems. As transgenes generally express higher in transient as opposed to transgenic settings (Garabagi et al., 2012), we showed that the amount of P19 protein produced by transient infiltration at concentrations of OD600 = 0.02 and lower did not significantly enhance trastuzumab levels. On the other hand, higher P19 concentrations (OD600 = 0.2) caused a 15-fold increase in antibody levels (Figure 4a). Our results thus support the idea of dose-dependent functionality of P19 in the context of boosting recombinant protein expression. However, the titre at which P19 exerts its boosting effect causes a hypersensitive response in most tobacco cultivars (Figure 5c).

Recent reports on the induction of the hypersensitive response in N. tabacum cvs. Samsun and NC95 describe a physiological response that involves local induction of RNAi silencing against the viral RNA, and a necrotic defence response at the site of infection (Angel et al., 2011). This response is likely triggered by the product of a putative R gene (Angel et al., 2011). Conely et al. (2011) had speculated on the effect of suppressors of gene-silencing proteins in different Nicotiana hosts, but this notion was not tested to this date. We identified a tobacco cultivar, Little Crittenden, which did not trigger a hypersensitive response when exposed to high titres of P19 (Figure 5a,c). All other tested tobacco cultivars showed a marked decrease in antibody production in the presence of P19 compared to the expression of the antibody vectors alone, likely due to cell death that is triggered by the hypersensitive response. Our results also indicate a dominant mode of inheritance for this trait, also in accordance with previously described characteristics of R genes (Moffett, 2009) and that the LCR cultivar has a recessive mutation in this gene. We speculate that this tobacco genotype and others that have similar R gene mutations will lend themselves to transgenic expression of recombinant proteins using P19 with a system capable of generating high titres of the protein. LCR can be used as a model for determining the number of genes involved in the hypersensitive response to P19.

Despite the ability of plants to carry out complex glycosylation, a possible bottleneck for their use as a versatile expression platform for therapeutic antibodies is the presence of plant-specific N-glycan residues, that is xylose and core α1,3-fucose. Such nonmammalian oligosaccharides might change the biological activity of a given protein or might even induce unwanted adverse side effects upon therapeutic application. As expected, during our experiments, plant-typical N-glycosylation was detected on trastuzumab. To circumvent this problem, RNAi technology had been used to generate a plant line that lacks unwanted plant-specific N-glycan residues (Strasser et al., 2008). Monoclonal antibodies produced in this mutant (∆XTFT) carry complex human-like N-glycans lacking plant-specific glycosylation. Moreover, mAbs with such a glycoengineered profile have also shown increased effector functions compared with their mammalian cell-derived counterparts (Forthal et al., 2010; Zeitlin et al., 2011). Thus, such glycosylation mutant plants may serve as valuable expression platforms for the generation of therapeutic mAbs. However, whether the use of P19 would perturb the silencing of XT and FT in ΔXTFT mutants has not been investigated yet. Based on previous reports on the cellular targets of P19, including a documented case of interference with the siRNA-silencing pathway in a transgenic plant line (Ahn et al., 2011), we expected P19 to interfere with the RNAi-induced gene silencing of XT and FT in glycomodified N. benthamiana. Surprisingly, mAbs expressed in ΔXTFT with and without P19 were in both cases completely devoid of xylose and fucose residues within the harvesting period of 5–7 days dpi (see Figure S1). The virtually identical glycan profiles of the two antibodies indicate that the silencing of FT and XT is unaffected by P19 (Figure 6). Thus, we conclude that P19 can be used effectively in combination with RNAi-based glycomodified hosts for the production of therapeutic glycoproteins in both transient and stable expression systems.

Experimental procedures

Plant expression vectors

Except for the 35S promoter [PCR-amplified from plasmid pMM29 (McLean et al., 2007)] and the nopaline synthase 3′ UTR and terminator region (PCR-amplified from plasmid pCAMBIA-1305.2; GenBank accession: AF354046), the genetic elements for five expression cassettes, namely 102–106, were synthesized de novo and subcloned into the T-DNA region of pICH14011, creating plasmids p102–p106 for producing recombinant proteins in Nicotiana species. The structures of the expression cassettes are depicted in Figure 1. Expression cassettes 103–105 carry an 800-bp fragment containing a doubled 35S promoter and 5′UTR of the Cauliflower mosaic virus, while 106 contains 1040 bps of the promoter and 5′ UTR of ribulose bisphosphate carboxylase (rbc) small subunit gene from Chrysanthemum morifolium (GenBank accession: AY163904.1). Cassette 103 contains 280 bps of the 3′ UTR and terminator sequences from the nopaline synthase (nos) gene of Agrobacterium (GenBank accession: V00087.1), cassette 104 contains 485 bps of the 3′ UTR and terminator sequences from the osmotin (osm) gene of Oryza sativa (Genbank accession: L76377.1), and cassettes 105 and 106 carry 496 bps of the 3′ UTR and terminator sequences from the rbc gene of C. morifolium. Expression cassette 102 mAb contains both antibody chains on one T-DNA. The heavy chain was driven by a 1403 bp actin2 promoter and 5′ UTR from Arabidopsis thaliana (GenBank accession: NM_112764), coupled with a 1109-bp fragment containing the 3′ UTR and terminator of the same gene. The light chain was driven by a 1025-bp chimeric octopine and mannopine synthase promoter (GenBank accession: EU181146.1) fused to the 379 bp 5′UTR, and 546 bps of the 3′ UTRs and terminator region of A. thaliana ubiquitin 10 (ubq10) gene (GenBank accession: L05361). The heavy and light chains of trastuzumab were also cloned in TMV/PVX-based vectors from ICON genetics, as described by Grohs et al. (2010). In all cases, the coding sequences for the heavy and light chains were fused to the signal sequence from the basic chitinase gene of A. thaliana (GenBank accession: AY054628) for secretion into the apoplast.

The coding sequence for the P19 protein from Tomato bushy stunt virus (TBSV; GenBank accession: M21958) was codon optimized for expression in N. benthamiana, synthesized de novo and cloned in cassette 103. The heavy and light chains of trastuzumab (Grohs et al., 2010) were cloned separately in expression cassettes 103–106. All protein sequences were codon optimized for expression in N. benthamiana.

Bacterial transformation and culture

Competent Agrobacterium tumefaciens A136 cells were transformed with the above-mentioned expression cassettes by a standard heat-shock method. Bacterial cultures were grown overnight at 28 °C in YEP medium (10 g Bacto peptone, 10 g yeast extract, and 5 g sodium chloride per litre, pH 7.0) supplemented with antibiotics. Unless otherwise stated, dense overnight cultures (up to OD600 = 2.5) for each vector were adjusted to an OD600 of 0.2 in Agro-infiltration buffer (AIB) containing 10 mm 2-(4-morpholino)-ethanesulphonic acid (MES), 10 mm MgSO4, pH 5.5 and mixed together to create an Agrobacterium infiltration cocktail (AIC) for plant treatment.

Transient expression assay and sample preparation

Plants were grown in the greenhouse and fed high nitrate fertilizer (N : P : K = 20 : 8 : 20) daily at 1 g/L (Plant Products, Brampton, ON, Canada), adjusted to pH 6.0 with 20% H3PO4. To transiently express trastuzumab, an AIC containing two Agrobacterium strains, each harbouring one of the antibody chains, were used to infiltrate plant leaves. All plants were treated at the 4- to 6-week stage either by spot or by whole-plant infiltration (Garabagi et al., 2012). Shortly after treatment, the plants were placed back in the greenhouse for a certain period of time prior to harvest, depending on the expression vector. During this period, plants were only fed water. When harvesting wholly infiltrated plants, newly emerged leaves were discarded and the infiltrated leaves were separated from the stems and stored at −80 °C until further processing. For spot infiltrations, 100 mg of leaf tissue from the infiltrated area was weighed and stored at −80 °C until further processing.

Approximately, 100 mg of leaf tissue was mixed with 300 μL of extraction buffer containing phosphate-buffered saline (PBS: 8 g NaCl, 0.2 g KCl, 1.44 g Na2HPO4 and 0.24 g KH2PO4 per litre) and 10 mm EDTA, pH 6.8 (PBSE buffer). Samples were disrupted in a 2-mL microfuge tube containing two stainless steel ball bearings for 5 min using a TissueLyser (Qiagen, Toronto, Ontario, Canada) to prepare a crude protein extract, which was aliquoted into small volumes and stored at −20 °C.

Western blotting

A frozen aliquot was thawed and spun in a refrigerated bench-top centrifuge at >18 000 g for 1 min to clarify the crude extract for protein quantitation. Either a Bradford or a BCA (Pierce) assay was used to determine protein concentration of the once-thawed crude extracts. For antibody detection, thirty micrograms of total soluble protein (TSP) per sample was loaded in each well on an 8% SDS-polyacrylamide gel. The separated proteins were blotted on a polyvinylidene fluoride (PVDF) membrane and probed for antibody presence with a mix of alkaline phosphatase conjugated anti-human γ and κ antibodies (Cat# A3312 and A3813; Sigma-Aldrich, St. Louis, MO), diluted to 1 : 10 000 in PBS (pH 7.4) using NBT/BCIP (Cat# 34042; Thermo Fisher Scientific, Waltham, MA) as substrate. Blots were developed for 2–5 min, depending on the experiment.

Detection of α1,3-fucose and β1,2-xylose was carried out as described by Strasser et al. (2008). Briefly, 1 μg of reduced purified antibody per sample was loaded in each well on an 8% SDS-polyacrylamide gel. The separated proteins were blotted on a polyvinylidene fluoride (PVDF) membrane and probed for the presence of α1,3-fucose and β1,2-xylose sugars with a 1 : 15 000 dilution of an anti-HRP polyclonal primary antibody (Cat#P7899; Sigma-Aldrich), a 1 : 100 000 dilution of an HRP-conjugated goat anti-rabbit secondary antibody (Jackson ImmunoResearch Code 11-035-003) and SuperSignal West Pico chemiluminescent (Cat# 34077; Thermo Fisher Scientific) as substrate.


Enzyme-linked immunosorbent assay (ELISA) was used to quantitate the amount of antibody present in the crude protein extract of treated plants. ELISA plates were coated overnight at 4 °C with a mouse polyclonal anti-human IgG1 (Cat# I5885; Sigma-Aldrich) capture antibody at 0.6 μg/mL in PBS. Human IgG1 standard (Cat# 16-16-090707; Athens Research and Technology, Athens, GA) spiked in 5 μg of untreated crude protein extract was used as standard. The standard curve was generated using human IgG1, which allowed for antibody detection over a range spanning three orders of magnitude (0.1–100 ng/well). Crude extract from each treated sample was loaded on an ELISA plate as twofold dilutions in triplicate. Final antibody concentration was calculated by averaging the mean antibody concentration for three crude extract dilutions. A second anti-human antibody conjugated to HRP from rabbit (Cat # ab6759; Abcam, Cambridge, MA) was used for detection, using TMB-ELISA (Cat# 34022; Thermo Fisher Scientific) as substrate. The plates were allowed to develop for 15 min before the reaction was stopped with 0.5 m H2SO4.

Antibody purification

Antibodies were purified essentially as described by Grohs et al. (2010).

Semi-quantitative RT-PCR

RNA was extracted from 100 mg of plant tissue using an RNA extraction Kit (Qiagen). One microgram of total RNA was used to generate cDNA with SuperScriptII reverse transcriptase (Invitrogen; Life Technologies subsidiary, Burlington, ON, Canada); 1/20 of the RT reaction was used as template for PCR, which was carried out for 20 cycles with specific primers (Table 2). The primers used for TBSV P19 (P19-F and P19-R), and for the heavy and light chains (abc-F, HC-R, and LC-R) of trastuzumab, amplified the full length mRNA, while the primers for Actin (Act-F and Act-R) amplified a 900-bp internal region of the gene. Owing to identical 5′ leader sequences, the same forward primer was used to amplify both HC and LC.

Table 2. List of primers used for RT-PCR
Primer namePrimer sequence

N-glycosylation analyses

N-glycan analyses of purified mAbs were carried out by liquid–chromatography–electrospray ionization–mass spectrometry (LC–ESI–MS) of tryptic glycopeptides as recently described (Stadlmann et al., 2008). Briefly, the purified samples were submitted to reducing SDS PAGE and the 55-kD band corresponding to the HC was cut from the gel, S-alkylated, digested with trypsin, eluted from the gel fragment with 50% acetonitrile and separated on a Biobasic C18 column (150 × 0.32 mm; Thermo Fisher) with a gradient of 1%–80% acetonitrile containing 65 mm ammonium formate pH 3.0. Positive ions were detected with a Q TOF Ultima Global mass spectrometer (Waters, Milford, MA). Summed and deconvoluted spectra of the glycopeptides elution range were used for identification of glycoforms. This method generates two glycopeptides that differ by 482 Da (glycopeptide 1, EEQYNSTYR; glycopeptide 2 TKPREEQYNSTYR).


ΔXTFT N. benthamiana seeds were kindly provided by Dr. Herta Steinkellner, Department of Applied Genetics and Cell Biology, BOKU-Vienna, Austria. We would also like to thank Dr. Steinkellner for a critical review of the manuscript. Wild-type N. tabacum seeds were kindly provided by Dr. Jim Todd, OMAFRA, Simcoe Resource Centre, Canada. This project was funded by grants from OMAFRA, NSERC and CRC. We would like to acknowledge Josephine Grass (BOKU-VIENNA) for N-glycan analyses, and John Teat and Jessica Rouleau (PlantForm, Canada) for their help with the P19 time-course. The authors have no conflict of interest to declare.