A 5' oligo capping method was used to construct an EST library of genes enriched in those upregulated on drying in A. avenae: an initial screen of 984 clones by reverse Northern dot blots, using cDNA from both dried and control nematodes in parallel experiments, yielded 88 ESTs which appeared to be induced by desiccation. From these, a subset of 30 ESTs showing significant BLAST matches to sequences from other organisms, and 12 ESTs representing apparently novel sequences, were selected for further study (Table 1).

The ESTs include a novel homologue of known Group 3 LEA protein genes: this cDNA sequence (accession number EF026241) is full length, as indicated by the presence of a spliced leader sequence (SL1c) at its 5' end and a polyA tail at the 3' end. The cDNA is 405 bp long, excluding the polyA tail, and its longest open reading frame predicts a protein of 85 amino acids with Mr 9990 and pI 6.4. The closest BLASTX match is with lea-1 of Caenorhabditis elegans and, although the BLAST score is not highly significant (0.001), other features of the predicted protein sequence identify it as a Group 3 LEA protein. For example, the predicted protein sequence has a section of K11 periodicity, with two interlaced runs of three lysines at residues 38, 40, 49, 51, 60 and 62; POPP analysis 35 places it cleanly within the Group 3 LEA protein cluster; it lacks cysteine and tryptophan residues; it is hydrophilic along its entire length according to a hydropathy plot based on the Eisenberg Consensus scale, with a grand average hydropathy (GRAVY) score of -1.83 indicating extreme hydrophilicity (the average score for Group 3 LEA proteins is -0.97; 36 ). Predictors of protein disorder suggest the protein is natively unfolded: the VL-XT algorithm at pondr.com indicates 95% disorder, while Foldindex and IUPred both return predictions of 100% disorder; and a Uversky plot of mean net charge against mean scaled hydropathy places the protein firmly among the set of intrinsically disordered proteins (Figure 1). All the above properties are consistent with this example being a new Group 3 LEA protein and therefore we designate it AavLEA2 (as the second such protein from A. avenae after AavLEA1; 24 ) and its gene Aav-lea-2.

A second example from the EST dataset, EF026240, also encodes a hydrophilic protein but it is less clearly identifiable as an LEA protein. The cDNA is again full length, with spliced leader (SL1a) and polyA tail, whose 464 bp insert encodes a protein of 102 amino acids in its longest open reading frame, with Mr 11656 and pI 6.53. The protein has a GRAVY score of -1.38, and a hydropathy plot shows it to be hydrophilic along almost its entire length. Disorder prediction algorithms are consistent with it being unfolded - VL-XT, Foldindex and IUPred predict 78%, 100% and 100% disorder respectively - and the Uversky plot positions the protein within natively unfolded space (Figure 1). POPP analysis shows clustering with Group 2 and Group 3 LEA proteins, but only at the level of band 3, indicating that this association is rather weak. Furthermore, there is no K11, and limited A11, periodicity and the predicted protein sequence includes one cysteine and one tryptophan, which it is unusual to find in Group 3 LEA proteins. A BLASTX run against the C. elegans genome returns lea-1 as the top hit, although the score is weak (0.002); a BLAST enquiry against the whole sequence collection returns as best hit (0.029) an LEA-like sequence from the Antarctic bacterium Polaromonas sp. JS666. Intriguingly, though, using BLAST to compare the EF026240 protein against AavLEA2, described above, gives a score of 5E-14, indicating a match of respectable significance. When aligned, the two predicted protein sequences are given as 34.3% identical with the ALIGN algorithm. In summary, the bioinformatics are consistent with this example being a hydrophilic, intrinsically disordered protein which probably has a distant evolutionary relationship to the Group 3 LEA proteins. For the moment, therefore, we consider it to be an LEA-like hypothetical protein.

While the presence of LEA and LEA-like protein gene sequences among the dehydration-induced EST dataset is in line with a presumed role for these polypeptides in the response to water stress, several other genes associated with stress responses were also found. For example, genes encoding: myo-inositol-1-phosphate synthase which controls synthesis of an osmolyte, myo-inositol, implicated in desiccation tolerance 37 38 ; AHA-1, a stress-regulated activator of the ATPase activity of molecular chaperone Hsp90 39 ; and glutathione peroxidase, which catalyses the reduction of hydroxyperoxides by glutathione and is important in the response to oxidative stress (reviewed in 40 ).

The control of the response to desiccation is very poorly understood and therefore it is of interest that a number of genes involved in signal transduction and the regulation of gene activity were recovered from the screen. For example, three zinc finger protein genes were identified (Table 1), encoding a C2H2 zinc finger family member (GR463901), a GATA-type zinc finger (GR463910), and a RING finger domain protein (DR121009). The latter is most likely involved in the regulation of protein activity, rather than gene regulation directly; it gives a best BLAST hit with mnat-1, a C. elegans gene whose protein is probably involved in the ubiquitination pathway as an E3-type ubiquitin ligase. It is also implicated in stabilising the CDK-activating kinase intrinsic to cell cycle progression through its MAT1 domain and is functionally associated with TFII, the transcription/DNA repair factor (http://www.wormbase.org; 41 ). The zinc finger proteins represented by GR463901 and GR463910 would merit further investigation as candidate anhydrobiotic gene set regulators: several transcription factors were found to be upregulated by water loss in Steinernema feltiae 8 , S. carpocapsae 9 and Plectus murrayi 25 , and a GATA-type transcription factor (ELT-2) has been found to regulate C. elegans transcription during the innate immune response and osmotic stress 42 ; in plants, many gene regulators have been characterised in the response to drought (e.g. reviewed in 43 ). Another gene, represented by GR463906, encodes a dual specificity phosphatase able to hydrolyse phosphate from Tyr and Ser/Thr residues; such phosphatases are key players in signal transduction, for example, in MAPK pathways, and in cell cycle regulation 44 . DR121003 encodes a homologue of C. elegans UNC-16 which is primarily involved in vesicle transport, but whose protein is also known to interact with the MAPK, JNK, and also with JNK kinases; it thus functionally intersects with signal transduction pathways. Further investigation of these genes could therefore lead to improved understanding of the dehydration stress response in nematodes.

Twelve of the selected ESTs encode novel proteins that have no matches in the GenBank or WormPep databases (Table 2). Six of these are predicted to be unusually hydrophilic (hydropathy scores <-0.43, the GRAVY score of bovine serum albumin) or largely disordered (>50%) or both, and all are predicted to be charged at physiological pH. (The pHi of resting intestinal cells of C. elegans has been determined to be 7.53; 45 .) Although they are clearly not LEA proteins, these novel proteins are reminiscent of anhydrin, another highly hydrophilic, disordered protein previously described in A. avenae, whose gene expression is induced by drying 24 . The prevalence of such proteins, albeit from a very limited sample, is intriguing and invites further investigation in the context of desiccation tolerance.

Homologues of two polygalacturonase genes in the fungi Aspergillus awamori and Penicillium griseoroseum 46 47 were found in the A. avenae EST set. These genes seem to be derived from the A. avenae genome and are not a result of fungal contamination since preliminary Southern hybridisation experiments show they are not present in Rhizoctonia solani, the food source used (data not shown). Polygalacturonases (EC:3.2.1.15) hydrolyse 1,4-alpha-D-galactosiduronic linkages in pectin, a major component of the cell walls of plants 48 . Plant pathogenic bacteria and fungi secrete hydrolytic enzymes such as polygalacturonases and cellulases to degrade plant cell walls. Plant parasitic nematodes also secrete a great diversity of plant cell-wall degrading enzymes including cellulases and galacturonases. Phylogenetic analyses support the interpretation that the cellulases and polygalacturonases present in cyst and root-knot nematodes have been acquired by horizontal gene transfer from bacteria 49 50 51 52 53 . In contrast, the polygalacturonase genes of A. avenae appear to be derived by horizontal transfer from fungi (Additional File 1; Table 1). Interestingly a polygalacturonase gene found in the genome of the rice weevil Sitophilus oryzae also appears to have been incorporated into the genome of this insect by horizontal gene transfer from a fungal source 54 . While the cellulase and polygalacturonase genes of the root-knot nematode M. incognita are most closely related to bacterial sequences, genome-wide analysis indicates that three other plant-cell wall degrading/modifying enzymes in this nematode (pectate lyase, arabinase and expansin) may be of fungal origin 52 , and it has also been shown that the cellulases of the pine wood nematode are most similar to cellulases from fungi 55 . A. avenae is routinely cultured in the laboratory on fungi such as Rhizoctonia solani or Botrytus cinerea, but A. avenae is also capable of growing and reproducing in vitro on tissue cultures prepared from several plant species 56 and it has also been shown that A. avenae will feed and multiply on plant roots, and suppress growth 57 . The occurrence of a plant cell wall degrading enzyme in A. avenae provides further evidence that in addition to feeding on fungi A. avenae might also utilise plant tissues as a food source in nature.

Spliced leaders derive from specific snRNAs and are coupled to pre-mRNAs through a trans-splicing mechanism during mRNA maturation 58 59 . Previously, 22-nucleotide spliced leader sequences representing four variants of the SL1 sequence characterised in C. elegans have been described at the 5' ends of trehalose-6-phosphate synthase (tps) mRNAs in A. avenae 13 . All four of these SL1 sequences were found in the ESTs described here, together with eight additional variants (Table 3). Uncovering a total of 12 different sequences from a relatively small sample of mRNA sequences suggests a degree of spliced leader polymorphism in A. avenae similar to that reported for Trichinella spiralis 60 . Interestingly, four variants (5, 6, 7 and 8) show striking similarity to SL2, the second main form of spliced leader found in C. elegans, which is associated with the resolution of polycistronic pre-mRNAs resulting from operon transcription. SL2 is postulated to have first evolved in the rhabditine lineage which includes C. elegans 59 , but the data of Table 3 suggest that this family of spliced leaders is also present in the Tylenchina, which includes A. avenae, and therefore must have arisen no later than the common ancestor of the two sister groups within Nematoda. It will be interesting therefore to investigate whether SL2-like trans-splicing is used for processing of operon transcripts in A. avenae, like C. elegans, instead of SL1, as found in all other nematodes to date outside the Rhabditina 59 .

The expression of 42 A. avenae genes in response to evaporative water loss (90% RH, 24 h), increased osmotic potential (1 M sucrose, 24 h), cold (4°C, 24 h) and heat (32°C, 24 h) was evaluated by quantitative PCR (Table 1; Table 2). Two thirds of the recognisable genes were found to be upregulated by drying, thus validating the selection procedure and implicating the EST set in anhydrobiosis. Interestingly, the large majority of genes analysed were also responsive to osmotic stress and cold (83% and 77%, respectively), probably reflecting similar effects on water activity and overlapping stress responses to all three environmental conditions. In contrast, heat stress caused relatively few genes to be induced, with 30% responding to elevated temperature, implicating a rather different response mechanism.

The degree of expression varied widely between genes although ANOVA analysis showed that, in most cases, a 1.5-fold increase in expression levels was significant (Table 1). Nevertheless, some genes were highly upregulated: the LEA protein and LEA-like genes (EF026241 and EF026240) showed a 43-fold and 36-fold increase in expression level, respectively, in response to desiccation, and a response to osmotic stress in excess of 20-fold. These genes were also substantially upregulated on exposure to low temperature (8-fold and 47-fold respectively). Other genes were not induced to the same levels in most cases, but notable results include that of a glutathione peroxidase (GR463914) whose mRNA concentration increased 32- and 14-fold in response to desiccation and osmotic stress, respectively; a DJ-1-like gene (GR463903) showed a 10-fold induction on desiccation; and an AHA1 activator gene (GR463897) also showed high relative expression levels following exposure to desiccation and osmotic stress. The above results are consistent with a requirement for increased antioxidant and molecular shield/chaperone activity during water stress, as has been noted previously (e.g. 4 ). Of the novel genes (Table 2), four showed more than 10-fold upregulation on osmotic stress and two displayed over five-fold induction in response to desiccation.

Although the selective cloning procedure and the expression data of Tables 1 and 2 implicate the large majority of the gene set in anhydrobiosis, it would be desirable to obtain more direct evidence that they are required for survival of the dry state. To this end, we sought to establish RNA interference (RNAi) in A. avenae as a tool to selectively eliminate or reduce gene activity. Populations of A. avenae in which expression of particular genes had been reduced by gene silencing could then be assessed for desiccation tolerance; where survival was compromised, the respective genes would be strongly implicated in anhydrobiosis. RNAi occurs in C. elegans after injection of double-stranded RNA (dsRNA) corresponding to the target gene into worms 61 , after soaking nematodes in a dsRNA solution 62 or after feeding nematodes on bacteria expressing dsRNA 63 . The last of these methods is not available to us since A. avenae do not feed on bacteria. Therefore, attempts were made at injecting dsRNA into A. avenae but these were unsuccessful due to technical difficulties relating to the small size of the worms. The injection method is also laborious and is not amenable to studies at the population level, which is usually required for assessment of desiccation tolerance. A soaking method was therefore attempted using dsRNA corresponding to the RNA polymerase II large (α) subunit gene, ama-1, which is known to give a lethal phenotype when silenced in C. elegans and other nematodes such as the human parasitic species Brugia malayi 64 . Two cDNA sequences corresponding to the A. avenae ama-1 transcript were cloned using primers directed against conserved peptide sequences identified in an alignment of nematode and other RNA polymerase II large subunit genes 65 . Soaking of A. avenae for 24 h in 1 mg/ml ama-1 dsRNA did not, however, show a phenotype compared to negative control experiments (data not shown). Positive technical controls were performed using C. elegans and cognate ama-1 sequences, resulting in sick animals within 24 h. Further controls were also carried out using unc-22 and GFP dsRNA on wild-type and GFP-expressing C. elegans (PD4251) to confirm the effectiveness of RNAi.

To determine whether dsRNA molecules were taken up by A. avenae, nematodes were soaked in fluorescently-labeled dsRNA (GFP sequence) for a period of approximately 24 h. Wild type C. elegans was used as a positive control. Fluorescence was observed in the gut of C. elegans by confocal microscopy but no signal was observed in A. avenae, suggesting that dsRNA is not ingested efficiently in this species (Figure 2). Addition of the neuromodulator, octopamine, at concentrations up to 50 mM, which has been reported to improve dsRNA uptake in some parasitic nematodes 66 67 , did not stimulate ingestion in A. avenae. Therefore, perhaps due to its feeding mechanism which involves stylet puncture of target fungal or plant tissues 68 , it is likely that A. avenae is not readily amenable to RNAi by the soaking method.

One strategy which might partially circumvent difficulties with RNAi in A. avenae is cross-species gene silencing using A. avenae sequences to reduce expression of homologous transcripts in another anhydrobiotic nematode. Cross-species RNAi has been shown using sequences from the animal parasitic nematode, Ascaris suum, in C. elegans, for example 69 . Recent work by one of our groups (AMB) has demonstrated efficient systemic RNAi in Panagrolaimus superbus, a nematode in the same phylogenetic group as A. avenae with a marked resistance to desiccation. Indeed, since P. superbus is a bacterivore, it is susceptible to large scale RNAi in nematode populations using a feeding method very similar to that used routinely in C. elegans 70 .

From the A. avenae EST set, 20 candidates (indicated in bold in Tables 1 and 2) were selected for RNAi in P superbus. The majority of these were chosen because they showed a high degree of similarity to sequences in C. elegans and/or Caenorhabditis briggsae, the rationale being that they were then more likely to have similar counterparts in P. superbus: both P. superbus and A. avenae are tylenchid nematodes, and thus more closely related in evolutionary terms than either is to the genus Caenorhabditis 70 . For comparison, three of the 20 ESTs chosen were novel sequences (GR463919, GR463921, GR463927).

P. superbus were fed bacteria expressing dsRNA corresponding to individual A. avenae sequences, after which worms were first pre-conditioned at 90% RH for 24 h followed by 24 h desiccation at 10% RH. The worms were then rehydrated overnight in buffer before scoring for survival. Worms fed dsRNA corresponding to GFP and treated similarly were taken as negative controls. Prior to desiccation, nematodes were also subjected to phenotypic analysis, but none of the experiments performed revealed any extraordinary change in morphology or behaviour, either of adults, larvae or embryos (Additional File 2).

This first drying experiment was used as a preliminary screen; four candidates which showed a reduction in desiccation tolerance of at least 10% compared to controls were chosen for a second round of RNAi and desiccation: GR463899 (ggr-3, a predicted member of the GABA family of ligand-gated chloride channels); GR463913 (col-14, a collagen structural gene); GR463914 (a glutathione peroxidase); and GR463921 (unknown function). Survival of P. superbus after RNAi treatment and desiccation was compared to nematode viability after mock desiccation, in which worms were subjected to 100% RH; as previously, a GFP negative control was also performed (Figure 3). In this experiment, RNAi treatment with both GR463899 and GR463913 resulted in no significant reduction in survival following desiccation compared to non-dried and GFP controls. Therefore, neither was considered to have a measurable RNAi effect. However, GR463914 and GR463921 did show significant (P < 0.01) reduction in survival of drying after RNAi to 66% and 59%, compared to 84% and 75%, respectively, without desiccation. These results therefore suggest that glutathione peroxidase and an additional uncharacterised gene (GR463921) are contributors to desiccation tolerance in P. superbus, and by extension in A. avenae.

If the cross-species strategy used above is valid, then genes that are upregulated in response to desiccation in A. avenae, and that have well-conserved sequence counterparts in P. superbus, should be part of an anhydrobiotic gene set in both nematodes. Therefore, we attempted to identify genes in P. superbus which were homologues or paralogues of the two genes indicated by the cross-species RNAi experiments, in order to carry out similar experiments with cognate P. superbus sequences. Consequently, an EST library of ~9,000 P. superbus cDNAs (Tyson et al., in preparation) was searched for sequences corresponding to A. avenae GR463914 and GR463921. Although no matches were found for the novel sequence represented by GR463921, two different sequences showing similarity to GR463914 were obtained, Ps92 (accession number GR881192) and Ps114 (accession number GR881191), both of which are similar to glutathione peroxidase sequences in GenBank and other databases. For example, both Ps92 and Ps114 gave highly significant matches with the C. elegans glutathione peroxidase gene C11E4.1, showing BLASTX scores of 2E-97 and 1E-86, respectively. The ESTs probably represent different genes in P. superbus since they are only 66% identical using the ALIGN tool.

RNAi experiments were therefore carried out in P. superbus with these sequences, together with A. avenae GR463914 (glutathione peroxidase) and GFP sequences as positive and negative controls, respectively. Figure 4 confirms that RNAi using the A. avenae glutathione peroxidase sequence reduces desiccation tolerance in P. superbus. Moreover, the two glutathione peroxidase sequences from the P. superbus EST library also showed a pronounced RNAi effect on survival after drying. These results strongly implicate a role for glutathione peroxidases in nematode anhydrobiosis and provide the first genetic evidence for the importance of antioxidants in metazoan desiccation tolerance.

Aphelenchus avenae was grown in the dark at 20°C on the fungus Rhizoctonia solani, itself grown on a substrate of wheat 102 . After a growth period of 2-3 weeks (before swarming of the nematodes in the jars), nematodes were rinsed with tap water and brought to a concentration of approximately 3,300 worms per ml. Nematodes were cleaned in 30% (w/v) sucrose solution, followed by three washes in tap water and the nematodes were transferred to 25 mm Supor^® Membrane Disc Filters (0.45 μm, Pall Corporation, Michigan, USA; 30,000 nematodes per filter) by vacuum filtration. These were placed in a desiccation chamber at 20°C for 24 h over a saturated solution of MgSO₄.7H₂O (90% RH), or over a saturated solution of KCl (85% RH) 103 .

For the qPCR experiments the nematodes were grown and harvested as described above but they were cleaned by placing the nematode pellet in a sieve containing a layer of Kleenex^® allowing them to crawl through the paper (for 2-3 h) into the water filled dish in which the sieve had been placed. A. avenae populations were exposed to individual stresses as follows. Heat stress: one replicate consisted of 1 ml of the above worm suspension placed in a 30 mm Petri dish incubated at 32°C for 24 h. Cold stress: one replicate consisted of 1 ml of worm suspension in a 30 mm Petri dish incubated at 4°C for 24 h. Desiccation: one replicate consisted of 1 ml of worm suspension vacuum filtered onto a 25 mm Supor^® membrane disc and placed in a desiccation chamber set at 90% RH (using a saturated solution of MgSO₄.7H₂O) for 24 h. Osmotic stress: one replicate consisted of 1 ml of worm suspension added to 1 ml of 1 M sucrose solution in a 30 mm Petri dish and placed at 20°C for 24 h. Control nematodes consisted of 1 ml of worm suspension in 30 mm Petri dishes incubated at 20°C for 24 h. Three biological replicates were set up for each stress and for control worms. These stress conditions yielded an 80% survival rate for each stress compared to a 90% survival rate for the control worms.

P. superbus and C. elegans were grown on NGM plates and fed Escherichia coli strain OP50, or strain HT115 carrying an empty L4440 plasmid, in the dark at 25°C.

Total RNA was extracted from desiccated and control nematodes using TRI^® Reagent (Sigma-Aldrich GmbH, Steinheim, Germany). Total RNA from desiccated nematodes was converted to 5'-oligo-capped cDNA using the GeneRacer™ Kit (Invitrogen, Carlsbad, California, USA), following the manufacturer's instructions. The GeneRacer™ Oligo dT primer was used in the reverse transcription reaction. PCR was carried out on the 5'-oligo-capped cDNA with the GeneRacer™ 5'-primer (CGACTGGAGCACGAGGACACTGA) and the GeneRacer™ 3'-primer (GCTGTCAACGATACGCTACGTAACG) using Platinum^® Taq DNA Polymerase (Invitrogen). The product of this reaction was cloned into the pCR2.1 TOPO vector using a TOPO-TA Cloning kit (Invitrogen).

Recombinant clones were PCR amplified in a 96 well plate using standard M13 forward and reverse primers and the presence of an insert was confirmed by electrophoresis on a 1% agarose gel. The PCR products were denatured by adding NaOH to a total concentration of 0.4 M and EDTA to a total concentration of 0.04 M and heating at 99°C for 10 min. The denatured DNA was blotted onto Hybond™-N+ transfer membrane (Amersham Biosciences, Buckinghamshire, UK) in duplicate using a SRC 60-D Dot Blot Minifold (Schleicher & Schuell, Dassel, Germany). The membranes were air dried at 55°C for 30 min and UV-crosslinked using a UV Stratalinker^® 1800 (1200 mJ; Stratagene, California, USA).

Total RNA was extracted from desiccated (90% RH for 24 h) and control nematodes as before 24 . The RNA was resuspended in 30 μl DEPC-treated H₂O, concentration determined by measuring absorbance at 260 nm and adjusted to 1 μg/μl. 5 μg of RNA was converted to cDNA using the SuperScript™ First-Strand Synthesis System (Invitrogen). Experimental (desiccated) and control probes were constructed by radiolabelling this cDNA with [³²P]-dCTP using the Prime-a-Gene^® Labelling System (Promega, Southampton, UK) in accordance with the manufacturer's instructions. One replicate of each dot blot membrane was hybridized with the experimental probe and the corresponding replicate was hybridized with the control probe. Hybridisation was carried out in hybridisation buffer (5% dextran sulfate, 1 M NaCl, 1% SDS in water) at 60°C for 18 h in glass tubes with gentle rotation. The membranes were washed twice in wash buffer (10× SSC, 0.5% SDS in water) for 4 min at room temperature followed by two washes for 15 min at 60°C. The membranes were then exposed on X-OMAT AR photography paper (Kodak, New York, USA) for 24 h. The control and experimental dot blots were digitally scanned and analyzed using the ImageQuant TL Array Analysis software (Amersham Biosciences). Clones considered to be upregulated in response to desiccation were selected for sequencing and sent to Agowa GmbH (Berlin, Germany). The resulting sequences were assembled into contigs using the online sequence assembly program CAP3 http://pbil.univ-lyon1.fr/cap3.php, yielding a dataset of 88 unique ESTs, enriched in those upregulated in response to desiccation.

The POPPs 35 is a suite of applications, one of which clusters protein sequences based on similarities in their peptide compositions, in this case to see whether the input sequence clusters with any of the LEA protein groups and, if so, which. The GRAVY calculation was undertaken by a small Python application which uses the Eisenberg Consensus scale 104 in preference to the Kyte-Doolittle scale 105 , which has well known anomalies. Three predictors of unfolded proteins were used in this study: Foldindex ( 106 ; http://bip.weizmann.ac.il/fldbin/findex/, VL-XT ( 107 ; http://www.pondr.com) and IUPred ( 108 ; http://iupred.enzim.hu). Uversky plots 109 of charge vs. hydropathy were performed at pondr.com and raw data downloaded to construct graphical output in GraphPad Prism version 4.0b (GraphPad Software, California, USA). An application called Perdix was used to search the input sequences for amino acid periodicity, i.e. amino acids that recur every N amino acids. So, for example, in the peptide GAGPG the G recurs with period 2. The application Perdix, and the earlier application for computing hydrophobicity and other peptide statistics, both currently unpublished, can be obtained from Wise. The pairwise sequence alignment application ALIGN 110 is part of the San Diego Supercomputer Workbench http://workbench.sdsc.edu.

The replicate worm samples from each individual stress treatment were pooled and RNA was extracted using Trizol reagent for each stressed and control worms followed by treatment with DNAse I (RNase free, Invitrogen). The concentration and quality of RNA was determined using a Thermo Scientific Nanodrop Spectrophotometer. Total RNA (1 μg per reaction) was converted to cDNA using the Roche Transcriptor First Strand cDNA Synthesis Kit. Primers were designed for each gene to amplify a fragment of approximately 125 bp (see Additional File 4). Real time qPCR reactions were carried out with a Roche LightCycler 480 thermocycler using Roche SYBR I Master Mix with 0.0025 pmol primers (each), 500 ng of sample cDNA template and 0.5 U heat-labile uracil-N-glycosolase per reaction. Relative expressions were calculated using the second derivative maximum method with A. avenae 5.8S rRNA and ama-1 genes as reference genes. Statistically significant differences in expression levels were confirmed using ANOVA with a confidence level of 99.9% and Dunnett's comparison test with α = 1.

Two ama-1 cDNA fragments were obtained from A. avenae by RT-PCR. The short fragment (322 bp), was obtained with the primers AaAMA1s (5'-TTTTTCATGCAATGGGCGGTC-3'; forward) and AaAMA1as (5'-ACATAAACCTCTCTCACGACG-3'; reverse). Reverse transcription was performed using SuperScript II (Invitrogen) according to manufacturer's instructions, and the PCR was done under the following conditions: 95°C 5 min; then 35 cycles of 95 C 30 s, 57°C 30 s and 72°C 1 min; with a final extension at 72°C for 10 min. A longer version of ama-1 cDNA (800 bp) was cloned using the same RT-PCR conditions with primers Aa.AMA1.s2 (5'-ACGTCGACTGATTAAGGCCAT-3'; forward) and Aa.AMA1.as3 (5'-TGATGATCTCCTTCAGACGAG-3'; reverse). PCR products underwent a second amplification round using T7-tagged primers. These new amplicons were then used for in vitro transcription using Ampliscribe T7 Flash Transcription Kit (Epicentre) according to manufacturer's instructions. For RNAi attempts, worms were soaked in 20 μl soaking buffer (5 mM Tris-HCl pH7.0) containing dsRNA (1 μg/μl) for 24 h and then plated on LB agar plates (50 μg/ml carbenicillin). Worms were placed over a small spot of fungus (Botrytis cinerea) previously grown on the LB agar plate for 24 h. Survival analysis was performed by observation under a microscope. C. elegans was treated in a similar fashion except worms were incubated with GFP and unc-22 dsRNA for 24 h at ambient temperature and recovered on E. coli-seeded NGM plates until phenotypes were observed. For dsRNA uptake analysis, dsRNA molecules were first synthesised by in vitro transcription from undigested L4440-GFP plasmids. Then 5 μg dsRNA was labelled using Label IT CX-rhodamine Nucleic Acid Labeling Kit according to manufacturer's instructions (Mirus/Cambridge Bioscience). Worms were soaked in 20 μl soaking buffer for 22 h at 15°C in the dark. Uptake efficiency was assessed using a confocal microscope (Zeiss LSM510 META).

Plasmid DNA of selected A. avenae EST clones (in TOPO vector; Invitrogen) were digested using Sac I (NEB) and Xho I (NEB), inserts were gel-purified and ligated into Sac I/Xho I -digested L4440 feeding vector 80 . Competent HT115 cells were transformed with ligation products and selected on LB ampicillin (100 μg/mL). Subcloning was confirmed by digesting L4440 plasmids with Sac I/Xho I. Transformed HT115 clones containing individual plasmids for RNAi were each grown in 10 ml LB ampicillin (50 μg/mL) overnight, recovered by centrifugation and resuspended in 300 μl plain LB medium. Bacteria were transferred to Petri dish plates (9 cm diameter) containing NGM agar plus IPTG (240 μg/ml) and ampicillin (50 μg/ml). Plates were left for 24 h at ambient temperature to allow dsRNA induction. P. superbus was prepared for RNAi as follows: a large (140 mm) NGM plate containing mixed stage animals with many gravid adults was harvested, worms washed 3× in M9 buffer and synchronised by bleaching (3 min in 1% NaOCl, 0.5 M NaOH). After bleaching, worms were washed 3× in M9 and left to hatch in M9 on a shaker for 48 h. Approximately 1500 worms were placed onto each feeding plate (three per test gene) where they remained for approximately 12 days in the dark at 25°C before desiccation. Worms were then collected, washed 3× in M9 buffer, counted and placed onto three 25 mm Supor^® Membrane Disc Filters by vacuum filtration (approximately 300 RNAi-treated worms per filter) and processed as follows: 24 h at 90% RH (saturated BaCl₂); 24 h at 10% RH (dry silica gel) and overnight in M9 buffer. Non-desiccated control worms were processed as follows: 48 h at 100% RH (UHP H₂O) and 24 h in M9 buffer. All treatments were in 350 ml sealed plastic boxes. After rehydration, viability was assessed under the microscope, where 80 worms were randomly chosen for analysis from each filter (spontaneously moving worms were considered alive). C. elegans fed with a clone expressing unc-22 dsRNA was used as a positive control of RNAi treatment (twitching phenotype). Negative controls consisted of P. superbus and C. elegans worms fed with a clone expressing GFP (green fluorescent protein) dsRNA. Statistical relevance was determined by one-way ANOVA and a Tukey post hoc test using InStat3 (GraphPad Software, San Diego, CA).