The heat shock factor 20-HSF4-cellulose synthase A2 module regulates heat stress tolerance in maize

Abstract Temperature shapes the geographical distribution and behavior of plants. Understanding the regulatory mechanisms underlying the plant heat stress response is important for developing climate-resilient crops, including maize (Zea mays). To identify transcription factors (TFs) that may contribute to the maize heat stress response, we generated a dataset of short- and long-term transcriptome changes following a heat treatment time course in the inbred line B73. Co-expression network analysis highlighted several TFs, including the class B2a heat shock factor (HSF) ZmHSF20. Zmhsf20 mutant seedlings exhibited enhanced tolerance to heat stress. Furthermore, DNA affinity purification sequencing and Cleavage Under Targets and Tagmentation assays demonstrated that ZmHSF20 binds to the promoters of Cellulose synthase A2 (ZmCesA2) and three class A Hsf genes, including ZmHsf4, repressing their transcription. We showed that ZmCesA2 and ZmHSF4 promote the heat stress response, with ZmHSF4 directly activating ZmCesA2 transcription. In agreement with the transcriptome analysis, ZmHSF20 inhibited cellulose accumulation and repressed the expression of cell wall-related genes. Importantly, the Zmhsf20 Zmhsf4 double mutant exhibited decreased thermotolerance, placing ZmHsf4 downstream of ZmHsf20. We proposed an expanded model of the heat stress response in maize, whereby ZmHSF20 lowers seedling heat tolerance by repressing ZmHsf4 and ZmCesA2, thus balancing seedling growth and defense.


Introduction
Global warming is responsible for the increasingly frequent occurrence of intense heat waves characterized by very hot days and nights (Sinha et al. 2021).Crops and other plants are greatly impacted by this rise in global average temperatures, which threatens plant growth and thus agricultural yields (Zhao et al. 2017).Maize (Zea mays) is one of the three most important food crops worldwide and is severely affected by heat stress (HS), with a 7% value for drop in yield with each 1 °C increase in temperature (Tigchelaar et al. 2018).Hence, it is urgent to explore the genes that comprise the regulatory networks behind plant tolerance of HS to produce more climate-resilient varieties of maize, as well as other crops.
Heat shock factors (HSFs) are key regulatory factors of the plant HS response.When plants experience HS, HSFs bind to the cis-regulatory element known as heat shock element (HSE, 5′-nGAAnnTTCn-3′) present in the promoters of downstream genes to regulate the transcription of heat shock protein (HSP) genes (Andrasi et al. 2021).HSFs are divided into three subfamilies, A, B, and C, based on their specific structures (von Koskull-Doring et al. 2007).Among the HSFA subfamily, in Arabidopsis (Arabidopsis thaliana), HSFA1 is the master regulator that activates transcriptional networks in response to HS (Yoshida et al. 2011).In addition, SlHSFA7 affects plant heat tolerance by regulating the expression of SlHSFA1 in tomato (Solanum lycopersicum) (Mesihovic et al. 2022).In wheat (Triticum aestivum), TaHSFA1 functions in heat tolerance, and interaction with ubiquitin-binding proteins alters TaHSFA1 activity (Wang et al. 2023).Alternative splicing modulates the extent of transcriptional activation in response to HS in wheat, with alternative splicing of TaHSFA6e producing two major forms of TaHSFA6e that are distinguished by their binding affinity for downstream HSP genes (Wen et al. 2023).The overexpression of FaHSFA2c in tall fescue (Festuca arundinacea) enhances plant HS by inducing the transcriptional activation of HS responsive genes (Wang et al. 2017).
The HSFC and HSFB subfamilies also have important functions in plants.Arabidopsis plants heterologously expressing FaHSFC1b displayed enhanced tolerance of HS compared to control plants, suggesting that FaHSFC1b enhances heat tolerance in Arabidopsis (Zhuang et al. 2018).Moreover, OsHSFC1b contributes to the salt tolerance response in rice (Oryza sativa) (Schmidt et al. 2012).Notably, the heat tolerance conferred by HSFB subfamily members varies among plant species.In Arabidopsis, AtHSFB1 and AtHSFB2b are negative regulators of heat tolerance and their expression is essential for acquired heat tolerance (Ikeda et al. 2011).Similarly, Arabidopsis and rice plants heterologously expressing ZmHsfB2b displayed decreased tolerance for heat mediated by a mechanism affecting the accumulation of reactive oxygen species (Qin et al. 2022).By contrast, in the wild tomato Solanum peruvianum (also known as Lycopersicon peruvianum), LpHsfB1 is a coactivator that cooperates with LpHsfA1 to positively regulate plant heat tolerance (Bharti et al. 2004).In grape (Vitis vinifera), VvHSFB1 acts as a positive regulator of HS responses (Chen et al. 2023a(Chen et al. , 2023b)).The regulatory mechanism of HSFs has not, however, been explored in maize.
In addition, there is very little information about the functions of non-HSF proteins responsible for HS in crops, especially in maize as compared to Arabidopsis (Ohama et al. 2017).It is known that HS conditions can reduce protein stability, leading to the accumulation of misfolded proteins in the endoplasmic reticulum (ER) (Howell 2013).For example, ZmbZIP60 improves the heat tolerance of maize seedlings by increasing the expression of the gene heat upregulated (Hug1), which influences the stability of proteins in the ER (Li et al. 2020;Xie et al. 2022).Overexpression of ZmHsp101 in maize anthers results in robust microspores under HS (Li et al. 2022); similarly, loss of invertase alkaline neutral 6 (ZmInvan6) function affects the progression of maize meiosis at high temperature and lowers pollen fertility, thus influencing yield (Huang et al. 2022).The transcription factor Necrotic upper tips1 (nut1) participates in heat and drought stress responses by directly influencing cellulose biosynthesis and apoptosis during protoxylem development in maize flowering (Dong et al. 2020).The mitogen-activated protein kinase ZmMPK20 is phosphorylated by its upstream kinase MAPK kinase 9 (ZmMKK9) and regulates the stability of RPM1-interacting protein 2 (ZmRIN2) to enhance maize thermotolerance (Cheng et al. 2023).
Overall, much remains to be investigated about the regulatory mechanism underlying HS response in maize.Here, we identified ZmHSF20 as a central regulator of the maize HS response.We demonstrated that knocking out ZmHsf20, overexpressing ZmHsf4 (a member of the ZmHSFA subfamily), or overexpressing the cellulose synthase gene ZmCesA2 all improve the heat tolerance of maize seedlings.We defined the two submodules ZmHSF20-ZmHSF4-ZmCesA2 and ZmHSF20-ZmCesA2 as playing central roles in shaping maize heat tolerance mediated by cell wall-related genes.Understanding of these regulatory modules will elucidate the function of ZmHsfs in HS tolerance in maize.

Identification of key transcription factors related to the HS response
To identify putative transcription factors (TFs) with a role in the response to HS, we performed transcriptome deep sequencing (RNA-seq) of B73 seedlings exposed to heat treatments of 45 °C for 5, 15, 30 min, 2, or 8 h, together with matching control seedlings maintained under control conditions.We conducted three replicates for each condition, whose results were all highly correlated (Supplementary Data Set 1).We then performed weighted gene co-expression network analysis (WGCNA) by using differentially expressed genes (DEGs), and classified them into five modules (Fig. 1A).Gene ontology (GO) analysis of these five modules revealed that the turquoise module displays an enrichment in the GO terms "response to heat" and "heat acclimation", which are likely to be directly related to the HS response (Fig. 1B; Supplementary Data Set 2).Furthermore, members of the HSF and ERF families are significantly overrepresented in the turquoise module (Fig. 1C).
We constructed a co-expression regulatory network using the eight HSF TF genes (five HsfA genes and three HsfB genes) present in our data, and we observed that ZmHsf20 was among the top three HSFs with the largest number of connections (Fig. 1D); then, by checking their expression patterns under heat treatment, we found ZmHsf20 was strongly induced by HS (Supplementary Fig. S1).Hence, we considered ZmHsf20 as a potentially essential TF in the HS response.Phylogenetic analysis of the HSF protein families of Arabidopsis, rice, and maize revealed that ZmHSF20 belongs to the HSFB2 subfamily (Supplementary Fig. S2).We observed no transcriptional autoactivation activity for ZmHSF20 in yeast (Saccharomyces cerevisiae) (Supplementary Fig. S3).

ZmHSF20 negatively regulates heat tolerance
To explore the function of ZmHSF20 during HS, we obtained two knockout lines (Zmhsf20-1 and Zmhsf20-2) through clustered regularly interspaced short palindromic repeat (CRISPR)/CRISPR-associated nuclease 9 (Cas9)-mediated genome editing.The Zmhsf20-1 mutant contains a 252-bp deletion and the Zmhsf20-2 mutant contains a 24-bp deletion and a 25-bp deletion in the coding region of ZmHsf20; both alleles result in a frameshift mutation (Fig. 2A).Compared to KN5585 (wild type, WT) controls, both Zmhsf20 mutants displayed a heat-tolerant phenotype after exposure to 45 °C for 3 d followed by a 3-d recovery (Fig. 2C), with a survival rate of about 40% for the mutants   and only about 10% for WT (Fig. 2E).We measured higher ROS levels in WT than in the Zmhsf20 mutants after HS treatment (Fig. 2I).Moreover, ion leakage assays indicated that the plasma membrane of Zmhsf20 mutants suffered less damage upon HS than the WT (Fig. 2G).In addition, we assessed the expression level of three classical heat response genes, ZmHSP20, ZmHSP70-1, and ZmHSP70-2.All these genes exhibited significantly increased transcript levels in the Zmhsf20 mutants relative to the WT (Supplementary Fig. S4, A to C).
To validate the role of ZmHsf20 in response to HS, we generated an overexpression line for ZmHsf20 by driving the ZmHsf20 coding sequence from the maize Ubiquitin (Ubi) promoter in the KN5585 background (ZmHsf20-OE).We chose two ZmHsf20-OE lines with substantially elevated ZmHsf20 transcript levels, named ZmHsf20-OE #1 and ZmHsf20-OE #2, for analysis (Fig. 2B).Both lines showed a heat-sensitive phenotype compared to WT (Fig. 2, D and  F), accompanied by increased ion leakage (Fig. 2H), and more ROS accumulation (Fig. 2J), and relatively low expression levels of three classical heat response genes (Supplementary Fig. S4, D to F). Together, these data suggest that ZmHSF20 negatively regulates HS in maize.
To determine the subcellular localization of ZmHSF20, we cloned the full-length coding sequence of ZmHSF20 in frame and upstream of the sequence encoding green fluorescent protein (GFP).We infiltrated the construct into the leaves of Nicotiana benthamiana plants and either grew them under normal growth conditions or exposed them to HS treatment.We detected a strong green fluorescent signal in the nucleus in normal and different HS treatments, with full colocalization with the regions stained by a DNA dye, 4′, 6-diamidino-2-phenylindole dihydrochloride (DAPI) (Supplementary Fig. S5).We conclude that ZmHSF20 constitutively localizes in the nucleus regardless of the application of HS.

ZmHSF20 modulates heat tolerance partly by targeting several ZmCesA and ZmHsfA genes
To identify the direct targets of ZmHSF20, we performed RNA-seq using the Zmhsf20-1 mutant and WT.We detected 1,097 DEGs between the mutant and WT, with 641 genes upregulated and 456 downregulated in Zmhsf20-1 mutant (abs (log 2 fold-change) ≥ 1, P-value < 0.05, Supplementary Data Set 3).A GO term enrichment analysis revealed that the upregulated genes are primarily enriched in the terms "dephosphorylation", "plant-type secondary wall", "cellulose synthase (UDP-forming) activity", along with several other categories (Fig. 3A).
In a complementary approach, we conducted DNA affinity purification sequencing (DAP-seq) with recombinant purified ZmHSF20 incubated with an adapter-ligated genomic DNA library in two biological replicates.In total, we detected 77,145 ZmHSF20-binding peaks (q < 0.05), of which 4.99% mapped to promoter regions (Supplementary Fig. S6).GO analysis indicated that the genes whose promoter regions contain ZmHSF20-binding peaks are primarily enriched in "protein folding", "response to heat", and "cellulose synthase (UDP-forming) activity" (Fig. 3B).Indeed, we observed upregulation in the transcript levels of ZmCesA genes in the Zmhsf20 mutants compared to WT, based on reverse-transcription quantitative PCR (RT-qPCR) analysis (Supplementary Fig. S7).Taking these results together, we reasoned that ZmCesA genes are candidate direct targets of ZmHSF20.
In agreement with this hypothesis, published maize ZmCesA2 overexpression lines exhibit improved cold tolerance (Zeng et al. 2021).We, therefore, asked whether ZmHSF20 physically binds to the ZmCesA2 promoter by performing Cleavage Under Targets and Tagmentation followed by qPCR (CUT&Tag-qPCR) and electrophoretic mobility shift assays (EMSAs).We observed a significant enrichment for the P1/P2 region of the ZmCesA2 promoter among the immunoprecipitated chromatin by CUT&Tag-qPCR using ZmHsf20-OE #1 seedlings and an anti-MYC antibody, especially after HS treatment (Fig. 3C).We also used region P2 as a probe for EMSA and detected the specific binding of recombinant ZmHSF20 to a labeled probe containing the P2 region that could be competed away by co-incubation with unlabeled probe (Fig. 3D).These results demonstrate that ZmHSF20 binds to the ZmCesA2 promoter in vitro and in vivo.
To validate the effect of ZmHSF20 on ZmCesA2 transcription, we performed a transient expression assay in N. benthamiana leaves by co-infiltrating an effector construct consisting of the full-length ZmHsf20 coding sequence and the ZmCesA2 promoter driving the firefly luciferase (LUC) reporter gene as reporter construct.We noticed that co-expression of ZmHsf20 with the ProZmCesA2:LUC reporter decreased LUC activity relative to leaf regions infiltrated only with ProZmCesA2:LUC, indicating that ZmHSF20 directly represses the promoter activity of ZmCesA2 (Fig. 3, E and F).These results indicate that ZmHSF20 suppresses ZmCesA2 expression by directly binding to its promoter.
Previous studies have shown that some HSFB genes affect HS tolerance by modulating the expression of certain HSFA genes (Ikeda et al. 2011).However, an interaction between HSFBs and HSFAs in maize has been elusive.To investigate whether ZmHSF20 might regulate ZmHsfA family members, we performed a yeast one-hybrid (Y1H) assay to test the binding of ZmHSF20 to the promoters of seven individual ZmHsfA genes that are strongly induced by HS treatment (absolute value of log 2 fold-change ≥ 2, P-value < 0.05, Supplementary Data Set 2, Supplementary Fig. S1, B to H).Indeed, we determined that ZmHSF20 binds to the promoters of the ZmHsfA genes ZmHsf4, ZmHsf12, ZmHsf13, ZmHsf14, ZmHsf17, and ZmHsf24 (Supplementary Fig. S8).We used the CUT&Tag-qPCR approach to establish that ZmHSF20 binds only to the promoters of ZmHsf4, ZmHsf12, and ZmHsf17 in vivo, with greater binding after heat treatment (Fig. 4A; Supplementary Fig. S9).These results indicate that ZmHSF20, a member of the HSFB subfamily, may directly bind to the promoters of some HsfA subfamily members.

ZmHSF4 positively regulates heat tolerance and functions downstream of ZmHSF20
ZmHsf4, like other ZmHsfA genes, was highly induced by HS (Supplementary Fig. S1H) and belongs to the HSFA2 subfamily (Supplementary Fig. S2).We demonstrated that ZmHSF20 binds to the ZmHsf4 promoter in vitro and in vivo (Fig. 4, A  and B) and inhibits the transcriptional activity of the ZmHsf4 promoter in an N. benthamiana LUC assay (Fig. 4, C and D).From these results, we concluded that ZmHSF20 inhibits ZmHsfA transcription.
To explore the role of ZmHsf4, we obtained two knockout mutants (Zmhsf4-1 and Zmhsf4-2) via CRISPR/ Cas9-mediated genome editing.The Zmhsf4-1 mutant harbored a 137-bp deletion and Zmhsf4-2 carried a 7-bp deletion in the first exon of ZmHsf4 (Supplementary Fig. S10), both leading to early translation termination.These mutant lines were more susceptible to HS than WT, as evidenced by their lower survival rates (Fig. 4, E and G).In agreement, the Zmhsf4 mutants accumulated more ROS than WT under HS but not under normal growth conditions, as determined by DAB staining (Supplementary Fig. S11).Similarly, the ion leakage of the Zmhsf4 mutant lines was much higher than that of WT under heat treatment (Fig. 4I).As with ZmHSF20, we determined that ZmHSF4-GFP localizes to the nucleus under both normal and HS conditions (Supplementary Fig. S12).
In addition, we generated lines overexpressing ZmHsf4.We chose two ZmHsf4-OE lines with much higher ZmHsf4 transcript levels, named ZmHsf4-OE #1 and ZmHsf4-OE #2 (Supplementary Fig. S13).The WT showed a heat-sensitive phenotype compared to ZmHsf4-OE #1 and ZmHsf4-OE #2 (Fig. 4, F and H), which was accompanied by decreased ion leakage and lower ROS accumulation in the overexpression lines (Fig. 4J; Supplementary Fig. S11).Furthermore, we analyzed the expression level of HS-responsive HSP genes in the Zmhsf4 mutants and ZmHsf4-OE plants.The expression levels of ZmHSP20 and ZmHSP70 were significantly lower in the Zmhsf4 mutants and higher in the ZmHsf4-OE lines compared to WT following HS treatment (Supplementary Fig. S14).These results indicate that ZmHSF4 positively regulates the HS response in maize.
To decipher the genetic relationship between ZmHsf20 and ZmHsf4, we generated the Zmhsf20-1 Zmhsf4-1 double mutant by crossing Zmhsf20-1 to Zmhsf4-1.After 2 d of WT and ZmHsf20-OE seedlings were exposed to heat treatment at 45 °C HS treatment for 24 h or maintained at 28 °C/ 22 °C for CK (control check).CUT&Tag was performed using an anti-MYC antibody.Immunoprecipitated DNA was quantified by qPCR using primers specific to regions within the ZmHsf4 locus.The relative enrichment of ZmHSF20 binding to the ZmHsf4 promoter was normalized to TUBULIN2 (TUB2).Schematic diagrams of ZmHSF20 binding cis-elements in the promoter regions of ZmHsf4 (P1).Triangles represent the position of P1.Each experiment was performed at least three times with similar results.The values are means ± SD (n = 3 independent experiments).ns, not significant.***P < 0.001, one-way ANOVA.B) EMSA showing the binding of recombinant purified ZmHSF20 to the ZmHsf4 promoter in vitro.The "+" and "−" symbols represent the presence and absence of components, respectively.Biotin: GCCTTCCAGAACCTTCCTTTCCCGAAATAG.Mutant: GCCTTCCAAAAAAAAAAAAAAAAAAATAG.heat treatment at 45 °C, the Zmhsf4-1 and Zmhsf20 Zmhsf4 double mutants had a survival rate of about 20%, while that of the WT was about 40%.Conversely, the Zmhsf20-1 mutant displayed strong heat tolerance, with a survival rate at ∼80% (Fig. 5, F and G), and the loss of ZmHSF20 function did not suppress the heat-sensitive phenotype of the Zmhsf4-1 mutant.These genetic results demonstrate that ZmHSF4 functions downstream of ZmHSF20 in the HS response.

ZmCesA2 is a direct target of ZmHSF4 and confers heat tolerance
As we demonstrated that ZmHSF20 can affect the expression of some ZmCesA under HS (Supplementary Fig. S7), while ZmHsf4 is an HSF family member like ZmHsf20, we speculated that ZmHSF4 might also affect ZmCesA gene expression under HS.We thus analyzed the expression levels of some ZmCesA genes in the Zmhsf4 mutants and WT, which revealed that ZmCesA2, ZmCesA, ZmCesA12, and ZmCesA13 are significantly downregulated in the Zmhsf4 mutants, compared to WT, upon heat treatment (Supplementary Fig. S15).Considering that ZmHSF20 directly targets ZmCesA2 (Fig. 3, C to F), we asked whether ZmHSF4 would also physically interact with the ZmCesA2 promoter via CUT&Tag-qPCR and EMSA.Indeed, we obtained evidence that ZmHSF4 binds to the ZmCesA2 promoter in vivo and vitro (Fig. 5, A and B).We also co-infiltrated a LUC reporter consisting of the ZmCesA2 promoter driving LUC (ProZmCesA2:LUC) together with the empty effector vector (as control) or a ZmHsf4 effector construct into N. benthamiana leaves.We detected low LUC activity from the ProZmCesA2:LUC reporter alone but higher LUC activity when it was co-expressed with ZmHsf4 (Fig. 5, C and D).Overall, these results suggest that ZmHSF4 directly targets ZmCesA2 and activates its transcription.
To determine whether ZmCesA genes regulated by ZmHSF20 and ZmHSF4 contribute to HS tolerance in maize, we obtained two published overexpression lines for ZmCesA2 (ZmCesA2-OE lines), which was previously shown to regulate cold stress in maize (Zeng et al. 2021).Strikingly, ZmCesA2-OE lines displayed higher survival rates after HS compared to the WT (LH244) (Fig. 5E).These results suggest that ZmCesA2 enhances HS tolerance in maize.Since several studies have shown that there is some association between cellulose and HS (Le Gall et al. 2015), we measured the cellulose content of WT, the ZmHsf20-OE, ZmHsf4-OE, and CesA-OE lines, and the Zmhsf20 and Zmhsf4 mutants.We determined that the Zmhsf20 mutants and the ZmHsf4-OE and CesA-OE lines accumulate more cellulose than WT, suggesting that the presence of more cellulose might improve heat tolerance (Supplementary Fig. S16).
In a previous study, ZmCesA genes and cell wall-related genes were shown to affect the formation of the cell wall, which forms a physical barrier that might influence basic plant tolerance of abiotic stress (Penning et al. 2019).We looked at the transcript levels of several cell wall-related genes mentioned in this study, such as GALACTURONOSYLTRANSFERASE-LIKE 5 (ZmGATL5, Zm00001d028824), a homolog of Arabidopsis GAlactUronosylTransferase 1 (GAUT1) and GAUT7, which play essential roles in plant cell wall pectin biosynthesis (Sterling et al. 2006;Atmodjo et al. 2011); PHENYLALANINE AMMONIA-LYASE 1 (ZmPAL1, Zm00001d017274), a potentially key enzyme involved in the biosynthesis of cell wall polymer lignin (Huang et al. 2010) and FERULIC ACID 5-HYDROXYLASE (ZmF5H, Zm00001d032467) which was involved in lignin biosynthesis (Anderson et al. 2015), in ZmCesA-OE, Zmhsf20, Zmhsf4, and WT (Supplementary Fig. S17).
All of these genes exhibited increased transcript levels in the Zmhsf20 mutants, ZmHsf4-OE, and ZmCesA2-OE.Further, by employing the transmission electron microscope (TEM) analysis, we checked the structure of cell walls in the WT, Zmhsf20 mutants, and ZmHsf4-OE lines.Under HS condition, the structure of cell walls of WT seems to be distorted, while the Zmhsf20 mutants and ZmHsf4-OE lines possessed relatively smooth and uniform cell wall structure; by contrast, no obvious difference was detected among them under normal condition (Supplementary Fig. S18).These results together suggest that Zmhsf20 mutants and ZmHsf4-OE lines may directly or indirectly mediate the cell wall development under HS, thereby improving maize heat tolerance.

Discussion
Only a few HSF family genes have been previously reported to be related to abiotic stress in maize, such as ZmHsf08, a class A member conferring greater tolerance to salt and drought stress (Wang et al. 2021), and ZmHsf07 (named as ZmHsf11), a class B2b member, whose heterologous expression in Arabidopsis and rice led to reduced heat tolerance (Qin et al. 2022).In this study, we report that ZmHSF20, a member of the B2a class of maize HSFs, confers heat tolerance at the seedling stage.Furthermore, ZmHSF20 acts upstream of ZmHSF4 and ZmCesA2, both of which positively regulate heat response by modulating cellulose content and cell wall-related gene expression in maize.
A phylogenetic analysis of HSF proteins from several plant species indicated that ZmHSF20 belongs to the HSFB subfamily (Supplementary Fig. S2).Although the HSFB subfamily is the most highly conserved of the three HSF subfamilies, individual members are involved in a variety of biological pathways.In Arabidopsis, the clock protein PSEUDO-RESPONSE REGULATOR7 (PRR7) regulates rhythmic responses to abiotic stress by repressing the expression of HSFB2b (Kolmos et al. 2014).In the Arabidopsis hsfb1 hsfb2b double mutant, loss of disease resistance arises through misregulation of the plant defensin genes PDF1.2a and PDF1.2b (Kumar et al. 2009).In addition, HSFB2b positively regulates salt tolerance in soybean (Glycine max) (Bian et al. 2020), reflecting the diversity of HSFB functions and the complexity of their downstream genes.This study adds ZmHSF20 to the small list of ZmHSFB family members that negatively regulate heat tolerance in maize.
In our study, we discovered that the expression of 9 of 16 ZmHsfA genes either is not induced by HS or responds only weakly to it, suggesting that these maize HSFAs may be involved in other stress responses and not HS (abs(log 2 foldchange) < 2, P-value < 0.05, Supplementary Data Set 2, Supplementary Fig. S19).In Arabidopsis, AtHSFA1 is the main factor contributing to heat tolerance (Liu et al. 2011), and AtHSFA2 also is most strongly expressed under HS (Schramm et al. 2006).AtHSFA2 and AtHSFA3 largely bind to the same regions in the Arabidopsis genome and may create a transcriptional memory of HS exposure by forming complexes that deposit hypermethylation marks at the promoters of their target genes (Friedrich et al. 2021;Kappel et al. 2023).In addition, Arabidopsis AtHSFB2b and AtHSFB1 are reported to inhibit the expression of HSFA2 and HSFA7, thereby influencing heat tolerance (Kappel et al. 2023).By contrast, our results showed that ZmHSF20, an HSFB2a subfamily member, affects the transcript levels of ZmHsf12, ZmHsf17, and ZmHsf4 of the ZmHsfA2 subfamily (Fig. 4A; Supplementary Fig. S9).In maize, a previous report indicated that ZmbZIP60, a TF, modulates heat tolerance by regulating ZmHsf13, from the ZmHsfA2 subfamily (Li et al. 2020), while the exact mechanism how ZmHSFA2 regulates HS has not been explored.In our study, we demonstrated that ZmHSF4 positively regulates heat tolerance.Therefore, we concluded that the ZmHSFA2 subfamily also has an important function in heat tolerance in maize, albeit ZmHSFA2 targeted by different TFs.
The relationship between cellulose synthase and the HS response in maize is unclear.Previous studies revealed that TFs from the NAC and MYB families regulate the expression of CesA genes (Lampugnani et al. 2019).In Arabidopsis, HSFA7b induces CesA expression, thereby affecting salt tolerance (Zang et al. 2019).In this study, we established that members of the ZmHSF family regulate the transcription of ZmCesA genes.Overexpression of ZmCesA2 improved the HS tolerance of maize seedlings.Notably, ZmCesA2 overexpression also improves cold tolerance in maize (Zeng et al. 2021), suggesting that it may enhance basal tolerance of abiotic stress.Thus, ZmCesA2 and its associated genetic pathway may be important tools for the future engineering of crop plants to resist a range of abiotic stresses.
The plant cell wall is composed of polysaccharides, which protect plants from various external stresses (Wang et al. 2016).Although cell wall remodeling is an important component of plant responses to HS, the exact nature of the connection between cell wall remodeling and HS response is unclear.The structures of pectin and lignin undergo structural changes in response to HS (Wu et al. 2018).Previous studies showed that CesA genes are upregulated and cellulose content increases under HS (Le Gall et al. 2015).CesA is crucial in the biosynthesis of cellulose, a major component of plant cell walls.Higher temperatures have been shown to affect the processing speed of CesA, with extremely high temperatures resulting in lower contents of crystalline cellulose (Fujita et al. 2011).In alfalfa (Medicago sativa), the expression of several CesA genes increases with higher temperature (Guerriero et al. 2014).Thus, CesA genes and other cell wall-related genes influence the formation and composition of the cell wall, which in turn would be expected to affect plant stress tolerance (Penning et al. 2019).
We speculate that the structure of cell wall change may explain the heat tolerance effects observed in maize.In our study, ZmCesA genes were upregulated in Zmhsf20 mutants and ZmHsf4-OE lines under HS conditions (Supplementary Figs.S7 to S15) leading to increased cellulose production (Supplementary Fig. S16).Moreover, the expression of some cell wall-related genes were upregulated between Zmhsf20 mutants, ZmHsf4-OE lines and WT (Supplementary Fig. S17).Correspondingly, the structure of cell walls of Zmhsf20 mutants and ZmHsf4-OE lines seems to be more robust to adapt to HS compared to WT (Supplementary Fig. S18).Nevertheless, the exact mechanism how ZmCesA2 regulates HS is to be investigated.
The Zmhsf20 mutants displayed enhanced tolerance of HS compared to WT at the seedling stage, but seed setting rate was comparable between these genotypes at the mature stage.We hypothesize that ZmHSF20 may promote maize growth by repressing heat tolerance during the seedling stage.We did not observe any clear differences in the seedsetting rates of the mutants and WT in plants grown at Yazhou, Sanya, Hainan, which experienced normal temperatures during the winter of 2022, or at Zhuozhou, Hebei, where plants were exposed to HS in July 2023 (http://data.cma.cn/) (Supplementary Figs.S20 and S21).However, a 1-year experiment is not sufficient to draw strong conclusions.Hence, these experiments should be repeated in the future.
We conclude that ZmHSF20, a member of the maize HSFB subfamily, may decrease heat tolerance by negatively regulating the expression of cellulose synthase-related genes (Fig. 6).HSFs regulate plant cell wall remodeling, thereby providing targets and potential strategies for breeding to facilitate plant adaptation to HS.

Plant growth conditions
All transgenic maize (Z.mays) plants and CRISPR/Cas9 mutants were generated by WIMI Biotechnology Co. Ltd.The constructs were transformed into the inbred line KN5585.Maize seeds were planted in pots (40 × 30 × 15 cm, length × width × depth) containing vermiculite, and Pindstrup soil mix (1:1, v:v) and grown at 28 °C/22 °C (day/night) under a 16 h light/8 h dark photoperiod with 150 µmol m −2 s −1 white light and 40% relative humidity.Seeds of N. benthamiana were germinated on soil and seedlings were grown in the same growth chamber as above.Four-wk-old N. benthamiana plants were used for experiments.
For heat treatment, T 2 generation maize seedlings were grown in pots (40 × 20 × 15 cm, length × width × depth) to the V2 stage with irrigation before being exposed to 45 °C under a 16 h light/8 h dark photoperiod for 2 to 4 d using a heat chamber.The control plants were grown in the similar chamber except for heat treatment.After heat treatment, the seedlings were allowed to recover at 28 °C/ 22 °C (16 h light/8 h dark) for 3 d prior to imaging.

Physiological assays
The ion leakage assays were performed as described previously (Zeng et al. 2021).Leaves from V2 stage maize seedlings then being exposed to 45 °C 1 d under a 16h light/8 h dark were placed in 15 mL centrifuge tubes with 10 mL double distilled water (ddH 2 O).The solution was vacuumed for 30 min, and the conductance of the water was measured as S0.After being shaken at room temperature for 1 h, the solution was detected as S1.Then the samples were boiled for 30 min and shaken at room temperature for cooling, followed by detection of the conductance of the water, defined as S2.The results were calculated as follows: ion leakage (%) = (S1 − S0)/(S2 − S0).
To detect H 2 O 2 accumulation in situ, staining with 3,3diaminobenzidine (DAB) was used as described (Jambunathan 2010).Briefly, V2 stage maize seedlings were exposed to 45 °C for 1 d before being transferred to 1.0 mg/mL DAB solution for 10 h and vacuum-infiltrated at 37 °C for 30 min.Subsequently, the leaves were cleared with 95% ethanol (v:v) until colorless.
The full-length coding sequence of ZmHsf20 was cloned into the pGADT7 vector (EcoRI and BamHI) to generate the AD-ZmHsf20 construct (Shen et al. 2021).The ZmHsf4, ZmHsf11, ZmHsf12, ZmHsf13, ZmHsf14, ZmHsf17, and ZmHsf24 promoters were individually amplified from KN5585 genomic DNA and cloned into the pHIS2 vector (XmaI) to generate the recombinant pHis2-promoter construct (Shen et al. 2021).The pET28a vectors (HindIII and BamHI) were used to produce His-ZmHSF20 and His-ZmHSF4 recombinant proteins (Shen et al. 2021).To obtain the Super:ZmHsf20-GFP and ZmHsf4-GFP plasmids, the full-length coding sequences of ZmHsf20 and ZmHsf4 were individually cloned into the pSuper:1300-GFP vector (HindIII and KpnI) (Tian et al. 2022).The 2.0-kb promoter fragments of ZmHsf4 and ZmCesA2 were amplified and cloned into the pGreenII 0800-LUC vector (HindIII and BamHI) to obtain the ProZmHsf4:LUC, and ProZmCesA2:LUC constructs, respectively (Shen et al. 2021).The ProZmHsf4:LUC, and ProZmCesA2:LUC plasmids were transformed separately into Agrobacterium (strain GV3101) together with the helper plasmid (pSoup-P19) for subsequent assays.All constructs were generated using a Seamless Assembly Cloning kit (Clone Smarter).The primers and restriction enzymes used for plasmid construction are listed in Supplementary Data Set 4.

RNA extraction and RT-qPCR
Total RNA was extracted from the leaves of V2 stage maize seedlings using a plant RNA kit (ER301; TransGen Biotech, China).Total RNA (1 µg) was reverse transcribed into firststrand complementary DNA (cDNA) using M-MLV reverse transcriptase according to the manufacturer's instructions.RT-qPCR analyses were performed as previously described (Zhang et al. 2019).ACTIN was used as internal reference to normalize the expression value of each sample.Primers used for qPCR are shown in Supplementary Data Set 4. The experiments were independently performed at least three times (three biological replicates from different plants).

RNA-seq analysis
For RNA-seq, the leaves and stems of V2 stage maize B73 seedlings grown at 28 °C/22 °C (day/night) or exposed to heat treatment at 45 °C for 5, 15, 30 min, 2, or 8 h, with matching control samples grown at 28 °C/22 °C (day/night) for each heat treatment time point were collected to extract total RNA with TRIzol reagent.For RNA-seq, the second leaves of V2 stage maize seedlings (Zmhsf20-1 and WT) exposed to heat at 45 °C for 24 h were used to extract total RNA with TRIzol reagent.The sequencing library was constructed using 1 µg total RNA from each sample.Libraries were prepared using the TruSeq RNA Library Preparation kit (Illumina, USA) and sequenced on an Illumina Novaseq platform, and 150-bp paired-end reads were generated.fastp (Chen et al. 2018) was used to remove the adaptor, low-quality bases (Q30), and sequences containing >10% undetermined bases.The filtered reads were aligned to the maize reference genome B73_AGPv4 (Jiao et al. 2017) using Hisat2 (Kim et al. 2015) with default parameters.Subsequently, Stringtie (Pertea et al. 2015) was used to quantify the unique alignment reads (FPKM), and the Pearson correlation coefficient was used to evaluate the repeatability between samples.

Figure 1 .
Figure 1.Co-expression analysis and gene regulatory networks of the maize HS response constructed by WGCNA.A) Cluster dendrogram, module detection, and heatmap representation of gene expression of heat response genes at different times during HS treatment.B) GO term enrichment analysis for genes enriched in the turquoise module.C) The HSF TF gene family is significantly overrepresented in the turquoise module.D) Co-expression network based on eigengenes in the turquoise module.Genes are shown as circles, and the gray lines indicate gene-gene associations (false discovery rate < 0.01).The hub genes (TF genes of the Hsf family) with |weight| > 0.4 are indicated by diamonds.Diamond sizes means |weight| value.

Figure 2 .
Figure 2. Heat tolerance is modulated by ZmHsf20.A) Construction of CRISPR/Cas9-mediated Zmhsf20 knockout lines.Two sgRNAs that specifically target ZmHsf20 were designed, leading to the identification of two mutants, Zmhsf20-1 and Zmhsf20-2.black rectangles exons, white rectangles Un-Translated Regions (UTRs) and horizontal black lines rectangles introns.B) Relative ZmHsf20 transcript levels in the leaves of V2 stage seedlings of ZmHsf20-OE and wild type (WT, KN5585).ACTIN was used as the internal control.The error bars are based on three independent experiments.The values are means ± SD (n = 3 independent experiments).***P < 0.001, one-way ANOVA.C) Representative photographs of V2 stage seedlings of WT, Zmhsf20-1, and Zmhsf20-2 grown at 28 °C/22 °C (top) or exposed to 45 °C HS treatment for 3 d followed by a 3-d recovery (bottom).Scale bar = 2 cm.D) Representative photographs of V2 stage seedlings of WT, ZmHsf20-OE #1, and ZmHsf20-OE #1 grown at 28 °C/22 °C (top) or exposed to 45 °C for 2 d followed by a 3-d recovery (bottom).Scale bar = 2 cm.E, F) Survival rate of seedlings after recovery at 28 °C/22 °C for 3 d in (C, D).The error bars are based on three independent experiments.The values are means ± SD (n = 3 independent experiments).*P < 0.05, **P < 0.01, ***P < 0.001, one-way ANOVA.G, H) Ion leakage rate of V2 stage seedlings grown at 28 °C/22 °C exposed to 45 °C for 1 d.The error bars are based on three independent experiments.The values are means ± SD (n = 3 independent experiments).*P < 0.05, ***P < 0.001, one-way ANOVA.I, J) Representative photographs of leaves from V2 stage seedlings grown at 28 °C/22 °C, exposed to 45 °C for 1 d, and stained with DAB, scale bar = 1.5 mm.

Figure 3 .
Figure 3. ZmHSF20 binds to the ZmCesA2 promoter and activates its transcription.A) GO term enrichment analysis of ZmHSF20-regulated genes following HS treatment.B) Top GO terms enriched among ZmHSF20-bound genes as identified by DAP-seq.C) CUT&Tag assay showing the binding of ZmHSF20 to the ZmCesA2 promoter in vivo.WT and ZmHsf20-OE seedlings were exposed to heat treatment at 45 °C HS treatment for 24 h or maintained at 28 °C/22 °C for CK (control check).CUT&Tag was performed using an anti-MYC antibody.Immunoprecipitated DNA was quantified by qPCR using primers specific to regions within the ZmCesA2 locus.The relative binding of ZmHSF20 to the ZmCesA2 promoter was normalized to TUBULIN2 (TUB2).Schematic diagrams of ZmHSF20 binding cis-elements in the promoter regions of ZmCesA2 (P1 and P2).Triangles represent the position of P1 and P2.Each experiment was performed at least three times with similar results.The values are means ± SD (n = 3 independent experiments).ns, not significant.**P < 0.01, ***P < 0.001, one-way ANOVA.D) EMSA showing the binding of recombinant purified ZmHSF20 to the ZmCesA2 promoter in vitro.The "+" and "−" symbols represent the presence and absence of components, respectively.Biotin: ATTTACTACTGAAATTTATAAGGATTTGCA.Mutant: ATTTACTACTAAAAAAAATAAGGATTTGCA.E) Interaction assays between ZmHSF20 and the ZmCesA2 promoter by transient expression in N. benthamiana leaves, based on a luciferase reporter assay.Similar results were obtained for three biological repeats.F) Quantitative analysis of the LUC activity in (E), the values are means ± SD (n = 3 independent experiments).**P < 0.01, one-way ANOVA.

Figure 4 .
Figure 4. Heat tolerance is modulated by ZmHsf4, a downstream target gene of ZmHSF20.A) CUT&Tag assay showing the binding of ZmHSF20 to the ZmHsf4 promoter in vivo.WT and ZmHsf20-OE seedlings were exposed to heat treatment at 45 °C HS treatment for 24 h or maintained at 28 °C/ 22 °C for CK (control check).CUT&Tag was performed using an anti-MYC antibody.Immunoprecipitated DNA was quantified by qPCR using primers specific to regions within the ZmHsf4 locus.The relative enrichment of ZmHSF20 binding to the ZmHsf4 promoter was normalized to TUBULIN2 (TUB2).Schematic diagrams of ZmHSF20 binding cis-elements in the promoter regions of ZmHsf4 (P1).Triangles represent the position of P1.Each experiment was performed at least three times with similar results.The values are means ± SD (n = 3 independent experiments).ns, not significant.***P < 0.001, one-way ANOVA.B) EMSA showing the binding of recombinant purified ZmHSF20 to the ZmHsf4 promoter in vitro.The "+" and "−" symbols represent the presence and absence of components, respectively.Biotin: GCCTTCCAGAACCTTCCTTTCCCGAAATAG.Mutant: GCCTTCCAAAAAAAAAAAAAAAAAAATAG.C) Interaction assay between ZmHSF20 and the ZmHsf4 promoter by transient expression in N. benthamiana leaves, based on a luciferase reporter assay.Similar results were obtained for three biological repeats.D) Quantitative analysis of the LUC activity in (C).The values are means ± SD (n = 3 independent experiments).**P < 0.01, one-way ANOVA.E) Representative photographs of V2 stage seedlings of WT, Zmhsf4-1, and Zmhsf4-2 grown at 28 °C/22 °C (top) or exposed to 45 °C for 2 d followed by a 3-d recovery (bottom).Scale bar = 2 cm.F) Representative photographs of V2 stage seedlings of WT, ZmHsf4-OE #1, and ZmHsf4-OE #2 grown at 28 °C/22 °C (top) or exposed to 45 °C for 3 d followed by a 3-d recovery (bottom).Scale bar = 2 cm.G, H) Survival rate of seedings to recovery at 28 °C/22 °C for 3 d in (E, F).The error bars are based on three independent experiments.The values are means ± SD (n = 3 independent experiments).***P < 0.001, one-way ANOVA.I, J) Ion leakage rate of V2 stage seedlings grown at 28 °C/22 °C exposed at 45 °C for 1 d.The error bars are based on three independent experiments.The values are means ± SD (n = 3 independent experiments).*P < 0.05, one-way ANOVA.
Figure 4. Heat tolerance is modulated by ZmHsf4, a downstream target gene of ZmHSF20.A) CUT&Tag assay showing the binding of ZmHSF20 to the ZmHsf4 promoter in vivo.WT and ZmHsf20-OE seedlings were exposed to heat treatment at 45 °C HS treatment for 24 h or maintained at 28 °C/ 22 °C for CK (control check).CUT&Tag was performed using an anti-MYC antibody.Immunoprecipitated DNA was quantified by qPCR using primers specific to regions within the ZmHsf4 locus.The relative enrichment of ZmHSF20 binding to the ZmHsf4 promoter was normalized to TUBULIN2 (TUB2).Schematic diagrams of ZmHSF20 binding cis-elements in the promoter regions of ZmHsf4 (P1).Triangles represent the position of P1.Each experiment was performed at least three times with similar results.The values are means ± SD (n = 3 independent experiments).ns, not significant.***P < 0.001, one-way ANOVA.B) EMSA showing the binding of recombinant purified ZmHSF20 to the ZmHsf4 promoter in vitro.The "+" and "−" symbols represent the presence and absence of components, respectively.Biotin: GCCTTCCAGAACCTTCCTTTCCCGAAATAG.Mutant: GCCTTCCAAAAAAAAAAAAAAAAAAATAG.C) Interaction assay between ZmHSF20 and the ZmHsf4 promoter by transient expression in N. benthamiana leaves, based on a luciferase reporter assay.Similar results were obtained for three biological repeats.D) Quantitative analysis of the LUC activity in (C).The values are means ± SD (n = 3 independent experiments).**P < 0.01, one-way ANOVA.E) Representative photographs of V2 stage seedlings of WT, Zmhsf4-1, and Zmhsf4-2 grown at 28 °C/22 °C (top) or exposed to 45 °C for 2 d followed by a 3-d recovery (bottom).Scale bar = 2 cm.F) Representative photographs of V2 stage seedlings of WT, ZmHsf4-OE #1, and ZmHsf4-OE #2 grown at 28 °C/22 °C (top) or exposed to 45 °C for 3 d followed by a 3-d recovery (bottom).Scale bar = 2 cm.G, H) Survival rate of seedings to recovery at 28 °C/22 °C for 3 d in (E, F).The error bars are based on three independent experiments.The values are means ± SD (n = 3 independent experiments).***P < 0.001, one-way ANOVA.I, J) Ion leakage rate of V2 stage seedlings grown at 28 °C/22 °C exposed at 45 °C for 1 d.The error bars are based on three independent experiments.The values are means ± SD (n = 3 independent experiments).*P < 0.05, one-way ANOVA.

Figure 5 .
Figure 5. ZmHSF4 binds to the ZmCesA2 promoter and activates its transcription.A) CUT&Tag assay showing the binding of ZmHSF4 to the ZmCesA2 promoter in vivo.WT and ZmHsf4-OE seedlings were exposed to heat treatment at 45 °C HS treatment for 24 h or were maintained at 28 °C/22 °C for CK (control check).CUT&Tag was performed using an anti-MYC antibody.Immunoprecipitated DNA was quantified by qPCR using primers specific to regions within the ZmCesA2 locus.The relative enrichment of ZmHSF4 binding to the ZmCesA2 promoter was normalized to TUBULIN2 (TUB2).Schematic diagrams of ZmHSF4 binding cis-elements in the promoter regions of ZmCesA2 (P1 and P2).Triangles represent the position of P1 and P2.Each experiment was performed at least three times with similar results.The values are means ± SD (n = 3 independent experiments).ns, not significant.*P < 0.05, ***P < 0.001, one-way ANOVA.B) EMSA showing the binding of recombinant purified ZmHSF4 to the ZmCesA2 promoter in vitro.The "+" and "−" symbols represent the presence and absence of components, respectively.Biotin: GCCTTCCAGAACCTTCCTTTCCCGAAATAG.Mutant: GCCTTCCAAAAAAAAAAAAAAAAAAATAG.C) Interaction assay between ZmHSF4 and the ZmCesA2 promoter in N. benthamiana leaves, based on a luciferase reporter assay.Similar results were obtained for three biological repeats.D) Quantitative analysis of the LUC activity in (C).The values are means ± SD (n = 3 independent experiments).***P < 0.001, one-way ANOVA.E) Representative photographs of WT, ZmCesA2-OE #3, and ZmCesA2-OE #6 grown at 28 °C/22 °C (top) or exposed to 45 °C for 4 d followed by a 3-d recovery (bottom).Scale bar = 2 cm.F) Representative photographs of V2 stage seedlings of WT, Zmhsf20-1, Zmhsf4-1, and Zmhsf20-1 Zmhsf4-1 grown at 28 °C/22 °C (top) or exposed to 45 °C for 2 d followed by a 3-d recovery (bottom).Scale bar = 2 cm.G) Survival rate of seedlings after recovery at 28 °C/22 °C for 3 d in (F).The values are means ± SD (n = 3 independent experiments).Different lowercase letters indicate statistically significant differences (adjusted P < 0.05, one-way ANOVA).

Figure 6 .
Figure 6.Proposed model of ZmHSF20 function in conferring tolerance to HS in maize.Under HS, ZmHsf20 transcripts and ZmHSF20 protein accumulate, leading to the direct transcriptional repression of ZmHsf4 and ZmCesA2 expression.In parallel, ZmHSF4 normally promotes the expression of ZmCesAs, such that heat treatment further decreases cellulose content in a ZmHSF20-and ZmHSF4-dependent manner.Flat represents inhibition.Pointed represents activation.Thicknesses represents cellulose content.