Revealing Differentially Expressed Genes and Identifying Effector Proteins of Puccinia striiformis f. sp. tritici in Response to High-Temperature Seedling Plant Resistance of Wheat Based on Transcriptome Sequencing

In the present study, we performed transcriptomic analysis to identify differentially expressed genes and effector proteins of Puccinia striiformis f. sp. tritici (Pst) in response to the high-temperature seedling-plant (HTSP) resistance in wheat. Experimental validation confirmed the function of the highest upregulated effector protein, PstCEP1. This study provides a key resource for understanding the biology and molecular basis of Pst responses to wheat HTSP resistance, and PstCEP1 may be used in future studies to understand pathogen-associated molecular pattern-triggered immunity and effector-triggered immunity processes in the Pst-wheat interaction system.

transcriptome sequencing (RNA-seq) and analyzed Pst differentially expressed genes (DEGs) during the pathogen incubation period in relation to the HTSP resistance in XY6. In total, 25 DEGs and 34 secreted proteins were identified; one of these proteins (Pst candidate effector protein 1, PstCEP1) was validated for its function as a candidate effector. The transcript profile of PstCEP1 was analyzed under different temperature treatments. PstCEP1 functions were identified with the bacterial type three secretion system (TTSS)-mediated overexpression and barley stripe mosaic virus (BSMV)mediated host-induced gene silencing (HIGS). The results indicate that PstCEP1 is a candidate effector and responds to the HTSP resistance in XY6.

RESULTS
Overview of the Pst transcriptome. The Pst transcriptome was profiled and analyzed at two time points (0 and 24 h) on wheat plants inoculated with Pst, which were subjected to three treatments: (i) normal temperature (N), (ii) normal-high-normal temperature (NHN), and (iii) high temperature (H). A total of 15 samples, each treatment with three biological replicates, were analyzed with the same sample used for both NHN and N at 0 h. First, we compared the mixture of wheat and stripe rust (CYR32) samples with Chinese Spring wheat genome and then obtained the host genes for the HTSP resistance. Second, we obtained the genes of CYR32 involved in the HTSP resistance by the reads compared with the Pst-78 genome. We have also calculated the proportion of mapping of host and pathogen according to their genome, respectively. We obtained a total of 340,757,732 reads with an average of 22,717,182 reads per sample. The average mapping rate to Chinese Spring genome was 64.51% (see Table S1 in the supplemental material). In addition, 13,526 genes were predicted when the reads were mapped onto the Pst-78 genome with an average mapping rate of 80.27% (Table S1; Fig. S1). Gene expression levels were expressed as fragments per kilobase of gene per million mapped fragments (FPKM). For nearly a quarter of the genes, the expression levels were in the range of 10 and 100 (FPKM) for each treatment. About 15.21% of genes had high expression levels with FPKM Ͼ 100 for the inoculated plants under the N treatment at 24 h (I-N-24), compared with the corresponding value of 5.82% for the NHN treatment ( Table 1). The first two principal components (PCs) explained 64.1% and 18.0% of the total variation, respectively. The I-N-0 and I-N-24 samples differed from those high-temperature-inoculated samples (I-H-24 and I-NHN-24) (Fig. 1), indicating that the temperature treatments significantly affected Pst gene expression.
Identification of DEGs and effector candidates of Pst in response to the HTSP resistance in XY6. We used the EdgeR package as implemented in version 2.0.6 to identify DEGs of Pst in response to the HTSP resistance in XY6. The false discovery rate (FDR) Ͻ 0.05, the absolute value of log 2 ratio Ͼ Ϯ2, and mean abundance logCPM (log 2 counts per million) Ͼ Ϫ2 were used as the threshold to identify DEGs between the NHN and H treatments. Twenty-five DEGs were identified, and their functions were annotated based on the Nr database (  To verify the gene expression profiles from RNA-seq analysis, eight transcripts were randomly selected from those 25 DEGs for qRT-PCR analysis. The I-N-0 sample was used as a control when calculating relative expression levels. The amplification efficiency and dissolution curve for each gene are given in Fig. S2. The qRT-PCR results were significantly correlated (P Ͻ 0.0001) with the RNA-seq data ( Fig. 2A) with a correlation coefficient of 0.79 (Fig. 2B). We searched the 25 DEGs against the KEGG database for classification and functional annotation. These DEGs were mainly involved in immune response, signal transduction, and protein transport pathways (Q Ͻ 0.05) (Table 3). Similarly, the results of GO enrichment showed that these DEGs were mainly involved in membrane protein, ribonucleotide binding protein, synthesis of nitrogen compounds, and thiamine biosynthesis (Fig. S3, S4, and S5).
Small secreted proteins have potential roles in the plant-microbe interactions. We analyzed secreted proteins based on the gene expression during the Pst infection stage. A total of 1,053 candidate secretory proteins were predicted from 13,526 mapped genes, among which 34 were predicted to have virulence or pathogenicity functions according to the PHI (pathogen-host interaction) database (Table 4). From these 34 genes, we selected the highest-upregulated gene, PstCEP1 (18.68-fold), as a candidate effector for functional validation.
Cloning and sequence characterization of PstCEP1. A 1,321-bp cDNA fragment was isolated from Pst via reverse transcription-PCR (RT-PCR) and rapid amplification of cDNA ends (RACE). This cDNA fragment encodes a small protein of 243 amino acids, containing four Cys. The first 25 amino acids at the N terminus of PstCEP1 are predicted to be a signal peptide based on the detection with SignalP ver. 5.0 software (35). The yeast signal sequence trap system (36) was used to confirm the secretory function of this signal peptide. Yeast mutant strain YTK12 cannot grow on the complete minimal plates lacking tryptophan (CMD-W), nor can it grow on the yeast extract, peptone, raffinose, and antimycin A (YPRAA) medium. When the pSUC2T7M13ORI (pSUC2) (36) empty vector was transferred to the YTK12 strain, the tryptophan and nonsecretory sucrase could be encoded, and therefore YTK12 could grow on the CMD-W medium but not on the YPRAA medium. When the pSUC2 vector was linked to a signal peptide that possesses a secretory function, the sucrase could be activated and secreted into the YPRAA medium to hydrolyze the raffinose into glucose, required by strain YTK12. The predicted signal peptide of Phytophthora sojae effector Avr1b (37) was used as a positive control. The results showed that the putative signal peptide of PstCEP1 enabled YTK12 to grow on both the CMD-W and YPRAA media, indicating that PstCEP1 is a secretory protein (Fig. S6).
Transcriptional expression analysis of PstCEP1. A previous study (38) demonstrated that the HTSP resistance in XY6 was activated by exposure of inoculated plants to 20°C for 24 h. Two treatments were therefore performed to confirm transcriptomic results: (i) inoculated plants from 15°C were transferred to 20°C at 192 h postinoculation (hpi) for 24 h and then returned to 15°C, i.e., the NHN treatment, and (ii) the plants remained at 15°C, i.e., the N treatment. The transcriptional expression level of PstCEP1 was determined in the leaf samples collected at 0 (prior to inoculation), 48,96,192,194,198,204,216,240,264, and 312 hpi with qRT-PCR. The expression level of PstCEP1 increased gradually with the incubation time before the NHN treatment (0 to 192 hpi). After the NHN treatment, the expression level of PstCEP1 was greater (P Ͻ 0.05) at 194, 198, and 216 hpi than in the N samples, with a peak at 194 hpi (ca. 33.47-fold increase) (Fig. 3).
Overexpression of PstCEP1 suppresses PCD caused by BAX/INF1 in Nicotiana benthamiana. BAX is a cell death-promoting protein of the Bcl-2 family in mouse, and INF1 is a pathogen-associated molecular pattern (PAMP) from Phytophthora infestans (39)(40)(41). Both proteins can cause programmed cell death (PCD) in N. benthamiana similar to the plant defense-related hypersensitivity reaction (HR). We determined whether overexpressing PstCEP1 in N. benthamiana suppresses PCD caused by BAX/ INF1. PstCEP1 and BAX/INF1 were coexpressed in the N. benthamiana leaves, which were assessed 5 days after infiltration (Fig. 4A). Overexpressing PstCEP1 inhibited PCD caused by BAX/INF1, but the control did not (Fig. 4B).

Subcellular localization of the eGFP-PstCEP1 fusion protein.
To confirm the subcellular localization of PstCEP1 in host cells, we cloned the open reading frame (ORF) region of PstCEP1 into the pBINGFP vector. The control (pBINGFP) and pBINGFP-PstCEP1 vectors were then transformed into Agrobacterium tumefaciens strain GV3101 and transiently expressed in N. benthamiana leaves. Fluorescence of enhanced green fluorescent protein (eGFP) was evenly distributed throughout the tobacco cells, and the eGFP-PstCEP1 fusion protein might be located in the cytoplasm (Fig. 5A). The results of Western blotting indicated that both eGFP and eGFP-PstCEP1 fusion protein were successfully expressed in tobacco leaves (Fig. 5B).
Effect of transient silencing of PstCEP1 on Pst virulence in response to HTSP resistance. BSMV-mediated HIGS was performed to knock down PstCEP1. The 336-bp specific segment of PstCEP1 was inserted into BSMV:␥ vector and designated as BSMV:PstCEP1-as. All of the BSMV-inoculated plants of XY6 showed stripe mosaic symptoms, and BSMV:PDS-inoculated plants showed bleaching phenotype 12 days after BSMV inoculation (Fig. 6A). Rust lesions were observed on the inoculated leaves of XY6 14 days post-Pst inoculation (dpi). The relative expression level of PstCEP1 indicated that the silencing was successful, and the expression level of PstCEP1 was suppressed by 43 to 68% by HIGS (Fig. 6D). BSMV:PstCEP1-as reduced sporulation compared with BSMV:00 (control) leaves in both the N and NHN treatments ( Fig. 6B and C). The transcriptional expression level of PstCEP1 was upregulated after the NHN treatment for 12 h compared with those in nonsilenced leaves. Although the NHN treatment induced PstCEP1 expression in the silenced leaves, the expression level was much lower than in the nonsilenced leaves under the NHN treatment at 24, 48, and 120 h post-temperature treatment (hptt). The results of qRT-PCR were consistent with the observed phenotypes (Fig. 6E).
The number of haustorial mother cells and hyphal length were determined at 48 and 120 hpi ( Fig. 7A to D). There were no differences at 48 hpi between PstCEP1silenced plants and nonsilenced plants, but the differences (P Ͻ 0.05) were observed at 120 hpi ( Fig. 7E and F). In addition, the linear colony length was assessed at 0 and 24 hptt. Colony length in the PstCEP1-silenced leaves was shorter (P Ͻ 0.05) for the NHN treatment than for the N treatment, but there were no significant differences between the PstCEP1-silenced leaves and nonsilenced leaves (Fig. 7G). There were fewer (P Ͻ 0.05) uredinia in the PstCEP1-silenced leaves than in the nonsilenced leaves (Fig. 7H).

PstCEP1 responds to both the PTI/ETI and HTSP resistance by EtHAn-mediated overexpression in wheat.
To test whether PstCEP1 responds to the host defenserelated PAMP-triggered immunity (PTI)/effector-triggered immunity (ETI) and HTSP resistance of the hosts, we overexpressed PstCEP1 in wheat by bacterial TTSS of the Pseudomonas fluorescens effector-to-host (EtHAn) strain (23,(42)(43)(44). EtHAn is a non- pathogenic strain of wheat but can cause callose accumulation, which is a characteristic of plant PTI defense response (42). PstCEP1 was cloned into the effector detect vector 6 (pEDV6), which can release nonbacterial effectors to host cells based on its TTSS function, and then injected into the leaves of wheat cv. Mingxian169 (MX169) mediated by EtHAn (45). MgCl 2 was used as the blank control, EtHAn and pEDV6-dsRed as the negative control, and pEDV6-AvrRpt2 as the positive control. Inoculation of EtHAn, pEDV6-dsRed, and pEDV6-PstCEP1 did not result in phenotypic changes on the leaves of MX169, but pEDV6-AvrRpt2 did (Fig. 8A). EtHAn, pEDV6-dsRed, pEDV6-AvrRpt2, and pEDV6-PstCEP1 caused callose deposition, but the amount of callose deposition caused by pEDV6-PstCEP1 was smaller (P Ͻ 0.05) than pEDV6-dsRed ( Fig. 8B and C). pEDV6-AvrRpt2 could lead to greater callose deposition and cause obvious chlorosis, indicating that PstCEP1 and AvrRpt2 are not the same types of effectors. Furthermore, bacterial counts indicated that EtHAn grew normally in wheat leaves (Fig. 8D).
Additionally, we overexpressed PstCEP1 in XY6 to study its role in response to the HTSP resistance. The EtHAn-, pEDV6-dsRed-, and pEDV6-PstCEP1-inoculated plants of XY6 were inoculated with Pst and incubated at 15°C for the first 192 hpi before being transferred to 20°C for 24 h to activate the HTSP resistance (i.e., the NHN treatment). Samples at 24 hpi, 96 hpi, 216 hpi (24 hptt), and 240 hpi (48 hptt) were taken for histological assessment. Reactive oxygen species (ROS) was produced at 24 hpi, and the differences in ROS accumulation between pEDV6-dsRed and pEDV6-PstCEP1 were not significant. However, pEDV6-PstCEP1 suppressed ROS accumulation at 96 hpi compared with pEDV6-dsRed (P Ͻ 0.05). After the NHN treatment, the ROS of pEDV6-dsRed began to burst, but the ROS accumulation of pEDV6-PstCEP1 remained at the same level as before the NHN treatment. pEDV6-dsRed and pEDV6-PstCEP1 differed (P Ͻ 0.05) in ROS accumulation at 216 and 240 hpi (Fig. S7). In addition, overexpressing PstCEP1 increased the number of uredinia and caused more severe rust development compared with EtHAn and pEDV6-dsRed (P Ͻ 0.05) (Fig. 9). Since the HTSP resistance of XY6 was activated after the NHN treatment, the wheat leaves inoculated with EtHAn and pEDV6-dsRed showed HR and had fewer uredinia compared with inoculation with pEDV6-PstCEP1, indicating that PstCEP1 can respond to HTSP resistance and enhance Pst pathogenicity.

DISCUSSION
Wheat stripe rust occurs frequently in the eastern area of Northwest China where climatic conditions (higher temperature in spring) may have caused the evolution of wheat cultivars containing high-temperature resistance to Pst. XY6 is a cultivar with typical HTSP resistance and has maintained this resistance since the 1970s (46). HTSP resistance was expressed in both the seedling-plant and adult-plant periods of wheat growth when seedlings were exposed to 20°C only for 24 h during the initial Pst incubation stage (7,47). In the present study, a total of 25 DEGs of Pst in response to the HTSP resistance of XY6 were identified through transcriptomic analysis of the RNA-seq data. KEGG analysis found that the most significant enrichment was the antigen presentation pathway and thiamine metabolism, including several heat shock  proteins, N-myristoyl transferase, and thiazole biosynthetic enzyme. DEGs in these pathways may be involved in macromolecular substance synthesis and metabolism in response to the HTSP resistance in XY6 (48,49).
Recently, Pst began to gradually adapt to higher temperatures (5). However, the HTSP resistance in XY6 remains effective. Previous studies suggested that several TaWRKY transcription factors, protein kinases (TaXa21), and resistance proteins (TaRPM1) are involved in the HTSP resistance in XY6 (28,46,50,51). Transcriptomic analysis of the HTSP resistance in XY6 was also performed (38). However, the HTSP resistance is a large and complex network, and its molecular mechanism is still not completely clear. Thus, the most important reason why XY6 maintains the resistance to Pst is the complex nature of HTSP resistance, making it difficult for Pst to overcome. However, Pst attempts to overcome HTSP resistance, and the secretion of effector proteins may be one of the most possible ways. Effectors play important roles as self-binders, self-modifiers, inhibitors, or activators in the host plants (52). Therefore, we identified 34 secreted proteins of Pst according to the transcriptome analysis in the present study. An important candidate effector gene, PstCEP1, was found to be upregulated 18.68-fold in response to the host HTSP resistance. We measured the relative expression levels of PstCEP1 during the infection stages, knocked down PstCEP1 by BSMV-mediated HIGS, and overexpressed PstCEP1 in wheat by EtHAn-mediated TTSS. The transcriptional expression level of PstCEP1 was upregulated 2 hptt while the HTSP resistance of XY6 was activating. The HTSP resistance of XY6 was activated after NHN treatment; thus, the wheat leaves inoculated with EtHAn and pEDV6-dsRed showed HR and a small number of uredinia but resulted in a large number of uredinia by pEDV6-PstCEP1, indicating that PstCEP1 can respond to HTSP resistance and enhance the pathogenicity of Pst. These results show that PstCEP1 is highly expressed and involved in the host HTSP resistance process, and they provide significant insight into the pathogenesis of Pst. In the previous studies, PEC6, Pst_8713, PstGSRE1, and PSTha5a23 have demonstrated suppression of plant PTI and ETI as a type of effector proteins of Pst (22)(23)(24)(25). Sixty-nine of 91 haustorial secreted proteins of Pst have been proved to suppress PCD in tobacco (53). Moreover, it has been proved that effectors of bacteria, oomycetes, and fungi can suppress PTI-associated PCD and callose deposition, as well as ETI-associated ROS accumulation (22,26,54,55). Our results indicate that PstCEP1 shares a similar function with these effector proteins. PstCEP1 can respond to the HTSP resistance in XY6 and enhance the pathogenesis of Pst. However, the HTSP resistance in XY6 is non-race specific and durable (28,38). XY6 still maintained some level of resistance after overexpressing PstCEP1 compared with the N treatment. The results suggest that the molecular mechanism of the HTSP resistance in XY6 is complicated and not likely to be controlled by a major R gene.
HTAP resistance to Pst in wheat has been successfully used in the United States for many years (7). Several Yr genes have been characterized as conferring HTAP resistance, such as Yr18, Yr29, Yr36, Yr39, Yr48, Yr52, Yr59, Yr62, Yr78, and Yr79 and several quantitative trait loci (QTL) (56)(57)(58)(59)(60)(61)(62)(63)(64)(65)(66)(67). However, most of these resistance genes have not been verified as effective or widely used in breeding in China. Another type of resistance is called all-stage resistance, which is usually qualitative and controlled by a single R gene (68). Thus, all-stage resistance in wheat will be overcome in the process of coevolution with Pst. In contrast, HTSP and HTAP resistances are non-race specific and durable. HTAP resistance has been demonstrated to be a quantitative trait, which is often controlled by multiple genes (68). Similarly, our results suggest that the phenotype of the HTSP resistance in XY6 is more like a quantitative trait, because a single effector provides only partial virulence against the HTSP resistance in XY6. In addition, conditions for activating the HTSP resistance are exposure to 20°C only for 24 h at the seedling stage, which has the advantage of being easier to identify compared with HTAP resistance. One possible research area in the future is to conduct a cross between XY6 and another cultivar completely susceptible. Breeding wheat cultivars with HTSP resistance is of great importance for control of stripe rust. Our findings provide strong evidence for the understanding of the interaction mechanism between the host HTSP resistance and Pst pathogenicity.

Conclusions.
In the present study, we identified 25 DEGs of Pst in response to the HTSP resistance of wheat by RNA-seq. Functional annotation and classification found that these DEGs were related to membrane proteins, mRNA binding proteins, cell membrane transport, and synthesis of cell nitrogen compounds. In addition, we identified 34 secreted proteins, and the highest-expression gene, PstCEP1, was used for functional verification. The results show that PstCEP1 is a candidate effector, which has potential virulence function. Furthermore, PstCEP1 is involved in Pst response to the HTSP resistance of XY6. Overexpression of PstCEP1 in XY6 can improve the pathogenicity of Pst. However, the single effector provides only partial virulence and the HTSP resistance in XY6 is still effective. Breeding wheat cultivars with HTSP resistance is of great importance for control of stripe rust. The present study improves our understanding of the molecular mechanisms of the Pst-wheat system.

Plant materials, fungus, bacterial strains, and sample collection.
In the present study, wheat cv. Mingxian169 (MX169) with susceptibility and wheat cv. Xiaoyan6 (XY6) with HTSP resistance again Pst  Agricultural Science. The EtHAn strain was identified and provided by the College of Plant Protection, Northwest A&F University, Yangling, Shaanxi, China. Propagation and preservation of the materials were done in our lab. Tobacco plants were grown in crocks (10 ϫ 10 ϫ 10 cm 3 ) under 16-h light/8-h dark at 25°C. Wheat plants were also grown in crocks (10 ϫ 10 ϫ 10 cm 3 ) with planting distance 1.5 cm under rust-free conditions. For infection assay, urediniospores of CYR32 were suspended to a ratio of ϳ1:6 to 9 (vol/vol) in sterile distilled water. Two-leaf stage wheat seedlings were brush inoculated with the urediniospore suspension at the first leaf stage and kept in growth chambers (Percival E-30B; Perry, IA, USA) for 24 h in the dark at 10 Ϯ 1°C with relative humidity of 80% (50). To determine the optimal high temperature to induce wheat resistance to CYR32 (the selected temperatures were 20°C), three types of temperature treatments were used to test for resistance to Pst: (i) for normal temperature (N), Pstinoculated wheat plants were incubated at persistent normal temperature (15 Ϯ 1°C); (ii) for normalhigh-normal temperature (NHN), Pst-inoculated wheat plants were kept at 15 Ϯ 1°C until 8 dpi and then switched to high temperature (20 Ϯ 1°C) for 24 h; and (iii) for high temperature (H), wheat plants were incubated at constant high temperature (20 Ϯ 1°C) after Pst inoculation. The beginning of leaf sampling time (0 h) for RNA-seq started when seedlings were moved into high-temperature treatment from normal temperature treatment (8 dpi). Leaves of XY6 inoculated with CYR32 under different temperature treatments were sampled. The samples at 8 dpi (0 hptt) and 9 dpi (24 hptt) under different temperature treatments were used for RNA-seq. All samples were frozen immediately in liquid nitrogen and then stored at Ϫ80°C until use.
RNA-seq analysis. A total of 15 RNA samples including three biological replicates from the three temperature treatments were isolated using TRIzol reagent (Invitrogen, CA, USA) and then treated with DNase I (Thermo Fisher, MA, USA) at a concentration of 1 U/g. The quality and concentration of the RNA samples were detected using an Agilent 2100 Bioanalyzer (Agilent, Chandler, AZ, USA). After total RNA was extracted, mRNA was enriched using oligo(dT) beads (Epicentre). Then, the enriched mRNA was fragmented into short fragments using fragmentation buffer and transformed into cDNA with random primers by reverse transcription. Second-strand cDNA was synthesized by DNA polymerase I, RNase H, dNTP, and buffer. The cDNA fragments were purified with QIAquick PCR extraction kit and end repaired, poly(A) was added, and the fragments were ligated to Illumina sequencing adapters. The ligation products were size selected by agarose gel electrophoresis, PCR amplified, and sequenced using the Illumina HiSeq 2000 platform by Macrogen (Seoul, South Korea). Paired-end reads were checked and scored according to the Q30 level standard (average Q30 Ͼ 90%). To create a comprehensive transcriptome, data trimming and adapter clipping were performed using Trimmomatic (v 0.33) software (69). Reads obtained from the sequencing machines included raw reads containing adapters or low-quality bases which affected the following assembly and analysis. Thus, to get high-quality clean reads, reads were further filtered according to the following rules: (i) removing the adapters of reads, (ii) removing reads containing more than 10% of unknown nucleotides, and (iii) removing low-quality reads containing more than 50% of the low-quality bases. Then, all reads were assembled by mapping to the Pst-78 genome. The DEG analyses were evaluated using (v3.2.2) software EdgeR after adjusting the data for batch effects (batch as a factor). In order to find DEGs from Pst responding to the HTSP resistance in wheat, the difference of temperature, time, and individual plant needed to be eliminated. X 1 and X 2 represent 0-and 24-h samples under N treatment, respectively. X 3 and X 4 represent the 0-and 24-h samples under NHN treatment, respectively. X 5 and X 6 represent 0-and 24-h samples under H treatment (20°C), respectively. To reduce the effects of individual differences, X 1 and X 3 used the same samples. X represents gene expression in each sample. ␣ h (X 4 Ϫ X 3 ) represents the value of the gene expression differences between 24 h after high-temperature treatment. ␤ T , [(X 5 ϩ X 6 ) Ϫ (X 2 ϩ X 1 )]/2, indicates the value of the gene expression differences between the NHN and H treatments. ␥ , [(X 6 ϩ X 2 ) Ϫ (X 5 ϩ X 1 )]/2, represents the value of gene expression differences of individuals under the same time point. Thus, the differences between X 4 and X 6 samples are DEGs of Pst in response to the HTSP resistance of XY6 (see Fig. S8 in the supplemental material). Probability (P) values were adjusted for multiple comparisons using false discovery rate (FDR) ␣ Ͻ 0.05 (70). Additionally, DEG sets must have possessed a log 2 fold change (Ͼ2 or ϽϪ2) under NHN versus H treatment to be considered.
Functional annotation and enrichment analysis. For function annotation, transcripts were subjected to BLASTx (BLAST v2.2.28, E value Ͻ 1E Ϫ5 ) analysis against protein databases including Nr, Swiss-Prot, KEGG, and COG. After Nr annotation, the Blast2GO program was used to get GO annotation (71). To investigate the metabolic pathway of annotated transcripts, we aligned the transcripts to the KEGG database. GO terms and KEGG pathways with FDR-corrected P values of Ͻ0.05 were considered statistically significant.
Identification of effector candidates. In order to identify the effector candidates of Pst in response to the HTSP resistance in wheat, we extracted all the corresponding protein sequences from the transcripts. The conditions for predicting the effector candidates are as follows. (i) Perl Script was employed to screen candidate proteins with a length of 50 to 400 amino acids; (ii) SignalP v4.1, TargetP v1.1, and TMHMM v2.0 were used to screen candidate secretory proteins with signal peptides and without transmembrane domains, respectively; (iii) the predicted secretory proteins were analyzed by cysteine statistics (number of Cys Ͼ 3) and Pathogen-Host Interaction Database (PHI) BLAST (E value Ͻ 1e Ϫ5 ); (iv) the predicted secretory proteins belonging to DEGs in response to the HTSP resistance in wheat based on RNA-seq were identified as effector candidates.
Total RNA extraction and qRT-PCR analysis. Total RNA of wheat leaves inoculated with Pst race CYR32 at 0, 48,96,192,194,198,204,216,240,264, and 312 hpi was extracted using the SV Total RNA isolation system (Promega, Madison, WI, USA). The RNA concentration and quality were detected using Nanodrop 2000 and electrophoresis. First-strand cDNA was synthesized from 1 g total RNA using the PrimeScript RT reaction system (TaKaRa, Tokyo, Japan). PstCEP1 was cloned from the cDNA by TransStart KD Plus DNA polymerase (Transgene, Beijing, China). qRT-PCR was performed by iQ5 (Bio-Rad, Hercules, CA, USA) using UltraSYBR mixture (Cowin, Beijing, China) in a volume of 25 l consisting of 12.5 l mixture, 9.5 l ddH 2 O, 2 l primers, and 1 l cDNA. Housekeeping genes EF1 and ACT from Pst were used as internal reference genes. Data were collected from three independent biological replicates, each consisting of at least three reactions, and negative controls without templates were detected in case of contamination. The amplification efficiency (80 to 100%) of primers was determined (Fig. S2) by LinReg PCR (72). The expression ratio of each gene was calculated by using the relative expression software tool of REST (v2.0.13) (73).
Secretory function verification of the signal peptide of PstCEP1. The signal peptide of PstCEP1 was predicted by SignalP Ver 5.0 and validated using the yeast signal trap system (36). DNA fragments encoding the signal peptide were amplified using specific primers (Table S2) and cloned to pSUC2T7M13ORI (pSUC2) vector by ClonExpress II One Step cloning kit (Vazyme, Nanjing, China) (36). The pSUC2-PstCEP1 vector was transformed into the yeast strain YTK12 and screened on the CMD-W (lacking tryptophan) medium; only the YTK12 strain carrying a pSUC2 vector could grow on the CMD-W medium. Positive colonies were replica plated on the YPRAA (using raffinose as carbohydrate source) medium plates for invertase secretion. YTK12 transformed with pSUC2-Avr1b and the empty pSUC2 vector were used as positive and negative controls, respectively.

Identification of PstCEP1 suppression of BAX/INF1-induced programmed cell death based on A. tumefaciens infiltration assays.
The open reading frame (ORF) region of PstCEP1 was amplified using specific primers (Table S2) and cloned to vector PVX106 using the ClonExpress II One Step cloning kit (Vazyme, Nanjing, China). The recombination plasmids of PVX106-PstCEP1, PVX106-eGFP, BAX, and INF1 were each introduced to A. tumefaciens strain GV3101 by electroporation. The harvested A. tumefaciens cultures containing PstCEP1, eGFP, BAX, or INF1 were collected, washed with 10 mM MgCl 2 3 times, resuspended in an infiltration medium (10 mM MgCl 2 ) to an OD 600 of 0.6, and then incubated at 28°C in the dark for 2 h prior to infiltration. The A. tumefaciens suspension carrying PstCEP1 and eGFP was infiltrated into N. benthamiana leaves, and the A. tumefaciens suspension containing BAX/INF1 was infiltrated into the same site 24 h later. Phenotypical observation was performed 5 days after infiltration of BAX/INF1. The leaves were decolorized by heating in absolute ethanol. Each assay consisted of at least three leaves on three independent tobacco plants.
Subcellular localization of the eGFP-PstCEP1 fusion protein.
To in planta express the eGFP-PstCEP1 fusion protein, A. tumefaciens strain GV3101 was used to deliver transgenes into N. benthamiana leaves. The ORF region of PstCEP1 was cloned into the pBINGFP vector using the ClonExpress II One Step cloning kit (Vazyme, Nanjing, China) via BamHI (New England BioLabs, Hitchin, United Kingdom) digestion (Table S2). The recombination vector pBINGFP-PstCEP1 and the control vector pBINGFP were transformed into A. tumefaciens strain GV3101 (Weidibio, Shanghai, China) and cultured on the LB medium for 1 to 2 days, respectively. A. tumefaciens cells carrying pBINGFP and pBINGFP-PstCEP1 were collected, washed 3 times with 10 mM MgCl 2 , and resuspended in an infiltration medium (10 mM MgCl 2 , 10 mM morpholineethanesulfonic acid [MES], and 200 mM acetosyringone, pH 5.6) to an OD 600 of 0.4. The suspensions were infiltrated into 8-week-old N. benthamiana leaves. Samples were collected 3 to 4 days after infiltration, and microscopic observation was performed using an FV3000 microscope (Olympus, Tokyo, Japan). The total protein of N. benthamiana leaves was extracted using a plant protein extraction kit (Solarbio, Beijing, China) and stored at Ϫ80°C until use. Protein extracts were separated by 10% SDS-PAGE, and anti-GFP antibody (Beyotime, Shanghai, China) was used to detect eGFP.
Functional verification of PstCEP1. EtHAn TTSS-mediated overexpression of PstCEP1 was used to detect whether it responds to the PTI/ETI defense and the HTSP resistance of wheat. The ORF region encoding mature protein without the signal peptide of PstCEP1 was cloned to pEDV6 vector using Gateway cloning (entry vector pDONR 221; primers are listed in Table S2). The recombination vector pEDV6-PstCEP1, positive-control pEDV6-AvrRpt2, and negative-control pEDV6-dsRed were transformed into the EtHAn strain and cultured on KB medium (2% peptone, 1% glycerol, 0.15% K 2 HPO 4 , 0.15% MgSO 4 , and 2% agar) for 1 to 2 days. The harvested wild-type EtHAn culture and EtHAn cultures carrying pEDV6-PstCEP1, pEDV6-AvrRpt2, and pEDV6-dsRed were collected, washed with 10 mM MgCl 2 for 3 times, and resuspended in an infiltration medium (10 mM MgCl 2 ) to an OD 600 of 0.8. The wild-type EtHAn culture and resuspended EtHAn cultures carrying pEDV6-PstCEP1, pEDV6-AvrRpt2, and pEDV6-dsRed were infiltrated into the second leaves of wheat cv. MX169 using a syringe with the needle removed. Samples of each treatment were collected at 48 hpi, and phenotypes of wheat leaves were examined at 72 hpi. The callose deposition in leaf samples was observed using fluorescence microscopy after 0.05% aniline blue staining (75). The callose deposition levels were determined by counting the number of fluorescent spots per mm 2 with 30 random sites, all of which were derived from three biological replicates. Bacterial growth levels in XY6 were measured by cutting 50-mg tissues around the infiltrated point and homogenizing them in 200 ml of the inoculation buffer. The bacterial suspensions were diluted and plated on KB solid medium with 50 mg/ml chloramphenicol and gentamicin.
The wild-type EtHAn culture and EtHAn cultures carrying pEDV6-PstCEP1 or pEDV6-dsRed were infiltrated into the second leaves of XY6 to determine the role of PstCEP1 in response to HTSP resistance. The method of infiltration was the same as previously described, and the wheat plants were inoculated with Pst race CYR32 at 24 h after the infiltration of EtHAn. The inoculated wheat plants were maintained in a growth chamber at 15°C, transferred to 20°C at 192 hpi for 24 h, and then returned to 15°C. Samples for histological observation were collected at 24, 96, 216 (24 hptt), and 240 (48 hptt) hpi. Diaminobenzidine (DAB) coloration was used to detect hydrogen peroxide accumulation (76). H 2 O 2 accumulation per unit area in wheat leaves inoculated with CYR32 at 24, 96, 216, and 240 hpi after infiltration with pEDV6-dsRED or pEDV6-PstCEP1 was determined using a BX-51 microscope (Olympus, Tokyo, Japan). Disease phenotypes were observed, and urediniospore quantification was performed in the EtHAn wild-type, pEDV6-dsRed (control), and pEDV6-PstCEP1 inoculated wheat plants at 14 dpi.
Data analyses. Software programs SAS v8.01 (SAS Institute Inc., Cary, NC, USA) and SPSS v25.0 (IBM, Chicago, IL, USA) were used to analyze the experimental data. Duncan's test at P ϭ 0.05 was used for determining the differences in the related expression of PstCEP1 as well as histological statistics between treatments at each temporal point.
Data availability. The raw data used in the present study for transcriptome assembly and gene expression analysis have been submitted to the NCBI Sequence Read Archive (SRA) database under accession numbers SRR5580869, SRR5580870, SRR5580871, SRR5580872, SRR5580873, SRR5580874, SRR5580875, SRR5580876, SRR5580877, SRR5580878, SRR5580881, SRR5580882, SRR5580883, SRR5580884, and SRR5580886. The sequence of PstCEP1 has been submitted to GenBank under accession no. MN431201.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.