bHLH35 mediates specificity in plant responses to multiple stress conditions Authors: Ranjita Sinha, Sara I. Zandalinas, María Ángeles Peláez-Vico, Devasantosh Mohanty, Abdul Ghani, Mather A. Khan, Sai Preethi Induri, Ahmad Bereimipour, Tara Ghandour, Andrew Ogden, Shao-Shan Carol Huang, Rajeev K. Azad, David Mendoza-Cózatl, Trupti Joshi, Felix B. Fritschi, Ron Mittler Published: 2025 ABSTRACT Climate change and enhanced pollution levels are subjecting plants and crops to an increased number of different stressors, simultaneously or sequentially, generating conditions of multifactorial stress combination (MFSC). Although MFSC was shown to severely diminish plant growth, yield, and survival, how plants acclimate to increased levels of stress complexity is largely unknown. Here, we reveal that the Arabidopsis thaliana transcriptional regulator basic helix-loop-helix 35 (bHLH35) is required for plant acclimation to a specific set of MFSC conditions that includes a combination of salinity, excess light, and heat, occurring simultaneously (but not to each of these stresses applied individually or in any other combination). Under the three-stress combination, bHLH35 interacts with no apical meristem/transcription activator factor/cup-shaped cotyledon 69 (NAC069), binds the promoter of lateral organ boundaries domain 31 (LBD31), and regulates the expression of transcripts involved in flavonoid metabolism and ethylene signaling. Our findings uncover a high degree of specificity in plant responses to stress combination, suggesting that different conditions of MFSC could require the function of specific genetic programs for acclimation. INTRODUCTION Climate change and increased levels of industrial pollutants are altering our environment, negatively affecting the growth and yield of crops, and threatening food security (1–13). Especially notable are the elevated day and night temperatures and the increase in the frequency, duration, and intensity of weather events, such as heat waves, floods, and droughts. In many instances, some of these weather events occur simultaneously or sequentially, subjecting plants to conditions of “stress combination” (1–4, 14), for example, when heat waves occur during periods of drought, when drought, salinity, and heat stresses (HSs) combine, or during cycles of drought and waterlogging stresses (1–3, 5, 14, 15). While previous studies determined the responses of plants to different abiotic stresses, such as heat, salinity, or drought, applied individually (the “reductionist” approach), much less is known about plant responses to different combinations of these conditions (a “holistic” approach that takes into consideration the high complexity of environmental stress conditions occurring in nature). This lack of knowledge is likely to severely hinder attempts to develop agronomically important crops with heightened resilience to climate change, essential for our survival in the coming years (2, 5, 9, 12). Recent studies focusing on the effects of three or more abiotic/biotic stresses affecting a plant simultaneously or sequentially, i.e., multifactorial stress combination (MFSC), revealed an alarming and new phenomenon, termed the “MFSC principle” (4, 16–19). The MFSC principle states that with the increase in the number of stresses simultaneously affecting plants (i.e., increase in stress combination complexity), the growth, yield, and survival of plants will markedly decline, although the impact of each individual stress involved in the MFSC is minimal or insignificant, when applied separately. In several other studies, a similar effect of multiple simultaneously occurring stresses, caused by anthropogenic activities, termed global change factor combination (GCFc), was found on soil microbiomes, plant diversity, and overall ecosystem/agroecosystem health (20–22). Although elevated levels of CO2 may compensate for some of these effects (23), the increase in the number, duration, intensity, and overall complexity of environmental stressors on our planet, under elevated temperatures, is likely to pose a substantial danger to our society (4, 16–19). To date, we lack, however, mechanistic understanding of plant acclimation to MFSC, as well as the identification of candidate genes and regulators involved in plant resilience to complex environmental conditions/MFSC. This knowledge will allow us to begin strengthening our crops and make them more resilient to MFSC/GCFc. While studying the physiological and/or molecular responses of thale cress (Arabidopsis thaliana), soybean (Glycine max), rice (Oryza sativa), tomato (Solanum lycopersicum), and corn (Zea mays) to conditions of MFSC, we found that each different combination of stresses comprising the MFSC resulted in a unique molecular and/or physiological response and that, with the increased complexity of MFSC conditions, the photosynthetic activity, stomatal aperture, transpiration, growth, and/or overall survival of plants declined (16, 18, 19, 24). However, whether the molecular responses of plants to each different combination of MFSC are driven by a unique genetic program is largely unknown. Here, we report on the identification of a transcription factor (TF; bHLH35; At5g57150) (25), required for plant acclimation to a specific set of MFSC conditions, i.e., a combination of salinity, excess light (EL), and HS, occurring simultaneously. bHLH35 was not required for plant acclimation to salinity, EL, or heat, applied individually or in any other combination. Under the three-stress combination conditions, we found that bHLH35 is required for the expression of transcripts associated with the regulation of flavonoid metabolism and ethylene signaling. Our findings uncover a high degree of specificity in plant responses to stress and reveal that the molecular response of plants to different stress combinations is likely driven by different genetic programs that are specific to each different subset of MFSC conditions. RESULTS Identification of bHLH35 as a TF involved in the MFSC response of Arabidopsis We previously studied the transcriptomics and proteomics responses of A. thaliana, soybean, Chlamydomonas reinhardtii, and rice to different MFSC conditions (16, 18, 19, 26). Our findings revealed that different combinations of different stresses induced transcriptomics or proteomics responses that were unique to them. However, we also identified groups of transcripts and proteins that were altered in their expression in response to most or all of the different stress conditions included in each of these MFSC studies. To determine the role of different plant pathways that regulate responses to compound stress conditions, we focused on transcripts that are significantly altered in their expression (up or down) in response to all MFSC conditions tested in A. thaliana (4, 16). We reasoned that some of these transcripts could represent common responses to MFSC in plants and that these could be used in future studies to augment the resilience of different plants to conditions of MFSC. We identified over 130 transcripts, including 12 that encode TFs, significantly altered in their expression (elevated or suppressed) in response to combinations of two, three, four, five, and six different abiotic stress conditions applied simultaneously (Fig. 1A and fig. S1; fig. S1 shows only elevated transcripts) (16). Of these, we focused on basic helix-loop-helix 35 (bHLH35), as the function of this TF is largely unknown in A. thaliana with only a few studies identifying it (25, 27, 28), suggesting that it has no identified role in responses to different environmental stress conditions applied separately or in combination. In addition, orthologs of bHLH35 can be found in other plants including different crops (data S1) (29–36), suggesting that studying it could have important biotechnological applications. The TF bHLH35 is required for plant survival under conditions of stress combination. (A) A model illustrating the concept of plant MFSC (left), and expression (log2 fold change) of A. thaliana TFs in response to different stresses/MFSCs (right) [arrow indicates bHLH35; data were obtained from (16); n = 3]. (B) Map of the bHLH35 gene with the positions of tDNA insertions in the two independent bhlh35 mutants (top) and survival of wild type (WT) and the two independent bhlh35 mutants (data S2) under control (CT) and stress combination conditions [salt (S), the herbicide paraquat (PQ), cadmium (Cd), excess light (EL), and heat stress (HS); in all possible combinations] (bottom) (n = 3). (C) Survival of WT and the two bhlh35 mutants subjected to control, S, HS, and/or EL at different stress levels (treatments 1 to 4; n = 3). (D) Complementation of bhlh35_2 with the bHLH35 cDNA expressed under its native promoter (n = 3). Survival of WT and the two bhlh35 mutants subjected to a range of individual S, EL, or HS is shown in fig. S4. Two-way analysis of variance (ANOVA) followed by a post hoc Fisher’s least significant difference (LSD) test was used to determine significant difference between means (different letters donate statistically significant difference; P ≤ 0.05). A, acidity; ACT, aspartate kinase, chorismate mutase, and TyrA; C, complementation; H, high; L, low; M, medium. Phenotypic characterization of bhlh35 mutants To functionally characterize the role of bHLH35 in plant acclimation to MFSC, we subjected wild type (WT) and two independent loss-of-function, with potentially altered function, alleles of bHLH35 (bhlh35_1 and bhlh35_2; see Materials and Methods, Fig. 1B, fig. S2A, and data S2) to five different low-level abiotic stresses in all possible combinations [i.e., salt (S; 50 mM NaCl), EL (700 μmol m−2 s−1), HS (33°C), the herbicide paraquat (PQ; 0.05 μM), and/or the heavy-metal cadmium (Cd; 5 μM CdCl2); Fig. 1B, figs. S2B and S3, and data S3; acidity that was used in (16) and Fig. 1A was not added to the four- and five-stress combinations]. Unexpectedly, although the expression of bHLH35 was up-regulated in response to all these stresses/stress combinations (Fig. 1A and fig. S2B) (16), bhlh35 mutants were exclusively impaired in their survival under a combination of S, EL, and HS (i.e., S + EL + HS; bhlh35 mutants were also impaired in their survival when subjected to four- or five-stress combinations that included S + EL + HS, but under these conditions, WT seedlings also exhibited reduced survival; bhlh35 mutants were not, however, impaired in their survival in response to a combination of S + PQ + Cd + EL or S + PQ + Cd + HS that did not include S + EL + HS; Fig. 1B). In contrast to their susceptibility to the combination of S + EL + HS that used low-level stress conditions (Fig. 1B) compared to WT, bhlh35 mutants were not impaired in their survival and root length when exposed to the individual S, EL, or HS conditions applied at higher stress levels (fig. S4). The unique susceptibility of bhlh35 mutants to S + EL + HS (Fig. 1B and fig. S4) also occurred under a range of S, EL, and HS conditions (Fig. 1C and data S3), underscoring the specificity of bHLH35 for acclimation to this stress combination (applied at different doses). Using the highest settings of S + EL + HS, we also demonstrated that bhlh35_2 could be genetically complemented with the bHLH35 cDNA (exons 1 to 5; Fig. 1B, top) expressed under the control of its native promoter (Fig. 1D). In addition, we showed that overexpressing the bHLH35 cDNA (exons 1 to 5; Fig. 1B, top) using the CaMV35S promoter specifically enhanced tolerance of A. thaliana to the S + EL + HS combination (Fig. 2A and fig. S5). The studies described above revealed that although the expression of bHLH35 is up-regulated in response to multiple stress conditions and their combinations (Fig. 1A and fig. S2A) (16), this TF is uniquely required for acclimation to a specific set of three-stress combination conditions (i.e., S + EL + HS; Figs. 1, B to D, and 2A and fig. S4). These results suggest that the response of plants to different combinations of stresses is unique and may require complex genetic programs that require some TFs to function in different pathways during different stress combination conditions. bHLH35 expression level is associated with plant survival under conditions of MFSC. (A) Survival of WT and three bHLH35 gain-of-function lines (bHLH35 cDNA expressed under the control of the CaMV35S promoter) subjected to CT, S, HS, and/or EL conditions in all possible combinations (n = 3). (B) Survival of WT and the two bhlh35 mutants subjected to control or S + EL + HS under simulated field conditions of S + EL + HS (figs. S6 and S16; n = 3). (C) Survival of different A. thaliana ecotypes with high (left) or low (right) expression level of bHLH35 under CT or S + EL + HS conditions (n = 3). (D) Correlation between SNPs in the bHLH35 gene and environmental conditions obtained from the GenoCLIM v2.0 database of geoclimatic variable associations for 879 A. thaliana accessions. Significant associations were found with high irradiance, HS, and ozone. Two-way ANOVA followed by a post hoc Fisher’s LSD test was used to determine significant difference between means (different letters donate statistically significant difference; P ≤ 0.05). OE, overexpressor; v, variant. As the phenotypic characterization of bHLH35 was conducted using seedlings grown on plates (Figs. 1, B to D, and 2A and figs. S3 to S5), we broadened our analysis of bHLH35 and its role in plant acclimation to MFSC. Thus, we studied the susceptibility of the bhlh35 mutants to S + EL + HS in a simulated field environment (Fig. 2B, fig. S6, and data S3). These studies revealed that in the absence of any added stress, the survival of the bhlh35 mutants was similar to that of WT. In contrast, compared to WT, under conditions of S + EL + HS, the bhlh35 mutants displayed reduced survival (Fig. 2B). In support of the simulated field experiments, A. thaliana ecotypes with high expression level of bHLH35 were more resilient to conditions of S + EL + HS compared to ecotypes with low expression level of bHLH35 (Fig. 2C) (37), and at least one single-nucleotide polymorphism (SNP) in the bHLH35 gene is significantly correlated with geoclimatic variables measuring solar radiation (World Bank diffuse horizontal irradiation), atmospheric nitrogen dioxide [ozone monitoring instrument (OMI) nitrogen dioxide in spring and summer], and vegetation indices indicative of temperature and moisture conditions [advanced very-high-resolution radiometer (AVHRR) smoothed NDVI in spring and AVHRR vegetation health index in summer] in a geoclimatic variable associations of 879 A. thaliana accessions (Fig. 2D) (38). These results support a role for the bHLH35 gene under natural growth conditions and highlight its importance for plant survival in nature. Transcriptomics analysis of the bhlh35 mutant subjected to MFSC To dissect the molecular mechanisms underlying bHLH35 function in plants, we focused on the combination of S + EL + HS and conducted a transcriptomics analysis of WT and one of these mutants (bhlh35_2; could be genetically complemented with the bHLH35 cDNA; Fig. 1D; referred herein as bhlh35) subjected to S, EL, and HS, in all possible combinations (Fig. 3 A and B, fig. S7A, and data S3). This analysis revealed marked differences between the transcriptomics responses of WT and bhlh35 to the different stress conditions (Fig. 3B; fig. S7, B and C; and data S4 to S7). Comparing the transcriptomics responses of WT to that of bhlh35 under each of the different stress conditions applied revealed that only 10 different transcripts were common to all bHLH35-dependent responses of WT (indicated in bold in Fig. 3B) to each of these stress conditions (arrow in Fig. 3C and data S7). This finding suggested that bHLH35 could play different roles under different stress conditions and supported the specific survival phenotype observed for the bhlh35 mutants during the S + EL + HS combination (Figs. 1, B to D, and 2A and fig. S4). Transcriptomics analysis of WT and bhlh35 subjected to S, EL, and HS in all possible combinations. (A) A model showing the transcriptomic analysis conducted to determine the role of bHLH35 in the response of A. thaliana to S, EL, and/or HS. (B) Venn diagrams showing the overlap between transcripts significantly altered in their expression in WT and bhlh35_2 in response to S, EL, and HS in all possible combinations (n = 3). (C) Upset plot showing the overlap between all bHLH35-dependent transcripts expressed in WT under the different stress conditions [from (B), bold]. Arrow indicates transcripts common to all bHLH35-dependent transcripts expressed in WT under the different stress conditions (n = 3). Negative binomial Wald test followed by Benjamini-Hochberg correction (P < 0.05) was used for statistical significance. Considering bHLH35 specificity (Figs. 1 to 3, figs. S3 to S7, and data S4 to S7), we focused on the combination of S + EL + HS and defined two classes of transcripts: transcripts that were altered in their expression in WT specifically under the S + EL + HS combination but did not respond at all in bhlh35 (class 1; Fig. 4A, top, and data S8) and transcripts that were significantly altered in their expression in bhlh35 in response to the three-stress combination but did not respond at all in WT (class 2; Fig. 4A, top, and data S9). While class 1 transcripts were enriched in ethylene- and biotic stress–response transcripts [Gene Ontology (GO) analysis; Fig. 4A, bottom], class 2 transcripts were enriched in development- and general stress–response transcripts (GO analysis; fig. S8). Following up on the GO analysis of class 1 (Fig. 4A, bottom), we studied the response of mutants impaired in ethylene signaling (data S2) to MFSC. This analysis revealed that ethylene signaling mutants [ein2-1 (CS3071), an ethylene insensitive mutant, and etr1-1 (CS237), a mutant in which the ethylene receptor 1 (ETR1) receptor escapes from inactivation leading to constitutive receptor signaling] (39) had altered survival under conditions of EL + HS and S + EL + HS (ein2-1) or just S + EL + HS (etr1-1) (Fig. 4B) and therefore did not have a specific survival response to S + EL + HS. We next identified transcripts that had a bHLH binding domain (G- and E-boxes) in their promoters from each of these two groups (Fig. 4A, top) and conducted a cluster analysis of these transcripts to identify those that had a unique expression pattern specific to the combination of S + EL + HS in each group (Fig. 4C and fig. S9). The rational for focusing on these transcripts was that they were significantly altered in their expression only in response to the three-stress condition in WT but did not respond to any of the stress conditions in the bhlh35 mutants, paralleling the unique survival phenotype of the bhlh35 mutants that was highly specific for the S + EL + HS combination (Figs. 1, B to D, and 2A and fig. S4). This pattern of expression (Fig. 4C and fig. S9) might therefore be associated with the unique function of bHLH35 during MFSC (Figs. 1 to 3). bHLH35 is required for the expression of different stress- and ethylene-response transcripts under a combination of S, EL, and HS. (A) The proportion of transcripts encoded by genes with a bHLH DNA binding domain (DNA-BD) in their promoter and the proportion of the latter with a unique expression pattern to S + EL + HS (data S8 and S9), among the bHLH35-dependent transcripts expressed in WT or bhlh35_2 under a combination of S + EL + HS (top). GO analysis of bHLH35-dependent transcripts (1268 from class 1) identifying stress- and ethylene-response transcripts (bottom) (GO analysis of the 2606 transcripts from class 2 is shown in fig. S8; n = 3). (B) Survival of ethylene biosynthesis and sensing mutants (data S2) under conditions of S, EL, and HS, in all possible combinations (n = 3). (C) Heatmap showing a partial cluster of bHLH35-dependent transcripts that are specifically expressed in WT under conditions of S + EL + HS (complete list is shown in fig. S9A; log2 fold change; n = 3). Two-way ANOVA followed by a post hoc Fisher’s LSD test was used to determine significant difference between means (different letters donate statistically significant difference; P ≤ 0.05), and negative binomial Wald test followed by Benjamini-Hochberg correction (P < 0.05) was used for statistical significance. DUF, domain of unknown function; LEA, late embryogenesis abundant; SAUR, small auxin upregulated RNA; TPR, tetratricopeptide repeat. Phenotypic analysis of mutants deficient in the expression of bHLH35-dependent transcripts To follow up on our transcriptomics study (Figs. 3 and 4), we conducted MFSC survival analysis of mutants (two independent alleles) representing randomly selected transcripts included in class 1 and class 2 (Fig. 5A and fig. S10; a total of 16 genes were tested with two independent mutants each; data S2). This analysis identified ETHYLENE RESPONSE FACTOR 111 (ERF111) (40) and REPRESSOR ACTIVATOR PROTEIN 2.3 (RAP2.3) (41), involved in ethylene signaling, bHLH42 (TT8; involved in flavonoid biosynthesis) (42), ATYPICAL CHLOROPLASTIC THIOREDOXIN 4 (ACHT4; a redox-response His-rich thioredoxin) (43), and LATERAL ORGAN BOUNDARIES DOMAIN 31 (LBD31; a TF involved in development and secondary metabolism) (44), as essential for plant survival specifically under the combination of S + EL + HS (similar to the bhlh35 mutant; Figs. 1, B to D, and 5B and fig. S10). Of these, only LBD31 was specifically expressed under the three-stress combination condition in WT but not in bhlh35 (Figs. 4C and 5A). Our follow-up MFSC survival mutant analysis (Fig. 5B and fig. S10) therefore identified additional genes that might be specifically involved in plant acclimation to an MFSC of S + EL + HS. These could be controlled by, or associated with, bHLH35, as their expression is altered in the bhlh35 mutant during conditions of S + EL + HS. The expression of different bHLH35-dependent transcripts and bHLH35 interactors is specifically required for survival under a combination of S, EL, and HS. (A) Expression of bHLH35 and transcripts that depend on bHLH35 for expression under a combination of S, EL, and HS (log2 fold change; n = 3). (B) Two independent mutants for each of the genes encoding transcripts that require bHLH35 for expression under conditions of S, EL, and/or HS [from (A); data S2] are deficient in survival specifically under the three-stress combination conditions (specific survival of bhlh35 under conditions of S + EL + HS is shown in Fig. 1, B and C; survival of other bHLH35-dependent transcripts is shown in fig. S10B; n = 3). (C) Selected yeast two-hybrid (Y2H) results for TFs that interact with bHLH35 (full results are shown in fig. S11A; n = 3). (D) Survival of two independent mutants for each of the TFs shown in (C) to interact with bHLH35 under conditions of S, EL, and HS in all possible combinations (data S2 and S3; n = 3). Two-way ANOVA followed by a post hoc Fisher’s LSD test was used to determine significant difference between means (different letters donate statistically significant difference; P ≤ 0.05). A, adenine; −HUT, YNB medium deficient in histidine-uracil-tryptophan. Protein-protein and protein-DNA binding activities of bHLH35 bHLH TFs typically interact with other TFs to bind their target DNA and alter transcriptional activity (45). To identify potential interactors of bHLH35, we therefore conducted a yeast two-hybrid (Y2H) analysis of bHLH35 against an A. thaliana whole-genome TF library (46). This screen revealed that bHLH35 interacted with several other TFs (Fig. 5C and fig. S11A), including myeloblastosis viral oncogene homolog 12 (MYB12), no apical meristem/transcription activator factor/cup-shaped cotyledon 69 (NAC069), and myelocytomatosis transcription factor 2 (MYC2). Functional analysis of mutants for some of these interactors revealed that, although none of them had a unique expression pattern such as bHLH35 (fig. S11B), mutants (two independent alleles) of MYB12 (47) and NAC069 (48) displayed a similar survival phenotype to that of bhlh35, whereas mutants (two independent alleles) of MYC2 (49) were specifically required for survival under the combinations of EL + HS and S + EL + HS (Figs. 1, B to D, and 5D and data S2). As LBD31 was the only TF with an expression pattern (Fig. 4C), as well as a survival phenotype (Fig. 5B), unique to the three-stress combination condition, we focused on this TF and conducted a targeted yeast one-hybrid (Y1H) and electrophoretic mobility shift assay (EMSA) analyses to determine whether bHLH35 binds to its promoter. While in the Y1H assay, bHLH35 showed binding to the LBD31 promoter (Fig. 6A), in the EMSA study, it needed to interact with NAC069 before it could bind the LBD31 promoter (Fig. 6B). An interaction between bHLH35 and NAC069 was also found in an in planta split luciferase assay (Fig. 6C). The studies described above provided initial insights into the molecular function of bHLH35 during MFSC and identified some of the different pathways that bHLH35 could potentially be regulating during the MFSC condition of S + EL + HS. bHLH35 interacts with NAC069 and binds the promoter of LBD31 and a weighted gene coexpression network analysis of bHLH35. (A) Y1H analysis showing that bHLH35 binds the promoter (LBD31 promoter 1 fragment) of LBD31 (EV, empty vector; data S13). Position of fragments 1 and 2 on the LBD31 promoter is shown below (n = 3). (B) Gel mobility shift assay showing that the in vitro binding of bHLH35 to the promoter 1 fragment of LBD31 requires NAC069 (n = 3). (C) Split luciferase assay showing in planta interaction between bHLH35 and NAC069 (n = 3). (D) The different modules identified through the weighted gene coexpression network analysis (WGCNA) of bHLH35. A submodule (the blue module; fig. S12 and data S10) in the WGCNA of bHLH35 that regulates responses to heat and other stresses is shown on the right. Dashed arrows, coexpression and promoter binding; solid arrows, coexpression and significant motif enrichment. Two-way ANOVA followed by a post hoc Fisher’s LSD test was used to determine significant difference between means (different letters donate statistically significant difference; P ≤ 0.05). Two-tailed Student’s t test (*P ≤ 0.05). EVprotein, empty pDEST vector; EVpromoter, empty pHIS vector with minimal His promoter. Gene regulatory network analysis of bHLH35 function during MFSC To further dissect the molecular function of bHLH35 during MFSC, we conducted gene regulatory network (GRN) analyses (50, 51) of its function (Figs. 6D and 7A, figs. S12 and S13, and data S10 and S11). These studies identified flavonoids, abscisic acid (ABA), and HEAT SHOCK TRANSCRIPTION FACTORs (HSFs) and other stress response transcripts, as associated with bHLH35 function (Figs. 6D and 7A, figs. S12 and S13, and data S10 and S11). Our weighted gene coexpression network analysis (WGCNA; Fig. 6D and fig. S12) identified 21 coexpressed modules (Fig. 6D and data S10) and revealed that bHLH35 and LBD31 were coexpressed with DEHYDRATION-RESPONSIVE ELEMENT BINDING 19 (DREB19), HSF2B, HSFB2A, and HSFB4 (blue module; Fig. 6D), associated with heat and oxygen stress responses. In contrast, target genes in the tan module were significantly enriched for terms related to flavonoid and other metabolic processes, while those in the magenta module were significantly enriched in terms related to stress and environmental responses (e.g., water, cold, oxygen/hypoxia, and ABA stimulus) and included MYC2 (fig. S12). These findings suggested that bHLH35 could be coregulating genes associated with different biological pathways (as identified by the different modules, as well as the mutant analyses described above; Figs. 4 to 6). Our differential inference analysis of expression (DIANE; Fig. 7A, fig. S13, and data S11) study agreed with our WGCNA analysis (Fig. 6D and fig. S12), further highlighting flavonoid metabolism, ethylene signaling, and general stress responses as linked to bHLH35 function. Dissecting the promoter of bHLH35, we identified multiple binding elements for different TFs that could function upstream of bHLH35 (data S12). Further examination of the expression pattern of these TFs in our RNA sequencing (RNA-seq) dataset (fig. S14) revealed that some TFs involved in ethylene and ABA signaling might be involved in regulating the expression of bHLH35 during different stresses and their combination. These included ERF111 that is required for survival under the S + EL + HS combination (Fig. 5B) and might play a feedback regulatory loop role in bHLH35 expression. Further studies are needed to determine the temporal dynamics of bHLH35 transcript and protein expression during and following different stresses (i.e., onset, maintenance, and recovery of individual and combined stresses). At least in our hands, transcripts encoding bHLH35 accumulated within the first 30 to 180 min of S + EL + HS application (fig. S2B). bHLH35 drives both ethylene- and flavonoid-associated pathways during a combination of S, EL, and HS. (A) GRN (DIANE) analysis of bHLH35 reveals links to stress-, flavonoid-, and ethylene-associated transcripts (data S11). (B) The flavonoid naringenin (Nar) rescues the survival phenotype of lbd31 (top), but not rap2.3 (bottom) mutants, under a combination of S, EL, and HS (i.e., S + EL + HS; n = 3). (C) Accumulation of reactive oxygen species (ROS) in WT, bhlh35, and lbd31 seedlings grown under control conditions or subjected to a combination of S + EL + HS in the presence or absence of naringenin (n = 3). (D) A hypothetical model for the function of bHLH35 during MFSC. bHLH35 is shown to be required for the expression of LBD31, bHLH42 (TT8), and MYB12, associated with flavonoid metabolism, and specifically required for the survival of A. thaliana seedlings under conditions of S + EL + HS combination. bHLH35 is also shown to be required for the expression of RAP2.3, ERF111, and ACHT4, associated with ethylene and stress metabolism, and required for survival under conditions of S + EL + HS combinations. bHLH35-regulated ethylene signaling and flavonoid responses are proposed to suppress ROS accumulation during MFSC and contribute to plant acclimation. Two-way ANOVA followed by a post hoc Fisher’s LSD test was used to determine significant difference between means (different letters donate statistically significant difference; P ≤ 0.05). Complementing flavonoid metabolism with naringenin rescues lbd31, but not rap2.3, under conditions of MFSC As transcripts involved in flavonoid metabolism were identified by our GRN analyses (Fig. 7A and figs. S12 and S13), bHLH42 (42) was essential for plant survival under the combination of S + EL + HS (Fig. 5B), and homologs of bHLH35 and LBD proteins were previously associated with flavonoid metabolism in different plants (34–36, 52), we tested whether augmenting flavonoid metabolism by treatment of plants with naringenin (53, 54) could rescue the lbd31 mutants. Naringenin application could rescue the lbd31 phenotype specifically under the three-stress combination (S + EL + HS; Fig. 7B, top). In contrast, naringenin could not rescue the rap2.3 mutants involved in ethylene signaling under the three-stress combination (S + EL + HS; Fig. 7B, bottom). Under conditions of S + EL + HS, bHLH35 could therefore regulate a unique set of transcripts that are specific to the three-stress combination condition and associated with flavonoid biosynthesis and ethylene signaling pathways, required for plant survival. A role for flavonoids in suppressing reactive oxygen species (ROS) accumulation was previously demonstrated (53, 54), and elevated levels of ROS were measured during MFSC in Arabidopsis (16). As ROS accumulation can cause cell death (55), enhanced ROS levels were associated with MFSC (16), and the bhlh35 mutants were rescued from dying by augmenting flavonoid metabolism with naringenin (Fig. 7B), we measure ROS levels in WT, bhlh35, and lbd31 mutants under control conditions and in response to a combination of S + EL + HS (in the presence or absence of naringenin; Fig. 7C). Compared to WT, both bhlh35 and lbd31 mutants accumulated higher ROS levels under control conditions or in response to the S + EL + HS stress combination, suggesting that they are deficient in the regulation of ROS levels (Fig. 7C). Treatment with naringenin reduced ROS levels in the bhlh35 and lbd31 mutants to levels that were similar to those of WT, under both control and the S + EL + HS stress combination (Fig. 7C). These findings suggest that, under the combination of S + EL + HS, flavonoids could suppress ROS levels in WT plants and that the bhlh35 and lbd31 mutants, which were impaired in the expression of flavonoid biosynthesis transcripts (Figs. 4 to 6 and Fig. 7A), could not suppress ROS accumulation. In addition, they provide further support to our hypothesis that bHLH35 and LBD31 function in the same pathway (Fig. 7D). DISCUSSION The increased complexity of environmental stress conditions on our planet, driven by anthropogenic activities, subjects plants and crops to conditions of MFSC (4, 17, 22, 56). These can cause massive reductions in yield, negatively affect food security, and destabilize societies (11–13). Although the responses of plants to conditions of MFSC were previously studied (16, 18, 19, 24), how these responses are regulated is poorly understood. Here, we reveal that specific genetic programs (e.g., those controlling flavonoid metabolism and ethylene signaling) that are controlled by different TFs (i.e., LBD31, Transparent Tesla 8, TT8, and MYB12 for flavonoid metabolism and RAP2.3 and ERF111 for ethylene signaling) are regulated by the TF bHLH35 during MFSC (Fig. 7D). We further reveal that at least one of these genetic programs (i.e., flavonoid metabolism) is required for plant survival under a specific set of environmental stress combination conditions, i.e., a combination of S + EL + HS, but not under conditions S, EL, and/or HS in any other combination (Figs. 1, B to D, and 5B and fig. S4). Our study therefore unravels a high degree of complexity in plant responses to stress combination and suggests that responses to MFSC are controlled by different combinations of different genetic programs that are specific to each of the different stress combination conditions included in an MFSC. Our findings also highlight an interesting, and mostly unexplored, aspect of plant gene expression regulation during stress, revealing that some TFs, up-regulated in response to multiple different stress conditions and their combinations (considered general stress response genes/regulators), could have different roles in each of the different stresses/combinations they respond to. Thus, bHLH35, which was up-regulated in response to multiple stress conditions and their combinations (Fig. 1A) (16), was associated with or required for the expression of different sets of transcripts in each of these stress conditions/combinations (Fig. 3). Moreover, although bHLH35 was up-regulated in response to all the different stress conditions/combinations studied here (Fig. 1A), it was required for plant acclimation only to a specific set of these stress combination conditions (Fig. 1, B to D, and fig. S4). These findings could suggest that other so-called general stress response TFs (e.g., HSFs) (57, 58) might have a similar behavior during different stresses and their combinations. Further studies are required to untangle the landscape of plant responses to different stresses and their combinations/MFSCs and identify the roles of different regulators during single and multiple stress conditions. Additional studies are also required to define the process of plant acclimation to stress, as this process may require the integration and coordination of multiple pathways under different stress conditions. The identification of regulators of flavonoid metabolism and ethylene signaling, as required for plant acclimation to a combination of S + EL + HS (Figs. 5 and 7, B and C), is intriguing. Different types of abiotic stress conditions were shown to result in the enhanced accumulation of flavonoids, and in some instances, this enhancement was proposed to be driven by ethylene signaling (59–61). In our hands, however, boosting flavonoid metabolism by the addition of naringenin rescued the flavonoid-associated lbd31 mutants but failed to rescue the ethylene-associated rap2.3 mutants (Fig. 7B), suggesting that these two pathways might not be linked under the specific three-stress combination condition studied here. Flavonoids and/or ethylene signaling were previously found to regulate ROS levels in plants, with some flavonoids proposed to have an antioxidative function (53, 54, 62). Our previous studies of MFSC in A. thaliana revealed that with the increased complexity of MFSC, ROS levels in plants increased (16). It is therefore possible that a disruption in flavonoid metabolism in the bhlh35, lbd31, and tt8 mutants during the S + EL + HS stress combination led to overaccumulation of ROS and that this overaccumulation had an adverse impact on the ability of these mutants to acclimate (Figs. 5 and 7, B to D). In support of this possibility, compared to WT, the expression of several ROS-scavenging enzymes (e.g., superoxide dismutases) was suppressed, while at the same time, the expression of several ROS-producing enzymes (e.g., respiratory burst oxidases) was enhanced in the bhlh35 mutant during the combination of S + EL + HS (fig. S15). In addition, compared to WT, bhlh35 and lbd31 mutants accumulated more ROS when subjected to the S + EL + HS combination, and application of naringenin suppressed this accumulation (Fig. 7C). Further studies are needed to untangle the relationships between flavonoid metabolism, ethylene signaling, and ROS levels during MFSC in general, as well as more specifically during the combination of S + EL + HS. Together, our findings suggest that different combinations of different environmental stresses require different genetic programs for survival. The survival of A. thaliana seedlings under conditions of S + EL + HS is shown here to require a specific set of TFs that controls a specific set of transcripts that are not required for survival under S, EL, and/or HS in all combinations other than S + EL + HS. We further show that a distinct combination of different pathways is required for survival under each condition of stress combination. Thus, ethylene signaling pathways were required for EL + HS acclimation, while flavonoid metabolism and ethylene signaling pathways were required for S + EL + HS acclimation (Fig. 7D). This specificity suggests that plants, as well as other organisms, could use genetically encoded “stress combination–specific responses” that could only be revealed using a holistic approach of testing all possible stress conditions in different combinations. Alternatively, a “stress dose” effect could explain some of our observations. Thus, three stresses applied together (i.e., S + EL + HS) may have a higher impact on plants compared to two stresses combined (i.e., S + EL, EL + HS, or S + HS) and require additional pathways (e.g., flavonoid accumulation) for acclimation. As humans, microbiomes, animals, and plants are subjected to a multitude of different stress conditions in an environment that continues to increase in its complexity with each year (4, 11, 56, 63), untangling the complex regulatory networks that control gene expression during MFSC is highly important. Studies of MFSC or GCFc recently revealed that with the increased complexity of stress conditions affecting a plant, microbiome, or ecosystem, their functions and services markedly decline (4, 16–22, 24, 26). A significant shift in the way we conduct experiments (toward a more holistic approach) might therefore be needed if we want to untangle the different networks that control plant responses to MFSC/GCFc and develop resilient crops that will allow us to overcome our future environmental challenges on this planet. Because orthologs of bHLH35 can be found in many other plants including different crops (data S1) (29–36), bHLH35 could be used in breeding and other gene editing efforts to enhance crop resilience to MFSC/GCFc. MATERIALS AND METHODS Plant growth, stress treatments, and mutant screening A. thaliana WT Col-0 and two independent loss-of-function, with potentially altered function, alleles of bHLH35 (bhlh35_1 and bhlh35_2; data S2) were used in this study. Although low levels of bHLH35 transcripts containing exons 1 to 4 could be detected in the bhlh35_1 and bhlh35_2 mutants (figs. S2A and S7A) and these could function as dominant-negative variants of the protein (due to the presence of the DNA binding domain (DNA-BD) of the protein but without the full activation domain; fig. S7A), the complementation of the bhlh35_2 mutant phenotype with the bHLH35 cDNA (exons 1 to 5; Fig. 1B, top) expressed under the control of the bHLH35 or CaMV35S promoters (Figs. 1D and 2A) suggest that the low levels of transcripts present in the mutants do not have a dominant-negative function and could be nonfunctional. WT and bhlh35_1 and bhlh35_2 plants were subjected to the following individual treatments and their different combinations on plates containing one-half Murashige and Skoog (one-half MS; Caisson Labs, UT, USA, MSP09) medium (pH 5.8), with 1% phytagel (data S3): CT (one-half MS, 21°C, 50 μmol m−2 s−1, pH 5.8), Cd (one-half MS, 21°C, 50 μmol m−2 s−1, pH 5.8, 5 μM CdCl2), EL (one-half MS, 21°C, pH 5.8, 700 μmol m−2 s−1), HS (one-half MS, 33°C, 50 μmol m−2 s−1, pH 5.8), S (one-half MS, 21°C, 50 μmol m−2 s−1, pH 5.8, 50 mM NaCl), and PQ (one-half MS, 21°C, 50 μmol m−2 s−1, pH 5.8, 0.05 μM PQ) (data S3) (16). All other experiments comparing WT and A. thaliana loss-of-function mutants (data S2) or gene complementation lines for the bhlh35 mutant were performed with S, EL, and HS and all their combinations as described in (16) (see also data S3). For experiments with different levels of S, EL, and/or HS, A. thaliana WT and the two bhlh35 mutants were screened for survival under low (35 mM NaCl, 31°C HS, 500 μmol m−2 s−1), mild (40 mM NaCl, 32°C HS, 600 μmol m−2 s−1), high (50 mM NaCl, 33°C HS, 700 μmol m−2 s−1), and low/high (35 mM NaCl, 34°C HS, 800 μmol m−2 s−1) after 5, 7, and 10 days post–stress application, as described in (16). For A. thaliana ecotype stress experiments, A. thaliana ecotypes (The 1001 Genomes Consortium; https://1001genomes.org) (37, 38) with high or low steady-state bHLH35 transcript expression levels (based on GSE80744) (37) were screened for survival under CT and S + EL + HS (50 mM NaCl, 33°C HS, 700 μmol m−2 s−1), as described in (16). For survival of A. thaliana WT and the two bhlh35 mutants under simulated field conditions, seeds of the different lines were germinated side by side in plastic pots (10 cm by 10 cm by 10 cm) filled with field soil (Mexico silt loam; fine, smectitic, and mesic Vertic Epiaqualf) obtained from the Bradford Research Center (Missouri Agricultural Experiment Station, Columbia, MO, USA, 38°53′N, 92°12′W) (64). Soil for CT pots was watered with plain water, while soil for S + EL + HS was supplemented with 50 mM NaCl once at the beginning of the experiment. Seedlings were grown in a greenhouse (https://plantgrowthfacilities.missouri.edu/eastcampus.htm) under natural light shaded to 150 μmol m−2 s−1 at a temperature of ~22°C for day and ~15°C for night. Five days postgermination, CT non–S-treated plants were kept under the conditions described above (CT conditions), while S-treated pots were subjected to EL and HS (for a combination of S + EL + HS), by removing them from shading, placing them under a heater (Kalglo Infrared Heater HS-2420 I 65″ I 240 V I 2000 W, Fogelsville, PA, USA) (65), and supplementing the natural daylight (500 to 600 μmol m−2 s−1) with light-emitting diode (LED) lights (CRAFTSMAN 9000 Lumen LED twin light, model CMXELAYMPL1029, CRAFTSMAN, Mississauga, ON, Canada; an additional 150 to 200 μmol m−2 s−1 above natural conditions) for a total of 650 to 800 μmol m−2 s−1 (fig. S16). Four to 5 days post–stress application, the survival of CT- and S + EL + HS–treated plants was determined as described in (16). For survival experiment in the presence or absence of the flavonoid naringenin (Indofine Chemical Company, NJ, USA), one-half MS plates with or without S (50 mM) were supplemented with 100 μM naringenin as described in (53), and the experiment was conducted as described above and in (16) (data S3). Each treatment was repeated in at least three technical repeats and conducted in at least three biological replicates. For survival and root growth studies of WT and the two bhlh35 mutants subjected to elevated levels of individual S, EL, or HS conditions, seedlings were subjected to EL (one-half MS; 21°C; pH 5.8; 700, 800, or 900 μmol m−2 s−1), HS (one-half MS; 33°, 37°, or 41°C; pH 5.8; 50 μmol m−2 s−1), or S (one-half MS; 21°C; pH 5.8; 50 μmol m−2 s−1; 50, 75, 100, or 150 mM NaCl). Seedlings were then scored for survival and root length as described in (16). As controls for plants sensitive to S or EL, seeds of the respiratory burst oxidase homolog D and F mutant (rbohDF; data S2) were placed side by side on plates with WT and the two bhlh35 mutants, and as controls for plants sensitive to EL, seeds of the catalase 2 mutant (cat2; data S2) were placed side by side on plates with WT and the two bhlh35 mutants (fig. S4). Genetic variation and geoclimatic variable associations Association between genetic variation of bHLH35 and geoclimatic variables were downloaded from the A. thaliana GenoClim 2.0 database (38). The scores for association strength (negative logarithm of the P value) computed by a mixed model that corrects for population structure (AMM_scores) were plotted in the heatmap with the indicated q values corrected by the Benjamini-Hochberg method as described in (38). RNA-seq analysis About 125 to 150 A. thaliana WT Col-0 and bhlh35 (bhlh35_2) seedlings were grown horizontally for 6 days on separate sets of plates containing one-half MS with and without S (50 mM NaCl) in three biological repeats. After 6 days, seedlings were subjected to the individual and combined stresses of S, EL, and HS as described above, except that EL and/or HS were applied for 1.5 hours before sampling. Whole seedlings for each stress treatments and control were sampled (flash freeze in liquid nitrogen) in three replicates after the 1.5 hours of EL and/or HS (16). Total RNA was isolated using RNAeasy plant mini kit (QIAGEN, MD, USA, 74904). RNA libraries for sequencing were prepared by Novogene Co. Ltd. (https://en.novogene.com/website, Sacramento, CA) using standard Illumina protocols. RNA-seq was performed using NovaSeq 6000 PE150 by Novogene Co. Ltd. Quality control, read trimming, read alignment, transcript assembly, quantification, and differential expression analysis of RNA-seq data was performed as explained in (66). Negative binomial Wald test followed by Benjamini-Hochberg correction (P < 0.05) was used to determine significant transcript expression (66). As a result of low quality, one sample (bhlh35 treated with EL, replicate 3) was removed from the analysis. Functional annotation, quantification of overrepresented GO terms (P < 0.05), and Kyoto Encyclopedia of Genes and Genomes enrichment were conducted in g:Profiler (67). The presence of bHLH G-box or E-box motifs in the promoter [500 base pairs (bp) upstream] of genes was performed in MEME Suit 5.5.4 (https://meme-suite.org/meme/index.html) using the FIMO motif scanning tool (68). UpSet plots were generated using the UpSetR (69) package. Heatmaps were generated using Morpheus (https://software.broadinstitute.org/morpheus/). Venn diagrams were created in VENNY 2.1 (BioinfoGP, CNB-CSIC; https://bioinfogp.cnb.csic.es/tools/venny/index.html). RNA-seq data files are available at Gene Expression Omnibus (GEO; GSE281968). GRN and weighted gene coexpression network analyses GRN analysis was performed using DIANE (51). RNA-seq data were normalized, and low-expression genes were filtered out, retaining genes with robust regulatory signals to infer gene-gene interactions based on differential expression profiles, leveraging both coexpression and causative regulatory evidence to capture direct and indirect regulatory relationships. DIANE parameters were optimized, with a significance threshold set to a P < 0.05, and multiple testing was addressed through the Benjamini-Hochberg correction. The resultant output, an edge list of significant gene interactions, was prepared in Cytoscape 3.10 (70). Additional node and edge attributes, such as gene functional annotations and interaction confidence scores, were added for a comprehensive view (71). A WGCNA network was constructed in R (50) using blockwiseModules function. The input matrix consisted of the 47 replicates and 9260 genes, with expression values transformed using variance-stabilizing transformation from DESeq2 to normalize the count data. The parameters for blockwiseModules were (i) soft threshold power of 18, as established by pickSoftThreshold, (ii) “signed” network and topological overlap matrix types to preserve correlation directionality, (iii) minimum module size of 40 genes, and (iv) module merging threshold of 0.18 (to merge highly similar modules). This resulted in a WGCNA network comprising 21 modules (data S10). For the promoter sequence analysis of the genes in WGCNA network, TAIR10 genome assembly of A. thaliana and their gene annotations from Ensembl Plants release 49 (https://plants.ensembl.org/Arabidopsis_thaliana/Info/Index) were used. Promoter regions were defined as spanning 1 kb upstream to 300 bp downstream of the transcription start site. Validated TF binding profiles were retrieved from JASPAR2024 (72). Connection to JASPAR2024 was made using the RSQLite package (https://rsqlite.r-dbi.org and https://github.com/r-dbi/RSQLite), and position weight matrices (PWMs) corresponding to the A. thaliana species code (3702) were retrieved by TFBSTools (73). Motif enrichment within WGCNA gene modules was computed using the calcBinnedMotifEnrR function from the monaLisa R package (74). To ensure that motif enrichment was evaluated within biologically relevant coexpression networks, we assigned each promoter sequence to its respective WGCNA coexpression module, creating module-specific bins. Enrichment calculations were performed relative to all promoter sequences across all modules by setting the background to “allBins.” This approach allowed us to assess the overrepresentation of TF binding motifs within each module’s promoters while accounting for background motif frequencies, thereby identifying WGCNA modules potentially enriched for specific TF targets. This binned motif enrichment analysis returned the false discovery rate–adjusted P values (−log10 scale) of the enrichment of each TF motif in each bin (WGCNA module) relative to its occurrences in the background sequences. TF regulators for the modules of interest (blue, magenta, and tan) were identified as TFs whose motifs were enriched for the genes in the module at adjusted P value threshold of 0.05. Matches of specific TF motifs in target gene promoters were found using findMotifHits function in monaLisa with the query parameter set to the PWMs of the TFs of interest, the subject parameter defined as the set of promoter sequences corresponding to all genes within the WGCNA module of interest, and motif matching method set to the matchPWM method at a minimum score threshold of 6.0. The visualizations were done with Cytoscape (70). DNA binding elements in the promoter of the bHLH35 gene were identified with (https://planttfdb.gao-lab.org/). Whole-genome TF Y2H assay To identify TF partners of bHLH35, we implemented a high-throughput Y2H screen using A. thaliana bHLH35 as bait against a whole-genome A. thaliana TF library described in (46). The coding sequence (CDS) of bHLH35 was cloned into a modified pGBKT7 (Takara, CA, USA, 630443) vector harboring the GAL4 DNA-BD where the uracil (URA) marker was replaced by a histidine (HIS) selection marker using In-Fusion cloning kit (Takara, CA, USA, 638955). The resulting DNA-BD-bHLH35 plasmid was transformed into the yeast two-gold (Y2G) yeast strain (Takara, CA, USA, 630498) using the lithium acetate method (75). Y2G cells carrying the CDS of bHLH35 were mated with the Yα1867 strain carrying the A. thaliana TF library, and diploid strains were selected on a Yeast Nitrogen Base (YNB) lacking histidine (−HIS), uracil (−URA), tryptophan (−TRP), with adenine (+ADE) medium (Sunrise Science Products, TN, USA, 1500 and 1051). This Y2H screen relies on the Ade marker to detect protein-protein interactions; therefore, after selecting positive mating colonies carrying both the bait and preys, diploid strains were pinned on quadruple dropout medium (YNB-His-Ura-Trp-Ade; Sunrise Science Products, TN, USA, 1500 and 1159) using an HDA ROTOR robot (Singer Instruments, UK), and after 2 days of incubation, white colonies were selected for further confirmation through a secondary screen and Sanger sequencing. Details of the primers used for cloning are listed in data S13. Y1H assay To test whether bHLH35 binds to the promoter region of LBD31 (AT4G00210), we performed a targeted Y1H assay. Briefly, the CDS of bHLH35 was cloned into vector pDEST22 by Gateway LR recombination reaction (Invitrogen, CA, USA, 11791020). Next, promoter regions comprising 500 bp upstream of the transcription start site and 183-bp 5′ untranslated region (5′UTR) were cloned as two fragments called LBD31 promoter 1 (183-bp 5′UTR + 200-bp promoter upstream of the transcription start site) and LBD31 promoter 2 (300-bp promoter; 200- to 500-bp upstream of the transcription start site). The two fragments (data S13) were amplified from A. thaliana genomic DNA using a high-fidelity polymerase (New England Biolabs, MA, USA, MO530S) and cloned into a Gateway-compatible version of the integration plasmid pHISi-1 (https://www.addgene.org/) using the In-Fusion cloning reaction following the manufacturer’s protocols. The pHIS plasmid containing the promoter fragments was linearized with Xho I and integrated into the yeast Y1H-aS2 (his3-Δ1) provided by the Walhout laboratory as described in (76). Positive colonies were selected on minimal medium deficient in His (YNB-His, Sunrise Science Products, TN, USA). Then, the Y1H-aS2 strains carrying the LBD31 promoter fragments were transformed with pDEST22-bHLH35, and positive colonies were selected on YNB-His-Trp medium (Sunrise Science Products, TN, USA). The interaction of bHLH35 with LBD31 promoter fragments was tested on selective medium at varying concentrations of 3-aminotriazole (3AT; Thermo Fisher Scientific, MA, USA, 264571000). Details of the primers used for cloning and the promoter fragment sequences are listed in data S13. Protein isolation and EMSA analysis CDSs of the bHLH35 (AT5G57150) and NAC069 (AT4G01550) were cloned into pET28a+ bacterial gene expression vectors downstream of the 6×HIS tag. The clones and empty vector were transformed along with a vector containing molecular chaperones pG-KJE8 (Takara, CA, USA) into Escherichia coli BL21 strain and transformants were selected on LB plates containing kanamycin (50 μM ml−1; Gold Bio, MO, USA, K-120-5) and chloramphenicol (25 μM ml−1; Gold Bio, MO, USA, C-105-5). Recombinant proteins were isolated using purification protocol in (77). Eluted protein was mixed with glycerol (10% final concentration) and stored at −80°C until further use. Protein quantification was performed using Bradford reagent (Thermo Fisher Scientific, MA, USA). Details of the primers used for cloning are in data S13. EMSA analysis was performed with biotin-labeled LBD31 promoter probes and LBD31 mutant promoter probes (E-box replaced with random sequences; data S13) using the LightShift Chemiluminescent EMSA Kit (Thermo Fisher Scientific, MA, USA, 20148) following the manufacturer’s protocol. For the binding reaction, probes (data S13) were incubated with bHLH35 and/or NAC069 protein (isolated as described above) in 1× binding buffer and 2.5% glycerol. Probe-protein mix was separated on 6% polyacrylamide gel, transferred to nylon membrane (Amersham Hybond-N, GE Healthcare, IL, USA) and cross-linked at 120 mJ cm−2 (UV Crosslinker, FB-UVXL-1000, Thermo Fisher Scientific, NH, USA). Chemiluminescence-based detection of biotin-streptavidin conjugate was performed following the LightShift Chemiluminescent EMSA Kit’s protocol instructions. Details of probes are in data S13. Mutant complementation The bhlh35_2 mutant was transformed with the bhlh35 CDS driven by the native bHLH35 promoter (1000-base fragment). The promoter of the bHLH35 gene (AT5G57150) and CDSs of monomeric green fluorescent protein (mGFP) and bHLH35 (in the following order: bHLH35 promoter-mGFP-bHLH35 gene) were cloned into pCAMBIA1302 binary vector. Cloning was performed using In-Fusion cloning kit (Takara, CA, USA, 638955). After confirming the sequence, the recombinant vector was transferred to Agrobacterium tumefaciens GV3101. A. thaliana plants were transformed by floral dip. T1 seeds were screened on one-half MS medium supplemented with hygromycin (12 μg ml−1) as described in (78). Ten different homozygous transgenic lines were generated (T3) and tested for survival under conditions of stress combination as described above. To create A. thaliana bHLH35 overexpression (bHLH35-OE) lines, bHLH35 CDS was cloned into pCAMBIA1302 (replacing mGFP) under CaMV35S promoter. A. tumefaciens GV3101-mediated transformation of A. thaliana WT (Col-0) plants was performed as described in (78). Details of the primers used for cloning are in data S13. Split luciferase complementation assay Split luciferase complementation assay based on Agrobacterium-mediated transient expression of luciferase in Nicotiana benthamiana was used to investigate protein-protein interactions in vivo (79). Full-length WT CDS of bHLH35 and NAC069 were polymerase chain reaction (PCR) amplified and sequenced (primers used are listed in data S13) from cDNA synthesized from WT seedlings subjected to S + EL + HS, respectively. bHLH35 PCR product was cloned in pCAMBIA1300-CLuc between the Kpn I and Bam HI sites, and NAC069 was cloned in pCAMBIA1300-NLuc between the Kpn I and Sal I sites as described in (79). The final constructs were 35S:NAC069-NLuc and 35S:CLuc-bHLH35. Following confirmation by sequencing, recombinant vectors were electroporated into A. tumefaciens strain GV3101. N. benthamiana plants were grown on soil at 28°C under 14/10-hour light/dark cycle. Four-week-old N. benthamiana leaves were infiltrated with Agrobacterium harboring the plasmids (79). Agrobacterium transformed with 35S:CLuc-GFP was used as control in the appropriate combinations. Three days postinfection, N. benthamiana plants were fumigated with 1 mM d-luciferin potassium S (Abcam, Waltham, MA, USA) prepared in 10 ml of 10 mM phosphate buffer (pH 7.4) containing 0.002% Silwet L-77 (Lehle seeds, Round Rock, TX, USA) (80). Postfumigation, the leaves were imaged using the IVIS Lumina S5 platform (PerkinElmer, Waltham, MA, USA) as described in (80). The relative luminescence units of regions of interest (ROIs) were calculated using the Living Image 4.7.3 software (PerkinElmer). ROS imaging Seedlings of Arabidopsis WT (Col-0) and two independent mutants each of bhlh35 and lbd31 were grown on one-half MS plates with and without S and naringenin as described above. Seven-day-old seedling were fumigated with 10 ml of 50 μM DCFH-DA (2′,7′-dichlorodihydrofluorescein diacetate; excitation/emission, 480/520 nm; Sigma-Aldrich, Saint Louis, MO, USA) solution prepared in 10 mM phosphate buffer (pH 7.4) containing 0.002% Silwet L-77 (Lehle seeds, Round Rock, TX, USA) in a glass chamber for 30 min as described in (81–83). Seedlings subjected to S stress were exposed to EL (700 μmol m−2 s−1) and HS (33°C) for 30 min as described above. Control plates were kept under control conditions for 30 min. Post–stress exposure, plates were imaged using the IVIS Lumina S5 platform using Living Image 4.7.3 software in acquisition mode (PerkinElmer, Waltham, MA, USA) for 10 min as described in (81–83). Images were further processed using the Living Image 4.7.3 software (PerkinElmer) as described in (81–83), and the raw integrated density of ROIs was calculated using ImageJ (81–83). RNA isolation and reverse transcription quantitative PCR Total RNA isolated for RNA-seq analysis was converted to cDNA using PrimeScript RT Master Mix (Takara Bio Inc., CA, USA, RR036A). Real-time reverse transcription quantitative PCR was performed with bHLH35 gene specific primers (data S13) and EF1a as internal reference (data S13) using the CFX Connect Real-Time PCR Detection System (Bio-Rad), as previously described (84). Data analysis Two-way analysis of variance (ANOVA) followed by a post hoc Fisher’s least significant difference (LSD) test (different letters donate statistically significant difference; P ≤ 0.05), two-tailed Student’s t test (*P ≤ 0.05), negative binomial Wald test followed by Benjamini-Hochberg correction (P < 0.05), or a one-way ANOVA followed by a Tukey post hoc test (different letters denote statistical significance at P ≤ 0.05) was used to determine significant differences.