Transcriptomics is a powerful approach for the global analysis of plant-pathogen interactions. Using the cDNA-AFLP technique, we observed widespread modulation of transcriptional activity, with 4.6% of all transcripts showing some form of differential expression. The gene expression patterns revealed by the cDNA-AFLP and qRT-PCR analyses were largely consistent with the physiological and biochemistry changes corresponding to the Pst infection events in the wheat leaf tissue.
About 73% (187) of the 255 differentially expressed genes in wheat were up-regulated during the infection process. Most of these genes peaked at 12~24 hpi, possibly reflecting the exploitation of cellular resources and/or the activation of defense responses [36, 31]. The up-regulated genes were similar to that of a recent study reported by Coram et al. . They reported that 64 genes specifically involved in the incompatible reaction between the Yr5 single gene line with a US Pst pathotype, PST-78, and these genes were up-regulated and peaked at 12-24 hpi. In these study, we identified 94 genes preferably induced in the incompatible interaction, also around 12-24 hpi. The most of the genes identified in both studies were characterized in the same functional categories. In contrast, Coram et al. identified only one gene down-regulated in the incompatible interaction . Comparatively, 68 TDFs of 255 differentially expressed genes showed down-regulation in the present study. Such difference might be due to the fact that the probes of Affymetrix GeneChip were designed based on known wheat genes. Meantime, the transcriptional profiling obtained by cDNA-AFLP technique largely covered overall wheat transcriptome. Moreover, different genotypes of wheat or Pst pathotype, as well as different temperatures used in the tests, might also contribute to the difference. Given that genetic manipulation for Pst is unavailable, along with unstable wheat transformation system, the putative functions of a large number of genes identified in this study have only been predicted by bioinformatical approaches combined with altered expression patterns. Of the 255 sequenced TDFs, 113 had relatively clear functions in various categories when searching the non-redundant protein database. Thus, these genes can be valuable resources for understanding molecular changes in the incompatible interaction.
A fascinating discovery in this study is the quenching of divergent expression of Pst-regulated genes in both incompatible and compatible interactions in the middle stages of Pst infection. Similar to the results of our previous study , the expression of nearly all wheat genes that were differentially regulated at the early time frame returned to the levels of the mock-inoculted plants by 48 hpi. The low level of expression remained up to 120 hpi. The lack of differential gene expression at the period from 72 to 120 hpi could be because Pst might initiatively inhibit the early host responses in both incompatible and compatible interactions. Haustorium-forming fungi and oomycetes secrete many proteins from the haustoria into the extrahaustorial matrix during the parasitic stage of host infection, subsequently, a subset of proteins are further transported into the host cell . Presumably, to establish infection, these proteins enable the pathogens to obtain nutrients, or to evade or manipulate host defenses [37, 38]. It is thought that the oomycete Phytophthora also forms haustoria and secretes molecular signals that functions in the plant apoplast and cytoplasm to reprogram molecular host defenses . The response of resistant plants at 48-72 hpi suggests that an avirulence gene is recognized by a resistance gene prior to this time, which appears to lead to a depression of host defenses. Our results may reflect that the resistance gene can be recognized by avirulence gene prior to 48 hpi in the interaction of wheat-Pst. Similar to our results, a study of the soybean response to Asian soybean rust (ASR) controlled by the Rpp2 gene also found many ASR-regulated genes responded early during the infection (6 to 36 hpi), followed by a period (24 to 72 hpi) in which expression levels returned to the mock levels and a new differentiation in gene expression during late infection (72 to 168 hpi) .
The development of Pst on resistant and susceptible wheat cv. Suwon 11 was found to be similar in urediniospore germination, appressorium formation, and penetration (foundation of substomatal vesicle, infection hypha, haustorial mother cell, and haustorium initial). However, after penetration (24 hpi), distinct differences in fungal spread between the compatible and incompatible interaction could be observed. In the compatible interaction, hyphae of CYR31 rapidly colonized host tissues intercellularly and numerous haustoria in the adjacent host cells were formed. In the incompatible interaction, the host cells showed hypersensitive cell death and the density of the intercellular hyphae and the number of haustoria were greatly reduced compared to the compatible interaction . These results suggested that resistance to stripe rust in wheat cv. Suwon11 is executed after penetration has occurred. Yet, for host response to Pst, H2O2 accumulation was detected in host guard cells as early as 6-8 hpi , therefore, the perception of the Pst fungus by wheat and the ability of the pathogen to avoid or overcome the host's defense imply a complex, dynamic network of communication, a series of signal events should be operated before the resistance gene is expressed. Relative specifically expression of the 94 genes in the incompatible interaction and their diverse putative functions in various metabolisms support the hypothesis. However, how and when the signal is perceived by the host and transduced is still poorly understood. In this study, we focused on genes that accumulated preferentially in the incompatible reaction before 48 hpi. Dissection of these genes and their involved biochemical pathways in the future studies might provide answers to the questions.
Because the regulation of gene expression is a dynamic process, the expression profiling was presented over a time course by cDNA-AFLP, which allowed us to study the dynamic behavior of gene expression and characterize their changes over time. The induction and signal transduction of defense responses specific to the interaction require up- or down- regulation of many genes. We were primarily interested in genes whose expression might be used to distinguish incompatible from compatible interactions in wheat. A different analysis was conducted to achieve the goal by comparing gene expressions in Suwon 11 challenged with CYR23 (avirulent) or CYR31 (virulent). The comparison analysis of 255 TDFs in the incompatible interaction with those in the compatible interaction as previously described  showed that of the 255 transcripts induced during the incompatible interaction, 161 TDFs (63%) were also induced during the compatible interaction, and thus were classified as basal defense-related. 94 TDFs were expressed preferably in the incompatible interaction. The large proportion of TDFs were shared in both interactions, these results were similar to the reports of Coram et al.  with the same pathosystem. Coram et al.  reported 51 genes commonly induced in both incompatible and compatible interactions between wheat and Pst. Tao et al.  also found that plant responses in compatible and incompatible interactions are qualitatively similar but quantitatively different soon after infection. Another study of the barley response to powdery mildew controlled by the Mla6, Mla13 and Mla1 single resistance genes also provided evidence for a shared response between compatible and incompatible interactions up to the point of pathogen penetration .
Of our special interest is that 11 of 94 TDFs were up-regulated in the incompatible interaction, but did not change or were repressed in compatible interaction through the qRT-PCR validation. Of the 11 TDFs, 6 (S11_CY23_193-4, S11_CY23_128-2, S11_CY23_181-5, S11_CY23_Contig73, S11_CY23_Contig105 and S11_CY23_12-1-3) have unknown functions and 5 (S11_CY23_397p-3, S11_CY23_469-1, S11_CY23_351-6, S11_CY23_Contig99 and S11_CY23_148-5) encode Leucine Rich Repeat family protein, CBL-interacting protein kinase, Serine/threonine Kinase, ethylene-responsive RNA helicase and protein phosphatase type 2C, respectively. Protein phosphatase type 2C is a negative regulator of ABA responses. Gosti et al.  reported that suppressor mutants were more sensitive to applied ABA than the wild type and displayed increases in seed dormancy, whole-plant drought tolerance, and drought rhizogenesis intensity. However, ABA is required for plant defense. Adie et al.  measured ABA hormone levels in wild-type and JA/ET/SA/ABA-related mutants after Pythium irregulare infection to determine whether ABA is required for overall plant resistance. They found that ABA mutants showed an increased susceptibility to P. irregulare compared with the wild-type background, indicating that ABA is a positive signal involved in the activation of effective defenses against this pathogen. However, several reports showed that ABA increases susceptibility by counteracting SA-dependent defenses, and ABA-dependent priming of callose biosynthesis promotes enhanced resistance to some pathogens . These results supported that ABA should have a negative effect on resistance. Our results also indicated that ABA should be expected to play a negative role in response to Pst. DEAD-box RNA helicases had been reported to play an important role during development and stress responses in various organisms [46–48]. Rice OsBIRH1 encoding DEAD box RNA helicase was shown to function in defense responses against pathogen and oxidative stresses . STRS1 (Stress response suppressor 1) and STRS2 encoding DEAD box RNA helicases were shown to function as negative regulators of ABA-dependent and ABA-independent signaling networks . Zegzouti  isolated an Ethylene-responsive 68 (ER68, corresponding to Arabidopsis thaliana RNA helicase 20), their results indicated the potential for ER68 RNA helicase activity to be involved with ethylene-regulated gene expression at either the transcriptional or post-transcriptional level. Similar to their results, S11_CY23_contig99 encoding ethylene-responsive RNA helicase, was only induced in incompatible interaction, which should attribute to resistance to Pst, however, the precise role played by S11_CY23_contig99 still needs to be further elucidated.
Caffeoyl-CoA O-methyltransferases (CCoAOMTs) is an important enzyme that participates in lignin biosynthesis especially in the formation of cell wall ferulic esters of plants. CCoAOMT was proposed to play a pivotal role in cell wall reinforcement during the induced disease resistance response. Lignin is often deposited at the sites of wounding or pathogen invasion, which may provide a physical barrier for protection of adjacent tissues from further damage. In the previous study , immunogold localization of lignin revealed a markedly higher labeling density in host cell walls of the infected wheat leaves of the resistant cultivar than in cell walls of the infected wheat leaves of the susceptible cultivar. In this study, S11_CY23_360-3 encoding CCoAOMT, induced at 24 hpi, and peaked at 48 hpi in the incompatible interaction, the transcripts accumulations occurred after the resistance gene was triggered, which indicated that lignification appears to be also an active resistance mechanism in the wheat-Pst panthosystem.
Suwon 11 showed HR to CY23 infection. Plant cells involved in the HR generate an oxidative burst by producing reactive oxygen species (ROS), superoxide anions, hydrogen peroxide, hydroxyl radicals, and nitrous oxide . Peroxidases were thought to play an important role in ROS production [51, 52]. In the present study, two genes (S11_CY23_contig32 and S11_CY23_ contig46) were predicted to encode peroxidase and peroxisomal membrane protein, respectively. Their transcripts peaked as early as 12 and 24 hpi, respectively. Our gene expression data were coincident with the previous report of a rapid increase of O2
-- and H2O2 at infection sites and a strong accumulation of H2O2 in mesophyll cells from 12~24 hpi using histochemical staining in the Suwon 11 leaves inoculated by CYR23 . In contrast, O2
- and H2O2 could not be detected in most of the infection sites in the compatible interaction between Suwon 11 and CYR31. In the present study, we found that the expression of these two genes were much less in the compatible interaction than in the incompatible interaction.
Several studies have provided evidence supporting that the PR-5 protein plays an active role in resistance mechanisms in cereals [53–56]. Transcripts of four PR protein genes were analyzed during Fusarium graminearum infection, with PR-5 transcript accumulated as early as 6 to 12 hpi and peaked at 48 hpi . We also found S11_CY23_112 homologous to a wheat PR-5-like protein gene. S11_CY23_112 transcript accumulated strongly at 24-48 hpi. A stronger induction of this gene could be observed in the incompatible interaction than the compatible interaction, suggesting a general role of this protein in wheat resistance to stripe rust. Similarly, gene S11_CY23_274-5 from wheat was deduced to encode a receptor related to antifungal PR proteins. The predicted protein contained an extracellular domain related to the PR 5 protein, a central transmembrane spanning domain, and an intracellular protein serine/threonine kinase. Wang et al. isolated a PR5K gene from Arabidopsis thaliana and found that PR5K transcript accumulated at low levels in all tissues examined . They suggested a possible interaction of PR5K with common or related microbial targets. Nevertheless, the interrelation of PR5K and PR5 protein during the interaction between wheat and Pst need further studies.
Protein kinase is known to play a central role in signaling during pathogen recognition and the subsequent activation of plant defense mechanisms . We identified six TDFs encoding different protein kinases. S11_CY23_contig90 was highly homologous to the Arabidopsis MKP1 gene, which was showed to be induced at the transcriptional level during the interaction of wheat-Pst. The Arabidopsis genome contains 20 genes encoding mitogen-activated protein kinases (MAPKs), which interact with MKP1 . Using expression profiling, a specific group of genes that probably represent targets of MKP1 regulation was also identified . Surprisingly, the identity of these genes and interacting MAPKs suggested involvement of MKP1 in salt stress responses. Indeed, mkp1 plants have increased resistance to salinity . Accordingly, the gene S11_CY23_contig90 may play a role in the integration and fine-tuning of plant responses to stripe rust pathogen challenging.
This study uncovered a number of new candidate genes possibly involved in the interactions of wheat and Pst. More than 42% of the sequenced TDFs had no homologous sequences in the EST databases. De Torres et al.  reported that the plant response to pathogen challenge evoked a large number of transcriptomic components not yet present in EST libraries. Over 60% of the differentially regulated transcripts in their cDNA-AFLP were absent from standard 8,200 feature Affymetrix Gene Chips . Therefore, cDNA-AFLP analysis is a suitable tool for discovering new potential genes that are differentially expressed during the wheat-Pst interactions. Because most of the molecular mechanisms involved in the pathosystem interactions are yet to be determined, the large number of TDFs identified in this study will serve as candidates for further studies to determine their functions and dissect the molecular networks involved in the plant-pathogen interactions.