Expressed Sequence Tags from the oomycete Plasmopara halstedii, an obligate parasite of the sunflower

Expressed sequence tags analysis

After two rounds of subtraction hybridization, cDNAs were cloned into pGemT-easy vector and the bacteria arrayed in 96 well plates. To estimate the average size of the obtained clones, 40 clones were randomly chosen and their inserts were amplified using the SP6 and T7 universal primers present on the vector. The amplification products were then separated by agarose gel electrophoresis. The estimated sizes were between 400 and 800 bp with an average of about 500 bp. Subsequently, 602 clones were randomly chosen and single pass sequenced using the T7 universal primer. The length of good quality sequences was on average between 400 and 500. 51 sequences were of a poor quality or too short (<100 bases) thus were excluded from further analysis.

PCR amplification and origin of ESTs

Because the 230 unisequences could originate either from P. halstedii or represent induced genes in sunflower, PCR primers were designed to amplify these gene fragments from DNA isolated either from P. halstedii sporangia or from Helianthus annuus. All the 230 primer pairs tested amplified fragments either from P. halstedii DNA or from Helianthus annuus DNA; 145 primer pairs amplified fragments only from P. halstedii DNA, indicating that these ESTs were derived from P. halstedii genes that are expressed during infection in sunflower. Figure 1 shows an example of PCR amplifications from DNA isolated from P. halstedii sporangia or from Helianthus annuus.

Homology search

Functional classification of P. halstedii ESTs

ESTs were assigned to putative cellular roles using the categories defined by Bevan et al.[19] and the Expressed Gene Anatomy Database (EGAD) [20]. Two categories (elicitor and pathogenecity, and cell defence) are added as described by Kamoun et al.[21]. A summary of the assignment of non-redundant ESTs to functional categories as well as their relative abundance is listed in Table 3. The majority of the identified cDNAs were related to protein synthesis, cell metabolism, signal transduction, and cell stress.

Protein-signature scanning and identification of putative secreted proteins

Identification of pathogenicity-related genes

Functional annotation of the ESTs identified at least 4 that could be potentially involved in the infectious process of P. halstedii. One EST [GenBank:CB174657] showed homology with a Kazal-like serine protease inhibitor from Phytophthora infestans [24]. Alignment of these sequences (Figure 2) shows that the P. halstedii Kazal-like protein contains the conserved cysteine backbone and the motif C-X3-C-X7-C-X10-C-X6-C-X9-C defining the Kazal family signature [24].

A second EST [GenBank:CB174713] showed a strong homology with a Phytophthora infestans Cystatin-like protein [25]. The P. halstedii cystatin-like protein displays 42% identity with the P. infestans cystatin-like protein (Figure 3). Interestingly, as in the P. infestans protein, the P. halstedii protein contains a potential signal-peptide and conserved domains, including the N-terminal trunk (NT), first binding loop (L1) and second binding loop (L2) [25], suggesting that these proteins may have similar functions.

Comparison with true fungi and other microbes for conserved virulence factors

Comparison with other Oomycetes

Search for homologous sequences in the ESTs collections or whole genomes sequences of different oomycetes showed that 117 out of the 145 P. halstedii ESTs share similarity with sequences in at least one oomycete taxon with a BlastN E-value < E^-5. When an E-value cutoff < E^-20 is used, 82 P. halstedii ESTs still shared similarity with sequences in the other oomycetes. Neverthless, 11 P. halstedii showed no significant homology with any sequence in the other oomycetes or any other sequence in the public databases. Thus, these sequences are unique to P. halstedii and may represent genes specific to this pathogen. Unfortunately, these 11 ESTs are annotated as hypothetical and no obvious motif was detected within them using Interproscan and Smart programs.

Fifteen sequences were highly similar (BlastN E-value < E^-20) with sequences in at least 5 out of the 8 oomycetes taxa. These sequences included 3 ESTs coding for unknown proteins ([GenBank:CB174649], [GenBank:CB174661] and [GenBank:CB174701]) and 12 ESTs coding for differents proteins such as calmodulin [GenBank:CB174628], actin [GenBank:CB174582] or ribosomal proteins ([GenBank:CB174588] and [GenBank:CB174589]).

Redundancy

The 230 unisequences correspond to 153 clones present as singletons and 398 clones corresponding to redundant cDNA which formed 77 contigs ranging from 2 to 78 ESTs. This redundancy rate of 72% is higher than those obtained from other EST sequencing programs, for example, 49% of 1409 clone of N. crassa [28], 53% of 4809 clones from the cambial tissue of poplar [29], and 37% of 1000 clones from Phytophthora infestans [21]. This redundancy rate of 72% could be reduced by sequencing more clones after a differential screen with the most represented clones, elongation factor of P. halstedii and Asparagine synthase of Heliantus annuus which were represented in 78 and 56 copies respectively and account for up to a third of redundancy observed.

Origin of the EST

The 230 unisequences may correspond either to sunflower genes induced upon the infection by P. halstedii or sequences originating from the parasite. To overcome this difficulty, 230 primer pairs were designed and used to amplify the corresponding fragment with sunflower and P. halstedii DNA. This analysis resulted unambiguously in the identification of 145 EST originating from P. halstedii which corresponds approximately to 63% of the total of unisequences. The sequences of these primers are deposited along with the sequences of the ESTs in the Genbank. When the EST belongs to P. halstedii, amplification product was observed only when the DNA of P. halstedii extracted from infected sunflowers is used as template. Conversely, when the EST is originating from sunflower, a faint amplification product is often observed with DNA from P. halstedii (Figure 1). This is due to the contamination of sporangia with sunflower cells when being collected from infected cotyledons. Although this PCR strategy is robust, it is not suitable for high throughput sequencing project. Alternatively, in a similar work, Thara et al. [30] confirmed the fungal origin of several EST from the obligate basidiomycete Puccinia tritici by hybridization with radioactively labelled total genomic DNA. When the whole genome sequence of the host is available such as in the model plant A. thaliana, it can be exploited to distinguish between the EST originating from the host and those originating from the parasite. This strategy has been used by van der Biezen et al.[14] to identify 7 genes from the oomycete Peronospora parasitica. The G+C content of the EST also has been used to distinguish Phytophthora sojae from soybean cDNA [31]. The average G+C content of soybean EST was 46% whereas the G+C content of Phytophthora sojae was 58%, and plotting of the ESTs from infected soybean produced two distinct peaks of G+C percentage [31]. In the present study, the percentage G+C contents of ESTs from P. halstedii and from sunflower were entirely overlapping with averages of 47% and 45% respectively (data not shown), making this criterion inappropriate to uncover the origin of an EST in this pathosystem.

In contrast, all the sequences showing homology with sequences from other oomycetes such as Phytophthora species proved to be originating from P. halstedii. Therefore, the growing number of sequences produced in different oomycete EST sequencing projects and the availability of Phytophthora and H. parasitica genome sequences should facilitate the analysis of pathogen genes expressed during host-dependent stages.

Functional classification of P. halstedii ESTs

Many of the ESTs identified in this study are associated with basic metabolisms such as energy production, carbohydrates metabolism, nucleotides and protein synthesis. Protein synthesis process is highly represented which may indicate that this process is actively involved during infection. The most represented EST [GenBank:CB174619] shows significant similarity with the tef1 elongation factor from Phytophthora infestans [32], which is highly expressed during spore germination and mycelium formation [33]. Interestingly, this gene has affinity for actin and tubulin and may be involved in the regulation of the cytoskeleton [33].

The EST [GenBank:CB174624] shows homology with cyclophilin which has a cis-trans isomerase activity [34]. However, this gene has also been identified as a virulence factor in the rice blast fungus, Magnaporthe grisea [35]. It should be interesting to test whether this gene has conserved functions in different plant pathogen species as it was hypothesized by Thara et al. [30].

The homology found between one EST [GenBank:CB174646] and an elicitor from P. megasperma [36] indicates that the EST approach can be a rapid way to generate sequences potentially involved in the pathogenicity process. However, whether this gene has a similar function in P. halstedii has still to be experimentally demonstrated. Additionally, many ESTs share homology with stress-related genes such as superoxide dismutase [GenBank:CB174636] or gluthatione peroxidase [GenBank:CB174637] which may indicate that P. halstedii faces a hostile environment within the sunflower tissues and that such enzymes may detoxify compounds released by the host such as hydrogen peroxide. For example, in Mycobacterium tuberculosis, a superoxide dismutase enzyme contributes to the resistance of the parasite to the oxidative burst in macrophages [37].

Identification of P. halstedii genes potentially involved in the pathogenesis process

Two of the ESTs obtained share significant homology with protease inhibitors from P. infestans. The first one [GenBank:CB174657] contains a Kazal-like domain and is similar to the serine protease inhibitor EPI9 from P. infestans [24]. At least 35 Kazal-like Serine protease inhibitors have been reported in different oomycetes and two of these secreted proteins (EPI1 and EPI10) have been shown to interact and inhibit the apoplastic pathogenesis-related Protease P69B, a subtilisin-like serine protease of tomato [24, 38]. Interestingly, Catanzariti et al. [39] showed that the flax rust avirulence gene AvrP123-A encodes a Kazal-like protein that is recognized by P1 and P2 resistance genes in flax. The putative P. halstedii Kazal-like protein possesses all the conserved domains defining the Kazal-like family. The second P. halstedii protease inhibitor like protein [GenBank:CB174713] shares similarity with a P. infestans Cystatin-like protein [38]. Both sequences possess all the signatures sequences of the Cystatin-like protease inhibitors, including the N-terminal trunk, the first loop and the conserved Trp within the second loop. It is likely that these sequences similarities may reflect conserved physiological functions. Thus, it should be interesting to test experimentally whether theses proteins could similarly inhibit proteases of infected sunflowers.

Identification of potentially shared factors with true fungi

We exploited the recently developed PHI-base [40] a database that catalogues the phenotypes resulting from mutations in defined genes of both plant and animal pathogens [26]. Homology search using the P. halstedii ESTs identified 11 sequences with significant matches in this database. Mutation of the identified genes led to reduced virulence in the respective hosts. For instance, the P. halstedii [GenBank:CB174644] is highly similar to a thiol peroxidase (PHI:386) from the basidiomycetous fungus Cryptococcus neoformans. As peroxidases, this gene acts to remove peroxides and provide defence against oxidative damage. Mutation of this gene significantly reduced virulence in mice [41]. We have shown that resistance of sunflower to P. halstedii is associated with an oxidative-like burst within the hypocotyls [3]. Thus, it should be interesting to test whether the P. halstedii thiol peroxidase gene plays a similar role in detoxifying peroxides in sunflower. Sequence homology does not necessarily imply a conserved function, yet many animal and plant pathogens appear to utilize common signaling cascades and protective compounds during their development and pathogenesis [42].

P. halstedii isolate and culture conditions

One isolate identified as the physiological race 300 of P. halstedii on sunflower differentials was used. This isolate was provided by Dr D. Tourvieille (INRA-Clermont-Ferrand, France). It has been collected from infected fields in the south of France in 1995 and maintained by asexual reproduction on the sunflower genotype Peredovick, susceptible to all known races of P. halstedii in a containment culture chamber [45]. Frequently, sunflower differentials are infected to assess the behaviour of the isolate. The infection method of sunflower germinated seeds and growing conditions were those described by Mouzeyar et al.[2].

Preparation of mRNA and Suppression Subtraction Hybridization library construction

Total RNA was extracted from 15-day-old infected and non-infected sunflower plants using the method described by Bogorad et al.[46] and the polyadenylated mRNA with the PolyATract mRNA Isolation System (Promega France). Using the Clontech PCR-select™ cDNA Subtraction kit [16], second-strand cDNAs are prepared from the two mRNA populations under comparison. To enrich the cDNA library with clones specifically from P. halstedii, the cDNA from infected plants was used as tester and the cDNA from non infected plants was used as driver. After subtraction, the PCR-amplified cDNA were cloned into pGemT-easy vector and transformed into E. coli JM 109 strain (Promega France).

Sunflower and P. halstedii DNA extraction

Sunflower DNA was extracted from healthy young leaf tissues by Nucleon PhytoPure kit (Amersham™). To extract P. halstedii DNA, 15-day-old infected seedlings were placed in a plastic bag for 2 days to induce sporulation on cotyledons and leaves. Zoosporangia were collected by gently washing out the cotyledons and leaves with sterile water and DNA extracted using the Nucleon PhytoPure kit (Amersham™ France).

DNA sequencing

Clones from the subtracted library were selected randomly and grown in 96-well plates. Each EST was single-pass sequenced using the Dye-Terminator method and the T7 primer (Genome Express, France).

Sequence analysis

Sequences with low quality bases (Phred score less than 20) and short sequences (< 100 nucleotides) were removed from further analysis. The vector and polylinker sequences were manually trimmed. Sequences were then assembled and arranged into unisequences using the BlastN algorithm [17] and CAP3 program [18]. Similarity searches were done using the BlastX program [17] against current version of NCBI "nr" non-redundant amino acid database.

Comparison with true fungi and other microbes for conserved virulence factors

The PHI-base containing the description of curated genes involved in pathogenicity both in animals and in plants were searched for homology with the P. halstedii ESTs [26]. The version 2.31 containing 592 proteins was downloaded and used in BlastX search using the BioEdit package v7.0.8 [47]. The hits with an E-value < E^-5 were considered as significant.

Primers design and PCR amplification

Primer pairs were designed using the primer3 program [48] and used to amplify each EST from DNA isolated from P. halstedii or from DNA isolated from sunflower line. The PCR amplifications were carried out with 50 ng DNA in the presence of 0.2 mM of each dNTP, 1 U of Taq DNA polymerase (Advantage 2, Clontech), 1 × Taq polymerase buffer and 0.5 μM of each primer. PCR was carried out in a 9600 Perkin-Elmer thermocycler under the following conditions: 35 cycles of 10 s at 94°C (denaturation), 30 s at 60°C (primer annealing), and 1 min 30 s at 72°C (primer extension). PCR products were separated using standard TAE agarose gel electrophoresis.