From gene to harvest: insights into upstream ... - Wiley Online Library

Plant Biotechnology Journal (2015) 13, pp. 1094–1105

doi: 10.1111/pbi.12438

Review article

From gene to harvest: insights into upstream process development for the GMP production of a monoclonal antibody in transgenic tobacco plants Markus Sack1,*, Thomas Rademacher2, Holger Spiegel2, Alexander Boes2, Stephan Hellwig2, Juergen Drossard2, Eva Stoger3 and Rainer Fischer1,2 1

Institute for Molecular Biotechnology, RWTH Aachen University, Aachen, Germany

2

Fraunhofer Institute for Molecular Biology and Applied Ecology IME, Aachen, Germany

3

Department of Applied Genetics and Cell Biology (IAGZ), University of Natural Resources and Life Sciences, Vienna, Austria

Received 11 March 2015; revised 12 June 2015; accepted 16 June 2015. *Correspondence (Tel +49-241-608513440; fax +49-241-6085-10000; email [email protected]) MS and TR contributed equally to this work

Keywords: upstream process development, GMP/GACP, plant-made pharmaceuticals, 2G12, HIV, transgenic tobacco, manufacturing licence, glasshouse.

Summary The EU Sixth Framework Programme Integrated Project ‘Pharma-Planta’ developed an approved manufacturing process for recombinant plant-made pharmaceutical proteins (PMPs) using the human HIV-neutralizing monoclonal antibody 2G12 as a case study. In contrast to the wellestablished Chinese hamster ovary platform, which has been used for the production of therapeutic antibodies for nearly 30 years, only draft regulations were initially available covering the production of recombinant proteins in transgenic tobacco plants. Whereas recombinant proteins produced in animal cells are secreted into the culture medium during fermentation in bioreactors, intact plants grown under nonsterile conditions in a glasshouse environment provide various ‘plant-specific’ regulatory and technical challenges for the development of a process suitable for the acquisition of a manufacturing licence for clinical phase I trials. During upstream process development, several generic steps were addressed (e.g. plant transformation and screening, seed bank generation, genetic stability, host plant uniformity) as well as productspecific aspects (e.g. product quantity). This report summarizes the efforts undertaken to analyse and define the procedures for the GMP/GACP-compliant upstream production of 2G12 in transgenic tobacco plants from gene to harvest, including the design of expression constructs, plant transformation, the generation of production lines, master and working seed banks and the detailed investigation of cultivation and harvesting parameters and their impact on biomass, product yield and intra/interbatch variability. The resulting procedures were successfully translated into a prototypic manufacturing process that has been approved by the German competent authority.

Introduction Plant-based expression systems are robust, scalable and costefficient platforms for the production of recombinant proteins of pharmaceutical interest (Fischer et al., 2013; Waheed et al., 2012). Under the umbrella term molecular farming, the plantbased production of vaccine subunits and virus-like particles (D’Aoust et al., 2010; Daniell et al., 2009; Farrance et al., 2011; Landry et al., 2010), therapeutic enzymes (Aviezer et al., 2009) and antibodies (Marzi et al., 2012) has been achieved using transient expression (Rybicki, 2010; Yusibov et al., 2011) and stable transformation in plants or plant cells (Fischer et al., 2004; Stoger et al., 2014). Even though several studies have demonstrated that plants can produce high-quality complex pharmaceutical proteins with high yields (Paul and Ma, 2011), the translation of these promising plant-made pharmaceuticals (PMPs) from proof of concept (Fischer and Schillberg, 2006) into an approved and economically feasible manufacturing process is a hurdle in the road towards commercialization (Sparrow et al., 2007). 1094

One major objective of the EU Sixth Framework Programme Integrated Project ‘Pharma-Planta’ was to develop a regulatory pathway suitable for manufacturing the HIV-neutralizing human monoclonal antibody 2G12 (Stiegler et al., 2002; Trkola et al., 1996) for a first-in-human double–blind, placebo-controlled, randomized, dose-escalation phase I safety study of a single vaginal administration. The production of recombinant proteins in transgenic plants under nonsterile conditions in contained environments such as glasshouses is associated with several logistic, scientific and regulatory challenges throughout the process chain, including tasks such as the generation and characterization of stable plant lines, master and working seed banks, and the development of robust standard operating procedures (SOPs) for sufficiently reproducible cultivation and harvesting conditions, as well as downstream steps such as the efficient extraction of soluble proteins from plant biomass (Buyel and Fischer, 2013), extract clarification (Buyel and Fischer, 2014) and host cell protein removal. Although the product characterization and quality control steps (e.g. host cell DNA and protein removal) could generally be

ª 2015 Society for Experimental Biology, Association of Applied Biologists and John Wiley & Sons Ltd

GMP/GACP upstream production of a human IgG in tobacco 1095 deduced from existing guidelines and processes for biopharmaceutical production in other expression systems, it was necessary to develop the upstream process from first principles in accordance with the European Medicines Agency (EMA) guideline ‘Points to consider on quality aspect of medicinal products containing active substances produced by stable transgene expression in higher plants’, which was available as a draft at the beginning of the project (CPMP/BWP/764/02) and later updated to a final document (CPMP/BWP/48316/06). In this article, we describe the relevant steps and analysis procedures used to define a GMP-compliant upstream process from gene to harvest.

Results and discussion The host plant Tobacco (Nicotiana tabacum L. Petit Havana cultivar SR1) was chosen as the host plant because it was one of the first genetically modified plants (Deblock et al., 1984), and the expression of pharmaceutical proteins in different subcellular compartments has been achieved using a variety of expression vectors and transformation methods (Ma et al., 2003). The inbred variety has a good growth characteristic in the glasshouse and produces up to 50 000 seeds per plant.

Transgene expression constructs The HIV-neutralizing antibody 2G12 is a human gp120-specific monoclonal antibody that has already been investigated in a phase I study in asymptomatic HIV patients using clinical material produced in Chinese hamster ovary (CHO) cells (Stiegler et al., 2002). The safety and tolerated dose has already been established and this was an important criterion for selecting this antibody for expression in tobacco. The 2G12 coding regions were kindly provided by Polymun Scientific (Vienna). The antibody chains were cloned separately in pTRA expression vectors, verified by sequencing and tested for functionality after transient expression and purification by protein-A chromatography. The coding regions for the heavy and light chains were then cloned in series within a single T-DNA to facilitate the generation of homozygous and nonsegregating double transgenic lines, reducing the workload and development time compared with introducing the genes separately. The pat and DsRed genes were also included in the same T-DNA for selection and visible screening, the latter supporting the selection of transgenic events after transformation and facilitating the breeding steps for production line development. Furthermore, we hypothesized that DsRed fluorescence could be used to predict the expression level of the antibody because the DsRed gene would be subjected to the same position effects, thus facilitating the early identification and selection of elite events. We deliberately placed the DsRed gene under the control of the same regulatory elements (CaMV 35S promoter/3ʹ UTR and TEV 5ʹ UTR) to provide opportunities for the detection of epigenetic effects that affect transgene expression. To avoid overloading the endoplasmic reticulum, we targeted the DsRed protein to the plastids. All cloning steps were confirmed by restriction endonuclease and PCR analysis. The expression of a functional 2G12 antibody using the expression vector pTRAp2G12-Ds (Figure 1a, Table 1) was validated by Agrobacteriummediated transient expression and subsequent detailed analysis of the product purified by protein-A chromatography (data not shown) as previously described (Holland et al., 2010; Rademacher et al., 2008). We also generated 2G12-producing transgenic lines

by crossing plants that were separately transformed with the heavy and light chain genes without the DsRed gene, and these did not produce more antibody than the lines generated using the tandem construct approach, indicating that the coexpression of the fluorescent marker protein is unlikely to negatively affect the antibody yield.

Generation and screening of plant lines Stably transformed tobacco plants were generated using a modified leaf-disc method and recombinant Agrobacterium tumefaciens carrying the validated pTRAp-2G12-Ds plasmid. Transgenic shoots (Figure 1b,c) and T0 transformants (Figure 1d) were initially screened for macroscopic DsRed fluorescence. Plants showing high and uniform DsRed expression were then analysed for the presence of 2G12HC and 2G12LC by Western blot using antibodies specific for the human heavy and light chains (data not shown). We sought lines producing an excess of the antibody light chain to avoid recovery of partially assembled H2 and H2L antibody species by protein-A chromatography. A total of nine events were selected for self-pollination. The T1 generation was analysed for the segregation of DsRed fluorescence and herbicide resistance to identify lines with a single transgene locus. These lines were then analysed for 2G12 accumulation, and two lines (pGFD2 and pGFD13) were selected as potential candidates for the establishment of seed banks. Four siblings of each primary event were taken into the next generation by self-pollination, and at least 40 descendants were analysed for 2G12 accumulation by surface plasmon resonance (SPR) spectroscopy, as well as the uniformity of macroscopic DsRed fluorescence, the morphological phenotype, growth behaviour and vigour, seed production and germination rate. The zygosity of the transgene locus and the stability of DsDed expression in subsequent generations were followed by segregation analysis on selective media and wholeplant macroscopic fluorescence analysis (Figure 1f). We have found that macroscopic DsRed fluorescence is a valuable tool for the development of upstream processing because it provides additional quality control opportunities for the identification, selection and validation of transgenic production lines and helps to identify epigenetic effects (Figure 1e, and S1). The method is inexpensive and straightforward and provides extremely valuable information.

Generation of master and working seed banks In the T4 generation, three plants from line pGFD13 and four plants from line pGFD2 were selected as candidates for producing the seeds for the master seed bank. The plants were grown in a physically isolated and contained phytotron chamber. Seeds were collected separately from each plant, the subsequent generation was analysed as described above, and the T5 master seed bank was finally derived from a single self-pollinated plant (master seed bank = seeds from plant pGFD13-9-17-25-11). The T6 working seed bank was then derived from 25 self-pollinated plants grown from the master seed bank. These plants were grown in a physically separated and contained glasshouse chamber. The numbered aliquots were stored in a dedicated, locked cabinet in a controlled-access room (GMP Released Goods Storage). The temperature was controlled at a setpoint of 20 °C with alarm limits at 15–25 °C. Relative humidity was not controlled. Temperature and relative humidity were constantly monitored. In the last 5 years, the medium temperature was 19–21 °C and the medium relative humidity was 40%–50% and was at noncondensing levels at all times.

ª 2015 Society for Experimental Biology, Association of Applied Biologists and John Wiley & Sons Ltd, Plant Biotechnology Journal, 13, 1094–1105

1096 Markus Sack et al. (a)

(b)

(c)

(d)

(e)

(f)

(g)

Figure 1 pTRAp-2G12-Ds expression vector and DsRed as a visual marker. (a) Map of the plant expression vector pTRAp-2G12-Ds. Open reading frames are shown as arrows. Sequences transferred to the plant genome are shown in green. LB and RB: left and right border of T-DNA from A. tumefaciens nopaline Ti-plasmid, Pnos: nopaline synthase gene promoter, pat: Streptomyces hygroscopicus gene for phosphinothricin acetyltranferase, SAR: scaffold attachment region of the N. tabacum Rb7 gene, P35SS: CaMV 35S promoter with duplicated transcriptional enhancer, TL: 5ʹ untranslated region from Tobacco etch virus, pA35SS: CaMV 35S polyadenylation signal, SPg: signal peptide sequence of the human Ig gamma chain, 2G12HC: 2G12 gamma-1 heavy chain gene, SPk: signal peptide sequence of the human Ig kappa chain, 2G12LC: 2G12 kappa light chain gene, TP: H. vulgare GBSSI transit peptide sequence, DsRed: Discosoma sp. red fluorescent protein gene (R2G mutant), RK2 ori: origin of replication, bla: b-lactamase gene, ColE1 ori: origin of replication. (b) DsRed fluorescent shoots after regeneration of primary transformants (T0).(c) Visual DsRed inspection of T0 plantlets growing in bags. (d) Wild-type (right) and pTRAp-2G12-Ds transgenic (left) tobacco plants inspected under green light using a red filter for visualization of DsRed fluorescence. (e) DsRed transgenic A. thaliana plants showing different nonhomogeneous red fluorescent phenotypes. (f) Homogenous DsRed fluorescence of master seed bank (T5, top right), working seed bank (T6, top left) and three progenies of the working seed bank (T7, bottom). (g) DsRed fluorescence in the working seed bank and three subsequent generations (T7, T8, T9).

Characterization of master and working seed banks The T5 master seed bank, the T6 working seed bank/production generation (which are the same for tobacco) and the T7 generation (which are one generation beyond the production generation) were characterized in detail to confirm their genetic stability. The transcription units of the antibody heavy/light chain and DsRed genes were amplified by PCR using primers annealing in the CaMV 35S promotor and terminator region. The template was genomic DNA from three individual plants grown from the master and working seed banks, respectively. Only PCR products corresponding to the expected sizes (HC = 1673 bp; LC = 959 bp; DsRed = 1158 bp) were observed (Figure S2a). The integrity of the antibody gamma and kappa chain PCR products was confirmed by sequencing and Southern blot analysis (NcoI and XbaI fragments), and the latter also confirmed that the gene dosage was identical in the T5–T7 generations. The copy number was determined by comparing the signal intensity to a plasmid control, revealing two copies of the HC sequence, six copies of the LC sequence (Figure S2b), one copy of the pat

marker gene and seven copies of DsRed (data not shown). The integrity of the transcription units was further corroborated by northern blot analysis, which showed that the transcripts for the 2G12 heavy and light chains were of a uniform and distinct size (Figure S2c). Segregation analysis suggested a single transgene locus (Table S1), and this was confirmed by additional self-pollination and back-crossing into nontransgenic tobacco cultivars (data not shown). No segregation of herbicide resistance was observed in the T3 and subsequent generations (Table S1). The 2G12 antibody genes, macroscopic DsRed fluorescence and herbicide resistance co-segregated in all analysed samples. The antibody yield, plant height and leaf biomass were directly compared in T1–T6 plants grown at the same time under identical conditions. The results showed that the general plant phenotype and growth characteristics were normal and uniform, consistent with the observation that no significant differences in the plant height and leaf biomass were evident. The antibody accumulation increased from the T1 to the T3 generation as anticipated from the increasing gene dosage, and remained stable in subsequent generations


1102 Markus Sack et al.

Figure 6 Development of biomass, 2G12 levels and yield. The development of leaf biomass, antibody accumulation and yield per plant were monitored over a period of 18 days starting from 30 days postsowing (dps) for topped and flowering plants (five plants for each type and time point). Triangles: 2G12 accumulation (lg/ mL), solid (topped plants), open (flowering plants). Squares: biomass (g) solid (topped plants), open (flowering plants). Circles: yield (lg/plant), solid (topped plants), open (flowering plants).

to the ebb and flow tables. In order to produce a homogenous batch, care was taken to use plantlets of the same size, the remainder being discarded. Because freshly transplanted plants are sensitive to full sunlight, the chambers were shaded during transplanting and for one additional day if required. The plants were watered with tap water by hand with a fine spray for the following few days. When the first roots reached the bottom of the pot, irradiation-dependent automated fertigation (irrigation with a fertilizer solution) was started with 0.1% Ferty 2 MEGA (16 + 6+26 + (3.4)), a highquality fertilizer consisting of fully water-soluble minerals including trace elements. The irrigation water in the reservoirs was controlled and adjusted twice a week. The tobacco plants were cultivated without any additional shading and received supplemental light of 20 klux when the external irradiance dropped below 35 klux (Table 2). When the leaves expanded over the rim of the pot, the plants were spaced to the final distance (eight plants/m²). At a height of 30–40 cm, all plants were supported with a bamboo stick to prevent tipping over. Depending on the season, the first plants started to produce flowers ~40 days after sowing. To prevent flowering, the inflorescences were removed by breaking them manually (topping) when the first coloured petal was visible. From this point on, the plants were checked daily and suckers were removed until harvesting. The onset of flowering was also an indicator for batch homogeneity (more than 90% should have set flowers in less than 1 week) and was used as timer for harvesting. The batch was harvested no later than 2 weeks after the first plants started to flower. Plants with unusually delayed flowering were excluded from the harvest. Approximately 1 week ahead of the planned harvesting date, 7% of the plants were analysed individually for biomass and 2G12 accumulation for quality control. Before the batch was finally approved for processing, the plants were checked for macroscopic DsRed fluorescence and nonfluorescent plants were discarded. Apart from the small round green spots with a diameter of a few millimetres only (Figure S1b), we so far have observed only a single case where a larger leaf area was nonfluorescent (Figure S1c) for >30 000 plants over a period of 7 years. Preliminary analysis indicated that antibody accumulation was also reduced in the green patches (data not shown). The leaves were hand-picked and inspected, and only healthy leaves were used for further processing. An additional set of measures (Tables S4 and S5) was defined to ensure the identity of the plant material and to avoid and monitor contamination with pathogens.

Table 2 Standardized cultivation conditions Parameter

Value

Temperature

25 °C/22 °C day/

Relative humidity

70%

Light

16 h/8 h day/night

Sowing compost

Einheitserde type

Supplier (if applicable)

night

(min 20 klux) (trays) Potting compost (2.5-L pots)

VM Einheitserde type T 1.5

Balster Einheitserdewerk, €ndenberg, Germany Fro Balster Einheitserdewerk, €ndenberg, Germany Fro

Fertigation

0.1% Ferty 2 MEGA

€ngemittel, Planta Du

Whitefly traps (Sticky

Gelbtafeln

Neudorff, Emmerthal,

Confidor WG 70

Bayer CropScience, Monheim,

Regenstauf, Germany yellow traps) Insecticide (if

Germany

required)

Germany

Summary and outlook This report describes the efforts undertaken to define and implement an upstream manufacturing process compatible with the GMP production of a human monoclonal antibody for a clinical phase I safety study, covering expression vector construct design, the generation and characterization of suitable transgenic plant lines as well as master and working seed banks, the investigation of production conditions, yield development, and the realization of several production-scale engineering batches to refine and verify the cultivation and harvesting conditions, to derive acceptance criteria for production parameters and to implement and validate measures required for GMP-compliant manufacturing. The upstream process development has established the foundation for the manufacturing process that led to the regulatory approval and a first-in-human phase I clinical trial of a monoclonal antibody produced in transgenic tobacco plants as described by Ma et al. in this issue. The process was repeated and refined in the context of the ERC project Future-Pharma (269110) aiming for a clinical dose-escalation phase I study of intravenously injected plant-derived 2G12. The presented study is intended to serve as a practical example for implementing the general regulatory requirements (Fischer et al., 2012; Medrano


1098 Markus Sack et al. (a)

(c)

(b)

(d)

(e)

(f)

Figure 2 2G12 yield variations. (a) Analysis of intraleaf 2G12 accumulation. A stack of 20 leaves of identical age and size (35 cm) was cut into 14 segments, each 2 cm wide. Pooled samples for each segment were analysed for 2G12 accumulation. (b) Intraday variation of 2G12 levels was investigated at 2-h intervals over 26 h by sampling leaves at positions +3 and +4 from 12 plants. Pooled samples for each leaf and time point were analysed for 2G12 accumulation. (c, d and e) In planta distribution of 2G12. Antibody accumulation and biomass were measured 47 and 57 days after sowing for 12 leaf positions from bottom (–3) to top (+9) by pooling leaves from 17 plants (c) Biomass in grams per leaf, (d) Accumulation in lg/mL, (e) Yield in lg (per leaf). (f) Contribution of different plant parts (stems, different leaves, flowers) to total biomass yield as a percentage of total plant biomass.

represented. The analysis was performed for two different leaf positions (+3 and +4, see Figure S4 for the numbering scheme). The mean 2G12 levels observed at the two leaf positions were 17.1 1.2 and 18.8 1.2 lg/mL (Figure 2b). The coefficients of variation were only 7% and 6.5%, lower than observed for intraleaf variation. The sample taken at 20:00 showed a peculiar drop in the 2G12 levels, but in general the 2G12 levels did not show a significant circadian variation.

Pot size selection In initial experiments, three different pot sizes (1, 2.5 and 5 L) were compared to determine their effect on biomass yield. When using a slow-release depot fertilizer (ED73, Balster Einheitserdew€ndenberg, Germany), antibody accumulation levels were erk, Fro comparable for the different pot sizes, but there were significant differences in biomass yield (data not shown) with the highest biomass achieved in the largest pot size (100% more biomass than plants grown in the smallest pots). Based on these results, the fertilization strategy was changed to automatic fertigation, which boosted the biomass produced in the smallest pots to the

same levels achieved in the largest pots with the slow-release fertilizer (data not shown). The 2.5-L pots were finally selected based on the optimal combination of standing stability and the need to reduce the volume of soil required, that is the costs for disposal of GMO containing material after harvest.

Product accumulation and distribution in planta Transgenic plants are whole-organism production systems that encompass many different differentiated tissues. It is important to understand the accumulation of 2G12 in different parts of the plant for several reasons. The accumulation kinetics and in planta distribution are further potential sources of process variability. It is also necessary to define which parts of the transgenic plant are harvested and processed. This is not only important for subsequent downstream processing, for example the type of biomass that needs to be processed, but also has direct consequences for the upstream production, for example by defining the optimal cultivation conditions. The production of transgenic plant biomass is generally not limited by costs and typically exceeds the downstream processing capacity. Therefore, the cultivation and


GMP/GACP upstream production of a human IgG in tobacco 1099 harvesting strategies may be tailored to economic and downstream processing requirements. We monitored the accumulation of 2G12 over the cultivation period by sampling the plants when they reached ~10 cm in height and at 4-day intervals thereafter on average (Figure S5). The highest 2G12 levels (11.9 lg/mL) were recorded on day 45, the earliest sampling time point, and progressively declined to 6.1 lg/mL. At the same time, the biomass increased rapidly to 125 g per plant until the flowers emerged and slowly increased further to almost 150 g per plant. Reflecting these two trends, the antibody yield peaked at 59 days after sowing, reaching 2.8 mg per plant, and declined slightly afterwards. Overall, this trend analysis showed that although the antibody levels were highest in young plants, the yield per plant peaked at the time of flowering. Apparently, flowering is an important event that was further investigated in the engineering batches, and it may also be particularly relevant for determining the optimal harvest time point. The in planta distribution of 2G12 was investigated to facilitate the selection of a suitable leaf set for harvesting. Antibody levels were determined at two time points (47 and 57 days after seeding) in different plant parts, including the stems, leaves and flowers of 17 individual plants. In our leaf numbering system, the round bottom leaves were named 3, 2 and 1, whereas the next leaves were numbered +1 to +9 from bottom to top (see Figure S4). Figures 2c–e show the accumulation, biomass and yield for all leaf positions. Figure 2f shows the relative contribution of different plant parts (stems, leaves, flowers and developing side shoots (suckers)) to the total antibody yield. The complete data set is provided in Tables S2 and S3. The leaves +1 to +8 were subsequently defined as the harvesting leaf set. The round bottom leaves were excluded because they may contact the soil or pot and often are damaged. The flowers were excluded because these tissues differ significantly from the leaves, potentially introducing heterogeneity into the 2G12 product. The uppermost leaves were not included for practical reasons, because they are almost always removed when topping the plants, that is when the inflorescence is removed mechanically.

total harvested biomass, (ii) the variation in the leaf-disc samples was higher than in the total harvested biomass, and (iii) there was no correlation between the sampling methods (Figure S6). This directly implies that single leaf-disc samples cannot be used reliably to estimate the antibody yield per plant. Although more advanced leaf-disc sampling regimes could be used (Piotrzkowski et al., 2012), we decided that analysing the entire harvested biomass is the most reliable method to determine the yield per plant in transgenic lines, during development genetics and when characterizing the production line.

Leaf-disc vs whole-plant sampling

The analysis of the engineering batches also provided insight into process variability, reproducibility and seasonal effects. The increases observed in the first three batches also reflected incremental improvements realized by fine-tuning the procedures and staff training. Figure 5 shows the average leaf biomass per plant for the different batches (Figure 5a) and the corresponding 2G12 antibody yield (Figure 5b). The sampling sizes represented ~7% of the total number of plants for the first four engineering batches (2000, 2000, 1200 and 1500 plants). The antibody concentration fell below 10 lg/mL and the yield per plant fell to below 2 mg in the fifth batch due to seasonal effects. This batch was grown in winter between January and March and was characterized by a long delay between sowing and potting and by a prolonged growth phase. The harvested leaf biomass was not affected because the plants were allowed longer to grow. Importantly, the variability within this batch was significantly higher, and the coefficient of variation for the 2G12 yield per plant was 56% compared with 18–28% for the other five batches. Clearly, such a batch should be rejected and acceptable variability for these process variables must be defined. The acceptance criteria for these critical process parameters (CPPs) were defined as (i) the concentration of 2G12 in the leaf extract must be within the range 10–20 mg/mL, (ii) the harvested leaf biomass must be 100–200 g per plant, and (iii) the coefficient of

We have repeatedly noticed large variations between samples when analysing leaf discs. In addition to the gradients across and between leaves and the differences in accumulation during the cultivation, leaf-disc sampling also has a greater likelihood of introducing technical errors during homogenization and extraction. However, the product yield per plant is an extremely important parameter for upstream processing and the evaluation and selection of transgenic lines. We therefore performed a comparative analysis of 2G12 levels determined in leaf-disc samples and in the entire harvested biomass to determine whether the values determined by these two approaches correlated. Samples were generated at the harvesting stage for 20 plants each from the T7 postproduction generation of the 2G12 production line and the T8 generation of candidate line GFD-2. The leaf discs were taken from leaf #8 and homogenized using an electropistil, and the total harvested leaf biomass was homogenized using a blender. Proteins were extracted in 2 mL of buffer per gram biomass, and the 2G12 concentration in the supernatants was measured by SPR after binding to immobilized protein-A. Plotting the leaf-disc levels against the levels determined for the total harvested leaf biomass showed that (i) the 2G12 levels were approximately threefold higher in the leaf-disc samples than in the

Engineering batches Following the previously described initial ‘small-scale’ experiments focusing on variability, productivity and sampling issues, a series of engineering batches was manufactured to establish and finetune the upstream and downstream process workflow and to investigate the process under full-scale production conditions (Figure 3).

Intrabatch variability In the first full-scale engineering batch, the uniformity of transgenic plants grown at different locations in the glasshouse facility was investigated. From ~2000 plants, we sampled a total of 135 plants (7%), corresponding to five plants from each of the 27 tables located in 13 separate growth chambers (Figure S7). For each plant, leaves +1 to +8 were harvested, completely homogenized and sampled for SPR analysis. When looking at the concentration of 2G12, harvested leaf biomass and the 2G12 yield per plant sampled at five different table positions (Figure 4b, d), no significant differences were observed. Nevertheless, there might be a subtle trend for marginally lower values at the top and bottom table positions (Figure 4c). Small differences were noticed for plants sampled from different tables and chambers (Figure 4a). The observed differences mainly fit within the generally observed variance. The observed differences were minor and were considered acceptable.

Interbatch variability



Figure 3 Overview of the engineering batches. Time line and details for six engineering batches produced over 13 months. Numbers in coloured boxes represent the number of days growing in trays (seed to plantlet) and in pods (plantlet to mature plant). Numbers next to the boxes represent the number of sampled plants and the total number of plants grown for this batch.

variation of the 2G12 product yield per plant must be below 35%.

materials of herbal origin (GACP) and implemented and supervised under the Quality System of the Fraunhofer IME good manufacturing practice (GMP) production unit. This Quality System covers all aspects of GMP-compliant production for clinical-grade bulk active pharmaceutical ingredients (APIs) in terms of personnel, premises, equipment, material flow, traceability, monitoring and controls, documentation and auditing, and its routines and principles have been applied or adapted to GACP. The same personnel are involved in GACP and GMP production. They have at least basic GMP and hygiene training and understand quality as it applies to the manufacture of biopharmaceuticals. The heads of manufacturing, quality control and quality assurance (and the Qualified Person) are involved in the GACP and GMP processes. Unit operations such as gowning, access, preparation and cleaning of the glasshouses, seeding, cultivation, quality control of the plants, harvesting and washing of leaves are described in standard operational protocols (SOPs), which are treated in the same way as GMP SOPs in terms of form, release, review and training. The glasshouse chambers for seeding, potting and cultivation until harvest are controlled, but

Accumulation, biomass and yield development under optimized cultivation conditions Antibody accumulation, biomass and yield were monitored 30– 48 days after sowing under optimized cultivation conditions was using five plants per time point (Figure 6). To investigate the influence of flowering, a second set of plants was modified by removing all flowers (topping) as they appeared during the late phase of the investigated growth period. In the 18-day period, a 10-fold increase in biomass was observed with no significant difference between flowering and topped plants. Because the antibody accumulation remained constant over time, the biomass increase resulted in a proportional increase in antibody yield.

The resulting upstream process The work described herein was carried out according to the principles to good agricultural and collection practice for starting

(a)

(c)

(b)

(d)

Figure 4 Intrabatch variations and the influence of table position. (a) Intrabatch (engineering batch 1) 2G12 yield distribution for different tables located in different glasshouse chambers. Different shades of green indicate the chambers (1-7 and A1-6). Numbers on the x-axis indicate the individual tables. Five plants per table were used for the analysis of biomass and 2G12 accumulation and to derive the yields. (b) Schematic drawing of a growth table. Open circles represent pot positions, and green circles represent positions that were sampled. (c) Influence of table position on 2G12 yield: 27 plants were sampled and analysed per table position. (d) The locations of table positions in the glasshouse chamber.


GMP/GACP upstream production of a human IgG in tobacco 1101 (a)

(b)

Glasshouse facility The biosafety level 1 glasshouse facility is subdivided into 14 climate-controlled chambers (six large and eight small) with a total area of 741 m² equipped with 28 ebb and flow tables providing 252 m2 of cultivation area (Figure S6). Due to the glasshouse structure, the light intensity inside the glasshouse is approximately 50% of natural light. All chambers are equipped with a lighting installation (1 : 1 ratio of metal halide and highpressure sodium lamps) to provide supplemental light of 20 klux. Depending on the chamber size, between one and four air conditioners with a spray injection system for elevating the relative humidity and for adiabatic cooling have been installed. The air inlets are cover with screens and filters, and the roof vents are shielded with insect screens. The chamber doors were closed tightly and are lockable. The access to the whole glasshouse complex is electronically controlled.

Cultivation and harvesting

Figure 5 Interbatch variations. (a) Interbatch biomass. (b) 2G12 yield distribution for six engineering batches. For details of the batches (months, plant numbers, sample sizes), refer to Figure 3.

not classified areas. Measures to monitor and control insects and rodents have been defined and applied. The microbial and particle status of the chambers is monitored for information only. All unit operations downstream from harvesting the leaves are described and documented in the batch production instruction and protocols. The equipment for washing, extraction and filtration has been formally qualified (DQ, IQ, OQ, Change control). The entire process including the ‘green’ upstream part has been described to the competent authorities enacting the German Medicines Law in great detail, and regulatory approval has been granted following several onsite inspections.

Glasshouse sanitation A strict sanitation procedure was applied to avoid plant pathogens. First, residual plant debris and substrate were removed from the tables and the floor, then the entire chamber was thoroughly cleaned with a high-pressure cleaner. After drying, the tables and the irrigation system were cleaned and disinfected with neutralized peracetic acid using the Wofasteril combination method (1.5% (v/v) alcapur E + 0.5% (v/v) Wofasteril E400). The chamber itself was than disinfected by fogging (15% (v/v) alcapur E + 5% (v/v) Wofasteril E400). Finally, the tables and the irrigation system were thoroughly flushed with tap water to remove residual acetic acid. After cleaning, all personnel entering the chambers were required to wear protective clothing (overall, overshoes, gloves, hair net and beard net as appropriate). All the equipment that comes into contact with the plants (trays, pots and sticks) was steam-sterilized before use. Only municipal water was used for irrigation.

A standard operational procedure was developed to ensure the reproducible and controlled production of approximately 200 kg leaf biomass of the pGFD production line every 5 weeks. A small chamber was designated for germination and seedling production. Clean and sterilized polypropylene trays with perforated bottoms were filled with the same amount of substrate (~15 L €ndenberg, GerEinheitserde VM; Balster Einheitserdewerk, Fro many) and slightly compressed to achieve an even surface. After watering, ~100 seeds per tray were evenly sown by hand. A 0.3-g aliquot of the pGFD working seed bank containing ~3500 seeds was used to produce a batch of 1350 plants (~2.5 times more than finally planted). The seeds were not covered (tobacco is a light germinator) and were watered with a soft spray to achieve good contact to the substrate. In order to ensure a faster and more homogenous germination, the heating was set to 27 °C and the opening temperature for the vents was set to 29 °C. The humidity was set to 70% and the lighting to a 16-h/8-h day/night cycle. Sticky yellow traps (Gelbtafeln) were positioned as close as possible above the trays to monitor insects (one trap per 4 m²). Each chamber was equipped with two boxes containing a defined number (20) of wheat grains and inspected weekly to detect rodents. In the germination phase, the trays were controlled daily and kept evenly moist. After 1 week, the heating temperature was set to 27 °C/25 °C day/night to promote the hardening of the plants. After 14–17 days, the seedlings were thinned out to ~50 per tray and the first checkpoint was implemented: if the seedlings are smaller than 2 cm diameter at 14 days, then the batch must be discarded. The heating was reduced further to 25 °C/22 °C day/night and the vent opening temperature to 27 °C. After 21–25 days, the second checkpoint was implemented: only red fluorescent plantlets with a diameter of ~10 cm were approved for planting. Nonfluorescent plantlets were observed at a very low frequency of ~1 : 10 000. The plantlets were transplanted into 2.5-L pots filled with Einheitserde T 1.5 €ndenberg, Germany). The pots (Balster Einheitserdewerk, Fro were filled on a table in a cleaned chamber, transported to the final growth chambers and placed on the ebb and flow tables. The temperature was set to 25 °C/22 °C day/night, the vent opening to 27 °C and the humidity to 70%. The day/night cycle was set to 16/8 h. The substrate was thoroughly watered before transplanting. The trays with the plantlets were brought to the cultivation chambers and the plantlets were transferred directly



Figure 6 Development of biomass, 2G12 levels and yield. The development of leaf biomass, antibody accumulation and yield per plant were monitored over a period of 18 days starting from 30 days postsowing (dps) for topped and flowering plants (five plants for each type and time point). Triangles: 2G12 accumulation (lg/ mL), solid (topped plants), open (flowering plants). Squares: biomass (g) solid (topped plants), open (flowering plants). Circles: yield (lg/plant), solid (topped plants), open (flowering plants).

to the ebb and flow tables. In order to produce a homogenous batch, care was taken to use plantlets of the same size, the remainder being discarded. Because freshly transplanted plants are sensitive to full sunlight, the chambers were shaded during transplanting and for one additional day if required. The plants were watered with tap water by hand with a fine spray for the following few days. When the first roots reached the bottom of the pot, irradiation-dependent automated fertigation (irrigation with a fertilizer solution) was started with 0.1% Ferty 2 MEGA (16 + 6+26 + (3.4)), a highquality fertilizer consisting of fully water-soluble minerals including trace elements. The irrigation water in the reservoirs was controlled and adjusted twice a week. The tobacco plants were cultivated without any additional shading and received supplemental light of 20 klux when the external irradiance dropped below 35 klux (Table 2). When the leaves expanded over the rim of the pot, the plants were spaced to the final distance (eight plants/m²). At a height of 30–40 cm, all plants were supported with a bamboo stick to prevent tipping over. Depending on the season, the first plants started to produce flowers ~40 days after sowing. To prevent flowering, the inflorescences were removed by breaking them manually (topping) when the first coloured petal was visible. From this point on, the plants were checked daily and suckers were removed until harvesting. The onset of flowering was also an indicator for batch homogeneity (more than 90% should have set flowers in less than 1 week) and was used as timer for harvesting. The batch was harvested no later than 2 weeks after the first plants started to flower. Plants with unusually delayed flowering were excluded from the harvest. Approximately 1 week ahead of the planned harvesting date, 7% of the plants were analysed individually for biomass and 2G12 accumulation for quality control. Before the batch was finally approved for processing, the plants were checked for macroscopic DsRed fluorescence and nonfluorescent plants were discarded. Apart from the small round green spots with a diameter of a few millimetres only (Figure S1b), we so far have observed only a single case where a larger leaf area was nonfluorescent (Figure S1c) for >30 000 plants over a period of 7 years. Preliminary analysis indicated that antibody accumulation was also reduced in the green patches (data not shown). The leaves were hand-picked and inspected, and only healthy leaves were used for further processing. An additional set of measures (Tables S4 and S5) was defined to ensure the identity of the plant material and to avoid and monitor contamination with pathogens.

Table 2 Standardized cultivation conditions Parameter

Value

Temperature

25 °C/22 °C day/

Relative humidity

70%

Light

16 h/8 h day/night

Sowing compost

Einheitserde type

Supplier (if applicable)

night

(min 20 klux) (trays) Potting compost (2.5-L pots)

VM Einheitserde type T 1.5

Balster Einheitserdewerk, €ndenberg, Germany Fro Balster Einheitserdewerk, €ndenberg, Germany Fro

Fertigation

0.1% Ferty 2 MEGA

€ngemittel, Planta Du

Whitefly traps (Sticky

Gelbtafeln

Neudorff, Emmerthal,

Confidor WG 70

Bayer CropScience, Monheim,

Regenstauf, Germany yellow traps) Insecticide (if

Germany

required)

Germany

Summary and outlook This report describes the efforts undertaken to define and implement an upstream manufacturing process compatible with the GMP production of a human monoclonal antibody for a clinical phase I safety study, covering expression vector construct design, the generation and characterization of suitable transgenic plant lines as well as master and working seed banks, the investigation of production conditions, yield development, and the realization of several production-scale engineering batches to refine and verify the cultivation and harvesting conditions, to derive acceptance criteria for production parameters and to implement and validate measures required for GMP-compliant manufacturing. The upstream process development has established the foundation for the manufacturing process that led to the regulatory approval and a first-in-human phase I clinical trial of a monoclonal antibody produced in transgenic tobacco plants as described by Ma et al. in this issue. The process was repeated and refined in the context of the ERC project Future-Pharma (269110) aiming for a clinical dose-escalation phase I study of intravenously injected plant-derived 2G12. The presented study is intended to serve as a practical example for implementing the general regulatory requirements (Fischer et al., 2012; Medrano


GMP/GACP upstream production of a human IgG in tobacco 1103 et al., 2012; Rybicki et al., 2012; Sparrow et al., 2013; Yusibov et al., 2011) and to allow comparisons with previously published reports for both transgenic (Boothe et al., 2010; Colgan et al., 2010; He et al., 2011; Valdes et al., 2003) and transient processes (D’Aoust et al., 2008; Klimyuk et al., 2014; Lai and Chen, 2012; Pogue et al., 2010). However, we want to point out that in each system, each product has unique features that have to be taken into account when translating the presented approach. Future reports are expected to add to the knowledge base for plant-made pharmaceuticals and thereby further progress and mature the field.

Materials and methods Host plant Certified tobacco seeds (Nicotiana tabacum L. Petit Havana SR1; PI 552516, USDA-ARS GRIN database) were obtained from Lehle (Round Rock, TX). This host plant shows monopodial growth, has natural plastid-based resistance to streptomycin and can be infected by Tobacco mosaic virus. Petit Havana is a small variety with day-length-independent flower induction. The generation time (seed to seed) is 3–4 months. The cultivar is a homozygous inbred line that is propagated by self-pollination.

Plant expression construct design The expression cassettes for the two 2G12 antibody chain genes and for the visual marker DsRed consist of the duplicated CaMV 35S promoter, the TEV 5ʹ untranslated region and the CaMV 35S 3ʹ UTR and transcriptional terminator. The expression cassette for the herbicide resistance gene (pat) comprises the nos promoter and the nos transcriptional terminator and was placed at the left border of the T-DNA. The expression cassettes are separated by scaffold attachment regions (SARs) to prevent interference and to isolate them from the surrounding host cell genomic DNA. The SAR downstream of the rb7 gene has also been described to be a transcriptional enhancer (Allen et al., 1996). The full-size 2G12 antibody is encoded by genes for the 2G12 gamma-1 heavy chain and the 2G12 kappa light chain, each fused in-frame to their signal peptide sequences (SPg and SPk). The coding regions were not further optimized for tobacco codon usage and are identical to the 2G12 produced in Chinese hamster ovary (CHO) cells, previously used in a human phase I trial (Stiegler et al., 2002). The T-DNA also contains a gene encoding the red fluorescent protein (DsRed) from the reef coral Discosoma sp. under the control of the same genetic elements regulating the antibody genes. DsRed was targeted to the plastids by a transit peptide from the barley (Hordeum vulgare) granule-bound starch synthase gene. A schematic representation of the pTRAp-2G12-Ds constructs is shown in Figure 1a and the identities of the genetic elements are listed in Table 1.

Plant transformation Recombinant A. tumefaciens were generated as described previously (Sack et al., 2007). Briefly, strain GV3101 carrying the binary vector pTRA-2G12-Ds and the helper plasmid pMP90RK were selected on YEB-agar plates containing 50 lg/mL carbenicillin, 50 lg/mL rifampicin and 25 lg/mL kanamycin. The presence of the binary vector was confirmed by PCR and transient expression of the functional 2G12 antibody and DsRed. Transgenic tobacco plants were generated using a modified leaf-disc method. Two fully developed leaves from a six-week-old glass-

house-grown wild-type plant were infiltrated with the recombinant bacteria resuspended in MS medium at an OD600 of 1.0 (Kapila et al., 1997; Vaquero et al., 1999). The accumulation of DsRed was followed by macroscopic fluorescence, and after 3 weeks, discs from bright section were cut from the leaf, surface-sterilized and placed on classical regeneration medium. The plates contained 100 lg/mL cefotaxime to kill the remaining bacteria and 100 lg/mL phosphinothricin for the selection of transformed cells. Shots with intense DsRed fluorescence were separated from the callus and transferred onto rooting medium also containing cefotaxime and phosphinothricin. Rooted plantlets were planted in soil and cultivated in the glasshouse. The accumulation and in planta distribution of DsRed was checked regularly. Plants were analysed for antibody accumulation, morphological phenotype, growth behaviour and vigour. To prevent cross-pollination, selected plants were covered with transparent, gas-permeable plastic foil once floral buds developed and before the first flowers opened. During the inbreeding phase, the seeds from up to eight mature seed pods were collected and stored dry in paper bags at 20 °C in the dark. In the first generation, seeds were germinated on phosphinothricin-supplemented agar plates until the propagated lines were nonsegregating for herbicide resistance. In each generation, selected lines were analysed for the segregation of DsRed fluorescence and validated for good seed production and high germination rates.

Plant sampling for the analysis of production parameters Intraday antibody levels were determined by sampling leaf discs from 12 leaves (positions +3 and +4) at 2-h intervals over 26 h, followed by SPR analysis. In a similar manner, intraleaf variations were addressed by the analysis of 14 samples (the mid-rib was excluded) distributed over the length of 10 leaves with a length of 35 cm. The in planta distribution of 2G12 was analysed by sampling material from stems, flowers and leaves at different positions from top to bottom. To investigate variations between different growth chambers, tables and table positions, leaves from entire plants (five per table for chamber comparison, and 27 per table position) were analysed for leaf biomass and 2G12 yield. Once the harvesting of the leaf biomass was defined, the entire biomass destined for processing was collected from individual plants. The biomass was homogenized in two volumes of extraction buffer using a blender. A 1.5-mL sample was taken and centrifuged, and the supernatant was used for further analysis.

Visual DsRed epifluorescence analysis The excitation spectrum of DsRed peaks at 558 nm (green) and the emission maximum peaks at 583 nm (amber/orange). For macroscopic DsRed analysis of green tissues, a long pass suppression filter is sufficient because interfering excitation of chlorophyll fluorescence is low under green light. The green excitation wavelength is not detrimental for the plants or for humans. In addition, green excitation light can easy penetrate plant tissues and is absorbed less by plant compounds than blue or UV light, the latter often used for GFP excitation. DsRed fluorescence was visualized using a portable device that allows the examination of whole plants in the glasshouse. For DsRed excitation, we used a cold light source with a green excitation filter and 1 m fibre optics (KL2500LCD, Schott, Germany), and red lighting filter foil (Lee filters, UK) was used to cover glasses and the digital camera objective for inspection and documentation.



Surface plasmon resonance spectroscopy SPR spectroscopy was used to determine the quantity, assembly and quality of 2G12 in plant extracts and purified samples using protein-A, protein-L and gp120 surfaces and binding assays as previously described (Holland et al., 2010; Rademacher et al., 2008). Assays were carried out at 25 °C on a BIACORE T100 instrument (GE Healthcare Europe GmbH, Freiburg, Germany) using HBS-EP as running buffer. Plant extracts, CHO-derived 2G12 (Polymun Scientific, Vienna) and the tobacco-derived 2G12 reference were diluted so that the binding signals were in the linear range. The binding curves were evaluated as previously described (Holland et al., 2010; Rademacher et al., 2008).

PCR, Southern and northern blot analysis Genomic DNA from transgenic pGFD tobacco plants kept in the dark for 7 days was extracted using the DNeasy Plant Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. PCR was used to confirm the presence of the transgenes, with primers hybridizing to the CaMV 35SS promoter and terminator to amplify the 2G12 heavy chain, light chain and DsRed coding regions simultaneously. For Southern blot analysis, the genomic DNA was digested with NcoI and XbaI, which excise the complete coding regions of both antibody chains. The expected size of the transgene coding sequence for the heavy and light chains is 1421 and 713 bp, respectively. We separated 7 lg of the digested DNA by agarose gel electrophoresis and transferred the fractionated DNA onto a nitrocellulose membrane. Probes were generated, and the targets detected using the nonradioactive Amersham AlkPhos Direct Labeling and Detection System with CDP-star (GE Healthcare) and chemiluminescence was measured with the ImageQuant LAS-3000 (GE Healthcare). The probes were homologous to the 3ʹ ends of the sequences for the heavy and light chains and were 393 and 313 bp in length, respectively. Defined amounts of digested plasmid (pTRAp-2G12Ds) containing the transgenes were used as a standard to estimate the number of transgene copies. RNA was extracted using the RNeasy Plant Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions, separated by agarose gel electrophoresis and blotted onto a nitrocellulose membrane. The expected sizes (from the transcriptional start site to the polyadenylation site) for the heavy and light chain transcripts are 1751 and 1037 b, respectively. The transcripts were detected using the same probes and protocol described for the Southern blots.

Acknowledgements The Pharma-Planta project was funded by the European Union (LSHB-CT-2003-503565). We also would like to thank Fraunhofer Society for strategic funding and all scientists involved in the Pharma-Planta project for scientific input. Finally, we thank members of the Biologics Working Group at EMA and regulatory experts at the MHRA for valuable discussions. We thank Richard Twyman for critically reading the manuscript.

References Allen, G.C., Hall, G. Jr, Michalowski, S., Newman, W., Spiker, S., Weissinger, A.K. and Thompson, W.F. (1996) High-level transgene expression in plant cells: effects of a strong scaffold attachment region from tobacco. Plant Cell, 8, 899–913.

Aviezer, D., Brill-Almon, E., Shaaltiel, Y., Hashmueli, S., Bartfeld, D., Mizrachi, S., Liberman, Y., Freeman, A., Zimran, A. and Galun, E. (2009) A plantderived recombinant human glucocerebrosidase enzyme–a preclinical and phase I investigation. PLoS ONE, 4, e4792. Boothe, J., Nykiforuk, C., Shen, Y., Zaplachinski, S., Szarka, S., Kuhlman, P., Murray, E., Morck, D. and Moloney, M.M. (2010) Seed-based expression systems for plant molecular farming. Plant Biotechnol. J. 8, 588–606. Buyel, J.F. and Fischer, R. (2013) Processing heterogeneous biomass: overcoming the hurdles in model building. Bioengineered, 4, 21–24. Buyel, J.F. and Fischer, R. (2014) Scale-down models to optimize a filter train for the downstream purification of recombinant pharmaceutical proteins produced in tobacco leaves. Biotechnol. J. 9, 415–425. Colgan, R., Atkinson, C.J., Paul, M., Hassan, S., Drake, P.M., Sexton, A.L., Santa-Cruz, S., James, D., Hamp, K., Gutteridge, C. and Ma, J.K. (2010) Optimisation of contained Nicotiana tabacum cultivation for the production of recombinant protein pharmaceuticals. Transgenic Res. 19, 241–256. Daniell, H., Singh, N.D., Mason, H. and Streatfield, S.J. (2009) Plant-made vaccine antigens and biopharmaceuticals. Trends Plant Sci. 14, 669–679. D’Aoust, M.A., Lavoie, P.O., Couture, M.M., Trepanier, S., Guay, J.M., Dargis, M., Mongrand, S., Landry, N., Ward, B.J. and Vezina, L.P. (2008) Influenza virus-like particles produced by transient expression in Nicotiana benthamiana induce a protective immune response against a lethal viral challenge in mice. Plant Biotechnol. J. 6, 930–940. D’Aoust, M.A., Couture, M.M., Charland, N., Trepanier, S., Landry, N., Ors, F. and Vezina, L.P. (2010) The production of hemagglutinin-based virus-like particles in plants: a rapid, efficient and safe response to pandemic influenza. Plant Biotechnol. J. 8, 607–619. Deblock, M., Herreraestrella, L., Vanmontagu, M., Schell, J. and Zambryski, P. (1984) Expression of foreign genes in regenerated plants and in their progeny. EMBO J. 3, 1681–1689. Farrance, C.E., Rhee, A., Jones, R.M., Musiychuk, K., Shamloul, M., Sharma, S., Mett, V., Chichester, J.A., Streatfield, S.J., Roeffen, W., van de VegteBolmer, M., Sauerwein, R.W., Tsuboi, T., Muratova, O.V., Wu, Y. and Yusibov, V. (2011) A plant-produced Pfs230 vaccine candidate blocks transmission of Plasmodium falciparum. Clin. Vaccine Immunol. 18, 1351– 1357. Fischer, R., Schillberg, S. (2006) Molecular Farming: Plant-made Pharmaceuticals and Technical Proteins. (Fischer, R. and Schillberg, S., eds), ISBN: 978-3-52760581-1, 338 pages, March 2006, Wiley-Blackwell. http://eu.wiley.com/ WileyCDA/WileyTitle/productCd-3527605819.html Fischer, R., Stoger, E., Schillberg, S., Christou, P. and Twyman, R.M. (2004) Plantbased production of biopharmaceuticals. Curr. Opin. Plant Biol. 7, 152–158. Fischer, R., Schillberg, S., Hellwig, S., Twyman, R.M. and Drossard, J. (2012) GMP issues for recombinant plant-derived pharmaceutical proteins. Biotechnol. Adv. 30, 434–439. Fischer, R., Schillberg, S., Buyel, J.F. and Twyman, R.M. (2013) Commercial aspects of pharmaceutical protein production in plants. Curr. Pharm. Des. 19, 5471–5477. Harmer, S.L. (2009) The circadian system in higher plants. Annu. Rev. Plant Biol. 60, 357–377. He, Y., Ning, T., Xie, T., Qiu, Q., Zhang, L., Sun, Y., Jiang, D., Fu, K., Yin, F., Zhang, W., Shen, L., Wang, H., Li, J., Lin, Q., Sun, Y., Li, H., Zhu, Y. and Yang, D. (2011) Large-scale production of functional human serum albumin from transgenic rice seeds. Proc. Natl Acad. Sci. USA, 108, 19078–19083. Holland, T., Sack, M., Rademacher, T., Schmale, K., Altmann, F., Stadlmann, J., Fischer, R. and Hellwig, S. (2010) Optimal nitrogen supply as a key to increased and sustained production of a monoclonal full-size antibody in BY2 suspension culture. Biotechnol. Bioeng. 107, 278–289. Kapila, J., DeRycke, R., VanMontagu, M. and Angenon, G. (1997) An Agrobacterium-mediated transient gene expression system for intact leaves. Plant Sci. 122, 101–108. Klimyuk, V., Pogue, G., Herz, S., Butler, J. and Haydon, H. (2014) Production of recombinant antigens and antibodies in Nicotiana benthamiana using ‘magnifection’ technology: GMP-compliant facilities for small- and largescale manufacturing. Curr Top Microbiol Immunol., 375, 127–154. Lai, H. and Chen, Q. (2012) Bioprocessing of plant-derived virus-like particles of Norwalk virus capsid protein under current Good Manufacture Practice regulations. Plant Cell Rep. 31, 573–584.


GMP/GACP upstream production of a human IgG in tobacco 1105 Landry, N., Ward, B.J., Trepanier, S., Montomoli, E., Dargis, M., Lapini, G. and Vezina, L.P. (2010) Preclinical and clinical development of plant-made viruslike particle vaccine against avian H5N1 influenza. PLoS ONE, 5, e15559. Ma, J.K.C., Drake, P.M.W. and Christou, P. (2003) The production of recombinant pharmaceutical proteins in plants. Nat. Rev. Genet. 4, 794–805. Marzi, A., Yoshida, R., Miyamoto, H., Ishijima, M., Suzuki, Y., Higuchi, M., Matsuyama, Y., Igarashi, M., Nakayama, E., Kuroda, M., Saijo, M., Feldmann, F., Brining, D., Feldmann, H. and Takada, A. (2012) Protective efficacy of neutralizing monoclonal antibodies in a nonhuman primate model of Ebola hemorrhagic fever. PLoS ONE, 7, e36192. McClung, C.R. (2006) Plant circadian rhythms. Plant Cell, 18, 792–803. Medrano, G., Dolan, M.C., Condori, J., Radin, D.N. and Cramer, C.L. (2012) Quality assessment of recombinant proteins produced in plants. Methods Mol. Biol. 824, 535–564. Paul, M. and Ma, J.K. (2011) Plant-made pharmaceuticals: leading products and production platforms. Biotechnol. Appl. Biochem. 58, 58–67. Piotrzkowski, N., Schillberg, S. and Rasche, S. (2012) Tackling heterogeneity: a leaf disc-based assay for the high-throughput screening of transient gene expression in tobacco. PLoS ONE, 7, e45803. Pogue, G.P., Vojdani, F., Palmer, K.E., Hiatt, E., Hume, S., Phelps, J., Long, L., Bohorova, N., Kim, D., Pauly, M., Velasco, J., Whaley, K., Zeitlin, L., Garger, S.J., White, E., Bai, Y., Haydon, H. and Bratcher, B. (2010) Production of pharmaceutical-grade recombinant aprotinin and a monoclonal antibody product using plant-based transient expression systems. Plant Biotechnol. J. 8, 638–654. Rademacher, T., Sack, M., Arcalis, E., Stadlmann, J., Balzer, S., Altmann, F., Quendler, H., Stiegler, G., Kunert, R., Fischer, R. and Stoger, E. (2008) Recombinant antibody 2G12 produced in maize endosperm efficiently neutralizes HIV-1 and contains predominantly single-GlcNAc N-glycans. Plant Biotechnol. J. 6, 189–201. Rybicki, E.P. (2010) Plant-made vaccines for humans and animals. Plant Biotechnol. J. 8, 620–637. Rybicki, E.P., Chikwamba, R., Koch, M., Rhodes, J.I. and Groenewald, J.H. (2012) Plant-made therapeutics: an emerging platform in South Africa. Biotechnol. Adv. 30, 449–459. Sack, M., Paetz, A., Kunert, R., Bomble, M., Hesse, F., Stiegler, G., Fischer, R., Katinger, H., Stoeger, E. and Rademacher, T. (2007) Functional analysis of the broadly neutralizing human anti-HIV-1 antibody 2F5 produced in transgenic BY-2 suspension cultures. FASEB J, 21, 1655–1664. Sparrow, P.A., Irwin, J.A., Dale, P.J., Twyman, R.M. and Ma, J.K. (2007) Pharma-Planta: road testing the developing regulatory guidelines for plantmade pharmaceuticals. Transgenic Res. 16, 147–161. Sparrow, P., Broer, I., Hood, E.E., Eversole, K., Hartung, F. and Schiemann, J. (2013) risk assessment and regulation of molecular farming – a comparison between Europe and US. Curr. Pharm. Des. 19, 5513–5530. Stiegler, G., Armbruster, C., Vcelar, B., Stoiber, H., Kunert, R., Michael, N.L., Jagodzinski, L.L., Ammann, C., Jager, W., Jacobson, J., Vetter, N. and Katinger, H. (2002) Antiviral activity of the neutralizing antibodies 2F5 and 2G12 in asymptomatic HIV-1-infected humans: a phase I evaluation. AIDS, 16, 2019–2025. Stoger, E., Fischer, R., Moloney, M. and Ma, J.K. (2014) Plant molecular pharming for the treatment of chronic and infectious diseases. Annu. Rev. Plant Biol. 65, 743–768.

Trkola, A., Purtscher, M., Muster, T., Ballaun, C., Buchacher, A., Sullivan, N., Srinivasan, K., Sodroski, J., Moore, J.P. and Katinger, H. (1996) Human monoclonal antibody 2G12 defines a distinctive neutralization epitope on the gp120 glycoprotein of human immunodeficiency virus type 1. J. Virol. 70, 1100–1108. Valdes, R., Gomez, L., Padilla, S., Brito, J., Reyes, B., Alvarez, T., Mendoza, O., Herrera, O., Ferro, W., Pujol, M., Leal, V., Linares, M., Hevia, Y., Garcia, C., Mila, L., Garcia, O., Sanchez, R., Acosta, A., Geada, D., Paez, R., Luis Vega, J. and Borroto, C. (2003) Large-scale purification of an antibody directed against hepatitis B surface antigen from transgenic tobacco plants. Biochem. Biophys. Res. Commun. 308, 94–100. Vaquero, C., Sack, M., Chandler, J., Drossard, J., Schuster, F., Monecke, M., Schillberg, S. and Fischer, R. (1999) Transient expression of a tumor-specific single-chain fragment and a chimeric antibody in tobacco leaves. Proc. Natl Acad. Sci. USA, 96, 11128–11133. Waheed, M.T., Gottschamel, J., Hassan, S.W. and Lossl, A.G. (2012) Plantderived vaccines: an approach for affordable vaccines against cervical cancer. Hum. Vaccin. Immunother. 8, 403–406. Yusibov, V., Streatfield, S.J. and Kushnir, N. (2011) Clinical development of plant-produced recombinant pharmaceuticals: vaccines, antibodies and beyond. Hum. Vaccin. 7, 313–321.

Supporting information Additional Supporting information may be found in the online version of this article: Figure S1 Visual inspection of transgene expression. Figure S2 Molecular characterization of master (MSB) and working (WSB) seed banks. Figure S3 Phenotypic characterization of different generations. Figure S4 Leaf numbering scheme. Figure S5 Investigation of the ‘harvesting window’. Figure S6 Leaf disc vs. total harvested leaf sampling. Figure S7 Production greenhouse overview. Figure S8 Visual plant inspection and detection of transgene during batch production. Table S1 Overview of the pGFD13 generations leading to the seed banks and final production line. Table S2 Yield and Biomass analysis from a production batch at day 47. Table S3 Yield and Biomass analysis from a production batch at day 57. Table S4 Measures to ensure the identity of the plant material. Table S5 Measures to avoid and monitor contamination of the plants with pathogens.
