Big progress for small subunits: new Rubisco mutants in Arabidopsis.

Nearly 75 years after Rubisco was first isolated from spinach leaves, the enzyme’s function and regulation remain an important research topic and longstanding target to improve plant productivity. In plants, Rubisco is a complex made up of eight large subunits (LSUs) encoded in the chloroplast genome (

Engineering increased photosynthetic capacity has shown improvements in both model and food crops under field conditions, reigniting strategies to optimize or redesign photosynthesis to increase crop yields sustainably (Kromdijk ; Ermakova ; South ). A persistent target for improvement aims to overcome the inefficiencies of the central carbon-fixing enzyme in photosynthesis, Rubisco. Rubisco has a slow catalytic rate and also catalyses a competing oxygenation reaction, promoting the energetically intensive photorespiration cycle. Accordingly, optimizing the regulation and amount of Rubisco can result in improved crop growth and yield (Parry ; Salesse-Smith ; Yoon ). In recent years, surveys of both plant species and other photosynthetic organisms indicate that there is also exploitable variation in Rubisco performance that could be an important target for crop improvement (Sharwood ; Young ; Orr ; Galmés ), but our understanding of the structure–function relationships underpinning this variation remains fairly cryptic, and limits our ability to mine this resource.

Evidence across plant and algal Rubiscos has demonstrated that structural changes in both the LSU, where the catalytic site resides at the interface between dimers, as well as the SSU can impact the Rubisco carboxylation rate, substrate specificity, and multimer assembly (well reviewed in the accompanying article). A long-awaited goal for the Rubisco research community was realized with the expression of a recombinant Arabidopsis Rubisco in Escherichia coli (Aigner ). This has accelerated the ability to explore structure–function relationships across Rubisco, and has in fact been harnessed to demonstrate that the kinetic performance of recombinant tobacco Rubisco varies depending on the SSU composition (Lin , Preprint). However, recombinant enzymes might vary slightly in performance due to as yet unreconciled post-translational modifications that occur in vivo, and cannot inform the impact of organ-specific localization or developmental-specific expression of specific Rubisco isoforms (Laterre ; Lin , Preprint). In Arabidopsis, the SSU-encoding gene family comprises four members: rbcS1A, rbcS1B, rbcS2B, and rbcS3B. Only one gene (1A) is expressed in root tips, while another (1B) is exclusively expressed in the lower side of the leaf, is unresponsive to light pulses, and lacks the light regulatory elements found in the promotors of the other three (Dedonder ; Sawchuk ). Since rbcS1B also shows the smallest contribution to overall SSU expression and has been replaced with a duplicate copy of rbcS2B in several global accessions, it is easy to discount any functional importance of this isoform (Schwarte and Tiedemann, 2011). However, the crystal structure of Arabidopsis Rubisco captured a homogenous SSU composition of only this supposed low abundant isoform (Valegård ). Khumsupan et al. were unable to isolate a 1B mutant in conjunction with any other SSU isoform in Arabidopsis, presenting compelling evidence that there is a key contributory role for this least expressed 1B isoform during early development.

Measuring Rubisco in the leaf

Perhaps the most exciting aspect of the plants generated by Khumsupan is their potential to be used to examine the impact of homogenous SSU populations on Rubisco activity in vivo. This bypasses not only issues of recombinant enzyme performance, but also the assumptions that in vitro assays make about in vivo conditions, such as pH, CO2 concentration, and molecular crowding environment at the site of carboxylation. The widely used parameterization of C3 photosynthesis in response to temperature has been determined using both antisense tobacco and Arabidopsis plants containing ~15% and 40% Rubisco content, respectively (Bernacchi ; Walker ). However, apart from reductions in Rubisco content, do all small subunit knockdowns have the same impact on enzyme performance? Although a full kinetic parameterization requires measurements of CO2 response curves over a range of oxygen partial pressures, the in vivo rate of Rubisco carbon assimilation (kcatCO2) may be estimated if both the maximum rate of carboxylation and the Rubisco content of the leaf are known (Box 1). This representation normalizes the carboxylation activities calculated by Khumsupan et al. to account for the amount of Rubisco in the leaves and suggests that, in the absence of differences in Rubisco activation, Rubisco carboxylation activity could be increased by 6–13% over the wild-type activity in 2b3b, 1a2b, and 1a3b plants (Box 1). As previously shown (von Caemmerer ; Walker ), rates of Rubisco carboxylation estimated in vivo are greater than the in vitro measurements from similar T-DNA insertion lines (Atkinson ), and probably drastically overestimate the in vivo activity of the 1a2b3b line containing only the 1B isoform. These differences could reflect unreconciled changes in activation state in the mutant lines, but this cannot explain the much higher estimate from line 1a2b3b. The Rubisco turnover rate required to maintain the leaf carboxylation rate (VCmax) observed for the 1a2b3b line is more than triple that of any other line. Though estimates of in vitro VCmax (i.e. Rubisco content×in vitro kcatCO2) are much lower than observed VCmax, suggesting that the in vitro kcatCO2 may be different for the 1B only isoform, they are unlikely to be as different as predicted here (Box 1). The discrepancy could reflect slight differences between the leaves sampled for biochemical analysis and those used for physiological measurements, and may be further compounded by the estimation method Khumsupan used to derive VCmax from their A/Ci curves.

Here, the maximum rate of carboxylation (VCmax) of lines described by Khumsupan et al. are plotted against Rubisco content. The dotted line represents the predicted VCmax at each site concentration for the estimated in vivo kcatCO2 inferred for wild-type (WT) plants. Barring differences in activation state between lines, in vivo kcatCO2 can be estimated as VCmax/[Rubisco], while in vitro VCmax can be estimated as [Rubisco]×in vitro kcatCO2 (3.5 s–1; Boyd ). All four mutant lines targeting more than one small subunit isoform fall above this line, which could be due to increased in vivo kcatCO2.

Estimated increases in the Rubisco carboxylation rate among the double mutants are small relative to the wild type and are obscured by the accompanying 37–61% reductions seen in leaf Rubisco content. However, when their impact is modelled based on similar Rubisco concentrations in the leaves, as could be achieved through simultaneous overexpression of specific Rubisco SSUs and accessory proteins (Salesse-Smith ; Yoon ), then the impact of these small differences in Rubisco activity on leaf carbon gain is apparent over a broad temperature range (Box 2). Apart from the triple mutant, this limited analysis suggests that Arabidopsis Rubiscos lacking the SSU 1A isoform may provide the most beneficial impact on modelled leaf assimilation, which is notable as this is the SSU isoform targeted via the antisense construct in plants used to derive our current in vivo Rubisco temperature responses in Arabidopsis (Walker ).

Theoretical impact of differences in Rubisco in vivo kcatCO2 on Rubisco-limited net photosynthesis if modelled at an equal leaf Rubisco concentration. Temperature responses are modelled here at 400 p.p.m. CO2 using the in vivo modelling parameters and temperature responses for Arabidopsis from Walker , and assuming an equal Rubisco site concentration, activation state, and stomatal and mesophyll conductance among all lines. The inset shows the response of the triple mutant generated and described by Khumsupan .

Future perspectives

In using CRISPR/Cas9 to establish plants with a homogenous Rubisco composition, Khumsupan have presented additional evidence that 1B has a greater impact on Rubisco than previously considered, which suggest as yet unexplored roles for Rubisco isoforms in plant development. Importantly, the generation of novel mutants described here offers a compelling new tool to address the longstanding question of the impact on Rubisco SSU composition on holoenzyme activity in vivo.

	Rubisco concentration (µM catalytic sites m–2)	V  Cmax (µmol CO2 m–2 s–1)	Estimated in vivo kcatCO2 (s–1)	Estimated in vitro VCmax (µmol CO2 m–2 s–1)
WT	10.0	55.5	5.5	35.0
2b3b	6.3	36.9	5.9	22.1
1a2b	5.8	36.3	6.2	20.3
1a3b	3.9	24.6	6.3	13.7
1a2b3b	0.29	5.4	18.6	1.02