The Synthesis of Starch

Following the uptake of carbon into the amyloplast, starch synthesis proceeds variously via (i) plastidial phosphoglucomutase and plastidial ADPglucose pyrophosphorylase, (ii) only via plastidial ADPglucose or (iii) via no intermediate steps prior to the polymerising reactions of starch synthases and branching enzymes (Smith et al., 1997). The exact route depends on the nature of the imported carbon source (see Figure 25.3). The first reaction of plastidial starch metabolism both in the potato tuber (Tauberger et al., 2000) and in the pea embryo (Hill and Smith, 1991) is the interconversion of glucose-6- and glucose-1-phosphate catalysed by plastidial phosphoglucomutase. Compelling evidence for the involvement of this enzyme in pea starch synthesis was provided by studies on the rug3 mutant which revealed that this locus encodes a plastidial phosphoglucomutase and that mutation at this locus results in a severe depletion of starch levels in pea embryos (Harrisson et al., 1998). In addition the depleted starch accumulation in transgenic potato plants exhibiting reduced levels of phosphoglucomutase (Tauberger et al., 2000) and the reduced gravitropic response of roots of the TC7 mutant of Arabidopsis (also deficient in this enzyme; Kiss et al., 1996) highlight its involvement in starch synthesis in other species. Moreover, transgenic potato plants with a reduction in the cytosolic isoform of phosphoglucomutase also exhibited reduced levels of starch, most probably due to a reduction in the available glucose 6-phosphate for uptake into the plastid (Fernie et al., 2001b).

The next reaction on the path to starch synthesis, which is catalysed by plastidial AGPase, has received much attention for a number of years. This reaction is often considered to be the first committed step of starch synthesis. AGPase utilises ATP and produces pyrophosphate, which is then hydrolysed by pyrophosphatase to yield 2Pi. The hydrolysis of PPi serves to remove the ADPglucose pyrophosphorylase reaction away from equilibrium. A cDNA encoding a soluble inorganic pyrophosphatase has been cloned from potato (Du Jardin et al., 1995), however, a functional assessment of the in vivo role of this protein is yet to be performed. Some evidence of a role for a pyrophosphatase activity was provided in experiments in which potato tuber discs were treated with fluoride (Viola and Davies, 1991), however, caution is required while interpreting these data as fluoride is a relatively promiscuous inhibitor. In many species including pea embryos, soybean cell suspension cultures and cauliflower buds, AGPase appears to be located exclusively in the plastid (MacDonald and ap Rees, 1983; Journet and Douce, 1985; Smith, 1988) and this isoform thus plays an important role in mediating the flux of carbon to starch. On removal or severe reduction of the AGPase activity in Arabidopsis or potato the level of starch was found to be dramatically reduced in all tissues (Lin et al., 1988a, 1988b; Müller-Röber et al., 1992). In direct contrast, when a non-regulated bacterial AGPase was expressed in various plant tissues the starch levels were dramatically increased (Stark et al., 1991). Plant AGPases are multisubunit proteins and expression studies in which the potato tuber enzyme way expressed in E. coli revealed that maximal activity can only be achieved on expression of both the large and the small subunit (Iglesias et al., 1993). Moreover, they are also allosterically regulated, being activated by 3-PGA and inhibited by Pi (Preiss, 1988; Sowokinos and Preiss 1992; Ballicora et al., 1995), and there is a clear evidence that changes in these metabolites are involved in the regulation of starch synthesis within leaves allowing the coordination of carbon assimilation, sucrose synthesis and starch synthesis (Stitt, 1997). There is also increasing evidence of a strong correlation between the 3PGA and ADPglucose levels and the rate of starch synthesis in potato tubers under a wide range of conditions (Geigenberger et al., 1997; 1998a). It is worth noting that using mutagenesis, 3PGA insensitive forms of the plant enzyme have also been created (Greene et al., 1996). However, since there are also several reports in the literature that the plant plastidial AGPase is clearly regulated, at least in vitro, by redox status (Fu et al., 1998; Ballicora et al., 2000), it is probable that this should also be taken into account in the design of future strategies intent on increasing starch content.

Whilst there is a wealth of information on the regulation of the plastidial isoform of AGPase detailed knowledge of the cytosolic isoform is very much limited. Several import studies provide evidence that the ADPglucose produced in the cytosol can be taken up by the plastid (Pozueta-Romera et al., 1991a, 1991b; Tetlow et al., 1994; Möhlmann et al., 1997), a process most probably mediated by the Brittle 1 protein (Sullivan et al., 1991). From these studies and from characterisation of maize Brittle 1 mutants which accumulated ADPglucose to 13 times the level found in wild-type plants (Shannon et al., 1996) it would seem that the Brittle 1 gene encodes for an amyloplastidial ADPglucose transporter. Despite these findings the physiological significance of cytosolic ADPglucose production remains unclear for a range of species. Thus, calculations of Denyer et al. (1996) demonstrate that the AGPase activity of the plastid is insufficient to account for measured rates of starch synthesis in barley endosperm, suggesting that at least some of the ADPglucose required for this process is provided by cytosolic production.

The recent purification of an ADPglucose pyrophosphatase from a range of plant species (Rodriguez et al., 2000) complicates matters further. This enzyme is believed to be co-localised with AGPase and to compete with starch synthase thus markedly blocking starch synthesis. Moreover, in studies in E. coli it was found that when this protein was reduced by insertional mutagenesis the level of glycogen marginally increased, suggesting this protein could play an important role in the regulation of carbohydrate storage (Moreno-Bruna et al., 2001). Furthermore, recent studies of 14-3-3 proteins within starch granules of Arabidopsis chloroplasts (Sehnke et al., 2001) indicate that these proteins may also be involved in the regulation of starch metabolism. However, to date no substantial data have been reported to characterise the physiological role of these proteins in plants and it is therefore unclear how important these proteins are in the regulation of starch metabolism. Despite this note of caution, it is clear that such regulatory genes could represent important future strategies. This is especially true when it is considered that the modification of pathway enzymes often has less than the desired effect.

Whilst the involvement of the above enzymes in starch biosynthesis are strictly species dependent, the starch polymerising activities are ever present and responsible for the formation of the two different macromolecular forms of starch, amylose and amylopectin. Starch synthases catalyse the transfer of the glucosyl moiety from ADPglucose to the non-linear end of an αα-1,4 glucan. The various starch synthases are able to extend 1, 4-glucans in both amylose and amylopectin. At least four different classes of starch synthases exist, designated as GBSS (granule-bound starch synthase), SSI, SSII and SSIII, which vary greatly in molecular weight, need for primers, substrate affinities and antigenic properties (for a review see Sivak and Preiss, 1998). It seems likely that most plant species contain the four different classes of starch synthase; however, the extent to which they contribute in vivo probably differs considerably between species (Denyer et al., 2001). Starch branching enzymes (SBE) are responsible for the formation of αα-1,6 branch points within amylopectin. Although there are more than two isoforms present in most plant species, all isoforms can be separated into two classes—most simply designated as A and B forms (Burton et al., 1995). The precise mechanism by which this is achieved is unknown, however it is thought to involve cleavage of a linear αα-1,4 linked glucose chain and reattachment of the chain to form an αα-1,6 linkage (Kossmann and Lloyd, 2000). The combined action of starch synthases and branching enzymes play an important role in determining the structure of starch which will be described in detail below. Other enzymes of starch synthesis and degradation are less well understood. Disproportionating enzyme (D-enzyme) is able to synthesise αα-1,4 glucans from maltose and has been suggested to be a candidate as a source of the malto-oligosaccharide primers required for starch synthesis. However, several lines of evidence suggest that this is unlikely to play a major role in starch synthesis in vivo. The maltose present in plant tissues is almost exclusively derived from starch (Kossmann and Lloyd, 2000) and transgenic plants exhibiting reduced D-enzyme expression had no effect on starch content (Takaha et al., 1998). Recent studies on an Arabidopsis mutant deficient in D-enzyme reveal a minor decrease in starch under certain conditions. However, they indicate that this enzyme primarily plays a role in the removal of malto-oligosaccharides during starch degradation (Critchley et al., 2001).

A further protein with a possible role in starch synthesis is the R1 protein. This protein has long been thought to be involved in the phosphorylation of starch since starch isolated from transgenic potato plants in which the R1 protein was reduced by antisense repression displayed only 10% of the phosphate content in wild-type potatoes (Lorbeth et al., 1998). Consistent with this proposal the enzymatic function of the protein was recently proved to be a starch water dikinase (Ritte et al., 2002).

To fully comprehend factors that determine starch biosynthesis knowledge of both synthetic and degradative functions is required. Currently, understanding of the roles of the starch degradative enzymes is relatively rudimentary. However, on the basis of several recent studies it has been proposed that several of the enzymes once regarded as operating exclusively in starch degradation also have a role in the synthesis of starch. The proposed catabolic and anabolic roles of debranching enzymes (isoamylases and pullanases), starch phosphorylase and αα- and ββ-amylases will be discussed below and their role in starch structure will be covered in greater detail in the later sections.

The endosperm from the Sugary-1 (Su-1) mutant of maize contains a second type of branched glucan other than amylopectin that is known as phytoglycogen. This mutant was shown to be deficient in an isoform of debranching enzyme (Pan and Nelson, 1984). This has since been confirmed when the Su-1 gene was cloned and found to encode an isoamylase-type enzyme (James et al., 1995). A similar phytoglycogen-accumulating Su-1 mutant has also been found in rice which exhibits changes in the activities of several enzymes of starch metabolism. The most dramatic of these by far was a 90% reduction in total debranching enzyme activity (Nakamara et al., 1996). That this reduction was specific to pullulanase was demonstrated immunologically, and pullulanase activity was found to correlate closely to phytoglycogen accumulation across rice lines producing different concentrations of soluble sugars (Nakamura et al., 1997). The sta-7 mutant of the monocellular green algae Chlamydomonas rheinhardtii has been found to contain no starch but a small amount of phytoglycogen (Mouille et al., 1996). Studies of this mutant revealed that the only activity of starch metabolism missing was that of a debranching enzyme. In combination these data present compelling evidence of a role of debranching enzymes in starch synthesis. However, the exact mechanism for this remains controversial. In the last few years two models have been proposed for amylopectin synthesis (summarised in Figure 25.4). In the model of Ball et al. (1996) glucans are synthesised within amylopectin until they reach a certain regular length which allows branching enzymes to act on them. Branching enzymes subsequently produce an uncrystalline glycogen-like polysaccharide (preamylopectin) on the outside of the linear chains. Debranching enzymes are then proposed to trim back the preamylopectin to leave amylopectin and in the process regenerate primer molecules to trigger a further cycle of synthesis and degradation. The second model of Zeeman et al. (1998a) suggests that starch is made exclusively by starch synthases and SBE and phytoglycogen is made as a byproduct of this process . They argue that phytoglycogen is subsequently degraded by a suite of enzymes including debranching enzymes and the products of this degradation can be used to support starch synthesis. Whichever of these models is correct it is clear that debranching enzymes play an important, albeit perhaps indirect, role in the process of starch biosynthesis.

The role of starch phosphorylase is less clear, since it catalyses a reversible reaction whereby glucose 1-phosphate is liberated from or incorporated into the non-reducing end of a glucan chain. Previously, based on the assumption that there was only a small amount of glucose 1-phosphate in the amyloplast, it was thought that the degradative reaction was favoured in vivo (Preiss and Levi, 1980; Steup, 1988). Plants contain both plastidial and cytosolic isoforms of this enzyme. On germination of pea embryos the plastidial isoform decreases 10-fold in activity whilst the cytosolic isoform remains unchanged (van Berkel et al., 1991). These data seem to preclude a major degradative role for the plastidial isoform and suggest that it is possible that the cytosolic isoform gains access to the starch as the amyloplastid membrane degrades. However, when the cytosolic isoform or either of the plastidial isoforms of this enzyme in potato is reduced by antisense repression no changes are observed in the rate of starch degradation (Duwenig et al., 1997a).

The degradation of starch by αα- and ββ-amylases is perhaps better characterised. Alpha-amylases are endoamylolytic, being able to break αα-1,4 bonds in amylose and amylopectin. These enzymes have been studied extensively, particularly in cereal endosperm, and are very varied with respect to both degradative ability and the extent of post-translational modification that they are subjected to (see Kossmann and Lloyd, 2000). Furthermore, in many species they appear to be gene families of αα-amylases with at least 10 in rice (Huang et al., 1990) and five in potato (Gausing and Kreiberg, 1989). It is thought that this great diversity may reflect the different roles for these enzymes in different tissues. Alpha-amylases appear to play a role in starch mobilisation during seed germination especially in cereals. In these crop plants it is thought that only hydrolytic enzymes have a role in the degradation of starch since large amounts of maltose and glucose accumulate and these substances are not known to be produced by the other degradative enzymes (Beck and Ziegler, 1989). Although the effects of αα-amylase during cereal seed germination has been well characterised for other plants its role is less clear. Studies on an Arabidopsis mutant with a reduced capacity to degrade leaf starch demonstrated that several by αα-amylases were extra-plastidial (Zeeman et al., 1998b). However, in this mutant it was one of the plastidial located αα-amylases that had a greatly reduced activity indicating that this isoform is responsible for starch degradation in Arabidopsis leaves. The role of αα-amylases in potato is currently unknown. However, in vitro studies of an αα-amylase purified from the potato starch granule revealed that at least one potato isoform is able, and correctly located, to degrade starch (Witt and Sauter, 1996).

In contrast to the αα-amylases, the ββ-amylases are exoamylolytic and liberate maltose residues progressively from the non-reducing ends of amylose and amylopectin until they react an αα-1,6 branch point. Their role in starch degradation is unclear as many isoforms have been found to be located in the vacuole. However, their activity increases during germination of seeds of mustard, maize and rice (Okomoto and Akazawa, 1980; Subbaramaiah and Sharma, 1989; Wang et al., 1997) and also on cold sweetening in potatoes (Hill et al., 1996; Nielsen et al., 1997). Since both these processes are associated with a time of active starch breakdown, it follows that there is at least some role for ββ-amylases during starch mobilisation.

The above studies have largely focussed on pathways as individual linear entities, however metabolism is in fact highly branched and better represented as a network since many pathways are linked by common metabolites and co-factors. In plant systems most prominent amongst such molecules are pyrophosphate and the adenylate and uridinylate pools. In addition other important factors that influence the regulation of carbohydrate metabolism should be considered before we begin discussing biotechnological strategies in earnest.

Pyrophosphate is an essential co-factor in starch storing organs since following SuSy-dependent sucrose cleavage, sucrose mobilisation requires the reaction catalysed by the pyrophosphate-dependent UDPglucose pyrophosphorylase (Zrenner et al., 1993). Since one molecule of pyrophosphate is required for each sucrose cleaved by the SuSy-dependent pathway and given that, in many heterotrophic cells the majority of the incoming sucrose is converted to starch, there are only two possible sources for the necessary pyrophosphate to fuel this reaction. It is either recycled across the amyloplast membrane from the starch biosynthetic pathway or it is produced by a cycling process involving pyrophosphate: fructose 6-phosphate, 1-phosphotransferase (PFP) or the tonoplast pyrophosphatase (Stitt, 1998). Correlative evidence for the former proposal includes (i) the observation that pyrophosphatase activity does not increase during the developmental switch from elongating stolon to growing tuber (Appeldoorn et al., 1999), (ii) the description of a pyrophosphate transporter in chloroplasts (Lunn and Douce, 1993) and (iii) measurements of pyrophosphate contents and the rates of sucrose degradation and starch synthesis in a range of transgenic lines exhibiting altered levels of pyrophosphate (Farre et al., 2000). Furthermore, analysis of metabolite levels in transgenic lines exhibiting reduced levels of PFP suggest that this enzyme is operating in the glycolytic direction and thus consuming pyrophosphate in vivo (Hajirezaei et al., 1994) and, therefore, is incapable of supporting sucrose degradation. Moreover, there was no difference in the rate of sucrose degradation (or resynthesis) in heterotrophic transgenic tobacco cells exhibiting increased levels of fructose 2,6-bisphosphate and therefore elevated in vivo activity of PFP (Fernie et al., 2001c). Taken together these data indicate that it is unlikely that PFP supplies pyrophosphate for sucrose degradation and it seems possible that the pyrophosphate level provides an important link between the catabolic reactions of the cytosol and the anabolic reactions of the plastid.

Direct evidence to support this model is however unfortunately lacking. When pyrophosphate levels were depressed by the expression of a bacterial pyrophosphatase, initial experiments revealed an inhibition of sucrose breakdown and a reduction in starch accumulation (Jelitto et al., 1992), whereas subsequent experimentation carried out with plants at a different developmental stage showed the exact opposite (Geigenberger et al., 1998b). A further discrepancy in the behaviour of plants exhibiting low levels of pyrophosphate is that they have been variously reported to be characterised by accelerated (Farre et al., 2001a) and delayed (Hajirezaei and Sonneewald, 1999) sprouting. However, different independent transgenic lines were used in these two studies and it is possible that different pyrophosphate levels trigger different effects. Despite these discrepancies, the combination of results to date suggest that pyrophosphate is clearly capable of effecting a variety of processes and, therefore, has a central, if somewhat enigmatic, role in the regulation of the sucrose to starch transition in heterotrophic tissues.

Studies on the effects of increasing the levels of adenylates (Tjaden et al., 1998; Loef et al., 2000; Geigenberger et al., 2001) and uridinylates (Loef et al., 2001) on biosynthesis, although limited to date, have revealed that these compounds are also important in the regulation of the starch biosynthetic pathway, with increases in cellular ATP (Loef et al., 2001) or of ATP supply to the plastid (Tjaden et al., 1998; Geigenberger et al., 2001) stimulating starch synthesis. Similarly, feeding potato tuber slices with precursors of uridinylate synthesis resulted in increase of uridinylates which stimulated starch synthesis and resulted in an increased partitioning of carbon towards starch.

In addition to their role as intermediates in metabolic pathways recently much attention has focused on the potential role of sugars as regulatory signals (Koch et al., 1990; Purcell et al., 1998). Driven by compelling evidence from the yeast system the primary candidate for a signal is glucose acting via sensing mechanisms involving either hexokinase or SNF1. In potato tubers the expression of invertase in the cytosol resulted in an accumulation of glucose and a resultant shift in partitioning from starch towards glycolysis (Trethewey et al., 1998). A similar metabolic shift was observed following the incubation of wild-type tuber discs in high-glucose concentrations (Geiger et al., 1998). However, the expression of a bacterial sucrose phosphorylase and the supertransformation of invertase expressing tubers with a bacterial glucokinase (Trethewey et al., 1998; Trethewey et al., 2001) displayed essentially the same metabolic phenotype despite the fact that they were not characterised by an increased glucose concentration. Furthermore, modulation in the activities of either isoform of potato tuber hexokinase had no major impact on either tuber morphology or on metabolism (Veramendi et al., 1999; 2001) thereby strongly arguing against a role for glucose-dependent signalling processes in the tuber.

Thus, the subject of sugar sensing remains highly controversial and there is reasonable evidence that sugar carriers or other factors at the plasma membrane may play a role in the regulation of heteotrophic metabolism (Sonnewald et al., 1997; Lalonde et al., 1999; Fernie et al., 2000, 2001a; Roitsch et al., 2000). Firstly, potato tubers expressing invertase in the apoplast are characterised by an increased rate of cell division and display a different metabolism to those expressing the invertase at a cytosolic location (Sonnewald et al., 1997) despite the fact that the hexoses released at both sites are able to enter metabolism (Fernie et al., 2000). Secondly, the rate of sucrose degradation and starch synthesis is stimulated when the unmetabolisable sucrose analogue palatinose is supplied to isolated tuber discs despite the fact that the uptake of palatinose into the tuber parenchyma is negligible (Fernie et al., 2001a). Whilst these studies allow us to speculate that these responses are mediated by a factor(s) localised in the plasma membrane (Lalonde et al., 1999), a lot of work is needed using a variety of approaches before the precise nature and role of sugar sensing in heterotrophic tissues can be defined and usefully manipulated. They do, however, remain an attractive alternative approach for metabolic manipulation since if such regulatory processes can be harnessed by the metabolic engineer as an opportunity to orchestrate rather than merely modify metabolism may arise.

Uptake of Carbon into Amyloplasts

The form in which carbon crosses the amyloplast membrane and enters into starch biosynthesis has been the subject of considerable debate. Categorical evidence that carbon enters potato tuber, Chenopodium rubrum, maize endosperm, wheat endosperm and tobacco amyloplasts in the form of hexose monophosphates (or nucleosides), rather than triose phosphates was provided by determination of the degree of randomisation of radiolabel in glucose units isolated from starch following incubation of the various tissues with glucose labelled at the C1 or C6 positions (Keeling et al., 1988; Viola et al., 1991; Hatzfeld and Stitt, 1990; Fernie et al., 2001c). These data are in agreement with the observation that potato tubers lack plastidial fructose 1, 6-bisphosphatase activity (Entwistle and ap Rees, 1990) and the failure to find expression of plastidial FBPase in tubers (Kossmann et al., 1992).

Although it is clear that triose-phosphates are not the substrate taken up by amyloplasts to support starch synthesis there has been considerable debate as to whether glucose 1-phosphate (Naeem et al., 1997; Tetlow et al., 1994; Tyson and ap Rees, 1988) or glucose 6-phosphate (Schott et al., 1995; Wischmann et al., 1999) is the preferred substrate for uptake. More recently, particularly in cereals, the uptake of cytosolically produced ADPglucose has also been much discussed (Pozeuta-Romera et al., 1991a, 1991b; ap Rees, 1995). The necessary pathways to support starch synthesis presuming uptake of one of these three substrates are presented in Figure 25.3. The results of recent transgenic and immuno-localisation experiments have indicated that the substrate for uptake is most probably species specific with clear evidence of the predominant route of uptake in the developing tuber in the form of glucose-6-phosphate, whereas in barley, wheat, oat and possibly maize, the predominant form of uptake is as ADPglucose (Denyer et al., 1996; Thorbjornsen et al., 1996; Shannon et al., 1998).

The cloning of a hexose monophosphate transporter from potato and the finding that the cauliflower homologue is highly specific for glucose 6-phosphate provides strong support for the first theory (Kammerer et al., 1998). Furthermore, when this observation is taken together with in vivo evidence that transgenic potato lines, in which the activity of the plastidial isoform of phosphoglucomutase was reduced by antisense inhibition, were characterised by a large reduction in starch content (Tauberger et al., 2000), then there are compelling grounds for asserting that glucose 6-phosphate is the major form in which tuber amyloplasts import carbon from the cytosol. Since these antisense plants were not starchless we cannot, however, exclude the possibility that glucose 1-phosphate makes some contribution to the flux to starch. Nor should we, in light of recent findings of extra-plastidial isoforms of ADPglucose pyrophosphorylase (Beckles et al., 2001), overlook the possibility of production of ADPglucose by a cytosolically localised enzyme and its subsequent transport in to the plastid to supplement starch synthesis.

However, this seems unlikely since following non-aqueous fractionation of potato tuber tissue AGPase activity always co-localised with pyrophosphatase activity which is known to be located exclusively in the plastid (Farre et al., 2001b) and expression of a bacterial AGPase in the cytosol of potato tubers did not result in an altered starch content (Stark et al., 1991). Furthermore, results from recent comprehensive studies in which the ratio of ADPglucose to UDPglucose was determined in a wide range of species suggest that the presence of a cytoplasmic AGPase isoform is limited to Graminaceous endosperms and is not a general feature of starch-storing organs (Beckles et al., 1991). This rationale behind these measurements is that the metabolite ratio is expected to be high in organs in which UDPglucose and ADPglucose are both mainly produced in the cytosol since the reactions of AGPase and UGPase will be coupled and close to equilibrium.

The results from this study are in direct contrast to earlier immunolocalisation studies using antisera against AGPase which suggested that there was an extraplastidiary isoform of AGPase in tomato fruit (Chen et al., 1998), but not in maize endosperm (Miller and Chourey, 1995; Brangeon et al., 1997). However, it is possible that the immunogold-labelling patterns seen in these studies do not accurately reflect the in vivo situation. Further studies by Beckles and Smith (2001) indicated that this is indeed probably the case since the proportion of the total activity of AGPase that was confined to the plastid was similar to that of the total activity of enzymes known to be confined to the plastid. When samples of plastid and total homogenate fractions were subjected to immunoblotting with an antisera raised against AGPase, most or all of the protein detected was plastidial.

The utilisation of UDPglucose as a substrate for starch synthesis has not been discussed here as it is unlikely to be a major route of starch synthesis in plants for several reasons. Firstly, unlike certain glycogen synthases that can efficiently utilise UDPglucose as a substrate, plant starch synthases are either specific for ADPglucose, or have affinities for this nucleoside that are far in excess of those for UDPglucose (Smith, 1990). Furthermore, since UGPase appears to be absent from the plastid (Entwistle and ap Rees, 1988) there is no route other than through starch synthases by which UDPglucose can support starch synthesis. Secondly, as described above the reduction of ADPglucose production by decreasing the AGPase activity in a wide range of species by approaches of mutagenesis or transgenesis reduces starch accumulation (Tsuai and Nelson 1966; Lin et al., 1988a, 1988b; Smith et al., 1989; Müller-Rober et al., 1992). Additionally, as described above genetic manipulation of SuSy which produces UDPglucose had no effect on starch synthesis.

Production of Hexose Phosphates in the Cytosol

The circular pathway operating between sucrose and hexose phosphates is well established and has been intensively investigated (Figures 25.1 and 25.2). In the last decade a battery of transgenic plants expressing sense or antisense constructs targeted against most of the genes involved in this pathway (and ectopic expression of proteins specifically directed at many of the metabolic intermediates) have been generated. Moreover, in the last 2 years the genes for plastidial and cytosolic phosphoglucomutase (Tauberger et al., 2000; Fernie et al., 2001b) and sucrose phosphate phosphatase (Lunn et al., 2000) have been cloned suggesting that the jigsaw puzzle of genes is now complete.

Sucrose delivered to the tuber can be cleaved in one of three ways (i) as described above in the apoplast by acid invertase or in the cytosol by either (ii) alkaline invertase or (iii) sucrose synthase (SuSy). As indicated in Figures 25.1 and 25.2 the primary route of sucrose cleavage mirrors the mechanism of unloading with invertase activities being high during the early stages of tuber initiation whilst SuSy dominates in the developing tuber (Appeldorn et al., 1999), whereas the opposite is true for the developing tomato fruit (Damon et al., 1988; Robinson et al., 1988; DemnitzKing et al., 1997). Cereal seeds tend to have high apoplasmic invertase activities to facilitate unloading, however, as mentioned above these are not a prerequisite for unloading hence it is most likely that both pathways have a role to play in sucrose mobilisation within this sink organ (Schmalstig and Hitz, 1987).

The products of sucrose cleavage enter into metabolism by the concerted action of fructokinase and UDPglucose pyrophosphorylase (Zrenner et al., 1993) or fructokinase and hexokinase (Smith et al., 1993; Veramendi et al., 1999, 2002) in the case of the SuSy and invertase pathways, respectively. The hexose phosphates produced by these reactions are equilibrated by the action of cytosolic isoforms of phosphoglucose isomerase and phosphoglucomutase (Fernie et al., 2001b). Sucrose mobilisation has been subject to intense investigation in all crop species, however, perhaps the greatest scrutiny has been applied to the potato.

Experiments of Fu and Park (1995a) (Fu et al., a,b) revealed that potato contains two differentially expressed classes of genes encoding for SuSy in potato which they named SuS3 and SuS4. Both classes were shown to contain 13 introns, including a particularly long leader intron, and their coding regions were found to be 87% identical at the nucleotide level. Using GUS fused to the 5′ flanking sequences they found that SuS3 is strongly expressed in vascular tissues of leaves, stems, roots and tubers—implying a possible role in energy provision in phloem cells: whereas SuS4 genes are strongly expressed in sink tissues such as root tips, basal tissues of the shoot and potato tubers. The SuS4 genes correspond to the cDNA for the T-type isoform first cloned by Salanoubat and Belliard (1987) who proposed that this isoform plays a dominant role in the metabolism of sucrose within the tuber. The T-type isoform was subject to antisense inhibition using the 35S promoter (Franck et al., 1980), and a reduction in SuSy activity was only found in the tubers (Zrenner et al., 1993). In tubers a reduction in activity of up to 95% resulted in a reduction in starch and storage protein content of mature tubers but surprisingly no changes in the level of sucrose. There was, however, a significant increase in the level of hexoses which was in keeping with an observed 40-fold increase in the invertase activity of these lines. The exact reason for this compensatory increase is unknown but it serves to provide further evidence of the flexibility inherent within plant metabolism. Such metabolic flexibility appears to be a central feature of plant metabolism and probably accounts for the absence of major metabolic effects of many transgenic manipulations (e.g., Burrell et al., 1994; Hajirezaei et al., 1994; Fernie et al., 2001c). Thus, when taken together the results of the studies of Zrenner and co-workers strongly indicate that SuSy plays a significant role in determining potato tuber sink strength—at least when activities are reduced. The fact that the enzyme is likely to operate close to equilibrium in vivo (Geigenberger and Stitt, 1993), probably explains the failure to increase sink strength via overexpression strategies (Howard et al., 2001). Further transgenic studies confirmed the pre-eminance of the SuSy route of sucrose cleavage within the developing tuber. Potato plants repressed in the activities of hexokinases (Veramendi et al., 2002, 1999) or of acid invertase (Zrenner et al., 1996) exhibiting very little difference from wild-type tubers implying that these enzymes do not play such a crucial role in metabolism during this stage of the tuber life cycle.

Although the net flux in the tuber is one of sucrose degradation, the rate of sucrose (re)synthesis within this and other tissues is considerable (Wendler et al., 1991; Geigenberger et al., 1997; 1999a; Fernie et al., 2001c; Nguyen-Quoc and Foyer, 2001). This process can also proceed via two different pathways: the reverse of the SuSy degradative pathway or the reactions catalysed by sucrose phosphate synthase and sucrose phosphate phosphatase (Geigenberger and Stitt, 1993). There is clear evidence from feeding experiments with labelled sugars that both pathways contribute to sucrose (re)synthesis within the developing tuber (Geigenberger and Stitt, 1993). It is thought that the combined operation of these pathways with the degradative pathway allows the cell to respond sensitively to both variations in sucrose supply and the cellular demand of carbon for biosynthetic processes (Hatzfeld and Stitt, 1990). The importance of SuSy is demonstrated following antisense repression of this enzyme which resulted in a reduced starch content, coupled to a reduced tuber dry weight and a reduction in storage proteins. Furthermore, when SuSy is bypassed either by introduction of a yeast invertase (Sonnewald et al., 1997; Trethewey et al., 1998) or a bacterial sucrose phosphorylase (Trethewey et al., 2001) there are similar dramatic repercussions on metabolism.

In contrast, sucrose synthesis via sucrose phosphate synthase is relatively well defined (Geigenberger et al., 1997). Although the reaction catalysed by sucrose phosphate synthase is not irreversible, the efficient removal of its product by sucrose phosphate phosphatase means that sucrose phosphate synthase is the first commited reaction in the sucrose synthetic pathway. In photosynthetic tissue sucrose phosphate synthase has been found to be regulated at a variety of levels including allosteric activation by glucose 6-phosphate and inhibition by inorganic phosphate (which allows sucrose synthesis to proceed at times when substrate is plentiful) and deactivated by protein phosphorylation (Reimholz et al., 1994). There is now a growing body of correlative evidence that the potato tuber enzyme is regulated in an analogous manner to the leaf enzyme (Geigenberger et al., 1994, 1999a; Fernie et al., 2001b). The complex regulation of this enzyme suggests that it plays an important metabolic role, a view that is supported by antisense studies which reveal only a minor influence on starch metabolism but a major role in sucrose synthesis in response to water stress (Geigenberger et al., 1997).

In sharp contrast to the lack of changes observed following antisense repression of the majority of enzymes involved in sucrose degradation in the potato are the severe changes seen following repression of the same enzmyes in tomato. With tomato plants altered in invertase (Dickinson et al., 1991; Klann et al., 1996), SuSy (D'Aoust et al., 1999) and hexokinase (Dai et al., 1999) activities display both dramatic phenotypes and marked shifts in metabolism. Transgenic tomatoes expressing a yeast invertase in their apoplast were severely repressed in their growth (Dickinson et al., 1991), however, this transgene was under the control of the tissue-constitutive 35S promoter, and it is most likely that this phenotype results from a block in photoassimilate export. Interestingly, when the expression of one of the tomatoes' six endogenous invertase genes, TIV1, was reduced constitutively by antisense repression of the fruit size was also reduced (Klann et al., 1996) and the precise role of this or any of the other tomato invertases remains to be elucidated. In contrast, the role of SuSy in tomato fruit is well defined and analogous to that in the potato tuber, it seems to be an important determinant of sink strength as indicated by the decreased fruit setting and sucrose unloading capacity of plants exhibiting reduced expression of the enzmye (D'Aoust et al., 1999). Marked phenotypic changes were also observed on the overexpression of an Arabidopsis hexokinase gene in transgenic tomato plants which displayed dramatic growth reduction and early senescence changes which were not seen in the potato (Veramendi et al., 1999). However, it is likely that many of these changes result from the dramatic reduction of leaf photosynthesis in these plants (Dai et al., 1999). In addition to these phenotypic changes, changes in the sugar content and its subsequent mobilisation via metabolism were observed in all the above examples and also in situations wherein the activities of sucrose phosphate synthase (Nguyen-Quoc et al., 1999) were altered transgenically and various invertase activities were altered by conventional plant breeding (Chetelat et al., 1995; Fridmann et al., 2000; Husain et al., 2001). One clear reason for the differences observed on manipulating certain enzyme activities in tomato as opposed to potato is that the tomato fruit goes through large developmental changes whereas the potato tuber does not. This is also reflected in their metabolism. For example starch accumulation in potato proceeds linearly with time (Moorby and Milthorpe, 1975) but is merely transient in tomato fruit (Schaeffer and Petroikova, 1997).

The phenotypes observed in maize seeds on the modification of pathways of sucrose mobilisation are equally severe. We have described the minature1 mutant of maize deficient in cell wall invertase above. Maize further contains two isoforms of SuSy SH1 and SUS1 encoded by the Shrunken1 (Sh1) and Sucrose synthase I (Sus1) loci, respectively. Although both genes are expressed in the developing endosperm, SH1 contributes the majority of the total SuSy enzyme activity. Seeds of the shrunken1 mutant have a mild starch deficiency and degenerate storage cell whereas those of the sus1 mutant shows no phenotypic change (Chourey et al., 1998). Furthermore, distinct phenotypes have been observed on studies of transgenic carrot plants wherein the activities of either invertase or SuSys were altered in the taproots. When invertase activity is severely reduced the carrot plants fail to develop roots suggesting a crucial role for this enzyme in the regulation of carbon portioning (Tang et al., 1999). Reduction of the SuSy activity of carrot roots results in decreased root growth. The transgenic plants also displayed a reduced sucrose utilisation but an increase in starch and cellulose accumulation which was interpreted to suggest that SuSy played an important developmental role within the carrot plant (Tang and Sturm, 1999).

When taken together these examples highlight that plant species vary dramatically with respect to the importance of various routes of both sucrose delivery and utilisation and that successful manipulation at a certain metabolic loci in one species does not imply that such a manipulation will be possible in other species.

It is also important to note that hexose phosphates have a crucial mediatory role in the sucrose to starch transition; they also serve as precursor substrates for glycolysis and oxidative pentose phosphate pathway. However, these pathways have also been the subject of many transgenic studies in heterotrophic tissues and this subject area has been extensively reviewed elsewhere in recent years (Stitt and Sonnewald, 1995; Neuhaus and Emes, 2000; Given, 1999).

Pathways of Heterotrophic Carbohydrate Metabolism

The Supply of Sucrose

In photosynthetic and gluconeogenic tissues sucrose is predominantly exported from cells, most probably by facilitated diffusion, and subsequently taken up by the phloem complex through a specific sucrose/H+ co-transport mechanism (Riesmeier et al., 1994; Frommer and Sonnewald, 1995). Once in the phloem complex, sucrose is transported to cells of heterotrophic ‘sink’ organs. Sucrose obtained through translocation can enter a cell via the symplasm (Figure 25.1) or the apoplasm (Figure 25.2) and in many species the nature of the predominantly used route is hotly debated. Several studies using assymetrically labelled sucrose suggest that carbon obtained by heterotrophic cells moves primarily through the symplasmic route and is not cleaved to glucose and fructose during transport. It seems likely that cells of many species receive most of their sucrose by such a route (Patrick, 1990; Tegeder et al., 1999; Lalonde et al., 1999).

However, in certain tissues it is clear that sucrose must be supplied through the apoplasm. This is certainly the case in developing seeds in which protoplasmic connections between maternal and embryonic tissue simply do not exist. Thus, studies on the pathway of uptake of sucrose from the apoplast revealed that there is not a single route of uptake. Hydrolysis of sucrose precedes uptake by developing seeds of maize, sorghum and pearl-millet, whereas in wheat, rye and barley sucrose appears to be transferred without cleavage (Thorne, 1985; Weschke et al., 2000). However, even in species where apoplasmic hydrolysis of sucrose occurs this does not seem to be a prerequisite for uptake since the invertase-resistant sucrose analogue 1-fluorosucrose is taken up by maize seeds at similar rates to that of sucrose (Schmalstig and Hitz, 1987). Studies on the minature-1 mutant of maize, deficient in apoplasmic invertase activity, revealed that seeds were only one fifth the normal weight (Miller and Chourey, 1992), suggesting that apoplasmic hydrolysis of sucrose may play an important role in the maintenance of source to sink sucrose gradients.

The pathway of phloem unloading in the tuber has been the subject of much debate. It has been clear for some time that plasmodesmatal connections between the phloem and the surrounding parenchyma cells exist in the tuber (Oparka and Prior, 1987) and that plasmolysis of growing tubers has an inhibitory effect on the flux of sucrose into the tuber (Oparka and Wright, 1988) suggesting that unloading occurs via a symplastic mechanism. However, isolated tuber discs display a substantial capacity to take up sucrose supplied to the surrounding media (Geigenberger et al., 2000; Fernie et al., 2001a). Furthermore transgenic expression of a yeast-derived invertase in the apoplast under the control of the tuber specific B33-patatin promoter significantly altered tuber yield (Sonnewald et al., 1997) indicating the importance of apoplastic sucrose.

Recent studies using a combination of confocal microscopy, autoradiography and biochemical analyses have provided definitive evidence that unloading in the potato tuber is predominantly apoplastic during stolon elongation and becomes primarily symplastic during the initial phases of tuberisation (Viola et al., 2001). This is in direct contrast to the situation observed in the developing tomato fruit in which sucrose unloading is predominantly symplasmic during early, starch accumulating, stages of development (Damon et al., 1988; Ruan and Patrick, 1995) and apoplasmic during later, hexose accumulating, stages (Patrick et al., 1990; Ruan and Patrick, 1995).

Given that sucrose unloading is essentially symplastic in the developing potato tuber the impact on tuber morphology following expression of a heterologous invertase in the apoplast at this developmental stage (Sonnewald et al., 1997; Hajirezaei et al., 2000) is intriguing. However, despite the large morphological changes apoplastic expression of invertase had no effect on the levels of cellular metabolites (Hajirezaei et al., 2000) and the role of sugars in the apoplastic space of the tubers remains unsolved. Interestingly, when the rate of glucose consumption in these transgenic lines was increased by the cytosolic expression of a bacterial glucokinase the total hexose content of the tuber was reduced implying that apoplastic hexose is somehow able to enter cytosolic metabolism (Fernie et al., 2000). Comparison of transgenic plants exhibiting apoplastic with those exhibiting cytosolic expression of the invertase reveals that a completely different phenotype is produced depending upon the compartment to which the enzyme is targeted. This observation together with results from biochemical studies, suggests that the route of entry of hexoses into metabolism differ according to whether they are generated in the cytosol or the apoplast.

For sugars synthesised in the apoplast this could imply either an endocytotic-like mechanism of transport to the vacuole and subsequent release to the cytosol or delivery into the cytosol by a specific hexose transporter in the plasma membrane which have a signalling capacity (Lalonde et al., 1999; Fernie et al., 2000). It is clear that an apoplasmic unloading mechanism needs the presence of one or both types (monosaccharide and sucrose) of transporter at the plasma membrane. Tables 25.1 and 25.2 list the currently sequenced monosaccharide and sucrose transporters of agronomically important plant species and of Arabidopsis. Since these transporters have been the subject of two excellent recent reviews (Lalonde et al., 1999; Lemoine, 2000) and to our knowledge there has been very little success, from a biotechnological standpoint, in the genetic manipulation of these transporters we will not discuss them in detail here. It is however worth pointing out that although the transport mechanism of the much studied potato sucrose proton transporter SUT1 has been characterised by expression in Xenopus oocytes (Boorer et al., 1996) its precise role in planta is yet to be fully elucidated. Since this is one of the best characterised transporters it, therefore, follows that much work is required before the factors controlling the intracellular movement of sugars within the heterotrophic cell can be fully resolved.

Carbohydrate Metabolism in Plants

The yield from crop plants has been subject to constant improvement through conventional breeding and refinements in agricultural practice for many decades. In the case of the potato, which will form the major focus of this chapter, the harvest index (ratio of dry weight of harvestable organs to the dry weight of the entire plant) has been increased from 0.09 in wild species up to 0.81 in modern cultivars (Inouhe and Tanaka, 1978), with smaller, yet dramatic, improvements found in other crop species (Ellen, 1993; Hay, 1995).

Although such improvements have proved both dramatic and revolutionary they were also time consuming and slow. The emergence of molecular-assisted breeding (see Chapter 6) and plant transformation technologies (see Chapters 8 and 9) offers the possibility of manipulating metabolism using a more rapid, targeted approach. Indeed since the advent and widespread adoption of transgenesis approaches some 15 years ago gave rise to the discipline of molecular plant physiology, much information has been obtained concerning the potential to manipulate plant metabolism. In this chapter we intend to review the many previous studies of genetic manipulation of heterotrophic carbohydrate metabolism in plants.

It is clear that the successful manipulation of plant metabolism requires detailed understanding of the underlying factors that regulate it. For this reason we intend to describe the current understanding of the central pathways of carbohydrate metabolism. Some stress will be put on the sucrose to starch transition since this pathway has received great attention over the past few years. In the case of the potato (Solanum tuberosum) tuber all the genes believed to be directly involved in the sucrose to starch transition have been cloned, the final gaps being filled within the last couple of years (Veramendi et al., 1999; Tauberger et al., 2000; Fernie et al., 2002; Veramendi et al., 2002).

Furthermore, an impressive variety of transgenic lines have been generated where the activities of most of the individual genes have been modulated, alone or in combination. In addition, a large range of transgenic potato lines has been created where deregulated alternatives to endogeneous enzymes have been introduced. These studies have allowed the confirmation of many longstanding hypotheses which were previously based upon indirect methodologies all of which will be summarised within this chapter. In addition, this chapter will also cover the synthesis of fructans and the introduction of novel carbohydrates into plants and will review recent advances in understanding and influencing structural properties of starch. As stated above we intend to split this review into two major sections. The first of these is a description of the pathways operating in various agronomically important plants following the path of carbohydrate from sucrose transported from source tissues to its assimilation into storage carbohydrate in sink tissues.

The metabolism of the major forms of carbohydrate: sucrose, starch and fructans will be covered in some detail. The second section provides a review of strategies taken to manipiulate these pathways for commercial gain with particular prominence given to alteration of starch functionality and to increasing end product accumulation of starch and sucrose and finally, the manipulation of minor and novel sugars are reviewed.

Map of Plant Protein Interactions

An international team of scientists has described their mapping and early analyses of thousands of protein-to-protein interactions within the cells of Arabidopsis thaliana -- a variety of mustard plant that is to plant biology what the lab mouse is to human biology.

"With this one study we managed to double the plant protein-interaction data that are available to scientists," says Salk Institute plant biologist Joseph Ecker, a professor in the Plant Molecular and Cellular Biology Laboratory. "These data along with data from future 'interactome' mapping studies like this one should enable biologists to make agricultural plants more resistant to drought and diseases, more nutritious, and generally more useful to mankind."

The four-year project was funded by an $8 million National Science Foundation grant, and was headed by Marc Vidal, Pascal Braun, David Hill and colleagues at the Dana Farber Cancer Institute in Boston; and Ecker at the Salk Institute. "It was a natural collaboration," says Vidal, "because Joe and his colleagues at the Salk Institute had already sequenced the Arabidopsis genome and had cloned many of the protein-coding genes, whereas on our side at the Dana Farber Institute we had experience in making these protein interaction maps for other organisms such as yeast."

In the initial stages of the project, members of Ecker's lab led by research technician Mary Galli converted most of their accumulated library of Arabidopsis protein-coding gene clones into a form useful for protein-interaction tests. "For this project, over 10,000 'open reading frame' clones were converted and sequence verified in preparation for protein-interaction screening," says Galli.

Vidal, Braun, Hill and their colleagues systematically ran these open reading frames through a high quality protein-interaction screening process, based on a test known as the yeast two-hybrid screen. Out of more than forty million possible pair combinations, they found a total of 6,205 Arabidopsis protein- protein interactions, involving 2,774 individual proteins. The researchers confirmed the high quality of these data, for example by showing their overlap with protein interaction datafrom past studies.

The new map of 6,205 protein partnerings represents only about two percent of the full protein- protein "interactome" for Arabidopsis, since the screening test covered only a third of all Arabidopsis proteins, and wasn't sensitive enough to detect many weaker protein interactions. "There will be larger maps after this one," says Ecker.

Even as a preliminary step, though, the new map is clearly useful. The researchers were able to sort the protein interaction pairs they found into functional groups, revealing networks and "communities" of proteins that work together. "There had been very little information, for example, on how plant hormone signaling pathways communicate with one another," says Ecker. "But in this study we were able to find a number of intriguing links between these pathways."

A further analysis of their map provided new insight into plant evolution. Ecker and colleagues Arabidopsis genome data, reported a decade ago, had revealed that plants randomly duplicate their genes to a much greater extent than animals do. These gene duplication events apparently give plants some of the genetic versatility they need to stay adapted to shifting environments. In this study, the researchers found 1900 pairs of their mapped proteins that appeared to be the products of ancient gene-duplication events.

Using advanced genomic dating techniques, the researchers were able to gauge the span of time since each of these gene-duplication events -- the longest span being 700 million years -- and compare it with the changes in the two proteins' interaction partners. "This provides a measure of how evolution has rewired the functions of these proteins," says Vidal. "Our large, high-quality dataset and the naturally high frequency of these gene duplications in Arabidopsis allowed us to make such an analysis for the first time."

The researchers found evidence that the Arabidopsis protein partnerships tend to change quickly after the duplication event, then more slowly as the duplicated gene settles into its new function and is held there by evolutionary pressure. "Even though the divergence of these proteins' amino-acid sequences may continue, the divergence in terms of their respective partners slows drastically after a rapid initial change, which we hadn't expected to see," Vidal says.

In the July 29 issue of Science researchers from the Arabidopsis interactome mapping study reported yet another demonstration of the usefulness of their approach. Led by Jeffery L. Dangl of the University of North Carolina at Chapel Hill, they examined Arabidopsis protein interactions with the bacterium Pseudomonas syringae (Psy) and a fungus-like microbe called Hyaloperonospora arabidopsidis (Hpa). "Even though these two pathogens are separated by about a billion years of evolution, it turns out that the 'effector' proteins they use to subvert Arabidopsis cells during infection are both targeted against the same set of highly connected Arabidopsis proteins," says Ecker. "We looked at some of these targeted Arabidopsis proteins and found evidence that they serve as 'hubs' or control points for the plant immune system and related systems."

Ecker and his colleagues hope that these studies mark the start of a period of rapid advancement in understanding plant biology, and in putting that knowledge to use for human benefit. "This starts to give us a big, systems-level picture of how Arabidopsis works, and much of that systems-level picture is going to be relevant to -- and guide further research on -- other plant species, including those used in human agriculture and even pharmaceuticals,"Ecker says.

The "Arabidopsis Interactome Mapping Consortium" consists of over 20 national and international laboratories that contribute to this study with support from a number of funding agencies including the National Science Foundation and the National Institutes of Health.

Plants as a Green Pharmacopoeia

We all routinely consume plant secondary metabolites—such as caffeine—without necessarily considering the complex biochemistry that is behind their production. Secondary metabolites have been used as medicines or as components of industrial processes for millennia. The methods used to manipulate their production are discussed by Yazaki in Chapter 43. Some secondary metabolites are limited in supply because the natural source is either hard to cultivate or difficult to synthesise chemically. Therefore, an effort has been made to produce these compounds in vitro or increase their production through the metabolic engineering of plants.

Yazaki first discusses how in vitro culture can be exploited to produce secondary metabolites. Thereafter, the strategies for metabolic engineering—that is, the use of genetic engineering methods to manipulate metabolism—are discussed in detail. It is evident from the review that understanding the biochemistry of their biosynthesis is the first step towards engineering the production of metabolites. Metabolic engineering differs from most molecular farming techniques for therapeutic protein production, because the aim is to manipulate complex plant metabolic pathways and to express proteins that are unlikely to have an effect on the plant cell itself. The successes in metabolic engineering reported by Yazaki are impressive.

The most telling and inspiring result so far has been the creation of ‘Golden Rice’, a transgenic rice line that has been metabolically engineered to produce high levels of the vitamin-A precursor ββ-carotene. This result exemplifies the potential of metabolic engineering to turn plants into living pharmacies.

Green Vaccines

Vaccination is one of the most successful developments in preventive human healthcare. In their contribution, Mor and Mason describe how edible vaccines that can be delivered in food plants may make that protection more widely accessible.

Despite their efficacy, most modern vaccines, which are inactivated or attenuated strains of the pathogen delivered by injection, have limitations. This creates constraints on the use of vaccines because many pathogens are difficult, and prohibitively expensive, to culture. Second, the majority of vaccines are delivered to patients by injection. This requires skilled staff and a sophisticated medical infrastructure, which limits their use in the developing world. Therefore, alternative routes for vaccination as well as alternate sources of vaccination antigens have been explored. Orally delivered vaccines that contain a subunit of the pathogen and can elicit a protective immune response are one such alternative. Using transgenic plants for both production and delivery is the major focus of this chapter.

The authors provide a historical overview of the development of strategies for production of recombinant antigens in plants, most of which have been achieved in the laboratories of Arntzen and Mason. They describe their success in producing immune responses in humans with plant-produced hepatitis B surface antigen, the labile toxin B subunit of enterotoxigenic Escherichia coli and the capsid protein of Norwalk virus. Success with these proteins in Phase I/II trials has now prompted larger scale clinical trials. What is compelling is that their approach is successful for proteins from widely different pathogens and indicates that orally delivered vaccines may be successful against a wide range of pathogens. Mor and Mason then discuss the rapid increase of research in the edible-vaccine field and point out that plants can be used to create multicomponent vaccines that can protect against several pathogens at once. This is an aspect of the edible-vaccine approach that further strengthens its impact.

After discussing strategies for vaccine expression, the authors turn their attention to the use of orally delivered antigens both as immunocontraceptive vaccines and in the treatment of autoimmune diseases. They end their chapter with a careful discussion of where the technical challenges in edible-vaccine technology lie, and how they may be solved.

We share their view that orally delivered vaccines are a proven technology, which holds great promise for development within mainstream pharmaceuticals. By their use, entire populations in the developing world will be able to share the same protection from disease that we take for granted in industrialised countries.

Hijacking the Factory Management—Using Viral Vectors for Protein Expression

There are, of course, alternative ways of exploiting the protein synthesis machinery of plants for the production of foreign proteins. The earliest proponents of this process, however, were not plant molecular biologists, but rather the plant viruses that hijack plant cells and use them for self-reproduction. These viruses provide a direct method for the transient production of recombinant proteins in crop plants, and this is the subject of the chapter written by Grill (see Chapter 40). By inserting target genes into the viral genomes, the hijacker can be turned into a molecular farmer.

In discussing how plant expression can be achieved using stably transformed plants or transient viral expression, Grill uses as examples two plant-produced proteins that have already entered clinical trials in the United States. The protein made in stably transformed plants was a humanised version of the Avicidin antibody for the treatment of prostate cancer, while the transient viral expression approach was used for the production of personalised vaccines produced from the cancerous B cells of non-Hodgkins lymphoma patients. What these examples illustrate is the flexibility of the viral vector system. Essentially, using this approach, personalised plant-based production of treatments for individual patients becomes possible, which is not feasible using stably transformed plants.

Grill then gives a detailed description of how viral vectors work in practice. Most plant viruses have single-stranded positive sense RNA genomes, and those that are assembled into rod-like particles can accommodate large inserted genes because it is possible to increase the length of the virion. Using recombinant viruses to infect plants and then produce the chosen recombinant proteins is rapid, compared to standard transgenic approaches, and has been tested in the field since the early 1990s. Grill argues that the rapid high-level expression of proteins is what sets viral vectors apart from the classical methods. Scaling up the methods for plant inoculation with viral vectors from the laboratory to the field has involved innovative solutions. An example, shown in the review figures, illustrates plants being inoculated using high-pressure sprays containing an abrasive and the recombinant viruses. A further interesting aspect of viral vectors is that there is no requirement for the cultivation of transgenic plant lines in the field. However, proteins transiently produced by viral infection have to be harvested from green leafy tissue and cannot be stored as easily as the seeds produced by transgenic plants.

Grill puts the case, and we agree, that while viral vectors compare well with transgenic plants overall, their real advantage is the ability to produce proteins rapidly. Indeed, the use of viral vectors to produce patient-personalised vaccines against non-Hodgkins lymphoma is such a compelling prospect that we feel viral vectors must remain an important tool for farming molecular medicines in the future.

A Perennial Production System

In their contribution, D'Aoust et al. describe the use of perennial plants for the production of recombinant proteins. Perennial plants are interesting as production systems because, with several harvests possible from the same crop, the plants can be used for repeated extraction of the protein.

The authors argue the case for forage legumes and, in particular, discuss the merits of using alfalfa. Because of the difficulties that had to be overcome to enable protein expression and plant transformation, forage legumes are relatively new to molecular farming. The authors describe the advances in promoter technology that have led to the construction of efficient inducible or strong leaf-specific expression cassettes. The features of the available transformation methods are discussed for alfalfa, which was first transformed more than 15 years ago, with D'Aoust et al. stating that reliable and efficient Agrobacterium-mediated transformation has been made possible by the optimal choice of genetic background and the use of standardised transformation protocols.

Because alfalfa is a fodder crop for ruminants, high-protein-yield varieties have been obtained and large-scale protein extraction techniques have been established by the animal feed industry. The authors describe how they are adapting these techniques for the production of plant-derived vaccines and, through the description of a case study, illustrate successful antibody expression in alfalfa.

Harvesting Recombinant Proteins from Food Crops

Many recombinant proteins have been produced in those plants that are ideally suited for laboratory experiments, such as tobacco or Arabidopsis thaliana, but these are not suited to long-term or large-scale production. The global agricultural industry has focused on high-yield crops that have been optimised over generations to be cost-effective for the large-scale production of food, animal fodder and other products, such as fibres for clothing and pulp for paper-making. Therefore, research on molecular farming has moved towards evaluating these commercial species for the production of pharmaceuticals and other recombinant proteins.

In their contribution, Stoger et al. first discuss the nature of the recombinant proteins expressed in plants before progressing to a detailed discussion of the issues that are related to choosing the most suitable species for protein expression. These include a meticulous review of the decisions that need to be taken during protein-expression projects and an indication of where the potential advantages lie with each crop species. Their contribution provides a solid framework and entry point for researchers new to molecular farming.

Industrial ‘Plants’ of the Future

In her chapter, Hood describes the exploitation of transgenic plants as a production system for proteins that are of industrial importance. These include recombinant enzymes, the use of which would benefit industrial production processes that are currently based on synthetic chemistry. The high specificity and activity of enzymes is what makes them attractive to industry, and the argument presented in the contribution is that plants, with their high biomass and large-scale production potential, are ideal for mass production of these proteins. The central thesis is that using plants will reduce the cost of these proteins and make them available for a wide number of applications in industries where they were previously unaffordable.

The use of plants to produce industrially relevant enzymes has been shown to be practical in a number of species. For example, phytase and αα-amylase have been produced in tobacco. However, the author argues that the expression of proteins in leaf tissue is not ideally suited to molecular farming. This is primarily because of the expense and the difficulties involved in extracting the proteins from leaves. She discusses, therefore, using other plant species, such as alfalfa or oilseed crops, to produce industrially important proteins. The author makes the case for maize (Zea) as a production system, her well-reasoned argument being based on a comparison of the production costs of various crops. Although alfalfa has the greatest potential for the production of recombinant protein per hectare, this is offset by the need to extract the protein from leaf material. Of the crops surveyed, soybean has the lowest cost for protein production, but the methods required for transformation are labour intensive and expensive. The cereal crops, such as rice and wheat, are shown to have positive advantages for expression but, for pragmatic reasons, maize was their selected production platform. First, maize is the largest crop produced in North America and its two major advantages are the low cost and ease of large-scale production. The rationale for the use of maize rests on the sophisticated existing production and harvesting infrastructure and on the advantages of using a seed-based production system.

One contentious and difficult issue covered by Hood is that of containment, which is of serious concern given the large amounts of maize produced within the United States for both human and animal consumption. The author presents a series of measures that can be used to control inadvertent mixing of corn destined for food or fodder with the transgenic variety. It can be argued that this is where industry has to be at its most vigilant, because the consequences of transgenic crops entering the food chain are potentially very large. Overall, however, with close control over its use, maize could become an acceptable and profitable species for use as an ‘industrial plant’ of the future.

An Introduction to Industrial and Pharmaceutical Protein Production in Plants

Advances in plant molecular biotechnology do not merely mean that farmers and research scientists alike can now contemplate a more than adequate harvest from newly sown crops. During the last decade, transgenic plants have been shown to provide a practical and feasible method of producing recombinant proteins. This technology has now progressed to the point where plants can be used as a platform for the expression of proteins intended for use in the treatment, or diagnosis, of a number of diseases. Such proteins include recombinant antibodies, cytokines and blood substitutes.

This research area—the combination of molecular biotechnology and agriculture, which is referred to as ‘molecular farming’—focuses on producing valuable proteins in plants, and forms the subject of the following contributions to this handbook. Drawing on expertise from both industry and academia, we present reviews of both the plant species and strategies that are being used to transfer molecular farming from the research laboratory to the field.

It is widely predicted that the world capacity for recombinant protein production will soon be exceeded by the demand and that this demand will continue to increase. Yet, the justification for the use of plants for recombinant protein production may not be immediately obvious. While transgenic cell lines, animals and microbes will continue to have significant roles to play as expression hosts for recombinant pharmaceuticals, the future role of plants should not be underestimated. One reason for this is that plants constitute a mass-production platform that can be used for the economical, large-scale production of proteins for industrial use in processes that were previously unaffordable. This makes them particularly relevant for the production of the recombinant proteins that will be required to treat the diseases we shall be challenged by in the 21st century.

Molecular farming is a fast-developing research area where fundamental research into protein expression and purification is coupled to the practicalities of plant growth and harvesting. This is reflected in the following contributions of how to express proteins, where to express them and how to choose the most appropriate host plant for protein expression. At present, there is no consensus on either the ideal expression method or choice of species, which has, therefore, to be determined empirically in individual cases. It should be emphasised that crops producing recombinant proteins have been in commercial production since 1997 (see Chapter 41) and crops producing recombinant therapeutics have already entered clinical trials (see Chapter 40). The development of edible vaccines using plants, as described by Mor and Mason (see Chapter 39), will have a great impact on world health and protection from disease. It is certain that the impact of molecular farming will increase as the technique develops both scientifically and commercially.

Below, we briefly introduce the contributions to this section and focus on what we regard as the most interesting issues covered in each review.