Production of renewable polymers from crop plants

Authors


*(fax +41 21 692 4195; e-mail yves.poirier@unil.ch).

Summary

Plants produce a range of biopolymers for purposes such as maintenance of structural integrity, carbon storage, and defense against pathogens and desiccation. Several of these natural polymers are used by humans as food and materials, and increasingly as an energy carrier. In this review, we focus on plant biopolymers that are used as materials in bulk applications, such as plastics and elastomers, in the context of depleting resources and climate change, and consider technical and scientific bottlenecks in the production of novel or improved materials in transgenic or alternative crop plants. The biopolymers discussed are natural rubber and several polymers that are not naturally produced in plants, such as polyhydroxyalkanoates, fibrous proteins and poly-amino acids. In addition, monomers or precursors for the chemical synthesis of biopolymers, such as 4-hydroxybenzoate, itaconic acid, fructose and sorbitol, are discussed briefly.

Introduction

Climate change and resource constraints, particularly for fossil fuels, have given renewed impetus to the development of plants for the sustainable synthesis of a large spectrum of chemicals and materials required by mankind. While the current focus is on using plants for biofuels, such as bioethanol and biodiesel, plants are a potential source of a much wider range of useful chemicals and biomaterials. Biopolymers currently occupy a very small but growing share of the polymer market. Some of these materials have unique properties that make them superior to synthetic polymers, particularly in medicine, where biocompatibility and biodegradability are strong assets. However, biomaterials often lack desired qualities such as the durability, strength and low price required for their use in large-scale low-value consumer products. Thus, with the notable exception of natural rubber, cellulose and suberin (as found in cork), only a small fraction of the potential material market is presently covered by biomaterials. To improve the competitiveness of biomaterials, new biopolymers and production technologies are required.

Although several novel polymers have been produced successfully in plants, including polyhydroxyalkanoates (PHAs) and cyanophycin (Nawrath and Poirier, 2007; Neumann et al., 2005), the production of polymers in transgenic plants remains at a relatively early stage of development. While feasibility studies have often been successful in demonstrating significant accumulation of a specific biopolymer in model plants, attaining high production in crop plants without negatively affecting crop yield and properties, and devising cost-effective processing methods, remain very challenging tasks.

The first section of this paper discusses the general considerations that should guide the choice of production strategy, biopolymer and crop. Three groups of polymers are then discussed in more detail, namely protein-based biomaterials, PHAs and natural rubber. The final section discusses the synthesis of monomers and building blocks useful for polymer synthesis. Not included in this review are biopolymers such as starch and cellulose, which are covered elsewhere (Pauly and Keegstra, 2008; Smith, 2008), as well as a wide range of minor biopolymers that have applications in food, cosmetics and medicine.

Production strategies

In general, there are three basic production strategies for renewable biomaterials (Figure 1). The traditional option is the direct use of biomass (wood, straw, cork) or biomass components (fibers, natural rubber, starch, cellulose, sugars, oils). A second option is the conversion of these resources to new compounds by industrial biotechnology involving fermentation (also named white biotechnology) or by chemical methods. For example, PHA, biocellulose, xanthan, silk and polythioesters can be produced by recombinant or wild-type micro-organisms in fermentation processes using carbohydrates from plants (Thakor et al., 2005), while polyesters such as polylactic acid (PLA), polycaprolactone, polytrimethylene terephthalate (e.g. Sorona® by Dupont) and polybutylene succinate (e.g. Bionolle® by Showa), are produced using chemical polymerization of substrates that are, at least in part, renewable and generated by fermentation (Mecking, 2004). The third option, which is the focus of this review, is the production of biopolymers or polymer building blocks in transgenic plants. As white or green biotechnology compete in the market for production of biopolymers, the advantages and disadvantages of these two technologies must be compared.

Figure 1.

 Production cycles.
Route A: direct production of materials from plants. Route B: conversion of plant products to polymers by industrial biotechnology or chemical methods. Route C: production of new chemical building blocks or polymers in transgenic plants. In all cases, the cycle is closed by composting, incineration or biodegradation.

Lifecycle assessments have indicated that biopolymers produced in plants may offer considerable savings in terms of energy use and greenhouse gas emissions over petrochemical equivalents and biopolymers produced from white biotechnology if the product concentration in the plant is high enough and the remaining biomass available after extraction of all other potentially valuable components is used to generate energy (Kurdikar et al., 2001). Furthermore, agricultural crops offer the possibility of producing biopolymers on a larger scale than is possible by microbial biotechnology, which is an important factor for biopolymers used in commodity products. However, valid comparisons between biopolymers produced in bacteria and transgenic plants are difficult because transgenic plants created for biopolymer or platform chemical production have not yet reached the commercial stage. If it is assumed that product concentrations in plants approach the same level as in bacteria, and processing is not appreciably more difficult, direct production in plants obviously involves fewer steps and should be cost-effective compared to industrial biotechnology. However, so far, current experience in synthesis of PHA and cyanophycin indicates that production levels are lower, making downstream processing more difficult (Nawrath and Poirier, 2007; Neumann et al., 2005).

The level of control over key factors affecting biopolymer properties (molecular weight, polydispersity and monomer composition) is likely to be lower in transgenic plants compared to bacterial fermentations or chemical synthesis, which involve closed systems with extensive control over the production conditions. In contrast, plant metabolism can be difficult to manipulate without affecting plant health and yield, while external environmental conditions can change fluxes through metabolic pathways. The synthesis of polymers in plants also raises the issue of seasonal harvesting and processing, and consequently the stability of the polymer in tissues between harvest and extraction. Fermentation and chemical synthesis are more flexible, with continuous production throughout the year or peaks of synthesis depending on demand being possible. A partial remedy is the production of polymers in plant organs that can be easily stored over a long time period and in which polymers are likely to be stable, such as in dry seeds. Alternatively, crops that can be harvested throughout the year, such as sugar cane, may be used. However, some plant biopolymers, such as 1,4-cis-polyisoprene (natural rubber), cellulose and starch, are successful materials that cannot be produced cost-effectively by other means or with the same level of control over regio- and stereochemistry.

The fact that the in planta production of chemicals and biopolymers requires GM technology is also of some significance. Although the technical possibilities of transgenic plants have been amply demonstrated, relatively few GM crops have reached the field (Wenzel, 2006). The process of developing a transgenic crop is more complicated and costly than is generally realized, in part because of regulatory issues and the lack of public acceptance (Figure 2; Devine, 2005). Toxicity testing of a transgenic plant alone can cost in excess of US$500 000. Regulatory clearance, clearance for import and variety registration are all difficult steps, and may have to be repeated from country to country. For a single country, the costs may be in the range of US$2–3 million, while global registration may cost upwards of US$5 million. The overall costs of discovery, research and development, breeding, production, admission and other regulatory matters may run into several tens of millions of US dollars (Devine, 2005). Moreover, the development of new registered crop varieties may take 8–12 years, making it very difficult to respond to short-term market demands using GM crops (Wenzel, 2006). Thus, only a few novel products or characteristics can be developed to marketable crop varieties, which obviously should bring a high return on investment, either through the high value of the product or because of its high market volume.

Figure 2.

 Approximate time lines for the development of a new transgenic crop, from gene discovery to commercial seed sales.
The time lines may vary depending on speed of success and regulatory requirements (adapted from Devine, 2005, reproduced with permission).

Which biopolymers should be produced in which transgenic crops?

Considering the relative strengths and weaknesses of plant versus white biotechnology, good targets for biopolymer production in plants are polymers that can either be produced more efficiently in plants, have better properties when synthesized in plants, or are needed on a large-scale at low cost. Currently, only few biopolymers can compete with their petrochemical equivalents based on price and performance. Natural rubber is unique in this respect as this polymer cannot be substituted by synthetic equivalents for many of its applications. Biopolymers that could compete with petrochemical plastics used in consumer products (e.g. polyethylene and polypropylene) would also be good targets for high-volume low-value biopolymers (or biopolymer building blocks) produced in plants. The bioplastic market is currently very small, estimated at 50 000 tonnes year−1 in the EU or near 0.1% of the volume of petroleum-based plastics (40 million tonnes year−1 in the EU; http://www.european-bioplastics.org). Assuming a future target of 10% of bioplastics based on renewable resources, production of 4 million tonnes of bioplastics per year would be required, and such an amount could be met by synthesis in crop plants. For example, synthesis of a biopolymer at a level of 10% dry weight in leaves of sugar beet would yield 0.5 tonnes of biopolymer ha−1. With 2 million hectares of sugar beet currently grown in the EU, this single crop could yield 1 million tonnes of biopolymers year−1 in addition to 24 million tonnes of sugar year−1. PHA and PLA are good examples of biopolymers that could capture a substantial part of the bioplastic market. PLA is produced by chemical condensation of lactic acid obtained by fermentation, and is marketed by Natureworks, while bacterial PHA is expected to reach the market soon as Mirel®, produced by Metabolix and Archers Daniels Midland. Substantial advances have been accomplished with regard to PHA synthesis in plants, making its synthesis in crops a realistic goal (see below).

Another important criterion for the success of biopolymer synthesis in plants relates to the choice of crop. Although the use of transgenic plants is well accepted in North America, this is not true for other regions of the world, particularly Europe. A way forward for production of biomaterials may be the choice of industrial crops or non-food crops as production platforms, preferably those lacking close relatives in the cultivation area. This clearly reduces the risks (perceived or real) to consumers and the food chain. Industrially processed crops include sugar beet, sugar cane, certain potato varieties and fiber crops. However, some of these crops belong to the same species as food varieties, and the occurrence of mix-ups, out-crossing and/or weedy variants may be inevitable. Miscanthus and switchgrass are being developed as biomass crops largely for fuel and energy. In the context of a biorefinery, they could also be used for biopolymer synthesis. Tobacco has also interesting advantages as a production platform, because it is a non-food crop that is native to South America, and lacks close relatives in Europe and North America. It does not persist in Northern climates, and seed production can be prevented by modern genetic techniques (cytoplasmic male sterility), but also by regular harvesting of biomass before flowering. Furthermore, plastid transformation is very well established for tobacco, unlike most other crop plants. For propagation, seeds can easily be produced in greenhouses, as tobacco is a prolific seed producer. Yields of leaf biomass can be increased by using different agronomic practices compared to growing tobacco for smoking, and yields of up to 14 tonnes ha−1 have been reported, which compares reasonably well with sugar beet at 18 tonnes ha−1 (average yield in Europe) but is lower than sugar cane or Miscanthus at 20–25 and 20–30 tonnes ha−1, respectively (van Beilen et al., 2007).

Proteins

Proteins are unique materials in that the exact composition and order of the 20 natural monomers (i.e. the amino acids) can be controlled at will. This implies that the properties of the resulting material can be influenced and to some degree predicted. With further progress in predicting structure–function relationships, materials with the desired properties can be designed and produced in appropriate hosts. The potential applications of such materials range from tissue engineering, drug carriers, hydrogels, coatings and glues to elastomers and fibers. Important target proteins are the so-called fibrous proteins, non-ribosomally produced poly-amino acids such as cyanophycin, and plant proteins obtained as co-products of starch, vegetable oil or biofuels.

Fibrous proteins

Fibrous proteins typically contain short blocks of repeated amino acids and can be regarded as elaborate block co-polymers with unique strength-to-weight, elastic or adhesive properties (Huang et al., 2007; Sanford and Kumar, 2005; Scheibel, 2005). Well-known fibrous proteins are silk, collagen, elastin, mussel adhesive proteins, keratin, wheat glutenin and resilin (Kiick, 2007). By combining repeat sequences of the various natural fibrous proteins or even completely synthetic sequences, and changing the linker elements between the repeat sequences, a tremendous combinatorial range is available (Holland et al., 2007; Nagapudi et al., 2005). Obviously, this could also include sequences optimized for production in plants.

Significant effort has gone into the heterologous production of spider silks in micro-organisms, cell cultures, animals and plants. Silks represent a broad class of polymers that can be loosely defined as externally spun fibrous protein secretions. They are mainly produced by a variety of insects and spiders, and interest in the development of novel silk-based fibers has mainly focused on the silks produced by the golden orb-weaving spider Nephila clavipes. This spider synthesizes several kinds of silks for different purposes, such as the construction of webs, the weaving of cocoons, or as a dragline (Hinman et al., 2000; Vollrath and Knight, 2001). These various spider silks have distinct properties, ranging from Lycra®-like elastic fibers to Kevlar®-like superfibers. Of particular interest are the silks of the dragline (the main structural web silk and the spider’s lifeline) because of their exceptional mechanical properties. Dragline silks are stronger than steel, when compared on a weight basis, and have similar strength but are more elastic than Kevlar® (Hinman et al., 2000). Dragline spider silk sequences are composed of repeated sequence blocks of various types (Huang et al., 2007). The GPGXX (often GPGQQ) motif is thought to form a β-turn spiral, while the GGX motif probably forms a 310 helix. These two hydrophilic domains are thought to confer reversible extensibility to the silk. In contrast, the (GA)n and poly(A) repeats form hydrophobic crystalline domains that are responsible for the high tensile strength of the fiber. Spacers of variable lengths connect these different motifs. Furthermore, the dragline spider silk proteins have blocks of mostly hydrophobic non-repetitive amino acids at the N- and C-termini (Ayoub et al., 2007). The C-terminus is thought to be essential for protein aggregation, an essential step in the spinning of a fiber.

Effective high-level and stable expression of silk-like proteins in various organisms has been hindered by problems such as gene size limitations, clone instability due to homologous recombination of repetitive sequences, the formation of inclusion bodies, and distinct codon usage. Nevertheless, synthetic silk genes have been successfully expressed in transgenic tobacco, potato and Arabidopsis thaliana, and transgenic plants have been cultivated in greenhouses and in field trials (Menassa et al., 2004; Scheller and Conrad, 2005). In the case of tobacco and potato leaves, targeting to the endoplasmic reticulum yielded 2% of total soluble protein (TSP; Scheller et al., 2001). Expression in leaf apoplasts of A. thaliana yielded 8.5% TSP, while targeting to seed endoplasmic reticulum yielded 18% TSP (Yang et al., 2005). Although determination of the level of accumulation of such proteins with an unusual amino acid composition is difficult, and is often based on semi-quantitative methods, the expression levels of silk proteins reported in plants are close to the level of 10% and 30% TSP reported in Escherichia coli and Pichia pastoris, respectively (Fahnestock and Bedzyk, 1997; Fahnestock and Irwin, 1997). Synthesis of recombinant silk protein has also been accomplished by secretion from cultured mammalian cells (Lazaris et al., 2002). Expression of collagen, of a synthetic protein made from repeats of a motif found in elastin, and of a chimeric protein made of silk and elastin domains, has also been reported in tobacco or potato (Guda et al., 2000; Ruggiero et al., 2000; Scheller et al., 2004).

Improvements in fibrous protein synthesis in plants may require several approaches, including optimization of the amino acid and tRNA pools for the amino acids that are over-represented in these proteins (such as glycine and alanine in spider silk), co-expression of several fibrous proteins as found in natural silk, and improved targeting to subcellular compartments and tissues that are optimal for protein synthesis and storage. Information on how to increase the pool of essential amino acids, such as lysine and threonine, as well as on targeting pharmaceutical proteins to specific subcellular compartments, could be used to increase the amount of silk protein produced in plants (Conrad and Fiedler, 1998; Galili et al., 2005). In addition to strategies to optimize expression, a remaining challenge is that effective synthesis of a fibrous protein does not ensure production of a good fiber. The properties of silk fibers depend to a large extent on correct assembly of the different types of proteins by spinning. In the spider or silkworm glands, silk proteins are maintained at a concentration of 30% without precipitation or aggregation, and are spun together via an elaborate liquid spinning process, which has not been reproduced in the laboratory (Huang et al., 2007). Recombinant spider silk obtained from mammalian cells has been spun into filaments that show similar toughness to dragline silk but with a lower tenacity (Lazaris et al., 2002). Thus, while processing of fibrous proteins is facilitated by the fact that proteins such as spider fibroins are extremely heat-stable and acid-soluble (Scheller et al., 2001, 2004), advances in micro-spinning and other processing technologies are essential to produce good fibers from fibrous proteins.

Assuming that expression and processing limitations can be resolved, heterologous expression in plants would enable production on a much larger scale than is possible in bioreactors. However, it must be determined whether any of the fibrous proteins has a (potential) market size that would justify the significant cost and time required for the development of a transgenic crop germplasm. One of the most appealing aspects of the fibrous proteins is the ability to specify properties through the DNA template. This allows tailoring of the protein to specific applications and processing methods, but at the same time clearly favors production in more flexible organisms, such as bacteria or yeast, in which the genetic engineering, downstream processing, and time from strain optimization to industrial use are easier and quicker. Such materials are likely to be used in high-value engineering or medical applications, where fermentation costs are less relevant than the material properties. Thus, plants are better hosts only if the fibrous protein is to be used at a commodity-scale, for example as a technical or textile fiber. For this scenario, production costs have been estimated at 10–50% of the cost for production in bioreactors (Scheller et al., 2001).

Non-ribosomally produced poly-amino acids

Cyanophycin is a nitrogen-storage polymer that is synthesized by a range of cyanobacteria. It comprises a polyaspartate backbone with arginine side chains attached via their α-amino group to the β-carboxy group of each aspartate (Figure 3). It is synthesized by cyanophycin synthetase without the involvement of ribosomes, and deposited as granules (Berg et al., 2000; Oppermann-Sanio and Steinbüchel, 2002). Although cyanophycin is not a polymer with good material properties, it is of interest mainly as a source of polyaspartate, which could replace the chemically synthesized material used as a super-adsorbant or anti-scalant (Oppermann-Sanio and Steinbüchel, 2002). The potential market for polyaspartate could be as high as US$450 million year−1, contingent on a cheap source of l-aspartic acid (Tsao et al., 1999).

Figure 3.

 Structure of cyanophycin, and potential products derived from cyanophycin.

Alternatively, the constituent aspartate and arginine from cyanophycin could serve as a starting point for the synthesis of a range of chemicals (Figure 3). In fact, aspartate was identified as one of 12 ‘top added value chemicals from biomass’ (Werpy, 2004), and a range of possible derivatives was suggested that can be obtained by reductions, dehydrations, polymerization, decarboxylation and deamination reactions (Scott et al., 2007; Werpy, 2004). For example, reduction of the carboxylic acids of aspartic acid would produce 3-aminotetrahydrofuran and 2-amino-1,4-butanediol, analogs of high-volume chemicals used in the polymer industry. Arginine appears to be less versatile, based on the number of reactions described, but could be converted to 1,4-butanediamine, which can be used for the synthesis of nylon-4,6. The critical question remains whether cyanophycin production levels in plants can reach economically viable levels. Transgenic tobacco and potato have been created that contain cyanophycin up to 1.1% dry weight in leaves through expression of cyanophycin synthase in the cytoplasm (Neumann et al., 2005). Some deleterious effects of cyanophycin accumulation in these transgenic plants were observed, such as changes in leaf morphology and decreased growth. However, a recent study revealed that translocation of cyanophycin synthesis from the cytoplasm to plastids led to an increase in cyanophycin content up to 6.8% of dry weight in tobacco leaves without visible adverse effects to the plants (Hühns et al., 2008). With regard to competing technology, recombinant E. coli has already been developed to produce cyanophycin up to 29% of the cell dry weight on protamylasse, a waste product of starch production from potato (Elbahloul et al., 2005). As with fibrous proteins, improvement in cyanophycin synthesis in plants may require optimization of the pathways involved in supplying aspartic acid and arginine, as well as engineering the cyanophycin synthase for maximal activity in the plant cell environment.

Other poly-amino acids, such as polylysine (Shih et al., 2006) and polyglutamate, (Buescher and Margaritis, 2007) have not been studied in as much detail. They are currently used in food, but have many potential non-food applications ranging from hydrogels, biochip coatings and drug carriers to cryoprotectants. As the genes encoding their synthesis machinery have been cloned, these are available for expression in plants. Glutamate is also one of the 12 ‘top added value chemicals from biomass’ with potential as a novel building block for five carbon polymers (Werpy, 2004), while lysine could be converted to ε-caprolactam, the building block for nylon (Scott et al., 2007).

Protein co-products of biofuels

The amount of protein co-products from future large-scale biofuel production could greatly exceed the amount that can be absorbed by the food and feed markets, possibly enabling the development of a protein-based bioplastics industry (Sanders et al., 2007). Assuming that 10% of transportation fuels will be bio-based, up to 100 million tonnes of proteins will be produced (Scott et al., 2007). Although a proportion of these proteins may be absorbed as feed, because the acreage of protein crops such as soybeans would diminish in response to a protein glut, a substantial amount of proteins would be available for other uses.

Proteins derived from starch or oil crops have been converted to materials in the past. Plastics can be produced from zein (the major corn protein; Lawton, 2002; Shukla and Cheryan, 2001), soy protein (Mohanty et al., 2005) and wheat gluten (Pallos et al., 2006; Woerdeman et al., 2004) by cross-linking with glutaraldehyde, formaldehyde or other chemicals. In 1950, about 2700 tonnes year−1 of zein plastic, a glossy, scuff-proof, grease-proof material for coatings, was produced along with 2200 tonnes year−1 of Vicara®, a fiber that is also based on zein. If produced on the same scale as in the 1950s, and as a by-product of ethanol production, zein would cost about €2.50 kg−1 (Lawton, 2002) – the actual cost of zein is currently 10 times higher. In the 1930s, Henry T. Ford used soy protein as a source of bioplastics to construct car parts. However, petroleum-based plastics soon replaced protein-based materials, partly because of poor properties of these protein fibers, including susceptibility to microbial degradation and water permeability (Mohanty et al., 2005), and because of their price. Because these proteins are likely to be derived mainly from food crops, and are side-products, it is not likely that the properties of these proteins will be optimized for material applications by genetic engineering. Preparation of protein co-products has been discussed for leafy crops (lucerne, grasses), oilseed crop (rape) and a starch crop (potato; Dijkstra et al., 2003).

Polyhydroxyalkanoates

Polyhydroxyalkanoates (PHAs) are biological polyesters that are produced by a wide variety of bacteria as osmotically inert carbon and energy storage compounds that accumulate in the form of granules. Almost all naturally occurring PHAs consists of 3-hydroxy fatty acids with a chain length of 4–16 carbons polymerized by a PHA synthase using R-3-hydroxyacyl-CoA intermediates. Under special circumstances or using specific feed compounds, bacterial PHAs may also contain 4-, 5- and 6-hydroxy fatty acids or monomers substituted with various groups such as aromatic rings, methyl or hydroxy groups, and ether or double bonds (Steinbüchel and Valentin, 1995). However, except for PHAs containing partially unsaturated monomers, these have not been produced in plants, and will not be discussed in this review.

Depending on the composition and resulting properties, PHAs have many and wide-ranging potential applications. Due to their impermeability to water and air, PHAs are considered very suitable for consumer products such as bottles, films and fibers. The simplest PHA, poly-3-hydroxybutyrate (PHB), is a relatively hard and brittle material with a melting point that is slightly below the thermal decomposition temperature (Lenz and Marchessault, 2005). While the use of C5 monomers to produce poly(3-hydroxybutyrate-co-3-hydroxyvalerate) [P(HB-co-HV)] improves the properties of the material to some extent, inclusion of a small proproportion of longer monomers (C6 and longer) has resulted in materials that are much easier to process, and are similar to polypropylene (Noda et al., 2005). PHAs consisting of higher-molecular-weight monomers (C6–C16, referred to as medium-chain-length PHA; mclPHA) are typically rubber-like materials with an amorphous soft/sticky consistency. These polymers still have to find large-scale applications, which may include the production of enantiopure R-3-hydroxy-carboxylic acids (Witholt and Kessler, 1999). The range of properties and applications may also be greatly extended by chemical modification of the polymer after extraction (Hazer and Steinbuchel, 2007). In general, PHAs are considered for products that end up in the environment, such as flowerpots used in planting, foils, bags, fishing lines and nets, which should decompose if lost, material used in biomedical applications, and for other disposable items such as bottles, cups, plates and cutlery that can be composted but not recycled (for detailed reviews on potential applications, see Philip et al., 2007; Van der Walle et al., 2001).

Thus, PHAs are a possible large-scale commodity. As starch, sugars and oils produced in plants are cheaper than typical petrochemicals, PHA production in plants has been proposed as an economically viable option (Poirier et al., 1995). This approach was first successfully demonstrated in 1992 for the synthesis of PHB from acetyl-CoA using transgenic Arabidopsis expressing acetoacetyl-CoA reductase and PHB synthase from Ralstonia eutropha in the cytoplasm (Poirier et al., 1992; β-ketothiolase, the first enzyme of the PHB pathway, is present in the cytosol of plants; Figure 4). The transgenic A. thaliana plants produced 0.1% PHB at most in the cytosol, nucleus or vacuoles, and showed strong growth retardation and reduced seed production, indicating that significant further work is required to improve production levels and the health of the transgenic plants. PHB synthesis in the cytosol of other plants, including tobacco, showed similar low accumulation (Table 1; Nawrath and Poirier, 2007).

Figure 4.

 Metabolic pathways leading to the production of PHB, P(HB-co-HV) and mclPHA in plants.
The pathway for PHB synthesis from acetyl-CoA (top left) has been implemented in the cytosol, plastid and peroxisome. The pathway for P(HB-co-HV) synthesis in the plastid was created by combining the PHB biosynthesis pathway with a pathway generating 3-propionyl-CoA via a threonine deaminase and pyruvate decarboxylase (top right). Synthesis of P(HB-co-HV) in the cytosol involves an unidentified source of propionyl-CoA or 3-hydroxyvaleryl-CoA. Other PHA co-polymers, such as mclPHA, have been mainly synthesized in the peroxisome using the 3-hydroxyacyl-CoA intermediates generated by the β-oxidation cycle (bottom left). Synthesis of mclPHA from the conversion of R-3-hydroxyacyl-ACP to R-3-hydroxyacyl-CoA is achieved via a bacterial 3-hydroxyacyl-CoA–ACP transacylase (bottom right).

Table 1.   Polyhydroxyalkanoates produced in transgenic plants
Plant speciesSubcellular compartmentTissuePHA producedPHA yield (% dry weight)Reference
  1. PHA, polyhydroxyalkanoate; P(3HB), poly-3-hydroxybutyrate. P(3HB-co-HV), poly(3-hydroxybutyrate-co-3-hydroxyvalerate); mclPHA, medium-chain-length PHA; Scl-mclPHA, short-chain-length to medium-chain-length PHA.

Arabidopsis thaliana PlastidShootP(3HB)14–40Bohmert et al. (2000), Nawrath et al. (1994)
PlastidShootP(3HB-3HV)1.6Slater et al. (1999), Valentin et al. (1999)
CytoplasmShootP(3HB)0.1Poirier et al. (1992)
CytoplasmWhole plantP(3HB-co-3HV)0.6Matsumoto et al. (2005)
PeroxisomeWhole plantmclPHA0.4Mittendorf et al. (1998)
PeroxisomeWhole plantscl-mclPHA0.04Arai et al. (2002), Matsumoto et al. (2006b)
AlfalfaPlastidShootP(3HB)0.2Saruul et al. (2002)
CornPlastidShootP(3HB)6Poirier and Gruys (2001)
PeroxisomeCell suspensionP(3HB)2Hahn et al. (1999)
CottonCytoplasmVascular bundlesP(3HB)0.3John and Keller (1996)
PlastidVascular bundlesP(3HB)0.05John and Keller (1996)
PotatoPlastidShootP(3HB)0.02Bohmert et al. (2002)
CytoplasmCell linemclPHA1Romano et al. (2003)
PlastidShootmclPHA0.03Romano et al. (2005)
Oilseed rapeCytoplasmShootP(3HB)0.1Poirier and Gruys (2001)
PlastidSeedP(3HB)8Houmiel et al. (1999)
PlastidSeedP(3HB-co-3HV)2.3Slater et al. (1999)
TobaccoCytoplasmShootP(3HB)0.01Nakashita et al. (1999)
PlastidShootP(3HB)<1.7Arai et al. (2001), Bohmert et al. (2002), Lössl et al. (2003)
PlastidLeavesmclPHA0.005Wang et al. (2005)
Sugar beetPlastidHairy rootsP(3HB)5Menzel et al. (2003)
Sugar canePlastidsLeavesP(3HB)1.88Petrasovits et al. (2007)
FlaxPlastidsStemP(3HB)0.005Wrobel et al. (2004), Wrobel-Kwiatkowska et al. (2007)

One explanation for the deleterious effects of PHB production in the cytosol is the diversion of acetyl-CoA and acetoacetyl-CoA away from the endogenous flavonoid and isoprenoid pathways. These pathways are responsible for the synthesis of a range of plant hormones and sterols, and minor changes could potentially have strong effects on plant growth. Interestingly, several of the phenotypes observed in plants downregulated for the cytosolic ATP-citrate lyase, the main enzyme responsible for acetyl-CoA synthesis in the cytosol, resemble the phenotypes observed in transgenic plants producing PHB in the cytoplasm (Fatland et al., 2005). As the plastids maintain a high metabolic flow of acetyl-CoA for use in fatty acid synthesis, and are able to accumulate relatively large amounts of starch, these organelles provide a more suitable environment for PHB production, provided that a β-ketothiolase, which is not present in plastids, is co-expressed. An approach in which the required enzymes were targeted to the plastid using signal sequences increased the amount of PHB to a maximum of 14% of dry weight in leaves without significant effects on plant growth, despite evidence of leaf chlorosis (Nawrath et al., 1994). Use of triple gene constructs targeting the PHB pathway in the leukoplast led to PHB accumulation in seeds of oil rape up to 8% dry weight without signs of deleterious effects on seed vigor (Houmiel et al., 1999). Even higher amounts (30–40% shoot dry weight) were obtained in A. thaliana leaves from a similar triple construct, but this resulted in dwarf plants unable to produce seed (Bohmert et al., 2000). Limited metabolite analysis of these later plants revealed complex changes in numerous metabolites, such as organic acids, amino acids and sugars, but remarkably not fatty acids. Expression of the PHB pathway in plastids of alfalfa and tobacco typically yielded lower amounts (<0.5% dry weight), while intermediate results were obtained with corn leaves (6% dry weight), sugar beet hairy roots (5% dry weight) and recently with sugar cane leaves (2% dry weight; Table 1). Attempts to directly express the PHB synthesis route from the plastid genome in tobacco led to relatively low amounts of PHB (<1.7% dry weight) that accumulated mainly in the tissue culture phase and was associated with deleterious effects on growth and male cytoplasmic sterility (Lössl et al., 2003). It should be noted that these synthesis routes are the most complex metabolic pathway to be introduced into the plastid genome so far (Bock, 2007). Finally, synthesis of PHB up to 2% dry weight from the peroxisomal acetyl-CoA pool has been reported for corn cell-suspension cells expressing the three PHB enzymes (Hahn et al., 1999).

With regard to potential applications, PHA containing only C4 monomers (i.e. PHB) may not be the best polymer. Inclusion of 3-hydroxyvalerate or longer 3-hydroxy fatty acids reduces crystallinity and yields a more flexible material. For production of P(HB-co-HV) co-polymers, a supply of propionyl-CoA was introduced using a threonine deaminase (Figure 4). This enzyme yields 2-ketobutyrate, which could be converted to propionyl-CoA by the endogenous pyruvate dehydrogenase complex (Slater et al., 1999; Valentin et al., 1999). Co-expression in the plastid of the three PHB biosynthetic proteins along with a threonine deaminase from E. coli led to P(HB-co-HV) accumulation up to 2.3% dry weight in seeds of oil rape and 1.6% dry weight in A. thaliana leaves (Slater et al., 1999). Interestingly, there was an inverse relationship between the amount of PHA and the proportion of the HV monomer, indicating a bottleneck in providing 3-hydroxyvaleryl-CoA to the PHA synthase. This bottleneck is thought to be caused by inefficiency of the pyruvate dehydrogenase complex in converting 2-ketobutyrate to propionyl-CoA. Theoretically, several other potential pathways could be used to generate propionyl-CoA in plants, such as the degradation pathway of isoleucine or valine, or the conversion of succinyl-CoA via a methylmalonyl-CoA or alanine via acrylyl-CoA (Slater et al., 1999). However, none of these pathways are as direct as the threonine deaminase pathway, and some pathways involve genes that have not yet been identified, enzymes that are oxygen-sensitive or require co-factors not present in plants, or substrates that are found in low amounts in plants. Interestingly, expression in the cytosol of Arabidopsis cells of a mutated PHA synthase from Aeromonas caviae capable of polymerizing C4 and C5 3-hydroxyacyl-CoAs led to accumulation of P(HB-co-HV) up to 0.6% dry weight, with a HV content of 0.2–0.8 mol% (Matsumoto et al., 2005). These results suggest an unidentified source of either propionyl-CoA or 3-hydroxyvaleryl-CoA in the cytosol.

Various approaches were tested to produce mclPHAs in plants (Figure 4). Using the 3-hydroxyacyl-CoA intermediates generated in the peroxisome by the fatty acid β-oxidation pathway, a mclPHA consisting of C6–C16 monomers was produced to a level of 0.4% dry weight in A. thaliana seedlings (Mittendorf et al., 1998). These polymers consisted of more than 40–50 mol% C12 and longer monomers, and also contained a significant amount of unsaturated monomers. As a consequence, the polymer was soft and sticky. Arabidopsis mutants deficient in fatty acid desaturation reduced the number of double bonds in the mclPHA, while co-expression of a medium-chain acyl carrier protein (ACP) thioesterase from Cuphea lanceolata increased the C8–C12 monomer content in both leaves and seeds (Mittendorf et al., 1999; Poirier, 1999). In addition to the biotechnological aspects, synthesis of mclPHA in plants has been used as a valuable tool to study carbon flux through the β-oxidation cycle, as PHA accumulation reflects the carbon flux through the β-oxidation pathway, while monomer composition provides information on the type of fatty acids entering the cycle and how they are degraded (Poirier, 2002). Producing mclPHAs containing longer-chain monomers in plastids using conversion of the fatty acid biosynthetic intermediate 3-hydroxyacyl-ACP into 3-hydroxyacyl-CoA via expression of a bacterial 3-hydroxyacyl-ACP–CoA transacylase led to the accumulation of only very low amounts of mclPHA (below 0.03% dry weight) in potato leaves (Romano et al., 2005).

Probably the most useful PHA would be a polymer containing primarily 3-hydroxybutyrate with a fraction of longer-chain monomers (C6 or higher). This type of polymer has been developed by Proctor & Gamble and Kaneka under the trade name Nodax® (Noda et al., 2005). Enzymes that accept both types of monomers include the native PHA synthases from A. caviae and Pseudomonas sp. 61.3. In addition, these enzymes have been engineered to broaden their substrate range (Nomura and Taguchi, 2007). Several mutants of the Pseudomonas sp. 61.3 and A. caviae PHA synthases were tested in Arabidopsis expressing the PHA synthases in the peroxisomes. However, the maximum amount of PHA produced was only 0.02–0.04% plant dry weight (Arai et al., 2002; Matsumoto et al., 2006b). An interesting result was that, although expression of these PHA synthases in E. coli led to PHA containing mainly C6–C12 monomers (Matsumoto et al., 2006a), expression of the same synthases in plant peroxisomes produced a polymer containing predominantly C4 and C5 monomers. This means that, in the plant peroxisome, the supply of shorter monomers exceeds that of longer monomers. Until now, the amount of PHA co-polymer synthesized from β-oxidation intermediates has been low in plants, although use of the same route in bacteria such as Pseudomonas oleovorans led to over 50% PHA accumulation (Lageveen et al., 1988). The differences in the β-oxidation pathway that accounts for this discrepancy are not known, but could include the presence of enzymes such as a 3-hydroxyacyl-CoA epimerase, enabling more efficient synthesis of the PHA substrate R-3-hydroxyacyl-CoAs from the β-oxidation intermediate S-3-hydroxyacyl-CoAs. Furthermore, metabolic channeling of intermediates of β-oxidation has been observed for the rat mitochondrial β-oxidation pathway, and the presence of similar channeling in peroxisomal β-oxidation may well strongly limit the availability of the intermediate 3-hydroxyacyl-CoA to the PHA synthase (Ishikawa et al., 2004).

From the experience gathered in targeting the PHA pathway to various subcellular organelles, the best location in terms of PHA quantity appears to be the plastid, although clearly more work needs to be done to understand how synthesis of the precursors (e.g. acetyl-CoA, propionyl-CoA, 3-hydroxyacyl-ACP) is regulated and how it could be more efficiently channeled towards PHA without affecting plant growth. Very little work has been done so far to understand how carbon flux is affected by the introduction of a new carbon sink such as PHA, and it is likely that genomics, metabolomics and flux analysis would provide very valuable information in this context.

Commercially viable PHA production in plants is not only determined by the costs and technological possibilities of production in plants, but also depends on the competing production methods. PHA production through bacterial fermentation is now rapidly approaching commercialization (by the joint venture between Metabolix and ADM in the USA, and by producers in China and Brazil). In addition to the general features of bacterial fermentation described above, synthesis of PHA via fermentation has other advantages. The PHA accumulation level in bacteria reaches 50% for mclPHAs and up to 85% for PHB and related short-chain PHAs, which is probably unattainable in plants. Such high-level PHA accumulation leads to easier recovery of the polymer than is likely to be possible in plants. Although several methods for PHA extraction in plants are found in the (largely prophetic) patent literature, actual experimentation on PHA extraction from plants is very limited and clearly deserves more attention (Poirier and Gruys, 2001). An outstanding feature of bacterial PHAs is the enormous flexibility with regard to properties as a function of the nature and ratio of the various monomers. Bacterial fermentation allows the synthesis of a wide spectrum of PHA with various physical properties (based on the choice of polymerase, host, feedstock and conditions), with some PHAs being suitable for low-value commodity applications, while others are suitable for specific high-value niche market applications, such as medical implants (Philip et al., 2007). Reaching similar levels of control and producing a wide spectrum of PHA types is unrealistic in the context of transgenic crops. It is thus clear that synthesis of PHA in crops must be limited to one or perhaps two types of PHAs that would be used for large-scale, low-cost bulk applications, such as a substitute for plastics used in consumer products, while bacterial PHA would be used for higher-value, lower-volume applications. In view of the current experience regarding PHB synthesis in plants, development of a crop plant producing 15% dry weight or more of PHB homopolymer, or of a PHA co-polymer based on 3-hydroxybutyrate, appears realistic. Obtaining the spectrum of physical properties required for large-scale commodity products would require compounding and blending PHB and 3-hydrobutyrate-based co-polymer with other additives or polymers (including perhaps other bacterial PHA co-polymers).

The cost of synthesis of PHA in plants versus bacteria remains somewhat speculative. Previously, PHAs produced by microbial fermentation were considered too expensive and lacking in environmental benefits (Gerngross, 1999). However, the rapid development of biogas technology to utilize waste biomass for the production of process heat and electricity makes PHA production by fermentation more energy- and CO2-efficient (Kim and Dale, 2005). According to some sources, PHAs could be produced by fermentation for approximately US$2 kg−1, still substantially higher than the cost of polypropylene and polyethylene at less than US$1 kg−1 (Philip et al., 2007). Competitive PHA production in plants must assume that PHA is only one of several valuable products obtained from the crop. For example, strategies such as developing Miscanthus, sugar beet or sugar cane as crops for production of biofuel, commodity chemicals and PHA would considerably enhance the economic benefit of PHA in crops. A typical sugar beet harvest yields 12 tonnes of sugar per hectare. If all of this is converted to PHA, with a yield factor of 33% (Anderson and Dawes, 1990), 4 tonnes of PHA could be obtained per hectare. In contrast, as sugar beet produces 17 tonnes of dry root biomass per hectare, a transgenic sugar beet would need to contain at least 23% PHA in total root biomass, which is perhaps a feasible target considering that 5.5% PHB has been synthesized in sugar beet hairy root cultures (Menzel et al., 2003). Alternatively, the economics of PHA production could be improved by synthesizing PHA in a part of the plant that is currently not harvested or used, such as the stover of corn or the leaves of sugar beet or sugar cane. One obvious advantage would be that handling and extraction of the crop for starch, sugar or oil would be unaffected, but new value would be found in a part of the plant that is normally poorly exploited. Such a strategy, however, could have implications that should be carefully examined with regard to the amount of nutrients and litter that is returned to the soil.

Sufficient added value from PHA may also be created more indirectly. For example, low-level PHB accumulation in vascular bundles of flax resulted in increasing strength, Young’s modules and energy for failure of the harvested flax fibers (Wrobel et al., 2004; Wrobel-Kwiatkowska et al., 2007). Similarly, PHB accumulation in cotton fibers led to enhanced insulating properties (John, 1997; John and Keller, 1996).

Natural rubber

One of the most significant biological materials used in non-food applications is natural rubber, a polymer consisting of isoprene units linked together in a 1,4-cis configuration (Figure 5a). Although rubber is produced in over 2500 plant species, commercial production of rubber is almost exclusively from Hevea brasiliensis, the Para rubber tree. The annual production level of natural rubber is nearly 107 tonnes, over 90% of which is harvested in South-East Asia, in particular Indonesia, Thailand and Malaysia.

Figure 5.

 Structure and synthesis of natural rubber.
(a) Structure of natural rubber. In this representation, farnesyl diphosphate is the initiator molecule and approximately 18 000 isoprene units are included in the polymer.
(b) Schematic pathway for rubber biosynthesis from isopentenyl diphosphate (IPP). Natural rubber is produced from a simple side branch of the ubiquitous isoprenoid pathway, with 3-hydroxy-methyl-glutaryl-CoA as the key intermediate derived from acetyl-CoA via the general mevalonic acid pathway. IPP is converted to dimethylallyl diphosphate (DMAPP) by IPP isomerase. IPP is then condensed in several steps with DMAPP to produce geranyl diphosphate (GPP) and farnesyl diphosphate (FPP) by the action of trans-prenyltransferases. The cis-1,4-polymerization is catalyzed by a rubber transferase (also called cis-prenyltransferase or rubber polymerase), which uses the non-allylic IPP as substrate.

Natural rubber has unique properties due to its structure, molecular weight and some poorly defined components, such as proteins, lipids, carbohydrates and minerals, that are present in the latex. These properties include resilience, elasticity, abrasion and impact resistance, efficient heat dispersion and malleability at cold temperatures (Cataldo, 2000; Cornish, 2001a). Although synthetic rubbers, such as styrene–butadiene or acrylonitrile–butadiene co-polymers, are commercially produced to levels similar to natural rubber, none of them match the price–performance ratio of natural rubber. The unique properties of natural rubber make it essentially irreplaceable for many applications, such as heavy-duty tires for trucks, buses and airplanes, as well as latex products for medical applications (Cornish, 2001a). While some plants can produce other types of polyisoprenes, none of them have material properties or value similar to Hevea rubber. For example, gutta-percha and balata are flexible but inelastic materials from Palaquium gutta and Manilkara bidentata, respectively, formed of isoprene units linked together in a 1,4-trans configuration (Polhamus, 1962). The inertness of gutta-percha or balata to biodegradation made them useful as an impermeable coating for undersea cables, and gutta-percha is still used in dentistry as a filling material. Chicle is a polyisoprene from Achras sapota that contains both cis and trans bonds in a 1:2 ratio (Polhamus, 1962). Chicle is still used by some companies for chewing gum.

Despite the economic importance of natural rubber, not all steps in its biosynthesis have been characterized in detail (van Beilen and Poirier, 2007a; Cornish, 2001b). Rubber synthesis starts with cytosolic acetyl-CoA, which is converted via 3-hydroxy-glutaryl-CoA to isopentenyl diphosphate (IPP). Farnesyl diphosphate is the likely physiological initiator molecule for rubber synthesis (Figure 5b; Stubbe et al., 2005). Rubber polymerase, or cis-prenyltransferase, is thought to be embedded in the membrane monolayer surrounding the rubber granules, and adds isoprenyl units from IPP to form the rubber polymer. A major question remaining with regard to rubber synthesis relates to the nature of the protein (or protein complex) required for the synthesis of high-molecular-weight rubber. A cis-prenyltransferase thought to be involved in rubber synthesis has been cloned from H. brasiliensis and expressed in E. coli, but was found to catalyze the formation of only short-chain cis-polyisoprenes (Ko et al., 2003). Long-chain rubber polymer could only be formed when the purified cis-prenyltransferase was combined with washed rubber particles (i.e. rubber granules with associated proteins and lipids; Asawatreratanakul et al., 2003), and it has not been established whether the rubber particles contain the sought-for long-chain rubber polymerase or another protein or co-factor that enables the expressed cis-prenyltransferase to produce long-chain rubber. Studies are complicated by the fact that the dominant proteins attached to rubber particles differ radically in number, size and predicted function based on sequence comparisons between the various plant species that produce rubber (van Beilen and Poirier, 2007a; Cornish, 2001a).

Two main problems are associated with the production of rubber from H. brasiliensis. The rubber tree used in commercial production in South-East Asia has a very narrow genetic basis, coming essentially from a few seeds collected at one location in Brazil by Dr Henry Wickham in 1876 (Polhamus, 1962). It is very sensitive to the pathogen Microcyclus ulei, the causative agent of South American leaf blight (Le Guen et al., 2007). South American leaf blight has essentially ended large-scale Hevea rubber production in South America, and can be expected to have a similar devastating effect if it spreads to Asia (Davis, 1997). Currently, the only barrier to prevent the spread of South American leaf blight to Asia is strict quarantine. It is thus essential to further understand the biology of South American leaf blight from both the pathogen and host perspectives in order to devise adequate defense strategies, which may, in the long term, involve the development of a genetically modified rubber tree. An emerging issue with natural rubber is the growing number of people worldwide who are allergic to certain proteins present in Hevea rubber. It is now estimated that 1–6% of the general population suffer from Hevea latex allergies, and up to 17% of healthcare workers are at risk of developing allergic reactions (Bousquet et al., 2006).

In view of the problems related to rubber from H. brasiliensis, the strategic importance of natural rubber and the uncertainties of its future supply, development of an alternative source of natural rubber has been identified as a clear priority (Mooibroek and Cornish, 2000). Large research programs established during World War II identified two main plants suitable for rubber production, namely guayule (Parthenium argentatum Gray) and the Russian dandelion (Taraxacum koksaghyz). Both plants produce a high amount of high-molecular-weight rubber, an essential determinant of rubber quality (van Beilen and Poirier, 2007a).

Guayule is a shrub native to Mexico and the south-western USA that can also be grown in southern Europe and northern Africa. One advantage of guayule rubber is that its latex contains fewer proteins, both in terms of diversity and total amount, than the latex from H. brasiliensis, and these proteins do not cross-react with immunoglobulins reactive against H. brasiliensis latex proteins (Siler et al., 1996). This makes allergenic reactions to guayule rubber by consumers sensitized to Hevea rubber unlikely. Indeed, Yulex Corporation is currently marketing guayule latex as a source of hypoallergenic rubber, which is particularly useful for medical applications. In guayule, rubber latex accumulates intracellularly in the bark parenchymal cells. Thus, the latex cannot be harvested by tapping, as with the rubber tree, but the guayule shoot material must be harvested and thoroughly disrupted to release the rubber particles (Cornish et al., 2006). Thus, while rubber harvest can be mechanized for guayule, extraction is technically more complicated than with H. brasiliensis. Productivity of rubber from guayule of up to 2000 kg ha−1 year−1 has been reported, while up to 3000 kg ha−1 year−1 can be produced from the Brazilian rubber tree. While high-value applications, such as hypoallergenic latex for medical applications, justify the higher cost of guayule rubber, large-scale use of guayule rubber as an alternative to Hevea rubber in lower-value items, such as tires, will require improvements at the level of the productivity of the varieties used and in agronomic and harvesting practices, as well as optimization of the extraction method (van Beilen and Poirier, 2007b).

Russian dandelion accumulates rubber in its roots and is considered an attractive annual crop for temperate regions, especially as the rubber is of high molecular weight. Although Russian dandelion has laticifers like the rubber tree, rubber must be harvested by homogenizing the roots. Yields of 150–500 kg of rubber ha−1 year−1 have been reported (van Beilen and Poirier, 2007a). Although rubber was produced from dandelion during World War II to make tires, limited progress has been achieved since then to improve production. Turning Russian dandelion into a viable crop for rubber production would require an increase of vigor and more favourable agronomic properties, such as larger roots that are easier to harvest, and increased rubber accumulation. In addition, the low yield per hectare, labour-intensive cultivation, crosses and seed contamination with other dandelions, and weed potential must be tackled (van Beilen and Poirier, 2007b).

For both guayule and Russian dandelion, modern molecular approaches are required to better understand and modulate rubber biosynthesis in these alternative crops. Although large gaps are present in our knowledge of the rubber polymerase, studies of the sterol and isoprenoid biosynthesis pathways have led to a good understanding of the enzymes involved in the synthesis of IPP and farnesyl diphosphate, and this knowledge could be used, via transgenesis, to overexpress or downregulate key genes involved in rubber biosynthesis and accumulation. This type of functional approach is easier to apply to Russian dandelion because of its short lifecycle, such that detectable rubber phenotypes can be obtained after six months. In contrast, for H. brasiliensis, several years of growth are needed before clear rubber phenotypes can be determined, while 1–2 years are required for guayule. Furthermore, to make rubber from either Russian dandelion or guayule competitive for large-scale low-value applications, it is essential to consider the plants in the context of a biorefinery, where all components, including waste material, would be used. Co-products from both guayule and Russian dandelion could include biogas or bioethanol from the cellulosic waste, or various specialty chemicals from secondary metabolites. Side-products of Russian dandelion processing could also include inulin, a polymer of fructose, which is the major storage sugar of dandelions (25–40% of dry root weight), and could be used directly in non-food applications, used to make the polymer building block 5-hydroxymethylfurfural (see below), or fermented to bioethanol (van Beilen and Poirier, 2007a).

Other plants have been examined as a potential source of natural rubber. These include golden rod (Solidago altissima), lettuce and sunflower. However, unfavorable characteristics, such as a very low yield of rubber or low molecular weight of the polymer, make them more difficult subjects for development as rubber-producing crops at this stage compared to guayule or Russian dandelion.

Production of monomers in planta

An alternative to the synthesis of biopolymers in planta is the synthesis of small molecules that can be extracted and used directly or indirectly to synthesize polymers, typically via chemical modification and polymerization. Most of these molecules, such as 4-hydroxybenzoate, sorbitol and fructose, are found naturally in some plants, and metabolic engineering would utilize the expression of genes from plants or other organisms as a mean to increase the synthesis and storage of these compounds. On the other hand, molecules such as methacrylate or itaconic acid are not found in plants, and metabolic engineering would be required to create the necessary pathways in plants (Figure 6).

Figure 6.

 Chemical building blocks that can be produced in plants for polymer synthesis.
(a) Pathway for the synthesis of 4-hydroxybenzoate (pHBA). Chorismate present in the plastid can be converted to pHBA by a bacterial chorismate pyruvate lyase (CPL). pHBA transits to the cytosol by an unknown mechanism where it is glycosylated by endogenous uridine diphosphate glycosyltransferases (UDP-GT) before being stored in the vacuole. The alternative pathway is the conversion of 4-coumaryl-CoA by a 4-hydroxycinnamoyl-CoA hydratase/lyase (HCHL) to 4-hydroxybenzaldehyde via a β-hydroxy-CoA thioester intermediate. The majority of the 4-hydroxybenzaldehyde is oxidized to pHBA by an unidentified plant enzyme.
(b) Examples of chemical building blocks and how they could be used for the synthesis of polymers.

Poly-4-hydroxybenzoate is an important liquid crystal polymer, a compound that combines the properties of polymers and ordinary liquid crystals. Although all plants normally synthesize small quantities of the monomer 4-hydroxybenzoate (pHBA) from the phenylpropanoid pathway, transgenesis has been used in order to increase the quantity that accumulates in plants. Two main pathways for pHBA synthesis have been exploited, namely the direct conversion of plastidial chorismate to pHBA by a bacterial chorismate pyruvate lyase, and conversion of cytosolic 4-hydroxycoumaryl-CoA to pHBA via a 4-hydroxybenzaldehyde intermediate by expression of a bacterial 4-hydroxycinnamoyl-CoA hydratase/lyase (Figure 6a). While expression from the nuclear genome of the E.  coli chorismate pyruvate lyase modified for plastid targeting led to accumulation of pHBA–glucose conjugates up to 0.5% and 1.5% dry weight in tobacco and sugar cane leaves, respectively (McQualter et al., 2005; Siebert et al., 1996), expression of the same gene from the plastid genome led to remarkable accumulation of pHBA–glucose conjugates up to 25% of the leaf dry weight without a significant effect on plant growth (Viitanen et al., 2004). These latter results demonstrate the high flux potential and flexibility of the shikimate pathway for chorismate synthesis. Expression of 4-hydroxycinnamoyl-CoA hydratase/lyase from Pseudomonas fluorescens in sugar cane led to accumulation of pHBA–glucose conjugates up to 7% dry weight in leaves without adverse effects despite a severe reduction in leaf chlorogenic acid, subtle changes in lignin composition, and compensatory upregulation of phenylalanine ammonia lyase (Brumbley et al., 2007; McQualter et al., 2005). In all cases, the pHBA–glucose conjugates are thought to accumulate in the vacuole.

Derivatives of furans have the potential to serve as substitutes for building blocks derived from petrochemistry that are normally used in the production of plastics. For example, 5-hydroxymethylfurfural (HMF) can be converted to 2,5-furandicarboxylic acid, which can be used as a replacement for terephthalic acid, which is used in production of the common polyester polyethylene terephthalate (PET). HMF can be synthesized from fructose, and recent advances in two-phase chemistry resulted in a process whereby 90% of fructose from a highly concentrated syrup can be converted to HMF (Roman-Leshkov et al., 2006). These results in the field of chemistry can now achieve their full value through combination with advances in the development of plants engineered for accumulation of a high amount of fructans (Ritsema and Smeekens, 2003).

Sorbitol is a versatile chemical that is also on the list of the top 12 most useful chemical building blocks (Werpy, 2004). Sorbitol is a polyol that is used as a sweetener and also as a monomer for PET-like polyesters such as polyethylene isosorbide. Sorbitol could be produced in sugar cane at levels up to 61% of soluble sugars in the leaves or 12% dry weight, but only 1% was found in the stalk pith (Fong Chong et al., 2007). Although biomass yields (30–40% less) and plant health (necrosis) were affected, this study demonstrates that metabolic intermediates of the sucrose biosynthetic pathway can be effectively diverted for the synthesis of alternative valuable compounds. Other potential polyols include inositol or galactinol, which are compatible solutes that are produced in the cytoplasm of plant cells in response to osmotic stress (Umezawa et al., 2006).

Methacrylate is the building block for synthesis of the transparent plastic polymethyl methacrylate, and is also used for the production of the co-polymer methyl methacrylate–butadiene–styrene. The related molecule itaconic acid is also used in the synthesis of co-polymers with acrylic acid and styrene–butadiene. Although no biological route for the synthesis of methacrylate is known, itaconic acid is currently synthesized through fermentation via a pathway that involves decarboxylation of aconitate (Willke and Vorlop, 2001). The US company Ceres Inc. has recently received a US$1 500 000 grant from the Joint Biomass Research and Development Initiative of the Department of Energy and the United States Department of Agriculture to develop a plant-based production system for methacrylate or methacrylate equivalents, such as itaconic acid, by genetically engineering existing metabolic pathways in cellulosic ethanol biomass crops such as switchgrass. Sugar cane produces aconitic acid, a precursor of itaconic acid, which is believed to cause problems in sugar refining (Tsao et al., 1999). Itaconic acid is also listed as one of the 12 top added value chemicals that may be derived from biomass (Werpy, 2004). Approximately 15 000 tonnes is produced annually in fermentation processes, and it is a rather expensive compound that is currently mainly used in specialties (Lucia et al., 2006).

Conclusion

Bringing about the production of renewable polymers from novel crop plants is a challenging task that will require scientific advances and successful interactions between several fields. A better understanding of the basic plant metabolic pathways, including how carbon flux is controlled, will be a key factor in creating plants that will accumulate larger amounts of interesting polymers or monomer building blocks without affecting plant health or yield. There is also a need for a better control of the expression of multiple genes as most applications will require stacking of transgenes. Success also hinges on advances in chemical engineering, as efficient extraction of polymers and monomers from plants and integration into biorefining facilities will be crucial. Polymer chemistry will play an important role in the cost- and energy-effective conversion of the building blocks into polymers and in the modification of the extracted polymers for applications in a wide range of products, either through adequate blending or chemical modifications. Finally, a key driver in this process is policy. Recognition of the inescapable need to reduce CO2 emission and our dependence on fossil fuels – not only for energy, but also for a wide range of chemicals – by policy makers and governments, is essential to provide the impetus and long-term funding required to support this scientific endeavor.

Acknowledgements

This work was supported by the Sixth Framework Programme of the European Commission (EPOBIO, SSP/022681).

Ancillary