Dietary Restriction during Development Enlarges Intestinal and Hypodermal Lipid Droplets in Caenorhabditis elegans

Dietary restriction (DR) extends lifespan in man species and modulates evolutionary conserved signalling and metabolic pathways. Most of these studies were done in adult animals. Here we investigated fat phenotypes of C. elegans larvae and adults which were exposed to DR during development. This approach was named “developmental-DR” (dDR). Moderate as well as stringent dDR increased the triglyceride to protein ratio in L4 larvae and adult worms. This alteration was accompanied by a marked expansion of intestinal and hypodermal lipid droplets. In comparison to ad libitum condition, the relative proportion of fat stored in large lipid droplets (>50 µm3) was increased by a factor of about 5 to 6 in larvae exposed to dDR. Microarray-based expression profiling identified several dDR-regulated genes of lipolysis and lipogenesis which may contribute to the observed fat phenotypes. In conclusion, dDR increases the triglyceride to protein ratio, enlarges lipid droplets and alters the expression of genes functioning in lipid metabolism in C. elegans. These changes might be an effective adaptation to conserve fat stores in animals subjected to limiting food supply during development.


Introduction
Dietary restriction (DR) also known as caloric restriction (CR) is defined as a significant reduction of energy and macronutrient intake in the absence of malnutrition [1]. In adult animals, DR is a fundamental nutritional intervention to reduce body weight and to decrease the risk of common diseases such diabetes type 2 or atherosclerosis [2,3]. It extends lifespan and health-span in many species, ranging from invertebrates to mammals [3]. In recent years, the short-lived model organisms Saccharomyces cerevisiae, Drosophila melanogaster and Caenorhabditis elegans were used to uncover key regulatory factors mediating DR induced longevity. These include the target of rapamycin TOR [4][5][6], the AMP-activated protein kinase AMPK [7] and the Insulin/IGF-1 signaling cascade [8]. In Drosophila, insulin-like signalling converges with DR and seems to be a downstream target of DR [9]. In C. elegans, insulin/ IGF-1 signaling and DR seem to be independent pathways in regulating lifespan [7,[10][11][12]. In addition, several transcription factors including the Forkhead transcription factor pha-4 [13], the Nrf2 transcription factor skn-1 [14] and the heat-shock transcription factor hsf-1 [15] were recognized as important regulators mediating longevity by DR in C. elegans. Of note, most of these pathways were identified in adult animals.
During the last years, C. elegans has been emerged as an important model to study the regulation of energy metabolism and lipid storage. As a great advantage, it enables the examination of the relationship of lipid metabolism, growth, reproduction and lifespan. Many of mammalian metabolic pathways, such as fatty acid (FA) synthesis, elongation and desaturation, mitochondrial and peroxisomal ß-oxidation of fatty acids are conserved in the nematode. A number of genes involved in pathways that regulate lipid homeostasis in mammals are assumed to control lipid storage as well in C. elegans. These include serotonin, insulin, transforming growth factor-ß (TGF-ß) and TOR signalling pathways [16][17][18][19][20]. In addition, many mammalian transcription factors involved in fat accumulation are present in C. elegans. For example, the C. elegans transcription factors SBP-1 and NHR-49 are homologues to the human sterol-regulatory-element-binding protein (SREBP) and peroxisome proliferator-activated receptor-a (PPARa), respectively [18,[21][22][23]. Thus, the worm seems to be an appropriate model to study lipid metabolism. However, only few studies have investigated the influence of DR on C. elegans fat phenotypes. For example, DR in liquid medium or dietary deprivation results in a pale appearance of adult worms suggesting a mobilization of intestinal fat stores [14,21].
A variety of protocols exist to subject C. elegans to DR. In the laboratory, C. elegans is usually fed on Escherichia coli (OP50) lawns cultivated on Nematode Growth Medium (NGM) agar plates. One common strategy to generate DR is the limitation of E. coli growth on solid medium by UV, heat or antibiotic treatment. The reduction of bactopeptone in the NGM agar plates is another strategy to control growth of E. coli [24][25][26]. Cultivation of C. elegans on plates in the absence of bacteria (dietary deprivation) is usually performed during adulthood [15,27,28]. In liquid media, DR is induced by dilution of E. coli in S-Basal medium [11,13,14,29], undefined axenic medium [30] and chemically defined liquid medium [31]. The use of feeding defective mutants such as eat-2 is a further approach to study DR in C. elegans [10]. However, all of these methods were primary established to examine effects of DR during adulthood.
Most studies performing DR in C. elegans have examined the effects and molecular mechanisms of DR on lifespan extension in adult worms. So far, effects of DR applied during development have not been extensively analyzed in C. elegans. Here, we investigated in C. elegans the impact of DR during development. For this purpose, we have optimized a solid medium based DR protocol [24,32,33], in which bactopeptone is excluded from the agar plates and the addition of antibiotics is not required. This DR regime was named ''developmental-DR'' (dDR). It increased the triglyceride to protein ratio and lipid droplet (LD) size in larvae and adult worms exposed to DR during development. To gain an insight into underlying mechanisms of dDR induced phenotypes, whole genome gene expression analysis was performed. Among the dDR responsive genes we found an enrichment of genes being involved in lipid metabolism. These genes might contribute to dDR induced fat phenotypes.

Results
Establishment of a dietary restriction (DR) protocol (developmental-DR, dDR) which allows the application of DR during development from hatching to adulthood Based on reported DR methods [24,32], here we established a modified solid medium based DR protocol in order to study the effect of DR on C. elegans larvae and adults exposed to DR during development ( Figure 1). This method was named ''developmental dietary restriction'' (dDR). It allows a standardized variation of the extent of dDR without starvation, dauer formation or arrest of the animals during development. In brief, adjusted optical densities (250 ml, OD 600 0.3 to 6.0, represents DR 0.3 to 6.0) of the E. coli strain OP50 were spread onto antibiotic-free agar plates and incubated for 16 h at 37uC. For dDR conditions, bactopeptone as sole carbon source for bacterial growth was omitted from the Nematode Growth Medium (NGM). This leads to different amounts of bacteria per agar plate depending on the OD of seeded OP50 (e. g. dDR 1.5: 0.5660.05 OD/plate; dDR 0.7: 0.3260.03 OD/plate). For ad libitum (AL) condition, standard (NGM) was used resulting in a thick bacteria lawn (25.85 (mean)61.19 (SD) OD/plate). To standardize food availability per worm, 500 synchronized embryos were sorted onto AL and dDR agar plates by flow cytometry and were cultivated at 20uC until reaching the L2, L4 or adult stage. To exclude starvation, L4 larvae and adult worms were transferred daily to fresh agar plates. dDR reduces body size without substantial changes of developmental time, locomotion and feeding rate To evaluate our dDR protocol, body proportions of adult worms exposed to DR during development were determined based on bright-field microscopy images (Figure 2A-D, Table  S1). As expected, we obtained an inverse relationship between the extent of dDR (0.3 to 6.0) and reduction of body width ( Figure 2B), length ( Figure 2C) and volume ( Figure 2D) of adult worms. This demonstrates dose dependency of our dDR regimes. Interestingly, the body width decreased to a greater extent than the length (Table S1). At the most stringent dDR condition (dDR0.3), width declined to approximately 68%, whereas length was reduced to about 74% when compared with AL fed worms. The body volume of dDR0.3 fed worms (1.3 nl) decreased by a factor of 3 (34.862.1% of AL). Flow cytometry based TOF (time of flight) and extinction (ext) values, which serve as proxies for length and volume of the worms [34], confirmed the microscopic body proportion measurements in a large number of worms ( Figure 2E/F). Because dDR0.7 and dDR1.5 did not substantial influence the time point of the first egg lay (AL: ,72 h; dDR1.5: ,74 h; dDR1.5: ,76 h), locomotion (Figure S1A-C) and feeding rate ( Figure S2D), these conditions were used for further experiments.
dDR increases the triglyceride to protein ratio in larvae and adult worms To get insight into the effect of dDR on body composition triacylglyceride (TAG) and protein levels were determined using an enzymatic assay in L4 and adult worms exposed to DR during development. In L4 larvae and in adulthood, the TAG content per worm was increased, whereas protein content per worm was reduced under moderate and stringent dDR conditions ( Figure 3A/B). As a consequence, the resulting TAG to protein ratio was increased in L4 larvae and adult worms when compared to control animals ( Figure 3C). In agreement, thin layer Figure 1. Scheme of the established solid medium based dietary restriction method to apply dietary restriction during development (dDR) in C. elegans. Escherichia coli (E. coli) OP50, serving as food source for C. elegans, was cultivated at 37uC until an optical density (OD 600 ) = 1.5 was reached (OD 600 = 1.0<1.0610 9 cfu/ml). Different ODs (6.0, 3.0, 1.5, 1.0, 0.7, 0.3) were adjusted in M9 buffer. Standard Nematode Growth Medium (NGM) was used for ad libitum (AL) condition. For dietary restriction (dDR), bactopeptone was omitted from NGM agar plates to prevent bacterial growth. A defined amount (250 ml) of different OP50 suspensions was seeded on dDR plates, resulting in a gradient of colony forming units (cfu) per plate. For AL plates, an OP50 suspension of OD1.5 was used. Nematodes were synchronized by hypochlorite treatment, and 500 eggs per agar plate were sorted via cytometry-based COPAS Biosort system. Animals were cultivated at 20uC and harvested at second larval stage (L2, 24 h after hatching), fourth larval stage (L4, 48 h) or after reaching adulthood (adult, 72-78 h). At L4 stage, nematodes were transferred once to fresh agar plates. doi:10.1371/journal.pone.0046198.g001 Enlarged Lipid Droplets in C. elegans PLOS ONE | www.plosone.org chromatography revealed a higher TAG to phospholipid ratio in adulthood subjected to dDR ( Figure S2). Thus, biochemical measurements revealed that dDR induced a remarkable shift to a higher TAG to protein ratio in L4 larvae and adult worms.

dDR increases lipid droplet size in larvae and adult worms
We next asked whether the increased TAG to protein ratio was associated with changes in the lipid droplet (LD) appearance. Conventional fluorescence microscopy of fixative BODIPY 493/ . Data for body width (mm), length (mm) and volume (nl) of single worms (n.50 per condition) derived from bright-field microscopy images. The body volume was calculated using a worm adapted cylinder volume formula which includes area and perimeter of single animals. Comparative calculations between AL and dDR treated animals are provided in Table S1. (E, F) Time of flight (TOF, AU) and extinction (AU) values (n.500 per condition) were collected by COPAS Biosort system. Bars represent a mean 6 SEM from two to three independent experiments (***p,0.001). AU = arbitrary unit. doi:10.1371/journal.pone.0046198.g002 Enlarged Lipid Droplets in C. elegans 505-stained adult AL fed worms showed a high density of smallsized LDs ( Figure 4). Interestingly, dDR0.7 and dDR1.5 increased the size of intestinal and hypodermal LDs in worms at the first (Figure 4), second ( Figure S3) and eight ( Figure S4) day of adulthood. LD expansion under dDR was even visible in corresponding bright-field microscopy images (Figure 4, S3, S4) and was confirmed by BODIPY 493/505 based vital staining ( Figure S5) as well as by the Oil red O method ( Figure 4D-F). Moreover, enlarged LDs were also observed in eat-2 (ad465) mutants, a genetic model of DR ( Figure S6). Control experiments revealed no differences in LD size of adult N2 worms exposed during development to AL plates containing peptone or no peptone ( Figure S7). To investigate the LD expansion phenotype in more detail, scanning-laser confocal (SLC) microscopy of fixative BODIPY 493/503 stained worms was performed. The enlargement of intestinal and hypodermal LDs under moderate and stringent dDR condition was observed in L2 larvae ( Figure 5), L4 larvae ( Figure 6) and adult worms ( Figure 7). In summary, DR increased the LD size in intestinal and hypodermal cells of larvae and adult worms which were exposed to DR during development.
dDR increases the relative number of large-sized lipid droplets by a factor of about 2 to 15 depending on the developmental stage To quantify the observed enlargement of LDs in response to dDR, the number of fixative BODIPY-labeled LDs in pharynx and tail regions was calculated from single z-stacks of SLC microscopy images (Figure 8). Small-sized LDs,10 mm 3 in volume represented 88.8 to 97.9% of detected LDs at all developmental stages and at all feeding conditions. In comparison with AL condition, dDR increased the relative number of medium-sized LDs (10-25 mm 3 ) of L2 and L4 larvae by a factor of 2.1 to 3.3. The relative abundance of large-sized (25-50 mm 3 ) and very large-sized (.50 mm 3 ) LDs increased up to 12-fold in dDR L4 larvae compared with AL condition. In tail region of dDR L4 larvae, the relative number of very large LDs was actually up to 15.2-fold higher. The relative number of large-sized LDs (25-50 mm 3 ) was up to 2.5-fold increased in pharynx region of dDR adult worms. The percentage of very large LDs (.50 mm 3 ) was up to 7.1-fold higher in dDR adult worms. Taken together, the extent of dDR-induced LD expansion was dependent on the respective body region as well as on the developmental stage.
Under dDR up to 65% of the total LD volume was represented by large-sized LDs (.50 mm 3 ) The mean LD volume of all BODIPY-labeled LDs in pharynx and tail region was calculated from single z-stacks of CLS microscopy images. Moderate and stringent dDR (dDR1.5, dDR0.7) increased the mean LD volume of L2 larvae, L4 larvae and adult worms in comparison with respective AL condition ( Figure 9A-C). In pharynx region, the mean droplet size of dDR animals was 1.8 to 2.5-fold increased. Mean droplet volume of tail regions was up to 3.3-fold elevated. As a consequence, the surface to volume ratio of LD was reduced (14-32%) under dDR (Table  S2). For further comparison, the volume of large-sized LDs (.50 mm 3 ) was expressed as percentage of the total LD volume ( Figure 9D-F). This parameter displays the relative proportion of fat stored in large LDs. In comparison with AL condition, moderate and stringent dDR significantly increased the volume-% of LD fraction .50 mm 3 by a factor of 1.5 to 6.3 in L2 larvae and 3.0 to 5.3 in L4 larvae, respectively. In adulthood, LD fraction .50 mm 3 represented 42.6 to 43.3% (AL: 8.37%) and 60.7 to 64.9% (AL: 39.9%) of the total LD volume in pharynx and tail region of dDR fed worms, respectively. Moreover, maximumsized LDs were enlarged by a factor of about 2 to 5 in dDR animals when compared with AL ( Table S3). The largest LDs were detected in the tail region of dDR adult worms (dDR1.5: 609693 dDR0.7: 408668; mean6SEM of the five largest LDs). Together, in worms subjected to dDR, up to 65% of the total LD volume was represented by large-sized LDs (.50 mm 3 ).

Microarray analysis identifies 124 genes which are consistently responsive to dDR
To identify candidate genes that might be responsible for the dDR induced increase in the TAG to protein ratio and enlargement of LDs we performed gene expression profiling using microarrays. We compared genome wide mRNA stead-state levels of L4 and adult worms which were exposed to AL and dDR (0.7, Figure 3. Influence of dDR on triglyceride (TAG) and protein content. (A) TAG content per worm (ng), (B) protein content per worm (ng/w) and (C) TAG/protein ratio of AL and dDR (dDR1.5, dDR0.7) fed wild-type worms were compared. Animals were grown until L4 larval stage or first day of adulthood. Results are represented as mean 6 SEM of three independent experiments (*p,0.05, **p,0.005, ***p,0.001). doi:10.1371/journal.pone.0046198.g003 Enlarged Lipid Droplets in C. elegans PLOS ONE | www.plosone.org 1.5) during development. To analyze and interpret gene expression data, we considered all genes that were significantly regulated (fold-change.2.0; p,0.05, t-test) under both dDR conditions relative to AL treated control group ( Table 1). Genes implicated in lipid metabolism were also selected ( Table S4).
We found 263 genes that were significantly up-regulated or down-regulated in L4 larvae cultivated under dDR1.5 and dDR0.7 (Set I, Figure 10). At the adult stage, 2736 genes were significantly regulated under dDR0.7 and dDR1.5 (set II, Figure 10A). The combination of these two gene sets identified 124 shared genes that were significantly regulated at both developmental stages and both dDR conditions ( Figure 10B). These genes were considered as dDR response genes. Examination of the predicted molecular functions using the gene ontology annotation for C. elegans revealed assignments for 72 of the 124 dDR response genes ( Table 1). Related annotations were combined into broader categories including fatty acid metabolism and other metabolic processes, lipid transport and storage, regulation of lifespan, regulation of transcription, immune response, detoxification and defense, transport and other functions ( Figure 10C). dDR regulates genes of the fatty acid metabolism and other metabolic pathways 14 of the 72 annotated genes regulated in response to dDR are predicted to encode enzymes involved in fatty acid metabolism ( Table 1, see also table S4). We identified several genes of mitochondrial and peroxisomal ß-oxidation up-regulated under dDR in L4 larvae and adults. These include the mitochondrial acyl-CoA synthetase encoding gene acs-2, one carnitine acyltransferase (T20B3.1), an acyl-CoA dehydrogenase (K09H11.11) and two splice variants of a putative acyl-CoA oxidase (F58F9.7). Four genes encoding triacylglycerol lipases were down-regulated in response to dDR. One lipase encoding gene (lips-6) was upregulated and showed increased fold changes at L4 compared to the adult stage.
The diacylglycerol acyltransferase (DGAT) gene (Y53G8B.2) was significantly up-regulated in dDR L4 larvae and adults. DGAT is a key enzyme of de novo synthesis of triacylglycerols (TAG). Expression levels of fat-5 encoding a D9 desaturase were significantly up-regulated under dDR relative to AL condition. Further, expression of Y48A6B.9 gene was stimulated in response to dDR. Y48A6B.9 is predicted to encode a putative trans-2-enol-CoA reductase participating in fatty acid elongation. Fatty acid/ retinol binding protein encoding far-3 was up-regulated and exhibited increased fold changes at L4 in comparison to the adult stage. Two other dDR response genes, F22E5.1 and K02D7.5 (swt-1), are predicted to function in lipid storage. A strong downregulation (up to 49.5 fold in dDR0.7 fed adults) was observed for vit-1, which encodes a lipoprotein functioning in lipid transport. We also found that dDR altered the expression of genes involved    in other metabolic pathways. For example, of five UDPglucuronosyl/UDP-glucosyl transferase encoding genes expected to function in carbohydrate metabolism and lipid glycosylation, three genes were repressed, whereas two genes were up-regulated. Further, the expression of two genes predicted to function in amino acid metabolism, one putative cystathionine gamma-lyase (cth-1) and one asparagine synthase (asns-2), was up-regulated under stringent and moderate dDR in L4 larvae and adult animals.
dDR regulates genes involved in life span extension, stress response and transcription As expected, dDR altered the expression of genes predicted to function in the determination of lifespan (Table 1). Furthermore, we identified several dDR responsive genes involved in immune response ( Table 1): seven C-type lectin encoding genes, three genes encoding a CUB-like bearing protein, saposin B encoding spp-17 and ilys-5, which encodes an invertebrate lysozyme family member. Interestingly, expression of these genes was consistently down-regulated under dDR with exception of C-type lectin  encoding gene clec-68, which was up-regulated during the L4 and adult stage. dDR also influenced the expression of six genes related to detoxification and defense, including cytochrome P450 family members and scl-6, which encodes a defense-related protein. Of these genes, five were down-regulated, whereas only cyp-34A1 was up-regulated under dDR. Another group of dDR-response genes is involved in the transcriptional regulation. We found that three nuclear hormone receptor (NHR) encoding genes (nhr-74, nhr-117, nhr-244) were consistently up-regulated in dDR L4 larvae (Table 1). In contrast, dDR led to the repression of nhr-74 and nhr-244 in the adult stage. Further, expression of djr-1.2 encoding a DJ-1 (mammalian transcriptional regulator) related protein was stimulated in response to dDR, whereas O-acyltransferase homolog oac-20, an, was down-regulated.

Discussion
In comparison to other DR protocols on solid medium [7,24,32] our method was established to study effects of DR during development. This approach was named ''development-DR (dDR)''. It offers several advantages. First, our method minimizes possible side-effects of antibiotic or heat treatment due to the fact that exclusion of peptone and serial dilution of living E. coli is the sole intervention. Second, we were able to control the food availability per worm using flow-cytometry based sorting of a defined number of embryos per plate. We show that serial dilution of the food source results in a dose-dependent reduction of body size indicating standardized dDR conditions. Third, our method enables the application of DR during development from hatching throughout development. Several other regimes, including diluting bacteria in liquid medium, axenic media, dietary deprivation or DR on agar plates, initiate DR in C. elegans at L4 stage or during adulthood to avoid detrimental effects on development or surviving [7,11,14,[26][27][28][29]32,35]. Using our method, the onset of DR from hatching throughout larval development did not induced larval arrest, dauer stage or and an adult reproductive diapause [33]. However, developmental time was slightly delayed under dDR (first egg lay: AL: ,72 h, dDR1.5: ,74 h; dDR0.7: ,76 h).
In summary, the use of living bacteria, the controlled application of the extent of dDR and the applicability during development are the main advantages of our dDR protocol. Of course, an influence of peptone or metabolic changes of living bacteria growing on AL and DR plates cannot be excluded. In control experiments using AL plates containing a thick bacterial lawn we observed not influence of peptone on LD size.
A reduction in body size under nutrient-limiting conditions has been observed in several species including D. melanogaster and mice [36,37]. In line with this, we found an inverse linear relationship between body size and the extent of dDR in adult worms exposed to DR during development. The observed body sizes of AL and dDR treated worms were similar to those reported in other studies performing DR in C. elegans [25,38].
The dDR induced reduction in body size was accompanied by a reduced protein content of adult worms. A close correlation between body size and protein content in C. elegans was also observed by using different E. coli strains [39]. It has been proposed that body size is controlled by the insulin/IGF-1 signaling pathway, which responds to the nutritional state and promotes cell growth by increasing the protein synthesis. DRdependent (plate, killed bacteria) reduction of body size in adult C. elegans is mediated by the sensory EGL-4/SMA/MAB pathway which regulates hypodermal endoreduplication [25]. The reduction in body size during dDR might also arise from these pathways and/or might be linked or accompanied by a decreased synthesis of phospholipids. In C. elegans, the glycerophospholipids phoshoethanolamine (PE) and phosphatidylcholine (PC) account for 55% and 32% of the total phospholipids, respectively [40]. Both, de novo synthesis of phospholipids and triglycerides (TAGs) require diacylglycerol (DAG) as substrate. Conversion of DAG into TAG is catalyzed by diacylglycerol transferase (DGAT) [41]. Gene expression analysis revealed that moderate as well as stringent dDR led to an up-regulation of the DGAT encoding gene Y53G8B.2. Assuming an increased de novo TAG synthesis by DGAT, this might cause substrate deficiency for the synthesis of phospholipids which are the major component of cell membranes. Thus, dDR-dependent reduction in body size might be mediated by a reduced polyploidy, synthesis of protein and/or phospholipid de novo synthesis. In C. elegans, fat is mainly stored in LDs of the intestine and hypodermis [42,43]. Surprisingly, moderate as well as stringent DR during development led to a remarkable LD expansion in intestinal and hypodermal cells. The LD expansion was found in L2 larvae, L4 larvae and adult worms which were subjected to DR during development. Of note, the observed LD phenotype under dDR is considerable distinct from the phenotype of fasted animals. Upon fasting, L4 larvae and adult worms mobilize TAGs from intestinal fat stores, which causes a drastic decrease in the number and size of LDs [14,21,44,45].
Gene expression analyses identified several lipogenesis genes that were up-regulated under moderate (dDR 0.7) and stringent dDR (dDR 1.5) in L4 larvae and adult worms. These genes are involved in lipid synthesis (i. e. Y53G8B.2/DGAT), lipid binding (i. e. far-3) and fatty acid desaturation (i.e. fat-5/D9 fatty acid desaturase). Assuming that gene expression is translated into enzymatic activity, up-regulation of these genes might increase de novo synthesis of TAGs. This hypothesis is supported by our finding that dDR led to an increased TAG to protein ratio in L4 larvae and adult worms. In line with this, mild calorie restriction in mice led to fat accumulation [46].
Studies indicate that storage efficiency of TAGs into LDs is likely dependent on the generation of monounsaturated fatty acids (MUFAs) [47,48]. For example, it has been demonstrated in chinese hamster ovary cells and in Drosophila that supplementation with oleic acid (C18:1n9) leads to TAG accumulation [49,50]. Further, supplementation with vaccenic acid (C18:1n7) increased TAG levels and LD size in C. elegans peroxisomal ß-oxidation mutants [43]. Here, we found an up-regulation of the D9 desaturase encoding gene fat-5. FAT-5 catalyzes the desaturation of palmitic acid (16:0) into palmitoleic acid (C16:1), which serves as substrate for elongation to vaccenic acid [51,52]. Thus, increased fat-5 expression may lead to enhanced vaccenic acid levels, which might promote TAG accumulation and enlargement of LDs under dDR. The DGAT encoding gene, which is upregulated under dDR, also links TAG synthesis to LD size. It has been shown, that mammalian DGAT translocates to the LD surface when fatty acid enter the cells in order to promote lipid production and storage [53]. Taken together, dDR might enlarge LDs via up-regulation of lipogenesis genes such as fat-5 and DGAT.
The ratio of surface phospholipids to core neutral lipids might be an important determinant of LD size. Depletion of phosphatidylcholine (PC), which represents the most abundant phospholipid in the LD monolayer, leads to enlargement of LDs ( [54][55][56]. PC can be synthesized either from choline by the Kennedy pathway or by S-adenosylmethionine (SAM) dependent methylation of phosphatidylethanolamine in mammals or of phosphoethanolamine in nematodes and plants [57][58][59]. Knockdown of the key enzyme of the Kennedy pathway, CTP:phosphocholine cytidylyltransferase (CCT), leads to decreased PC levels and to an drastically increased LD size in Drosophila S2 cells [49,55]. Likewise, repression of genes (sams-1, pmt-1) mediating methylation-dependent PC synthesis in C. elegans leads to large intestinal LDs [56,60]. More recently, SREBP-1 has been identified as a transcriptional activator of genes involved in the one-carbon cycle and LD accumulation [60]. SREBP-1 target genes include sams-1, pmt-2, folic acid transporter folt-2 and several other genes. Because key genes of the PC synthesis are not consistently regulated under dDR in larvae and adult worms, dDR induced enlargement of LD seems to be not mediated by a perturbed SAM-dependent PC synthesis.
It has been shown [56] that LD expansion in sams-1 and pmt-1 depleted C. elegans was associated with reduced PC levels and increased TAG content indicating a reciprocal regulation of PC and TAG synthesis. Thus, we speculate that up-regulation of DGAT under dDR may indirectly reduce de novo synthesis of phospholipids which leads to the LD enlargement under dDR. One model explaining expansion of LDs under conditions of reciprocal regulation of TAGs and phospholipids involves the fusion of LDs [49,55,61]. It has been reported that during incorporation of lipids into LDs, PC homeostasis at the expanding monolayer is essential to stabilize the organelles and to prevent their coalescence [55,62]. In turn, PC deficiency during LD growth leads to fusion of LDs because of instability [55,62]. We therefore speculate that LD expansion by fusion could be a dDRinduced mechanism to reduce the surface to volume ratio of LD, which might protect stored TAGs against lipolysis.
In C. elegans, only a small number of genes modulating LD size were identified. It has been shown, that mutations in genes encoding enzymes of the MAOC-1/DHS-28/DAF-22 peroxisomal ß-oxidation pathway cause LD expansion in C. elegans intestinal cells [43,63]. Functional loss of ACS-3/acyl-CoA synthetase resulted in the formation of large intestinal LDs due to an elevated fatty acid uptake and increased de novo lipid synthesis [42]. LD expansion has been also reported for C. elegans deficient for the KLF-3/Krüppel-like transcription factor [64]. Moreover, it has been shown that deletion of lbp-5 encoding a lipid-binding protein (LBP) leads to enlarged intestinal LDs in the worm [65]. Since all of these known C. elegans genes modulating LD size are not consistently regulated by dDR in larvae and adult worms we suggested that they play no substantial role for the observed enlargement of LDs under dDR. However, we cannot exclude a post-transcriptional effect of dDR on these genes.
LD expansion in C. elegans mutants were observed in the intestine but not in other fat storage sites [42,43,45,64,65]. In the present study, we found that dDR causes an expansion of LDs in the intestine as well as in the hypodermis. The formation of enlarged LDs has been functionally associated with an increased resistance of stored lipids to lipolysis. Studies indicate that due to the reduction of the LD surface relative to its volume, lipids are less accessible to membrane associated lipases [43,49]. In Drosophila, large LDs were slower metabolized than smaller ones and provided a survival advantage during starvation [43]. In C. elegans, peroxisomal ß-oxidation dhs-28 mutant exhibited enlarged LDs in the intestine that were resistant to fasting and lipase induced lipolysis [43,63]. Over-expression of adipose triglyceride lipase atgl-1 (C05D11.7) caused a drastic reduction of TAG levels in wild-type but to a much lesser extent in dhs-28 animals. We observed a reduction of the surface to volume ratio of LD under dDR in comparison to AL condition. Moreover, of dDR response genes four out of five lipase genes were down-regulated (lipl-5, lips-6, lips-12 and C40H1.8), which may indicate decreased lipolysis rates under dDR. Thus, we assume that expansion of LDs in the one case and reduced activity of lipases in the other case might be a protective mechanism to prevent excessive breakdown of stored lipids and to release energy stores more slowly during dDR.
In conclusion, using an optimized DR protocol we demonstrated that DR subjected to C. elegans during development increases the triglyceride to protein ratio, enlarges lipid droplets and alters the expression of genes functioning in lipid metabolism. These changes might be an effective adaptation to conserve fat stores in animals which were exposed to dietary restriction during development.

Strains and Maintenance
Wild-type C. elegans Bristol N2 and mutant strains daf-2(e1370) and eat-2(ad465) were used. Nematodes were cultivated at 20uC on Nematode Growth Medium (NGM) agar plates seeded with Escherichia coli OP50 as food source [66]. Strains were obtained from the Caenorhabditis Genetics Center (Minnesota, USA).

Dietary restriction on agar plates
To induce dDR on agar plates bactopeptone (BD, Heidelberg, Germany) was omitted from the standard NGM recipe [32,66]. E. coli OP50 suspension was cultivated in DYT medium at 37uC until reaching an optical density (OD 600 nm /ml) of 1.5. A serial dilution of E. coli OP50 suspended in M9 buffer was prepared ranging from OD 600 = 6.0 to OD 600 = 0.3. For dDR conditions, 250 ml of the respective E. coli suspension was seeded onto the dDR plates. For ad libitum (AL) condition, standard NGM agar seeded with E. coli OP50, OD 600 = 1.5 was used. Plates were incubated at 37uC for 16 h. For all experiments, nematodes were synchronized by hypochlorite treatment of gravid adults. To standardize food availability, 500 eggs per plate were sorted via a cytometry-based object parametric analysis and sorting system (COPAS Biosort, Union Biometra, Geel, Belgium) and cultivated at 20 C until reaching L2, L4 or adult stage. Animals were transferred daily onto fresh plates at L4 and adult stages to prevent starvation.

Preparation of worm homogenates and biochemical measurements
Wild-type nematodes were cultivated under AL and two different dDR conditions (dDR0.7, dDR1.5) until reaching the L4 and adult stage, respectively. For each replicate, 2000 to 2500 L4 larvae and 800 to 1000 adult worms were collected via the COPAS Biosort system. Animals were homogenized in 100 ml buffer (150 mM NaCl; 1 mM EDTA; 50 mM Tris-HCl, pH 7.5; 0.5% CHAPS) using the Precellys 24 homogenizer (Peqlab, Erlangen, Germany) and ceramic beads (1.4 mm diameter). Cell debris was removed by centrifugation at 21,000 g for 20 minutes. Homogenates were used in 96-well format for colorimetric determination of the triacylglycerol (TAG) and protein content as well as for quantification of auto fluorescent 'age pigment'. The triacylglycerol (TAG) content was determined using an enzymatic assay (Analyticon diagnostics, Lichtenfels, Germany) and a TAG standard (Biovision, Hannover, Germany) according to the manufacturer's directions. The protein content was measured using the PierceH BCA protein assay kit (Thermo Fisher Scientific, Bonn, Germany) and bovine serum albumin (BSA) as protein standard.

Determination of the body proportion
Gravid adult wild-type worms cultivated under AL and different dDR conditions (dDR0.3 to dDR6.0) were harvested and anesthetized in 2% NaN 3 . Bright-field microscopy images were obtained using a Zeiss Axio Observer.D1 inverted microscope equipped with an Axiocam MRm camera (Zeiss, Jena, Germany). Magnification of animals was 50-fold (56/0.12 objective). Length (mm), width (mm), perimeter (mm) and area (mm 2 ) of single worms was obtained using the AxioVision software (Release 4.8, Zeiss, Jena, Germany). The body volume (nl) was calculated using an adapted cylinder volume formula which includes the worm's area and perimeter [69,70]. In addition, time of flight (TOF; arbitrary unit, AU) and extinction (Ext, AU) of single nematodes was measured via a COPAS Biosort. TOF and Ext values are approximate values for the axial length and volume of the worms, respectively [71]. Using flow cytometry, 500 to 2,000 animals were analyzed for each condition in three independent experiments.

Motility and pharyngeal pumping rate
For analysis of animal motility and pharyngeal pumping rate, 10 animals per agar plate were cultivated under AL and two different dDR conditions (dDR0.7, dDR1.5), respectively. Single nematodes were analyzed at the first day of adulthood using a Zeiss SteREO Discovery V8 binocular microscope (Zeiss, Jena, Germany). For analysis of animal motility, single worms were recorded for 20 seconds at 8.0-fold magnification using a Zeiss AxioCam ICc 1 and Zeiss AxioVision software (Release 4.8). A worm tracking software (WormTracker 2.0.25, Thomas Bornhaupt, Neustadt adW., Germany) was used to calculate the body bending frequency (Hz), whole animal motility (mm s 21 ) and head motility (mm s 21 ) [72]. For measurements of pharyngeal pumping rate, the pharynx was filmed for 40-60 seconds at 63-fold magnification using a Canon camera (Legria HF20). The pharyngeal pumping rate (pumps/min) was counted on slowed films. 20 randomly chosen animals were recorded for each condition in at least three independent experiments.

Oil Red O staining
AL fed and dDR (dDR0.7, dDR1.5) wild-type worms were harvested at day one of adulthood. Animals were washed in PBS buffer and fixed with 4% paraformaldehyde (PFA) for 15 min. After three freeze and thaw cycles in liquid nitrogen, worms were washed in PBS buffer, followed by a dehydration step in 60% isopropanol. Animals were stained in filtered Oil Red O staining solution (60% Oil Red O stock solution (5 mg/ml isopropanol)/ 40% distilled water) over night. After washing in PBS, worms were analyzed using an Axio Observer D1 inverted microscope (Zeiss, Germany). Photographs of 10-15 worms per feeding condition were taken by AxioCam MRm camera (Zeiss, Germany) and a 206/0.50 M27 objective at a fixed exposure time. All experiments were performed three times.

Determination of number and size of lipid droplets
To determine the number and size of lipid droplets, BODIPY TM 493/503 positive structures in pharynx and tail region of L2, L4 and adult staged wild-type animals were imaged using scanning-laser confocal (SLC) microscopy (Leica TCS SP). A HCX PL APO CS 63.06/1.32 oil immersion objective and Leica LAS AF software was used to collect z-stacks with a step size of 0.5 mm (image format: 1586158 mm 2 , 102461024 pixels). Zstacks consisted of 20 to 110 plane images depending on position and developmental stage of the animal. Images of eight to ten animals per condition and developmental stage were taken using identical settings and exposure times. Z-stacks of pharynx and tail regions were analyzed using ImageJ software (version 1.42q, Object Counter3D plugin). The BODIPY TM 493/503 positive droplets were automatically identified via adaptive thresholding. The volume of each droplet was calculated by summing up the voxels. The total droplet number, total droplet volume and mean droplet volume in pharynx and tail region was calculated using

Whole genome gene expression analysis
Synchronized wild-type worms cultivated under AL or dDR conditions (dDR0.7, dDR1.5) were harvested at L4 and adult stage. After washing in M9 buffer, worms were resuspended in 350 ml RTL buffer, disrupted in the Precellys 24 homogenizer (Peqlab, Erlangen, Germany) and subjected to an additional homogenization step using QIAshredder spin columns (QIAGEN, Hilden, Germany). Total RNA was isolated using the RNeasy Mini Kit (QIAGEN, Hilden, Germany). On-column DNAase digestion was performed to eliminate genomic DNA (RNase-Free DNase Set, QIAGEN, Hilden, Germany). The quality and yield of the preparation was assessed using a 2100 Bioanalyzer (Agilent Technologies, Waldbronn, Germany).
Labeled cRNA was generated, hybridized and processed by imaGenes expression profiling service (Berlin, Germany) using a customized 8660 K C. elegans Agilent microarrays (imaGenes/ SourceBioscience, Steffen Hennig). Normalization was done by 'quantile normalization' using the R-package [73]. After normalization, each data set included 26,843 gene expression values of four biological replicates for dDR and AL treated control group. Fold-changes of intensities were calculated from the arithmetic mean of gene expression values between dDR and AL group (L4 stage: dDR1.5vsAL, dDR0.7vsAL; adult stage: dDR1.5vsAL, dDR0.7vsAL). The significance was calculated using an unpaired t-test with unequal variance (Welch-test). P-values.0.05 were regarded as not significant. Genes with a fold-change .2.0 were regarded as differentially regulated.

Statistical analysis
Statistical analysis was performed with Microsoft Exel (2003) and GraphPad Prism (Version 4.0). Significances were calculated using one-way ANOVA and unpaired t-test (two tailed). Welchcorrection was used if variances were different. Logrank t-test was used for lifespan analysis. Differences were considered statistically significant at p,0.05 (*), p,0.01 (**) and p,0.001 (***). Figure S1 Influence of dDR on motility and pumping rate. Body bend frequency (A), whole animal motility (B) and head motility (C) of dDR restricted (dDR1.5 and dDR0.7) and AL fed wild-type worms were analyzed using a worm tracking software (see Experimental procedures). Nematodes were analyzed at first day of adulthood. (A) Results for body bend frequency (Hz) are represented as mean 6 SEM of three experiments with 20-30 animals each. (B, C) Bars represent mean motilities (mm s 21 ) 6 SEM of three experiments with 20 individuals each (***p,0.001). (D) Mean pharyngeal pumping rate of adult wild-type worms cultivated at either AL or dDR condition (dDR1.5, dDR0.7) was obtained form three experiments with 15-20 individuals each. Error bars represent a SEM. Significant decrease in pharyngeal pumping of dDR1.5 worms is indicated by asterisks (**p,0.01). (TIF) Figure S2 Influence of dDR (0.7) in adult N2 worms on the triglyceride (TAG) to phospholipid (PL) ratio determined by thinlayer chromatography (TLC). TAG and PL contents were measured by TLC. Data are shown as mean 6 SEM (n = 3). Significant differences to the AL (ad-libitum) group were detected using an unpaired two-tailed t-test (*** p,0,001).