Introduction
Metabolism is the basis of all cellular processes. Metabolic systems must be optimized to maintain function, regardless of the environmental conditions experienced. Plants are exposed to environments which fluctuate over timescales varying from seconds to decades, with environmental parameters such as light and temperature directly impacting metabolism. As such, plants provide an excellent example in which to understand how metabolic systems can be optimized to tolerate environmental change. To understand such optimization, we can use approaches taking form systems engineering.
In systems processes, the quality of a function or product is referred to as its reliability. The quantity, on the other hand, is dependent on the system’s robustness (King and Jewett, 2010). As an analogy, we can consider a business process where the aim of the process is to produce as many high-quality products as possible, despite fluctuations in operating conditions. A reliable process ensures that the quality of the products is consistently high, whereas a robust process ensures that as many products as possible are being produced. If all low-quality products are discarded then, to make a profit, the system must first be reliable and then robust (Bakera et al. , 2008; King and Jewett, 2010). This is comparable to plants optimizing seed production (Sadras 2007). Only viable seeds will germinate. Thus, features of reliability, which ensure that seeds germinate, are of primary concern. Features of robustness, which ensure that as many seeds as possible are produced, are secondary. In fact, in the model plant species Arabidopsis thaliana the trade-off between seed size and number is minimal (Gnanet al. , 2014), suggesting that these traits may be selected for independently.
This distinction between robustness and reliability is considered crucial in many disciplines, including systems engineering and product development (Yassine, 2007; King and Jewett, 2010). With regards to metabolism, the two concepts are rarely distinguished, and the reliability of metabolic networks has not been formally quantified. However, reliability engineering frameworks exist which can be applied to metabolic networks.
Failure mode and effect analysis (FMEA) is a framework that is commonly applied to study the reliability of a product of a process, and is considered to be one of the most effective ways of conducting a failure analysis (i.e. collecting data on why a system might fail). FMEA breaks the system down into individual components or single processes and quantifies the risk that each of them poses to the system (Stamatis, 1995; Rausanda and Øienb, 1996). The risk of each component (or process) is assessed by studying the potential causes of failure and effects that these would have on the system. To quantify this risk, each component is assigned a probability of failure (P ) and a severity score in the event of failure (S ). The risk (R ) of each component is then calculated such that R = P × S . Components which are essential for the system to function are considered high-risk. A well-engineered system has control mechanisms to regulate the functioning of the high-risk components, to minimize the impact of failures. Such control mechanisms can be inherent features of the system (e.g. the closing of one valve automatically opens another) or can be sensed and regulated (e.g. a sensor perceives that one valve is closed and opens another).
The benefit of the FMEA framework is that it can be adapted to conduct a failure analysis of any system, making it also suitable for studies of metabolic systems. We can think of individual metabolites as the components that constitute a system. The metabolic system may represent the whole of metabolism or a subset of metabolites, depending on the processes of interest. The components (i.e. the metabolites) need to be present within specific concentration ranges, or the system will not function correctly. Failure of the system might mean a failure to produce certain products at an appropriate rate (e.g. a failure to synthesise sufficient ATP to support cell maintenance) or the accumulation of metabolites to toxic concentrations. We would expect the concentration of high-risk metabolites to be tightly regulated to avoid systems failures. The probability of failure of an individual metabolite, i.e. the probability that its concentration falls outside an acceptable range, could be considered a function of upstream metabolites and the severity of failure as a function of downstream metabolites. Thus, the risk score for each metabolite arises from the underlying metabolic network structure, the highest risk metabolites being those which are heavily dependent on the production of other metabolites whilst at the same time supporting the production of many others.
Reliability engineering has potential in trying to understand how metabolic systems respond to environmental stress. For example, it is predicted that, due to anthropogenic climate change, plants will increasingly be exposed to extreme weather events, including both high and low temperature periods (Changnon et al ., 2000; Trenberth, 2012). These extreme conditions will affect crops yield, threatening food security for a growing world population (Lesk et al. , 2016; Powell and Reinhard, 2016). Thus, the identification of traits which determine the reliability of key acclimation processes becomes a priority. In this context, plant responses to single temperature stresses have been well characterized (e.g. Hikosaka et al. , 2006; Yamori et al. , 2014; Ding et al. , 2020). A few studies have also compared the effects of warm and cold temperature stresses on Arabidopsis leaves, showing that the two processes share common features (Kaplan et al. , 2004; Zhang et al. , 2019). Nevertheless, how metabolism adjusts across the entire physiological temperature range of a plant is still largely unknown.
When exposed to short term environmental changes, plants canregulate enzymes to ensure metabolic function is maintained. Over longer periods, plants acclimate to their environment, altering the concentration of different enzymes to match the prevailing conditions (Webber, 1994; Stitt and Hurry, 2002). For example, when a plant is transferred from warm to cold conditions, enzyme activities are slowed, reducing metabolic activity. This would be reflected, for example, in a reduction in the light-saturated rate of photosynthesis. Following acclimation, plants transferred to the cold typically increase the concentration of key enzymes, allowing them to achieve a similar rate of photosynthesis at low temperature to that previously seen in the warm (Hurry et al. , 2000; Strand et al. , 2003; Dysonet al. , 2016). This results in an increase in the photosynthetic capacity (Pmax ) (Athanasiou et al. , 2010; Dyson et al. , 2016). Conversely, in response to increased temperature, plants decrease their Pmax , allowing them to reallocate resources to other processes (Herrmann et al. , 2019b).
During the day, plants fix atmospheric carbon through photosynthesis. A large proportion of that fixed carbon is accumulated in the leaf during the photoperiod. Overnight this is then remobilised to support growth and cell maintenance. Under different environmental conditions, plants vary the form in which photoassimilates are stored. Arabidopsis stores carbon primarily in the form of the polysaccharide starch and the organic acids fumarate and malate (Chia et al ., 2000; Smith and Stitt, 2007; Zell et al ., 2010; Pracharoenwattana et al ., 2010). Starch accumulation and degradation have been studied extensively as this is the major non-structural carbohydrate in plants (Smith and Stitt, 2007; Streb and Zeeman, 2012). Less is known about malate and fumarate. Both metabolites are intermediates of the tricarboxylic acid cycle in the mitochondria; however, it has also been shown that they accumulate in the vacuole diurnally (Kovermann et al ., 2007; Fernie and Martinoia, 2009). In the cytosol, malate is made from or converted to oxaloacetate, citrate, isocitrate, cis-aconitate, fumarate, α-ketaglutarate, glutamic acid and pyruvate (Arnold and Nikoloski, 2014). It provides carbon skeletons for various biosynthetic pathways, including amino acid synthesis (Zell et al. , 2010). In the cytosol, fumarate is synthesised by a cytosolic isoform of the enzyme fumarase, FUM2 (Pracharoenwattana et al ., 2010). It is made from malate during the day and is assumed to be converted back to malate during the night (Pracharoenwattana et al ., 2010; Zellet al ., 2010; Dyson et al. , 2016). Fumarate accumulation in the cytosol has previously been linked to growth on high nitrogen (Pracharoenwattana et al ., 2010) and photosynthetic acclimation to cold (Dyson et al. , 2016); however, its role in these processes is unclear (Chia et al. , 2000).
The fum2 Arabidopsis mutant is a cytosolic fumarase knockout (Pracharoenwattana et al ., 2010). Previously, we showed that fumarate and malate accumulation increase in response to cold, and that the fum2 knockout mutants, which do not accumulate fumarate, are unable to acclimate photosynthetic capacity in response to low temperature (Dyson et al ., 2016). The mutant accumulates higher levels of both malate and starch under cold stress. It has also been shown that plants which accumulate more fumarate accumulate less starch (Chia et al ., 2000; Pracharoenwattana et al ., 2010; Dysonet al., 2016) and that fumarate accumulation is proportional to biomass production at low temperature (Scott et al., 2014). Therefore, fumarate accumulation participates in the cold acclimation response, although its specific role in this remains to be elucidated.
In this study, we demonstrate that FUM2 activity is essential not only for low temperature acclimation but also for high temperature responses. Using three genotypes, with varying capacities to accumulate fumarate, we have examined acclimation across the physiological temperature range. We have measured photosynthetic and metabolic responses of the Col-0 wild-type, which accumulates high levels of fumarate, a fum2knockout, which does not accumulate fumarate (Pracharoenwattana et al ., 2010), and the C24 wild-type, which accumulates intermediate levels of fumarate compared to Col-0 (Riewe et al ., 2016). To understand their responses to temperature treatment, we combine the novel application of FMEA with a more widely used kinetic modelling approach (Saa and Nielsen, 2017; Herrmann et al., 2019a), identifying cytosolic malate as a high-risk component and its metabolic neighbour, cytosolic fumarate, as a low-risk one. We propose that fumarate synthesis acts as a fail-safe, maintaining malate concentrations within safe limits and ensuring that metabolic functions continue across a wide range of conditions. We discuss that FMEA is a valuable approach to understanding metabolic system with possible widespread applications in biotechnology and systems biology.