Review article: Commercialization of whole-plant systems for biomanufacturing of protein products: evolution and prospects
Correspondence (fax +1 859 323 1077; email firstname.lastname@example.org)
Technology for enabling plants to biomanufacture nonnative proteins in commercially significant quantities has been available for just over 20 years. During that time, the agricultural world has witnessed rapid commercialization and widespread adoption of transgenic crops enhanced for agronomic performance (herbicide-tolerance, insect-resistance), while plant-made pharmaceuticals (PMPs) and plant-made industrial products (PMIPs) have been limited to experimental and small-scale commercial production. This difference in the rate of commercial implementation likely reflects the very different business-development challenges associated with ‘product’ technologies compared with ‘enabling’ (‘platform’) technologies. However, considerable progress has been made in advancing and refining plant-based production of proteins, both technologically and in regard to identifying optimal business prospects. This review summarizes these developments, contrasting today’s technologies and prospective applications with those of the industry’s formative years, and suggesting how the PM(I)P industry’s evolution has generated a very positive outlook for the ‘plant-made’ paradigm.
Introduction—the origins of PM(I)P
The invention of plant transformation in the early 1980s (reviewed in Newell, 2000) was initially accompanied by an agronomic focus of application, the relatively low ‘metabolic’ level of transgene expression being suitable for nonnative enzymes, which would detoxify herbicides, and for xenobiotic proteins that would provide defence against insects. The choice of these agronomic traits, and the accompanying massive development process that included patent protection, formulation, on-farm utilization methods, pricing, branding, regulatory oversight, etc., paid off handsomely, with these simplest applications of agricultural biotechnology becoming ‘the fastest adopted crop technology in the recent history of agriculture’ (ISAAA, 2009).
The initial breakthrough in plant transformation was rapidly followed by scientific advances that added considerable sophistication, such as the ability to suppress endogenous plant gene expression, to regulate transgenes with externally applied compounds, to express many nonnative genes simultaneously, etc. Among these developments, there emerged the capability to express a nonnative (‘foreign’) gene in a plant such that the resulting protein would accumulate to sufficient enrichment in the tissues that it could be productively extracted. This concept envisaged the plant as a ‘bioreactor’ for production of the protein in bulk, effectively an alternative to using contemporary microbial fermentation. While its very earliest practical demonstration is now difficult to trace, the idea of commercial application of what was then referred to as ‘molecular farming’ to manufacture valuable proteins in quantity is readily apparent in patenting activity dating from 1986 to 1990 (e.g. Goodman et al., 1987; filed 1986) and publications at that time (e.g. Hiatt et al., 1989; Vandekerckhove et al., 1989; Sijmons et al., 1990). Thus, the plant-made protein concept has been available, and steadily advanced, for just over 20 years. During this time, there have been many demonstrations of the expression in plants of a wide variety of protein-based pharmaceuticals, enzymes and other potential products, often with impressive yields and functional fidelity (Horn et al., 2004; Streatfield, 2007). However, in marked contrast with the transgenic agronomic traits, these ‘value-added’ applications are still in process of becoming established commercially, sales of PM(I)Ps having been thus far limited to a few research- and manufacturing-grade products (e.g. aprotinin produced in tobacco plants for nonclinical applications: KBP, 2010); evidently, the path to commercialization has been longer and more challenging than was originally anticipated.
This review examines the evolution of PM(I)P commercialization from a combined technological and business perspective, offers an explanation for the comparatively long development time of this biotechnology sector, and suggests that the industry is currently presented with new and much more appropriate growth opportunities than were possible in the previous two decades. For reasons of space limitation, the article considers only the whole-plant applications. Reviews of the continuing and exciting progress towards commercialization of in-vitro plant-cell systems (including algae and nonvascular plants) for biomanufacturing of proteins can be found elsewhere (e.g. Decker and Reski, 2007; Mayfield et al., 2007; Huang and McDonald, 2009). Similarly, the field of ‘edible vaccines’, wherein the intact (or minimally processed) transgenic plant tissue is the final product, is sufficiently different from a business perspective to warrant separate treatment; a convenient overview of the edible-vaccines concept was provided by Mor et al. (2004).
Overview of the underlying technology
The PM(I)P protein-biomanufacturing strategy is based on specialized gene-expression technologies. The history, technical details and capabilities of these inventions have been frequently and recently reviewed (e.g. Fischer et al., 2004; Davies, 2005; Streatfield, 2007).They conveniently fall into a limited number of different categories according to the mechanism employed to introduce the nonnative (foreign) gene into the plant’s cells (Table 1). All these strategies have a common capability that defines them as being suitable for PM(I)P applications, namely the generation of very large amounts of gene product (protein), while each type achieves this so-called high-level expression in a different way. Nuclear genome-based expression may rely on inherent stability of the final protein product (e.g. Verwoerd et al., 1995) or potency of a particular genetic promoter (e.g. Dai et al., 2000), plastid expression primarily achieves its productivity through organelle abundance and gene copy-number (reviewed in Grevich and Daniell, 2005), and the virus-derived and replicon systems borrow the strategies that viruses naturally use to command and control the plant’s transcription and translation machinery for maximum output (reviewed in Gleba et al., 2004).
Table 1. ‘Enabling’ gene-expression technologies for PM(I)P applications
|Stably into plant’s nuclear genome||Usually less than c. 5%||Avidin in corn seed (Hood et al., 1997), endoglucanase in tobacco leaf (Dai et al., 2000)|
|Stably into plant’s plastid genome||Up to c. 45%||Industrial polymer in tobacco (Guda et al., 2000), insecticidal protein in tobacco (De Cosa et al., 2001)|
|Transiently into nucleus via recombinant Agrobacterium vector, with viral suppressor of gene silencing as enhancer||1–7%||Test proteins in N. benthamiana (Voinnet et al., 2003), lettuce (Joh et al., 2005)|
|Transiently into cytoplasm via recombinant viral vector, producing free polypeptide product or virion-presented protein||1–10%||Virion-presented single-chain immunoglobulin in N. clevelandii (Smolenska et al., 1998), rice amylase (free polypeptide) in N. benthamiana (Kumagai et al., 2000)|
|Transiently into nucleus or cytoplasm using components of deconstructed plant viruses||Up to c. 40%||Test protein in N. benthamiana (Marillonnet et al., 2005), human antitrypsin in N. benthamiana (Plesha et al., 2008)|
Table 1 also shows some of the highest enrichments of protein that these different expression systems have achieved, relating target-protein yield to total soluble protein in the tissue (% TSP) according to PM(I)P convention. A % TSP of 1–10 was once considered an arbitrary ‘benchmark’ for commercializing these technologies, but as several systems have subsequently achieved considerably in excess of 10% TSP one now begins to wonder about the upper limit for practical application. This is probably determined either by cytotoxic effects of the ‘foreign’ protein or by the ability of the cell to withstand diversion of energy and material resources to production of the nonnative product at the expense of normal cellular constituents without compromising plant health and overall yield.
Different versions of the expression systems exhibit individual attributes that can be important for certain applications, e.g. viral vectors will be relatively specific for the plant species that are the natural hosts for the corresponding viruses, and thus will be used in PM(I)P applications of those plants. Another important distinguishing feature is the dependence of some systems on external delivery of the expression vector into each plant which is to be used, in every generation of plants that are grown (Gleba et al., 2005). Such a ‘transfection’ procedure presumably obviates economical outdoor production of the PM(I)P crop, yet the resulting strong performance of the vector is highly advantageous once indoor production is accepted, ensuring that every single plant is made as productive as possible.
Table 2 lists some random examples of the multitude of proteins having diverse medical and industrial relevance that have been expressed in plants, illustrating the breadth of potential applications. The high proportion of biomedical products relative to industrial ones reflects the balance of product types that have been promoted commercially.
Table 2. Examples of some plant-made proteins
|Anti-malaria, presented-epitope vaccine||Turpen et al., 1995|
|Secretory antibody (IgA) to combat tooth decay||Ma et al., 1995|
|Anti-threat-agent (anthrax), antigen-type vaccine||Koya et al., 2005|
|Veterinary vaccine: e.g. anti-canine parvovirus||Molina et al., 2004|
|Anti-cancer (lymphoma) therapeutic protein||McCormick et al., 1999|
|Blood component: human serum albumin||Sijmons et al., 1990|
|Hormone therapy: human growth hormone||Leite et al., 2000|
|Leech anticoagulant, hirudin||Parmenter et al., 1995|
|Nutritional supplement: e.g. lactoferrin||Nandi et al., 2002|
|Glucocerebrosidase (therapy for Gaucher’s disease)||Cramer et al., 1996|
|Human interferon-alpha (therapy for hepatitis)||Zhu et al., 1994|
|Alpha-galactosidase (therapy for Fabry’s disease)||Gelderman et al., 2004|
|Human insulin (therapy for diabetes)||Nykiforuk et al., 2006|
|Wound-healing agent: human factor XIII||Gao et al., 2004|
|Clot-disruption agent: human t-pa||Hahn et al., 2009|
|Fermentation-system reagent: bovine aprotinin||Tissot et al., 2008|
|Surgical polymer: collagen||Stein et al., 2009|
|Drug delivery, tissue repair agent: elastin||Floss et al., 2010|
|Industrial polypeptide (synthetic gene)||Guda et al., 2000|
|Industrial enzyme: e.g. Clostridium xylanase||Herbers et al., 1995|
Commercialization over the past 20 years
As might be expected, early commercial interest in the 1980s took the form of investment by large established companies (e.g. Monsanto), investment by smaller and relatively new biotechnology companies (e.g. Calgene) and creation of PM(I)P-dedicated ‘start-up’ companies (e.g. CropTech, Biosource, etc.). Over the ensuing 20 years, some of these commercial interests have ceased to exist (e.g. CropTech, Chlorogen) or apparently de-emphasized PM(I)P work (e.g. Phytomedics, Monsanto), while others have been re-organized (e.g. acquisition of the technological assets of Large Scale Biology by Kentucky Bioprocessing) or acquired by larger concerns (e.g. acquisition of Icon Genetics by Bayer Innovation). New companies have continued to appear (e.g. recently Plantform) or to enter the PM(I)P field (e.g. Bayer Innovation), bringing new technology or product-focus to the industry. The reader interested in a general sense of current commercial activity is referred to the accompanying Supporting Information available online (based on that provided by Murphy, 2007, with permission).
Given the substantial investments being made in biotechnology and the life sciences from the early 1980s through the mid-2000s, one can ask why PM(I)P companies were unable to progress products significantly to market over that timeframe. There is no doubt that many of the expression strategies were technically successful and exhibited potential for scale-up, as evidenced by the considerable output of scientific publications, patent activity, field trials and presentations throughout that period. Positive tests of pharmaceutical efficacy in animal studies (for an early example see McCormick et al., 1999) and progress in human clinical trials (reviewed historically in Ma et al., 2005; Kaiser, 2008; recent examples include SemBioSys, 2009; Biolex, 2010) amply evidenced the remarkable potential of the new technology (note: the summary of clinical trials presented in Kaiser (2008) suggests that a product made by Cuban organization CIGB is ‘on market’, but this reviewer can find no additional material in support of this assertion. Moreover, the product in question appears to be a reagent-antibody used in vaccine purification rather than the pharmaceutical itself: Roumeliotis, 2006).
From the outset, the early PM(I)P companies were not the only means of producing recombinant proteins, as microbial fermentation was already harnessed for protein biomanufacturing in the hands of Genentech and other pioneers. Moreover, ‘vertical integration’, which proved so successful for the companies developing the transgenic agronomic traits, was not possible for those PM(I)P companies that did not own drug-candidates, novel enzymes, etc. And while commercial interest in plant-made proteins often emphasized pharmaceuticals over industrial products such as enzymes and structural polymers, presumably because of their higher earnings potential and appeal to investors, this focus posed an additional problem owing to the relative scarcity of protein-drug candidates during the industry’s early years, most pharmaceutical interests being slow to adopt ‘biologics’ (proteins) as prospective next-generation drugs. For example, by 1991 only five monoclonal antibody drugs had been approved in the United States, and by 2000 the total was still only 10, from a development history going back 20 years (Reichert, 2001). This contrasts markedly with subsequent growth, and today several therapeutic antibodies are successful market-leaders achieving annual sales of over $1 billion (Maggon, 2010). Combining U.S. FDA approvals for all recombinant biomedical proteins and antibodies, 15 biologics were approved from 1982 to 1991, but 54 were approved in the equivalent timeframe from 1992 to 2001 (Rader, 2009). From the early 2000s, protein drugs have featured increasingly in the pharmaceutical industry’s research and development programmes (Mullin, 2004; Wong, 2009), creating a much more favourable environment for PMP interests.
But perhaps the most difficult challenge for PMPs was that posed by the anticipated major benefit of the PMP strategy itself, i.e. the advantage of reduced manufacturing costs. Unfortunately, for a new drug being produced under patent protection by a single company, product-pricing must recover much more than the cost of its daily production (Kaiser, 2008; Masia, 2008). While solid financial data are always hard to obtain on this topic, it is generally acknowledged that drug pricing during the product’s patent life reflects, and attempts to recover, not only routine manufacturing expenses but also the investment that was made in the original research and development and in clinical trials, and also at least a portion of the company’s expenses in developing and testing other, ultimately unsuccessful, drug candidates (DiMasi et al., 2003). It seems likely that the minor degree to which manufacturing contributes to overall pricing explains at least some of the relative apathy of the pharmaceutical industry towards alternative production methods based on transgenic plants, transgenic animals or other novel systems, during the 1980s and 1990s. When a drug ‘comes off patent’ the situation potentially becomes much more favourable for PMPs, as will be discussed below.
While manufacturing costs may not be critical for a protein pharmaceutical ‘on patent’, the existence of adequate manufacturing capacity is absolutely essential. When the pharmaceutical industry began to take proteins more seriously their engagement of contract-fermentation companies quickly overwhelmed those operations, such that by 2001 many contract-manufacturing entities were quoting an unacceptable 2-year wait to new clients (Garber, 2001). This ‘protein crisis’ situation provided a potential opportunity for the alternative animal and plant platforms. For example, our own research programme on adaptation of tobacco agriculture for more efficient PM(I)P production, which had only attracted the attention of PMP companies until then, began to receive inquiries from pharmaceutical interests who were contemplating the plant-based option. However, the ‘protein crisis’ was overcome within just a few years through construction of yet more fermentation facilities quickly and economically, including some in countries outside the USA and Europe, leading some observers to question the need for transgenic plant and animal alternatives (Thiel, 2004). This facilities-expansion trend has continued subsequently, with contract cell-based manufacturing of ‘biologics’ in China and India expected to grow significantly (Van Arnum, 2009).
It is surprising that more attention was not paid to the possibility of producing pharmaceutical products for the veterinary markets (agricultural livestock and companion animals), given that their development typically entails lower regulatory-approval costs and shorter timelines compared with those for human medicines (Meeusen et al., 2007). The former Large Scale Biology Corporation’s development of a protective vaccine against feline parvovirus, produced as a coat-protein fusion to tobacco mosaic virus in tobacco plants, exemplifies commercial interest that was apparently not progressed to market (Pogue et al., 2004). Recombinant proteins are being accepted by the veterinary community today (e.g. recombinant human insulin is marketed for treatment of diabetic cats), so these applications of PMPs warrant continued consideration, especially for products whose manufacturing costs contribute significantly to competitive pricing.
The breadth of diversity of nonmedical applications for recombinant proteins in such products as enzymes, food ingredients, structural polypeptides, etc. precludes a generalized analysis, but it seems likely that factors and considerations similar to those encountered by PMPs also limited the commercialization of PMIPs. For example, food-grade enzymes that are used in the manufacturing of food products and as ingredients in formulated laundry detergents are the results of intensive, continuing and costly protein engineering research, the investment in which must be recovered through eventual sales. Thus, while the detailed economics of enzyme production are confidential, we can reasonably expect that the daily manufacturing cost has historically been a small contributor to the typically high market price of these recombinant products. This is further supported by the observation that although the major enzyme companies can, and do, compete to produce superior enzymes at attractive prices for specific applications and customers they have hitherto not embraced alternative production platforms such as plants as ways to enhance that cost-competitiveness.
The consolidation or ‘shakedown’ phase of the PM(I)P industry’s development, outlined at the beginning of this section, is quite typical in technology commercialization, and it happened during the commercialization of the transgenic agronomic traits as well, e.g. the early and then discontinued involvement of such familiar companies as Shell and Atlantic Richfield, the start-up and later acquisition of new companies such as DNA Plant Technologies, Plant Genetics, International Plant Research Institute, Plant Genetic Systems (PGS). A common reason for such consolidation is the continuing need for capital investment, especially if the initial development of technologies and products takes longer than expected and thus challenges the resources available from the initial investors. In the case of PM(I)P commercialization, the protracted and perhaps unforeseen timeframe-to-market caused by the issues discussed earlier surely contributed to ‘investor fatigue’, and thus anticipated consolidation. It is also interesting to recall that most of the independent PM(I)P companies remained (and still are) privately held, and thus did not realize the considerable injection of capital that can result from successful entry into the publicly traded stock market. Noteworthy exceptions include SemBioSys and the former Large Scale Biology. Those successful public stock offerings refute the notion that manufacturing ‘platform’ technologies are less appealing to investors than product-type technologies such as new medical drugs, but there are many reasons why companies remain privately held, such as the timing of their progress relative to market sentiment in technology stocks, success or failure with major technical benchmarks, etc.
The pioneering PM(I)P companies tested the market for plant-made proteins, exploring a wide range of product types such as high-value drugs for rare diseases, treatments for more common conditions, components to construct artificial blood plasma, enzymes for use in industrial processes. But most importantly, they also continued to advance the technology itself. As will be discussed later, their continuing innovation expanded the scope of the PMP strategy, and helped bring it closer to the standards and expectations of the pharmaceutical world, ready for the time when the economic benefit of PMP systems as a manufacturing platform would become more relevant and important. In hindsight, the continuing support and technological development of the PMP paradigm through the 1980s and 1990s, while protein-drugs came to be accepted and began to enter the marketplace, can now be seen as the pioneers’ most valuable contribution.
Technological advances that are improving the prospects for commercialization today
The original principle of using plants as a protein-manufacturing platform has been considerably extended through the subsequent development and application of a number of novel attributes, improvements, and ancillary technologies, as summarized in Table 3. Some of these developments such as enhanced expression levels, and ability to express multisubunit proteins, effectively increase the scope and appeal of the PMP approach. Others, such as customization of post-translational modifications in the plant, ability to accommodate genes for very large or very small proteins, and an emphasis on nonfood host plants, represent ways to address various limitations, concerns and criticisms that have been raised (conveniently summarized in Twyman et al., 2005). A third category of developments more directly addresses customer/client needs (Table 3).
Table 3. Examples of PM(I)P technology enhancements
|In overall system performance|
| Raise expression level of vector system||Reviewed in: Streatfield, 2007|
| Optimize host plant for vector performance||Sheludko et al.,2007|
| Optimize codon assignments of ‘foreign’ gene||Tregoning et al., 2003|
| Suppress gene silencing||Voinnet et al., 2003|
| Improve agronomic performance of host plant||Davies, 2005|
| Optimize field and indoor growing methods||McCarthy, 2005|
| Improve containment of host plant||Davies, 2005|
| Improve product stability; protein fusion||Gil et al., 2001|
| Improve product stability; tissue targeting||Lau and Sun, 2009|
|To broaden applications|
| Accommodate genes for larger proteins||Sainsbury and Lomonossoff, 2008|
| Make multigene proteins, e.g. antibodies||Ma et al., 1995|
| Make small proteins, e.g. peptides||Gil et al., 2001|
| Demonstrate accurate disulphide-formation||Stein et al., 2009|
| Enable ‘personalized’ drug products||McCormick et al., 1999|
|To address end-user specifications|
| Customize product glycosylation||Frey et al., 2009|
| Compartmentalize product for purification||Hassan et al., 2008|
| ‘Tag’ product for special purification method||Floss et al., 2010|
| Develop general bioprocessing efficiencies||Platis et al., 2008|
| Eliminate ‘adventitious presence’||Drummond, 2010|
PM(I)P application and performance enhancements
With expression levels of most PMP systems now attaining the ‘gold standard’ of 1–10% TSP and some systems substantially exceeding that enrichment (Table 1), it seems unlikely that further enhancements in this performance parameter will have a dramatic effect on prospects for successful commercialization. Nevertheless, to the extent that expression level influences the final yield of purified protein per plant, greenhouse or acre, its continued improvement (reviewed by Streatfield, 2007) may usefully contribute competitiveness within the industry, and in relation to the alternative production platforms. The demonstrated ability to express a wide range of product types, ranging from small peptides (e.g. Gil et al., 2001) to large molecules with complex quaternary structure such as secretory antibodies and collagen (Ma et al., 1995; Stein et al., 2009) has further expanded PMPs’ scope of competitive application. And optimization of host-plant germplasm and plant-production methods for responsiveness to the expression system and maximum biomass yield can also contribute to increased overall efficiency of the PM(I)P strategy (e.g. for tobacco: Davies, 2005; Sheludko et al., 2007).
A PMP ‘performance’ topic that has attracted considerable attention in relation to open-field production has been ‘containment’, i.e. the prevention of genetic and physical contamination of other, non-PM(I)P and nontransgenic crops. Genetic containment seeks to obviate any chance of transgene transfer via outcrossing to other plants, whether crops or naturalized ‘wild’ populations, while physical containment seeks to prevent accidental co-mingling of the harvested PM(I)P-plant material with other harvests. The frequent adoption and promotion of major food-plant species such as corn, rice, potatoes, alfalfa as the PM(I)P production hosts in the earlier years of commercialization was unfortunate in this regard. Although the use of fully developed agricultural crops offers many advantages this food-crop emphasis inevitably encountered the emerging opposition to ‘GM’ agriculture, to which it unwittingly added a further basis for objection, namely the imagined, unsettling prospect of contamination of the food supply with pharmaceuticals and industrial enzymes because of inadvertent co-mingling during production or harvesting. The concerns received further impetus from the much-publicized failure of PMP company ProdiGene to maintain adequate physical and genetic containment of its transgenic corn line that produced a veterinary vaccine, resulting in contamination of the next season’s (2002) non-PMP crop production on the same (and possibly adjacent) land (Fox, 2003). While this incident was intercepted long before any derived food products were affected, the resulting intensity of opposition to the use of food crops for PM(I)P applications began to include even pro-agricultural biotechnology observers (Editorial, 2004) and the food industry in general (GMA/FPA, 2007), effectively quenching future prospects for developing new PM(I)P applications in major food species like corn and soya beans.
Following those significant public objections to the perceived prospect of unintended contamination of food by pharmaceuticals and industrial proteins, the evolving PM(I)P industry has taken various steps to mitigate such possibilities (reviewed by Murphy, 2007). These actions have included selecting host-plant species that are not major food crops (e.g. SemBioSys Genetics’ emphasis on safflower, several companies’ interest in tobacco), growing the PMP crop in a region or location distant from that where the same plant is grown for food production (e.g. Ventria Bioscience’s location of transgenic rice production; Turner, 2006), exclusive use of indoor production (e.g. McCarthy, 2005; Bayer Innovation, 2008), and an emphasis on plastid-based and other gene-expression systems that reduce or obviate the chances of unintended transgene transfer to the same or related species through natural outcrossing events (e.g. Gleba et al., 2005; Grevich and Daniell, 2005). Even producers of nonfood crops may object to PMPs. For example, the traditional tobacco industry currently has a zero-tolerance standard for ‘GMO’ contamination in the crop (Dean Wallace, Council for Burley Tobacco, personal communication), and consequently we have examined the use of interspecific Nicotiana hybrids as alternatives to commercial tobacco varieties in PM(I)P applications (Davies, 2005). Such hybrids exhibit extremely low fertility, thereby reducing the chances of transgene transfer to conventional tobacco plants via pollination and of persistence of the transgenics in the environment through the setting and dispersal of seeds containing the transgene.
When transgenic plants are field-grown for PM(I)P applications, their developers will also need to incorporate other safety- or acceptance-related features which come to be expected or mandated for transgenic crops in general. The relevant technologies may have to be in-licensed by the PM(I)P company, potentially adding to the cost of development and affecting product price. As these aspects are not PM(I)P-specific they are not discussed in detail here, except to cite the drive to eliminate the use of antibiotic-resistance genes as a currently prominent example (technologies reviewed in Goldstein et al., 2005).
Governmental regulatory authorities in several countries have issued, and continue to develop, guidelines and regulations for the conduct of field trials and eventual production of PM(I)P crops (Spok, 2007; Chambers et al., 2008; Spok et al., 2008; USDA, 2008). The mandated regulatory requirements are equally applicable whether the new transgenic crops harbour products intended for medical or for industrial uses (USDA, 2008). The above-mentioned ‘containment’ technologies, together with appropriate growing and handling methods, contribute in the preparation of crop-production protocols that must be officially approved prior to any open-field planting. If there is reason to be concerned that the nonnative protein product may pose a health threat to organisms which are predatory on the plant species of choice, or which are consumers of plant-parts such as seeds or pollen, the regulatory authorities may also require an assessment or study of such ‘environmental’ risk. Such a study will take into account the potential toxicity and other properties of the pharmaceutical or industrial protein itself, and the physical and genetic containment prospects for the host plant and its cultivation. A convenient example of this process in the United States is provided by the ‘Finding of No Significant Impact’ (FONSI) issued by the USDA to PMP company Planet Biotechnology in relation to that company’s plans to field-grow tobacco plants expressing an antimicrobial antibody intended to treat dental decay (U.S. Federal Register, 2008).
It is reasonable to expect that stringent regulatory requirements will continue to exist for open-field PM(I)P deployment in the future. And this situation may not be limited to experimental plantings—the suggestion has been made that such crops may never be deregulated in the manner of today’s agronomically enhanced transgenics (Spok et al., 2008). It will, therefore, continue to benefit the industry to research and develop robust methods for both types of containment.
Optimization of PM(I)P production in relation to purpose and specifications of the product
In the opinion of this reviewer, it is the enhancements that directly address ‘fitness for purpose’, i.e. compatibility with the needs, standards, and expectations of the product’s end-user, that are most likely to influence the future commercial fortunes of the plant-made approach, especially for medical applications. Positioning the production platform as close as possible to the end-user’s requirements not only maximizes the chances of acceptance but also helps to contribute competitiveness with respect to the other protein-manufacturing systems. Currently, significant advances are being made in this category, both technologically (Table 3) and from a business perspective.
Naturally, production of authentic product, which exhibits the expected biological functionality and is not accompanied by improperly folded or assembled protein, is a basic requirement of any biomanufacturing system. Plants have proved themselves remarkably capable in this respect—for example, the accurate assembly of functional antibody molecules involves both post-translational disulphide bridging and noncovalent assembly of identical and different polypeptides, and this challenge has been met successfully for many different immunological products, including such complex molecules as pentameric secretory immunoglobulin A (Ma et al., 1995). But when biomanufacturing systems compete to make a pre-existing drug (as in the production of biosimilars or ‘follow-on’ protein drugs), the matter of product specification becomes especially critical. This is because the shortest, most economical route for regulatory approval of a pharmaceutical made by the new process will be the demonstration of product equivalence, i.e. showing that the product made in the new way is in all respects (structural, compositional, and functional) identical to the one made by the previously approved method (Woodcock, 2007; Schellekens, 2009). Authenticity of fundamental structure and biological activity may not be sufficient to achieve acceptance by the relevant pharmaceutical regulatory system; it will also be necessary to demonstrate that there are no plant-specific modifications such as covalent attachment of reactive secondary metabolites, species-specific glycosylation events etc., and to show that the molecule’s biological and pharmacological properties are sufficiently similar to those from the natural or previously approved source as to obviate any risks of unintended immunogenicity or toxicity (Woodcock, 2007; Schellekens, 2009). A recent example that illustrates a comprehensive structural and functional assessment of the plant-generated recombinant product is that of human collagen produced in tobacco plants (Stein et al., 2009). Authentic formation of collagen in a heterologous production system necessitates hydroxylation of lysyl and prolyl resides in the two different polypeptides, glycosylation of the hydroxylysyl resides with galactosyl and glucosyl substituents, formation of intrachain and interchain disulphide bonds, and assembly of the modified polypeptides into a triple helix. The PMP version of collagen was made by co-expression in the plant of the two collagen genes and two human enzymes to supplement the plant’s inherent mechanisms for post-translational modification of the polymer. Stein et al. (2009) employed a range of physical, biochemical and bioassay criteria to validate their plant-made collagen, and while their characterization does not appear to have been exhaustive, their emphasis on establishing product equivalence is both appropriate and timely.
If problematic issues such as spurious derivatization or proteolysis should occur in the plant or upon extraction of the product, the possibility theoretically exists to engineer the plant to eliminate the problem. As a hypothetical example, the activity of polyphenol oxidase, which by catalysing oxidation of phenols to reactive quinones can seriously compromise proteins during purification from some plant species, can be reduced through gene-suppression (Bachem et al., 1994). A variety of techniques for mitigating proteolysis were recently summarized (Benchabane et al., 2008), including the use of host plants with minimal protease activities against the product, subcellular targeting of the product to reduce its accessibility to proteases, ectopic expression of protease inhibitors, etc.
There is no more salient example of customization of the host-plant for generation of application-acceptable product than that of protein glycosylation. This topic is important because the carbohydrate moieties of glycoprotein drugs may contribute significantly to tolerance by the patient’s immune system, and to the proper functioning of antibodies (Wright and Morrison, 1997). Moreover, there has been some concern regarding the suitability of plants for the biomanufacturing of mammalian glycoproteins, as the glycosylation pathways in plants are sufficiently different from those in animals that mammalian-authentic glycan structures are not achieved, and the resulting ‘novel’ plant-made glycan moieties can be immunogenic (for a recent example of this issue see Jin et al., 2008). For example, after some debate on the matter it seems likely that most plants do not elaborate sialic acid, a common residue in animal glycoprotein oligosaccharides (Zeleny et al., 2006). Many other detail, but significant, differences occur in linkages and arrangements of the monosaccharide residues in plants vs. animals, as well as in overall heterogeneity of the protein-linked glycans (reviewed in Gomord et al., 2005). But, fortunately, the glycosylation sequence-motifs in protein primary structure are the same in plants and animals, as is much of the fundamental post-translational glycosylation mechanism, so that carbohydrate attachment tends to occur at the correct locations in the protein. Thus, the challenge of making a ‘humanized’ glycoprotein in a plant is principally one of adjusting the composition and structure of the oligosaccharide substituents.
Significant progress is being made towards this fascinating if ambitious metabolic engineering goal, as recently illustrated by Frey et al. (2009) and Castilho et al. (2010) and reviewed by Karg and Kallio (2009). The recent accomplishment of Castilho et al. (2010) is particularly striking; six mammalian genes were introduced into Nicotiana benthamiana to enable that plant to make and utilize sialic acid, with subsequent successful incorporation of sialate residues at the termini of the glycan moieties of a monoclonal antibody that was co-expressed with them. Meanwhile, an interesting nearer-term and less demanding alternative that may be applicable to some pharmaceutical products is to modify the glycan moieties until they are no longer immunogenic (Gomord et al., 2005) or to eliminate them altogether (Nuttall et al., 2005). It remains to be seen whether such modified molecules, or ‘humanized’ ones that still exhibit a percentage of variant structures as in the incompletely sialylated forms observed by Castilho et al. (2010), will be accepted as ‘biosimilars’ by regulatory authorities or regarded as sufficiently novel to require the same approval process as entirely new drug candidates.
Another aspect of the PMP strategy that can influence product quality and standards is postharvest bioprocessing. Given the considerable extent to which product-purification can be influenced and controlled in fermentation systems, e.g. ensuring that the product is secreted from the cells into the culture medium, engineering the cells to eliminate problem host constituents, etc., the challenge of purification from the protein complexity represented by whole-plant tissues and organs might seem to be a disadvantage, especially when one considers the very high standards of purity required of medical products. But this topic has received considerable attention (Menkhaus et al., 2004), and many strategies that facilitate efficient purification from other protein biomanufacturing systems have also been employed successfully with plants, such as those based on biological affinity between the product and molecules that can be immobilized on inert supports (e.g. protein A recognition by antibodies; Hober et al., 2007), or on expression of the product fused to an affinity-matrix-recognized protein or peptide ‘tag’ that can easily be removed after the chromatography step has been completed (e.g. Desai et al., 2002; Leelavathi and Reddy, 2003). Such sophisticated and potentially expensive methods may not be obligatory, despite the relative complexity of plant extracts; Platis et al. (2008) demonstrated that even the simplest, traditional techniques such as solvent partitioning and ion-exchange chromatography have potential in the quest to purify monoclonal antibodies from tobacco leaves in a minimum number of economical steps. And recently attention has been drawn to new strategies that employ fusion of the recombinant protein product with elastin so that purification can take advantage of the thermally triggered aggregation and precipitation of the latter (Floss et al., 2010), and with fungal hydrophobins that enable the fusion product to be separated from other proteins by solvent partitioning (Joensuu et al., 2010). These methods effectively achieve the level of specificity which is obtained with affinity chromatography but without the high cost incurred with the latter media and systems.
The PMP host and expression system may also be chosen and designed specially to facilitate downstream bioprocessing. Targeting the gene product to specific plant organs such as seeds and tubers from which purification will be easier because of a less complicated native-protein composition, and/or the absence of interfering secondary metabolites, is the most basic approach (examples listed in Horn et al., 2004; reviewed in Lau and Sun, 2009). SemBioSys’ PMP system directs the transgene product to the oil body of an oilseed plant, effecting an even closer integration with the bioprocessing method because the oil body is an organelle that offers the combined benefits of a simple native-protein composition and facile extraction from the seed (Markley et al., 2006). Similarly, Hassan et al. (2008) have explored the relationship between product localization within leaf tissue and ease of subsequent extraction. Recently, another in-cell sequestration system appears to be in development (Erabiotech, 2009). Other strategies that enable the product to be separated from the bulk of the host-plant constituents quickly and efficiently include fusion to virus particles (e.g. Smith et al., 2009), and natural secretion from roots into the surrounding medium (Borisjuk et al., 1999; Xuejun et al., 2002).
The development of economical postharvest bioprocessing also requires construction of pilot facilities which represent significant scale-up from laboratory-based demonstrations, and their use to research procedures for efficient isolation of product from substantial quantities of plant material to the required final specifications. To date, a few PM(I)P-oriented commercial ventures have made this investment, e.g. Kentucky Bioprocessing (USA), Medicago (Canada), Meristem Therapeutics (France), SemBioSys (Canada), Sigma-Aldrich Fine Chemicals (USA), Ventria Biosciences (USA). For pharmaceutical products, the bioprocessing pathway itself will have to meet government-mandated standards, and to do so in a completely reproducible and reliable manner. Establishment of the stringent quality control and quality assurance procedures that meet these requirements for plant-based production can only be achieved by working at the final production scale. Thus, an emphasis on postharvest bioprocessing is increasingly important, and the pioneering work that has been done on this topic so far is undoubtedly money well-spent.
Another topic of considerable importance in relation to postharvest bioprocessing of PMPs, and assurance of product quality, is the need to obviate ‘adventitious presence’. This term is commonly used in reference to inadvertent mixing of transgenic and nontransgenic foodstuffs, but in the PMP context it can mean the (unintended) presence in the harvest of biological or nonbiological materials that are not taken into account in the design of the bioprocessing protocol which isolates the pharmaceutical product. Insect parts, bird excrement, and pesticide drift from other crops are some of the many examples of materials which might contribute substances that could perhaps work their way through the purification sequence and thus contaminate the final protein product. Obviously, production platforms that use higher organisms face a greater challenge with this issue than do the fully contained fermentation systems. One way in which some PMP companies are addressing it is to eschew field-based production in favour of indoor growing (e.g. Bayer Innovation, 2008; Drummond, 2010; other companies via personal communication), a change that also eases the earlier-mentioned concerns about containment and public acceptance. A further advantage of indoor planting is that it enables year-round production, as opposed to the seasonal format of the field (production year-round may also be achieved outdoors in regions of the world having suitable climates, but this option will not be acceptable for production of certain nationally strategic agents such as vaccines to combat bioterrorism). Because growing plants in greenhouses may be significantly more expensive than open-field agriculture, additional research will be necessary to optimize germplasm and production methods for maximum efficiency in the enclosed environment. Economies may be realized through the use of very small plants that can be grown in a highly mechanized, automated manner, such as Biolex’s choice of the aquatic plant Lemna (McCarthy, 2005). Whatever the eventual solutions, the PMP industry today is cognizant of the issue of adventitious presence, and this is a welcome contrast to the early years when the concept of making pharmaceuticals in the open field seemed to be promoted with little or no attention to this important topic.
The continuing development of technology enhancements will become increasingly important not only to strengthen the capability and applicability of the PM(I)P approach but also from a business perspective; as the lifetimes of patents covering the fundamental PM(I)P gene-expression technologies come to an end this ‘add-on’ intellectual property will help to sustain company competitiveness and investor interest. Examination of the scientific literature describing many of these innovations will show the considerable and continuing role of universities and other nonprofit organizations in conducting the underlying inventive research. Given the under-resourced status of the PM(I)P industry compared with the large multinational companies that market the ‘GM’ mainstream crops, this public-domain research is likely to remain critical to the commercial prospects for PM(I)Ps for some time to come.
Business strategies and opportunities that will facilitate commercialization now and in the future
The aforementioned topics of postharvest bioprocessing and prevention of adventitious presence exemplify the importance and benefits of productive dialog between the plant-based biomanufacturing industry and the product end-user. But parallel thinking by the PM(I)P company and its potential clients and customers predates these procedural aspects – it begins with the very choice of product itself. Few steps taken by a commercial manufacturing platform could be as important to its future success as the targetted product and market. For PMP and PMIP companies, the factors influencing product selection are many and varied, including suitability for the gene-expression technology and plant-based biosynthesis, the extent to which cell-based manufacturing is likely to be competitive, where the medical drug stands in development and clinical trials, whether licensing is necessary to access the product, as well as the more familiar ones like anticipated sales volume, pricing, lifetime and profit potential. At first, PM(I)P companies preferred not to take on the challenge of actually developing products in-house, i.e. to vertically integrate, opting instead to access product targets through licensing, collaboration, or selection of materials from the public-domain. This reflected their origins as inventors of gene-expression technology, not originators of drugs, enzymes, etc. Among the few examples of vertical integration is Planet Biotechnology, as the company’s antibody-type PMP products designed to combat the common cold and bioterrorism agents were developed in-house. The former Large Scale Biology’s promotion of its viral-vector PMP system as an efficient new way to ‘personalize’ a therapeutic vaccine (McCormick et al., 2008) further blurred the distinction between producer and developer of the product. In personalized medicine, customization to an individual patient is part of what defines the final drug molecule, and yet it is also a function of specialized biomanufacturing technology. Thus, in this strategy, the PMP platform and the drug itself became more intimately associated than merely as production system and product. A yet stronger link would be represented by proteins which, for technical reasons, could only be made in plants. These product targets would obviously be strongly promotional of the PMP paradigm, but although we have occasionally heard PMP companies speak of such plant-only candidates, it is difficult to find well-documented examples.
For PMP companies interested to produce protein drugs that have been developed and marketed by others, the product-availability landscape is changing significantly, placing the companies in a much stronger position relative to the early years. In the 1980s and early 1990s, as discussed earlier in this review, there were only a few protein drugs (‘biologics’) under serious development by the pharmaceutical industry. Moreover, even when the larger pharmaceutical interests began to accept the potential of monoclonal antibodies and other proteins, these new molecules were the subject of their active patents, and thus they were not openly accessible for production by other means (except via contract with the product’s owner). Without this competitiveness at the manufacturing level, and with eventual pricing of the drug only minimally influenced by daily manufacturing cost, there was little need or incentive to involve any new, alternative platforms, whether based on transgenic plants or animals. In contrast, today’s PMP industry is facing a very different and more favourable situation, as many protein pharmaceuticals progressively reach the end of their patent lives (17 years, or more recently 20 years, since the 1980s/1990s) and thus become accessible to competitive manufacturing as ‘(bio)generics’ (Christopher, 2006; Gaskin, 2008). This situation not only broadens the range of products available to the PMP companies but also provides an economic incentive as manufacturing cost becomes more critical to competitive pricing.
A government-approval process will be necessary if the production platform of an existing protein drug is changed; while Europe already has a procedure for review and approval of ‘biosimilars’ (the same protein, but made as a ‘generic’ product by a different organism and/or process) (Schellekens and Moors, 2010), such an approval process is only now (2010) beginning to be established in the United States (H.R. 3590, 2010). A PMP company intending to manufacture a biosimilar drug will need to price the eventual product to recover these approval costs as well as daily manufacturing expenses (Roth and Fleischer, 2009), but unlike the original inventor of the drug it will not be attempting to recover the development costs of other, unsuccessful drugs, or the original research and development cost of this one. Overall, therefore, the economics of producing off-patent drugs for which substantial markets still exist are likely to encourage competitive biomanufacturing. Thus, biosimilars represent significant new opportunities for the PMP industry because they are open to competitive production without a license and because their manufacturing cost will present a basis for competitively differentiating the producers. Within the next few years, we should expect to see PMP companies increasingly focused on biosimilars.
Proprietary proteins whose patents are about to expire are not the only biomedical products that offer these PMP-relevant business attributes. Protein pharmaceuticals that have existed off-patent for a considerable time may be considered, if sufficient market size still exists. For example, the PMP company SemBioSys has identified recombinant insulin as such a product (Markley et al., 2006). Prophylactic vaccines, for widespread immunization against natural diseases such as influenza, and potential bioterror threats such as anthrax also fit the aforementioned criteria, as the cost of their development and testing is borne substantially by the government rather than the manufacturer, and the government seeks cost-effective production through a competitive contracting process. At the present time, there is considerable interest in developing efficient, economical methods for large-scale vaccine production as alternatives to the use of virus-infected cells or the time-honoured chicken-embryo system (Ulmer et al., 2006). Recombinant protein vaccines, including antigen-based and virus-like particle types, are recognized as viable options in this search for a new national vaccine-production system (e.g. in the United States: DARPA, 2010), and PMP companies have been quick to respond to the call for demonstration of appropriate production methods. Examples of candidate vaccine production in higher plants are plentiful and diverse (reviewed recently by Daniell et al., 2009; D’Aoust et al., 2010; Rybicki, 2010), including animal studies that show successful immunization by the purified product, e.g. protection of mice against anthrax by an antigenic protein expressed and produced in tobacco chloroplasts (Koya et al., 2005), immunization of mink by a plant virus (cowpea mosaic) engineered to present an epitope of mink enteritis virus and produced in bean plants (Dalsgaard et al., 1997), protection of mice against West Nile virus utilizing a monoclonal antibody produced in Nicotiana benthamiana (Lai et al., 2010). The PMP industry is thus well equipped to respond to the current interest in alternative production platforms for prophylactic vaccines, and there is evidence of progression of this potential out of the laboratory and into pilot-scale production (Drummond, 2010).
While the increasing availability of off-patent and vaccine product targets appeals directly to the PMP strategy’s identity as a protein biomanufacturing platform, it should also be remembered that there will be competition from many alternative systems for these applications, including continuously improved cell-culture (fungal, plant, insect, mammalian) platforms, and the transgenic animal methods, which have experienced very much the same business-development history as PMPs. Indeed, a recombinant protein derived from the milk of transgenic goats has gained regulatory approval in both Europe and the United States (Kling, 2009). The PMP companies will need to demonstrate competitive advantage to capture a significant share of the new markets, and these advantages may include not only economy of production but also such distinctive benefits as the absence of animal-origin contaminants, faster and cheaper scale-up for very large-scale production, etc.
With respect to new market potential for PMIPs, the existence today of proven markets for high-grade enzymes at one extreme (reviewed in Kirk et al., 2002) and the sales of crude natural preparations that are claimed to be enzyme-based at the other (e.g. products for pesticide-spill detoxification, conditioners for ornamental lakes, etc.) raises the question of whether plants might usefully be employed in the economical production of intermediate-grade, semi-pure enzymes, perhaps for such applications as enzyme-based manufacturing of chemicals (so-called ‘green chemistry’), paper and other materials. Of course, the economics of plant-based production of the recombinant version might have to compete favourably with extraction of the native molecule from scaled-up production of the natural source (micro)organism. But the latter may not always be feasible. Thus far the emergence of significant business opportunities for economical recombinant enzymes has probably been stymied by the lack of their availability, and investment in their production via the PMIP route has presumably been hindered by the lack of markets. A compelling, urgent requirement for production of a customized enzyme economically and in large amounts could provide an escape from this paradox, and at the present time it is exciting to imagine that the enzymes utilized for processing of plant biomass into bioethanol and other biofuels may represent such a need. This enzyme application is very cost-sensitive (Geiver, 2010), and the required volumes are expected to be large - a rough estimate based on the U.S. annual consumption of automotive gasoline suggests that in the production of biomass-derived ethanol to replace just 1% of oil-derived fuel, the replenishment of just 1 mg of the enzyme catalyst used to make each gallon of ethanol would require more than 200 tons of enzyme to be supplied annually.
Meanwhile, it is pleasing to note that PMIPs are beginning to serve the markets for additives that enhance the performance of commercial mammalian-cell bioreactors, e.g. Kentucky Bioprocessing’s tobacco-based production of industrial-grade aprotinin (KBP, 2010) mentioned earlier, and Ventria Biosciences’s launch of the company InVitria to provide albumin, lactoferrin, lysozyme and other products for this application via expression in rice (InVitria, 2010).
It is possible to characterize the history of commercial PM(I)P endeavours as four distinct phases—initial invention and company start-up, followed by proof-of-concept and growth, then an industry consolidation, and recently a second commercialization round. Although the first ‘boom and bust’ cycle of the 1980s and 1990s may retrospectively appear like a failure, in actuality the technology was ahead of its time, as the pharmaceutical industry was slow in embracing proteins as the basis of new drugs, and as long as those protein drugs were still under patent protection the cost of protein manufacturing was inevitably an insignificant part of pricing. But, meanwhile, much progress was made in advancing the plant gene-expression systems, and in commencing work on very necessary complementary aspects such as choice of crops, containment, more efficient field and greenhouse production of plants and ‘humanization’ of the plant-made protein drug. The pioneering companies of the early years contributed substantially to the visibility and technological maturation of the PM(I)P concept, without which today’s ‘plant-made’ industry would be much less competitive.
While the subsequent industry ‘shakedown’ also coincided unfortunately with a tightening of available investment capital, it also ushered in new commercial players and stimulated more realistic and relevant business strategy. The more mature PMP industry began thinking of shifting to nonfood crops or to indoor production, and emphasizing products for which manufacturing costs are a more significant contributor to competitiveness and pricing (e.g. off-patent generics or ‘biosimilars’ like insulin and aprotinin, and government-underwritten vaccines). This evolution in the PMP industry has helped build confidence in its capability and potential, as evidenced in the recent investment by major life-sciences companies, i.e. Bayer Innovation’s acquisition of the German PMP company Icon Genetics, and Philip Morris International’s substantial investment in the Canadian PMP company Medicago. (Pfizer’s investment in plant cell culture–based Israeli company Protalix has also been noted, although some analysts speculate that the drug molecule is perhaps of more interest than the production platform in this instance; Ratner, 2010).
Progress with clinical trials is also being watched closely, as industry observers have often suggested that one whole-plant-made drug successfully brought to market will trigger much new interest in PMPs. Although plant-made proteins have been the subject of human trials in the past (e.g. McCormick et al., 2008), only one attained regulatory approval thus far, namely Planet Biotechnology’s antibody-based treatment for dental decay that was authorized for sale as a ‘medical device’ in Europe but has yet to be marketed (personal communication). Current clinical trials include Biolex’s Lemna-produced therapeutic for hepatitis C, now preparing to enter Phase III in the United States, and Bayer Innovations’ anti-idiotypic personalized antibody treatment for nonHodgkin’s lymphoma (Baynews, 2010).
The major push in the United States and Europe for large-volume, more efficient production of vaccines bodes especially well for today’s PMP industry. It is gratifying to see the PMP companies engaging this opportunity energetically. And while it is early yet to be confident about the pace of future development of plant-made industrial products, the oil-price ‘driver’ of new markets for sustainable, renewable, alternatives to petrochemicals must surely create a significant opportunity in the near future, such as in the economical biomanufacturing of the enzymes required for production of biofuels from cellulosic biomass.
Footnote about the author’s involvement in the PM(I)P field
The Kentucky Tobacco Research and Development Center is state-mandated to facilitate PM(I)P-based developments for agriculture in Kentucky, particularly (but not exclusively) in regard to the opportunities which they represent for tobacco farmers who are transitioning from the traditional crop to new commercial endeavours. Since the mid-1990s, this programme has engaged in collaborative research with many companies whose activities have emphasized tobacco-based PMPs, thus providing its researchers with a unique opportunity to observe and document the continuing development of that industry and of associated, relevant technologies.
The author appreciates the assistance provided by numerous acquaintances in academia and industry who have provided information, opinion and comment during preparation of this article. The constructive suggestions supplied by the anonymous reviewers were very helpful and are acknowledged, as is permission from Dr. Denis Murphy (University of Glamorgan, UK) for use of his earlier listing of companies as the basis for Table S1. Financial support was provided by the Kentucky Tobacco Research Board.