Transcriptome Analysis of Heat Shock Factor C2a Over-Expressing Wheat Roots Reveals Ferroptosis-like Cell Death in Heat Stress Recovery

Wheat (Triticum aestivum L.) growing areas in many regions of the world are subject to heat waves which are predicted to increase in frequency because of climate change. The engineering of crop plants can be a useful strategy to mitigate heat stress-caused yield losses. Previously, we have shown that heat shock factor subclass C (TaHsfC2a-B)-overexpression significantly increased the survival of heat-stressed wheat seedlings. Although previous studies have shown that the overexpression of Hsf genes enhanced the survival of plants under heat stress, the molecular mechanisms are largely unknown. To understand the underlying molecular mechanisms involved in this response, a comparative analysis of the root transcriptomes of untransformed control and TaHsfC2a-overexpressing wheat lines by RNA-sequencing have been performed. The results of RNA-sequencing indicated that the roots of TaHsfC2a-overexpressing wheat seedlings showed lower transcripts of hydrogen peroxide-producing peroxidases, which corresponds to the reduced accumulation of hydrogen peroxide along the roots. In addition, suites of genes from iron transport and nicotianamine-related gene ontology categories showed lower transcript abundance in the roots of TaHsfC2a-overexpressing wheat roots than in the untransformed control line following heat stress, which are in accordance with the reduction in iron accumulation in the roots of transgenic plants under heat stress. Overall, these results suggested the existence of ferroptosis-like cell death under heat stress in wheat roots, and that TaHsfC2a is a key player in this mechanism. To date, this is the first evidence to show that a Hsf gene plays a key role in ferroptosis under heat stress in plants. In future, the role of Hsf genes could be further studied on ferroptosis in plants to identify root-based marker genes to screen for heat-tolerant genotypes.


Introduction
Heat is a major abiotic stress factor that affects crop production in many regions of the world. Depending on the intensity and duration of heat stress and the physiological and developmental status of the plant, most plant species are negatively affected by heat stress. However, heat stress is especially problematic for cool season crops, such as wheat (Triticum aestivum L.). Because wheat is a crop that meets the calorie demand of a significant portion of the world's population, a reduced wheat production caused by heat stress can potentially be a significant threat to global food security. Indeed, according to one estimate, the average cereal yield across the globe declined by 7.6% due to heat stress during extreme weather events that occurred between 1964 and 2007 [1].
In addition to the elevated temperatures currently experienced in many wheat-growing regions, the increased atmospheric concentration of CO 2 and other greenhouse gases is Int. J. Mol. Sci. 2023, 24, 3099 2 of 17 predicted to cause even more intense heat waves and dramatically alter rainfall patterns [2]. Indeed, the frequency, duration, and severity of episodes with exceptionally high temperatures have already been seen to rise in recent years [3,4]. These unusual climatic changes, which are projected to become more severe in future, have the potential to further threaten global food security by severely limiting crop productivity [1]. Global wheat yield is projected to decline by 4.1 to 6.4% with every 1 • C rise in global temperatures [5]. Deryng et al. [6] predicted that strong heat stress during anthesis could be responsible for up to a 52% reduction from the expected yield of spring wheat by 2080.
Plants respond to heat stress by activating many genes involved in hormone biosynthesis, protein degradation, transcriptional activation and repression, and an unfolded-protein response [7][8][9]. In addition, heat stress induces the production of reactive oxygen species (ROS), which can damage membranes and other cellular systems through lipid peroxidation [10]. Distefano et al. [11] demonstrated that heat stress induces ferroptosis-like cell death in Arabidopsis (Arabidopsis thaliana (L.) Heynh.) roots. Ferroptosis is a non-apoptotic form of cell death and is dependent on intracellular iron and ROS [12][13][14][15]. Excess free iron (Fe) ions can generate ROS within the cell [16], resulting in lipid peroxidation and ferroptotic cell death, and the inhibition of this response was shown to have a beneficial effect on heat stress tolerance.
Heat stress also induces the production of heat shock proteins (HSPs) that act as molecular chaperones under stress conditions to maintain cellular homeostasis [17,18]. Heat shock factors (HsfA, HsfB and HsfC) play a key role in the regulation of heat-inducible genes, including HSPs [19]. HsfA and HsfB have a central role in the regulation of heat stress-related genes in plants [20]. However, the monocot-specific HsfC2 subclass HSPs have been poorly studied [21].
To date, most studies have considered the detrimental effects of heat stress on aboveground plant parts [22,23], although the roots can also be affected by high temperatures under natural conditions, especially when plant canopies are not dense enough and sunlight can directly contact the soil surface [24]. Given that heat and drought stress often occur simultaneously, maintaining a functional root system, that can adsorb water and nutrients from the soil, could be essential to alleviate the damaging effects of heat and drought stress, especially during the early developmental stages of plant growth. Indeed, reducing root heat stress was found to increase shoot biomass by 33 to 160% and grain yield by 18 to 147% in wheat under field conditions [25]. However, currently very little is known about the molecular responses of plant roots to heat stress. Recently, it has been reported that exogenous trehalose modulates gene expression and alleviates the oxidative damage of high temperature stress in wheat roots [26]. In our previous study, it has been shown that the survival rates of TaHsfC2a-B-overexpressing wheat seedlings were significantly improved (90%) relative to those of untransformed plants (15%) after exposure to heat stress (2 h at 43 • C) [27]. In addition, a significant correlation was observed between recovery rates and the average root dry weights of the recovered transgenic lines which were 8.7 and 7.8-fold higher than those of the recovered parental line, suggesting that physiological changes in the roots of transgenic plants may have contributed to the heat stress recovery phenomenon.
The aim of this study was firstly to investigate transcriptional responses of wheat roots to heat stress, and secondly to understand the potential roles of TaHsfC2a over-expressing roots in contributing to plant recovery after heat stress that can kill most untransformed control plants [27]. Therefore, this study undertook RNA-seq analyses to identify heat stress-inducible root genes that are potentially associated with heat stress tolerance in wheat. A comparative analysis of untransformed control plants (hereafter called wild-type) and TaHsfC2a root transcriptomes indicated that TaHsfC2a negatively regulates ROS levels under heat stress in roots through class III peroxidases. In addition, the reduced transcripts of Fe transport genes and nicotianamine synthases caused a lower accumulation of Fe in TaHsfC2a roots under heat stress. These results suggested the existence of ferroptosis-like cell death in wheat. To the best of our knowledge, no functional genes have been identified in the ferroptosis-like cell death mechanism seen under heat stress in plants. Therefore, this study is the first report that functionally characterizes a gene in the ferroptosis-like cell death pathway under heat stress in plants. Overall, these results provide new insights into heat stress tolerance in plants.

Root Transcriptome of Wild-Type Wheat Seedlings under Heat Stress
To identify root transcriptional responses to heat stress, the RNA-seq of wild-type roots with and without heat stress was performed. On average, 90% of the reads were aligned to the reference RefSeq v1.0 (Supplementary Table S1). RNA-seq analysis was carried out with log 2 (Fold Change, FC ≥ 2; p < 0.05), unless otherwise stated to identify the genes that are at least twofold induced or repressed with a p value of < 0.05. We found that 1666 and 1660 genes were up-and down-regulated, respectively, in wild-type roots after heat treatment (Figure 1; Supplementary Table S2).
To gain an insight into the nature of molecular responses to heat, gene ontology (GO) enrichment analysis of the differentially expressed genes (DEGs) in the root tissue was performed, and found that the genes belonging to 46 GO categories were up-regulated under heat stress in wild-type roots (Supplementary Table S3). Some of these GO categories include "response to heat, nicotianamine synthase activity, the nicotianamine biosynthetic process, the tricarboxylic acid biosynthetic process, protein folding, response to hydrogen peroxide, ROS, iron ion transport, glutathione transferase activity, DNA binding transcription factor activity, heat shock protein binding, transcriptional repressor activity, and RNA polymerase II proximal promoter sequence-specific DNA binding" (Supplementary Table S3). In the down-regulated gene sets, 86 GO categories were over-represented. Some of these categories include "ROS, oxidative stress and hydrogen peroxiderelated, cellular oxidant detoxification, peroxidase activity, antioxidant activity, cell proliferation and cell wall-related, cellulose and signalling-related" (Supplementary Table  S3). In addition, in wild-type roots, 182 and 75 transcription factors (TF) encoding genes were up-and down-regulated, respectively after heat stress ( Figure 2; Supplementary Table S4). The most abundant TF families up-regulated in wild-type roots were HSFs (24), ERFs (22), NACs (21), MYBs (15), and WRKYs (15) (Figure 2A) while those down-regulated in wild-type roots were bHLHs (11), bZIPs (10), and NF-YAs (8) ( Figure 2B).  To gain an insight into the nature of molecular responses to heat, gene ontology (GO) enrichment analysis of the differentially expressed genes (DEGs) in the root tissue was performed, and found that the genes belonging to 46 GO categories were up-regulated under heat stress in wild-type roots (Supplementary Table S3). Some of these GO categories include "response to heat, nicotianamine synthase activity, the nicotianamine biosynthetic process, the tricarboxylic acid biosynthetic process, protein folding, response to hydrogen peroxide, ROS, iron ion transport, glutathione transferase activity, DNA binding transcription factor activity, heat shock protein binding, transcriptional repressor activity, and RNA polymerase II proximal promoter sequence-specific DNA binding" (Supplementary Table  S3). In the down-regulated gene sets, 86 GO categories were over-represented. Some of these categories include "ROS, oxidative stress and hydrogen peroxide-related, cellular oxidant detoxification, peroxidase activity, antioxidant activity, cell proliferation and cell wall-related, cellulose and signalling-related" (Supplementary Table S3). In addition, in wild-type roots, 182 and 75 transcription factors (TF) encoding genes were up-and downregulated, respectively after heat stress ( Figure 2; Supplementary Table S4). The most abundant TF families up-regulated in wild-type roots were HSFs (24), ERFs (22), NACs (21), MYBs (15), and WRKYs (15) (Figure 2A) while those down-regulated in wild-type roots were bHLHs (11), bZIPs (10), and NF-YAs (8) ( Figure 2B).

Heat Stress Regulation of Genes in the Roots of TaHsfC2a-Overexpressing Wheat Seedli
As indicated above, previously, it has been reported that the TaHsfC2a-overexp ing plants exposed to heat stress (2 h of heat treatment at 43 °C) showed significa better recovery rates (> 90%) in the subsequent 3 weeks following the heat treatment wild-type plants (15%) [27]. The evidence presented by Hu et al. [27] also indicated the roots played a role in the recovery phenomenon. To understand the potential me nisms involved in the recovery of TaHsfC2a-overexpressing lines after heat stress, we formed an RNA-seq analysis of the roots. The analysis showed that 1,535 and 2,487 g were up-and down-regulated in transgenic roots in response to heat stress ( Figure 1; plementary Table S2).
In the up-regulated gene sets, several GO categories were observed, including ondary metabolic processes and responses to stress, and amine metabolic processes w overrepresented in the wild-type plants (Figure 3), which were also overrepresente the transgenic line. To further understand the differences in GO enrichment between w type and transgenic plants, we have selected highly significant GO categories (FD 0.001) from the BLAST2GO enrichment analysis. We found 38 categories over-represe relative to the wild-type in the transgenic up-regulated gene dataset, including the r lation of hydrogen peroxide metabolic process, the regulation of reactive oxygen sp metabolic process, glutathione metabolic process, flavonoid biosynthetic process, calc ion binding, transition metal ion transport, UDP-glucosyltransferase activity, and D binding and regulation of the RNA biosynthetic process (Supplementary Table S3). A of 42 GO categories down-regulated in the transgenic roots relative to the heat-tre wild-type, including the nucleosome and chromatin assembly, protein-DNA packa related, oxidation-reduction process, cell wall polysaccharide, metal ion binding, membrane transporter activity (Supplementary Table S3). Among the DEGs between

Heat Stress Regulation of Genes in the Roots of TaHsfC2a-Overexpressing Wheat Seedlings
As indicated above, previously, it has been reported that the TaHsfC2a-overexpressing plants exposed to heat stress (2 h of heat treatment at 43 • C) showed significantly better recovery rates (> 90%) in the subsequent 3 weeks following the heat treatment than wildtype plants (15%) [27]. The evidence presented by Hu et al. [27] also indicated that the roots played a role in the recovery phenomenon. To understand the potential mechanisms involved in the recovery of TaHsfC2a-overexpressing lines after heat stress, we performed an RNA-seq analysis of the roots. The analysis showed that 1535 and 2487 genes were upand down-regulated in transgenic roots in response to heat stress ( Figure 1; Supplementary  Table S2).
In the up-regulated gene sets, several GO categories were observed, including secondary metabolic processes and responses to stress, and amine metabolic processes were overrepresented in the wild-type plants (Figure 3), which were also overrepresented in the transgenic line. To further understand the differences in GO enrichment between wild-type and transgenic plants, we have selected highly significant GO categories (FDR ≤ 0.001) from the BLAST2GO enrichment analysis. We found 38 categories over-represented relative to the wild-type in the transgenic up-regulated gene dataset, including the regulation of hydrogen peroxide metabolic process, the regulation of reactive oxygen species metabolic process, glutathione metabolic process, flavonoid biosynthetic process, calcium ion binding, transition metal ion transport, UDP-glucosyltransferase activity, and DNA binding and regulation of the RNA biosynthetic process (Supplementary Table S3). A total of 42 GO categories down-regulated in the transgenic roots relative to the heat-treated wildtype, including the nucleosome and chromatin assembly, protein-DNA packaging related, oxidation-reduction process, cell wall polysaccharide, metal ion binding, and membrane transporter activity (Supplementary Table S3). Among the DEGs between the control and heat-treated seedlings, 137 TFs were up-regulated, and 128 TFs were down-regulated in the transgenic roots, respectively ( Figure 2; Supplementary Table S4). The most abundant TF families up-regulated in the roots of TaHsfC2a-overexpressing plants were HSFs (22), ERFs (17), C2H2 zinc fingers (13), NACs (12), bZIPs (12), and bHLHs (12) (Figure 2A), while the most abundant TFs down-regulated were bHLHs (17), MYBs (15), NACs (15), and NF-YA (12) ( Figure 2B). These results indicated that several TFs were differentially regulated in the transgenic roots.  Table S4). The most abundant TF families up-regulated in the roots of TaHsfC2a-overexpressing plants were HSFs (22), ERFs (17), C2H2 zinc fingers (13), NACs (12), bZIPs (12), and bHLHs (12 (Figure 2A), while the most abundant TFs down-regulated were bHLHs (17), MYBs (15) NACs (15), and NF-YA (12) ( Figure 2B). These results indicated that several TFs were dif ferentially regulated in the transgenic roots.

TaHsfC2a Negatively Regulates Root ROS Levels after Heat Treatment
As indicated above, the analyses found several GO categories related to ROS, such as those involved in the regulation of hydrogen peroxide metabolism, regulation of ROS metabolic processes, and glutathione metabolism, in both the wild-type and TaHsfC2a overexpressing plants (Supplementary Table S3). It is well known that heat stress gener ates ROS [28,29], and class III peroxidases have important functions in maintaining the ROS balance in the plant cell by consuming and/or producing ROS [30]. Therefore, a com parative analysis was performed, and it revealed the differences in peroxidase expressions between wild-type and the transgenic line after heat treatment. At least 70 class III perox idases were significantly down-regulated in the roots of TaHsfC2a-overexpressing plants while only four peroxidases significantly up-regulated when compared with the wild type after heat treatment (Table 1). To understand whether the down-regulation of perox idases in the roots of TaHsfC2a-overexpressing plants caused a difference in the accumu lation of ROS, hydrogen peroxide levels were measured in the roots of the wild-type and transgenic line before and after heat treatment and found that transgenic roots accumu lated lower hydrogen peroxide than wild-type roots ( Figure 4). These results indicated that the overexpression of TaHsfC2a may be negatively regulating hydrogen peroxide level in transgenic roots by down-regulating the hydrogen peroxide-producing class II peroxidases.

TaHsfC2a Negatively Regulates Root ROS Levels after Heat Treatment
As indicated above, the analyses found several GO categories related to ROS, such as those involved in the regulation of hydrogen peroxide metabolism, regulation of ROS metabolic processes, and glutathione metabolism, in both the wild-type and TaHsfC2aoverexpressing plants (Supplementary Table S3). It is well known that heat stress generates ROS [28,29], and class III peroxidases have important functions in maintaining the ROS balance in the plant cell by consuming and/or producing ROS [30]. Therefore, a comparative analysis was performed, and it revealed the differences in peroxidase expressions between wild-type and the transgenic line after heat treatment. At least 70 class III peroxidases were significantly down-regulated in the roots of TaHsfC2a-overexpressing plants, while only four peroxidases significantly up-regulated when compared with the wild-type after heat treatment (Table 1). To understand whether the down-regulation of peroxidases in the roots of TaHsfC2a-overexpressing plants caused a difference in the accumulation of ROS, hydrogen peroxide levels were measured in the roots of the wild-type and transgenic line before and after heat treatment and found that transgenic roots accumulated lower hydrogen peroxide than wild-type roots ( Figure 4). These results indicated that the overexpression of TaHsfC2a may be negatively regulating hydrogen peroxide level in transgenic roots by down-regulating the hydrogen peroxide-producing class III peroxidases.

Tahsfc2a-Overexpression Affected Iron Transport Genes and Iron Accumulation in Roots after Heat Treatment
In addition, Fe ion transport and nicotianamine-related GO categories were highly up-regulated in both wild-type and TaHsfC2a-overexpressing line transcriptomes after heat stress (Supplementary Table S3). Nicotianamine is a precursor of phytosiderophores, which are excreted by the roots of Poaceae species, and mediate the chelation and acquisition of Fe ions [31]. Recently, Distefano et al. [11] identified Fe and ROS-dependent cell death could be induced by heat stress in Arabidopsis roots. To explore a possible involvement of Fe-mediated cell death phenomenon in heat stress tolerance, we examined the expression profiling of genes related to Fe transport and nicotianamine synthase activity and found heat to down-regulate the expression profiles of these genes in TaHsfC2a over-expressing plants when compared with wild-type roots ( Figure 5). Further, the TaHsfC2a over-expressing plant roots were found to accumulate low Fe ions when compared with the wild-type roots after heat treatment ( Figure 6). Therefore, these results suggested that TaHsfC2aoverexpression was reducing Fe accumulation in the roots, and thus, in turn, reducing ROS-and iron-mediated cell death response, enabling transgenic roots to recover from the detrimental effects of heat stress.

Tahsfc2a-Overexpression Affected Iron Transport Genes and Iron Accumulation in Roots after Heat Treatment
In addition, Fe ion transport and nicotianamine-related GO categories were highly up-regulated in both wild-type and TaHsfC2a-overexpressing line transcriptomes after heat stress (Supplementary Table S3). Nicotianamine is a precursor of phytosiderophores, which are excreted by the roots of Poaceae species, and mediate the chelation and acquisition of Fe ions [31]. Recently, Distefano et al. [11] identified Fe and ROS-dependent cell death could be induced by heat stress in Arabidopsis roots. To explore a possible involvement of Fe-mediated cell death phenomenon in heat stress tolerance, we examined the expression profiling of genes related to Fe transport and nicotianamine synthase activity and found heat to down-regulate the expression profiles of these genes in TaHsfC2a overexpressing plants when compared with wild-type roots ( Figure 5). Further, the TaHsfC2a over-expressing plant roots were found to accumulate low Fe ions when compared with the wild-type roots after heat treatment ( Figure 6). Therefore, these results suggested that TaHsfC2a-overexpression was reducing Fe accumulation in the roots, and thus, in turn, reducing ROS-and iron-mediated cell death response, enabling transgenic roots to recover from the detrimental effects of heat stress.

Wheat Root and Leaf Transcriptomes Show Low Correlation under Heat Stress
To explore if wheat roots and leaves/shoots show similar transcriptional response to heat, we have compared our root transcriptional data to recently published leaf tran scriptome data [32]. The cultivars used in these two studies were different; however, th age of the seedlings and the temperature and duration of heat treatment were similar be tween the two experiments. Overall, the root transcriptome after heat treatment showed relatively low correlations with the transcriptome of 1 h (r = 0.42) and 6 h (r = 0.27) heat treated leaves when both up-and down-regulated genes were included in the analysi (Figure 7). Furthermore, our analysis also showed that 3331 genes were at least twofold or more differentially regulated (log2 FC ≥ 2) in the roots under heat stress. Of these 333 genes, only 474 (14%) and 546 (16%) genes were significantly differentially regulated in 1 and 6-h in heat-treated leaves, respectively (Supplementary Table S5). Surprisingly, 2 and 104 genes showed opposite transcript levels in the 1-and 6-h heat-treated leaves, re spectively (Supplementary Table S5). In addition, several NAS genes were up-regulated in the roots; hence, we looked at the transcript levels of the nicotianamine synthases (NAS in the leaves. However, none of the NAS-related genes were significantly up-regulated in the leaves (Table 2). These results indicated that the overall transcriptomic responses o heat-treated leaves and heat-treated roots may not be similar, and root-specific iron-re lated mechanism may exist under heat stress.

Wheat Root and Leaf Transcriptomes Show Low Correlation under Heat Stress
To explore if wheat roots and leaves/shoots show similar transcriptional responses to heat, we have compared our root transcriptional data to recently published leaf transcriptome data [32]. The cultivars used in these two studies were different; however, the age of the seedlings and the temperature and duration of heat treatment were similar between the two experiments. Overall, the root transcriptome after heat treatment showed relatively low correlations with the transcriptome of 1 h (r = 0.42) and 6 h (r = 0.27) heat-treated leaves when both up-and down-regulated genes were included in the analysis (Figure 7). Furthermore, our analysis also showed that 3331 genes were at least twofold or more differentially regulated (log 2 FC ≥ 2) in the roots under heat stress. Of these 3331 genes, only 474 (14%) and 546 (16%) genes were significantly differentially regulated in 1-and 6-h in heat-treated leaves, respectively (Supplementary Table S5). Surprisingly, 29 and 104 genes showed opposite transcript levels in the 1-and 6-h heat-treated leaves, respectively (Supplementary Table S5). In addition, several NAS genes were up-regulated in the roots; hence, we looked at the transcript levels of the nicotianamine synthases (NAS) in the leaves. However, none of the NAS-related genes were significantly up-regulated in the leaves (Table 2). These results indicated that the overall transcriptomic responses of heat-treated leaves and heat-treated roots may not be similar, and root-specific iron-related mechanism may exist under heat stress.   Table 2. Fold changes (heat/control ratio) observed at the transcript levels of nicotianamine synthase genes in the roots and the leaves of the wheat cultivars, wild-type (this study) and TAM107 [32], respectively.

Discussion
Heat stress is becoming an immense problem in crop production around the world. Although heat can have a drastic effect on roots [33], most studies have focused on aboveground tissues [23]. Here, we have analyzed the effects of heat shock on seedling roots in both the wild-type and a transgenic line over-expressing TaHsfC2a. Our analyses showed that heat stress does indeed have a major effect on the root transcriptome. Further, our heat-stressed root transcriptome data were compared with those of the above-ground plant parts reported by Liu et al. [32]. These results showed that although general categories of differentially expressed root genes were like those seen in wheat leaves [32,34], very little correlation, if any, could be found between the actual genes differentially expressed in different tissues (see also below). Organ-specific (roots as compared to leaves) expressions patterns of the TaHsf gene family between roots and leaves of wheat in response to heat stress were also reported in wheat [18]. In another study, heat-responsive TF expression patterns were found different in different wheat tissues, such as flag leaves and filling-grains [22]. Similar results have been reported in root hairs and striped roots in soybeans [35]. Furthermore, He et al. [36] showed that transcriptomic responses to heat stress differ between the different tissues of maize (Zea mays L.). The alterations of photosynthesis-related proteins in above-ground parts of wheat during heat-stress could be another difference between wheat roots and leaves under heat stress [37]. Another potential difference between wheat roots and leaf transcriptomes under stress is, found by Su et al. [38], the transcriptional regulation of zeatin, brassinosteroid, and flavonoid biosynthesis pathways during heat stress response in leaves [38] but not in the root transcriptome (this study).
The GO enrichment analysis of the DEGs showed that responses to environmental stimuli (heat, light, temperature), ROS, transcriptional regulation, protein-folding, and chromatin remodeling were affected significantly by heat treatment in both the wild-type and transgenic roots. Similar categories were reported from the transcriptome analysis of wheat roots under heat stress [26]. A transcriptome and protein analysis of soybean (Glycine max (L.) Merr.) roots under heat also found similar GO categories [35]. The categories related to stress, chromatin, and ROS were also reported in heat-treated wheat seedling leaves [32]. Recently, TaHsfA6b-4D was shown to contribute to unfolded protein responses under heat stress to maintain protein homoeostasis in the cells [39]. In tomatoes, HSFA1a plays a key role in chromatin spatial reorganization and enhances the expression of heat-stress responsive genes under heat stress [40]. Together with these studies and our RNA-seq results, it can be suggested that TaHSFC2a could be involved in protein homeostasis and chromatin reorganization. In addition, our root transcriptome study identified Fe-and nicotinamide-related GO categories, and similar categories were identified by Luo et al. [26] in wheat roots under heat.
Transcription factors can play a significant role in regulating cellular responses to abiotic stress. The TF families most affected by the heat treatment were HSFs, ERFs, NACs, bHLHs, and MYBs, WRKYs in both wild-type and transgenic roots, and these TFs were also reported in wheat leaves after heat stress [32]. TFs, such as HSFs, ERFs, and WRKYs, were also found in heat-treated soybean roots and maize seedlings [9,35]. When we compared our transcriptome data with the heat-treated leaf transcriptome of wheat [32], we found that G2-like, NF-X1, GRF, RAV, and LBD TF families were differentially regulated only in the roots. G2-like TFs were also differentially expressed in Arabidopsis roots and Brassica rapa whole-plant tissues under abiotic stress conditions [41,42]. NF-X1 was differentially expressed in Arabidopsis whole-plant tissues under heat stress [43], which is consistent with our transcriptome result on NF-X1. The grf7 Arabidopsis mutants were more tolerant to salinity and drought stress, and GRF was found to regulate the expression of DREB2A [44]. DREB2A promotes plant survival under severe environmental stress, including high temperatures [44]. At least six GRFs were down-regulated in the transgenic line; however, only four were down-regulated in wild-type roots. RAV1 and LBD genes were shown to respond to environmental stimuli in Arabidopsis [45,46]. These TFs were expressed in roots (this study) but not in leaves [32], suggesting that these TFs may be root specific, and may play a role in the heat stress tolerance of wheat.
Several heat shock proteins, including HSP70 and HSP90, were highly up-regulated after heat treatment in both the wild-type and transgenic roots, indicating a strong transcriptional response of root tissues (Supplementary Table S2). In wheat, the RNA-seq of seedling leaves after heat treatment increased the expression of several Hsf (class A, B and C), and HSP genes [47]. The Hsf TFs play a crucial role under heat stress by regulating HSPs and other stress-related genes. Previous studies have reported that the overexpression of HsfA induced the expression of a large array of stress-related genes [48][49][50]. The overexpression of TaHsfA2-1 and TaHsfA2e-5D in Arabidopsis was found to up-regulate HSP genes and confers stress tolerance [51,52]. HsfC2 members are monocot specific [19], and their effects on the whole transcriptome have not been studied before. Our results showed that TaHsfC2a overexpression affects a large array of genes under heat stress in roots when compared with wild-type plants. Although several studies observed that the overexpression of Hsfs increased the survival of plants and conferred stress tolerance [27,53], the underlying molecular mechanisms are unknown. For example, the overexpression of TaHsfA2-1, TaHsfA2e-5D, and TaHsfA2-7-AS in Arabidopsis enhanced heat tolerance [51,52,54]. However, the molecular mechanisms are yet to be studied.
Here, through transcriptional profiling, we aimed to decipher the mechanism behind the survival of TaHsfC2a-overexpressed transgenic plants after heat stress as reported by Hu et al. [27]. We found that a group of peroxidases were down-regulated in transgenic lines when compared with wild-type plants after heat treatment. The down-regulation of hydrogen peroxide-producing peroxidases would decrease the hydrogen peroxide accumulation in the transgenic roots. This may be one of the reasons for the increased survival of TaHsfC2a-overexpressing plants as increased ROS levels could trigger cell death in roots. Previous studies showed that the peroxidases gene expression possibly affected the level of hydrogen peroxide in Arabidopsis roots [55,56]. This view is consistent with the observation of the reduced accumulation of hydrogen peroxide in the transgenic roots in this study, as class III peroxidases are involved in generating hydrogen peroxide [57,58]. Furthermore, a recent study in hair grass (Agrostis scabra Willd.) showed that the heat-tolerant grass accumulated lower ROS levels in the roots than in the heat-susceptible creeping bent grass (A. stolonifera L.) under heat stress [59].
In addition to peroxidases, nicotianamine-related transcripts were less abundantly expressed in the roots of TaHsfC2a-overexpressing plants in response to heat when compared with wild-type heat responses. Nicotianamine is a precursor of phytosiderophores, which are important for the iron uptake by the Poaceae family plants, including wheat [31]. Using a chelation-based strategy to form soluble Fe (III) complexes, which are then taken up by the roots [60]. Recently, Distefano et al. [11] identified ferroptosis-like cell death in Arabidopsis roots in response to heat stress that is dependent on ROS and iron accumulation. Therefore, our results suggest that reduced ROS and iron levels may be critical factors contributing to the survival of TaHsfC2a-overexpressing plants after heat treatment through the reduction of ferroptotic cell death. Ferroptosis-like cell death has been identified in diverse species with distinct features of the disruption of iron and ROS-homeostasis [61]. Ferroptosis is an ancient evolutionary mechanism, which is conserved across animal and plant species [11]. In animal cell culture studies, the knockdown of HSF1 was found to enhance ferroptosis through the accumulation of intracellular iron and the associated lipid peroxidation in cancer cells and xenograft models [62], which indicates that TaHsfC2a plays a key role in ferroptosis in plants as both genes belong to the Hsf gene family.

Plant Materials and Growth Conditions
The spring wheat (Triticum aestivum L.) cultivar (cv.) Fielder (wild-type/untransformed control) and TaHsfC2a-B overexpressing transgenic line (C2a-17) in cv. Fielder background were described in Hu et al. [27]. Seeds were washed, kept on wet paper towel at 4 • C for 5 days, and then transferred to 12 • C to induce germination. Two days after germination, the seedlings retained within the moistened paper towel were transferred to room temperature for 1 day, and then transferred to plastic containers containing sterilized reverse osmosis water and allowed to grow for one further day at room temperature. Then, the 4-day-old seedlings were placed into a Hoagland and Arnon nutrient solution No.2 [63] containing plastic container and transferred to a controlled-environment facility (CEF; CSIRO, St. Lucia, Queensland, Australia), where the environmental factors, such as photoperiod, temperature, light, and relative humidity can be controlled. The seedlings were kept under a 16/8-h (day/night) thermoperiod of 22/16 • C with a matching photoperiod (500 µmol m −2 s −1 ) and 60 to 80% relative humidity for a further 3 days in the CEF. Seven-day-old seedlings in the container with Hoagland and Arnon nutrient solution No.2 were kept in a water bath at 43 • C. Once the nutrient solution reached 43 • C, the seedlings were heat-treated for 2 h at 43 • C, while untreated seedlings were kept in the CEF as control samples. The heat-treated and untreated seedlings were then kept in the CEF for one further day before being sequenced.

RNA Preparation and Sequencing
Eight-day-old seedling roots from the wild-type and C2a-17 lines were collected from heat-treated and untreated plants. Root tissues were immediately frozen in liquid nitrogen and then stored at −80 • C for further use. Roots from at least six seedlings, from each container with nutrient solution, were pooled to make up one replicate. Total RNA was extracted from the pooled root tissues, and per replicate using RNeasy Plant Mini Kit (Qiagen, Melbourne, Victoria, Australia) according to the manufacturer's instructions. Four biological replicates (biologically distinct samples from same treatment) per treatment were used for each of the treated and control samples of wild-type and overexpressing plants in this study. Nucleic acid quantity was measured with a NanoDrop ND-1000 UV-Vis Spectrophotometer (Nano Drop Technologies, Wilmington, DE, USA), and quality was determined using Agilent 2100 Bioanalyzer (Agilent Technologies, CA, USA).
A TruSeq stranded messenger RNA (mRNA) kit was used to generate 100 base pair (bp) paired-end libraries, according to the manufacturer's protocol (Illumina Inc., San Diego, USA). Libraries were barcoded prior to sequencing. Sequencing was performed using an Illumina HiSeq 2500 platform with four lanes, and all samples were run on each lane which made up four technical replicates (same sample run on four lanes) per sample. Library preparation and sequencing were performed at the Australian Genome Research Facility (AGRF; Melbourne, Victoria, Australia). Sequence files were deposited to the National Centre for Biotechnology Information (NCBI) Sequence Read Archive under bioproject PRJNA498129.

RNAseq Analysis
We used the wheat reference genomic sequence of cv Chinese Spring, RefSeq v1.0 [64], and the Tuxedo package, previously described [65], to detect genes that were differentially expressed between the control and heat-treated seedlings. Firstly, we excluded base-call errors arising during the sequencing process using the software package SolexaQA. RNAseq reads were trimmed so that all remaining bases had a PHRED score > 30 and their final read length was at least 70 bp with both right and left reads fulfilling these requirements. Filtered and trimmed paired reads were aligned to the Chinese Spring cv. RefSeq v1.0 annotation (files accessed from URGI on 13 March 2017) using Tophat2 v2.1.1 with Bowtie2 v2.2.9 as the aligner. Binary alignment maps (BAM files) were produced using the program SAMtools v1.3.1. Transcript fragments were assembled and normalized to yield fragments per kb per million reads using the package Cufflinks v2.2.2, prior to replicate concatenation by Cuffmerge, and a differential gene expression calling by Cuffdiff. A false discovery rate (FDR) and multiple comparison correction (Bonferroni correction) were performed within the Cuffdiff process, identifying which genes were significantly differentially expressed between the control and heat-treated seedlings for each genotype, the wild-type Fielder, and the TaHsfC2a-overexpressing line. Differentially expressed genes (DEG) were then filtered from each dataset using FDR adjusted p value < 0.05 and > log 2 twofold change cut-off values and divided into up-and down-regulated sets based on the positive or negative fold change values, respectively. Venn diagrams were produced using these filtered datasets with the aid of a webtool produced by the Bioinformatics and Evolutionary Genomics group based at the University of Ghent (http://bioinformatics.psb.ugent.be/webtools/Venn/) (accessed on 1 December 2019).

Gene Ontology Enrichment Analysis
The set of high confidence coding sequences were identified from our RNA-seq. Then, these sequences were annotated using the BLAST2GO [66] server based at CSIRO Agriculture and Food. This produced set of ca.138,000 annotated genes and was used as a background reference to define the wheat GO categories for the enrichment analysis. The up-and down-regulated genes (≥2 log 2 -fold and ≤−2 log2-fold) were used as separate test inputs. Standard parameters were applied for the BLASTx, mapping, and annotation steps and for functional enrichment testing using the Fisher's exact test module (FDR adjusted p value < 0.05). Highly significant GO terms (FDR; p value < 0.001) are presented in Supplementary Table S3. The heatmaps incorporated in this study were generated using the ggplot2 package (tidyverse 1.2.1) in R. We have used the BiNGO plugin for Cytoscape to identify significantly overrepresented GO categories from the wild-type, and used custom wheat GO term annotation as a reference (http://www.psb.ugent.be/ cbd/papers/BiNGO/ [67]). The differentially expressed gene sequences were used in PlantTFDB4.0 to identify putative TFs [68].

Hydrogen Peroxide Measurement Using Hydroxyphenyl Fluorescein (HPF)
At least 15 heat-treated and untreated roots from 8-day-old seedlings (similar treatments were applied as described for RNA sample collection) were incubated for 2 min in 0.1 M phosphate buffer, pH 7.4, containing 5 µM 3 -(p-hydroxyphenyl) fluorescein (Sigma, St. Louis, CA, USA) [69]. Roots were then mounted on a glass slide in a drop of buffer, covered, and observed under a fluorescence microscope (Leica MZ16FA) with exci-tation/emission of 480/510 nm. Fluorescence images of roots were immediately captured from roots, with the microscope setting unaltered for all the treatments. ImageJ was used to measure the staining intensity of the fluorescence images.

Histochemical Detection of Fe 3+
The histochemical study for the detection of Fe was carried out following the previous methods with slight modifications [70,71]. One-week-old seedlings in a container containing Hoagland and Arnon nutrient solution No.2 were exposed to heat stress in a water bath at 43 • C for 2 h, as described for the RNA sample collection, and then incubated in 7% potassium ferrocyanide and 3% HCl (1:1 v/v) for 15 h at room temperature. Roots were washed three times with running distilled water and samples collected from the root hair zone. The sections were mounted onto slides, stained with Prussian blue, covered with a cover slip, and observed under the microscope. Blue pigmentation in tissues caused by the presence of ferric ferrocyanides, formed by an interaction between Fe ions and ferrocyanides, was observed.

Conclusions
In conclusion, our current study showed that TaHsfC2a is modulating ROS and iron levels in response to heat stress in wheat roots. Consequently, this study found the existence of ferroptosis-like cell death in wheat, and TaHsfC2a could be a key player in this mechanism. Further studies in this area are highly warranted to test the utility of TaHsfC2a overexpression in developing thermotolerant varieties in wheat.