Lack of sexual reproduction and transformation system for the stripe rust pathogen has limited the use of most genetic and molecular techniques in studying genes of the pathogen involved in the interactions with its host plants. Interactions between the pathogen and its host plants have been studied almost restrictively by virulence analysis [1
]. Genes of the pathogen had not been reported until recently. In 2007, Ling et al. [21
] constructed a full-length cDNA library and reported the first group of 51 genes with putative functions for the stripe rust pathogen. Similarly, Zhang et al. [22
] identified 267 genes with putative functions from a germinated urediniospore EST library. These studies did not address expression changes of the rust genes. Using the microarray technique, Coram et al. [23
] identified genes involved in basal defense and different types of wheat resistance to stripe rust. However, the studies focused on plant genes using the wheat GeneChips and did not determine the pathogen genes involved in the compatible and incompatible interactions of the host and parasite. The present study was the first to determine expression of genes from both wheat and the stripe rust pathogen involved in a compatible interaction.
Using the cDNA-AFLP technique, we were able to detect 54,912 reproducible TDFs with 64 primer combinations. The number is within the same range as found in a previous study of the tomato-Cladosporium fulvum
pathosystem, in which 50,000 TDFs were detected with all possible 1,024 primer combinations [15
]. The fewer primer combinations producing more TDFs in the present study could be due to the larger genome size of wheat and the stripe rust fungus than those of tomato and C. fulvum
. Nevertheless, the present study demonstrated that cDNA-AFLP is a powerful technique to study genes involved in the wheat-P. striiformis
system. Because AFLP primers are universal, the cDNA-AFLP technique is applicable to any organisms or any host-pathogen systems for comprehensive transcript profiling [26
]. In fact, the technique has been successfully employed to identify expressed genes in various plant-pathogen systems [15
]. In comparison with microarray, cDNA-AFLP has low cost and does not require sequence information and expensive or sophisticated equipment. As showed in this study, cDNA-AFLP banding patterns are highly reproducible when compared with other techniques like differential display [29
The time points used in the present study were selected based on our detailed microscopic study of the infection process (as previously described). Interestingly, 48% and 35% of the differentially expressed genes showed different degrees of change between 6–24 and 120–168 hpi, respectively. In contrast, only 17% of genes were differentially expressed at 48–96 hpi. The expression changes of these genes corresponded quite well to the different infection stages of the pathogen. Compared with the early stages of infection, the proportion of cDNA derived from the fungus increased slightly and that from the plant decreased at the late stages, as the fungal biomass increased. These results were similar to the expression patterns in compatible interaction between wheat and P. triticina
] and were also supported by our histological study of compatible interaction between wheat and P. striiformis
(data not shown). Functions could not be determined for many of the TDFs due to the fact that fungal genes were more limited in the database than plant genes. Therefore, the genes of no hit are more likely from the stripe rust fungus.
In the present study, we identified 89 genes which were up-regulated and 97 down-regulated in a compatible interaction. The up-regulated genes were remarkably similar to that of a recent study by Coram et al. [25
]. They reported that 73 genes were up-regulated in a compatible interaction between wheat (cv. Avocet S) and a US wheat stripe rust pathotype (PST-78), which belonged to the major categories of functions as described in the present study. The expression of these genes also peaked at 24 hpi. In contrast, Coram et al. [25
] only identified two genes down-regulated for the compatible interaction. Such difference might be due to the fact that the Affymetrix GeneChip does not include all possible wheat genes. The cDNA-AFLP technique does not have such limitation. Different wheat genotypes and stripe rust pathotypes also might contribute to the difference.
Ideally, the mock treatment should be used for all time-points in the cDNA-AFLP experiments. However, the increase of cost and time for such experiments could be substantial as we used a large number of time points to represent all key time points of stripe rust infection. To reduce the experimental cost, only the 0-hpi treatment was included in the cDNA-AFLP study. Thus, not all transcript changes of the 186 genes can be attributed to regulations by stripe rust infection. Other factors such as plant growth could affect expression of some of the genes. However, we obtained a 96.6% confirmation rate by testing 29 genes in experiments including the mock treatment for every time point using the qRT-PCR technique. With such a high percentage of genes showing consistent expression patterns in both cDNA-AFLP and qRT-PCR experiments, it is conceivable to conclude that the most of the genes identified using cDNA-AFLP are involved in the compatible interaction between the host and pathogen, although further studies are needed to determine their roles in the interaction.
In a plant-pathogen pathosystem, differentially expressed genes from either the plant or pathogen may interplay in the interaction. In this study, we determined the origins of 18 genes that had high homologies to genes of the stem rust pathogen. Interestingly, 50% of the tested genes were from the stripe rust pathogen and another 50% from the wheat host. The origin for two of the genes amplified in both wheat and the pathogen were determined from wheat by sequencing the PCR fragments. Similarly, some genes identified from the plant based on the BLAST results might be from the pathogen. This suggests that BLAST analysis should be used only as an indication of potential origins, but definite determination of gene origins from infected leaves should be from PCR amplifications of genomic DNAs from the pathogen and host separately, and sequencing comparisons become necessary if a gene is amplified from both organisms. Thus, further studies are needed to determine origins of most other genes identified in this study.
Because systems for transformation and expression of stripe rust genes have not been developed and the number of putative wheat genes was large, we used bioinformatical approach to determine putative functions of genes with altered expression patterns. Of 186 sequenced TDFs, 74 had relatively clear functions in various categories. These genes can help us in understanding molecular changes in the compatible interaction.
Clone PST_C38 was predicted to encode wheat ATPase. Its expression was generally down-regulated upon infection. ATPase was suggested to control stomatal apertures in guard cells [31
]. The reduced transcript level of this gene in wheat leaves might indicate an attempt of the host to limit water loss by reducing stomatal openings. Clone PST_C71 was identified as a putative expansin and its transcripts accumulated rapidly during the rust infection. Expansin is a family of closely-related non-enzymatic proteins found in the plant cell wall with important roles in plant growth and response of plants to an adverse environment [33
]. Several genes including ribulose bisphosphate carboxylase (PST_63C-3), phosphoribulokinase (PST_C24) and carbonic anhydrase (PST_C50) encoding enzymes in the Calvin cycle were down-regulated during the rust infection. Chlorophyll a-b binding proteins (PST_C53) associated with photosynthesis-related functions were strongly down-regulated during the early stage of the infection. Transcriptional down-regulation of photosynthesis-related genes has been reported for compatible interactions between soybean and Phytophthora sojae
] and grapevine and powdery mildew [35
]. Plants infected by biotrophic fungal pathogens, such as rusts and powdery mildews, reduce their photosynthetic rates, possibly as a result of increased invertase activity that causes carbohydrate accumulation [7
A significant outcome of this study was the identification of ten TDFs with signal transduction functions, including serine/threonine kinase (PST_143-7), receptor kinase (PST_576-6, PST_298-3 and PST_274-5), phosphatase (PST_C16) and G-protein coupled receptor (PST_106B-4). The majority of these genes were up-regulated, indicating that they are essential in plant defense as previously reported [25
]. Several genes encoding enzymes were up-regulated in infected leaves, including (1,3;1,4)-beta glucanase (PST_C44), peroxidase (PST_C6) and quinone reductase (PST_68B-1). This observation was similar to cases for salt-stress induced hydrophobic peptide (PST_C88), protease inhibitor (PST_266-3) and subtilisin-chymotrypsin inhibitor (PST_266-4) [37
]. These genes have been assigned a function related to resistance in many pathosystems. This indicates the presence of a general, although weak, defense response in susceptible plants [25
]. However, this was not the case for a disease resistance gene (PST_267-4), encoding nucleotide-binding site leucine-rich repeat (NBS-LRR) protein. The expression of this gene was down-regulated in the compatible interaction between wheat and the stripe rust fungus, which was in agreement with recently developed models [41
Compared to TDFs of wheat origin as discussed above, relatively few genes were identified from the stripe rust pathogen, which could be due to higher abundance of wheat mRNA than fungal mRNA in the infected leaves. We clearly identified nine genes from P. striiformis
f. sp. tritici
. Because expressions of these genes were altered during the infection, they should be essential for the fungus to establish a compatible interaction with wheat plants. For example, ATPase was reported to reduce transpiration rates and stomatal apertures in rust fungus infections before uredia were fully developed [42
]. The pathogen gene, TDF PST_C86, has homology to a lipase in Aspergillus fumigatus
. Although little is known about the influence of lipases during infection of plants by fungi, intracellular lipases have been shown to be involved in appressorium formation of Magnaporthe grisea
]. The first evidence of involvement of secreted lipases in plant infection came from Comménil et al. [44
]. Eddine et al. [45
] reported a secreted lipase in the pea pathogen Nectria haematococca
. Voigt et al. [46
] found that transformation-mediated disruption of FGL1
encoding a secreted Fusarium graminearum
lipase led to reduced extracellular lipolytic activity in culture and reduced virulence to both wheat and maize. Because the lipase-like PST_C86 gene is from the stripe rust pathogen, its elevated expression during the infection process observed in this study suggests that the gene is important for the pathogen infection. More studies are needed to test the hypothesis and to determine the functions of the rust genes in the pathogen growth and interaction with wheat plants.
The present study was focused on a compatible interaction between the wheat and P. striiformis
pathosystem to identify genes essential for establishing disease. A parallel study was undertaken to focus on incompatible interaction using the same wheat genotype and a pathotype of the pathogen avirulent to YrSu
to identify genes essential for resistance and avirulence. As transformation and gene expression systems are currently under development in several laboratories, genes detected in the present study will be used in functional analysis for identifying genes involved in hypersensitive response, as done by Gabriël et al. [15
] with the tomato-C. fulvum