• Open Access

Production of pharmaceutical-grade recombinant aprotinin and a monoclonal antibody product using plant-based transient expression systems


Correspondence (fax +1 270 689 2571; e-mail gppogue@yahoo.com)


Plants have been proposed as an attractive alternative for pharmaceutical protein production to current mammalian or microbial cell-based systems. Eukaryotic protein processing coupled with reduced production costs and low risk for mammalian pathogen contamination and other impurities have led many to predict that agricultural systems may offer the next wave for pharmaceutical product production. However, for this to become a reality, the quality of products produced at a relevant scale must equal or exceed the predetermined release criteria of identity, purity, potency and safety as required by pharmaceutical regulatory agencies. In this article, the ability of transient plant virus expression systems to produce a wide range of products at high purity and activity is reviewed. The production of different recombinant proteins is described along with comparisons with established standards, including high purity, specific activity and promising preclinical outcomes. Adaptation of transient plant virus systems to large-scale manufacturing formats required development of virus particle and Agrobacterium inoculation methods. One transient plant system case study illustrates the properties of greenhouse and field-produced recombinant aprotinin compared with an US Food and Drug Administration-approved pharmaceutical product and found them to be highly comparable in all properties evaluated. A second transient plant system case study demonstrates a fully functional monoclonal antibody conforming to release specifications. In conclusion, the production capacity of large quantities of recombinant protein offered by transient plant expression systems, coupled with robust downstream purification approaches, offers a promising solution to recombinant protein production that compares favourably to cell-based systems in scale, cost and quality.


Cancer, infectious and chronic diseases continue to exact a high toll on human life. More than 560 000 people die each year from cancer and associated complications in the United States alone, and the emergence of a new and deadly swine flu strain in 2009 provides ample illustration of this reality (Jemal et al., 2008; Sym et al., 2009). While medical science has made great strides, traditional small molecule drugs do not adequately address many disease conditions requiring treatment. The advent of recombinant, biologically derived products (biologics) has revolutionized the practice of medicine through the use of monoclonal antibodies (mAbs), vaccines and other therapeutic proteins. Biologics offer new patient therapies and often show increased efficacy and less off-target, off-mechanism effects in comparison with small molecule therapies or chemotherapies (Szymkowski, 2004; Platis and Labrou, 2008). The desirability of these therapies is best characterized by biologic sales that have experienced annual double digit increases for most years since their introduction in the early 1980s and are predicted to increase from 26% of the pharmaceutical market to 40% by 2013 (Goodman, 2009). Although the biologics market continues to represent the fastest growing segment in the biopharmaceutical industry because of their new and broad clinical applications, the financial challenges facing the industry necessitate alternative cost-effective solutions.

Plants have been gaining market acceptance as an attractive alternative production system for biologics by overcoming several challenges facing the biopharmaceutical industry. In this article, we review two major market challenges and, utilizing two case studies, demonstrate the ability of transient plant expression systems to meet the stringent demands for high quality biologics at competitive scale and competitive cost of current manufacturing systems while overcoming some complications with current systems.

One imminent challenge to existing biologic products is competition from follow-on biologics. Currently, the market is partially insulated from follow-on biologic threat because of the ‘process is product’ stance of the US Food and Drug Administration (FDA) and the willingness of governments to provide extended product protection periods to biologics (Datamonitor, 2008). However, change is on the horizon. The European Agency for the Evaluation of Medicinal Products has issued formal guidance for follow-on biologics and has approved 13 follow-on products—67% of the applications filed since 2004 (Greb, 2009). In the United States, discussions have begun regarding follow-on biologic approval pathways by healthcare reform groups who desire significant cost reduction in pharmaceutical products (Greb, 2009). As pressure mounts to reduce the cost of biologics, biopharmaceutical companies will seek ways to recapture the high costs of research and development, along with the reduced profit margins associated with follow-on biologics.

Production costs represent a second challenge to successful biologic products which are manufactured primarily using microbial or mammalian cell-based expression systems. Cell-based systems are inherently more complex and expensive than the production of most small molecule drugs (Yina et al., 2007). Cell-based manufacturing requires considerable capital and time to construct the requisite facilities, including both upstream, cell-based fermentation and downstream production, purification and formulation, capabilities. The typical costs associated with these facilities are $300–$500 million and require from 4 to 5 years to complete construction, validation, and to gain regulatory approval (Thiel, 2004). Each facility has a basic production capacity that must be continually deployed for necessary amortization of construction costs; however, this capacity cannot be readily expanded without construction of replicate facilities. Such costs and capital commitment inevitably affect the costs of resulting products necessitating manufacturing systems that offer less capital-intensive scaling to accommodate product requirements (Garber, 2001).

Because of their eukaryotic protein processing and established success surrounding agricultural products, plants are viewed as an attractive alternative production system for many biologics (Ma et al., 2003; Floss et al., 2007; Lico et al., 2008; Plasson et al., 2009). Agriculture allows upstream manufacturing capacity to be scaled in a capital-efficient manner, offering both flexibility and cost savings that cannot be easily matched by fermentation technologies. Such efficiencies make plants particularly attractive with the threat of follow-on biologics and rising capital costs. Downstream handing of plant biomass requires unique biomanufacturing solutions (Plesha et al., 2009). In spite of different methods employed, the purified product must show the same quality as produced by traditional cell-based systems (Pogue et al., 2002; Sharma and Sharma, 2009). To this end, many competing technologies have been developed to produce recombinant proteins in plants using stable transformation methods that modify the genetic complement of the production plant species, such that prodigy inherits the foreign gene sequence and expression capacity (Floss et al., 2007; Sharma and Sharma, 2009). Although this method has shown robust expression of several pharmaceutically relevant proteins such as growth hormone, α-1 antitrypsin, various mAbs and recombinant vaccines (Ma et al., 2003), it does have notable drawbacks. The process of genetic transformation is slow, requiring months to years to derive sufficient seed for significant plantings. Horizontal transmission of the recombinant gene is a concern and has led to complex regulatory oversight to prevent pollen transfer or regrowth of transgenic crops through tissue or seed dispersal. Finally, many food or feed crops are often employed as production hosts leaving significant human food supply or livestock contamination concerns (Belson, 2000; Pogue et al., 2002) as described both in the media and in official documentation from agricultural regulatory authorities in Europe and the United States (European Food Safety Authority, 2009). The risks associated with non-food, non-feed crops are significantly less than consumable plants and are strongly favoured by regulatory authorities (Belson, 2000).

In the face of these challenges, transient protein expression strategies bring the significant advantages of plant-based bioreactors at considerably reduced costs to current cell-based manufacturing systems while avoiding the less desirable properties of stable plant transformation. Transient systems have been demonstrated as safe and environmentally friendly in both indoor and outdoor tests since 1991, and 16 products produced by transient systems were shown safe in early-stage human clinical trials as personalized vaccines administered to non-Hodgkin’s lymphoma patients (Pogue et al., 2002; McCormick et al., 2008). Additional advantages compared with traditional cell-based fermentation approaches include: (i) speed and low cost of genetic manipulation; (ii) rapid manufacturing cycles; (iii) no mammalian pathogen contamination; (iv) minimal endotoxin concentrations and (v) economical production (Pogue et al., 2002; Ma et al., 2003; Gleba et al., 2008; Lico et al., 2008; Vézina et al., 2009).

Two approaches dominate transient expression: standard integrative plant expression vectors and virus-based replicating systems (Ma et al., 2003; Floss et al., 2007; Lico et al., 2008; Sharma and Sharma, 2009). Standard integrative plant expression vectors are introduced into intact plants using an Agrobacterium tumefaciens-mediated transfer-DNA delivery system (Agro-infiltration; vacuum infiltration of aerial parts of the plant to introduce Agrobacterium cells containing expression vectors into the plant cells; Joh and VanderGheynst, 2006; Vézina et al., 2009). The flexibility of the Agro-infiltration system allows for the efficient expression of the biopharmaceutical protein of interest and provides the required cofactors to improve pharmaceutical protein yield and processing. The co-expression of silencing suppressor proteins has been shown to be a key factor for optimized yields (Mallory et al., 2002; Hellens et al., 2005; Azhakanandam et al., 2007). Such methods have been used to produce a range of biopharmaceutical proteins (Joh and VanderGheynst, 2006; Benchabane et al., 2009; Sourrouille et al., 2009) and offer strategies to modify the plant enzymatic machinery, producing more stable and ‘human’-like recombinant proteins, including glycan structures (Benchabane et al., 2008; Vézina et al., 2009).

Virus-based replicating systems offer advantages over standard integrative plant expression systems by exploiting the cytoplasmic replication cycle of the virus vector (Pogue et al., 2002; Gleba et al., 2008; Lico et al., 2008). The ability of virus-based systems to sequester host and virally encode enzymes facilitates the amplification of messenger RNA leading to increased pharmaceutical protein accumulation (Gleba et al., 2008; Lico et al., 2008). These systems also offer rapid and efficient expression characteristics, new genes can be tested for expression and test quantities of the recombinant protein can be obtained in as little as 4–8 weeks, while leaving no heritable changes to the production plant (no foreign gene is transmitted in the pollen or by insects and is therefore contained within its boundaries; Pogue et al., 2002). Virus-based replicating systems generally fall into two categories: ‘independent-virus’ or ‘minimal-virus’. Independent-virus vectors are inoculated as virus particles or viral RNA and exploit virus-encoded cell-to-cell and systemic movement activities to infect host plants. Replicating independent-viruses spread systemically from a small number of initially infected cells to infect the majority of the phloem sink tissue of a host. Expression of messenger RNAs encoding recombinant proteins is mediated by either the activity of virus subgenomic promoter or polyprotein translational expression mechanisms. Independent-virus systems have been derived from the genomes of potexviruses (including potato virus X; PVX), tobamoviruses (including tobacco mosaic virus; TMV), comoviruses (including cowpea mosaic virus), potyviruses, tobraviruses, closteroviruses and several others (Pogue et al., 2002; Lico et al., 2008).

In contrast, minimal-virus systems are capable of functions supporting RNA replication. This approach increases the genetic load carried by the minimal-virus systems allowing efficient expression of larger recombinant proteins (Giritch et al., 2006). Because these systems are incapable of movement in inoculated plants, they must be delivered to the majority of plant cell to produce meaningful amounts of recombinant proteins. This is usually accomplished through Agro-infiltration of host plants to launch the infection process (Gleba et al., 2005, 2007, 2008). Minimal-virus systems do not require the delays associated with systemic plant movement and have the ability to replicate to high levels, often yielding greater amounts of recombinant proteins in a shorter period of time than independent-virus systems (Gleba et al., 2007, 2008). Minimal-virus systems primarily exploit the subgenomic promoter activities and genomes of potexviruses and tobamoviruses (Gleba et al., 2008).

This review will focus on two case studies of transient, viral-based plant expression technologies: Kentucky BioProcessing, LLC’s (KBP) GENEWARE®, a independent-virus system and Icon Genetics (Halle, Germany), GmbH’s magnICON®, a minimal-virus system. We will consider the advantages offered by each and explore two specific biologic examples, recombinant aprotinin and a mAb binding the chemokine (C–C motif) receptor-5 (CCR5), each produced at multi-gram scale under current good manufacturing practices. These case studies will illustrate the flexibility and power of transient plant expression systems to provide recombinant protein products of the quality and quantity required for clinical development.

Case study I: GENEWARE® system

GENEWARE® is a hybrid replicon derived from TMV, principally strains U1 and U5. Tobamoviruses have a plus sense single-stranded RNA genome of ∼6400 nucleotides helically encapsidated in rigid rod-shaped particles composed of ∼2100 copies of the 17.5 kDa coat protein (CP). The viral proteins involved in RNA replication are directly transcribed from the genomic RNA, whereas expression of internal genes is through the production of subgenomic RNAs (Dawson and Lehto, 1990). The production of subgenomic RNAs is controlled by sequences in the tobamovirus genome, which function as subgenomic promoters. The CP is translated from a subgenomic RNA and is the most abundant protein, and RNA produced in the infected cell (Turpen, 1999). In a tobamovirus-infected plant, there are several milligrams of CP produced per gram of infected tissue.

GENEWARE® expression system takes advantage of independent-virus functions, including cell-to-cell and systemic movement activities mediated by movement protein (MP) and CP, respectively (Figure 1). GENEWARE® also exploits the strength and duration of the viral subgenomic promoter’s activity to reprogram the translational priorities of the plant host cells so that virus-encoded proteins are synthesized at similar high levels as the TMV CP (Shivprasad et al., 1999). A foreign gene encoding the protein for overexpression is added in place of the virus CP, so it will be expressed from the endogenous virus CP promoter [illustrated by green fluorescent protein (GFP) in Figure 1; Shivprasad et al., 1999]. A second CP promoter of lower transcriptional strength, divergent in sequence from the endogenous (TMV U1) CP promoter, is placed downstream of the heterologous coding region, and a virus CP gene is then added. This encodes a third subgenomic RNA allowing the virus vector to express all requisite genes for virus replication and systemic movement in addition to the heterologous gene intended for overexpression (Figure 1). GENEWARE® vectors infect various tobacco-related species (genus Nicotiana), including tabacum, benthamiana and a KBP-proprietary Nicotiana hybrid species, Nicotiana excelsiana (Fitzmaurice, 2002). The infectious vector RNA enters plant cells via wounds induced by an abrasive. The virus replicates in the initial cell, moves to adjacent cells to produce round infection foci and then enters the plants’ vascular system for transport to aerial leaves. There, it systematically infects the majority of cells in each infected leaf (as illustrated using GFP in Figure 1). The foreign gene is expressed in all cells that express other virus protein products, including the replicase, MP and CP. The foreign protein is deposited in the site dictated by its protein sequence, either naturally or purposely engineered (Turpen, 1999; Pogue et al., 2002).

Figure 1.

 Genomic structure of tobacco mosaic virus (TMV) and illustration of construction and utility of GENEWARE® expression system. (a) Shows the genomic organization of TMV and the positions of two subgenomic promoters (bent arrows) driving expression of subgenomic messenger RNAs encoding movement protein and coat protein, respectively. Replicase proteins are translated from the genomic RNA. GENEWARE® vectors are constructed by insertion of an additional subgenomic RNA promoter and multiple cloning site for insertion of foreign genes (b) such as the green fluorescent protein (GFP) shown. Infectious cDNA clones of the recombinant TMV genome are transcribed from the T7 bacteriophage RNA promoter, followed by infection of plants, such as Nicotiana benthamiana, with infectious RNA transcripts, shown in (c). The plants in (c) are shown 2 and 6 days post inoculation under white light (top) and ultraviolet light illumination (bottom). Expression of GFP and systemic spread of GFP carrying GENEWARE® are clearly visible under UV light.

A range of human enzymes, antimicrobials, cytokines, subunit vaccine components and immunoglobulin fragments have been produced using the GENEWARE® system. Selected examples of highly purified proteins that have been subjected to potency testing are shown in Table 1. The results obtained from these proteins expressed from the GENEWARE® system in Nicotiana hosts were extracted using either tissue homogenization and clarification methods or leaf infiltration and isolation of interstitial fluids (Pogue et al., 1998; Turpen, 1999) and purified through differential separation and standard chromatographic separations (see references in Table 1). Purity of the recombinant proteins was determined by densitometric analysis of overloading of Coomassie brilliant blue-stained sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) gels and high pressure liquid chromatography (HPLC), when appropriate. Potency determinations were made testing the specific activities of each product with appropriate enzymatic or cytokine controls. In each case, highly purified proteins with specific activities matching established controls were observed demonstrating the broad classes of proteins that can be effectively expressed and purified from plants treated with transient expression systems. In each of the examples presented above, the control proteins used were either obtained from established research vendors or were proteins manufactured under good laboratory practices.

Table 1.   Qualities and bioequivalence of GENEWARE® produced pharmaceutical proteins and peptides
ProductSize (kDa)Results
  1. *Protein name and species of origin is indicated.

  2. Gelderman et al. (2004).

  3. This report.

  4. §Kentucky BioProcessing, LLC, unpubl. data.

  5. O’Keefe et al. (2009).

  6. **Du et al. (2008).

  7. ††Grill et al. (2005); Palmer et al. (2006).

  8. ‡‡McCormick et al. (1999, 2003, 2008).

α Galactosidase A (human)*48.5>98% purity, comparable enzymatic activity with CHO cell-derived controls and preclinical efficacy demonstrated
Aprotinin (bovine)6.5>99% purity, comparable specific activity with pharmaceutical product Trasylol®
Granulocyte colony-stimulating factor (human)§18.8>95% purity, bioequivalence to Neupogen using specific cell proliferation activity assay
Griffithsin (Griffithsia)12.7>99% purity, bioequivalence with natural product and potent neutralization of 12 different human immunodeficiency virus strains
Hepatitis B core antigen (Hepatitis B virus)§31>95% purity, conservation of virus-like particle structure and immunoreactivity
Interferon α 2a (human)§19.3>99% purity, bioequivalence with WHO standard in antiviral and antiproliferative activity assays
Interferon α 2b (human)§19.3>99% purity, bioequivalence with WHO standard in antiviral and antiproliferative activity assays
Interleukin-2 (human)§15.4>97% purity, bioequivalence in cell proliferation assays with interleukin-2 standards
Lysosomal acid lipase (human)**50.6>99% purity, bioequivalence with standards and preclinical efficacy demonstrated
Lysozyme (bovine)§14>85% purity, comparable enzymatic activity with natural and yeast derived protein standards
Papillomavirus capsid fusion (Human Papillomavirus)††19>99% purity, preclinical efficacy demonstrated in two different models
Single chain antibody fragments (human)‡‡∼30>95% purity for 16 different human idiotypic proteins, preclinical efficacy and clinical safety and immunogenicity

Biologic example: aprotinin


Aprotinin is a 58 amino acid active serine protease inhibitor of bovine origin that is processed from a preproprotein precursor (Laskowski and Kato, 1980). The active protein conformation requires three disulphide bridges and appropriate processing from both N-terminal and C-terminal prepropeptides. Aprotinin has been explored for clinical applications for four decades for a variety of clinical indications (Beierlein et al., 2005). Bayer HealthCare Pharmaceuticals’ Trasylol®, natural aprotinin, was a FDA-approved product indicated for prophylactic use to reduce perioperative blood loss and the need for blood transfusion in patients undergoing cardiopulmonary bypass in the course of coronary artery bypass graft surgery (CABG; Munoz et al., 1999; Sedrakyan et al., 2004). The drug, manufactured from residual bovine lung materials, was approved in the United States in 1993. However, recent international studies have indicated increased risk of in-hospital death and 5-year mortality rates among aprotinin recipients when compared with non-recipients (Mangano et al., 2006, 2007). In late 2008, Bayer HealthCare announced that marketing of the product was temporarily suspended pending review of additional clinical studies (Stamou et al., 2009). In spite of the adverse events associated with the drug in CABG patients, clinical studies continue to explore the application of aprotinin in other indications, both prophylactic and therapeutic, where the control of pathophysiological inflammatory cascades is desirable. These ongoing studies suggest that the market for aprotinin could expand once again provided an alternative active pharmaceutical ingredient (API) to bovine tissue could be more reliably produced without raising concerns over animal-associated adventitious agents, such as bovine spongiform encephalopathy prions (Maffulli et al., 2008; Orchard et al., 2008; Rademakers et al., 2009).

To date, several groups have expressed and purified recombinant bovine aprotinin (r-aprotinin) from transgenic plant materials. The crude yields of r-aprotinin varied per system, examples include 0.17% total protein in the corn seed (Azzoni et al., 2002; Zhong et al., 2007), 0.65 mg/L plant media or 3.7% of secreted protein present in Spirodela (duckweed) growth media (Rival et al., 2008) and 0.5% total soluble protein in selected leaves in transplastomic tobacco (Tissot et al., 2008). These measurements of yield are in general difficult to compare because of the vastly different levels of protein present in the various targeted tissues or the efficiencies in extraction from these tissues. This is most clearly seen with r-aprotinin expressed in transgenic corn seed. Total soluble protein concentration is much lower when the entire seed is extracted compared with the germ (Zhong et al., 2007). Optimized extraction methods and selective extraction from the germ resulted in a >10-fold increase in recovered r-aprotinin activity from corn seed (Zhong et al., 2007). These results illustrate the critical nature of downstream processing efficiencies to ensure yield and quality of purified protein product (Zhong et al., 2007; Plesha et al., 2009). Further, purified proteins showed comparable protein size and trypsin inhibitory activity (Zhong et al., 2007; Rival et al., 2008; Tissot et al., 2008). However, extensive analysis of the identity, purity and potency of the product was not presented. In each cited study, the limited product accumulation required detection by immunoassay or activity assays—visualization of product in crude plant lysates by protein gel analysis was not provided. In contrast to transgenic expression strategies, transient plant expression vectors generally offer higher yield potential enabling product analysis in direct plant lysates and development of appropriate product release tests. Aprotinin serves as a promising product candidate well suited for transient plant expression.

GENEWARE® r-aprotinin production

A synthetic cDNA of the mature bovine aprotinin gene was constructed as an in-frame fusion with the Nicotiana benthamiana (N. benthamiana) extensin signal peptide (Figure 2). This genetic arrangement was chosen to simplify the post-translational processing of r-aprotinin with regard to its preproprotein structure. The expression cassette was sub-cloned into the TMV-based GENEWARE® vector under the control of the T7 RNA polymerase promoter to produce expression plasmid construct pKBP2602. RNA transcripts were prepared and inoculated on N. benthamiana. Characteristic viral symptoms, vein clearing and leaf curling, were noted ∼6–12 days post-inoculation (dpi). Representative plants were extracted at 14 dpi by total leaf and stem homogenization and analysed for the presence of r-aprotinin and activity. The pH of this homogenate was acidified and clarified by centrifugation. The expression of r-aprotinin was evaluated by reducing SDS-PAGE and showed the accumulation of the TMV CP and the r-aprotinin that co-migrates with the Trasylol® control (Figure 3a). The molecular mass of r-aprotinin in the clarified homogenate was the expected molecular mass of 6512 Da as determined by matrix-assisted laser desorption–ionization time-of-flight mass spectrometry. Further, significant inhibition of serum protease activity was also determined in the extract using trypsin inhibition assays and conversion of activity into trypsin inhibitory unit (TIU). Inhibition activity showed ∼7100 TIU/mg of extract protein, comparable to that observed for native bovine aprotinin (Table 2; Fritz and Wunderer, 1983).

Figure 2.

 The recombinant bovine aprotinin expression construct is described in both nucleic acid coding strand (a) and deduced amino acid sequence (b). The modified aprotinin gene sequence is shown with Nicotiana extensin signal peptide underlined (nucleic acid and deduced amino acid) and mature aprotinin sequence (nucleic acid and deduced amino acid) not underlined. The DNA sequence of the synthetic aprotinin gene was constructed using codon biases based on the tobacco mosaic virus coat protein sequence.

Figure 3.

 Expression and extraction of recombinant bovine aprotinin (r-aprotinin) in Nicotiana plants. Virion preparations containing tobacco mosaic virus expression vector encoded by plasmid pKBP2602 were inoculated on Nicotiana benthamiana plants. Plants were harvested 14 days post inoculation. (a) Initial plant homogenate is represented in lane 2, while the supernatant derived from clarified homogenate is shown in lane 3. Trasylol®, bovine-purified aprotinin (2 μg), is provided in lane 4 for control. Molecular weight marker is shown in lane 1 and relevant markers with known molecular weight provided at left. (b) Trasylol® (1.5 μg/lane) was loaded in triplicate in lanes 1–3. Purified r-aprotinin (1.5 μg/lane) was loaded in triplicate in lanes 4–6. Molecular weight markers containing known molecular weight proteins are loaded at right. Proteins were analysed using 4–12% Bis–Tris sodium dodecyl sulphate polyacrylamide gel electrophoresis gels and subjected to Coomassie Brilliant blue-staining.

Table 2.   Test methods and results of aprotinin comparisons
AssayComparative attributeR-aprotininTrasylol®
  1. RP-HPLC method separates non-oxidized and oxidized forms of r-aprotinin.

  2. MALDI-TOF MS, matrix-assisted laser desorption–ionization time-of-flight mass spectrometry; SDS-PAGE, sodium dodecyl sulphate polyacrylamide gel electrophoresis; RP-HPLC, reverse phase high pressure liquid chromatography; GC, gas chromatography; MS, mass spectrometry; KIU, Kallikrein inactivation unit; TIU, trypsin inhibitory unit; EU, endotoxin units; r-aprotinin, recombinant bovine aprotinin.

Identity by tryptic digest MALDI-TOF MS mass mappingConforms with bovine lung aprotinin predicted tryptic fragments and fragment derivatives (84% amino acid coverage)ConformsConforms
Identity by MALDI-TOF MS6512 Da ± 0.05%6512 Da6512 Da
Identity by amino acid analysisConforms with bovine lung aprotinin amino acid compositionConformsConforms
Purity by SDS-PAGEPurity>99%>99%
Purity by RP-HPLCPurity87.6% + 12.4(Ox)%86.3% + 5.7(Ox)%
Purity by GC/MS small molecular weight host toxicantsPurityComparable levels of target compoundsComparable levels of target compounds
Purity by appearanceClear, colorless, free of visible particlesClear, colorless, particle freeClear, colorless, particle free
Potency by specific activity>6500 KIU/mg protein or >5.0 TIU/mg protein7175 KIU or >5.7 TIU6859 KIU or 5.4 TIU
Endotoxin<1 EU/28 mg<1 EU/28 mg<1 EU/28 mg

Based on these promising results, large-scale manufacturing of aprotinin was conducted using plants grown under greenhouse conditions (N. benthamiana) or in open field cultivation (N. excelsiana). TMV virions were isolated from plants infected with transcripts derived from pKBP2602 plasmid DNA. The virion was confirmed for its ability to produce aprotinin in inoculated plants, and reverse transcriptase polymerase chain reaction was used to confirm the presence of the aprotinin expression cassette in the recombinant virus genome (data not shown). Virions were mixed with an abrasive and spray inoculated on either greenhouse-grown or field-grown plants. Plants were monitored for virus symptoms and were harvested in bulk 14 days post infection. R-aprotinin accumulated in the leaf and recovered by extraction of the interstitial fluid of leaves or through total leaf homogenization. Maximum yields (∼40% enhanced protein recovery; data not shown) were noted by total homogenization extraction; therefore, this approach was adopted for large-scale manufacturing. The leaves were homogenized and clarified followed by concentration of the r-aprotinin using ultrafiltration. The r-aprotinin was purified using cation exchange chromatography followed by reverse phase chromatography. The final product was concentrated across a 1 kDa molecular weight cut-off membrane (MWCO), pH adjusted, sterile filtered and vialed.

Because of differences in soluble protein content in various plant extracts, KBP and its collaborators report protein accumulation as milligrams per kilograms of fresh weight of extracted tissues. This approach takes into account variables in the extraction and the efficiency of the method used and provides a basic and relevant crude production level from which to base predictable economics. GENEWARE® production of r-aprotinin in greenhouse-grown Nicotiana plants showed crude and purified yields of ∼750 and 400 mg/kg, respectively. Field-produced plants showed crude and purified yields of ∼300 and 150 mg/kg, respectively. However, costs of plant agronomic practice were approximately fivefold less in open fields compared with greenhouse production plants modulating the reduction in absolute protein expression. In spite of the differences in methods of reporting protein yield, these results suggest transient expression offers superior yields than transgenic approaches (Azzoni et al., 2002; Zhong et al., 2007; Tissot et al., 2008). The exploitation of agriculture scale allows production of 1 kg of purified r-aprotinin from 2500 square feet of greenhouse space or 1.5 acres of field transfected Nicotiana plants. These results demonstrate that transient plant production systems can provide product quantity and economies of scale more competitive than traditional production systems.

GENEWARE® r-aprotinin characterization

The purified r-aprotinin was subjected to rigorous analytical testing. Table 2 shows the types of release tests performed on the plant-produced r-aprotinin lots and a comparison of results from greenhouse-produced product with that of Trasylol®. The identity of the proteins was virtually identical as determined by tryptic peptide analysis, amino acid analysis and reactivity with anti-aprotinin mAb (Table 2; data not shown). Further, the molecular mass of both proteins was found to be identical at 6512 Da (Table 2). The potency of the r-aprotinin was consistently higher than Trasylol®, as measured by Kallikrein inactivation units per milligram of purified protein (Table 2). Purity analyses showed no detectable protein impurities by overloaded SDS-PAGE, exact migration pattern on gels, reverse phase HPLC (RP-HPLC; Figure 3b; Table 2) and immunoassays (data not shown). Neither Trasylol® nor GENEWARE® r-aprotinin product showed any immunoreactivity with a polyclonal antibody generated against crude Nicotiana protein extracts demonstrating an absence of host-derived proteineous impurities in the final product (data not shown). Detailed RP-HPLC analysis revealed minor aprotinin variants. Truncated aprotinin species, including desAla58 and desAla58Gly57, and various oxidized aprotinin species were quantitated in the Trasylol® product as 8% and 5.7%, respectively. R-aprotinin showed no detectable truncated species, although it contained oxidized forms at 12.4% of the final product. It is well known that oxidation is common on methionine 52 (Concetti et al., 1989). Indeed, oxidation of the methionine residue was noted in chloroplast-produced aprotinin in specific plant lines (Tissot et al., 2008). The oxidated species did not exhibit reduced inhibition activity (data not shown; Concetti et al., 1989). Exploitation of changes in physicobiochemical behaviour of the oxidized protein allowed efficient removal of the oxidized forms using a second, subsequent reverse phase chromatography method (data not shown).

Comparison of field-produced r-aprotinin with greenhouse produced revealed virtually identical products (Table 3). Purity and identity analyses revealed identical results (examples provided electrospray ionization time-of-flight mass spectroscopy, appearance and SDS-PAGE). Protein concentrations of the final bulk drug differed before vialing because of degree of concentration, yet both were under predetermined bulk drug release specifications. The potency of the field product was comparable with that of the greenhouse-produced API (Table 3). The stability of the greenhouse-produced r-aprotinin, in liquid form, was monitored over a 31-month period with real-time storage at 4 °C (Table 4). No significant changes in the purity, protein concentration and specific activity were observed at any point in the 31-month test period or when the initiation point was compared with terminal time point (Table 4). These results demonstrate the consistency and quality of the GENEWARE® r-aprotinin produced from transiently transfected Nicotiana plants of different species and production conditions. Further, the data presented show comparable results with that of pharmaceutical products, such as Trasylol®, demonstrating the ability of the GENEWARE® transient plant-expression system to produce product matching those of FDA-approved biologics.

Table 3.   Test methods and results of r-aprotinin comparisons (greenhouse versus field grown)
AssayComparative attributeR-aprotinin (Greenhouse)*R-aprotinin (field)
  1. *Lot 07A0009.

  2. Lot O8A0025.

  3. ESI-TOF MS, electrospray ionization time-of-flight mass spectrometry; SDS-PAGE, sodium dodecyl sulphate polyacrylamide gel electrophoresis; TIU, trypsin inhibitory unit; EU, endotoxin units; r-aprotinin, recombinant bovine aprotinin.

Identity by ESI-TOF MSAverage molecular mass between 6508.2–6514.8 Da6511.4 Da6511.8 Da
Purity by SDS-PAGE (reduced)≥95% of r-aprotinin as determined by densitometry (% band)>99%>99%
Protein concentration by UV absorbance≥5.0 mg/mL21.3 mg/mL18.3 mg/mL
Purity by appearanceClear, colorless to amber, free of visible particlesClear, light yellow, particle freeClear, light yellow, particle free
Potency by TIU>5.0 (B) TIU/mg protein5.7 TIU/mg5.6 TIU/mg
Endotoxin<1 EU/28 mg<1 EU/28 mg<1 EU/28 mg
Table 4.   Stability testing of r-aprotinin product*
Assay (# months)Purity (%)Protein concentration (mg/mL)Potency§ (TIU/mg)
  1. r-aprotinin, recombinant bovine aprotinin.

  2. *Lot 07A0009.

  3. Purity—sodium dodecyl sulphate polyacrylamide gel electrophoresis and densitometry; release specification ≥95%.

  4. Concentration—OD280 and bicinchoninic acid method; release specification ≥5.0 mg/mL.

  5. §Trypsin inhibitory unit (TIU); release specification ≥5.0 TIU/mg.

3Not determined21.55.7
6Not determined21.35.4
12Not determined21.06.4
16Not determined21.05.1
24Not determined21.35.5

Case study II: magnICON® system

Independent-virus systems, such as GENEWARE®, must maintain all activities of the virus to successfully colonize an infected host plant, in addition to subgenomic promoter and RNA replication functions responsible for production of recombinant protein products. The necessity of MP and CP coding regions reduces the genomic capacity of the virus and reduces the size of proteins efficiently produced by such systems to <70 kDa. The magnICON® system represents a distinct minimal-virus approach for using tobamovirus-based vectors where systemic movement functions are eliminated to transiently express heterologous proteins in permissive hosts, such as N. benthamiana (Marillonnet et al., 2004; Gleba et al., 2005, 2007, 2008). In magnICON® vectors, the MP and CP genes may be eliminated through genetic deletion, and the gene encoding a pharmaceutical protein is placed under the control of the endogenous CP subgenomic promoter. This minimal-virus strategy provides increased genomic capacity to express larger proteins than typically compatible with independent-virus systems. The magnICON® system utilizes the Agro-infiltration system to introduce the plant viral vector expression system, as intact virus vectors or in distinct modules, including a module containing the gene(s) of interest (Marillonnet et al., 2004; Gleba et al., 2008). If the distinct module strategy is used, the components are assembled in planta and the resulting DNA is transcribed, spliced and translated, resulting in high yields of the expressed protein (Marillonnet et al., 2004). Numerous heterologous proteins have been produced using this system, including cytokines, interferon, bacterial and viral antigens, growth hormone, single chain antibodies and mAbs at levels of 1–10 g/kg (Giritch et al., 2006; Gleba et al., 2007, 2008).

Nicotiana benthamiana plants are ideally suited for the magnICON® expression system because it relies on Agrobacterium infection to mediate initial entry and introduction of the viral expression vectors. Nicotiana benthamiana is known to be nearly universally susceptible to plant viruses, partially based on a defective form of RNA-dependent RNA polymerase found in its genome (Yang et al., 2004). This viral susceptibility allows external viral replicases, delivered as part of the magnICON® expression system, to successfully replicate the delivered genes. The combination of ease of infection with bacterial and viral components and a long history of experimental use have made N. benthamiana a common host for the expression of many recombinant proteins. The flexibility of Agro-infiltration of N. benthamiana also offers the ability to introduce more than one expression vector into a host plant in a given treatment. The magnICON® system exploits this advantage to be an efficient system for the production of heteromeric recombinant proteins, such as mAbs. For the production of mAbs, the magnICON® system employs two non-competitive virus vectors: one based on turnip vein-clearing tobamovirus (TVCV) and the other based on PVX (Giritch et al., 2006; Hiatt and Pauly, 2006). In mAb production, two magnICON® virus expression vectors are delivered by Agro-infiltration into the same plant. Each vector replicates independently and expresses heavy and light chains (HC and LC) in the same cells. The two chains self-assemble into authentic and functional mAbs and are secreted to the apoplastic space at yields up to 1 g/kg fresh weight (Giritch et al., 2006; Gleba et al., 2007, 2008).

To date, most published studies concerning transient expression systems have detailed expression under laboratory conditions yielding milligram to gram levels of product (see references in this article and Floss et al., 2007; Sharma and Sharma, 2009). The ability to scale manufacturing to multi-kilogram quantities of plant material is a critical step to validate the use of plant systems for therapeutic protein production. To accomplish this task with transient expression systems, plant inoculation as well as protein extraction and purification methodologies must be adapted. Although plant processing and purification methods can be modelled from food processing and standard biomanufacturing systems (Doran, 2000; Pogue et al., 2002), inoculation methods for magnICON® vectors require adaptation of the traditional laboratory-based Agro-infiltration method to a robust, large-scale process.

Working in cooperation with Bayer Innovation, GmbH and Icon Genetics, KBP adapted the Agro-infiltration process to accommodate the infiltration of kilograms of plants per hour, allowing 25–75 g of antibody to be produced per greenhouse lot using the magnICON vectors (KBP Agro-infiltration system shown in Figure 4). The process begins with seeding plants in a tray system that, as plants grow through a hole in the tray lid, the aerial portion of the growing plant is physically separated from the soil and root components. The trays are grown in a controlled growth environment until reaching appropriate size and then manually loaded onto a conveyor, inverted 180° and moved through a vacuum rated autoclave with reservoirs containing the Agrobacterium solution. Sufficient vacuum is applied and then released to allow entrance of the Agrobacterium solution into the interstitial spaces of the submerged plant tissues. Upon completion of vacuum cycle, trays are placed into an upright position and transported back to a controlled growth environment (Figure 4). The system has been designed to operate with 450–750 kg of green biomass in an 8-h production cycle, depending on plant growth conditions and protein product design. Following Agro-infiltration, plants are grown in greenhouses for 7–14 days depending on product-specific optimization of plant biomass and yield (Figure 4). Following growth period, plants are harvested and subjected to standard protein extraction methods. To demonstrate the adaptability of the magnICON® plant virus transient expression system for large-scale, multi-gram, biomanufacturing, the production of a neutralizing mAb binding the CCR5 co-receptor follows.

Figure 4.

 ‘At scale’Agrobacterium tumefaciens-mediated transfer-DNA delivery (Agro-infiltration) system. (a) Plants are seeded in trays with specially designed lid to permit growth, yet provide a barrier for soil and root components. Following plant growth to appropriate size, ten trays are loaded on each of four conveyors to enter the vacuum-rated chamber, shown in (b) with both for and aft doors open and empty. Conveyors rotate 180° and enter the chamber (c), plants are submerged in Agrobacterium-containing solution and vacuum is applied and released. Plants are removed from chamber and rotated to upright position using conveyors and subsequently transferred to greenhouses for growth and product accumulation (d).

Biologic example: mAb


mAbs represent the fastest growing sector in the biopharmaceutical market ($35 billion in 2008 revenue; La Merie Business Intelligence, 2009) and are used therapeutically in many different clinical areas, including infectious disease, oncology, inflammation, allergy and cardiovascular (Hoentjen and van Bodegraven, 2009; Weiner et al., 2009). Many companies are exploring broader applications of mAbs, including their use to block the entry of viruses into cells to prevent infection (Trkola et al., 2001; Murga et al., 2006; Shearer et al., 2006; Jacobson et al., 2008). For example, CCR5 acts as a co-receptor for human immunodeficiency virus type 1 (HIV-1) entry into cells (Moore et al., 1997), and it has been suggested that a microbicide acting to block CCR5 may serve as a possible strategy for the prevention of sexual transmission of HIV-1 (Gaertner et al., 2008). Further, CCR5 appears to be non-essential for human health because individuals with CCR5-Δ32 alleles (essentially a CCR5 knockout) are healthy (Dean et al., 1996).

Anti-(α) CCR5 mAbs are currently in clinical development as HIV therapeutics (Jacobson et al., 2008) due to their potent blockage of CCR5-mediated HIV-1 cell entry in vitro (Trkola et al., 2001; Murga et al., 2006; Shearer et al., 2006). Despite the fact that small molecule CCR5-specific drugs are potent chemokine antagonists, neutralizing antiviral concentrations of CCR5 mAbs [inhibitory concentration (IC)50 0.1–1 μg/mL] did not block the natural activity of CCR5 in vitro, although CCR5 antagonism was observed at higher concentrations (IC50 of 45 μg/mL; Olson et al., 1999). Similarly, at concentrations ranging to 100 μg/mL, the mAbs had no effect on lymphocyte proliferation in response to mitogenic and allogeneic stimulation (Gardner et al., 2003) and did not mediate significant levels of antibody-dependent cellular cytotoxicity or complement-dependent lysis of CCR5-expressing cells. A CCR5 mAb has been shown to neutralize escape mutants raised against small molecule CCR5 inhibitors (Pugach et al., 2008). Systemic delivery (intravenous and subcutaneous) of an αCCR5 mAb has shown strong antiviral activity in Phase 1b (Jacobson et al., 2008) and later stage clinical trials (Olson and Jacobson, 2009) as well as a good safety profile. Although the potent anti-HIV activity of αCCR5 mAb suggests this molecule could be quite valuable in prophylactic as well as therapeutic applications to control HIV infection, the costs of αCCR5 mAb mammalian production are prohibitive.

Plants were first shown to correctly fold and produce antibodies in 1989 with continued demonstration of a variety of antibody candidates through the efforts of many investigators (see Ma et al., 2003 and references therein). However, production levels and characterization of these products have been slow to emerge in the published literature. Reported expression levels of mAbs expressed via transgenic plants are rather low (<30 mg/kg plant tissue; Fischer et al., 2003; Floss et al., 2007; Gaertner et al., 2008; Ma et al., 2003; Valdés et al., 2003). Indeed, a highly efficient process showing protein purity of >90% was demonstrated from transgenic tobacco plants with a yield of recombinant antibody of ∼25 mg/kg fresh weight tissues (Valdés et al., 2003). However, the time required to construct, select and grow these lines for large-scale production is predicted to be >24 months. Transient plant expression offers a solution to this challenge. Using the magnICON® system, the time and subsequent cost efficiency of agricultural-scale production of mAbs offer a viable manufacturing option for products, including mAbs as microbicides for the prevention of HIV-1 infection and a means to apply promising products to a broader range of individuals.

magnICON® mAb production

Mapp Biopharmaceutical, Inc. is developing a αCCR5 mAb as an intravaginal topical antimicrobial agent to reduce mucosal transmission of HIV-1. This humanized mAb specifically binds the ligand-binding domain of the human chemokine receptor and HIV co-receptor, CCR5. Using the magnICON® system, HC and LC of the αCCR5 mAb are inserted in two different virus expression vectors, TVCV and PVX (Figure 5). To express the αCCR5 mAb in plants, Working Cell Banks (WCB) of Agrobacterium cell lines, containing mAb LCs and HCs (HC; 31 160-LC, 26 211-HC, respectively), were derived from Master Cell Banks (MCB; see Figure 5 for flow diagram of αCCR5 MCB construction). WCBs were amplified, and overnight cultures were mixed and diluted in infiltration buffer. Nicotiana benthamiana plants were subjected to the KBP Agro-infiltration process using an infiltration buffer containing the two Agrobacterium cell banks (311 600-LC, 26 211-HC) and allowed to grow for 10 dpi (Figure 4). At this time, all aerial portions of the treated plants were harvested and αCCR5 mAb extracted and purified. Briefly, plant materials were homogenized (typically 40–60 kg of plant material/extraction). The homogenized materials were then subjected to a horizontal screw press to separate plant fibre and ‘green juice’ extract. The pH of the extract was adjusted and clarified using a plate and frame filter press. The clarified extract was then loaded on a protein A column, and bound antibody was further treated by filtration and multi-ion exchange resin column. The column eluant was pooled and diafiltered using a 30-kDa MWCO membrane, sterile filtered using a 0.2-μm filter and vialed.

Figure 5.

 Flow diagram of the anti-chemokine (C–C motif) receptor-5 (αCCR5) Master Cell Bank construction. Strain development process is presented and used for infiltration inoculation of Nicotiana benthamiana plants for the production of αCCR5 monoclonal antibody.

magnICON® mAb characterization

During the manufacturing process, the purity and protein concentration of the mAb product were monitored via SDS-PAGE gels and OD280 measurements. Presence of endotoxin was also monitored to ensure appropriate recovery and freedom from contaminating materials. Bulk drug substance for the mAb was vialed immediately and stored as drug product. The final αCCR5 mAb product was subjected to rigorous release testing in accordance with predefined acceptance criteria specifications (Table 5). Example results showed high purity obtained through this purification process, >99% by SDS-PAGE and 97% monomer by size exclusion HPLC. The potency was also measured using a CCR5-specific enzyme-linked immunosorbent assay and revealed a highly active mAb product with expected specific activity. Further, non-protein impurities, such as nicotine, were reduced to parts per billion levels (Table 5). The product showed no significant levels of endotoxin, and no detectable bioburden per millilitre (Table 5). These are important findings because the Agro-infiltration process involves infiltration of all aerial portions of the plants with a solution containing Agrobacterium strains encoding the production viruses. The copious quantities of bacteria would provide opportunity for the retention of these contaminants and impurities. Nevertheless, the purification strategy and aseptic environment led to efficient removal.

Table 5.   Release test specifications and results for anti-chemokine (C–C motif) receptor-5 monoclonal antibody
ParameterTest methodRelease specificationProduction batch results
  1. OD, optical density; PAGE, Isoelectric focusing polyacrylamide gel electrophoresis gels; pI, isoelectric point; SDS-PAGE, sodium dodecyl sulphate PAGE; HPLC, high pressure liquid chromatography; LMW, low molecular weight; IC, inhibitory concentration; EU, endotoxin units; CFU, colony forming units.

  2. *Isoelectric focusing PAGE.

AppearanceVisualClear, colorless to amber, liquidClear, colorless, liquid
Protein concentrationOD2800.7–1.3 mg/mL1.1 mg/mL
Identity*Isoelectric focusing4–5 bands pI range 8.4–9.75 bands pI 8.4–9.7
PuritySDS-PAGE≥95% (sum of heavy and light chain)>99%
PuritySize exclusion HPLC≥90% monomer97% monomer
≤10% aggregation0.44% aggregation
≤10% LMW2.63% LMW
PotencyViral neutralizationIC50 < 1 μg/mL0.08 μg/mL
Physical/chemical propertiespH5.5–6.5 pH units6.2 pH units
Physical/chemical propertiesConductivity9.15 mS/cm ± 0.59.15 mS/cm
SafetyEndotoxin<10 EU/mL0.5 EU/mL
SafetyBioburden<10 CFU/mL<1 CFU/mL
Impurities1-methyl-2-[3-pyridyl]-pyrrolidine (nicotine) concentrationFor information only<50 ppb
ImpuritiesResidual host cell proteinFor information only<0.2%

The αCCR5 mAb was produced in a scalable manner by the magnICON® system. Processing greenhouse-propagated plants provides for 25–75 g purified product lots at an expected yield of ∼250 mg/kg fresh weight of plant materials. These levels are ∼10-fold greater than production levels reported for transgenic systems (Valdés et al., 2003). The quality of the vialed product supports its use in early-stage clinical investigations as a novel, biologic microbicide to stem the tide of HIV-1 infection within at-risk populations and provides further proof-of-concept for transient plant expressions systems.


Plants have been touted as an attractive alternative for pharmaceutical protein production to the current mammalian or microbial cell-based systems. The potential for reduced production costs coupled with the low risk for contamination with human-tropic adventitious agents and other impurities have led many to hypothesize that agricultural systems may offer the next wave for pharmaceutical product production (Ma et al., 2003; Floss et al., 2007; Lico et al., 2008; Plasson et al., 2009). However, for this to be a reality, the quality of products produced at a relevant scale must match the common release criteria in the pharmaceutical industry. A review of the literature demonstrates the variety of recombinant proteins that can be produced in transgenic and transient plant virus expression systems, as well as the quality of the resulting purified products (Ma et al., 2003; Floss et al., 2007; Lico et al., 2008; Plasson et al., 2009; Sharma and Sharma, 2009). Detailed review of the GENEWARE® and magnICON® systems provides further demonstration that quality biologics, such as r-aprotinin and αCCR5 mAb, respectively, can be produced using non-food/feed, non-genetically modified plants.

In addition to the quality of proteins produced from transient plant expression systems, speed to develop the virus inoculum and MCB for the expression of a given protein provides significant advantages. As described by Hiatt and Pauly (2006), the ability of transient systems to produce milligrams of product can be as little as 2 weeks and production of grams may take only a few weeks more. These timeframes are much shorter than the requirements to transfect, select, establish and characterize mammalian cells, transgenic animal or traditional plant-based systems. Both the GENEWARE® and magnICON® systems, through the adaptation of both virus inoculation and Agro-infiltration methods to large-scale biomanufacturing systems, can yield productivity of recombinant proteins at levels of 200–1000 mg/kg fresh weight tissue in as little as 3 months. Further, the yields that can be expected from these systems can be quite high, ranging from 0.25 to 0.75 g/kg when extracting >100 kg of crude plant material. These values are 10-fold greater than production levels of the same proteins in transgenic plant systems. Lastly, these transient systems yield correctly folded monomeric and multimeric proteins that show release properties comparable with standard pharmaceutical products attesting to the robustness of plant expression capabilities. The data reviewed here strongly support the contention that transient plant expression systems have moved beyond the proof-of-concept stage in development and offer a legitimate cost-competitive alternative for recombinant protein production.


We appreciate the efforts of Terri Cameron, Mark Smith, Sarah Doucette, Steve Reinl, Long Nguyen, Amanda Lasnik, Lee Hamm, Hal Padgett, Wayne Fitzmaurice and Peter Roberts for contributing to GENEWARE® expression results. We also acknowledge the contributions of Jennifer Bleckmann, Cara Working, Josh Morton and Jennifer Poole in the production and characterization of the r-aprotinin product. We thank Dr David Montefiori (Duke University) for performing the HIV neutralization assays. This project was supported in part by Award Number U19 AI 62150 from the National Institute of Allergy and Infectious Diseases. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Allergy and Infectious Diseases or the National Institutes of Health.