An update on redox signals in plant responses to biotic and abiotic stress crosstalk: insights from cadmium and fungal pathogen interactions

Reactive oxygen and nitrogen species and redox signals are at the crossroads of plant responses to abiotic and biotic stress and mediate plant-induced resistance/sensitivity, which mainly depends on the specific interaction.


Introduction
Plants are routinely confronted with more than one stress either simultaneously or sequentially in the field, where a changeable environment exists, especially in the context of global warming, and where pathogens and herbivores are present . In fact, a study of transcriptome responses to different combinations of stresses in Arabidopsis has shown that plants have evolved to cope with combinations of stresses (Rasmussen et al., 2013). An understanding of specific and common biological and molecular responses of plants to different stresses is crucial for crop resistance in This paper is available online free of all access charges (see https://academic.oup.com/jxb/pages/openaccess for further details) the current environmental context. For this reason, in recent years, large-scale transcriptomic analysis involving microarray, RNA-seq, and metabolomic techniques has been used to study crosstalk between different signalling networks (Cheong et al., 2002;Mhamdi and Noctor, 2016;Cohen and Leach, 2019;Zandalinas et al., 2021). Furthermore, large-scale analysis involving 350 Arabidopsis accessions and various combinations of stresses has highlighted genome-wide associations with plant resistance and has identified target genes related to plant responses to multiple stresses (Thoen et al., 2017). Plant responses to more than one simultaneous stress are complex, with a balance between different pathways being required to enable plant survival (Makumburage et al., 2013;Suzuki et al., 2014;Thoen et al., 2017;Zandalinas et al., 2021). The many recent studies, comprehensive reviews, and special issues of scientific journals on different combinations of abiotic stresses highlight the importance of this topic (Loudet and Hasegawa, 2017;Lawas et al., 2018;Sehgal et al., 2018;Balfagón et al., 2019;Peck and Mittler, 2020;Zandalinas et al., 2020Zandalinas et al., , 2021. Interestingly, unique plant responses to combinations of abiotic stresses including heat stress induce specific transcription factor (TF) group patterns, which are not shared with other stress combinations . A recent exhaustive analysis of up to six combined stresses showed that an increase in the number of stresses negatively correlates with plant growth and survival (Zandalinas et al., 2021).
Combinations of abiotic and biotic stresses, and the ways in which adverse growth conditions affect plant responses to pathogens, have attracted less interest from researchers than combinations of different abiotic stresses. In fact, the variable behaviour and the diverse nature of plant infection mechanisms make it difficult to reach general conclusions. In this review, we evaluate the latest data on crosstalk between plant responses to biotic and abiotic stresses, with particular attention paid to the key regulatory role of reactive oxygen species (ROS), reactive nitrogen species (RNS), and redox signals. Analyses of transcriptomes related to plant responses to single and combined stresses will help to decipher plant responses to biotic and abiotic stresses commonly encountered in the field. The results obtained could be used to improve crop stress tolerance in the future. The relationship between plant hyperaccumulation of metals and pathogen defences, the availability of transcriptomes involving the heavy metal cadmium (Cd), and the presence in these transcriptomes of plant responses to biotic stresses, particularly fungal pathogens, enabled us to gain insights into the possible role of ROS/RNS and redox signals at the crossroads of plant responses to Cd and fungi.

Crosstalk between plant responses to abiotic and biotic stress
Protection of plants against disease using abiotic stress treatments previously appeared to be specific to the type of stress encountered and to the behaviour of the pathogen (Rasmussen et al., 2013;Bostock et al., 2014;Zhang and Sonnewald, 2017). Co-expression analysis has revealed a set of gene transcripts with similar profiles of responses to biotic and temperature stresses, mainly associated with the hormones ethylene (ET), jasmonic acid (JA), and/or salicylic acid (SA) (Rasmussen et al., 2013). In a recent genome-wide association mapping study of plant resistance to different biotic and abiotic stresses, genetic correlation analysis showed a strong relationship between plant responses to osmotic stress and root-feeding nematodes (Thoen et al., 2017). Nematodes alter cellular osmotic pressure and plant water potential (Baldacci-Cresp et al., 2015), which link the specific abiotic stress to the plant response to the infection mechanism of these parasites (Atkinson and Urwin, 2012). Heat stress undermines the resistance of tomato to nematodes, although little is known about the underlying mechanism involved (Marques de Carvalho et al., 2015). Insect damage is frequently associated with osmotic stress and drought stress, which appear to strongly overlap in phytohormone-dependent signalling (Ma et al., 2006;Pieterse et al., 2012;Thoen et al., 2017). Following sequential double-stress treatment in Arabidopsis involving a combination of Botrytis cinerea infection, Pieris rapae herbivory, and drought, changes in the transcriptome profile were very similar to those observed after the application of the second stress, although significant signatures, mainly related to hormones, from the first stress were also identified (Coolen et al., 2016;Fig. 1). The first stress also affected the timing of the regulation of specific biological processes (Coolen et al., 2016). In this case, prior treatment of Arabidopsis with herbivory, but not with drought stress, protected against B. cinerea lesion spread, again suggesting that protection is probably treatment-specific (Coolen et al., 2016). Some studies of simultaneous drought/heat and biotic stresses suggest that abiotic stress plays a predominant role, leading to increased plant susceptibility, although the precise mechanisms involved are not fully understood (Luo et al., 2005;Prasch and Sonnewald, 2013;Pandey et al., 2015;Gupta et al., 2020). Other studies suggest that abscisic acid (ABA) reduces plant tolerance to hemibiotrophic and biotrophic pathogens across species (reviewed in Zhang and Sonnewald, 2017). Plant protection against biotic stresses under salt-stress conditions depends on the specific pathogen, with salt-stressed tomato plants being more susceptible to Oidium neolycopersici (Kissoudis et al., 2014) and more resistant to B. cinerea (Achuo et al., 2006), while salt-stressed barley plants are more resistant to powdery mildew (Wiese et al., 2004). Salt stress has been shown to decrease SA-dependent responses to Pseudomonas syringae in tomato plants and to alter negative JA-SA interactions in response to the herbivore Trichoplusia ni without affecting resistance to either of these pathogens (Thaler and Bostock, 2004). Temperature changes also affect plant resistance, with low temperatures appearing to prevent gene silencing against viruses (Szittya et al., 2003) and high temperatures contributing to the spread of pathogens such as Fusarium (Madgwick et al., 2011). Furthermore, high temperatures induce conformational changes in tobacco mosaic virus R genes, leading to increased susceptibility of tobacco plants (Zhu et al., 2010). On the other hand, high temperatures have been found to contribute to increased resistance of wheat to Puccinia striiformis (Carter et al., 2009). This variability in reported results highlights the complexity of biotic and abiotic stress responses, as well as the specific nature of each interaction and situation (Zhu et al., 2010;Prasch and Sonnewald, 2013;Huot et al., 2017). Apart from temperature, other climate-change-related factors, such as increasing CO 2 emissions, may affect the resistance of crop species (Luck et al., 2011).

ROS, nitric oxide, and redox signals in plant responses to stress
Data collected over time strongly demonstrate that stress signalling in plants is organized in a complex network mediated by signals, some of which are commonly found in plant responses to abiotic and biotic stresses. Recent research on signalling components, which include calcium (Ca 2+ ) and other ions, mitogen-activated protein kinase (MAPK) cascades, hormones, and TFs, and function in biotic/abiotic crosstalk, have been widely reviewed ( Fig. 1; Gilroy et al., 2014;Choudhury et al., 2017;Zhang and Sonnewald, 2017;Bai et al., 2018;Zandalinas et al., 2020Zandalinas et al., , 2021. Some of these signalling molecules are ROS/RNS, key molecules that orchestrate crosstalk between plant responses to abiotic and biotic stress. In addition, the two key thiol/disulfide couples, reduced/oxidized glutathione (GSH/GSSG) and cysteine (Cys/CySS), and the ascorbic/dehydroascorbic acid couple (ASC/DHA), as well as a broad range of redox-dependent proteins, lie at the core of the cellular redox state (Bowler and Fluhr, 2000;Baxter et al., 2014;Sandalio et al., 2019;. ROS, which are by-products of the plant aerobic metabolism (Inupakutika et al., 2016) , which rapidly dismutates to H 2 O 2, is shorter than that of H 2 O 2 and 1 O 2 , but longer than that of ·OH (Halliwell and Gutteridge, 2007). Plants contain numerous ROS-generating pathways associated with different organelles, which are intimately linked to metabolic pathways and to plant function and development. ROS production in chloroplasts and mitochondria is mainly dependent on photosynthetic electron transport and the mitochondrial electron transport chain (Smirnoff and Arnaud,Fig. 1. Signal transduction pathways in plant responses to stress. (A) After stress perception, a complex and specific signalling pathway (indicated by the yellow colour) is activated to produce a response leading to plant survival, aimed at achieving a trade-off between acclimation and yield. Signalling pathways involve different factors such as ions/Ca 2+ , reactive oxygen and nitrogen species (ROS/RNS), mitogen-activated protein kinases (MAPKs), hormones, changes in proteins by post-translational modifications (PTMs), and transcription factors (TFs). All these factors need to be integrated to ensure a proper response. (B) Sequential double stress-induced changes are very similar to those observed after the application of the second stress (indicated by the blue colour), although significant signatures from the first stress (indicated by the yellow colour) are also identified. The application of the first stress may also affect the timing of the regulation of specific biological processes related to the second stress. (C) Simultaneous stresses induce unique plant responses to each combination of stresses (indicated by the green colour), which differ from the responses to stresses applied individually. 2019); ROS production in peroxisomes has been recently reviewed by Sandalio et al. (2021).
NADPH oxidase is the principal source of O 2 .− and derived H 2 O 2 in the apoplast (Suzuki et al., 2011), while peroxidases also contribute to ROS production (Daudi et al., 2012). Although high and uncontrolled levels of ROS can be dangerous, controlled concentrations of ROS play an important role as signals in the regulation of different developmental processes and responses to biotic and abiotic stresses. Antioxidant defences regulate the balance between ROS production and removal, which enables the signalling of these molecules to function. Superoxide dismutase (SOD) disproportionates O 2 .− to H 2 O 2 , and several isoforms of SOD, with different prosthetic metals, are present in all cellular compartments (Gill et al., 2015). H 2 O 2 is then removed by catalase, the ASC-GSH cycle and peroxiredoxins (Smirnoff and Arnaud, 2019). However, antioxidants do not merely defend against oxidants, but also regulate cellular redox biology. Using the term "ROS-processing systems" rather than "antioxidative systems", (Noctor et al. 2018) suggested that these molecules play a broad role in regulating and transmitting redox-derived signals.
The stability, diffusibility, and selective reactivity of H 2 O 2 make it an ideal signalling molecule. It can react with sulfurcontaining amino acids such as cysteine, leading to its reversible oxidation to sulfenic acid (-SOH; sulfenylation) and sulfinic acid (-SO 2 H; sulfinylation), while excessive ROS accumulation gives rise to an irreversible sulfonic acid (-SO 3 H) derivative (sulfonylation; Young et al., 2019). Sulfenylation and sulfinylation, as well as intra-and inter-molecular disulfide bond formation, are rapid and reversible mechanisms, which regulate protein function, stability, and location Young et al., 2019). Given their transient nature, these sulfur modifications, which can be reversibly reduced by thioredoxin and glutaredoxin pathways, are regarded as redox switches. The flexibility of these redox circuits favours rapid responses to changes in intracellular redox homeostasis caused by environmental changes, thus regulating metabolic pathways and facilitating signalling networks (Noctor et al., 2018;Sandalio et al., 2019;Young et al., 2019). There is some evidence that ROS production in different organelles, as well as temporary spikes in ROS, leave a specific imprint on the transcriptome response, which can be translated by the cell into specific cellular responses (Rosenwasser et al., 2011;Sewelam et al., 2014).
Nitric oxide (NO) is well known to be a global intra-and intercellular signalling molecule involved in the regulation of an enormous range of plant processes, from development to defence responses to biotic and abiotic stresses Sánchez-Vicente et al., 2019). Reductive and oxidative mechanisms have been reported to be involved in NO biosynthesis in plants, although this process remains unclear (reviewed in Chamizo-Ampudia et al., 2016;Astier et al., 2018;León and Costa-Broseta, 2020). NO production has been reported in peroxisomes (reviewed in Sandalio et al., 2021), cytosol, mitochondria, and chloroplasts, although the mechanisms involved are not fully understood (León and Costa-Broseta, 2020). NO is also produced in the plasma membrane and apoplast (Stöhr et al., 2001;reviewed in León and Costa-Broseta, 2020). Intracellular levels of NO are regulated by balancing its production, scavenging, and metabolism. NO can react with reduced glutathione (GSH), giving rise to S-nitrosoglutathione (GSNO), which in turn is regulated by GSNO reductase (GSNOR) or reacts with O 2 .− -producing peroxynitrite (ONOO − ) (reviewed in Arnaiz et al., 2021). NO levels can be regulated by globins, which are capable of metabolizing NO-producing nitrate (Perazzolli et al., 2006;Becana et al., 2020). The mode of action of NO in plants depends on covalent protein post-translational modifications (PTMs), the best known of which is S-nitrosylation (S-nitrosation); this PTM involves the formation of a nitrosothiol in a cysteine residue, which can modify the function, location, and stability of a large number of proteins (Romero-Puertas et al., 2013;Feng et al., 2019). Different TFs are targeted by S-nitrosylation, which affects their DNA-binding and gene-regulation capacities (Cui et al., , 2020Imran et al., 2018). NO interacts with most phytohormone metabolisms and/or signalling pathways through the S-nitrosylation of key enzymes, and also regulates ROS levels through the S-nitrosylation of ROSproducing and ROS-removing enzymes (reviewed in Sandalio et al., 2019). S-nitrosylation is a reversible process, which is partly regulated by thioredoxins (Mata-Pérez and Spoel, 2019). Another NO-dependent PTM, whose reversibility remains elusive, is nitration; nitration of proteins and fatty acids affects the functionality of a number of plant proteins and signalling pathways (Mata-Pérez et al., 2017;Arasimowicz-Jelonek and Floryszak-Wieczorek, 2019).

ROS/RNS and redox signals at the crossroads of plant responses to abiotic and biotic stresses
Virtually all abiotic and biotic stresses induce ROS/RNS production and redox changes, which in turn are connected with MAPK signalling, as well as hormone metabolism and signalling. Signalling mechanisms such as phosphorylation and ubiquitination are regulated by ROS/RNS, as are various TFs, leading to changes in gene expression (Vaahtera et al., 2014;Imran et al., 2018;Sandalio et al., 2019;Siauciunaite et al., 2019). A crucial challenge in redox biology is the identification of sensors that trigger different signalling mechanisms. Interestingly, stomatal movements, which are regulated under various abiotic stresses such as drought, light, ozone, and CO 2 (Devireddy et al., 2018Zhang et al., 2018;Gupta et al., 2020), and are also the entrance point for numerous pathogens (Melotto et al., 2006;Qi et al., 2018), may be involved in crosstalk between abiotic and biotic stresses. Stomatal movements are regulated by a complex signalling network involving ROS/RNS, Ca 2+ and other ions, channels, and transporters, as well as ABA. One of the first signs of stomatal closure is an increase in ROS in the apoplast and chloroplast (reviewed by Song et al., 2014;Sierla et al., 2016), and NO is also involved in stomatal movements (Van Meeteren et al., 2020). Systemic signalling in plant responses to abiotic stress, which is mediated by ROS mainly derived from NADPH oxidase D [respiratory burst oxidase protein D (RBOHD); Fichman et al., Zandalinas et al., 2020], constitutes another point of crosstalk between abiotic and biotic stresses. MYB30, one of the RBOHD-dependent transcripts regulated during systemic signalling, is involved in plant responses to abiotic and biotic stresses (Mabuchi et al., 2018;. Cell wall lignification, which is also ROS dependent Pan et al., 2021), may be another point of crosstalk between abiotic and biotic stresses, as various abiotic stresses induce lignin accumulation (Díaz et al., 2001), which is a physical barrier against specific pathogens such as Verticillium (Pomar et al., 2004).
Furthermore, a number of studies have analysed ROS/ RNS and redox signals at the crossroads of combined abiotic and biotic stresses. Narusaka et al. (2004) have reported that treatment of Arabidopsis thaliana with copper (Cu) and infection with the necrotrophic pathogens Alternaria alternata and Alternaria brassicicola cause a significant overlapping of regulation of cytochrome P450 genes, suggesting that common ROS signals trigger similar responses. Down-regulation of O 2 .− and induction of antioxidants are associated with an increase in the sensitivity of tobacco plants to the tobacco mosaic virus at high temperatures, although the mechanisms involved are not well understood (Király et al., 2008). While redox signals are key elements in networks of cross-tolerance to stresses, the role of NO in these networks remains unclear, although its role in plant responses to a single stress has been well documented Martínez-Medina et al., 2019;León and Costa-Broseta, 2020).

Crosstalk in plant responses to heavy metals and biotic stress
While some heavy metals (those with density ≥5.0 g cm −3 ), such as iron (Fe), manganese (Mn), and Cu, are essential elements needed for plants to achieve normal metabolism and to carry out physiological processes, other heavy metals, such as Cd, mercury (Hg), chromium (Cr), and the metalloid arsenic (As), are toxic even at low doses (Clemens and Ma, 2016;Terrón-Camero et al., 2019). Nevertheless, essential heavy metals may be toxic to plants at high concentrations, and excessive availability may result from global warming effects such as drought, high temperatures, and flooding. Currently, soil contamination with heavy metals poses a potential threat to the environment and to agriculture, and therefore to human health. The main sources of heavy metals in agricultural soils are anthropogenic activities such as wastewater irrigation from sewage sludge, limestone amendments, and application of inorganic fertilizers (Cao et al., 2016;Clemens and Ma, 2016). Heavy metals/metalloids also occur naturally in sediment deposits in, for example, soil and water (Peralta et al., 2020).
Apart from the risk of sudden pollution spills, plants growing in contaminated soils are already under threat and are likely to face other types of stress, particularly biotic stresses. Heavy metals therefore make for an interesting in-depth case study of crosstalk between abiotic and biotic stresses. It has been suggested that several plant species even capture high concentrations of metals from the soil as a defence mechanism against herbivores and pathogens (Poschenrieder et al., 2006;Llugany et al., 2019). These authors have identified at least five different modes of action induced by metals to counter biotic stress: (i) phytosanitary actions, as various metals are widely used as fungicides, which are detrimental to pathogen and herbivore growth (reviewed in Morkunas et al., 2018); (ii) metal therapy, as metals can activate defence signals to protect the plant against pathogens; (iii) possible trade-offs, whereby a metal defence strategy could save energy for organic defences; (iv) metal fortifications, induced either directly or indirectly through ROS/ RNS, with cell wall lignification providing a mechanical barrier against pathogens, as well as the induction of antioxidants and defence genes (Choudhury et al., 2017;Terrón-Camero et al., 2019), and (v) possible elemental defences, which enable metals to directly protect the plant against pathogens (Michaud and Grant, 2003;Coleman et al., 2005;Matyssek et al., 2005).
As explained earlier in the section "Crosstalk between plant responses to abiotic and biotic stress", signal transduction routes in plant responses to biotic and abiotic stresses, particularly those caused by heavy metals (Romero-Puertas et al., 2019), show several interaction points, mainly for short-term responses. MAPK signalling mechanisms, which are involved very early on in plant responses to various heavy metals such as Cu and Cd, differentially activate signalling routes (Suzuki et al., 2001;Jonak et al., 2004;Opdenakker et al., 2012;Cuypers et al., 2016). Extensive data are available on plant hormone responses to heavy metal stress (reviewed in Cuypers et al., 2016;Anwar et al., 2018;Demecsová and Tamás, 2019;Sharma et al., 2020;Betti et al., 2021). For example, ET signalling and biosynthesis are induced in both early and late responses to Cd in Arabidopsis (Herbette et al., 2006;Weber et al., 2006;Rodríguez-Serrano et al., 2009).The phytohormone JA is induced by Cd and Cu stress in various plant species, such as rice, Arabidopsis, pea, and Phaseolus coccineus (Maksymiec et al., 2005;Rodríguez-Serrano et al., 2006;Ogawa et al., 2009). Despite being associated with GSH and phytochelatins (Xiang and Oliver, 1998), JA is involved in the activation by metal toxicity of H 2 O 2 production via lipoxygenase (Maksymiec et al., 2005). SA, another phytohormone associated with plant responses to heavy metals, displays variable dynamics depending on the tissue and the experimental conditions (Rodríguez-Serrano et al., 2009), and also affects H 2 O 2 levels (Tao et al., 2013).
Tolerance to both heavy metals and biotic stress has long been a topic of research. Several studies show that ROS metabolism and/or the induction of defence signalling pathways are involved in heavy metal protection, although the mechanisms underlying these cross-tolerance processes are sometimes unclear. Changes in the expression of cytochrome P450 genes are commonly found in the responses of Arabidopsis to Cu, as well as to A. alternata and A. brassicicola, suggesting that heavy metals induce ROS signals that serve to enhance plant resistance to fungi (Narusaka et al., 2004). Pepper plants pre-treated with Cu show a phenotype that is more resistant to Verticillium dahliae Kleb. than plants grown under normal conditions (Chmielowska et al., 2010). This resistance could be partly due to the induction of peroxidase and defence genes such as PR1 and β-1,3-glucanase by treatment with Cu (Chmielowska et al., 2010). Interestingly, a positive feedback loop between H 2 O 2 , Ca 2+ , and the TF WRKY41 coordinates pepper responses to Ralstonia solanacearum and Cd exposure (Dang et al., 2019). Cu, which decreases pathogenic disease symptoms and is even used as a fungicide (Molina et al., 1998), induces an increase in sensitivity in a small number of interactions (Evans et al., 2007). Aluminium (Al) stress induces H 2 O 2 accumulation and activates SA-and NO-dependent signalling pathways, which correlates with a reduction in disease symptoms in susceptible potato plants infected with Phytophthora infestans (Arasimowicz-Jelonek et al., 2014). Interestingly, Arasimowicz-Jelonek et al. (2014) found that treatment with Al induces signalling mechanisms in distal tissue that are effective in combating biotic stress. Furthermore, Vitis vinifera pre-treated with Mn shows resistance to Uncinula necato due to the induction of SA, ABA, peroxidases, and defence proteins such as phenylalanine ammonia lyase, PR proteins, and an NBS-LRR analogue (Yao et al., 2012).

Metal hyperaccumulation and defence responses
Metal hyperaccumulation, defined as the capacity of some plants to accumulate abnormally high levels of a metal in the aerial parts without causing phytotoxic damage, is not very common (Poschenrieder et al., 2006;Krämer, 2010;van der Ent et al., 2013). Only approximately 700 taxa from distantly related families have been described as hyperaccumulators (Calabrese and Agathokleous, 2021). One hypothesis used to explain metal hyperaccumulation by plants is that metals can efficiently provide elemental defence against herbivores and pathogens (Poschenrieder et al., 2006;Rascio and Navari-Izzo, 2011;Fones et al., 2019). A well-documented example of this is the hyperaccumulation by Noccaea (formerly Thlaspi) caerulescens of zinc (Zn), whose toxicity is capable of reducing P. syringae pv. maculicola (Psm) growth (Fones et al., 2010). In addition, while N. caerulescens lacks a ROSand SA-dependent signalling capacity in response to Psm, Zn can induce an increase in O 2 .− production in non-threatened plants . The typical oxidative burst defence responses are shut down in N. caerulescens in response to Psm, probably due to its ability to use Zn for defensive purposes .
In fact, trade-offs between Zn tolerance and defence gene expression have also been described in relation to two N. caerulescens ecotypes (Plessl et al., 2010). Hyperaccumulation of Zn also replaces SA-and JA-dependent defence responses in N. caerulescens plants threatened by A. brassicicola (Gallego et al., 2017). Noccaea praecox, a Cd hyperaccumulator, is more sensitive to the powdery mildew pathogen Erysiphe cruciferarum at lower Cd concentrations, and low Cd supply also appears to prevent a pathogen-dependent increase in SA (Llugany et al., 2013). In a similar study, the nickel (Ni) hyperaccumulator Noccaea goesingense, which has higher SA content than the non-accumulators Arabidopsis and Noccaea arvense, showed greater sensitivity to E. cruciferarum infection and was unable to induce SA production following infection; this sensitivity to the pathogen is reduced by Ni hyperaccumulation (Freeman et al., 2005). Recent analyses of four N. caerulescens populations with different Zn accumulation capacities have shown that this species has different modes of action, such as metal toxicity, glucosinolate production, and cell death, in response to Psm, leading to tradeoffs and synergistic interactions that protect the plant. Metal availability appears to be one of the factors that triggers defence responses in this case (Fones et al., 2019). Trade-offs between glucosinolates and metal accumulation have also been described in relation to Streptanthus polygaloides and N. caerulescens when Ni and Cd are hyperaccumulated (Davis and Boyd, 2000;Asad et al., 2013). However, the complex relationship between metal accumulation and glucosinolates may depend on the hyperaccumulator species and may even vary between specific populations (Fones et al., 2019). Other factors, such as hormones and ROS, are also involved in the relationship between glucosinolates and metal accumulation, enabling hyperaccumulator plant defences to be fine-tuned, with an additional stage of regulation leading to possible joint effects that could explain hyperaccumulation (Rascio and Navari-Izzo, 2011;Kusznierewicz et al., 2012;Hörger et al., 2013;Gallego et al., 2017). Therefore, some evidence shows that hyperaccumulated metals contribute to plant defences in the case of at least some kinds of pathogens and herbivores (Cabot et al., 2019). However, the trade-offs and synergistic interactions between other signalling molecules, and how selection for resistance to disease relates to the environment during their evolution, are little understood (Hörger et al., 2013).

Cadmium and fungi: a case study
The heavy metal Cd is a non-essential element for life (Ismael et al., 2019;Zhang and Reynolds, 2019) and, at even low concentrations, is toxic to living organisms Zhang and Reynolds, 2019). Although Cd is not abundant in the earth's crust (0.08-0.1 ppm), Cd concentrations in soils have been increasing over the past 100 years due to human activity (Rudnick and Gao, 2003;Gupta and Sandalio, 2012;Cullen and Maldonado, 2013). However, a report by the European Environment Agency (2018) shows a decrease in Cd emissions of ~64% between 1990 and 2016, mainly due to a decrease in Cd concentrations in agricultural processes and waste. Nevertheless, in 2017, the Agency for Toxic Substances and Disease Registry (http: //www.atsdr.cdc.gov/) considered Cd to be the seventh most toxic heavy metal due to its toxicity and potential exposure of humans. The principal sources of Cd emissions are industrial energy consumption (29%), industrial processes and product use (28%), and the commercial, institutional and household sector (21%; European Environment Agency 2018).
Cd, which affects different ecosystems, causes atmospheric, terrestrial, and marine damage (Pinto et al., 2004;Gupta and Sandalio, 2012;Li et al., 2019). Following uptake by plant roots, Cd moves through the vascular bundles to other organs, including edible parts of the plant. Thus, by entering the food chain, Cd constitutes a human health hazard (Nawrot et al., 2006;Liu et al., 2010;Clemens et al., 2013). The type II oxidation capacity and electronegativity of Cd mainly explain its toxic nature; it can form complexes with a wide variety of ligands, mainly with weak donors such as sulfide, nitrogen, and selenium (Salt and Wagner, 1993;Ismael et al., 2019). One major toxic effect of Cd is redox imbalance due to disturbances of the antioxidant system, damage to the respiratory chain, and the induction of Fenton-type reactions (Cuypers et al., 2016;Romero-Puertas et al., 2019). Interestingly, one of the gene categories found in transcriptomic analyses of plant responses to Cd includes biotic stress responses, particularly to fungi, although little is known about crosstalk in the plant responses to Cd and fungal infections.
Pathogenic fungal microorganisms, which have been classified according to their mode of action, use a diverse range of mechanisms to infect plants. Necrotrophic pathogens use ROS/ RNS, toxins, and cell-wall-degrading enzymes, among other mechanisms, to obtain nutrients from dead tissues (Wolpert et al., 2002;Martínez-Medina et al., 2019). Some necrotrophic pathogens even induce the overproduction of NO to accelerate infection (van Baarlen et al., 2004;Sarkar et al., 2014;Floryszak-Wieczorek and Arasimowicz-Jelonek, 2016), which, depending on the intensity and timing of NO production, can activate plant defences (Asai and Yoshioka, 2009). Plants also activate other signalling pathways, such as JA-and ET-dependent signalling, to activate the expression of defence-related genes (Thomma et al., 2001;Kunkel and Brooks, 2002;Broekaert et al., 2006). Other phytohormones, such as gibberellins, play a key role in resistance to necrotrophic pathogens due to a degraded DELLA repressor, which activates plant growth (Achard et al., 2008) and interacts with a JA signalling repressor . Biotrophic fungal pathogens, which usually have a specific host, can induce effectors capable of suppressing plant immunity (Perfect and Green, 2001). In addition, fungi get their nutrients from living cells by maintaining host viability through specialized structural and biochemical relations (Gebrie, 2016). In some cases, fungi synthesize plant cytokinins to attract nutrients from the plant to infected tissues and to decrease the plant production of SA, thus activating plant defence biotrophic fungal genes (Choi et al., 2011;. Conversely, plants develop mechanisms to resist biotrophic fungal infections. These include a penetration resistance mechanism, which strengthens the cell wall and membrane to halt spore germination and to prevent the formation of haustoria. Plants can also activate programmed cell death accompanied by a ROS and NO burst, leading to a hypersensitive response in penetrated epidermal cells, to shut down the supply of nutrients to the fungus (Koeck et al., 2011). All of these plant defence signalling mechanisms could be points of crosstalk in plant responses to Cd and fungal pathogens; in fact, various studies have found that Cd treatments protect against fungal infections. For example, the induction of resistance to Fusarium oxysporum in Triticum aestivum by pre-treatment with Cd is related to GSH-induced glutathionylation, which protects proteins against oxidative damage (Mittra et al., 2004;Mohapatra and Mittra, 2017). In addition, ROS production and cell death decrease in Cd-treated Cajanus cajan which was further infected with Fusarium incarnatum, although this was not always associated with an increase in the antioxidant system (Satapathy et al., 2012). In Arabidopsis plants, increased resistance to B. cinerea following pre-treatment with Cd or Cu has been reported to be exclusively caused by the induction of defence genes such as PDF1.2 (Cabot et al., 2013).

Bioinformatic analysis of the redox footprint in plant responses to Cd and fungi
The large variability in treatments, tissues analysed, culture media, plant age, and other parameters in studies conducted so far makes it difficult to reach general conclusions concerning plant responses to Cd stress. However, bioinformatic analysis provides a straightforward way to identify and analyse a common set of transcripts in plant responses to different stresses, and to identify their specificity or otherwise to different parameters, which can be very useful for future research and to better understand the mechanisms and role of these transcripts in plant responses to stress. To obtain a deeper insight into the role of ROS/RNS and redox signalling in crosstalk between plant responses to Cd and fungal pathogens, we carried out a web search of the available transcriptome analyses relating to both stresses with the aid of the PubMed (https://www.ncbi.nlm.nih.gov/pubmed/), Gene Expression Omnibus (GEO) (https://www.ncbi.nlm.nih.gov/geo/), Recursos Científicos https://www.recursoscientificos.fecyt. es/), and Scopus https://www.scopus.com/home.uri) databases. When probe information for a dataset was available, no additional filters were applied, thus ensuring that data originally filtered by the authors were used. In five studies, the differentially expressed probe lists were acquired by reanalysing the data stored in GEO. We used the GEO2R web tool (http:// www.ncbi.nlm.nih.gov/geo/info/geo2r.html) with default options for differential analysis and gene list acquisition [false discovery rate (FDR) <0.05; fold change (FC) >2.0]. The search was narrowed to A. thaliana, which is a model plant with a larger number of available analyses, in response to Cd and a diverse range of fungi, such as F. oxysporum, Fusarium graminearum, and B. cinerea; these pathogens, which can infect over 150 economically important crops, are responsible for one of the highest reductions in crop productivity (Dean et al., 2012). We analysed 19 microarray/RNA-seq datasets from eight different studies related to A. thaliana responses to Cd (Table 1), and 12 datasets from five studies of responses to fungi ( Table 2).
The shortage of crop species data in some cases and barely identified transcripts in others, as well as the variability in the nomenclature used to define genes, are major barriers to carrying out bioinformatic meta-analysis. We used rice (Oryza sativa L.), one of the most important cereal crops, as a model monocotyledonous plant, although only 25% of the data published could be analysed in our meta-analysis. Rice, which is the principal food for almost half of the world's population, is usually grown in paddy fields under flood conditions, and is therefore more susceptible to heavy metals contamination . We identified four different profile analyses in three studies of rice responses to Cd and 15 profile analyses in five studies of rice responses to Magnaporthe oryzae, which causes blast disease and seriously affects rice yields (Sánchez-Sanuy et al., 2019) (Table 1 and Table 2).
Expression profiles of genes involved in ROS/RNS and redox-related categories according to the Gene Ontology (GO) resource (http://geneontology.org/) (Table 3) were analysed in the transcriptomes described in Tables 1 and 2. These categories include 210 genes in A. thaliana and 218 genes in O. sativa (see Table S1 at Zenodo Repository, https://zenodo. org/record/5040382#.YNrth5j7S71). A total of 82 RBOHDand H 2 O 2 -dependent genes in systemic responses to different stress conditions have also been analysed . Probes were annotated with locus identifiers using the TAIR Microarray Elements Search and Download tool for A. thaliana or were converted to ORF IDs using the UniProt (https://www.uniprot.org/) and NCBI GPL19274 (https:// www.ncbi.nlm.nih.gov/geo) databases for O. sativa. All probes were then categorized under the following headings: no data/ no change, increase, and decrease. After the first analysis, genes not expressed in any treatment were removed and the selected data were reanalysed. We then performed a hierarchical clustering analysis to objectively search for groups of probes in an unsupervised manner without specifying the number of clusters to be created. We used H-clustering, heatmaply, and htmlwidgets in the R software package to do this.

Arabidopsis thaliana
When analysing genes involved in ROS/RNS and the redox category (Table 3; Fig. S1 at Zenodo Repository, https://zenodo.org/record/5040382#.YNrth5j7S71), a group of A. thaliana genes that showed no changes in response to any of the stresses examined was removed. Further clustering analysis enabled us to find two clusters (I and II) for the stresses applied based on the induction or repression, respectively, of a group of 57 genes (group A; Fig. 2; Fig. S2, Table S2 at Zenodo Repository, https://zenodo.org/record/5040382#. YNrth5j7S71). Cluster I mainly involves the fungal pathogens The code of each paper appears in the first column and in the abscissa axis of B. cinerea and F. graminearum in plants growing in soil and the Cd treatment Cd_L_P_8, the longest treatment analysed (12 days) (Fig. 2). Cluster II involves most of the Cd treatments, F. oxysporum, and one study of B. cinerea with plants growing in sand supplemented with Hoagland solution. String analysis of these group A genes showed one main group, related to glutathione metabolism, to be the strongest KEGG pathway ( Fig. 3A; Table S2 at Zenodo Repository, https:// zenodo.org/record/5040382#.YNrth5j7S71), as well as genes associated with ASC metabolism, particularly those encoding dehydro-and monodehydro-ascorbate reductases. As H 2 O 2 has been shown to be directly related to glutathione status, different H 2 O 2 -dependent signalling pathways may be regulated by GSH (Noctor et al., 2012). Given its chemical properties, glutathione, which can undergo different redox reactions, is a key molecule involved in the regulation of the cellular redox network (Noctor et al., 2012). Genes related to glutathione metabolism from group A mainly include glutathione S-transferases (GSTs) and two glutathione peroxidases. GSTs are a diverse group of AT3G11630  AT1G78370  AT3G26060  AT4G09010  AT4G35090  AT5G64940  AT1G10360  AT2G37240  AT5G16400  AT1G64860  AT1G67740  AT5G06690  AT1G67840  AT3G04340  AT5G21100  AT1G05690  AT1G17190  AT1G54050  AT5G17220  AT2G47730  AT1G78380  AT1G77120  AT5G37670  AT5G39950  AT2G29440  AT3G43800  AT4G08390  AT1G03680  AT5G59720  AT3G06730  AT1G52990  AT1G48030  AT1G21350  AT1G43560  AT3G52960  AT4G36910  AT5G16710  AT1G10370  AT3G06590  AT4G35970  AT1G02950  AT5G63310  AT2G41680  AT3G51030  AT3G55040  AT5G06290  AT4G21870  AT4G31870  AT1G77490  AT5G58330  AT2G25080  AT3G02730  AT3G54660  AT5G07460  AT5G01600  AT1G20630  AT1G52760  AT1G59700  AT5G42850  AT4G14030  AT1G53580  AT1G57720  AT5G37770  AT4G37610  AT1G20620  AT1G49860  AT3G16950  AT1G76080  AT1G35580  AT1G63940  AT4G12570  AT1G18150  AT3G63080  AT5G45020  AT1G32230  AT5G10860  AT3G04120  AT1G09640  AT2G19310  AT2G18170  AT1G65980  AT3G29035  AT5G58940  AT1G76760  AT2G48150  AT3G09640  AT1G22770  AT1G59670  AT3G14395  AT2G47900  AT5G43940  AT5G37830  AT4G32320  AT3G52880  AT3G24170  AT1G07890  AT2G30860  AT5G42150  AT2G17420  AT1G27130  AT2G30870  AT4G34120  AT4G11600  AT1G78340  AT1G02940  AT1G27140  AT4G35460  AT2G31570  AT1G19550  AT1G08320  AT5G06839  AT5G39610  AT3G48360  AT5G63160  AT3G09940  AT5G05300  AT4G23810  AT3G10930  AT4G19880  AT2G29480  AT1G69930  AT1G65820  AT2G29490  AT1G65970  AT2G29470  AT1G16420  AT5G12020  AT1G60740  AT3G17240  AT5G24110  AT5G44990  AT1G69920  AT3G09270  AT2G29460  AT1G74590  AT5G02780  AT5G05410  AT2G26150  AT1G59860  AT2G29450  AT1G73260  AT1G75270  AT2G29420  AT1G17170  AT5G62480  AT1G17180  AT2G02390  AT3G17790  AT5G67480  AT5G03630  AT2G02930  AT1G02930  AT1G02920  AT1G53540  AT5G12030  AT2G29500  AT1G07400  AT3G46230 Table S1 at Zenodo related to ROS/RNS and redox categories from Arabidopsis, which show changes in response to the different stresses. Gene up-regulation and down-regulation are indicated in blue and brown, respectively. Data were obtained from plant responses to Cd and fungal pathogen stresses described in Tables 2 and 3. Unbiased hierarchical clustering showed two clusters, I and II. Genes from groups A and B (both framed in red) were differentially regulated in clusters I and II. The code for each study (shown at the bottom) is represented by the metal or pathogen used and is described in Tables 2 and 3. multi-functional proteins essential for protecting plants against oxidative damage, in what has been classified as a phase II detoxification system (reviewed in Gullner et al., 2018). GSTs catalyse the conjugation of GSH to a variety of electrophilic and hydrophobic substrates, including xenobiotic compounds, which are then sequestered in vacuoles to prevent substrate toxicity. GSTs are also involved in removing excess lipid hydroperoxides produced in response to stress (Gullner et al., 2018). Plant GSTs have been categorized into four classes: phi, tau, lambda, and dehydroascorbate reductase GSTs (Edwards and Dixon, 2005). Although the precise metabolic functions of GST isoenzymes in plant infection and abiotic stress have not been determined, their most important role, acting as glutathione peroxidases, could be to affect lipid hydroperoxides. GST transcripts have been reported to be up-regulated in response to stress conditions, such as fungal or bacterial infection (reviewed in Gullner et al., 2018), heavy metals, cold, salt, H 2 O 2 , UV, and light (reviewed in Kumar and Trivedi, 2018). However, their single-/multiple-stress responsiveness or possible redundant functions depend on the class of GSTs to which they belong (Sappl et al., 2009). We have identified a group of genes that are regulated under Cd treatment and fungal infection regardless of a wide range of experimental conditions. The induction of a group of GST-encoding genes suggests that the induction of Cd-stress-related genes could provide protection against fungal infection. Following string analysis, a smaller number of genes from group A were also grouped together on the basis of protein processing in the endoplasmic reticulum (ER) (Fig. 3A; Table S2 at Zenodo Repository, https://zenodo.org/record/5040382#. YNrth5j7S71) and, in particular, of ER-associated degradation (ERAD); this subgroup of genes encoded heat shock proteins. ERAD is involved in the degradation of terminally misfolded proteins. In fact, in Arabidopsis plants, low concentrations of ROS, acting as signalling molecules, have been shown to induce ER stress-related genes, whose regulation is dependent on the compartment from which the ROS originated, such as the chloroplasts, mitochondria, and peroxisomes (Ozgur et al., 2015). In our study, ERAD cluster I genes were repressed mainly by B. cinerea and long-term Cd treatment, while cluster II genes were induced. Repression of ERAD may induce ER stress, which activates signalling pathways or unfolded protein responses involved in ER protection, which, when insufficient to restore ER function, can lead to cell death by apoptosis.
Group B, containing 23 probes (Table S2 at Zenodo Repository, https://zenodo.org/record/5040382#.  Fig. 2) related to ROS/ RNS and redox metabolism and differentially regulated in clusters I and II. These genes showed one main group related to glutathione metabolism (in red), the strongest KEGG pathway, and a smaller group related to protein processing in the endoplasmic reticulum (in blue), as described in Table S2 at Zenodo. (B) String analysis of genes from group C (see Fig. 4) related to systemic RBOHD-and H 2 O 2 -dependent transcripts from Arabidopsis and differentially regulated in clusters I and II. These genes showed one main group related to responses to chitin (in red) and responses to chitin, as well as the cysteine-rich transmembrane (CYSTM) domain (in blue), the strongest KEGG pathway, as described in Table S2 at Zenodo. YNrth5j7S71), was induced in cluster I, but, unlike group A, no changes or distinct types of induction were observed in cluster II (Fig. 2). String analysis of group B did not show any clear interacting groups, although the genes involved appear to be mainly related to the glutathione metabolism by GSTs and to antioxidant-detoxification processes (Table S2 at Zenodo Repository, https://zenodo.org/record/5040382#. YNrth5j7S71). Our results show that both groups A and B were mainly related to genes encoding GSTs, with specific footprints being observed in both clusters. As described above, our experimental results indicate the important role played by these genes in plant protection against Cd and fungal stresses, as has previously been described with respect to wheat and F. oxysporum (Mittra et al., 2004;Mohapatra and Mittra, 2017). Therefore, glutathione metabolism, and particularly the GSTrelated metabolism, may be key players in the crosstalk between heavy metal and fungal pathogen stress responses. In fact, Arabidopsis mutants overexpressing GSTs show higher tolerance to fungal infection (Gullner et al., 2018) and to various abiotic stresses such as heavy metals, cold, and salt (Kumar and Trivedi, 2018).
When analysing systemic RBOHD-and H 2 O 2 -dependent transcripts, we also found two clusters (I and II) corresponding to a group of 30 genes (group C) that were induced or repressed, respectively, under the stresses applied ( Fig. 4; Fig. S3, Table S2 at Zenodo Repository, https://zenodo.org/record/5040382#.YNrth5j7S71). Clusters in this analysis were similar to those previously analysed except for the Cd_L_P_8 treatment, which is now included in cluster II with all the other Cd treatments. String analysis of the 30 group C genes found a main group based on the biological process: response to chitin (Fig. 3B, Table S2 at Zenodo Repository, https://zenodo.org/ record/5040382#.YNrth5j7S71). Perception of fungal pathogens by the plant occurs through the recognition of chitin, a polymer component of the fungal cell wall, followed by the activation of the plant immune response (Squeglia et al., 2017). Our bioinformatic analysis showed that gene group C is downregulated in cluster II, which is mostly composed of B. cinerea  Tables 2 and 3. Unbiased hierarchical clustering showed two clusters, I and II. Genes from group C (framed in red) were differentially regulated in clusters I and II. The code for each study (shown at the bottom) is represented by the metal or pathogen used and is described in Tables 2 and 3. treatments. The process of infection by B. cinerea includes an initial production of local necrotic lesions followed by lesion spreading at a later stage (Bi et al., 2021), suggesting that the plant response to the pathogen is repressed. Cd-induced genes related to responses to chitin may help to protect plants against fungal infection following Cd treatment, a process that requires further exploration. Interestingly, different plant culture conditions may affect the expression of the group C genes, as B. cinerea with plants cultured in river sand supplemented with Hoagland solution, as well as F. oxysporum with plants cultured in Murashige and Skoog medium supplemented with sucrose, showed an opposite trend in gene expression to that for fungi such as B. cinerea and F. graminearum with plants cultured in soil.

Oryza sativa
The clustering of data from O. sativa has been complicated, probably due to lower availability of data and the diversity of cultivars used; each transcriptomic analysis of Cd treatment was carried out with a different cultivar, and the behaviour of these different cultivars may differ under similar environmental conditions. In addition, different lines, which were either compatible or incompatible with the fungal pathogen M. oryzae, were analysed in the same cultivar. Despite these problems, clustering analysis of transcriptome changes in genes involved in ROS/RNS and redox categories (Table 3) in rice responses to Cd and M. oryzae enabled us to find two clusters (I and II) for the stresses applied, based on the induction or repression, respectively, of a number of genes (group D; Fig. 5; Fig. S4, Table S2 at Zenodo Repository, https://zenodo. org/record/5040382#.YNrth5j7S71). Cluster I involves both compatible and incompatible rice interactions M. oryzae, with different timings; this suggests that different induction/repression waves of redox-related genes take place during the treatment, which are associated with a type of interaction. Cluster II involves all the other treatments analysed, in most of which only a few genes underwent changes (Fig. 5). Cluster I and    Table S1 at Zenodo. Gene up-regulation and down-regulation are indicated in blue and brown, respectively. Data were obtained from plant responses to Cd and fungal pathogen stresses described in Tables 2 and 3. Unbiased hierarchical clustering showed two, clusters I and II. Genes from group D (framed in red) were differentially regulated in response to abiotic and biotic stresses. The code for each study (shown at the bottom) is represented by the metal or pathogen used and is described in Tables 2 and 3.
Cd_L_R_9 behaved similarly to a group of 32 induced genes, which were repressed in cluster II. String analysis of these genes showed no gene pooling; most of the genes were related to glutathione metabolism, the strongest KEGG pathway, mainly encoding GSTs (Table S2, Fig. S5 at Zenodo Repository, https://zenodo.org/record/5040382#.YNrth5j7S71). These results suggest that rice plants growing in Cd for short to medium periods of time may also show induction of GST activity and therefore be more resistant to fungal pathogens, similar to the findings with Arabidopsis plants and in previous studies of wheat (Mittra et al., 2004;Mohapatra and Mittra, 2017).

Conclusions and perspectives
Plant responses to certain stresses have been well characterized when applied individually, which has provided the basis for establishing models with key components involved in plant responses to stress. However, as plants are usually confronted with more than one stress in the field, we need to build similar models for serial and combined stresses, which would be unique for each combination. Combinations of abiotic and biotic stresses are of particular importance given the singular nature of each interaction between two or more organisms. Recent advances in the study of plant responses to combinations of stresses point to a role for key signalling molecules, including hormones, TFs, and, in particular, to ROS/ RNS and redox homeostasis, for selecting different pathways to achieve a trade-off between acclimation/survival and yield. Bioinformatic analyses of transcriptome changes in plant responses to Cd and fungal pathogens point to redox signalling at the crossroads of both these stresses, which is mainly related to the glutathione metabolism, particularly with respect to GST genes. We identified different groups of GST genes that are up-or down-regulated depending on the treatment (Cd/fungi). The results obtained indicate that genes encoding GSTs are a key gene family in relation to a broad range of species at the crossroads of plant responses to biotic and abiotic stresses. We identified other groups of genes, such as ERAD genes associated with heat shock proteins, as well as those involved in responses to chitin, which may also be involved in crosstalk between abiotic and biotic stresses, particularly Cd and fungal infections. Our bioinformatic findings should pave the way for more comprehensive future research into crosstalk between different stresses. The characterization of the key molecules identified in different stress combinations could lead to the development of new strategies to alleviate the effects of multifactorial stress conditions, especially in the current context of global climate change.