A directionally cloned cDNA library of desiccated nematodes was constructed and a total of 2,586 of clones were subjected to single pass sequencing from their 5' ends. Trimming of vector sequences, poly A/T tails, low quality, adaptor, and contaminating sequences provided a data set of 2,486 high quality (hq) ESTs with a minimum length of 100 base pairs (bp) (Table 1). Among 2,486 hq ESTs, 1,423 were assembled into a total of 324 contigs, and the remaining 1,063 ESTs were classified as singletons, suggesting a combined total of 1,387 putative unique transcripts (Table 1). The number of ESTs in the 324 contigs varied from 2 to 37, with the highest number of contigs being two ESTs, followed by more than 3 ESTs and the least number with more than 21 ESTs (Fig. 1). These hq ESTs ranged from 90–1125 bp with average lengths of 545 ± 156 bp. The average length of the contigs was higher than for singletons. The average GC content was higher in P. murrayi (44%) than in C. elegans (36%). All sequences have been deposited in the dbEST division of DDBJ/EMBL/GenBank under accession numbers [GenBank: FG618921] – [GenBank: FG621295], [GenBank: FG647736] – [GenBank: FG647869].

We used the 1,387 unique sequences to search a non-redundant protein data base using BLASTX 3334 (Table 1) and the Wormpep 190 database consisting of extensively curated C. elegans proteins from WormBase 35. A total of 782 unique sequences (56%) matched known proteins, including 523 unique sequences (38%) with significant match to C. elegans proteins at a cut-off expectation (E)-value of 10^-5 or below. The remainder of the unique sequences (44%) had no meaningful matches (E > 10^-5). We compared our unique sequences with the ESTs from other nematodes as well as non-nematodes using BLAST searches. Only 107 unique sequences (7%) matched other nematode ESTs and 153 unique sequences (11%) matched organisms other than nematodes at E < 10^-5 (Table 1). Of 1,387 unique sequences, 36 had homologues in C. elegans which could be silenced by RNAi. The RNAi phenotypes (as described by WormBase) included mig-15 (migration), lin-8 (lineage), unc-16, 89 (uncoordinated), dpy-6 (dumpy), rde-1 (RNAi defective), drh-2 (dicer related), nhr-67 (nuclear hormone receptor) and ard-1 (alcohol/ribitol dehydrogenase).

To identify transcripts differentially expressed (DE) during desiccation stress, subtractive hybridization was conducted between cDNA from gradually desiccated and fresh active nematodes (control). Two rounds of hybridization were done and DE clones were sequenced which resulted in 80 quality sequences above 100 bp (Table 2). The nucleotide sequences were analyzed and their putative functions identified by BLASTX search. The DE transcripts included 22 ESTs (28%) similar to metabolism related genes, 15 (19%) similar to environmental information processing genes, 23 (28%) similar to genetic information processing genes, 3 (4%) similar to hypothetical proteins of other organisms and 17 (21%) novel transcripts that had no identifiable similarity to known sequences in GenBank 36. Among the metabolism related genes, 13 ESTs (68%) were involved in carbohydrate metabolism, 2 transcripts (10%) of each in lipid metabolism, amino acid metabolism and protein folding, sorting and degradation. The environmental information processing category was dominated (53%) by stress related proteins. In the genetic information processing category, ribosomal proteins were the most abundant (42%) group followed by translation elongation factor (19%) (Table 2). All sequences have been deposited in the dbEST division of DDBJ/EMBL/GenBank under accession numbers [GenBank: FK670236] – [GenBank: FK670315] and [GenBank: GH196899; GH229101].

The most unexpected discovery among DE ESTs was Type II antifreeze protein (AFP) [GenBank: FK670242 which showed high similarity to that of Clupea harengus (Table 2), the Atlantic herring. This finding represents another case of an ice structuring protein from an Antarctic nematode, suggesting the possibility that Antarctic nematodes may use similar antifreeze proteins for stress adaptation heretofore observed only in one Antarctic nematode 37, some fishes 38, insects 39, plants 40 and fungi and bacteria 41.

A total of 23 contigs containing 384 ESTs were highly redundant. This accounted for more than 15% of the total high quality ESTs. The minimum and maximum number of ESTs that made up these highly redundant contigs was 7 and 37 respectively (Table 3). More than one third (9) of the highly redundant contigs, totalling 127 ESTs, had significant similarity to various genes involved in metabolism. One third (8) of the highly redundant contigs, totalling 139 ESTs, had significant similarity to various environmental information processing related genes, indicating high transcript abundance of stress related genes, as expected. Two of the contigs, totalling 48 ESTs, had significant similarity to ribosomal proteins while three of the contigs, totalling 37 ESTs, had significant similarity to genetic information processing related genes. One of the highly redundant contigs totalling 33 ESTs matched similar sequences derived from the mitochondrial cytochrome oxidase subunit (Table 3). The most redundant group of contigs were composed of 37 ESTs and had significant similarity to ribosomal protein from C. elegans, indicating higher activities of ribosomal protein genes during desiccation stress.

To categorize transcripts by putative function, we utilized the GO classification scheme (April 2008 release of GO database, Gene Ontology Consortium). GO provides a dynamic controlled vocabulary and hierarchy that unifies descriptions of biological, cellular and molecular functions across genomes 42. In this report, we relied on well-annotated GO information of C. elegans and other nematodes. GO representation of P. murrayi clusters is shown for each organizing principle of GO: molecular functions (Additional file 1a; Fig. 1a), cellular components (Additional file 1b; Fig. 1b), and biological processes (Additional file 1c; Fig. 1c). Additional file 1 and Fig. 2 provide a breakdown of representation by major GO categories. The highest GO term for molecular functions was protein binding, under 'ligand binding and carrier' categories, which had 87 unique sequences accounting for 18% of the total unique sequences matched in this category and 6% of the total unique sequences. The highest final GO term in cellular components was mitochondria (under the 'cytoplasm' category) with a total of 28 unique sequences, 17 of which are encoded on the mitochondrial genome, accounting for 12% of the total in this category. Similarly, the highest final GO term for biological processes was protein metabolism, under 'metabolism' categories, which had 44 unique sequences accounting for 12% of the total in this category and more than 3% of the total unique sequences. We found 13 unique sequences showing significant similarity to C. elegans signal transduction factors; 8 of them belonged to the receptor binding group and 5 sequences belonged to receptor and receptor signalling proteins (Additional file 1).

As an alternative method of categorizing unique sequences by biochemical functions, sequences were assigned to metabolic pathways via KEGG 43 using enzyme commission (EC) numbers as the basis for assignment. Only 281 unique sequences (36% of total) were assigned EC numbers and had 158 unique mappings to KEGG biochemical pathways (Table 4). The KEGG metabolic pathways that are well represented by P. murrayi unique sequences are carbohydrate metabolism (18 enzymes), amino acid metabolism (9 enzymes), lipid metabolism (8 enzymes), xenobiotic and bio-degradation metabolism (5 enzymes), and biosynthesis of secondary metabolites (3 enzymes). Of these, 12% of the unique sequences belonged to the environmental information processing (EIP) category, indicating higher activities of stress and chaperone related genes during desiccation. The KEGG pathways well-represented under EIP are membrane transport (15 enzymes), ligand-receptor interaction (15 enzymes), signal transduction (8 enzymes) and signalling molecules and interaction (9 enzymes). About 11% of the unique sequences belonged to the genetic information processing (GIP) category with most of them having roles in folding, sorting and degradation. The KEGG pathways well-represented under GIP are folding, sorting and degradation (25 enzymes), transcription (9 enzymes), translation (8 enzymes) and replication and repair (6 enzymes). Most of the sequences (49%) remained unassigned to any known functional pathway and 15% of the sequences were similar to C. elegans hypothetical proteins (Table 4). The lowest number of sequences mapped to the cellular processes category (3%), suggestive of developmental arrest during anhydrobiosis. The cell growth and death (5 enzymes) and cell communication (4 enzymes) pathways were also the well-represented categories under cellular processes (Table 4).

In order to validate our differential gene expression results and obtain more refined gene expression data, we designed gene-specific primers for 14 transcripts selected from Table 2 and analyzed their expression using quantitative real-time PCR (qRT-PCR) (Fig. 3). These genes were chosen to represent a variety of functional classifications. Among these 14 transcripts, 9 were significantly induced (fold-change > 2.0×, P value < 0.05) and 3 genes were reduced (fold-change > 2.0×, P value < 0.05) in response to desiccation. Significant desiccation-induced gene expression change ranged from -6.50-fold for antifreeze protein to 26.77-fold for trehalose-6-phosphate synthase. Heat shock protein 70 and 90 were also weakly induced (fold-change 1.7 and 1.94×), but lacked significant statistical support P > 0.05). Among the DE transcripts, putative homologs to trehalose 6-phosphate synthase protein showed highest induction (fold-change 26.77×, P value < 0.05) followed by the putative homolog to glycerol kinase (fold-change 25.04×, P value < 0.05). Interestingly, there was significant reduction and induction on the level of expression of two novel transcripts, [GenBank: FK670306 and FK670310] (fold-change -10.3× and 16.71× respectively, P < 0.05), suggestive of their possible roles in desiccation tolerance (Fig. 3). The mRNA copy number was calculated using an absolute quantification method 44. There was no significant difference (P value < 0.05) in copy number of transcripts encoding Hsp70 and Hsp90 on desiccated and control nematodes (Additional file 2).

Gene ontology has been widely used to characterize gene function annotation and classification 42. GO describes gene function using controlled vocabulary and hierarchy, including molecular function, biological processes, and cellular components (Fig. 2; Additional file 1). A large number of the unique sequences from our study mapped to molecular functions and ligand binding, enzymes, molecular transducer and transporter subcategories. Each one of these subcategories represents catalytic activities that could be argued are important for a cell to survive major metabolic perturbation during desiccation. The second most represented category was biological process, which includes unique sequences associated with cell growth or maintenance, development and cellular communications. Within the cell growth or maintenance category most unique sequences were associated with metabolism, localization, transport and response to stress. This distribution is not surprising for ESTs (unique sequences) derived from an organism undergoing metabolic changes such as desiccation. The subcategories under the least represented category, cellular process, also seem to reflect the nature of the cellular disturbances that result from desiccation. Under this category there is significant representation under intracellular and membrane subcategories. Within these subcategories representation is most significant in mitochondria, cytosol, ribosome, nucleus and ribonucleoprotein complex. The importance of protein synthesis and cellular homeostastasis during desiccation may explain the preponderance of unique sequences associated with mitochondrial and ribosomal structural components.

As an alternative method of categorizing clusters by biochemical function, clusters were assigned to metabolic pathways using the KEGG database 44 (Table 4). Only 36% of the unique sequences mapped to the currently known KEGG pathways with 158 unique mappings. The paucity of EC assignments limits this aspect of analysis but the mapping of unique sequences to the KEGG metabolic and other pathways still presents some useful perspectives on the metabolic, protein folding and degradation and cellular repair emphasis of desiccating cells. Most of the unique transcripts belonged to genetic information processing (GIP) with protein folding, sorting and degradation. Transcripts encoding a cathespin L-like protease [GenBank: FK670238] enzyme found in this category have long been recognized for their role in intracellular and extracellular protein degradation in a range of cellular processes. In C. elegans cathespin L-protease (Ce-cpl) has been demonstrated to play a critical role during embryogenesis, larval development and moulting 47. Considering the putative homology of the transcript encoding P. murrayi cathespin L-protease (Pm-cpl) with C. elegans, it might have a similar role in nematode molting and thus its possible role during desiccation survival needs further investigation.

Metabolism was the second most represented category and pathways well-represented by the P. murrayi clusters were carbohydrate metabolism, amino acid metabolism, lipid metabolism, biosynthesis of secondary metabolites, glyoxylate, and decarboxylation metabolism (Table 4). The lipid metabolism pathway in anhydrobiotes is one of the most active pathways as lipids are the main reservoir of energy and the most likely source of carbon for the synthesis of trehalose 21. The GIP category included the pathways involved in protein folding, sorting and degradation, transcription, translation, replication and repair. Desiccation stress coupled with oxidative stress results in lipid and protein damage, leading to impaired cellular survival and functioning. Nematodes are reported to respond to these damages and repair them with increased expression of stress-protective genes and antioxidant enzymes 48. The cellular processes category has the least number of transcripts. This finding is suggestive of developmental arrest during anhydrobiosis. Indeed, anhydrobiosis may be an exaptation that allows the individual to survive unfavourable conditions by staying in an arrested state of development and reproduction (Table 4).

A high level of representation in a cDNA library generally correlates with high transcript abundance in the original biological sample 49, although artefacts of library construction can result in selection for or against representation of some transcripts. The genes with the most abundant transcripts were mostly involved in metabolism, molecular chaperones, reactive oxygen species scavenging and genetic information processing (Table 3). Ribosomal protein (RP) (GenBank: FG618944) was the most abundant transcript expressed during desiccation, suggesting the possible involvement of transcriptional regulation of ribosomal proteins during desiccation stress. Differential expression of plant ribosomal protein genes has been observed during development and following various stress or hormone treatments, including desiccation stress 50. The 'ribosomal filter hypothesis' of Mauro and Edelman 51 proposes that particular combinations of ribosomal proteins or rRNA could favour the translation of specific mRNAs, thereby providing a mechanism for translational control. Interestingly, many ribosomal proteins were DE and one of them was up-regulated (Pm-rpl-4), suggesting that RPs may provide a means for selectively translating specific mRNAs required for the desiccation response as suggested for the insect parasitic nematode Steinernema carpocapsae 28. Similarly, it may be part of an adaptive response to maintain ribosomal function in dehydrated cytoplasm.

A number of metabolism related transcripts encoding cytochrome c oxidase subunit 2 (GenBank: FG618989), glutamate synthase (GenBank: FG620419), NADH-dehydrogenase subunit 1 (GenBank: FG619004) and aldehyde dehydrogenase (GenBank: FG619103) were also abundantly expressed during desiccation stress. Several lines of evidence have suggested that nematodes activate metabolic pathways in response to desiccation shortly after exposure to dehydrating conditions 26.

Transcripts of a number of stress related genes were also abundantly expressed, including heat shock proteins, a chaperonin containing sub-unit, multiple stress-responsive zinc finger proteins and oxidative stress responsive genes. Desiccation stress in nematodes reportedly produces a large number of molecular chaperones to facilitate the synthesis, folding, assembly and intracellular transport of proteins, reduce protein denaturation and aggregation, and aid in protein renaturation 1952. Transcripts of several oxidative stress related genes like glutathione S-transferase (GenBank: FG618951) and copper/zinc superoxide dismutase (GenBank: FG618942) were also abundantly expressed. Reactive oxygen species and other toxins produced by oxidative stress during desiccation of nematodes can damage membrane systems, proteins and nucleic acids. Therefore, the transcript abundance of several proteins that contribute to cellular survival after oxidative damage is not surprising.

Nematodes activate their metabolic pathways in response to desiccation shortly after exposure to dehydration 26. Several transcripts encoding metabolism related genes are differentially expressed by desiccation stress in P. murrayi. These genes include aldehyde dehydrogenase [GenBank: FK670237], trehalose-6-phosphate synthase [GenBank: FK670250], thymidylate synthase [GenBank: FK670263], glycerol kinase [GenBank: FK670258], glycogen synthase [GenBank: FK670264], ATP synthase [GenBank: FK670293], ADP/ATP translocase [GenBank: FK670265], and malate dehydrogenase [GenBank: FK670259]. Interestingly, a transcript encoding a bifunctional glyoxylate cycle protein (malate synthase), a distinct and anaplerotic variant of the tricarboxylic acid cycle, was also found to be highly expressed during desiccation (Table 2).

Nematodes are unique among animals in utilizing the glyoxylate cycle to generate carbohydrates from the beta-oxidation of fatty acids 53. The glyoxylate pathway, generally found in plants and micro-organisms, is similar to the citrate cycle, but relies on two critical enzymes, malate synthase and isocitrate lyase, to bypass two decarboxylation steps. Interestingly, the anhydrobiotic nematode A. avenae has been reported to use the glyoxylate cycle during induction of anhydrobiosis 54. One sequence unique to P. murrayi that mapped to a glyoxylate cycle protein [GenBank: FK670269] includes putative homologs of malate synthase (Pm-ms). The abundant expression of malate synthase transcripts in our EST collection and its up-regulation upon desiccation stress may provide experimental support for an active role of glyoxylate cycle proteins during induction of anhydrobiosis by P. murrayi.

Transcriptional regulation and intracellular signalling cascades for nematode stress response in general and secondary metabolism in particular, are poorly understood. A number of desiccation responsive P. murrayi ESTs encode putative signalling molecules or transcription factors. One of the P. murrayi ESTs was most similar to the unc-16 gene of C. elegans, which encodes a c-Jun N-terminal kinase (JNK)-interacting protein 55, and one of the members of JNK kinase family (jkk-1) [GenBank: FK670239] was DE during desiccation. JNK (also known as stress-activated MAP kinases or SAPK) is a member of the mitogen-activated protein kinases (MAPKs) that regulate cellular responses to a variety of extracellular signals, including desiccation stress 565758. Transcripts homologous to genes encoding Zinc finger protein (GenBank: FG619080) were abundantly present in the expressed genome of P. murrayi. Zinc finger proteins are cellular proteins which play a major role in transcriptional regulation by binding with high affinity to specific regions of DNA. In conjunction with leucine zipper domains these proteins may form hetero- or homodimers and activate transcription, either constitutively, or in a regulatory manner, through post-translational modifications in response to external stimuli (although some may also be cell specific or developmentally regulated) 59. The abundance of genes encoding zinc finger protein in response to desiccation suggests that these proteins may regulate further events in the stress-response cascade of P. murrayi. Three transcripts from P. murrayi were most similar to the C. elegans gene encoding a predicted neurotransmitter gated ion-channel (GenBank: FG620960) protein. Neuronal signal transduction in response to desiccation stress would be required to initiate the coiling response 6070 of desiccating P. murrayi.

Nematodes respond to desiccation stress by synthesizing a conserved set of proteins 232425. Our results demonstrate that for P. murrayi desiccation stress can significantly elevate stress related genes encoding trehalose 6-phosphate synthase, late embryogenesis abundant proteins, heat shock proteins, ubiquitin, c-type lectins, chaperone related proteins, and other stress responsive genes (Table 3). Three transcripts encoding trehalose 6-phosphate synthase, which synthesize the storage carbohydrate trehalose, were expressed during desiccation stress. A characteristic feature of anhydrobiotic organisms is their synthesis of high concentrations of non-reducing sugars during the induction of anhydrobiosis 2961. Trehalose protects membranes and proteins from desiccation damage by replacing structural water 61, and contributes to the formation of an intracellular organic glass 62 which is thought to stabilize the cell's contents. The up-regulation of trehalose during desiccation stress of P. murrayi could be part of an adaptive response to desiccation (Fig. 3).

Protein aggregation during desiccation is likely to be a major potential hazard for anhydrobiotes; late embryogenesis abundant (LEA) proteins acting as molecular chaperones or molecular shields play an important role in prevention of this aggregation 63. Transcripts similar to plant LEA related family members [GenBank: FK670249] of C. elegans were up-regulated during desiccation stress of P. murrayi. An LEA group 3 gene Aav-lea-1 was strongly induced in A. avenae during the induction of anhydrobiosis 24. The C. elegans genome encodes three LEA genes 20 and silencing of the lea-1 gene by RNA interference (RNAi) caused a marked reduction in desiccation resistance in dauer larvae 64. We thus assume that LEA proteins contribute to protection and recovery from desiccation stress in anhydrobiotic nematodes.

Molecular chaperones, such as the Hsp70 family and the Hsp60 chaperonin complexes, are commonly perceived as heat shock proteins (Hsps), being up-regulated by stress. The P. murrayi transcripts of 70 [GenBank: FK670245] and 90 kda heat shock protein [GenBank: FK670262] and small heat shock proteins were abundantly expressed following desiccation stress. Heat shock proteins have been implicated in response to desiccation in many nematodes 18, but they appear to be constitutively expressed in P. murrayi (Additional file 2). Though Hsps may contribute to enhanced stress resistance overall, our results showed no evidence that expression levels of these Hsps were altered by desiccation. It has been shown that other Antarctic organisms constitutively express Hsp70 and Hsp90, showing no or modest up-regulation of this gene in response to thermal stress 656667. Characteristically, Hsps are expressed not at the animal's normal habitat condition, but only as part of the organism's stress response. In fact, the expression of these genes is thought to be incompatible with ongoing protein synthesis and the progression of development 6869. It may be that P. murrayi has evolved a mechanism to maintain Hsp function without disrupting normal metabolism and growth that requires synthesis of other proteins. Consistent with the observed constitutive expression, it is possible that desiccation stress did not activate these genes, and the mild desiccation failed to boost Hsp expression. An alternative explanation for the continuous up-regulation of Hsps is that because Antarctic nematodes are frequently, although unpredictably, exposed to a variety of environmental stressors such as desiccation, high pH, extreme osmotic shock, freezing, and anoxia as well as temperature 70, their survival depends on maintaining continuous expression of molecular chaperones. Furthermore, because of the unpredictability and the potential rapidity of exposure to diverse environmental stresses, the continuous production of these molecular chaperones may be energetically justified.

Unexpectedly, one of the ESTs encoded a protein similar to the type II antifreeze protein (AFP) [GenBank: FK670242] of Atlantic herring (Clupea harengus). This finding is somewhat surprising since antifreeze proteins from phylogenetically divergent organisms (insects, fish, plants, etc.,) generally have very little similarity to each other 71. The transcript encoding antifreeze protein (AFP) (Pm-afp) was down-regulated during desiccation stress of P. murrayi. It is found that Pm-afp is up-regulated only during freezing and once nematodes are exposed to desiccation other stress response mechanisms are activated by down-regulating AFP (B.N. Adhikari, unpublished work). Many overwintering organisms, including insects, fish, bacteria, fungi, and plants, accumulate AFPs that bind to the faces of ice crystals during freezing and inhibit their growth 71. While the high degree of AFP similarity between P. murrayi and C. harengus raises the idea that it originated in either of the two organisms via horizontal gene transfer, we suggest that such speculation is premature, particularly given their geographic, ecological, and phylogenetic disparity, but also because such a small fraction of the phylogenetic diversity of nematode genomes has been explored. Further sequencing and characterization are currently underway to better understand the origin and evolution of this gene in P. murrayi.

Desiccation stress induces the generation of reactive oxygen species (ROS) in nematodes, and therefore it is important for nematodes to have effective ROS-scavenging mechanisms. A number of ESTs encoding proteins which detoxify reactive oxygen species like superoxide dismutase (SOD) [GenBank: FK670240], Ras-related protein [GenBank: FK670252], and glutathione S-transferase (GST) (GenBank: FK670241) were expressed in P. murrayi in response to desiccation (Table 2). The SOD enzymes are a family of metalloenzymes responsible for quenching the potentially deleterious effects of superoxide radicals. There was abundant expression of ESTs similar to C. elegans sod-1 that encode a copper/zinc superoxide dismutase (Table 3). Three transcripts encoding glutathione s-transferase 1 were expressed in the ESTs of P. murrayi and one of them was up-regulated following desiccation stress. GSTs are a diverse super-family of multifunctional proteins that play prominent roles in detoxification metabolism in nematodes 72. More than a dozen different GSTs have been isolated from C. elegans 73 and these detoxifying enzymes are reported to be involved in several functions, including xenobiotic detoxification and oxidative stress tolerance 7274. Differential expression and up-regulation of detoxifying enzymes like Pm-sod-1 and Pm-gst-1 suggests that P. murrayi has efficient ROS scavenging mechanisms under desiccation stress.

In this study, we identified several ESTs encoding proteins involved in transport facilitation. A total of 15 genes encoding ion and water transporters included ABC transporter proteins (GenBank: GH229101), water channel proteins (GenBank: GH196899), ATPase (GenBank: FG618924) and lipid transfer proteins (LTP) [GenBank: FK670243]. Though the mechanism of coupling ion and water flow through membrane channels is not well studied, there is evidence that up-regulation of such genes is correlated with sensitivity to different types of stress in nematodes 75. Sixty ABC transporters have been identified and functionally characterized in C. elegans. Members of this protein family are responsible for resistance to heavy metals [76]. In addition to short-term response and regulatory mechanisms, a functional system for reestablishing homeostasis is vital to desiccation tolerance. We found transcripts encoding proteins involved in ion homeostasis, such as vacuolar H+-ATPase and ATP synthase [GenBank: FK670298], showing that P. murrayi has efficient homeostatic pathways. Desiccation tolerance involves changes in the levels and composition of fatty acids of the major glycerolipids in nematodes [21]. Our data showed that gene encoding non-specific LTP was differentially regulated following desiccation, suggesting an active lipid metabolism and desiccation resistance.

In general, we observed larger changes in gene expression using qRT-PCR, likely reflecting the greater dynamic range of detection and sensitivity of this method for gene expression profiling. Desiccation stress caused significant up-regulation of transcripts encoding stress related genes like Pm-lea (late embryogenesis abundant protein) and Pm-tps (trehalose-6-phosphate synthase) and down-regulation of Pm-afp (antifreeze protein). Similarly, there was significant up-regulation of the ROS scavenging enzyme Pm-gst-1(glutathione s-transferase 1) and metabolism related genes like Pm-gpx (glutathione peroxidise) and Pm-ms (malate synthase), and down-regulation of Pm-afp and Pm-gsy (glycogen synthase). Similar gene expression patterns, except for antifreeze protein, were reported for the free-living mycophagous nematode A. avenae 20 and insect parasitic nematode Steinernema feltiae 26 during desiccation acclimation and survival (Fig. 3). Although the observed changes in gene expression are common to many nematodes (except for the antifreeze protein and heat shock proteins), we observed significant variability in the magnitude of transcript abundance in P. murrayi. It is likely that the expression of these genes in response to desiccation stress is influenced by the thermal history of the organism, including seasonal variation experienced in the field, and by phylogenetic constraints.

Live nematodes (P. murrayi) from Antarctica were reared in culture in order to produce sufficient quantities for experiments. Nematodes were cultured in sand agar plates containing Bold Modified Basal Freshwater Nutrient Media (BMBFN) (Sigma Aldrich Inc. USA) and stored at 15°C until used. Briefly, 20 ml/L BMBFN solution was mixed with 15 g/L Bacto-Agar in deionised water and autoclaved. The autoclaved Agar media was poured into Petri dishes and Standard Ottawa Sand (EMD Chemicals Inc., Gibbstown, NJ) was spread over each plate before the media solidified. About 5–10 nematodes were released into each plate, and a thin layer of water was maintained on the surface of the media for the duration of culture. The plates were incubated at 26°C for one to two weeks followed by 16°C for 3–5 weeks. In the laboratory culture at 16°C, nematodes completed their life cycle in about 6–8 weeks. Nematodes were also cultured in Nematode growth medium (NGM) plates with Escherichia coli OP50. About 15–20 nematodes were placed in one NGM plate and incubated at 16°C for about 5–8 weeks. Nematodes were sub-cultured every 3–4 weeks by chunking the agar with a sterile scalpel and moving a chunk of agar from an old plate to a fresh plate containing a lawn of E. coli OP50. To harvest the nematodes, the Agar media was cut into small pieces and poured into #40 standard testing sieve (Advantech Manufacturing, New Berlin, WI). Agar was removed by sieving and nematodes were washed twice with deionized water before collection by centrifugation.

Nematodes (about 3,000–5,000) were desiccated in glass desiccation chambers. The relative humidity (RH) was controlled by using saturated salt solutions as described by Winston & Bates 81 with some modifications. The requisite RH was maintained at 23°C ± 1°C in the desiccation chamber for 3 days prior to the addition of nematodes for equilibration. Relative humidity was maintained as 100% RH with distilled water vapour; 97% with K₂SO₄, 87% RH with HCl and 75% with NaCl. The treatments were an initial desiccation at 97% RH for 3 days followed by exposure to 87% for 2 days. All experiments were repeated under identical conditions using mixed stage populations of P. murrayi.

The desiccated and control nematodes were transferred to 10 volumes of Trizol Reagent (Molecular Research Center Inc., Cincinnati, OH) and exposed to freeze thaw cycles using liquid nitrogen and a 37°C water bath. The suspension was ground using mortar and pestle and vortexed. RNA was phase separated using chloroform, precipitated by isopropanol and pelleted. Total RNA was quantified and quality checked by spectrophotometer (NanoDrop, ND-1000 Spectrophotometer) and running agarose gel with RNA Century™ Plus Marker (Ambion Inc., Austin, TX).

Total RNA was extracted from the gradually desiccated nematodes and used as starting material for complementary deoxyribonucleic acid (cDNA) library construction. A full-length cDNA library was prepared using Creator™ SMART™ cDNA library construction kit (Clontech, Mountain View, CA) according to the manufacturer's protocol. Briefly, 2 μg of total RNA was reverse transcribed to cDNA using SMART™ IV Oligonucleotide and PowerScript Reverse Transcriptase and amplified using long distance (LD) PCR. The cycling parameters were 95°C, 20 s; two 20 cycles of 95°C, 5 s; and 68°C, 6 min. cDNA was analyzed using agarose gel electrophoresis, proteinase K digested and restriction digested by SfiI. cDNA was size fractionated using CHROMA SPIN-400 columns and analyzed via agarose gel electrophoresis. The size fractionated cDNA was directionally cloned into pDNR-LIB vector and incubated at 16°C for 15 h. Transformation was done at 37°C overnight, and colonies were picked and grown for 18 h. Template DNA was extracted using QIAprep Spin Miniprep Kit (Qiagen Inc., Valencia, CA), amplified, cleaned using ExoSAP-IT^® (USB Corporation, Cleveland, Ohio) and sequenced using an ABI PRISM 377 automated DNA sequencer (Applied Biosystems, CA) at the DNA Sequencing Center (DNASC) at Brigham Young University (BYU). Sequencing reactions were performed with 2 μl of ABI PRISM BigDye Terminators v3.0 (Applied Biosystems) and 3.5 pmol of primer in 12 μl reaction volumes, followed by a sequencing reaction clean up to remove residual dye and enzyme. Sequencing was with M13 forward (GTAAAACGACGGCCAG) and reverse (CAGGAAACAGCTATGAC) primers for 25 cycles of 96°C, 10 s; 50°C, 5 s; and 60°C, 4 min. The sequenced products were sorted and analyzed with Sequencher™ 4.8 (Gene Codes Corporation, Ann Arbor, MI).

Subtractive hybridization was performed with the PCR-Select™ cDNA subtraction kit (Clontech, Palo Alto, CA). Random hexamer primers (Clontech) were used to convert mRNA that had been extracted from desiccated and control nematodes into cDNA. The cDNA under investigation was designated tester, and the reference cDNA was designated driver. In the forward run cDNA from desiccated samples was designated as tester, in a reverse run control nematode cDNA was used as tester to investigate both up- and down-regulation of mRNA molecules. Second strands of driver and tester DNA were synthesized using T4 DNA polymerase and restriction digested by RsaI to create blunt-ended double stranded cDNA for subtraction. The tester cDNAs were subdivided into two portions and each group ligated to different adaptor sequence and subjected to two levels of subtractive hybridizations as described in the manufacturer's protocol. The DE cDNAs were selectively amplified by two PCR reactions using 50× Advantage cDNA Polymerase mix. The first PCR amplified double stranded cDNA with different adaptor sequences and the second PCR further reduced background and enriched for DE sequences. The PCR-amplified cDNA fragments generated by SSH were then ligated into the TOPO^® TA Cloning^® vector (Invitrogen Corporation, USA) and sequenced using ABI 3730xl DNA Analyzer (Applied Biosystems, CA) at the BYU DNASC.

Base calling was performed using PHRED software (versions 0.000925.c) 8283 with the quality cut-off set at PHRED 20. Raw sequences were then imported into the Vector NTI Advance™ 10 software (Invitrogen Corporation) and subjected to trimming of vector sequences and 5' adapter sequences (for subtractive hybridization) using default settings. Sequences with less than 100 quality bases (PHRED 20 or better) after trimming were discarded. Sequences having polyA tails of 100 bases or more were eliminated from analysis. EST sequences representing contamination from bacterial, yeast or fungal sources were identified using BLASTN algorithm 3435 and removed from further analyses. ESTs were aligned and assembled into contigs using CAP3 software 84 when the criterion of a minimum identity of 95% over 50 bp was met. When an EST could not be assembled with others in a contig, it remained as a "singleton". The contigs and the singletons should thus correspond to sequences of unique genes. The consensus sequences of the contigs and the sequences of the singlets were compared to the sequences in GenBank's non-redundant (nr) and Uniprot database using the TBLASTX and the BLASTX 3839 algorithms and C. elegans Wormpep 190 database 36 using the BLASTX algorithm 35. The cut-off for sequence similarity was E-value < 10^-5 for all analyses.

Gene ontology (GO) term annotation and function-based analysis 85 of unique sequences were performed using Blast2go (V 1.6.2) 86, a sequence-based tool to assign GO terms, extracting them for each BLAST hit obtained by mapping to extant annotation associations. GO terms for each of the three main categories (biological process, molecular function, and cellular component) was obtained from sequence similarity using the application default parameters. From these annotations, pie charts were made using 2nd level GO terms based on biological process, molecular function, and cellular component. Pathway assignments were carried out according to Kyoto encyclopedia of genes and genomes (KEGG) mapping 43. Enzyme commission (EC) numbers were assigned to unique sequences that had BLASTX scores with a cut-off value of E = 10^-5 or less upon searching protein databases. The sequences were mapped to KEGG biochemical pathways according to the EC distribution in the pathway database.

All the primers used in quantitative real-time PCR (qRT-PCR) were designed with IDT SciTools (Integrated DNA Technologies, Coralville, IA) by aligning EST sequences with similar sequences from NCBI. All the primers used in this study were synthesized by Operon (Operon Biotechnologies Inc., Huntsville, AL). The sequences of the primers and product sizes are listed in Additional file 3.

Total RNA extracted from dehydrated and control P. murrayi nematodes was reverse transcribed using ImProm-II™ reverse transcriptase (Promega corporation, Madison, WI) and subjected to qRT-PCR analysis using Light cycler 480 SYBER Green I mastermix and gene specific primers in a Light cycler 480 RT-PCR system (Roche Applied Science, Mannheim, Germany) equipped with light cycler 480 software. High-resolution gel electrophoresis was used to verify that the qRT-PCR amplification product from each examined gene was a single-band product. Thermal cycling was performed in accordance with the manufacturer's instructions for a total of 40 cycles at an annealing temperature of 58°C for each primer pair. Quantitative RT-PCR analysis was performed with Lightcycler 480 software, the threshold cycle was automatically calculated by the second-derivative maximum method, and the copy number of the specific mRNA in the experimental samples was calculated by extrapolation from the gene-specific standard curve.

The copy number of specific cDNA molecules present in the samples was determined by absolute quantification method of qPCR analysis 45. A range of six dilutions (10⁷–10² copies) of the cDNA was made and a gene-specific external standard curve was generated by using cDNA standards that were run simultaneously with the experimental samples. Change in target gene expression was calculated using equation 2^-ΔΔCT 87. The fold change in the target gene, normalized to 18S rRNA (Pm-18s) and relative to the expression of control, was calculated for each sample. A gene with a relative abundance of one is equal to the abundance of 18S rRNA in the same sample. An F-test at a significance level of P < 0.05 was used to compare the ratio of the mean gene expression of desiccated samples with that of the control. To minimize mRNA quantification errors, genomic DNA contamination biases and to correct for inter-sample variations, we used 18s ribosomal RNAs (rRNAs) of P. murrayi as an internal control.