New challenges in modelling photosynthesis: temperature dependencies of Rubisco kinetics



This article comments on: Temperature response of in vivo Rubisco kinetics and mesophyll conductance in Arabidopsis thaliana: comparisons to Nicotiana tabacum

Photosynthesis models have become useful and important tools for predicting actual CO2 fluxes in ecophysiological studies at the leaf level (Diaz-Espejo, Nicolás & Fernández 2007; Greer & Weedon 2012), or in canopy models and global circulation ,models used to predict weather forecast and climate (Sellers et al. 1997; Zhu, Portis & Long 2004). But photosynthesis models should not be seen as a means for prediction only: they are powerful tools to understand the mechanisms involved in the regulation of photosynthesis and its relationship with water use by plants. In this sense, in order to cover all the goals, it is necessary to use process-based models in which the parameters involved have full physiological meaning. The most widely used model, which grouped all these requirements, is the Farquhar, von Caemmerer and Berry model (Farquhar, von Caemmerer & Berry 1980), hereafter FvCB model. In this model, it is feasible to include diverse aspects of the biochemistry of photosynthetic processes or diffusion resistances that are explored in other studies. Among these are the diffusional limitations imposed by mesophyll conductance to CO2 (Flexas et al. 2008, 2012), or the effects of Rubisco kinetics (Savir et al. 2010) and Rubisco regulation (Galmés et al. 2013). The paper in this issue of Plant, Cell & Environment by Walker et al. (2013), on the temperature response of in vivo Rubisco kinetics and mesophyll conductance to CO2 (gm) comparing Nicotiana tabacum and Arabidopsis thaliana, highlights the importance of considering interspecific variability in photosynthesis models. In the following sections, we will briefly review the effect of temperature on two of the aspects addressed by Walker et al. (2013), gm and Rubisco kinetics, whose actual impact has been somehow ignored to date in the application of the FvCB model of photosynthesis.

Response of gm to Temperature

Several studies have reported the effect of temperature on gm. The first piece of evidence of the effect of temperature was brought up by Bernacchi et al. (2002) in tobacco. It was followed by new data from other species (Diaz-Espejo et al. 2007; Egea et al. 2011; Evans & von Caemmerer 2013; Flexas et al. 2008; Scafaro et al. 2011; Warren 2008; Warren & Dreyer 2006; Yamori et al. 2006a). Comparing all these data, we can conclude that the response of gm to temperature is species-specific, as is clearly demonstrated by Walker et al. (2013) in their comparison of two species, using the very same methodology and experimental conditions. The inclusion of gm in the FvCB model, that is to consider a finite internal resistance to CO2, has two main implications. Firstly, the CO2 concentration at the Rubisco carboxylation site (Cc) can be estimated, and therefore Vcmax, the maximum Rubisco activity, can be calculated more realistically [CO2]. Secondly, this makes it necessary to use Cc-based parameters in the FvCB model, and therefore to include in any model the response of gm to temperature. As Warren & Dreyer (2006) pointed out, when using previous temperature response equations of Vcmax and Jmax (the maximum electron transport activity) on a Ci-basis, we are simply accepting the use of an apparent Vcmax and Jmax. The response of gm to temperature acquires more complexity in some species, where it depends on the growth temperature. Yamori et al. (2006a) reported a strong acclimation of gm to temperature in Spinacia oleracea, as did Flexas et al. (2008) in Brassica oleracea. Another acclimation process is the differences among populations of a single species. Silim, Ryan & Kubien (2010) studied the acclimation to different growing temperatures in two ecotypes of Populus balsamifera collected from a northern and a southern population. Their results showed that gm was relatively insensitive to temperature below 25 °C, although it declined at 37 °C in cool-grown plants. This highlights once more the need to improve our knowledge on the mechanisms regulating gm.

In the first report on gm responses to temperature Bernacchi et al. (2002) concluded that the process was mediated by a protein-facilitated process. Since then, an increasing number of papers have demonstrated the importance of anatomical features in gm (Tomás et al. 2013; Tosens et al. 2012), especially chloroplast surface area, thickness of leaves and cell wall thickness. Although the two main candidates explaining the dynamic response of gm to environmental variables point to aquaporins and carbonic anhydrases (Flexas et al. 2012), a better knowledge of the role filled by the anatomical players may explain part of the picture, including the acclimation aspects.

The inclusion of the temperature dependence parameters for gm in photosynthesis models, especially at a global scale where many different plant groups are involved, requires the characterization of this response in several species representative of each plant group. Currently we still lack this information and more studies are demanded if we decide to adopt the use of photosynthesis models on a Cc-basis.

Response of Rubisco to Temperature

In most of the studies in which in vivo Rubisco parameters are used, it is assumed that species-specific differences are negligible (Yamori & von Caemmerer 2009; Galmés et al. 2011; Greer & Weedon 2012; Scafaro et al. 2012), or seen from another point of view, that all species behave as tobacco. The development of the antisense Rubisco small subunit tobacco plants (antirbcS; von Caemmerer et al. 1994) allowed Bernacchi et al. (2001, 2002) to estimate temperature response functions of Rubisco kinetics, both on a Ci-basis and Cc-basis. These functions have become by far the most used in later studies in any species, that is photosynthetic parameterization in all species and conditions has been done using empirical values found for antisense tobacco under the particular conditions of the studies of Bernacchi et al. The main reason for this monopoly in most of the studies published recently is the lack of alternative surveys where the temperature dependencies of Rubisco kinetics parameters in vivo had been measured for other species. This is precisely one of the main contributions of Walker et al. (2013), and the development of a second antirbcS for A. thaliana has allowed the exploration of the variability of Rubisco kinetics in vivo in a second species. Although small differences were found in most of the parameters estimated, suggesting that Rubisco of these two species does not differ that much, their combined use in the model produced significantly different modelled rates of photosynthesis. An important question arises at this point. Some works have demonstrated that in vitro Rubisco kinetics are not as conservative as supposed among species, with large differences found in some cases (Galmés et al. 2005). Why are these differences not reflected in in vivo estimations? Arabidopsis and Nicotiana seem to possess Rubisco of similar characteristics, as their in vitro kcat suggest (Walker et al. 2013). Would we find the same result in other species with contrasting Sc/o (=VcmaxKo/KcVomax)?

By the beginning of the current century, the Rubisco specificity factor (Sc/o) of around 100 C3 species has been measured and the specific variability in this parameter reported. As a result, we know that cyanobacteria and algae display much higher Sc/o than higher plants and diatoms (Jordan & Ogren 1981; Tortell 2000). However, it was Galmés et al. (2005) who demonstrated that significant variability in Sc/o exists among C3 higher plants (up to 30%), and that these differences were related to environmental factors associated mainly to water availability and hot environments. Similarly, Sage (2002) showed that the catalytic efficiency of Rubisco (kcat) of both C3 and C4 species originating in cool environments was higher than those from warm environments. This correlation of Rubisco kinetics with environmental factors drove Savir et al. (2010) to analyse the adaptation of Rubisco from organisms living in various environments. How these changes in Rubisco are achieved has been studied in Flaveria, a genus that includes C3, C4 and C3C4 intermediate species, making it a perfect model to study the evolution of Rubisco and the structural basis for its adaptation. Rubisco of C4 species had a reduced Sc/o compared with C3 species (Kubien et al. 2008), while Kapralov et al. (2011) found that most of the changes that produced these differences were localized in the large subunit of Rubisco. Given the acclimation of Rubisco to growing temperature conditions (Vu et al. 1997; Yamori et al. 2006b), or CO2 enrichment (Vu et al. 1997), it has been even suggested that Rubisco has the capacity to acclimate by modifying the gene expression of the small subunit of Rubisco (Yoon et al. 2001; Cavanagh & Kubien 2013). Finally, Whitney et al. (2011) were able to identify that the substitution of the amino acid methionine by isoleucine in position 309 in the large subunit of Rubisco acted as a catalytic switch between C4 and C3 catalysis.

Hence, it is clear that Rubisco presents some variability – including the capacity to acclimate to changing environmental conditions – probably caused by evolving under differential selection pressures. The question is: how feasible is the extrapolation of its kinetics from in vitro to in vivo? We can directly measure the first after extracting the enzyme, and estimate the second using the whole attached leaf environment as the matrix where Rubisco resides. But both approaches have uncertainties. In vivo has the main difficulty of a reliable estimation of gm (Pons et al. 2009; Tholen et al. 2012), in addition to problems resulting from leaf heterogeneity across the mesophyll. That is, we retrieve gas exchange values from the entire leaf, but these are dominated by the gas exchange occurring in the most illuminated chloroplasts. Because we use these rates as an ‘average’ from the entire leaf, differences in Rubisco distribution and/or Rubisco ‘environments’ (e.g. pH, ionic strength) between differently illuminated cells may lead to deviations from the ‘real average’ values. In vitro measurements are prone to errors associated with the conditions at which the activity assays are made. When working in vitro, we have to simply assume variables occurring in vivo like stromal pH and gm. Bernacchi et al. (2001) found that their estimations of the ratio Vcmax/Vomax at all temperatures, and Γ* at the highest temperatures were clearly higher than the ones measured in vitro by Badger & Collatz (1977). The differences were attributed to gm because, in that work, the kinetics of Rubisco were estimated on a Ci-basis. This was corroborated in later work where estimations of Rubisco kinetics were made on a Cc-basis (Bernacchi et al. 2002). However, in this work, the new estimation of Rubisco parameters was not compared with the previously mentioned in vitro work by Badger & Collatz (1977). If we compare Γ* estimated on a Cc-basis (Bernacchi et al. 2002) with in vitro values, we observe a good match at all temperatures. This would confirm the major role played by gm in extrapolating from in vitro to in vivo Rubisco kinetics. Rogers, Ellsworth & Humphries (2001) specifically studied the disparity between the in vitro and in vivo measurements for Rubisco activity, and concluded that the usual underestimations of in vitro as compared with in vivo values are due to insufficient extraction of Rubisco protein prior to activity analysis. However this disparity only affects the extrapolation of Rubisco activity, that is extrapolation to Vcmax values. Parameters of Rubisco kinetics estimated in vitro are independent of the amount of protein extracted. This would explain why when gm was taken into account in the estimations by Bernacchi et al. (2002) there was a good agreement between in vivo and in vitro estimates of Γ*. Another encouraging evidence of good extrapolation from in vitro to in vivo behaviour of Rubisco comes from the previously mentioned work by Whitney et al. (2011) in Flaveria with mutant Rubisco. The artificial modification of the large subunit of Rubisco was not only reflected in the in vitro performance of the mutant Rubisco, which had a faster Vc and lower Sc/o, but also when leaf photosynthesis rate or plant growth were compared with the native one.

But, have these differences in temperature dependencies on Rubisco kinetics and gm among species a significant impact on the modelled photosynthesis? Figure 1 simulates the ribulose bisphosphate saturated rate of CO2 assimilation rate comparing the Rubisco kinetics and gm temperature functions of two contrasting species: N. tabacum and Oryza sativa. O. sativa presents a kcat lower than N. tabacum, and a totally different temperature response (Fig. 1a & b). Another reason for choosing O. sativa for this comparison is that its gm temperature response (Scafaro et al. 2011) differs for that of N. tabacum at high temperatures (Fig. 1c). Kc, Ko and Γ* for O. sativa were estimated from the temperature response values of kcat reported by Sage (2002), and from the strong power functions among Rubisco parameters reported for a large variety of Rubiscos by Savir et al. (2010). These results must be considered with caution as Savir et al. used values at 25 °C only, but the results of the simulation show clearly the large difference in CO2 assimilation rate between both species when their specific parameters are used (thin and thick solid lines, Fig. 1d). This difference is mainly determined by the Rubisco kinetics differences with only a marginal role played by gm, at least under the conditions explore in this simulation.

Figure 1.

CO2 assimilation in response to temperature, comparing Rubisco kinetic parameters and mesophyll conductance to CO2 (gm) for Nicotiana tabacum and Oryza sativa. (a) Temperature response of the ratio Ko/Kc for N. tabacum (blue solid line, Walker et al. 2013) and O. sativa (red dashed line, Sage 2002). In the case of O. sativa, the temperature response of kcat was obtained from Sage (2002), and the extrapolation of kcat to Kc and Ko was calculated from power functions proposed by Savir et al. (2010), assuming that the functions hold for the whole range of temperatures. (b) Γ* obtained as in panel a. (c) Temperature response of gm obtained for N. tabacum from Walker et al. (2013) and for O. sativa from Scafaro et al. (2011). (d) Simulation of the ribulose bisphosphate saturated rate of CO2 assimilation in both species using temperature response function for Rubisco kinetics and gm characteristic of each species (solid lines), and their combination (dashed lines). In the simulation, Vcmax = 130 μmoL m−2 s−1, gm = 0.56 moL m−2 s−1 bar−1; gs = 0.45 moL m−2 s−1, all at 25 °C.

In conclusion, Walker et al. (2013) have confirmed that the inclusion of gm in a photosynthesis model makes it a more mechanistic and process-based tool, but at the expense of increasing its complexity and reducing its ease of use. This should not be an excuse for not continuing to work in that direction. In fact, what is really needed at this stage are more works exploring the diversity of response functions of gm in diverse functional plant types with the goal of improving our understanding of the mechanisms involved in its regulation. A similar conclusion can be reached about Rubisco. It is expected that larger differences emerge when other species with higher Sc/o or higher kcat are studied. The reported variability of Rubiscos and their optimal adaptation to the environment where they are found suggest that including this variability in models are necessary if we want to simulate and understand the photosynthetic performance of these species in their natural environments.