Previous studies have suggested biological roles for HPL-2 at larval stages of C. elegans growth, while no obvious defects in embryonic development were observed in its absence 24 . To detect changes relevant for the hpl-2 phenotypes, we carried out DNA microarray experiments comparing expression profiles of wild-type and hpl-2 loss-of-function worms grown at 20°C and synchronized at the embryonic or L3 larval stages.

To analyze the microarray data, we applied a statistical threshold with a cutoff of P < 0.05 (empirical Bayes statistics) and a fold change > 1.5 for mutant versus wild-type samples. The curated list of genes showed a total of 1,794 embryonic transcripts and 1,160 L3 stage transcripts that were misregulated at least 1.5-fold in the hpl-2 mutant background compared to wild-type animals (Additional file 1; NCBI [GSE33700]). To further validate the array findings, 12 genes with different developmental expression patterns and different biological functions were selected as targets for RT-PCR (Figure 1a). For all 12 genes, a good correlation between microarray data and quantitative PCR (qPCR) expression values was observed. Of the total number of responsive genes (2,954), only 5.6% were found to be misregulated at both developmental stages (Figure 1b), confirming that HLP-2 has distinct biological roles at different stages of development. Although the vast majority of commonly regulated genes encode proteins of unknown function, a notable exception is lin-13, which was up-regulated in both embryos and larvae. LIN-13 is a zinc finger protein required for HPL-2 recruitment to nuclear foci in embryos 23 . This indicates that a complex feedback control exists between HPL-2 and the important lineage factor LIN-13. Analysis of lin-13 expression in vivo using a green fluorescent protein (GFP) translational fusion revealed strongly increased GFP expression in hpl-2 mutant animals compared to wild type, further validating microarray results (Additional file 2).

In embryos, significantly more genes were down-regulated than up-regulated (1,344 and 450, respectively) in the absence of HPL-2, while approximately the same number of genes were found in each class in L3 larvae (615 and 545, respectively; Figure 1b). Although some of the down-regulated genes are likely to be indirect targets, these results suggest that HPL-2 may also play a significant role in gene activation, as reported for certain other HP1 family proteins in both Drosophila and mammalian cells 5 11 13 31 32 .

To analyze the responsive genes in more depth, they were divided into functional categories based on gene ontologies (GOs; Table 1). In embryos, up-regulated genes are slightly enriched in GO terms associated with lipid metabolic processes (Fisher exact test (FET) P = 0.043), while down-regulated genes are enriched in GO terms associated with RNA metabolic processes (FET P = 2 × 10^-6). L3 up-regulated genes are significantly enriched in GO terms associated with meiotic functions (FET P = 4 × 10^-6). The expression of many of these same genes is enriched in the germline (Additional file 3a) and is likely to reflect the role of HPL-2 and other chromatin-associated proteins in the repression of germ cell fate in the soma 28 33 . However, as our microarray analysis was performed on whole animals, some of these genes might also be up-regulated in the germline, where HPL-2 plays a role in germline transgene silencing and the related phenomenon of germline co-suppression 22 34 .

The genes that are down-regulated in L3 larva by loss of HPL-2 activity are enriched in GO terms associated with alcohol metabolic processes (FET P = 0,003), transport (FET P = 0,003), and lipid metabolism (FET P = 0,004). Intriguingly, visual inspection of the data revealed that four classes of genes (abu, nas, fipr, nspb) that are specifically down-regulated at this larval stage are tightly grouped together in mountain 29 of the C. elegans three-dimensional topographical expression map 35 (Additional file 3c). These genes may, therefore, be coordinately regulated by HPL-2. Because all of the phenotypes associated with hpl-2 loss of function concern post-embryonic development, we specifically focused our attention on genes showing altered expression at the L3 stage, as we considered these genes as those most likely to be directly responsible for stage-specific phenotypes.

Comparison of our expression data with HPL-2 ChIP-on-chip profiles obtained in wild-type L3 larvae (available through the modENCODE Consortium 29 ) revealed that a significant number of genes up- and down-regulated in the absence of HPL-2 are bound by HPL-2 (Additional file 4; up-regulated genes FET = 6.7 × 10^-8; down-regulated genes FET = 1.2 × 10^-4). ChIP-qPCR analysis confirmed HPL-2 enrichment on a subset of these genes (Figure 2). HPL-2 binding was found both within promoter and coding regions. No correlation was found between promoter or internal binding and either repression or activation of the bound gene in the absence of HPL-2. Therefore, HPL-2 may regulate gene expression by more than one mechanism and in a context-dependent manner, as observed for other HP1 family proteins 7 14 36 .

In most organisms HP1 binds methylated H3K9 1 2 . This interaction appears to be conserved in C. elegans, as heterochromatin-like transgenic arrays enriched in methylated H3K9 require HPL-2 for repression 22 37 38 . ChIP analysis confirmed that both H3K9me3 and HPL-2 are enriched on these arrays (Additional file 5). We therefore compared our expression data with ChIP-on-chip profiles of H3K9me1/2/3 in wild-type L3 larvae, also available through the modENCODE Consortium 29 . We reasoned that this approach might help identify genes whose expression is directly regulated by the binding of HPL-2 to methylated H3K9. For each gene list, we calculated the median H3K9me1/2/3 signal over the gene body or the promoter and compared this value to the values obtained for expression-level matched genes. We then plotted the distribution of these median signals as box-whisker representation. In this representation, the bar indicates the median for distribution over genes and the upper and lower edges of the box represent the 25th and 75th percentile, respectively. For the genes that require HPL-2 for full repression, we observe a significantly higher signal for H3K9 mono- and di-methylation, and to a lesser extent tri-methylation, compared to expression-matched promoters and gene bodies from wild-type cells (Additional file 6; Wilcoxon test P = 10^-89, 10^-83, 10^-34, respectively, for promoters, 10^-44, 10^-39 and 10^-18, respectively, for gene bodies). This suggests that at least part of the derepression in hpl-2 mutant larvae is indeed direct, that is, due to HPL-2 binding to methylated H3K9. For genes that show lower expression levels in the hpl-2 mutant, the correlation with methylated H3K9 was much weaker, suggesting indirect effects on gene expression, or possibly the existence of alternative mechanisms for recruiting HPL-2 to euchromatic sites. One such mechanism may be through the LIN-13 zinc finger protein 23 .

Visual inspection revealed that both up- and down-regulated transcripts in hpl-2 mutant animals show a significant overlap with genes regulated in animals undergoing dauer arrest 39 (Figure 3a, b; FET P < 2 × 10^-6). In a favorable environment, C. elegans develops into an adult through four larval stages (L1 to L4), but in response to starvation or overcrowding, larvae arrest development at the second molt (L2) to form dauer larvae. Dauer animals are non-feeding and stress-resistant, and can remain in this state for months until environmental conditions become favorable again. Dauer formation has been widely used as a model to study the regulatory mechanisms governing morphological change during organismal development 40 . The list of genes whose expression changes during dauer arrest was obtained by comparing expression profiles of wild-type larvae at the L2/L3 larval stage transition to animals undergoing dauer larva formation (pre-dauer L2d/dauer). Our data were obtained from a very similar developmental stage (early L3), which makes the observed overlap particularly relevant. Significantly, we also observed overlap between genes up-regulated during dauer exit 41 and genes down-regulated in the absence of hpl-2 (FET P = 2 × 10^-16; Additional file 3b).

The commonly regulated genes fall into four classes. The first class includes genes encoding collagens and metalloproteases involved in the synthesis and degradation of the extracellular matrix and cuticle physiology 39 42 43 . The second group consists of members of the Hedgehog (Hh)-related-family, which are down-regulated in both data sets. The nematode Hh-related genes encode secreted proteins mostly expressed in the epidermis, which secretes the lipid- and collagen-rich cuticle, and in neurons and neuronal support cells required for the response to the environment 44 45 46 . Consistent with a possible role of these proteins in cuticle formation, inactivation of some Hh-related genes causes cuticle abnormalities and molting defects 45 46 . The co-regulation of Hh-related genes and structural components of the cuticle suggest that these genes could regulate the production of the dramatically different cuticles of dauer and reproductive L3 47 through coordinated regulation by HPL-2. The third class of dauer and hpl-2-sensitive genes comprises a number of transporters whose expression is reduced upon dauer entry. These include members of the ABC transporter superfamily (pgp-5, 7 and 12), and members of the MSF1 Major facilitator superfamily (F09A5.1, F14D7.6 and F53H8.3). Finally, the fourth and last class consists of cytochromes, enzymes involved in protection from toxins, hormone biosynthesis and other functions. Of the 81 members of the P450 family in C. elegans, 8 are present in the hpl-2 microarray. Three of these (cyp-33C9, cyp-14A5, and cyp-37B1) are down-regulated in hpl-2 mutants and in dauer, while two (cyp-25A2 and cyp13A5) are up-regulated in hpl-2 mutants and upon dauer entry 39 48 . Genes from each of these classes are also induced upon dauer exit 41 , suggesting an important function in the regulation of both dauer entry and exit programs. HPL-2 binding and H3K9 methylation could be detected on a number of these dauer-regulated genes, including cdh-10, cdh-12, Y55F3AM.14, and srr-2 49 (Figure 2), suggesting that they are likely to be direct targets of HPL-2.

The above results prompted us to test whether hpl-2 mutants are associated with dauer phenotypes. High temperature (27°C), scarce food and a secreted dauer pheromone influence dauer formation. While higher temperatures alone failed to induce a dauer phenotype in hpl-2 animals, when plates with identical bacterial lawns were seeded with the same number of worms at 25°C, we observed that starved plates containing hpl-2 mutant worms reproducibly had many more dauers than wild-type plates (data not shown). While this is not a quantitative assay, the observation nonetheless suggested that hpl-2-deficient animals enter dauer more easily under starved and crowded conditions at 25°C. We therefore tested whether hpl-2 mutants are more sensitive to dauer pheromone, an indicator of population density produced by the worms, which induces dauer formation at high concentrations.

At pheromone concentrations that trigger dauer formation in 23% of wild-type animals after 48 hours at 25°C, 97% of hpl-2 mutant animals entered dauer (Figure 3c). Therefore, hpl-2 mutants are associated with a hypersensitivity to dauer pheromone, which becomes evident under strong dauer-inducing conditions. When shifted to 20°C following exposure to dauer pheromone, a large majority of wild-type worms exit the dauer stage after 2 to 3 days. In contrast, only 15 to 20% of hpl-2 mutant worms exit dauer under the same conditions. We conclude that hpl-2 mutants exhibit both dauer entry and exit defects, namely hypersensitivity for dauer entry and a failure to exit efficiently.

The second C. elegans HP1 homologue, hpl-1, acts redundantly with hpl-2 in postembryonic development 24 . However, hpl-1 does not appear to act either alone or redundantly with hpl-2 in dauer development, since dauer formation was not impaired in either hpl-1(tm1624) single or hpl-1(tm1624);hpl-2(tm1489) double mutants at any of the temperatures tested.

We next tested whether hpl-2 interacts genetically with either of the well-characterized dauer pathways: DAF-2/insulin-like signaling (IIS) and DAF-7/TGF-β. Mutations in either of these pathways have been isolated as dauer constitutive (daf-c), forming dauers even under non-dauer-inducing conditions 50 . We first constructed double mutants with the daf-2(e1370) allele, a well-established sensitized background for assaying a weak dauer phenotype 51 . While daf-2 single mutants do not form dauer at 20°C, 10 to 20% of double hpl-2;daf-2 mutants entered dauer at this temperature (Figure 3d, 20°C). This difference between single and double mutants was not detected at 22°C, a temperature at which single daf-2(e1370) mutants all enter dauer for approximately 3 days, then recover and resume development to become reproductive adults (data not shown). However, 3 days after egg-laying, when only 30% of the daf-2 single mutants remained as dauers, 98% of hpl-2;daf-2 double mutants were still in the dauer stage (Figure 3d, 22°C). Under these conditions, the block persisted for more than two weeks. At a slightly higher temperature, 25°C, 100% of the population remained in the dauer stage in both genetic backgrounds after 3 days (Figure 3d, 25°C). We conclude, therefore, that hpl-2 significantly enhances the dauer phenotype of daf-2 mutants.

Like daf-2(e1370) mutants, daf-7(e1372) mutants show a temperature-sensitive dauer constitutive phenotype 52 . At the semi-permissive temperatures of 15°C and 20°C, hpl-2;daf-7 double mutant animals show a greatly enhanced dauer arrest phenotype compared to single daf-7 mutants (Figure 3e). Our results thus suggest that HPL-2 acts in parallel or downstream to the IIS and TGF-β pathways to influence dauer through the regulation of genes important for both dauer entry and dauer exit processes. Given that this is a stress-induced differentiation event, our results implicate HPL-2 in communication between environment and a chromatin-controlled genetic response to stress.

The decision to enter dauer is triggered by chemosensory stimuli. Two major head sensory organs, called amphids, are critical to dauer formation and both DAF-28/insulin and DAF-7/TGF-β are expressed in a subset of sensory neurons to impact dauer development 53 54 55 . These hormones in turn regulate the activity of widely expressed downstream receptors and transcription factors, which regulate dauer remodeling in other tissues. To establish in which tissue HPL-2 acts to influence dauer development, we carried out tissue-specific rescue experiments on hpl-2;daf-2(e1370) animals raised at 22°C (Figure 3f). While at this temperature all daf-2 animals reenter the reproductive cycle after 7 days, less than 1% of the double hpl-2;daf-2 mutants ever exit the dauer stage. Expression of hpl-2 under the control of either the pan neuronal rab-3 promoter or the intestinal ges-1 promoter allowed more than 30% of hpl-2;daf-2(e1370) mutant animals to exit the dauer stage. By contrast, expression of hpl-2 under the control of the lin-31 promoter, which is expressed in vulval precursor cells, failed to promote rescue. Furthermore, we confirmed that GFP expression of the rab-3p::GFP::hpl-2 rescuing transgene was limited to neuronal cells in hpl-2;daf-2 mutants (Additional file 7). These results suggest that the rescue observed with the rab-3 and ges-1 promoters is unlikely to be due to ectopic expression in additional cell types, but reflects the tissue-specific expression of these transgenes. The fact that rescue is partial could stem from insufficient induction of these promoters during dauer and/or from additional expression requirements in other tissues. Neuronal and intestinal expression together did not result in better rescue, however, suggesting that in these tissues HPL-2 does not have additive functions, but acts in the same process.

Our results suggest that hpl-2 synergizes with the IIS pathway to control dauer development. Given that the IIS pathway is also a well-established regulator of lifespan in C. elegans, Drosophila and mice 56 , we performed lifespan analysis of hpl-2 null mutants at the permissive temperature (20°C). We observed a reproducible increase in the mean lifespan of hpl-2 animals by 17% on average (average lifespan: hpl-2(tm1489), 15.1 ± 0.31 days (n = 287); wild-type, 12.9 ± 0.25 days; log rank test P < 10^-3; Figure 4a). To confirm that the increased lifespan is due to inactivation of hpl-2 and not to a background mutation, we repeated lifespan assays of animals in which hpl-2 was inactivated by RNAi feeding. hpl-2(RNAi) only partially inactivates hpl-2 activity and compared to hpl-2 loss of function results in weaker phenotypes only apparent at higher temperatures 22 23 . However, similarly to the hpl-2 mutant, longevity of hpl-2(RNAi) animals raised at 20°C and shifted to 25°C at the L4 stage was extended by 18% on average compared to control animals (Additional file 8), confirming that HPL-2 is required for normal longevity. Lifespan extension in the absence of HPL-2 is unlikely to be related to a germline proliferation defect, as hpl-2 mutants have a wild-type brood size and germline cell number at 20°C 22 23 (Additional file 9).

To assess whether the lifespan increase observed in hpl-2 mutants depends on the IIS pathway, we analyzed the lifespan of hpl-2;daf-2 double mutants. The two-fold increase in lifespan of daf-2(e1370) mutants was unchanged in hpl-2;daf-2 animals (average lifespan: daf-2(e1370), 26.2 ± 0.85 days (n = 95); daf-2(e1370);hpl-2(tm1489), 24.8 ± 0.97 days (n = 84); log rank test P = 0.271; Figure 4b). Therefore, the hpl-2 lifespan phenotype may rely on IIS signaling. The transcription factor DAF-16/FOXO acts downstream of daf-2/IIR and is essential for lifespan extension by daf-2. We therefore tested whether the hpl-2 mutant lifespan phenotype also requires DAF-16/FOXO. hpl-2(tm1489);daf-16(mu86) mutants had a lifespan indistinguishable from that of daf-16(mu86) alone, which is 16% shorter than wild-type worms (average lifespan: daf-16(mu86), 11.9 ± 0.44 days (n = 112); daf-16(mu86);hpl-2(tm1489), 11.03 ± 0.27 days (n = 109); log rank test P = 0.05; Figure 4c). Furthermore, the hpl-2 effect on longevity is not due to altered expression of genes encoding proteins involved in the IIS pathway, as the levels of daf-16 and other major components of the pathway, including daf-2, age-1, akt-1, akt-2 and daf-18, were not altered in hpl-2 mutants (Additional file 1). Taken together, our results suggest that hpl-2 requires the activity of daf-16 to modulate longevity in C. elegans and that hpl-2 may be a novel co-regulator of daf-16-controlled genes.

To better define this relationship, we looked for overlap with the 514 genes that microarray analysis by Murphy et al. 57 found to be modulated by DAF-16; 46 of the genes that we found to be misregulated in the absence of hpl-2 activity overlap with genes regulated by DAF-16. Of these, 16 are genes repressed in hpl-2 and in long-lived daf-2 pathway mutants, and activated in daf-2;daf-16 mutants, reflecting DAF-16 dependence (Additional file 2e; P = 0.003). We also compared our dataset to a recent transcriptional profile to identify genes whose expression changes over the C. elegans lifespan 58 . Out of 1,217 ageing-associated genes, 66 were down-regulated in the absence of hpl-2, representing a significant enrichment (Figure S2f; P = 5 × 10^-7). Taken together, these analyses suggest that the effect of hpl-2 on longevity may be mediated, at least in part, through the regulation of longevity-associated genes, some of which may be DAF-16-dependent.

Similar to mammalian FOXO transcription factors, translocation of DAF-16 from the cytoplasm to the nucleus is required for activation, and has been observed in animals with reduced DAF-2/IIS signaling and in response to stresses such as heat shock or starvation 59 . Using a previously described DAF-16::GFP reporter 60 , we observed no change in DAF-16 localization in animals lacking HPL-2 (data not shown and Additional file 10). Therefore, HPL-2 does not appear to impinge on longevity by altering DAF-16 localization.

Many long-lived mutants also exhibit increased resistance to environmental stress. However, hpl-2 mutants are not significantly more resistant to oxidative stress, as measured on paraquat (Additional file 11), and are intrinsically thermosensitive. Therefore, the increased longevity of hpl-2 mutants appears to be uncoupled from increased resistance to environmental stressors, as already observed in other contexts 55 61 62 63 64 . Consistently, we did not observe increased expression of genes commonly associated with increased stress resistance - for example, the mtl-1 metallothionein gene and the mitochondrial sod-3 superoxide dismutase gene - while expression of the heat shock proteins hsp-16 and hsp-12 actually decreased 65 66 67 (Additional file 1).

Adipose tissue has been implicated in the regulation of longevity in a number of species 68 69 70 . In C. elegans, lipid hydrolysis in fat storage tissue was recently shown to prolong life span, connecting the metabolic functions of adipose tissue to lifespan control 71 . Genes associated with fatty acid metabolism are over-represented in the set of hpl-2 down-regulated genes (Table 1), prompting us to test whether hpl-2 mutants may have defects in lipid metabolism that could contribute to the longevity phenotype. Down-regulated genes include the fat-7 Δ9 fatty acid desaturase, acs-2, an acyl-CoA synthetase, and ech-1, an enoyl-CoA hydratase. hpl-2 mutants raised at the non-permissive temperature are transparent, resembling starved animals (data not shown). This 'clear' phenotype has been associated with decreased fat storage in the intestinal compartment 72 . However, using the neutral lipid dye Oil red O 73 , no reproducible difference in staining intensity was observed between wild-type and hpl-2 mutant animals at either permissive or non-permissive temperatures (Figure 5a, b and data not shown), ruling out any gross alteration in lipid content.

Quantitative lipid analysis following solid phase extraction 74 revealed more subtle, highly reproducible differences in lipid composition in hpl-2 mutant animals at the permissive temperature. Consistent with the Oil red O staining results, no difference in the total content and composition of neutral lipids was observed and levels of triacylglycerides and fatty acids were unchanged in hpl-2 animals compared to wild type (Additional file 12 and data not shown). In contrast, phospholipid analysis showed significant changes in phospholipid content (Figure 5d). Higher amounts of phosphatidylserine (PS) were observed in hpl-2 mutant animals, whereas no increase in phosphatidylethanolamine (PE), the decarboxylation product of PS, was observed. These results may therefore reflect changed expression or function of PS decarboxylase leading to lower synthesis of PE. Phosphatidylcholine (PC) is the predominant membrane lipid in eukaryotes and also functions in signal transduction pathways. Although amounts of PC were slightly reduced in hpl-2 mutant worms, the fact that we do not observe any obvious defects in hpl-2 mutants at the permissive temperature suggests either that the observed decrease in PC is not large enough to impinge on PC mediated processes, or that compensatory mechanisms exist.

Of the different classes of phospholipids analyzed, the most dramatic change was observed for sphingomyelin (SM), whose content was reduced by more than 60% in hpl-2 mutants. SM is a vital component of cellular membranes and plays an important role in signal transduction and membrane trafficking 75 . Synthesis of SM involves a transfer of phosphorylcholine from PC to ceramide by sphingomyelin-synthases (SMSs; Figure 5E). Expression of lagr-1, encoding a ceramide synthase (CerS), is decreased in hpl-2 mutant animals. This could potentially result in lower levels of ceramide, thereby influencing SM levels. However, the total content of ceramides (hydroxylated and nonhydroxylated) was not affected in the hpl-2 mutant context (Additional file 12). Therefore, the decrease in SM levels is more likely to be due to the down-regulation of de novo SM synthesis rather than enhanced degradation. C. elegans contains three SMS homologues, one of which (sms-3) is down-regulated in hpl-2 mutants. qRT-PCR analysis confirmed that sms-3 RNA levels are decreased approximately two-fold with respect to wild type in L3 animals, with a smaller effect in young adults (Figure 5f). Similarly, sms-1 levels also appeared to be slightly reduced in L3, but no significant change in sms-2 RNA levels was observed, nor were changes in sms-3 or sms-1 levels observed at the embryonic stage. As shown in Figure 2, ChIP experiments revealed a small but reproducible enrichment of HPL-2 binding within the sms-3 coding regions in L3 larvae. Therefore, it is likely that HPL-2 directly influences SM levels through the regulation of sms-3 transcription.

Strains were maintained according to the standard protocol 101 . The following mutant alleles and transgenic strains were used: LGI, daf-16(mu86); LGIII, hpl-2(tm1489), daf-2(e1370), daf-7(e1372); LGIV, zIs356 (daf-16::gfp). The Bristol strain N2 was used as wild type. To construct the hpl-2(tm1489); zIs356 (daf-16::gfp) strain, zIs356 (daf-16::gfp) hermaphrodites were mated with hpl-2 (tm1489) males. F2 worms were screened for the presence of GFP, and the presence of the hpl-2 deletion allele was confirmed by PCR analysis as previously described 23 .

Wild-type and hpl-2(tm1489) gravid adult animals were synchronized by bleaching. Half of the culture was collected immediately for RNA extraction from the embryos. The other half was left to hatch overnight in M9 without food. Synchronized L1 larvae were then distributed onto peptone plates, grown at 20°C and staged until they reached the L3 larval stage (between 28 and 29 hours). The developmental stage was confirmed by microscopic observation, using both vulval development and somatic gonad development as criteria.

Triple biological replicates were flash frozen in liquid N2 and RNA was extracted using the standard Trizol method followed by RNeasy column purification (Qiagen, France, #74104). RNA was labeled and hybridized to Affymetrix tiling arrays. Washing and scanning were performed following the manufacturer's instructions. Oligos from the tiling array were mapped to chromosome coordinates of the exons from Wormbase WS180. Any oligo that mapped to a gene on both the Watson and Crick strands was excluded. The remaining oligos were then grouped together (perfect match and mismatch) into probe sets and written out into an Affymetrix CDF file. The CDF file was converted into an R-package and loaded into R. The expression values were calculated using the justRMA function from Bioconductor. This used a Benjamini and Hochberg false discovery rate correction. The R code is available upon request.

cDNAs were synthesized with a mix of random hexamers and oligo dT primers using iScript cDNA synthesis kit and 500 ng of RNA (BIO-RAD, France). qPCR reactions on reverse transcription products diluted ten times were performed using FastStart Universal SYBR Green Master (Roche, France) and the StepOne Plus Detection System (Applied Biosystems, France). The qPCR conditions were 95°C for 4 minutes, followed by 40 cycles of 12 s at 95°C and 30 s at 60°C. Melting curve analysis was performed for each primer set at the end to ensure the specificity of the amplified product. Standard curves were generated for each primer set to calculate the efficiency of each set. Only primer sets with an efficiency between 1.9 and 2.1 were used for qPCR. Primer sets for the pmp-3 and cdc-42 transcripts were used as internal controls, and the RNA level of each gene of interest was normalized to the level of the mean of both these genes for comparison. qPCR experiments were repeated at least three times using independent RNA/cDNA preparations. Data were pooled and analyzed using the StepOne Plus Software (Applied Biosystems). Primer sequences are available upon request.

The P-values for overlaps between lists were generated by comparison to results of random simulation and by calculation based on the hypergeometric distribution as in 102 , using R.

For intestinal expression, the hpl-2 (K01G5.2a) full-length cDNA was cloned into the XbaI-SacII sites of pJM16 under the control of the ges-1 promoter 103 . For pan-neuronal expression, the cDNA was cloned into the NotI-SmaI sites of pHU004, under the control of the rab-3 promoter in fusion with GFP (gift from JL Bessereau). Both intestinal-specific and neuronal-specific constructs, named pCB15 and pSS05, respectively, were injected into N2 hermaphrodites at a concentration of 5 mg/ml using pRF4 as a co-transformation marker at a concentration of 200 mg/ml. For each construct, two to three independent transgenic lines were then crossed with hpl-2(tm1489);daf-2(e1370) males. F2 rollers were allowed to lay eggs at 15°C, then shifted to 25°C. The F3 were selected for both 100% dauer in the progeny and the presence of the hpl-2 deletion allele.

Lifespan tests were performed at 20°C as described in 104 . For each genotype, lifespan experiments were repeated two to four times. Survival analysis was carried out using the Kaplan Meier method and the significance of differences between survival curves calculated using the log rank test. The statistical software used was XLSTAT 2007 and all P-values < 0.05 were considered significant.

The ChIP protocol was based on 105 , with a number of modifications. L3 staged wild-type or transgenic animals carrying an integrated rescuing hpl-2::GFP construct 23 were used as starting material. Worm powder was incubated in phosphate-buffered saline containing 1.5 mM EGS for 20 minutes at room temperature, followed by the addition of formaldehyde (1% final) for 10 minutes at room temperature. After quenching with glycine (125 mM final), the worm pellet was washed in phosphate-buffered saline, resuspended in lysis buffer (50 mM HEPES/KOH pH 7.6, 1 mM EDTA, 0.5% sarkosyl, 150 mM KCl, protease inhibitors cocktail (Roche Biochemicals) and 1 mM PMSF) and sonicated in a Bioruptor water bath (Diagenode, Belgium). The homogenate was centrifuged at 12,000 rpm for 15 minutes at 4°C and the supernatant containing the sheared chromatin (chromatin extract) precleared twice for 30 minutes at 4°C with 80 μl of a 1:1 slurry of protein A agarose beads (Millipore #16157, France). The cleared extract was diluted with 2 volumes of FA buffer (50 mM HEPES/KOH pH 7.5, 1 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate, 150 mM KCl, protease inhibitors cocktail and 1 mM PMSF). Following incubation with antibodies (anti-HPL-2 or anti-GFP), 45 μl of a 1:1 slurry of protein A agarose beads was added to each tube for 1 hour. Beads were washed twice with FA buffer, once with FA-500 (50 mM HEPES/KOH pH 7.5, 1 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate, 500 mM NaCl), once with FA-1M (50 mM HEPES/KOH pH 7.5, 1 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate, 1 M NaCl), once with TEL (0.25 M LiCl, 1% NP-40, 1% sodium deoxycholate, 1 mM EDTA, 10 mM Tris-HCl, pH 8.0), and twice with TE (10 mM Tris/HCl, 1 mM EDTA). Elution was performed by adding two times 100 μl of elution buffer (10 mM Tris/HCl, 1 mM EDTA, 250 mM NaCl, 1% SDS) and incubating at 65°C for 10 minutes. Supernatants were pooled, incubated for 30 minutes with 1 μl of RNase A (10 mg/ml) and then incubated overnight at 65°C with 1 μl of proteinase K (15 to 20 mg/ml). DNA was recovered using a Qiaquick pure purification kit in 40 μl of PCR grade water for qPCR analysis. ChIP results were analyzed by qPCR using primers specific to the sms-3 locus. A primer corresponding to an intergenic region on chromosome IV was used as control. The anti-GFP antibody used was from Abcam, UK (#ab290).

The solid phase extraction procedure was performed as previously described 74 . Briefly, lipids were extracted using the Folch procedure 106 and loaded onto a 100 mg aminopropyl LC-NH2 cartridge pre-equilibrated with 3 ml of hexane, and placed onto a vacuum manifold apparatus (12 port Visiprep, Supelco, Sigma-Aldrich, France). Neutral lipids (sterols, sterol esters, methylated fatty acids, tri- and diacylglicerides; fraction 1) were eluted with 1.4 ml hexane:ethylacetate (85:15 v/v) at a solvent flow rate of 0.3 ml/minute obtained by applying negative pressure. Ceramides (fraction 2) were eluted with 1.6 ml of chloroforme:methanol (23:1 v/v). Free fatty acids (fraction 3) were eluted with 1.8 ml diisopropylether-acetic acid (98:5, v/v), and this fraction was pooled together with fraction 1. Glycosphingolipids (fraction 4) were eluted using 2.1 ml of acetone-methanol (9:1.35, v/v). Neutral and acidic phospholipids were eluted together in one fraction (fraction 5) using 2 ml of methanol. The fractions were dried under N₂ and resolved by thin layer chromatography using the following solvent systems: for separation of neutral lipids and free fatty acids, hexane:diethylether:acetic acid (80:20:1 v/v/v); for ceramides, chloroform:methanol (50:5 v/v); for glycosphingolipids, chloroform:methanol:water (65:25:4 v/v/v); for resolution of phospholipids by two-dimensional thin layer chromatography, tetrahydrofurane:acetone:methanol:water (50:20:40:6 v/v/v/v) in the first direction and chloroform:acetone:methanol:aceticacid:water in the second direction (50:20:10:15:5 v/v/v/v/v). Fractions 1, 2, 3 and 4 were detected after spraying with cupric sulfate and charred at 180°C for 5 minutes. Phospholipids were detected by iodine and scraped into glass tubes. Lipidic phosphorous was mineralized by warming the samples in 0.25 ml mixture of sulfuric acid:perchloric acid (2:1 v/v) containing vanadium tetraoxide (1 g/l) for 20 s above the flame and quantified by 5 ml of reagent containing 0.63 g of ANSA mixture, 0.2 g ammonium heptamolybdate in 100 ml of water (ANSA contains 60 g of sodium metabisulfate, 2 g of sodium sulfate and 1 g of 1-amino 2-naphtol 4-sulfonic acid). Samples were mixed by vortexing, incubated at 100°C for 8 minutes and absorbances were read at 830 nm. Standard curves were made by the use of monopotassium phosphate.

Microarray data are publicly available through NCBI, Gene Expression Omnibus accession number [GSE33700].