From chaperonins to Rubisco assembly and metabolic repair

Abstract

Ribulose‐1,5‐bisphosphate carboxylase/oxygenase (Rubisco) mediates the fixation of atmospheric CO₂ in photosynthesis by catalyzing the carboxylation of the 5‐carbon sugar ribulose‐1,5‐bisphosphate (RuBP). Despite its pivotal role, Rubisco is an inefficient enzyme and thus has been a key target for bioengineering. However, efforts to increase crop yields by Rubisco engineering remain unsuccessful, due in part to the complex machinery of molecular chaperones required for Rubisco biogenesis and metabolic repair. While the large subunit of Rubisco generally requires the chaperonin system for folding, the evolution of the hexadecameric Rubisco from its dimeric precursor resulted in the dependence on an array of additional factors required for assembly. Moreover, Rubisco function can be inhibited by a range of sugar‐phosphate ligands. Metabolic repair of Rubisco depends on remodeling by the ATP‐dependent Rubisco activase and hydrolysis of inhibitors by specific phosphatases. This review highlights our work toward understanding the structure and mechanism of these auxiliary machineries.

Abbreviations

AAA+

ATPases associated with various cellular activities

Raf

Rubisco accumulation factor

Rca

Rubisco activase

RbcL

Rubisco large subunit

RbcS

Rubisco small subunit

Rubisco

Ribulose‐1,5‐bisphosphate carboxylase/oxygenase

RuBP

ribulose‐1,5‐bisphosphate

XuBP

xylulose‐1,5‐bisphosphate.

Introduction: The Rubisco Enzyme

Ribulose‐1,5‐bisphosphate carboxylase/oxygenase, commonly known as Rubisco,1 is one of the most abundant proteins in the biosphere.2 Rubisco evolved 2–3 billion years ago and is found in cyanobacteria, algae and plants.3 Life on earth is dependent on the ability of these organisms to transform inorganic carbon from CO₂ into biomolecules. This process is called photosynthesis (Fig. 1): it uses the energy derived from sunlight to generate the energy‐rich molecules NADPH and ATP with the consumption of water and CO₂, resulting in the production of usable sugars and O₂ as a waste product. Thus, photosynthesis not only supplies the chemical building blocks of life, but also produces the oxygen we breathe. Rubisco is the key enzyme of the Calvin‐Benson‐Bassham (CBB) cycle of photosynthesis, mediating the fixation of atmospheric CO₂ by catalyzing the carboxylation of its five‐carbon sugar substrate ribulose‐1,5‐bisphosphate (RuBP).

Figure 1.

The process of photosynthesis converts sunlight into chemical energy, splits water to liberate O₂, and fixes CO₂ into sugar. The light reactions provide energy and reducing agents (ATP and NADPH), which are used in the light‐independent Calvin‐Benson‐Bassham (CBB) cycle. The enzyme Rubisco catalyzes the key step of fixation of atmospheric CO₂ by mediating the carboxylation of the 5‐carbon sugar substrate ribulose‐1,5‐bisphosphate (RuBP)

There are two distinct structural forms of Rubisco in photosynthetic bacteria and eukarya.4 Form I is a protein complex of ∼550 kDa, consisting of eight large (RbcL; ∼55 kDa) and eight small subunits (RbcS, ∼15 kDa) (Fig. 2A). Four RbcL antiparallel dimers form the core of the complex with four RbcS subunits located on top and four at the bottom. Each antiparallel RbcL dimer has two active sites at the interface for the binding of the substrate RuBP (Fig. 2). The simpler form II consists only of an antiparallel dimer of RbcL subunits (Fig. 2B). The initial discovery of form I Rubisco in plant leaves dates back to 1947, while the form II from the proteobacterium Rhodospirillum rubrum was first described only in 1974 (see ref. 7 for a historical perspective).

Figure 2.

Rubisco structures: (A) Structure of hexadecameric form I Rubisco from spinach (PDB: 1RCX).5 Side and top views shown in surface representation (RbcL subunits alternating in white and gray; RbcS subunits in yellow), with one antiparallel RbcL dimer (RbcL in green and RbcL' in blue) and the adjacent RbcS subunits in the side‐view shown in ribbon representation. The antiparallel RbcL dimer is shown with RuBP (in red) bound in the active sites. (B) Structure of form II Rubisco from the proteobacterium R. rubrum (PDB: 9RUB).6 The antiparallel RbcL dimer with RuBP bound in the active sites is shown in ribbon representation and colored as in (A)

Despite its fundamental role in biology, plant Rubisco is a remarkably inefficient enzyme with a catalytic rate of only 2–5 CO₂ molecules fixed per sec.8 Moreover, Rubisco can also accept oxygen as a substrate instead of CO₂—a vestige of having evolved in an atmosphere free of oxygen and rich in CO₂.9 Thus, Rubisco is an obvious target for reengineering to increase crop yields.10, 11, 12

The biogenesis of Rubisco has long been known to require the chloroplast chaperonin Cpn60 or its bacterial homolog GroEL.13, 14 I became interested in the biogenesis of Rubisco when we used the form II RbcL from the proteobacterium R. rubrum in our mechanistic studies of chaperonin‐assisted folding.15 While we could efficiently reconstitute the form II Rubisco with unfolded RbcL subunits and chaperonin, the form I Rubisco stubbornly resisted all efforts of in vitro reconstitution. I am grateful to John Ellis for pointing out the significance of this problem during a phone conversation in 2004, he said to me “Manajit you have not solved the initial problem and you have to do it.” This provoked my interest and ever since part of my research group has been working on understanding the cellular machinery for the folding and assembly of form I Rubisco. Here, I describe, from a personal perspective, the findings we have made during this exciting journey over the past 10 years.

Chaperonin Requirement for Folding of the RbcL Subunit

First evidence for the involvement of molecular chaperones in the assembly of Rubisco was reported in 1980 by Barraclough and Ellis, when they found that newly‐synthesized RbcL subunits in chloroplasts bound to a large protein complex of ∼800 kDa.16 This RbcL binding protein was later found to be homologous to the GroEL protein in bacteria17 and to Hsp60 in mitochondria,18 which are classified as chaperonins.17 A year later Ulrich Hartl and Art Horwich showed that Hsp60 was required for protein biogenesis in mitochondria by mediating protein folding in an ATP‐dependent mechanism.19, 20 The finding that a monomeric protein like dihydrofolate reductase utilized Hsp60 for folding provided definitive evidence for the new paradigm that cellular protein folding is a chaperone‐assisted process.20 Soon after, Lorimer and colleagues reconstituted the form II Rubisco of R. rubrum with the bacterial chaperonin system, GroEL and its cofactor GroES.21

Mechanism of Chaperonin‐Assisted Folding

Having been trained as a chemist and biophysicist, my interest in molecular chaperones, and the chaperonins in particular, was stimulated by marriage to Ulrich Hartl in 1987. I joined his research group in 1991 when we moved from Munich to the Sloan‐Kettering Institute in New York. For the next ∼15 years (since 1997 at the Max Planck Institute of Biochemistry) my research focused on elucidating the fascinating mechanism by which the bacterial chaperonin system assists protein folding (reviewed in ref. 22).

GroEL is a cylindrical complex of ∼800 kDa, composed of ∼60 kDa subunits arranged as two heptameric rings stacked back to back. Each subunit is composed of three domains: an equatorial ATP‐binding domain, an intermediate hinge domain, and an apical domain that forms the ring opening. The apical domains expose hydrophobic amino acids toward the ring center for binding a non‐native substrate protein (SP) or its cofactor GroES. GroES is a heptameric ring of ∼10 kDa subunits that binds to the ends of the GroEL cylinder and undergoes binding and release cycles regulated allosterically by the GroEL ATPase.22 Obligate substrates of GroEL, such as RbcL, typically populate kinetically‐trapped folding intermediates that are aggregation‐prone and bind to GroEL via exposed hydrophobic amino acid residues. The folding cycle starts when a nonnative SP binds to the open ring of the GroEL:GroES complex (Fig. 3A). ATP‐dependent binding of GroES to the SP containing ring results in the release of bound SP into an enclosed cage formed by one ring of GroEL and the GroES lid (the so‐called cis cavity) (Fig. 3A). ATP and GroES binding is accompanied by dramatic conformational changes in the interacting GroEL subunits, creating an enlarged, hydrophilic cage that is permissive for folding proteins up to ∼60 kDa. The SP is allowed to fold inside the cage for the time it takes the GroEL cis‐ring to hydrolyze its seven ATP molecules (a few seconds, depending on the temperature). Following hydrolysis, ATP binding to the trans‐ring sends a negative allosteric signal to the cis‐ring causing the release of ADP, dissociation of GroES and release of native SP (Fig. 3A).

Figure 3.

The GroEL/GroES chaperonin system. (A) Folding in the GroEL:GroES chaperonin cage. Substrate protein (SP) binding to GroEL may result in local unfolding.23 ATP binding then triggers a conformational rearrangement of the GroEL apical domains. This is followed by the binding of GroES (forming the cis complex) and SP encapsulation for folding. At the same time, ADP and GroES dissociate from the opposite (trans) GroEL ring, allowing the release of SP that had been enclosed in the former cis complex (omitted for simplicity). SP remains encapsulated, free to fold, for the time needed to hydrolyze the 7 ATP molecules in the newly‐formed cis complex (∼2–7 s, dependent on temperature). Binding of ATP and GroES to the trans ring causes the opening of the cis complex. Structural model based on PDB 1AON. Figure reproduced from ref. 24. (B) Accelerated chaperonin‐assisted folding as a result of confinement. Simple energy funnel diagrams are shown for a protein populating a kinetically trapped intermediate(s) during spontaneous folding that converts slowly to the native state (left). For the GroEL SPs that have been characterized, no significant population of fast folding molecules (dotted arrow) was observed to bypass the kinetically trapped intermediate.15, 25, 26, 27 This is due to a high entropic component to folding energy barrier. Confinement of SP in the hydrophilic environment of the GroEL:GroES cage (right) is suggested to avoid the accumulation of entropically stabilized intermediate. This is reflected in a smoothing of the energy landscape (red line). Figure modified from ref. 15

To understand the chaperonin mechanism it was important to distinguish its function in preventing protein aggregation from a possible role in actively promoting folding. The simpler dimeric form II Rubisco from R. rubrum allowed us to address this problem, because it can fold spontaneously at low concentration and temperature, conditions that minimize aggregation. In 2001, we showed that RbcL subunit folding inside the chaperonin cage is significantly faster than folding in free solution in the absence of aggregation.15 This led us to propose the concept that confinement of a dynamic folding intermediate in the chaperonin cage reduces the entropic penalty of folding, increasing the flux toward the native state. Subsequent studies with other SPs and at single‐molecule level confirmed the ability of GroEL:GroES to accelerate folding.23, 25, 26, 27, 28, 29, 30

We identified three features of the cis‐cavity to be crucial in accelerating folding: the volume of the cage relative to the size of SP;27 the net‐negatively charged cavity wall;25, 27, 28, 29 and the flexible C‐terminal Gly‐Gly‐Met repeat sequences that extend from the equatorial domains into the central cavity.27, 31 Theory predicts that confinement of a non‐native protein in a repulsive (charged) cage can accelerate folding by one to two orders of magnitude. It does this by restricting the conformational freedom of dynamic folding intermediates and rendering the formation of local and long range contacts, including those present in the folding transition state, more favorable.32, 33, 34 This mechanism would effectively smoothen the folding energy landscape (Fig. 3B). Indeed, we measured a rate acceleration of ∼10‐fold over the spontaneous folding rate for a model SP, DM‐MBP (double mutant maltose‐binding protein),25, 27 and up to ∼100‐fold (at 37°C in single molecule measurements) for the obligate in vivo substrate dihydrodipicolinate synthase (DapA),26 a TIM‐barrel protein like the RbcL subunit of Rubisco. Measurements of photoinduced electron transfer coupled to fluorescence correlation spectroscopy (PET‐FCS) subsequently provided direct evidence for the confinement effect on chain mobility during folding inside the GroEL:GroES cage.28

Assembly Chaperones of Form I Rubisco

The folded RbcL of form II Rubisco rapidly assembles to functional dimers upon release from chaperonin, allowing in vitro reconstitution.15 In contrast, the RbcL subunits of cyanobacterial form I Rubisco failed to assemble spontaneously into RbcL₈ core particles, even in the presence of folded RbcS. This suggested that additional factors are required for the assembly of the structurally more complex form I enzyme. We found that this failure to assemble is due to intrinsic properties of form I RbcL and is overcome only with the assistance of specific assembly factors. By contrast, RbcS refolds spontaneously in vitro to a state competent to assemble with the RbcL₈ core particle.35

As a possible assembly factor we considered the protein encoded by the rbcX gene, which is located in the intergenic space between the rbcL and rbcS genes of several cyanobacterial species, and is conserved in all plants.36 Indeed, in 2007 we made the exciting finding that RbcX functions in mediating the assembly of cyanobacterial form I Rubisco upon recombinant expression.37 The crystal structure, solved by Andreas Bracher, showed that RbcX functions as a boomerang‐shaped homodimer of ∼15 kDa subunits, with a narrow hydrophobic groove (Fig. 4A). This groove could be shown to specifically bind the C‐terminal RbcL sequence motif, EIKFEFD, which is conserved in many cyanobacteria and all plants (Fig. 4B).36 While the interaction of RbcX with RbcL is normally dynamic, a heterologous complex of RbcL and RbcX proteins from different cyanobacterial species was stable.39 The crystal structure of this complex revealed eight RbcX dimers bound to the RbcL₈ core, with the RbcX dimers functioning as “molecular staples” clamping each RbcL₂ unit at the top and bottom (Fig. 4C). This is achieved by the RbcX central groove binding the C‐terminal motif of one RbcL subunit and conserved residues at the periphery of RbcX making contact with the adjacent RbcL within the RbcL₂ unit. Using RbcX and GroEL‐GroES, three years later we succeeded in reconstituting the folding and assembly of the cyanobacterial RbcL₈S₈ holoenzyme40—a major advance defining the paradigm of assisted oligomeric assembly.

Figure 4.

Rubisco assembly chaperones. (A) Molecular representation of the RbcX dimer from the cyanobacterium Synechococcus sp. PCC7002 with the protomers colored white and blue, respectively (PDB: 2PEI).37 (B) Zoom‐in on the central hydrophobic cleft of RbcX with bound C‐terminal peptide (EIKFEFD) of form I RbcL. N‐ and C‐termini of the peptide are indicated. The important residues involved in the interaction between peptide and RbcX are labeled. Dashed lines represent hydrogen bonds. (C) Model of RbcX‐assisted assembly of cyanobacterial form I Rubisco. Upon folding by the GroEL/GroES chaperonin system, the released RbcL subunit exposes the flexible C‐terminal RbcL peptide. Whether RbcX recognizes and binds the monomeric or the spontaneously dimerized RbcL subunits is not clear. The antiparallel RbcL dimer is stabilized by RbcX dimers acting as “molecular staples.” The RbcL₂RbcX₂ units subsequently assemble to the RbcL₈RbcX₈ complex, in which a large portion of the RbcS binding interface is pre‐formed. RbcS binding structures the RbcL N‐terminus and the “60ies loop” of RbcL, causing displacement of RbcX and formation of the functional Rubisco holoenzyme. (D) Model of Raf1‐mediated Rubisco assembly of cyanobacterial form I Rubisco. After folding by the GroEL/GroES chaperonin system, the released RbcL subunits either dimerize spontaneously or do so with assistance by Raf1. The antiparallel RbcL dimer is stabilized by Raf1 and is in dynamic equilibrium with higher oligomers up to RbcL₈Raf1₄ complexes. RbcS binding shifts the equilibrium toward holoenzyme formation. Figure reproduced from refs. 37 and 38

However, the reconstitution experiment reached only ∼50% efficiency and RbcX was not sufficient for the reconstitution of plant Rubisco, suggesting that additional factors were required. Indeed, seminal work from the group of David Stern identified several so‐called Rubisco accumulation factors (Raf) by screening a library of photosynthetically defective maize mutants.41 One of these putative assembly factors, Raf1, is conserved in all photosynthetic organisms containing RbcX.38 Remarkably, our biochemical analysis showed that Raf1 alone was highly efficient in mediating the assembly of cyanobacterial Rubisco.42 Raf1, like RbcX, functions as a dimer. The Raf1 subunit (∼40 kDa) consists of an N‐terminal α‐helical domain and a C‐terminal β‐sheet dimerization domain. Using a hybrid approach of X‐ray crystallography and electron microscopy, we found that Raf1, similar to RbcX, also acts in stabilizing the RbcL₂ unit. The β‐sheet domains are located at the equator of the RbcL₂ unit, with the α‐helical arms hugging the RbcL dimer at the top and bottom edges (Fig. 4D).42 The end‐state of Raf1‐mediated assembly is an RbcL₈ core with four Raf1 dimers bound (RbcL₈Raf1₄) from which Raf1 is displaced by RbcS to form the holoenzyme. Thus, both RbcX and Raf1 have similar roles but use different interaction sites on RbcL.42 Whether both factors co‐operate in vivo for efficient Rubisco assembly remains to be investigated.

Metabolic Repair of Rubisco

Once assembled, the Rubisco holoenzyme is structurally stable,8 but its troubles nevertheless continue. The multistep catalytic reaction of the enzyme is error‐prone, and can result in the formation of so‐called “misfire” sugar phosphates that block the active site43 (Fig. 5A). Inactive Rubisco is also generated when the natural substrate, RuBP, binds to a catalytic site pocket that has not been activated by carbamylation.49 Furthermore, some plants under low light conditions generate the inhibitor 2‐carboxy‐D‐arabinitol‐1‐phosphate (CAIP) to deactivate Rubisco when it is not needed (Fig. 5A).50 In all these cases reactivation requires that the inhibitory sugar molecule be removed from Rubisco by a class of AAA+ (ATPases associated with various cellular activities) chaperones, called Rubisco activases (Rca).38, 47, 51, 52 Rca has been known since the 1980s53 and was originally assumed to be restricted to plants. However, since our discovery of Rca in the proteobacterium Rhodobacter sphaeroides in 2011,44 several additional prokaryotic Rcas have been identified.54, 55, 56, 57 Rca proteins display considerable variability in sequence and mechanism of action, reflecting the constraints of co‐evolution with their cognate Rubisco substrate (reviewed in refs. 47 and 51). Here, I focus on our structural and mechanistic analysis of R. sphaeroides Rca (RsRca).

Figure 5.

Rubisco regulation by Rca. (A) Regulation of Rubisco activity and inhibition by sugar phosphates. E, the non‐carbamylated enzyme; ECM, the carbamylated and Mg²⁺ ion‐bound enzyme; EI, the sugar phosphate inhibited E form; ECMI, the inhibited ECM form; Rca, Rubisco activase. Xylulose‐1,5‐bisphosphate (XuBP) and 2,3‐pentodiulose‐1,5‐bisphosphate (PDBP) are misfire products, generated during the multistep catalytic reaction of Rubisco. The inhibitor 2‐carboxy‐D‐arabinitol‐1‐phosphate (CA1P) is synthesized by some plants under low light and referred to as “nighttime” inhibitor.43 (B) Crystal structure of the Rca monomer from R. sphaeroides (PDB: 3SYL)44 shown in ribbon representation, with the domain structure shown schematically below. The positions of the canonical pore loop, ADP (cyan) and the allosteric regulator, RuBP (yellow), are also indicated. N‐ext., N‐terminal extension of RsRca. (C) Top and side views of the RsRca hexameric model superposed on the electron microscopic reconstruction, with alternating subunits shown in shades of blue (EMDB EMD‐1932; PDB 3ZUH).44 (D) Model of the putative storage form of RsRca44 and its conversion to active hexamer. In the absence of photosynthetic activity (dark period), the concentration of free RuBP is low and Rca populates a helical assembly with no ATPase activity, avoiding unnecessary ATP consumption. Activation of photosynthesis results in the accumulation of free RuBP.45 Free RuBP binds to Rca, inducing its rearrangement to the catalytically competent hexamer. (E) Model of the mechanism of repair of inhibited Rubisco by RsRca. The active Rca hexamer interacts with inhibited Rubisco via its highly conserved top surface and transiently pulls the extended C‐terminal tail of the RbcL subunit into the central pore. This initial action is mediated by the ATPase activity of Rca and results in the destabilization of the Rubisco active site, facilitating release of inhibitory sugar phosphate. Rca and Rubisco (PDB: 4F0K)46 are displayed in surface representation, with the RbcL subunits in shades of green, RbcS subunits in orange and RsRca in blue. The RbcL C‐termini are drawn as lines in red. Figure modified from refs. 47 and 58

All Rcas share the core subunit architecture of AAA+ proteins, consisting of an N‐terminal nucleotide binding domain and a C‐terminal α‐helical domain (Fig. 5B). Like most AAA+ proteins, the Rca enzymes function as donut‐shaped, hexameric rings (Fig. 5C), with their central pore implicated in threading specific peptides of Rubisco.47, 51 In the case of the RsRca, formation of the active hexamer requires ATP and binding of RuBP in the C‐terminal α‐helical domain (Fig. 5B,C). Allosteric regulation by RuBP, the substrate of Rubisco, ensures that RsRca is functional only when photosynthesis is active. In the absence of photosynthesis or at low levels of RuBP, RsRca is ATPase inactive, forming spiral‐shaped high molecular weight assemblies that may represent a storage form (Fig. 5D). The generation of millimolar concentrations of RuBP upon activation of photosynthesis45 would induce conversion of the storage form to catalytically competent RsRca hexamers (Fig. 5D).44, 47 Recently, using a hybrid approach of hydrogen/deuterium exchange, chemical crosslinking and cryo‐electron microscopy, we showed that RsRca interacts side‐on with the cube‐shaped Rubisco complex, positioning Rca over one catalytic site (Fig. 5E).48 This topology allows the Rca hexamer to pull the flexible C‐terminal tail of RbcL, covering the catalytic site, into the central pore. Following this initial interaction, the conserved top surface of RsRca contacts both RbcL and RbcS subunits, mediating the opening of the catalytic site (Fig. 5E).48 Importantly, this repair mechanism preserves the structural integrity of the Rubisco complex, allowing Rca to remodel successively the inhibited catalytic centers of the enzyme with minimal overall structural perturbation.

Hydrolysis of Inhibitory Sugar Phosphates

Inhibitory sugar phosphates generally have high affinity for Rubisco and must be hydrolyzed upon release by Rca to prevent competition with RuBP.58 Notably, XuBP is formed at a much higher rate than other misfire products (Fig. 5A) and is a potent inhibitor of Rubisco even when present at 100‐fold lower concentration than RuBP.59, 60, 61 We recently identified the XuBP phosphatase (XuBPase) of R. sphaeroides as the product of the cbbY gene.61 This gene is conspicuously localized downstream of the gene for RsRca (cbbX) in the Rubisco operon.62 Homologs to the XuBPase protein are widely distributed among photosynthetic organisms including plants. Our structural and enzymatic analysis revealed how XuBPase distinguishes between XuBP and RuBP, which differ only by their stereochemistry at the C3 position.61

XuBPase is a two‐domain protein of ∼26 kDa, consisting of a core‐domain and a flexibly attached cap‐domain, with the active site pocket located at the interface (Fig. 6A). To understand the mechanism, we solved the crystal structure of an inactive mutant, which showed that XuBP binds in a distinct bent conformation, positioning the phosphate group at C1 for efficient hydrolysis (Fig. 6B).61 Modeling RuBP into the active site in the same conformation resulted in a steric clash, explaining the high‐affinity of the phosphatase for XuBP (K_m ∼30 μM) over RuBP (K_m ∼3 mM). Combined with a ∼50‐fold higher turnover rate for XuBP, XuBPase hydrolyzes micromolar concentrations of XuBP in the presence of millimolar RuBP. The resulting D‐xylulose‐5‐phosphate (Xu5P) enters the CBB cycle to be converted back to RuBP (Fig. 6C).61

Figure 6.

Structure and function of the phosphatase XuBPase. (A) Crystal structure of the catalytically inactive XuBPase mutant (D10N) from the proteobacterium R. sphaeroides (PDB 4UAT)61 with XuBP trapped in the active site pocket, located at the interface of the cap (yellow) and core (blue) domains. Bound XuBP is indicated in green and the catalytic Mg²⁺ ion is indicated by a sphere. (B) Hydrogen‐bond interactions of XuBP, bound in a bent conformation, with residues of the cap (yellow) and core (blue) domains. Residue Asp10 in the wild‐type enzyme is positioned for efficient hydrolysis of the C1 phosphate of XuBP. (C) Functional cooperation of XuBPase and Rca in maintaining Rubisco activity in the presence of XuBP. As soon as XuBP is released from the active site pocket of Rubisco by Rca, XuBPase hydrolyzes it to Xu5P (xylulose‐5‐phosphate). Xu5P is converted to RuBP by the action of Ru5P (ribulose‐5‐phosphate) epimerase and Ru5P kinase in the CBB cycle. Figure reproduced from ref. 61

Concluding Remarks

The necessity to meet the food demands of an increasing human population has re‐invigorated the interest in strategies to improve Rubisco performance. A better understanding of the cellular machineries and mechanisms involved in Rubisco biogenesis and metabolic repair holds the promise for new bioengineering strategies based on a rational design. However, many challenges remain. Perhaps most importantly, despite considerable efforts, it has not yet been possible to recombinantly produce plant Rubisco in E. coli or to reconstitute the holoenzyme in vitro. It seems that Rubisco still holds many secrets and I look forward to future studies uncovering them. As aptly stated by Bob Tabita, Rubisco keeps on giving.62