RuBisCo

RuBisCO is the rate-limiting enzyme in plant photosynthesis. Under the present model for photosynthesis, it should be possible to increase CO2 fixation in C3 plants by about 20%, before entering RuBP- and Pi-limited phases (Sage, 1990; Sage et al., 1989). Since the PCR cycle is the major consumer of energy formed at the thylakoids (Heldt, 1997), alterations of the enzyme should guarantee that the PCR cycle would siphon off and productively utilize more energy with an improved enzyme. Several directions about how to accomplish such improvement have been discussed (Andrews and Whitney, 2003; Parry et al., 2003). However, another strategy would be to engineer a RuBisCO enzyme that continued to fix CO2 under drought conditions when stomata aperture is reduced. First, we need to know which partial reaction of the enzyme constitutes the limiting step and which residues might determine the enzymatic properties (Mauser et al., 2001). Second, based on the detection of naturally occurring RuBisCO enzymes that are superior to the plant enzyme, work may be directed to replace resident rbcL (and rbcS) gene in plastid and nuclear DNA with the genes coding for the superior enzyme (Andrews and Whitney, 2003; Parry et al., 2003). Integration of the information from research with these superior enzymes suggests the possibility to engineer a higher plant rbcL gene that incorporates sequences responsible for improved RuBisCO performance. However, incorporating such engineered chimeric genes into chloroplast DNA faces challenges and obstacles that need to be addressed.

Enzymatic Properties of RuBisCO
The turnover rate of catalysis in CO2 fixation by plant RuBisCO is as low as 3.3 s -1 per site (Woodrow and Berry, 1988). The rate is less than one-thousandth of the rate of triose phosphate isomerase, the reaction of which proceeds in a diffusionlimited manner (Morell et al., 1992). All RuBisCOs analyzed to date catalyze an oxygenase reaction in addition to the carboxylase reaction (Andrews and Lorimer, 1978). The Km values of plant RuBisCO for CO2 and O2 are close to the concentrations of these gases in water equilibrated at normal atmospheric pressure (Woodrow and Berry, 1988). These gases compete with each other for the accepter molecule, the endiolate of RuBP (Andrews and Whitney, 2003). The relative frequency of the carboxylation and oxygenation reactions can be expressed as Srel, that is, the ratio of the specificity of the carboxylase reaction to that of the oxygenase reaction (Laing et al., 1974). The ratio of the velocities of both reactions can be expressed as vc/vo = Srel · [CO2]/[O2], where vc and vo are the velocities of the carboxylase and oxygenase reactions, respectively, and Srel is (Vmax of carboxylase reaction/Km for CO2)/(Vmax of oxygenase reaction/Km for O2). Since the exact concentration of CO2 in the stroma has been estimated as 5–7 µM(Evans and Loreto, 2000), and the activation of RuBisCO in chloroplasts is not complete, only a quarter of the total RuBisCO molecules in the stroma can participate in CO2 fixation during active photosynthesis (McCurry et al., 1981). Thus, either conditions in the stroma are suboptimal with respect to the potential of RuBisCO’s performance, or the intrinsic enzymatic properties of RuBisCO are inadequate with respect to stromal gas concentrations. Evolutionarily, plants have counteracted these disadvantages by investing an inordinate amount of nitrogen in RuBisCO synthesis, up to a level at which the RuBisCO concentration reaches 50% of that of total soluble proteins or 0.2 g of RuBisCO protein ml -1 in the stroma (equivalent to 4 mM in the concentration of its active site) (Ellis, 1979; Yokota and Canvin, 1985). However, plants must still lose water from the leaf through the open stomata in order to incorporate enough CO2. On average, water loss through evaporation is 250- and 1000 times faster in both C4 and C3 plants than the rate of incorporation of CO2 through the stomata (Larcher, 1995).

An ideal RuBisCO that could make optimal use of the global environment in C3 plants would incorporate the following properties: a higher turnover rate, a higher affinity for CO2, and a higher Srel. In contrast, the photorespiratory carbon oxidation (PCO) cycle driven by the RuBisCO oxygenase reaction has been proposed to play an important role in several reactions that are quite possibly equally important: (1) salvaging 75% of the carbon deposited in 2-phosphoglycolate into PGA through the PCR cycle, (2) dissipating more energy than the PCR cycle during turnover and refixation of photorespired CO2, and (3) supplying glycine and serine (Douce and Heldt, 2000; Heldt, 1997). These points apply solely to C3 plants containing present-day RuBisCO.

To attempt to remove the oxygenase reaction from RuBisCO, even if possible, would be dangerous for plants, although a reduction in the concentration of O2 in the atmosphere increases net photosynthesis rate (Tolbert, 1994). However, the reduction decreases Je (RuBisCO) or the rate of utilization of electrons by the PCO cycle (Fig. 4.2). Figure 4.2B also shows that the significance of the PCO cycle increases with decreasing CO2 concentrations and, inversely, that increasing CO2 concentrations weaken the importance of the cycle. In addition, the fact that high CO2 concentration in the atmosphere increases plant productivity to some degree (Sage et al., 1989) supports the idea that the PCO cycle is dispensable for plants if the solar energy captured by chlorophyll is efficiently consumed by other metabolic events in chloroplasts. Under those conditions, serine and glycine are synthesized from PGA in metabolism through the glycolate pathway and/or phosphorylated serine pathway (Hess and Tolbert, 1966; Ho and Saito, 2001). RuBisCO of cyanobacteria does not meet two of the outlined three ideal conditions essential for desired plant photosynthesis (Badger, 1980). However, cyanobacteria grow photosynthetically, in the absence of a well-developed PCO cycle, but with the aid of an active CO2-pumping mechanism (Kaplan and Reinhold, 1999; Shibata et al., 2002).

FIGURE 4.2 Simulation of the rate of net photosynthesis and flux of electrons used by PCR and PCO cycles in the electron transport chain. The rates of the carboxylase (vc) and oxygenase (vo) reactions of RuBisCO are expressed as vc = (kc · [RuBisCO] Cc)/{Kc · (1 + Oc/Ko) + Cc} and vo = (ko · [RuBisCO] Oc)/{Ko· (1 + Cc/Kc) + Oc}, respectively, where kc, ko, Kc, and Ko are kcat’s of carboxylase and oxygenase reactions and Michaelis constants for CO<sub>2</sub> and O2, respectively (Miyake and Yokota, 2000). Oc and Cc are concentrations of O2 and CO<sub>2</sub>, respectively, around RuBisCO. [RuBisCO] is the mole number of the active sites of RuBisCO per unit leaf area. The rate of net photosynthesis (A) is expressed as follows: A= vc – 0.5vo - Rd = vc[1 – 0.5Oc/S<sub>rel</sub>·Cc] – Rd, where Rd is the rate of day respiration and was assumed as 0.5 µmol CO<sub>2</sub> m-2 s-1. The flux of electrons used by RuBisCO-related cycles in the electron transport chain, <em>J</em><sub>e</sub> (RuBisCO), corresponds to 4vc + 4vo. Light is assumed to be saturating for photosynthesis. (A) and (B) show the effects of lowering atmospheric O2 concentration on A and <em>J</em><sub>e</sub> (RuBisCO), respectively, in a C<sub>3</sub> plant undergoing photosynthesis with RuBisCO representative of the higher plant enzyme. The kinetic parameters of RuBisCO from C<sub>3</sub> plants were from the literature (Woodrow and Berry, 1988): S<sub>rel</sub>, 89; kc, 3.3 mol; CO<sub>2</sub> s-1 per site; ko, 2.2 mol CO<sub>2</sub> s-1 per site; Kc, 29.5 Pa; Ko, 43.9 kPa; [RuBisCO], 18.56 mmol catalytic site m-2. The concentration of O2 in the atmosphere was assumed to be 21 (circles) and 2 kPa (squares). The effects of variations in kinetic parameters of RuBisCO on A and <em>J</em><sub>e</sub> (RuBisCO) are simulated in (C) and (D), respectively. Parameters for simulations are the same as those in (A) and (B) except that S<sub>rel</sub> were varied as indicated below and [RuBisCO] was 9.28 µmol catalytic site m-2. Enzymatic properties of RuBisCO are changed as follows: Circles, S<sub>rel</sub>, 89, kc, Kc, ko, Ko; squares, S<sub>rel</sub>, 180, 2kc, Kc, ko, Ko; lozenges, S<sub>rel</sub>, 180, kc, 0.5Kc, ko, Ko; open triangles, S<sub>rel</sub>, 360, 2kc, 0.5Kc, ko, Ko; closed triangles, S<sub>rel</sub>, 360, kc, Kc, 0.5ko, 2Ko.
FIGURE 4.2 Simulation of the rate of net photosynthesis and flux of electrons used by PCR and PCO cycles in the electron transport chain. The rates of the carboxylase (vc) and oxygenase (vo) reactions of RuBisCO are expressed as vc = (kc · [RuBisCO] Cc)/{Kc · (1 + Oc/Ko) + Cc} and vo = (ko · [RuBisCO] Oc)/{Ko· (1 + Cc/Kc) + Oc}, respectively, where kc, ko, Kc, and Ko are kcat’s of carboxylase and oxygenase reactions and Michaelis constants for CO2 and O2, respectively (Miyake and Yokota, 2000). Oc and Cc are concentrations of O2 and CO2, respectively, around RuBisCO. [RuBisCO] is the mole number of the active sites of RuBisCO per unit leaf area. The rate of net photosynthesis (A) is expressed as follows: A= vc – 0.5vo - Rd = vc[1 – 0.5Oc/Srel·Cc] – Rd, where Rd is the rate of day respiration and was assumed as 0.5 µmol CO2 m-2 s-1. The flux of electrons used by RuBisCO-related cycles in the electron transport chain, Je (RuBisCO), corresponds to 4vc + 4vo. Light is assumed to be saturating for photosynthesis. (A) and (B) show the effects of lowering atmospheric O2 concentration on A and Je (RuBisCO), respectively, in a C3 plant undergoing photosynthesis with RuBisCO representative of the higher plant enzyme. The kinetic parameters of RuBisCO from C3 plants were from the literature (Woodrow and Berry, 1988): Srel, 89; kc, 3.3 mol; CO2 s-1 per site; ko, 2.2 mol CO2 s-1 per site; Kc, 29.5 Pa; Ko, 43.9 kPa; [RuBisCO], 18.56 mmol catalytic site m-2. The concentration of O2in the atmosphere was assumed to be 21 (circles) and 2 kPa (squares). The effects of variations in kinetic parameters of RuBisCO on A and Je (RuBisCO) are simulated in (C) and (D), respectively. Parameters for simulations are the same as those in (A) and (B) except that Srel were varied as indicated below and [RuBisCO] was 9.28 µmol catalytic site m-2. Enzymatic properties of RuBisCO are changed as follows: Circles, Srel, 89, kc, Kc, ko, Ko; squares, Srel, 180, 2kc, Kc, ko, Ko; lozenges, Srel, 180, kc, 0.5Kc, ko, Ko; open triangles, Srel, 360, 2kc, 0.5Kc, ko, Ko; closed triangles, Srel, 360, kc, Kc, 0.5ko, 2Ko.

These considerations teach us that C3 plants are able to grow photosynthetically using RuBisCO with or without a much slower oxygenase reaction. In this case, some conditions must be met. The Srel value is the ratio of specificity of the carboxylase reaction to that of the oxygenase reaction, and is varied by changing either or both of the specificities of the reactions. An increase in Srel by increasing the turnover rate of the carboxylase reaction and the affinity for CO2 twofold over that of the wild-type enzyme causes photosynthesis and Je (RuBisCO) to increase (Fig. 4.2C and D). In contrast, RuBisCO with a higher Srel value attained by lowering the specificity of the oxygenase reaction results in increased photosynthesis (Fig. 4.2C), but Je (RuBisCO) is lowered (Fig. 4.2D). Plants containing RuBisCO manipulated to have such properties would be distressed by excess energy in high light intensities. However, this does not entail that photorespiration is completely indispensable for C3 plants. If the excess energy caused by lowering the specificity of the oxygenase reaction could be used by the PCR cycle, that is, if the specificity of the carboxylase reaction were increased to a level equal to or greater than the point where the excess energy is compensated by the PCR cycle, such a RuBisCO enzyme would improve C3 photosynthesis without excess-light stress.

Naturally Occurring Diversity in RuBisCO Kinetics
RuBisCO homologues are widely distributed among organisms and have been classified into four forms (Hanson and Tabita, 2001). Form I consists of eight large and eight small subunits of about 53 and 13 kDa, respectively, and is widely distributed among photosynthetic organisms such as higher plants, green algae, chlorophyll b-less eukaryotic algae, and autotrophic proteobacteria. Form II is composed only of the large subunits and is found in some eukaryotic algae, such as dinoflagellates, and photosynthetic proteobacteria. Form III is composed of only large subunits that are intermediates between Forms I and II, and is found in some Archaea (Ezaki et al., 1999; Finn and Tabita, 2003). All three forms possess the amino acid residues known to be essential for catalysis of RuBisCO and, in fact, catalyze both carboxylation and oxygenation of RuBP. RuBisCO homologues found in Bacillus subtilis, Chlorobium tepidum, and Archaeoglobus fulgidus are classified as Form IV based on their primary sequences (Hanson and Tabita, 2001). Form IV lacks up to half of the amino acid residues essential for RuBisCO classical catalysis, and, in fact, has no RuBP-dependent CO2-fixation activity. The exact function of Form III RuBisCO of Archaea is not known, while the RuBisCO homologue in B. subtilis catalyzes the 2,3-diketo-5-methylthiopentyl-1-phosphate enolase reaction in the methionine salvage pathway (Ashida et al., 2003, 2005; Sekowska et al., 2004). Form II RuBisCO of Rhodospirillum rubrum has the ability to catalyze the same reaction at a much slower rate. It has been suggested that the Form IV enzyme may be an ancestor of photosynthetic RuBisCO (Ashida et al., 2003, 2005).

The Srel value of Form I RuBisCO enzymes from cyanobacteria and γ-proteobacteria is around 40 (Roy and Andrews, 2000; Uemura et al., 1996). The Km for CO2 of the cyanobacteria enzyme is 250 µM, the highest value among RuBisCO enzymes examined so far (Badger, 1980). The Srel value is around 60 for RuBisCO from green microalgae, around 70 in conjugates and green macroalgae, and 85–100 in higher plants (Uemura et al., 1996). β-Proteobacteria, and micro- and macroalgae in which an accessory pigment chlorophyll b is replaced by bile pigments, possess Form I RuBisCOs. These are developed from an ancestor separate from those that evolved into the higher plant enzyme through cyanobacterial and γ-proteobacterial ancestors in the phylogenetic tree of the primary sequence of the large subunit proteins. RuBisCOs grouped in the nongreen Form I branch have higher Srel values than those grouped with the higher plant enzymes (green FormI RuBisCO) (Uemura et al., 1996). One extreme is the nongreen FormI enzyme from a thermoacidophilic alga, Galdieria partita (Uemura et al., 1997). The Srel and Km for CO2 values are 238 and 6.6 µM at 25 °C, but the Srel value decreases to 80 at 45 °C (its growth temperature). The protein structure of this enzyme has been resolved at 2.4 Å (Sugawara et al., 1999). The high Srel value has been proposed to be due to the stabilization of a loop partially covering the active site, loop 6, by hydrogen bonding between the main chain oxygen of ValL-332 and amido group of GlnL-386 (the numbering of amino acid residues follows the sequence of spinach RuBisCO, and the superscript indicates a large subunit residue) (Okano et al., 2002). Generally speaking, for Form I RuBisCOs, an enzyme having a higher Srel value and a lower Km for CO2 has a lower turnover rate and vice versa (Andrews and Lorimer, 1981). The Srel value of Form II RuBisCOs is the lowest among all known enzymes, and it is possible that the assembly with small subunit proteins may be important to increase the value (Andrews and Lorimer, 1981). An exception is known in the Pyrococcus kodakaraensis KOD1 Form III RuBisCO, in which five L2 dimers make up the enzyme without any small subunits (Kitano et al., 2001). The Srel value in this enzyme has been reported as 300 at 90 °C but is 80 at 25 °C (Ezaki et al., 1999).

The turnover rate of RuBisCO varies according to the source organism. The plant enzyme is one of the slowest catalysts, RuBisCOs from cyanobacteria and photosynthetic bacteria have a rate of 8–12 s -1 per site (Badger and Spalding, 2000), while the green algal enzymes occupy an intermediate position (Seemann et al., 1984). The highest turnover rate has been recorded as 20–21 s -1 per site for a Form III RuBisCO from A. fulgidus (Finn and Tabita, 2003).

During the era in which photosynthetic bacteria and cyanobacteria evolved the PCR cycle and the RuBisCO enzyme, the earth’s atmosphere contained high concentrations of CO2 with a marginal level of oxygen (Badger and Spalding, 2000). Over time, CO2 concentration decreased and the atmospheric oxygen concentration increased as a result of photosynthesis, initially by cyanobacteria and later by green algae. Cyanobacteria seem to have optimized a ‘‘CO2-pumping mechanism’’ in preference over improving RuBisCO. The evolution in green algae moved partly toward improved RuBisCO properties and partly toward a mechanism that concentrated CO2 in chloroplasts. Considering the properties of RuBisCOs of green algae, conjugates, and green macroalgae (Uemura et al., 1996), and since terrestrial plants lack the CO2-pumping system of cyanobacteria and algae, it is probable that higher plants could not be terrestrial until the Srel value reached 80 and the Km for CO2 was lowered to 15 µM. Apparently, the turnover rate was sacrificed in favor of development of properties that improved RuBisCO properties. Evolutionarily, higher plants responded to the selection pressure imposed by a change in [CO2] by moderately changing the structural gene sequence of rbcL, and compensated for the resulting disadvantages by developing a powerful promoter for the RuBisCO small subunit gene with changes in the small subunit protein that stabilized the L protein only a few hundred million years ago. Such compensation was necessarily incomplete since RuBisCO concentration in the stroma of algae was already high (Yokota and Canvin, 1985) because of the inherently slower turnover rate of this enzyme. There may still be room, however, to explore sequences of subunit proteins that exist in unexplored species, or to engineer sequence alterations that have not resulted from natural evolution. This is the research basis from which present and future protein engineering technology should succeed in improving the enzymatic properties of plant RuBisCO.

Engineered Improvements of RuBisCO Enzymatic Properties
In attempts to understand the structure–function relationships of RuBisCO, many amino acid residues in both subunit proteins have been manipulated in both Forms I and II (Hartman and Harpel, 1994; Parry et al., 2003; Spreitzer and Salvucci, 2002). In order to identify residues responsible for activity in one step of a sequence of partial reactions of RuBisCO, the chemical nature of the side chain of either the residue or the length of the side chain is changed. In another approach, alignments can be done of the primary sequences of more than 2000 varieties of large subunits and 300 varieties of small subunits (Spreitzer and Salvucci, 2002). This may either suggest which residue(s) or sequence(s) are responsible for a range in the Srel value from 10 to 238, in Kms for CO2 value from 6 to 250 µM, and kcat’s from 2.5 to 20 s -1 per site.

RuBisCO engineering depends on the synthesis of native recombinant proteins. Recombinant bacterial Forms I and II RuBisCOs can be synthesized in Escherichia coli (Hartman and Harpel, 1994). The genes for eukaryotic RuBisCOs can be transcribed in E. coli, but synthesized proteins aggregate rather than form the soluble, active enzyme (Gatenby et al., 1987). This is thought to be due, at least in part, to the fact that large subunit proteins of the eukaryotic Form I RuBisCO are insoluble in the absence of the small subunit protein (Andrews and Lorimer, 1985), and partly due to E. coli chaperones being incompatible with large subunit proteins.

Engineering of an amino acid residue involved in a partial reaction step generally causes a loss in activity of the recombinant enzyme. Nevertheless, there are several instances in which RuBisCO properties have been successfully changed. These engineering successes could point toward rational engineering strategies for the improvement of plant photosynthesis in the near future. The recombinant Form II RuBisCO of R. rubrum in which SerL-379 is replaced by Ala shows no oxygenase activity, although the turnover rate in the carboxylase reaction decreases to less than one-hundredth of the wild-type enzyme (Harpel and Harman, 1992). The function of this residue has been confirmed using Form I RuBisCO from the cyanobacterium Anacystis nidulans (Lee and McFadden, 1992). The 21st and 305th residues of plant RuBisCOs are conserved lysines, which are replaced by arginine residues in many bacterial and algal enzymes (Uemura et al., 1998). Simultaneously changing ArgL-21 and ArgL-305 of Form I RuBisCO of the photosynthetic γ-proteobacterium Chromatium vinozum to lysine residues resulted in an increase of the turnover rate from 8 to 15.6 s -1 per site with a concomitant increase in Km for CO2 from 30 to 250 µM (Uemura et al., 2000).

The exact function of small subunit proteins in Form I RuBisCO is still unclear (Spreitzer, 2003). However, many residues in small subunits have been modified, resulting in altered catalysis of the holoenzyme, although no small subunit residue is located close to the active site on the large subunit proteins (Spreitzer, 2003). The most striking improvement was achieved by changing ProS-20 to alanine in the cyanobacterium Synechocystis sp., with the Srel value increasing from 44 in wild-type to 55 in the mutated enzyme without any change in the turnover rate (Kostiv et al., 1997). The engineered IleS-99-Val RuBisCO of the cyanobacterium had a higher affinity for CO2 with no change in the Srel value and a decrease in turnover rate (Read and Tabita, 1992a). Either GlyS-103- Val or PheS-104-Leu cause small increases both in the Srel value and the affinity for CO2. RuBisCO of diatoms belongs to red-Form I with an Srel value over 100. A hybrid enzyme composed of the large subunit of Synechococcus and the small subunit from a diatom Cylindrotheca exhibits a 60% increase in Srel compared to the original cyanobacterial enzyme (Read and Tabita, 1992b).

Obstacles to be Resolved for RuBisCO Engineering
RuBisCO engineering has not yet succeeded in increasing Srel values for cyanobacterial and Chlamydomonas RuBisCOs to levels observed in plant enzymes but the knowledge gained from engineering these enzymes has provided a blueprint to be applied to higher plant RuBisCO enzymes. This is expected to become possible because of our ability to manipulate the higher plant rbcL gene by chloroplast DNA transformation (Kanevski et al., 1999; Svab and Maliga, 1993; Whitney et al., 1999). Combination of this technical advance with the discovery of a RuBisCO enzyme with an extreme Srel value provides an important new start point for improving plant RuBisCO and thereby alters plant productivity (Whitney et al., 2001). The obstacles that still stand in the way are addressed here in a discussion of three strategies directed at changing the enzymatic properties of plant RuBisCO by genetic engineering.

The first strategy will be to introduce multiple mutations into higher plant rbcL genes, and then return the modified genes to their original locus in chloroplast DNA in a high-throughput fashion. This will circumvent the problem of either insolubility of large subunit proteins from higher plants in E. coli (Gatenby et al., 1987) or the stroma of Chlamydomonas chloroplasts (Kato and Yokota, unpublished). While chloroplast transformation schemes are time consuming, the magnitude of the problem and the potential benefit resulting from successful engineering justify such efforts. That this is possible has been documented. Tobacco rbcL has already been engineered resulting in a reduction of Srel and has been exchanged with the original rbcL in the tobacco chloroplast genome (Whitney et al., 1999). The characteristics of photosynthetic CO2 fixation of the transformant were consistent with Farquhar’s photosynthetic simulation model (Whitney et al., 1999).

A second strategy will be to clone genes for both large and small subunits for a RuBisCO, which is superior in Srel and Km for CO2, and introduce them into the rbcL locus of chloroplast DNA of the target plant. In a pioneering study to express the Form II RuBisCO gene from R. rubrum in tobacco chloroplasts, the foreign gene gave rise to an active enzyme (Whitney and Andrews, 2001a). However, the genes of cyanobacterial and Galdieria Form I RuBisCO did not result in soluble, active enzymes (Kanevski et al., 1999; Whitney et al., 2001). This lack of success has been ascribed to incompatibility between the foreign large subunit peptides, the resident small subunit proteins, and the system for folding of nascent peptides in tobacco chloroplasts.

A third strategy addresses a different topic. Information on mechanisms involved in protein synthesis and folding in chloroplasts is still fragmentary (Houtz and Portis, 2003; Roy and Andrews, 2000), and our lack of knowledge of the precise mechanisms thus impedes the successful manipulation of RuBisCO genes in plants. For example, synthesis of the large subunit was formerly believed to take place on stromal free polysomes (Minami and Watanabe, 1984). However, recent work showed that a majority of the large subunits are translated by thylakoid-bound polysomes (Hatoori and Margulies, 1986). Since the large subunit itself is insoluble in an aqueous environment and translated on polysomes, one can expect the involvement of various chaperones in association with the polysomes. Otherwise, large subunit peptides in the process of translation and nascent large subunit peptides still in the process of synthesis would aggregate into an insoluble form. In this context, the observation (Amrani et al., 1997) of translational pausing on polysomes is intriguing. Nascent large subunits released from polysomes assemble with lipids or membranes, the fatty acid composition of which is quite different from that of thylakoids (Smith et al., 1997). Chaperonin-60 is known to bind at this stage to large subunit proteins (Gatenby and Ellis, 1990; Roy and Cannon, 1988; Smith et al., 1997). The holoenzyme may then be assembled as an L8 core to which small subunit proteins are added, as in the case of the synthesis of cyanobacterial RuBisCO (Hebbs and Roy, 1993).

The chloroplast outer and inner envelope membranes have individual translocon complexes, Toc and Tic, respectively, that recognize and transfer precursor proteins synthesized in the cytosol (Jarvis and Soll, 2002). Precursor proteins in a plastid-targeting complex with Hsp-70 and other proteins are guided to Toc and incorporated through the Toc complex in an ATP/GTP-dependent manner (Schleiff et al., 2002). The precursor proteins are then passed to Tic. The transit sequence of the small subunit precursor is then cleaved and the N-terminal methionine of mature small subunits is methylated (Grimm et al., 1997). One Tic component, IAP100, associates with chaperonin-60 and methylated small subunits are passed to chaperonin-60 through IAP100 (Kessler and Blobel, 1996). The L8 core and the small subunit/chaperonin-60 complex meet to form the holoenzyme. The importance of small subunit methylation is emphasized by the fact that there is only limited incorporation into a holoenyzme of small subunits synthesized from a foreign rbcS gene in chloroplasts (Whitney and Andrews, 2001b; Zhang et al., 2002). However, successful accumulation of the RuBisCO protein has been achieved when the promoter of the chloroplast-located psbA gene and the 50-UTR-attached cDNA of a transcript encoding a small subunit protein was engineered into a transcriptionally active space of the chloroplast (Dhingra et al., 2004).

When rbcL and rbcS genes are coordinately expressed in E. coli, even in the presence of coexpressed chloroplast chaperonin-60, no holoenzyme is formed (Cloney et al., 1993). In addition to the involvement in RuBisCO assembly of known chaperonin proteins (Brutnell et al., 1999; Checa and Viale, 1997; Gutteridge and Gatenby, 1995; Ivey et al., 2000), there are probably several additional, still unknown, proteins in chloroplasts that participate in successful folding of the holoenzyme. Transcription and translation systems of chloroplasts are bacteria-like, and many foreign proteins can be synthesized and accumulated in an active form in chloroplasts (Daniell, 1999). One most important aspect requiring a solution is that the coordinate synthesis and assembly of RuBisCO subunit proteins is severely discriminated against by host chloroplasts of different species: chimeric rbcL/rbcS holoenzymes have not been reported.

In studying RuBisCO structure–function relationships, a tobacco rbcL insertion mutant has been useful (Kanevski and Maliga, 1994). In this study, the original chloroplast-localized rbcL gene was disrupted by insertion of a selection marker gene, aadA, into the gene. The rbcL-deficient transformant is then transformed with a different rbcL sequence fused at its N-terminus to a chloroplast transit peptide sequence under the control of a nuclear promoter. Another useful mutant plant is a tobacco mutant, SP25, where GlyL-322 has been replaced by serine (Shikanai et al., 1996), which led to dysfunctional assembly of the holoenzyme and only a small amount of RuBisCO accumulated in an aggregated form in the stroma (Foyer et al., 1993). An engineered rbcL gene may then be introduced into chloroplast DNA of SP25.

A serious obstacle to plant RuBisCO engineering had been the difficulty in chloroplast transformation in any major crop plant. Efficient chloroplast transformation has in the past been restricted to some species in the Solanaceae, that is, tobacco (Svab and Maliga, 1993), potato (Sidorov et al., 1999), and tomato (Ruf et al., 2001). However, recent success appears to have been achieved with chloroplast transformation in crop species (Daniell et al., 2005).