• Open Access

PHA Bioplastics, Biochemicals, and Energy from Crops


Correspondence (Tel 617 583 1729;fax +1 617 583 1767;

email snell@metabolix.com)


Large scale production of polyhydroxyalkanoates (PHAs) in plants can provide a sustainable supply of bioplastics, biochemicals, and energy from sunlight and atmospheric CO2. PHAs are a class of polymers with various chain lengths that are naturally produced by some microorganisms as storage materials. The properties of these polyesters make them functionally equivalent to many of the petroleum-based plastics that are currently in the market place. However, unlike most petroleum-derived plastics, PHAs can be produced from renewable feedstocks and easily degrade in most biologically active environments. This review highlights research efforts over the last 20 years to engineer the production of PHAs in plants with a focus on polyhydroxybutryrate (PHB) production in bioenergy crops with C4 photosynthesis. PHB has the potential to be a high volume commercial product with uses not only in the plastics and materials markets, but also in renewable chemicals and feed. The major challenges of improving product yield and plant fitness in high biomass yielding C4 crops are discussed in detail.

Why produce polyhydroxyalkanoates in crops?

In 1989, a publication with the headline ‘In search of the plastic potato’ (Pool, 1989) created great excitement within the scientific community with the prediction that plants could be engineered to produce the bioplastic polyhydroxybutyrate (PHB). PHB is the simplest member of the polyhydroxyalkanoate (PHA) family of microbial carbon and energy storage materials, and this natural thermoplastic polyester and its variants have properties similar to petroleum-based plastics yet are biodegradable. The initial excitement in this field was based on the potential to produce PHB in crops at costs similar to starches and vegetable oils, a feat which would help address needs for cost-competitive alternatives to petroleum feedstocks as well as waste management. In 1992, scientists reported the production of PHB in Arabidopsis thaliana (Poirier et al., 1992) and initiated a wave of research activity to optimize PHA production in plants that continues today.

Since this initial plant research milestone, there has been increased demand for petroleum-based products, including plastics, fuels, and chemicals, from high growth economies in China and India. This growth coupled with the depletion of existing oil fields, and a decline in the rate of their replacement has led to increases in the costs of fossil resources and energy. Some experts have noted that oil production has remained essentially constant from 2005 through 2011 and have suggested that the world's maximum production capability has already been reached (Murray and King, 2012). Increased costs and volatility in petroleum prices have and will continue to have a major effect on the global economy (Murray and King, 2012). Fluctuating oil prices have, however, also helped to drive the production and adoption of ethanol as a biofuel as well as major initiatives to develop sources of biobased chemicals as drop-in replacements for existing chemical building blocks. These developments have placed considerable strain on existing crop resources and driven massive new investment in basic biological science research. Advanced genomics and metabolic engineering technologies developed with these efforts are beginning to be deployed in dedicated biomass bioenergy crops to support the development of low-cost cellulose hydrolysates and sugars for biorefineries.

Biofuels and biochemicals produced from biomass can have a significant impact on meeting the world's growing demand for materials and energy. In the United States, it is estimated that total biomass production potential is one billion tons, enough to displace approximately 30% of the country's petroleum use (Perlack et al., 2005; US Department of Energy, 2011). Under high-yield scenarios, 1.6 billion tons of biomass could be available by 2030 with energy crops contributing the largest amount. Cost-competitive biorefineries would process this abundance of biomass to both fuels and value-added co-products, mimicking the production of multiple products by their petroleum counterparts. Leveraging the added value from these co-products would improve the overall economics of the biorefinery and justify investment in large production facilities (Snell and Peoples, 2009).

PHAs are an ideal value-added co-product for crops as they possess properties making them suitable replacements for many of the high-volume plastics that are currently produced from petroleum resources. In this review, we will highlight studies that have significantly contributed to the science knowledge base for the production of PHAs in plants with a focus on high-biomass-yielding C4 bioenergy crops. As many of the most productive plants in nature possess C4 photosynthetic pathways, these plants are attractive targets for engineering the production of biofuels and industrial materials where yield per acre dominates the economics and thus commercial viability. A comprehensive list summarizing other efforts to engineer production of PHAs in plants with noted highlights of the significance of each body of work is provided in Tables 1-3.

Table 1. Polyhydroxyalkanoate production in plastids of C3 plants
PlantPolymerTissuePHA contentaReference & Notes
  1. DW, dry weight; FW, fresh weight; ppm, parts per million; phaA; gene encoding β-ketothiolase; phaB, gene encoding reductase; phaCSCL, gene encoding PHA synthase with preference for short-chain-length substrates; phaP, gene encoding phasin; phaG, gene encoding 3-hydroxyacyl-acyl carrier protein-coenzyme A transferase; fabH, gene encoding 3-ketoacyl-acyl carrier protein synthase III; phaCMCL, gene encoding synthase with preference for medium-chain-length substrates; PHASCL, short-chain-length PHA; PHAMCL, medium-chain-length PHA; PHBV, copolymer of 3-hydroxybutyrate and 3-hydroxyvalerate; HB, 3-hydroxybutyrate; HV, 3-hydroxyvalerate; HH, 3-hydroxyhexanoate; HO, 3-hydroxyocanoate; HD, 3-hydroxydecanoate; HDD, 3-hydroxydodecanoate; HTD, 3-hydroxytetradecanoate; mol %, mole % of indicated monomer in copolymer; MW, molecular weight of polymer; PDI, polydispersity of polymer.

  2. a

    Values for top PHA-producing transgenic lines in each study are listed. % PHA is defined as g PHA/g biomass ×100. In cases where values have been reported with different units in the literature, the appropriate conversion has been made to report % PHA biomass as defined above.

AlfalfaPHBLeaves0.18% DWSaruul et al. (2002)
Arabidopsis PHBLeaves14% DWFirst demonstration of PHB production in plastids, 3 separate constructs transformed with subsequent crossing to obtain complete pathway, some chlorosis in leaves at higher PHB levels (Nawrath et al., 1994)
12%–13% DWCompared performance of separate transformations of single-gene constructs and crossing to transformation of one multi-gene construct, multi-gene constructs produced highest PHB (Valentin et al., 1999)
4% FW (∼ 40% DW)Highest level of polymer production in plants to date, transformation of multi-gene constructs (Bohmert et al., 2000)
13.2% DWChemical or somatic induction of phaA (Bohmert et al., 2002)
5.8% DWCo-expression of phaP with base PHB pathway, health of plants producing >0.15% FW compromised (Bohmert et al., 2002)
6.4% DWNovel thiolase–reductase gene fusion co-expressed with phaC (Kourtz et al., 2005)
14% DWChemical induction of all three genes of the PHB pathway, improved plant health observed (Kourtz et al., 2007)
PHBVLeaves0.07%–1.6% DWHV content 2–17 mol % (Slater et al., 1999); polymer MW of 5.5 × 105 with PDI of 1.8
0.2%–0.8% DWHV content 4–17 mol % (Valentin et al., 1999); polymer MW of 5.4 × 105 with PDI of 1.8
PHASCL/MCLWhole plants without roots0.18% DWHighly active mutant of phaCMCL expressed with phaA and phaB; HB, HV, HH monomers detected (Matsumoto et al., 2009)
0.14% DWMutants of fabH and phaCMCL encoding highly active enzymes expressed with phaA and phaB; HB, HV, HH, HO, HD, HDD, HTD monomers detected (Matsumoto et al., 2009).
Brassica napus PHBSeeds7.7% FWFirst demonstration of PHB production in seeds (Houmiel et al., 1999; Valentin et al., 1999)
PHBVSeeds0.7%–2.3% DWHV content 2.3–6.4 mol %; MW of polymer ranged from 0.9 × 106 to 1.2 × 106 with PDI of 1.9–2.4 (Slater et al., 1999; Valentin et al., 1999)
Camelina sativa PHBSeeds19.9% DW PHB in single seedsHighest levels reported in seeds to date; high levels of PHB-impaired survival of seedlings (Patterson et al., 2011a,b)
FlaxPHBStem0.005% FWImprovement in elastic properties of fibres; increase in cellulose, decrease in lignin and pectin content in stems; no change in cellulose, lignin, pectin of fibres (Wróbel-Kwiatkowska et al., 2007; Wróbel et al., 2004)
0.00045% FWField-grown PHB-producing flax; PHB production improved retting (fibre isolation) from flax (Wróbel-Kwiatkowska et al., 2009)
Oil palmPHBLeavesNo data reportedGenetic constructs assembled and plants transformed (Ismail et al., 2010; Masani et al., 2009)
MesocarpNo data reportedGenetic constructs with mesocarp-specific promoter used for transient expression (Omidvar et al., 2008)
PHBVLeavesNo data reportedGenetic constructs assembled and plants transformed (Ariffin et al., 2011; Masani et al., 2009; Parveez et al., 2008)
PoplarPHBLeaves3.69% DWChemical induction of PHB pathway (Dalton et al., 2011)
PotatoPHBLeaves0.009% DWChemical or somatic induction of phaA (Bohmert et al., 2002)
PHAMCLLeaves0.00026% DWphaG-mediated PHAMCL production (Romano et al., 2005)
Sugar beetPHBHairy root culture5.5% DWMW of polymer ranged from 1.63 × 106 to 1.75 × 106; PDI ranged from 1.34 to 1.35 (Menzel et al., 2003)
TobaccoPHBLeaves0.09%–0.15% DWPolymer values reported as 900–1500 ppm DW (Nakashita et al., 2001b)
0.16% DWMaximum polymer yields in senescent leaves (Arai et al., 2001)
0.16% DWIncreased PHB production with chemical inhibition of acetyl-CoA carboxylase (Suzuki et al., 2002)
0.08% DW(Bohmert et al., 2002)
0.04%–0.32% DWChemical or somatic induction of phaA (Bohmert et al., 2002)
0.0002%–0.0008% DWPlastid transformation; polymer values reported as 2–8 ppm DW (Nakashita et al., 2001a)
0.005%–0.016% DWPlastid transformation; polymer values reported as 50–160 ppm DW (Nakashita et al., 2001b)
1.7% DWPlastid transformation; up to 1.7% DW obtained in tissue culture, levels dropped significantly in soil, sterile plants obtained (Lössl et al., 2003)
0.005%–0.016% DWPlastid transformation; polymer values reported as 50–160 ppm polymer in tissue culture, levels dropped in soil (Arai et al., 2004)
0.14% DWPlastid transformation and chemical induction; polymer levels reported as 1383 ppm DW (Lössl et al., 2005
Up to 18.8% DW in leaves; 8.8% DW whole plantPlastid transformation; stable fertile plants obtained, isolated polymer possessed MW of 471 000 and PDI of 2.2 (Bohmert-Tatarev et al., 2011)
PHAMCLLeaves0.48% DWphaG-mediated PHAMCL production via plastid transformation, HO and HD monomers detected (Wang et al., 2005)
Table 2. Polyhydroxyalkanoate production in cytosol, peroxisomes and vacuoles of C3 plants and algae
PlantPolymerTissue/OrganellePHA contentaReference & Notes
  1. DW, dry weight; FW, fresh weight; ppm, parts per million; phaA; gene encoding β-ketothiolase; phaB, gene encoding reductase; phaC, gene encoding PHA synthase; PHASCL, short-chain-length PHA; PHAMCL, medium-chain-length PHA; HB, 3-hydroxybutyrate; HH, 3-hydroxyhexanoate; HO, 3-hydroxyoctanoate; PhaCMCL, synthase with preference for medium-chain-length substrates; FatB3, acyl-acyl carrier protein thioesterase; DGAT, sn-1,2-diacylglycerol acyltransferase. MW, molecular weight of polymer; PDI, polydispersity of polymer.

  2. a

    Values for top PHA-producing transgenic lines in each study are listed. % PHA is defined as g PHA/g biomass ×100. In cases where values have been reported with different units in the literature, the appropriate conversion has been made to report% PHA biomass as defined above.

Arabidopsis PHBLeaves/cytosol0.01% FWFirst demonstration of PHB production in plants, polymer granules observed in cytosol, nucleus and vacuole (Poirier et al., 1992)
0.05% FWPoirier et al. (1995)
Suspension culture/cytosol0.12% FWGranules observed in cytosol, nucleus and vacuole, isolated polymer possessed MW of 615 000 and PDI of 10.5 (Poirier et al., 1995)
PHASCLWhole plants without roots/cytosol0.61% DWPrimarily HB with small amount of HV monomer; strategy employed mutant PHA synthases with high activity (Matsumoto et al., 2005)
PHAMCLLeaves/peroxisome0.02%–0.4% DWMeasured PHA throughout plant life cycle; highest PHA levels observed in 7-day-old seedlings grown in culture media; monomer units from 6–16 carbons; polymer MW of 23 700 with PDI of 4.3; polymer observed in peroxisomes and vacuole (Mittendorf et al., 1998)
Seeds/peroxisome0.02% DWPeroxisome-targeted PhaCMCL (Poirier et al., 1999)
0.11% DWPeroxisome-targeted PhaCMCL, plastid-targeted FatB3 (Poirier et al., 1999)
0.06% DWPhaCMCL, reduced DGAT activity (Poirier et al., 1999)
PHASCLVarious plant developmental stages/peroxisome0.001%–0.044% DWHB, HV, HH monomer units; expressed gene encoding PhaCSCL only (Arai et al., 2002)
Up to 1.8% FWPredominantly HB with small amount of HV, highest levels observed in dark-grown seedlings, expressed phaA, phaB, phaC for maximum yield (Tilbrook et al., 2011)
Chlamydomonas reinhardtii (green algae)PHBCytosol6 × 10−4% DWChaogang et al. (2010)
CottonPHBCytosol0.003%–0.34% DWChange in thermal properties of cotton fibres (Chowdhury and John, 1998; John, 1998; John and Keller, 1996)
Phaeodactylum tricornutum (microalgae)PHBCytosol10.6% DWHempel et al. (2011)
RicePHBWhole plants/cytosol0.5% DWEndo et al. (2006)
SoybeanPHBSeed coat/vacuole0.36% DWStrategy employed seed coat-specific promoter and targeting signal to deliver PHB enzymes to vacuole (Schnell et al., 2012)
TamarixPHBWhole plants/cytosol0.32% DWEndo et al., (2006)
TobaccoPHBLeaves/cytosol0.09% DWIncreased PHB production with chemical inhibition of either acetyl-CoA carboxylase or 3-hydroxy-3-methylglutaryl-CoA reductase (Suzuki et al., 2002)
0.001% DWPolymer MW of 32,000 with PDI of 1.90 (Nakashita et al., 1999)
0.01% DW senescent leavesPolymer value reported as 100 ppm DW (Nakashita et al., 2001b)
approximately 0.06% DWCodon-optimized phaB and/or phaC (Matsumoto et al., 2011)
PHASCLLeaves/peroxisome0.0001% DWPolymer value reported as 1 ppm; HB, HV, HH monomer units (Nakashita et al., 2001b)
Seeds/peroxisome0.00001% DWPolymer value reported as 0.1 ppm; HB, HV, HH monomer units (Nakashita et al., 2001b)
PHASCL/MCLLeaves/peroxisome0.022% DW0.0036% DW observed in stems. HB, HV, HH, HO monomer units (Matsumoto et al., 2006)
Table 3. Polyhydroxyalkanoate production in C4 plants
PlantPolymerTissue/OrganellePHA contentaReference & Notes
  1. DW, dry weight; FW, fresh weight; ppm, parts per million; phaA, gene encoding β-ketothiolase; phaB, gene encoding reductase; phaC, gene encoding PHA synthase; phaJ2, gene encoding enoyl-CoA hydratase; fatB2, acyl-ACP thioesterase; kasA1, 3-ketoacyl-ACP synthase; PHASCL, short-chain-length PHA; PHAMCL, medium-chain-length PHA; HH, 3-hydroxyhexanoate; HO, 3-hydroxyoctanoate; MW, molecular weight of polymer; PDI, polydispersity of polymer.

  2. a

    Values for top PHA-producing transgenic lines in each study are listed. % PHA is defined as g PHA/g biomass ×100. In cases where values have been reported with different units in the literature, the appropriate conversion has been made to report % PHA biomass as defined above.

MaizePHBLeaves/plastid5.66% DWPHB primarily in plastids of bundle sheath cells (Mitsky et al., 2003)
5.73% DWPHB primarily in plastids of bundle sheath cells (Poirier and Gruys, 2002)
PHASCLSuspension culture/peroxisome0.2% FWMW of 100 000 with PDI of 2.69 (Hahn et al., 1999)
SugarcanePHBLeaves/plastid1.88% DWPHB found in all plastids except mesophyll in the leaves and thin-walled cortical cells in the stalk rind (Petrasovits et al., 2007)
1.77% DW leaves; 0.79% DW PHB whole plantPurnell et al., (2007)
4.8% DWPHB primarily in plastids of bundle sheath cells (Petrasovits et al., 2012)
Leaves/cytosolTrace levelsPetrasovits et al., (2007)
Leaves/mitochondrionNo detectable PHBPetrasovits et al., (2007)
PHASCLLeaves/peroxisome and vacuole0.4%–1.6% DWPredominantly HB with small amount of HV (Tilbrook et al., 2011)
PHASCL/MCLLeaves/peroxisome0.015% DWCo-expression of peroxisome-targeted phaA, phaB, phaC and phaJ2, and plastid-targeted fatB2 and kasA1. Monomers from 4 to 16 carbons observed. Mw of 111 000 with PDI of 1.2 (Anderson et al., 2011)
SwitchgrassPHBLeaves/plastid3.7% DW leaves; 1.2% DW whole tiller

PHB primarily in plastids of bundle sheath cells with some inclusions in mesophyll plastids (Somleva et al., 2008). MW of 2.57 × 106 to 3.76 × 106 depending

on line; PDI of 1.49–2.37 depending on line. Mw and PDI measured by gel permeation chromatography (Metabolix, unpublished results)

6.1% DW leaves; 2.3% DW whole tillerPlants propagated through immature inflorescence-derived callus cultures or nodal segments from PHB-producing primary transformants (Somleva and Ali, 2010)
7.7% DW leaves; 1.7% DW whole tillerCo-expression of a gene encoding bifunctional FBPase/SBPase with base PHB pathway (Somleva et al., 2012)

Properties and uses of PHAs

PHAs function as a reservoir of stored carbon that can be degraded when needed by a wide range of microorganisms capable of expressing intracellular or extracellular depolymerase enzymes (Jendrossek and Handrick, 2002; Madison and Huisman, 1999). This feature makes PHAs degradable in all biologically active environments including soil, rivers and oceans, compost and sewage, yet the plastics are very stable in many of the environments where they would encounter everyday use. PHAs can have different monomer units. The final composition of the polymer chain is determined by both the carbon source that is available to the organism and the substrate specificity of the host organism's PHA synthase, the enzyme that polymerizes hydroxyacyl-CoA units to polymer. In plants, unique pathways must be engineered to allow conversion of existing plant metabolites to monomers that can be polymerized by a PHA synthase. PHB is synthesized from the central metabolite acetyl coenzyme A (acetyl-CoA), which is converted to polymer in three simple enzyme-catalysed steps. Because of the relative simplicity of its biosynthesis, PHB has received the most attention as a target molecule for production in plants. PHB has some limitations as a plastic because it is a very hard and brittle material. This is not very different from the current leading compostable plastic polylactic acid (PLA) made through chemical polymerization of lactic acid produced from a fermentation process. PLA material is fairly low cost, and its uses in the plastic market have expanded based on the development of additive materials which improve its properties (Sherman, 2010). Applications of PHB can also be expanded by the addition of other materials and additives (Chen and Luo, 2009; Ha and Cho, 2002; Yu et al., 2006). Given the major volume potential of PHB as a plastic and its current natural prevalence in the environment in multiple ecosystems (Snell and Peoples, 2009), the authors are of the opinion that PHB is likely to be the first PHA to be produced directly from crops.

While PHAs have mainly been used as biodegradable bioplastics, the unique features of these polymers will allow their use in other applications (Figure 1). For example, heating PHB or biomass containing PHB under the appropriate thermolysis conditions can break the polymer chain releasing the chemical intermediate crotonic acid, which is easily recovered (van Walsem et al., 2011; Figure 2). Although a small-volume chemical today, crotonic acid can be transformed using existing chemistries to a number of drop-in commodity chemicals, including propylene by decarboxylation (Peterson and Fischer, 2010) and butanol by hydrogenation (Figure 2). The ability to sequester PHAs in a cell as an inert granular material makes it possible to produce large amounts of a readily convertible polymeric chemical precursor in a biological system where accumulation of the desired chemical itself might be toxic to the host. This broadens the range of renewable chemicals that can be produced from engineered plant feedstocks, and thermolysis provides a simple, scalable, highly efficient recovery option. PHB can also be used as a feed supplement and has been shown to have nutritional value and/or prebiotic effects in studies with broiler chicks, sheep, pigs, fish and prawns (Boon et al., 2010; De Schryver et al., 2010; Forni et al., 1999a,b; Najdegerami et al., 2012; Nhan et al., 2010; Peoples et al., 2001). This application is an exciting opportunity for oilseed crops, as the crushed meal before or after oil extraction could be used directly as a value-added feed. The diversity of uses for PHB is quite unique for a biologically produced molecule and can be broadly exploited to fully reap the benefit of the necessary technology investment to develop a direct crop-based production system.

Figure 1.

Diverse products can be obtained from PHB, including bioplastics, chemicals and feed supplements.

Figure 2.

Production of chemicals from PHB containing biomass using thermolysis procedures. FAST™ refers to a fast-acting selective thermolysis process developed at Metabolix to release chemicals from PHA polymers.

Overview of prior efforts to produce PHAs in plants

Since the first report describing the engineering of Arabidopsis to produce PHB in the cytosol (Poirier et al., 1992), numerous articles have been published describing the production of PHAs containing a range of monomer units in different plant species, tissues and organelles (Tables 1-3). PHB has been the target of most of this research as it requires the coordinated expression of only three bacterial genes (Madison and Huisman, 1999) to convert acetyl-CoA, a ubiquitous plant metabolite, into polymer. Acetyl-CoA is present at different levels in the cytosol, plastids, mitochondria and peroxisomes of plant cells, and because cellular membranes are impermeable to acetyl-CoA, each organelle has distinct metabolic pathways for its synthesis (Fatland et al., 2002, 2005). Acetyl-CoA is a key metabolite in catabolic and anabolic processes, including the tricarboxylic acid cycle, fatty acid biosynthesis and fatty acid β-oxidation, and it is also an intermediate for the biosynthesis of a wide range of secondary metabolites including isoprenoids, polyketides, certain amino acids and sterols (Fatland et al., 2005). The flux of acetyl-CoA is particularly high in plastids, where it is mainly used in de novo fatty acid biosynthesis for membrane lipids in leaf chloroplasts and for storage lipids in plastids of seeds. Production of PHB in these organelles has provided the most successful results to date. For every two molecules of acetyl-CoA that are converted to acetoacetyl-CoA in the first reaction of the pathway, a Claisen condensation catalysed by β-ketothiolase (Davis et al., 1987; Masamune et al., 1989), one reducing equivalent in the form of NAD(P)H, is required for the subsequent conversion of acetoacetyl-CoA to monomer R-3-hydroxybutyryl-CoA by a reductase (Ploux et al., 1988). The PHB synthase incorporates the monomer into the growing polymer chain in the third and final step (Gerngross et al., 1994). As in bacteria, PHA polymer chains have been shown to accumulate within plant cells as discrete granules (Figure 3). PHAs in general cannot be reutilized by the plant and thus represent a terminal carbon sink consuming carbon and reducing equivalents. The choice of organelle and strategy for engineering the PHB pathway therefore not only relies on the presence of acetyl-CoA in the targeted production site, but also on the potential effects of diverting acetyl-CoA and NAD(P)H away from the natural plant metabolism, a process that may result in unwanted deleterious effects on the plant.

Figure 3.

Transmission electron micrographs of leaf tissue from PHB-producing switchgrass and tobacco. Abbreviations are as follows: G, granule of PHB; PL, plastoglobule; G & PL, granule of PHB co-localized with plastoglobule; CW, cell wall; M, mitochondria.

Researchers have also succeeded in producing co-polymers of short-chain-length PHAs (PHASCL, 3–5 carbons) and medium-chain-length PHAs (PHAMCL, 6–14 carbons) by diverting carbon away from other pathways. However, due to the significance of the PHB opportunity (Figure 1) and the progress made with the production of this polymer in plants, we will focus this review on PHB. The production of other PHAs in plants has been extensively described in other reviews (Poirier and Brumbley, 2010; Poirier and Gruys, 2002; Snell and Peoples, 2002; Suriyamongkol et al., 2007), and the reader is directed to these articles as well as the comprehensive summary in Tables 1-3 for detailed information.

PHA production within biomass of plants with C3 photosynthesis

PHB synthesis in cytosol

The first demonstration of PHB production in plants was performed in Arabidopsis engineered to express genes encoding a reductase and a PHA synthase from Alcaligenes eutrophus (now known as Cupriavidus necator) from the 35S promoter from cauliflower mosaic virus (CaMV 35S; Poirier et al., 1992). A β-ketothiolase was not included as thiolase activity is known to be present in the cytosol of plants. Low levels of polymer [0.01% fresh weight (FW)] were detected, and transmission electron microscopy (TEM) analysis demonstrated that granules were located not only in the cytosol, but also in the nucleus and vacuole. Severe stunting of plant growth was also observed. Since these original studies, several other C3 plants have been engineered to produce PHB in the cytosol with little success. Most efforts have yielded trace amounts of polymer (Table 2) often with deleterious effects on the host plant. To date, the highest levels produced in the cytosol of vascular plants have been observed in Arabidopsis engineered with a mutant highly active PHA synthase [0.61% dry weight (DW) (Matsumoto et al., 2005)], suggesting that significant additional metabolic engineering beyond expression of the base PHB biosynthetic pathway is needed to increase polymer production. High levels (10.6% DW) have, however, been produced in the cytosol of the microalgae Phaeodactylum tricornutu, a diatom capable of depositing lipids in its cytosol, which could be indicative of a large acetyl-CoA pool (Hempel et al., 2011). In contrast, engineering of Chlamydomonas reinhardtii for cytosolic PHB production yielded only 6 × 10−4% DW PHB (Chaogang et al., 2010).

PHB synthesis in chloroplasts

The first efforts to produce PHB in chloroplasts employed Agrobacterium-mediated transformation of Arabidopsis separately with three individual genetic constructs, each containing an expression cassette for either thiolase, reductase or synthase from Alcaligenes eutrophus behind the CaMV 35S promoter. Each PHB pathway gene coding sequence was modified at the 5′ end with a DNA fragment encoding the transit peptide from the small subunit of Rubisco from pea (RbcS-TP) to target the encoded enzymes to the plastids (Nawrath et al., 1994). Subsequent crosses between transgenic plants yielded lines that expressed all three PHB enzymes. This work resulted in significantly elevated levels of PHB (14% DW) compared with cytosolic PHB production. The PHB content in older leaves was found to be higher than in younger leaves, and the transformants grew normally. TEM analysis demonstrated that the polymer granules were indeed in the chloroplasts but not in any other organelles.

Subsequent efforts for PHB production in plastids utilized multi-gene constructs containing the three PHB genes in individual expression cassettes. This allowed Agrobacterium-mediated delivery of the entire pathway into Arabidopsis in one step without the need for subsequent crossing. In one study, the multi-gene expression vector contained PHB pathway genes from Ralstonia eutropha (now known as Cupriavidus necator), modified at the 5′ end with DNA encoding a plastid targeting sequence (Valentin et al., 1999). Each expression cassette also contained the e35S promoter. Homozygous lines obtained upon transformation of the multi-gene constructs and subsequent rounds of self-pollination yielded plants with a maximum PHB content of 12%–13% DW (Valentin et al., 1999). Experiments by other researchers generated Arabidopsis lines producing up to 4% FW PHB (approximately 40% DW; Bohmert et al., 2000) using a multi-gene vector. This is the highest level of PHB reported to date in plants and was accompanied by extremely stunted growth, severe chlorosis and a loss of plant fertility. The multi-gene construct contained expression cassettes for PHB pathway genes from R. eutropha modified with the RbcS-TP from pea and controlled by the CaMV 35S promoter.

Although there were no reported growth or fertility issues in the first two publications describing plastid-targeted PHB production in Arabidopsis (Nawrath et al., 1994; Valentin et al., 1999) and only limited chlorosis was observed (Nawrath et al., 1994), subsequent work described detrimental effects on plant phenotype at levels above approximately 3% DW (Bohmert et al., 2000, 2002, 2004). Expression of the PHB pathway from inducible promoters has been used as a strategy to reduce the effects of high PHB levels on plants by allowing an initial period of normal plant growth prior to induction of polymer synthesis. In one study, the gene encoding the R. eutropha thiolase, modified with a sequence encoding the RbcS-TP from pea, was placed under the control of the salicylic acid-inducible prp-1 promoter from potato. Genes encoding R. eutropha reductase and synthase, modified with a sequence encoding RbcS-TP, were driven by the constitutive CaMV 35S promoter (Bohmert et al., 2002). The thiolase gene was chosen for inducible control as previous results demonstrated very low transformation efficiency of a thiolase only construct, especially in potato where no transgenic plants could be obtained (Bohmert et al., 2002). Use of this inducible construct allowed successful transformation of potato as well as an increased transformation efficiency in tobacco; however, polymer yields in both plants were still extremely low (0.009% DW in potato, 0.04% DW in tobacco). Transformation of Arabidopsis with this same construct yielded up to 1.16% FW (13.2% DW) PHB; however, the health of plants accumulating more than 0.6% FW was compromised (Bohmert et al., 2002).

Arabidopsis plants have also been transformed with a multi-gene construct in which expression cassettes for the PHA genes are under the control of an ecdysone inducible promoter (Kourtz et al., 2007). These promoters can be induced by ecdysone agonists such as the commercially available insecticides Mimic® and Intrepid® that are approved for field use on some crops. The transformation construct contained a gene encoding a hybrid Pseudomonas oleovorans/Zoogloea ramigera synthase, as well as genes encoding thiolase and reductase from R. eutropha, all modified with the RbcS-TP. Prior to induction, the best performing plants possessed essentially normal developing leaves accumulating a low level of polymer. After induction, some plants produced chlorotic stunted leaves where others produced more normal size leaves. These new leaves contained more polymer than the ones formed prior to induction. Screening of lines for the best controlled induction, polymer production and plant phenotype was still found to be important in these experiments. The best plant produced 14% DW in a young leaf and 7% DW in an older leaf and was a fairly normal-sized plant (Kourtz et al., 2007). These experiments demonstrate the potential power of an inducible system to enable high level of product formation without sacrificing biomass yield.

Other efforts to produce PHB in plastids have included expression of the transgenes from the plastome of tobacco, the plant which is most readily engineered by plastid transformation. Expression of transgenes from the plastome can provide high levels of protein accumulation, which can be controlled to some extent by the choice of 5′ and 3′ untranslated regions (UTRs) flanking transgenes (Eibl et al., 1999). Early studies using this method yielded low levels of polymer and even a loss of fertility with constructs containing slightly modified bacterial operons with minimal optimization of genetic elements for expression from the plastome (Arai et al., 2004; Lössl et al., 2005, 2003; Nakashita et al., 2001a). The highest levels produced were 1.7% DW in plantlets in tissue culture. This value dropped to an average of 0.002% DW (reported as 20 ppm) upon further growth, and plants were found to be sterile (Lössl et al., 2003). More recently, work to engineer a synthetic PHB operon specifically designed for optimal expression of transgenes from the plastome and increased stability of the insert was performed (Bohmert-Tatarev et al., 2011). Transgenes encoding the thiolase and synthase from Acinetobacter sp., and the reductase from Bacillus megaterium, were flanked with UTRs known to yield high levels of transgene expression, and care was taken to minimize the use of sequences with homology to the host plastome as this is known to provide unwanted recombination events. Plants with up to 18.8% DW PHB in leaf samples and 8.8% DW in a whole plant were obtained (Bohmert-Tatarev et al., 2011), levels that are approaching commercial targets (7.5%–15% DW; Kourtz et al., 2007) for whole-plant PHB production.

PHA synthesis in peroxisomes

Peroxisomal β-oxidation pathways are most active and thus potentially capable of supplying the most substrate for polymer synthesis in young seedlings, where fatty acids are degraded to support plant growth, and senescing leaves, where fatty acid-containing molecules such as membrane lipids are degraded to remobilize carbon and other nutrients. It is challenging to obtain high yields of polymer if peak product synthesis is confined to these plant developmental stages. Early work to produce PHAs in peroxisomes focused on the production of PHAMCL by expressing a PHA synthase with substrate specificity for medium-chain-length monomers (PhaCMCL). The results from these and other studies have been extensively described in other excellent reviews (Poirier, 2001, 2002), and the reader is directed to these articles as well as the summary in Table 2 for information. More recently, Arabidopsis has been engineered for peroxisomal PHB synthesis by expressing genes encoding the thiolase, reductase and synthase from R. eutropha from the CaMV 35S promoter using either multiple single-gene constructs or one multi-gene construct (Tilbrook et al., 2011). A control vector containing phaC was also prepared. The PHB genes in these vectors were fused at the C-terminus to peroxisomal targeting sequences. Up to 1.8% DW PHB was produced in 10-day-old dark-grown etiolated seedlings expressing all three transgenes. Plants engineered to express only PhaC produced less PHB, suggesting that available substrate pools contained more acetyl-CoA and/or acetoacetyl-CoA than R-3-hydroxybutyryl-CoA. While no reduction in seed germination or plant establishment was detected, reduced hypocotyl elongation was observed in polymer-producing seedlings (Tilbrook et al., 2011).

PHA production in crops with C4 photosynthesis

The superior biomass yield of C4 plants makes them ideal targets for the production of PHAs, especially non-food crops that can grow well on marginal land that is not considered prime food production acreage. C4 photosynthesis occurs mainly in the grass family, and economically important C4 grasses include sugarcane, maize, sorghum, switchgrass, Miscanthus, energy cane, sweet sorghum and pearl millet. C4 plants have a competitive advantage over plants possessing the more common C3 carbon fixation pathway under conditions promoting photorespiration due to the mechanism of CO2 concentration around Rubisco allowing the enzyme to reach its maximal catalytic activity. In most C4 plants, concentration of CO2 is achieved by the spatial separation of its fixation (C4 cycle) and reduction (Calvin cycle) in two highly specialized leaf cell types, mesophyll and bundle sheath. In differentiated leaves, these cells are arranged around the vascular bundles in a wreath-like structure (Kranz anatomy) and are connected to each other through plasmodesmata (Figure 4). Capture of CO2 occurs in the outer ring of mesophyll cells via carboxylation of the three-carbon molecule phosphoenolpyruvate to produce oxaloacetate. The fixed carbon is transferred in the form of a four-carbon molecule to the bundle sheath cells where CO2 is released by decarboxylase enzymes. As a result, Rubisco's oxygenase activity leading to photorespiration is strongly suppressed by the elevated CO2 concentrations, which are estimated to be three to eight times higher than those in C3 photosynthetic cells (Kanai and Edwards, 1999). In addition to the increased photosynthetic rate (up to tenfold higher CO2 assimilation rates than the most productive C3 plants), C4 plants use water and nitrogen more efficiently than C3 species (Ehleringer and Monson, 1993).

Figure 4.

Structural differences between C3 and C4 leaves. Light microscopy of a transverse leaf section from tobacco (a) and switchgrass (b). (c) TEM image of a switchgrass leaf showing mesophyll and bundle sheath cells. Abbreviations are as follows: S, starch, CW, cell wall.

The complex compartmentalization of C4 photosynthesis in bundle sheath and mesophyll cells, however, creates challenges for uniform accumulation of an engineered bioproduct such as PHAs. It is also difficult to compare and predict product yields in plants belonging to the different subtypes of C4 photosynthesis [i.e. NADP-dependent malic enzyme (NADP-ME), NAD-dependent malic enzyme (NAD-ME) and phosphoenolpyruvate carboxykinase (PCK)] with biochemically distinct steps for release of CO2 in the bundle sheath cells (Figure 5). To date, work to engineer C4 plants for PHA production has included maize (Zea mays L.) and sugarcane (Saccharum spp. hybrids), which both possess the NADP-ME subtype of C4 photosynthesis, and switchgrass (Panicum virgatum L.), an NAD-ME plant. An effort will be made to describe experiments performed with these crops in detail including some of the major challenges that researchers face.

Figure 5.

Biochemical pathways and their compartmentalization for (a) NADP-ME and (b) NAD-ME subtypes of C4 photosynthesis. Abbreviations are as follows. PEP, phosphoenolpyruvate; OAA, oxaloacetate; MAL, malate; PYR, pyruvate; RuBP, ribulose-1,5-bisphosphate; 3-PGA, 3-phosphoglyceric acid; ASP, aspartate; α-KG, α-ketoglutarate; GLU, glutamate; ALA, alanine. Select enzyme activities are as follows: 1, carbonic anhydrase; 2, PEP carboxylase; 3, NADP-malic enzyme, 4, Rubisco; 5, NAD-malic enzyme; 6, pyruvate dehydrogenase. The reader is directed to Kanai and Edwards, 1999 for description of the PCK subtype of C4 photosynthesis.


Engineering of PHB production in the stover of maize was the first demonstration of bioplastic production in a C4 crop. Genes encoding the PHB enzymes from R. eutropha were assembled in multi-gene transformation vectors, and different viral and plant promoters were used to drive their expression (Mitsky et al., 2003; Poirier and Gruys, 2002). Promoters tested included the CaMV 35S promoter, figwort mosaic virus promoter (FMV), rice actin promoter (rACT) and the promoter of the maize chlorophyll a/b-binding protein gene (cab-m5). All promoter sequences were fused to the maize hsp70 intron for enhanced transgene expression. The encoded enzymes were targeted to the chloroplasts using the RbcS-TP from Arabidopsis, and polymer levels up to 5.73% DW were detected. Strong preferential accumulation of PHB granules in the chloroplasts of the bundle sheath cells surrounding the vascular bundles was observed with little formation in the mesophyll chloroplasts. Based on these findings, the authors suggested that the substrate for PHB biosynthesis, acetyl-CoA, may be present at different levels in the plastids of various cell types. As with other plants [i.e. Arabidopsis (Nawrath et al., 1994), sugarcane (Purnell et al., 2007), switchgrass (Somleva et al., 2008) and tobacco (Bohmert-Tatarev et al., 2011)], in general, higher levels of polymer were produced in older leaves compared with younger leaves. A correlation between higher levels of PHB accumulation and leaf chlorosis was also observed (Poirier and Gruys, 2002).


PHB production in chloroplasts

The first work to engineer sugarcane for the production of PHB employed multiple single-gene vectors to express the genes from R. eutropha encoding PHB biosynthetic enzymes and the selectable marker nptII (Petrasovits et al., 2007). The strong constitutive maize polyubiquitin promoter (ubi1) was used to drive the transgene expression, and the encoded enzymes were targeted to the plastids using the RbcS-TP from pea. The single-gene vectors were introduced simultaneously into sugarcane callus cultures by particle bombardment. Approximately 20% of the 130 transgenic plants screened produced polymer at levels detectable by HPLC, and the highest PHB content measured was 1.88% DW. Significantly lower PHB content (up to 0.01% DW) was detected in stems.

PHB granules were present in plastids of bundle sheath cells of sugarcane leaves but were not visible in mesophyll plastids as shown by Nile blue staining and fluorescence microscopy and TEM analysis (Petrasovits et al., 2007). As the maize ubi1 promoter has previously been shown to drive the expression of transgenes in the mesophyll cells of sugarcane and maize, the authors suggested that unequal product distribution could be due to inefficient transport of enzymes to plastids of mesophyll cells with a C3 dicot chloroplast targeting signal or an inability to efficiently produce PHB in cells involved with the light reactions of photosynthesis (Petrasovits et al., 2007).

More detailed analyses of the spatiotemporal distribution of PHB production were performed with six independent transgenic sugarcane lines grown under greenhouse conditions (Purnell et al., 2007). In these studies, it was found that stalk height and weight, total aerial biomass and culm internode sugar content were not affected by polymer accumulation at levels up to 1.77% DW, the highest levels measured in the analysed lines. As there was low PHB production in stem tissue, the maximum whole-plant PHB yield was 0.79% DW. Although the integration pattern and copy number of the transgenes were not studied, good correlations between PHB production, the amount of transgene transcripts and the accumulation of PHB pathway enzymes were detected by molecular analyses. In addition to previously described differences in polymer formation with leaf age, a gradient of PHB content along an individual leaf was observed with polymer accumulation significantly increasing from the base of the leaf (the youngest part) to the tip (the oldest part; Purnell et al., 2007).

Efforts to improve PHB production in plastids of sugarcane were made by testing constructs in which the expression of the PHB biosynthetic genes was driven by different promoters (Petrasovits et al., 2012), including the maize (ubi1) and rice (rubi2) polyubiquitin promoters, the maize cab-m5 promoter and the Cavendish banana streak badnavirus promoter (Cv). For the ubi1 and Cv promoters, multiple single-gene vectors containing genes encoding the PHB enzymes from R. eutropha fused to the RbcS-TP from pea were introduced into embryogenic callus cultures by biolistics. For the rubi2 and cab-m5 promoters, single multi-gene constructs containing expression cassettes for the entire PHB pathway that had been previously tested in switchgrass (Somleva et al., 2008) were used. These constructs contained a gene encoding a hybrid P. oleovorans/Z. ramigera synthase, as well as genes encoding thiolase and reductase from R. eutropha, all modified with the RbcS-TP. In tissue culture, the highest levels of polymer were produced in plantlets when the Cv promoter was used. However, the activity of this promoter significantly decreased in mature tissues of soil-grown plants in all transgenic lines analysed resulting in reduced PHB production. Interestingly, the constitutive ubiquitin promoters from rice and maize provided similar levels and spatiotemporal patterns of polymer production in plants, with a modest increase in PHB content between 3- and 6-month growth in soil. In contrast, some of the sugarcane plants transformed with the cab-m5 promoter construct continued to accumulate polymer after transferred to soil reaching up to 4.86% DW PHB in mature leaves (Petrasovits et al., 2012), approximately 2.5 times higher than previously reported levels in sugarcane (Petrasovits et al., 2007). An analysis of PHB production in four different leaf types chosen based on their position along the stalk and/or morphology revealed that with all promoters tested, PHB content was higher in mature leaves than in young ones. The youngest most rapidly growing tissues had similar polymer content regardless of the promoter used, suggesting that PHB production in these plant parts is metabolically limited. In lines producing ≥2% DW PHB, formation of polymer granules was observed in all photosynthetic cell types, including mesophyll cells; however, the highest amount of PHB was still found to accumulate in the bundle sheath chloroplasts. Reduced biomass production and slight leaf chlorosis were observed in the highest PHB-producing lines (Petrasovits et al., 2012).

PHB production in peroxisomes

PHB has been successfully produced in the peroxisomes of sugarcane leaves with multiple single-gene constructs, or one multi-gene construct. Genes encoding the PHB pathway enzymes from R. eutropha were fused to a sequence encoding a peroxisomal targeting signal, and their expression was controlled by the maize ubi1 promoter (Tilbrook et al., 2011). Vectors were delivered into sugarcane callus using particle bombardment. Levels between 0.4% and 1.6% DW were produced in the oldest leaves of mature sugarcane plants, whereas no detectable PHB was observed in stalks. In contrast to plastid-targeted PHB production in sugarcane, where granules were predominantly observed in chloroplasts of bundle sheath cells (Petrasovits et al., 2007; Purnell et al., 2007), peroxisomal-targeted production resulted in formation of granules in peroxisomes of both mesophyll and bundle sheath cells. In addition, polymer was also observed in vacuoles, possibly due to pexophagy, a process of autophagic degradation of peroxisomes in the vacuole. A low amount of 3-hydroxyvalerate (1.9%–3.7% of the total polymer) was also found to be incorporated into the polymer.


Expression of the base PHB pathway

The first demonstration of PHB production in a C4 plant with the NAD-ME photosynthetic pathway was performed in switchgrass (Somleva et al., 2008), a high-yielding biomass crop that is receiving considerable attention as a bioenergy feedstock (Monti, 2012). Genes encoding the thiolase and reductase from R. eutropha and a gene encoding a hybrid P. oleovorans/Z. ramigera synthase were modified with the RbcS-TP from pea and assembled in one multi-gene transformation construct that contained an expression cassette for the marker gene bar. Either the rice rubi2 or the maize cab-m5 promoter, fused to the maize hsp70 intron, was used to drive the expression of the PHB genes. Vectors were introduced into highly embryogenic mature caryopsis-derived callus cultures by Agrobacterium-mediated transformation (Somleva, 2006; Somleva et al., 2002). Polymer production was monitored in more than 400 primary transformants in tissue culture, and the majority of the transgenic switchgrass plants (85% and 71.5% of the PCR-positive plants transformed with the cab-m5 and rubi2 constructs, respectively) produced detectable amounts of PHB in the range of 0.01%–1.82% DW. After 2 months of growth under greenhouse conditions, the polymer content in mature leaves of some cab-m5 lines was two to five times higher than in tissue culture with the highest plant producing up to 2.56% DW PHB. The majority of the primary transformants analysed contained 1–2 copies of the transgenes; however, some of the highest PHB producers identified had rather complex integration patterns and multiple copies (3–5) of the T-DNA inserts.

Detailed analyses of PHB distribution in leaf and stem tissues from vegetative and reproductive tillers demonstrated the presence of a gradient of polymer accumulation along a tiller and an individual leaf. As with maize and sugarcane (Petrasovits et al., 2007; Poirier and Gruys, 2002), in general the mature tissues in the base of the tillers and the tip of the leaves contained higher levels of PHB than the younger leaf and stem parts. Low polymer levels were measured in stem tissues, including leaf sheaths (up to 0.27%), nodes and internodes (0%–0.1%), and panicles (up to 0.17%). Tillers at a reproductive stage consistently produced more polymer than vegetative tillers. Data for the average PHB content in lines generated from transformations with the two multi-gene vectors suggested that the cab-m5 promoter maintained higher levels of transgene expression than the rubi2 promoter at later stages of plant growth and development. Primary switchgrass transformants producing up to 3.72% DW PHB in leaf tips and 1.23% DW in whole tillers were identified (Somleva et al., 2008). At the cellular level, the highest accumulation of PHB granules was observed in the chloroplasts of bundle sheath cells similar to maize (Poirier and Gruys, 2002) and sugarcane (Petrasovits et al., 2007). PHB production was also analysed in the T1 generation obtained from controlled crosses between transgenic switchgrass plants accumulating polymer at different levels. This is the first work demonstrating the transmission of the pha genes through both female and male gametes to the next generation in PHB-producing C4 monocots. T1 plants containing three- to six-fold more polymer (up to 4.19% DW in mature leaves) than the parent lines were obtained indicating that stable, high-level polymer production could be maintained in the next generation (Somleva et al., 2008). These results also demonstrate the potential for improvement of PHB production in switchgrass by conventional plant breeding.

Polymer levels have also been measured after in vitro propagation of PHB-producing switchgrass plants through immature inflorescence-derived cultures or nodal segments (Somleva and Ali, 2010). A highly efficient procedure for large-scale in vitro propagation produced more than 42 000 plants within 3–6 months after initiation of cultures from previously characterized primary transformants (Somleva et al., 2008). All of the in vitro propagated plants obtained were transgenic, and most of them accumulated polymer at levels higher than the donor lines (Figure 6). In soil, some of the micropropagated plants produced up to 6.09% DW PHB (Somleva and Ali, 2010). Although all of the propagated transgenic switchgrass plants appeared phenotypically normal and uniform in their growth under in vitro and greenhouse conditions, the possibility of somaclonal variation cannot be excluded. In fact, some adaptation to higher PHB levels in the in vitro cultures may have occurred as callus from a T0 donor plant that produced 2.56% DW PHB in soil and was slightly chlorotic with reduced growth yielded plants that accumulated 5%–6% DW PHB (Fig. 6b) and had normal development and growth. Induced epigenetic and/or genetic changes could be a valuable tool in developing plants with special characteristics by the methods of conventional plant breeding (Seliskar and Gallagher, 2000). In the switchgrass experiments (Somleva and Ali, 2010), somaclonal changes that may have occurred in the cultured tissues appear to have favoured the increased PHB production and improved phenotype of the in vitro propagated plants.

Figure 6.

PHB production in switchgrass plants propagated through immature inflorescence-derived callus cultures. (a) Polymer levels in leaves of plants in tissue culture. (b) PHB content in randomly selected soil-grown plants. T0, the primary transformant used for callus initiation; 1–79, individual propagated plants.

Increased carbon flow for PHB production

In efforts to increase polymer production beyond the levels achieved by expression of the base PHB pathway, a gene from Synechococcus encoding a bifunctional enzyme with both fructose-1,6-bisphosphatase (FBPase) and sedoheptulose-1,7-bisphosphatase (SBPase) activities (Tamoi et al., 1996) was introduced into PHB-producing switchgrass lines by re-transformation (Somleva et al., 2012). FBPase and SBPase are enzymes in the Calvin cycle and have been suggested to influence sugar partitioning and carbon fixation rate, respectively (Raines, 2003; Serrato et al., 2009; Tamoi et al., 2006). Previous studies have demonstrated that expression of genes encoding FBPase/SBPase or SBPase in tobacco enhances photosynthesis and plant growth (Lefebvre et al., 2005; Miyagawa et al., 2001; Tamoi et al., 2006). For re-transformation of PHB-producing switchgrass, immature inflorescence-derived callus cultures were initiated from previously characterized transgenic lines (Somleva et al., 2008). A vector containing the fbpase/sbpase gene (modified with the RbcS-TP from pea) under the control of the cab-m5 promoter, and the marker gene hptII (conferring resistance to hygromycin), was introduced into these cultures by Agrobacterium-mediated transformation. In tissue culture, polymer content was measured in more than 220 re-transformed plants and 100 control plants (regenerated from the initiated cultures without re-transformation) prior to transfer to soil, and no significant differences in polymer levels were detected. However, differences in PHB production between both sets of plants were observed after growth in a greenhouse for 2 months. The highest PHB levels detected in plants expressing the fbpase/sbpase gene were 7.69% and 5.24% DW in mature and developing leaves, respectively, values significantly higher than the 3.53% and 2.55% DW PHB obtained in mature and developing leaves, respectively, from plants containing only the PHB genes (Table 4; Somleva et al., 2012).

Table 4. PHB Production in switchgrass plants re-transformed with a gene encoding an FBPase/SBPase
TransgenesNo. of plants analysedPHB content (% DW)a
Mature leavesDeveloping leaves
  1. Data originally described in Somleva et al. (2012).

  2. a

    Polymer content was measured in mature and developing leaves of vegetative tillers from plants grown under greenhouse conditions for 2 months.

PHB genes 150.14–3.531.441.060.18–2.551.030.80
PHB genes + fbpase/sbpase 610.42–7.693.481.520.00–5.241.840.90

Challenges for commercialization

Despite the significant progress in engineering PHB production in plant biomass, increased product yield (>10% DW PHB) and reduced agronomic penalties are required for successful commercialization of a polymer-producing crop. To achieve these goals in a timely manner, the throughput of transformation experiments needs to be increased. The scope of these challenges is discussed in detail below.

Increased product yield

The highest levels of PHB accumulation in plant biomass have been achieved with plastid-targeted PHB production in Arabidopsis (approximately 40% DW; Bohmert et al., 2000) and expression of PHB genes directly from the plastome in tobacco (18% DW leaves and 8.8% whole plants; Bohmert-Tatarev et al., 2011). Arabidopsis has no agronomic value, and the costs of growing tobacco for the production of commodity items such as fuels and/or plastics are not economic. Duplication of these results in high-yielding C4 crops is therefore needed to create plants for commercial production. As robust procedures for plastid transformation of monocots are currently not available (Maliga, 2012), near-term experiments will require strategies for nuclear-encoded expression of additional transgenes to increase PHB levels and improve plant fitness. Surprisingly, despite the large number of prior efforts to produce PHB in plant biomass (Tables 1-3), the majority of studies have focused on engineering only the base bacterial pathways for polymer biosynthesis. A notable exception is the co-expression of fbpase/sbpase in switchgrass (Somleva et al., 2012). The recently developed genome-scale model for C4 plants (Gomes de Oliveira Dal'Molin et al., 2010) will allow a rational guided approach to future engineering efforts to increase carbon flow to polymer.

Reduced agronomic penalties of PHB production

PHB accumulation in plant tissues above certain levels causes undesired changes in plant development and metabolism, irrespective of whether the plant is a monocot or dicot or possesses a C3 or C4 photosynthetic pathway. Some of the phenotypic changes associated with polymer synthesis in green tissue are decreased biomass yield, chlorophyll deficiency and reduced fertility. The levels of PHB that trigger these changes and their severity depend on the plant species and cellular compartment(s) where the polymer is produced, as well as the tissue specificity and timing of transgene expression as exemplified by the extensive research with Arabidopsis. Although the causes of reduced plant fitness in high PHB producers are not known, there are several structural and metabolic factors that should be considered.

PHAs are osmotically and metabolically inert and thus theoretically could be stored without affecting the host plant; however, their presence in the form of a hydrophobic granule could be problematic. In native PHB producers, polymer granules are primarily coated with proteins called phasins that are thought to provide a barrier between the hydrophobic granule and the bacterial cytoplasm and prevent non-specific binding of other proteins (Jendrossek and Handrick, 2002; Neumann et al., 2008). Multiple studies have shown that proteins other than phasins can bind granules formed in vitro or in vivo in engineered bacterial strains that are not naturally PHB producers (reviewed in Grage et al., 2009); however, similar studies have not been performed with plants. Efforts have been made to co-express genes encoding plastid-targeted PHB enzymes and phasins in Arabidopsis; however, no improvement in plant phenotype was observed (Bohmert et al., 2002).

Transmission electron micrographs (TEMs) of switchgrass and tobacco that produce PHB in their chloroplasts have shown that granule formation is often co-localized with plastoglobules, and in some cases, the reduction in plastoglobule matter is visible (Figure 3). Plastoglobules are associated with thylakoids in chloroplasts, and while their role is still under investigation, they are believed to be involved in chloroplast development, senescence and stress response (Lundquist et al., 2012). Reduced grana stacking is apparent in many published TEMs (Bohmert et al., 2000; Nawrath et al., 1994; Petrasovits et al., 2007; Saruul et al., 2002; Somleva et al., 2008). While the cause of this phenomenon is unknown, it could be a result of association of PHB granules with plastoglobules or, alternatively, may be reflective of reduced synthesis of the components of the grana. Whatever the cause, these structural differences likely impact the efficiency of photosynthesis and thus plant health.

Some metabolic changes have been observed in PHB-producing plants, including differences in the levels of various organic acids, amino acids, sugars and sugar alcohols in Arabidopsis (Bohmert et al., 2000), reduced starch in switchgrass (Somleva et al., 2012) and decreased chlorophyll fluorescence in poplar (Dalton et al., 2011). Reduced expression of genes encoding light-harvesting chlorophyll a/b-binding proteins at the mRNA level in Arabidopsis (Choi, 2009) and protein level in switchgrass (Somleva, unpublished results) have also been detected. As PHB synthesis consumes reducing equivalents, changes in the plant's NADPH/NADH ratio are possible. In the chloroplasts, a shift in redox balance could affect the activities of multiple enzymes, including those involved in the Calvin cycle, the function of reductive anabolic processes such as fatty acid biosynthesis and the synthesis of PHB itself. Due to the complex interactions between the organelles in the plant cell, it is difficult to predict the overall metabolic consequences of PHB production.

The complexity of PHB biosynthesis and its effects on plant metabolism, growth and development are even greater in C4 plants because of the increased metabolic fluxes across the chloroplast membranes due to the high CO2 assimilation rates (Brautigam et al., 2008) and the constant exchange of metabolites between bundle sheath and mesophyll cells. In addition, consumption of acetyl-CoA, a molecule whose main precursor is pyruvate, for PHB synthesis may impair the C4 photosynthetic cycle itself (Figure 5). Although there is not enough experimental evidence to predict advantages of one C4 subtype over another for PHB production, some variations in the spatial distribution and rate of photosynthesis-related processes (Table 5; Byrt et al., 2011; Furbank, 2011) may be important. Cell-specific energy requirements are based on the difference in the decarboxylation mechanisms and transported metabolites. In NADP-ME C4 plants, malate, synthesized in the mesophyll cells, carries both the reducing equivalents (NADPH) and CO2 to bundle sheath chloroplasts. As NADP-ME plants such as maize, sugarcane and sorghum lack functional photosystem II (PSII) in differentiated bundle sheath chloroplasts, malate decarboxylation along with the triosephosphate shuttle can account for the reducing power required for carbon reduction in these organelles (Lasik and Edwards, 2009). In NAD-ME C4 plants, the mesophyll to bundle sheath transport of aspartate does not involve transfer of redox equivalents consistent with the presence of comparable photosystem I (PSI) and PSII activities in both types of photosynthetic cells.

Table 5. Major functional differences between NAD-ME and NADP-ME subtypes of C4 photosynthesisa
Metabolic processNAD-ME C4 plantsNADP-ME C4 plantsReference
  1. BS, bundle sheath; M, mesophyll; NDH, NAD(P)H dehydrogenase.

  2. a

    As no major crops with PCK C4 photosynthesis have been identified, this subtype is not included.

Active photosystems in BS thylakoidsPSI and PSIIPSI (PSII only in young leaves)Furbank (2011) and Romanowska and Drozak (2006)
Increased demands for ATPIn M chloroplasts (some deficiency in grana)In BS chloroplasts (deficient in grana)Kiirats et al. (2010)
Cyclic photophosphorylationIn M chloroplastsIn BS chloroplastsKiirats et al. (2010)
NDH expression (cyclic electron flow)High in M cellsHigh in BS cellsTakabayashi et al. (2005)
NADPH production in mature leavesIn BS and M chloroplastsIn M chloroplasts onlyIvanov et al. (2005)

As a consequence of the differences in the activities of PSI and PSII, the distribution of photosynthesis-related antioxidant compounds and enzymes between the bundle sheath and mesophyll cells is likely to be different in NAD-ME and NADP-ME C4 plants. In maize and probably in other NADP-ME plants, superoxide dismutase and ascorbate peroxidase are concentrated in the bundle sheath cells, while the antioxidants dehydroascorbate and oxidized glutathione (GSSG) have to be transported to the mesophyll for re-reduction, as glutathione reductase and dehydroascorbate reductase are localized only in the mesophyll cells (Kingston-Smith and Foyer, 2000). As a result, bundle sheath proteins are more sensitive to accumulation of reactive oxygen species. Any constraint on transport between the compartments under stress conditions could lead to oxidative damage of bundle sheath proteins, including the enzymes of the Calvin cycle. Although there are no reports on the distribution of the components of the antioxidant system in NAD-ME plants, the presence of active PSI and PSII in both the bundle sheath and mesophyll cells suggests similar function of the antioxidant network in the two types of photosynthetic cells.

In summary, the metabolic and photosynthetic changes detected in plants engineered for the production of PHB in the plastids could affect the substrate levels, redox balance, enzyme activities and metabolite (re)distribution in different plant cell organelles. In C4 plants, these changes could have negative effects on the interactions between the bundle sheath and mesophyll cells resulting in disrupted coordination between the C3 and C4 cycles. Because of the dependence of the Calvin cycle on the import of reducing power from the mesophyll cells typical for the NADP-ME subtype of C4 photosynthesis, preserving the integrity of the mesophyll–bundle sheath complex might be more critical in maize and sugarcane than in switchgrass and other NAD-ME plants. The constraints identified in the described studies provide targets for future metabolic engineering to improve product yield and plant agronomic performance. The available genome-scale models should significantly aid these efforts by providing a predictive whole-systems analysis of the engineered plant.

Increased throughput of experiments

One of the major limiting factors for successful genetic transformation of monocotyledonous plants is the low regeneration potential of tissue cultures, resulting in insufficient transformation efficiency, as well as the requirement of significant labour resources to produce and transform the appropriate explants. In addition, the tendency of freshly initiated callus cultures from most C4 grasses to form both embryogenic and organogenic structures (Somleva, personal observations) demands unique skills for identification and selective propagation of the desired type of callus. Despite these challenges, significant progress has been made in establishing reliable procedures for transformation of economically important cereals and grasses, species that have been considered recalcitrant until recently. Although both physical and biological methods facilitate the transfer of genes to the target explants, Agrobacterium-mediated transformation is preferred for stable integration and coordinated expression of multiple transgenes. Examples of the efficiency of this method for transformation of select C4 plants are shown in Table 6. Often the highest transformation efficiency reported for a given crop has only been achieved in a particular cultivar, variety, inbred line, genotype, etc. of this species.

Table 6. Agrobacterium-mediated transformation of C4 monocotyledonous plants
Host plantTarget tissueTE (%)aGene(s) of interestSelectable marker geneReference
  1. AB, axillary buds; EC, embryogenic callus; IE, immature embryos; II, immature inflorescences; II-EC, immature inflorescence-derived embryogenic callus; II-OC, immature inflorescence-derived organogenic callus; MC, mature caryopses; MC-EC, mature caryopsis-derived embryogenic callus; MC-OC, mature caryopsis-derived organogenic callus; MT, meristematic tissue; N, nodes; S, seedlings; S-EC, seedling-derived embryogenic callus; SA, shoot apex; SC, suspension cultures; SE, somatic embryos; TE, transformation efficiency; nd, not determined. Gene abbreviations: aco, 1-aminocyclopropane-1-carboxylate oxidase gene; als, acetolactate synthase gene; aphA, aminoglycoside phosphotransferase gene; bar, phosphinothricin acetyltransferase gene; cp4, glyphosate-insensitive EPSPS (5-enolpyruvylshikimate-3-phosphate synthase) gene from Agrobacterium tumefaciens strain CP4; cre, tyrosine recombinase gene from the P1 bacteriophage; cry1Ah, gene encoding an insecticidal protein; fbpase/sbpase, fructose-1,6-bisphosphatase/sedoheptulose-1,7-bisphosphatase gene; gfp, green fluorescent protein gene; gna, Galanthus nivalis agglutinin gene; gus, β-glucoronidase gene; hph, hpt, hptII, hygromycin phosphotransferase genes; manA, mannose-6-phosphate isomerase gene; npk1, nicotiana protein kinase 1 gene; nptII, gene encoding an aminoglycoside phosphotransferase conferring resistance to antibiotics such as kanamycin, neomycin, paromomycin, butirosin, gentamycin B and geneticin; phaA, β-ketothiolase gene; phaB, reductase gene; phaC, gene encoding PHA synthase; pinII, gene encoding a protease inhibitor protein from Solanum tuberosum; pmi, phosphomannose isomerase gene; ppo, polyphenol oxidase gene; pporRFP, red fluorescent protein gene; sbglr, gene encoding a genomic lysine-rich protein from Solanum tuberosum; si401, Setaria italica ortholog of the maize pollen-specific gene Zm401.

  2. a

    Transformation efficiency determined as the number of antibiotic- or herbicide-resistant plants recovered per explants inoculated.

  3. b

    Transformation efficiency determined as the number of PCR-positive plants recovered per explants inoculated.

MaizeIE0.8–50a,b gus, gfp bar, pmi, nptII, cp4RepAll pinII, NPK1, PPO, cre, als Frame et al. (2002), Gordon-Kamm et al. (2002), Huang et al. (2004), Ishida et al. (2007, 1996), Li et al. (2003), Negrotto et al. (2000), Shou et al. (2004), Yang et al. (2006), Zhang et al. (2003) and Zhao et al. (2002)
S-EC2–11b gfp nptII, cp4 Sidorov et al. (2006)
SugarcaneMT10–35a gus bar Enríquez-Obregón et al. (1998, 1997)
EC0.01–10.5a gus, gfp, gna, aco hptII, bar, aphA, hph Arencibia et al. (1998), Elliott et al. (1998, 1999), Joyce et al. (2010), Wang et al. (2009) and Zhangsun et al. (2007)
AB50a gus bar, nptII Manickavasagam et al. (2004)
MT, ECnd gus bar, hpt Carmona et al. (2005)
SorghumIE0–8.3a gus, gfp bar, hptII, manA, pmi Carvalho et al. (2004), Gao et al. (2005a,b), Gurel et al. (2009), Nguyen et al. (2007) and Zhao et al. (2000)
Sweet sorghumII-EC2.4b cry1Ah bar Zhu et al. (2011)
SwitchgrassSE, MC-EC, II-EC, S-EC0–97.3a gus bar Somleva et al. (2002)
MC-EC0–93b phaA + phaB + phaC bar Somleva et al. (2008)
MC-EC, II-EC, II-OC4.4–98.7a gus, gfp, pporRFP bar, hptII, hph Burris et al. (2009), Li and Qu (2011), Xi et al. (2009) and Xu et al. (2011)
N, II10a gus hptII Wang and Nandakumar (2011)
S0–5.97a gus bar, hpt, nptII Song et al. (2012)
Miscanthus xgiganteus II-EC0–26.9b phaA + phaB + phaC + fbpase/sbpase bar Somleva and Ali (unpublished results)
M. sinensis MC-EC, II-ECnd gus nptII Engler and Chen (2011)
MC-EC2.8b phaA + phaB + phaC bar Somleva and Ali (unpublished results)
Finger milletSA, MC-OC3.8a gfp hptII, nptII Antony Ceasar and Ignacimuthu (2011) and Sharma et al. (2011)
Pearl milletSA5.8b gus hptII Jha et al. (2011)
BermudagrassII-EC, SC, N4.8–6.1b gus hph, bar Hu et al. (2005), Li et al. (2005) and Wang and Ge (2005)
ZoysiagrassN6.8b gus hph Ge et al. (2006)
Setaria italica II-EC5.5–6.6a si401, sbglr hptII Liu et al. (2005), Qin et al. (2008) and Wang et al. (2011)
Setaria viridis MC-OCnd gus hptII Brutnell et al. (2010)

Switchgrass is a good example of the effort that has been necessary to develop and improve the enabling tissue culture and transformation methods for the application of gene technology to a C4 grass. Well-established protocols for callus initiation and plant regeneration from various explants of switchgrass (Alexandrova et al., 1996a,b; Denchev and Conger, 1994, 1995; Dutta Gupta and Conger, 1999) allowed the development of reliable procedures for transformation by biolistics (Richards et al., 2001) and Agrobacterium (Somleva et al., 2002). Switchgrass genotypes capable of producing plants at high frequency (up to 930 plants/g FW callus) were identified using a method for screening of numerous genotypes for their in vitro response (Somleva et al., 2009). A high frequency of Agrobacterium-mediated gene delivery was achieved in transformations of some of these genotypes with multi-gene vectors that contained expression cassettes for the three PHB genes and bar (Somleva et al., 2008). For example, the efficiency of transformation of cultures from the Alamo genotype 56 with a vector containing the PHB biosynthetic pathway was 42%, determined as the number of independent transformation events (confirmed by Southern blot analysis) recovered per explant inoculated. More recently, other highly regenerable switchgrass cultivars and genotypes have been identified including Performer and Colony (Li and Qu, 2011), ST-1 and ST-2 (Xi et al., 2009) and HR8 (Xu et al., 2011). Following the original procedure for Agrobacterium-mediated transformation (Somleva et al., 2002) with slight modifications, mature caryopsis-, node- and immature inflorescence-derived callus cultures from these and other switchgrass lines were successfully transformed with simple reporter–marker gene vectors (Table 6) as well as with vectors containing a gene of interest and a selectable marker for the engineering of different traits. Agrobacterium-mediated transformation of nodes and inflorescences or seedling segments directly without an intervening callus phase has also been reported (Song et al., 2012; Wang and Nandakumar, 2011). Other important tools for genetic modifications of switchgrass at large scale have been developed including the isolation of switchgrass promoters (Mann et al., 2011) and the construction of a set of vectors for transformation of switchgrass and other monocots (Mann et al., 2012). A reliable procedure for cryopreservation of switchgrass cultures (Somleva, unpublished results) and protocols for efficient co-transformation (Somleva, unpublished results) and re-transformation of embryogenic callus (Somleva et al., 2012) allow for the introduction of additional transgenes for stacking engineered traits into transgenic lines.

All of these achievements have contributed to the enormous progress in the development of switchgrass as a platform for the production of biofuels and plastics as demonstrated by engineering of this crop for the production of PHB (Somleva et al., 2008), increased photosynthetic activity for improved polymer production (Somleva et al., 2012), improved sugar release and bioethanol production (Chuck et al., 2011; Fu et al., 2011a,b; Saathoff et al., 2011; Shen et al., 2012), and increased biomass yield (Fu et al., 2012). Despite all of this work, transformation of switchgrass and other C4 crops is still considered labour-intensive, time-consuming and low throughput due to the sheer amount of manual tissue culture work required to initiate, maintain and transform cultures to generate the large numbers of transgenic lines necessary for cost-effective crop improvement. Progress is further reduced by the long growth cycles of many of these plants when analyses of product yield in mature tissues are required. Shortening either of these time periods or development of a reliable model system for C4 plants would have a significant impact on experimental throughput.

The reported basic protocols for tissue culture and Agrobacterium-mediated transformation of the annual diploid C4 plant Setaria viridis (green foxtail) have generated great excitement in the plant community due to Setaria's short life cycle, small size and prolific seed production and its possible use as a C4 model system (Brutnell et al., 2010; Li and Brutnell, 2011) with a known genome sequence (Bennetzen et al., 2012). However, the ease of use and efficiency of the current S. viridis transformation procedure (Brutnell et al., 2010) are still far from what is needed to make this an effective model system due to the low frequency of embryogenic callus formation and poor plant regeneration (Somleva, unpublished results). To accelerate progress in engineering C4 plants, a model system that provides the ease of use of the Arabidopsis floral dip (Clough and Bent, 1998) or a truly high-throughput, low-labour transformation method for the C4 crop of interest is needed.

Summary and future work

Significant progress has been made over the last 20 years in the engineering of PHB-producing plants that can provide low-cost, renewable bioplastics and fuels. While challenges remain to increase product yield and improve plant health, the continued development of cutting-edge plant biotechnology methods that allow predictive modelling of metabolic pathways, accelerate throughput and shorten timelines for generating results will allow the complex experiments required to solve these issues to be successful.

Gene containment systems will also be necessary as work progresses towards field trials. One of the benefits of using a crop such as switchgrass for the co-production of plastics, biofuels and/or chemicals is that it is a native species naturally adapted to North America such that concerns for domestic invasiveness are minimal (Keshwani and Cheng, 2009; McLaughlin and Kszos, 2005; US Department of Energy, 2011). However, as switchgrass is an exclusively outcrossing species, a reliable system for biological containment of the transgenes is crucial for the successful commercialization of genetically modified lines of this crop (Kwit and Stewart, 2012; Wang and Brummer, 2012). While progress has been made to develop systems for gene containment of perennial grasses, there are still serious limitations to most of the available systems (reviewed in Kausch et al., 2010).

PHB-producing crops are an exciting commercial opportunity due to the diversity of products that can be obtained from the polymer (Figure 1). These crops have the potential to significantly decrease petroleum consumption by providing a large-scale source of renewable fuels, plastics and chemicals. They can also have a significant impact on animal and fish feed, supplying higher feed conversion values and prebiotic effects. The remarkable utility of PHB, enabling its use in industrial material and chemical applications as well as feed, is a testament to this molecule's unique properties and commercialization potential.

Conflict of interest

All authors are employees of Metabolix, Inc.