Fluorescence-based fixative and vital staining of lipid droplets in Caenorhabditis elegans reveal fat stores using microscopy and flow cytometry approaches.

The proportions of body fat and fat-free mass are determining factors of adiposity-associated diseases. Work in Caenorhabditis elegans has revealed evolutionarily conserved pathways of fat metabolism. Nevertheless, analysis of body composition and fat distribution in the nematodes has only been partially unraveled because of methodological difficulties. We characterized metabolic C. elegans mutants by using novel and feasible BODIPY 493/503-based fat staining and flow cytometry approaches. Fixative as well as vital BODIPY staining procedures visualize major fat stores, preserve native lipid droplet morphology, and allow quantification of fat content per body volume of individual worms. Colocalization studies using coherent anti-Stokes Raman scattering microscopy, Raman microspectroscopy, and imaging of lysosome-related organelles as well as biochemical measurement confirm our approaches. We found that the fat-to-volume ratio of dietary restriction, TGF-β, and germline mutants are specific for each strain. In contrast, the proportion of fat-free mass is constant between the mutants, although their volumes differ by a factor of 3. Our approaches enable sensitive, accurate, and high-throughput assessment of adiposity in large C. elegans populations at a single-worm level.

Additionally, fl uorescence was quantifi ed using the cytometrybased object parametric analysis and sorting system (COPAS), Biosort (Union Biometrica), which measures time of fl ight (TOF; nematode length), decrease in laser light when a worm passes through the laser beam (extinction), and green fl uorescence (BODIPY fl uorescence) for each animal. The system is equipped with a 488 nm multiline argon laser. Green and red fl uorescence emissions were measured using a fi lter of 510 nm, with a bandwidth of 23 nm. Worms were washed from plates, placed in the COPAS sample cup, and analyzed. COPAS settings were as follows: gains: extinction (EXT), 1; green, 1; yellow, 1 and red, 1; thresholds: signal, 50 and TOF minimum, 100. Voltages of photomultiplier tubes for the green, yellow, and red channels were set to 600 V. Only objects with TOF values of adult nematodes and signals detected in the green channel were included in quantifi cation data. The background fl uorescence of unstained worms was subtracted in microscopy and COPAS analyses. Nematodes (250 to 2,000) in each of two to three independent experiments were analyzed using COPAS. Microscopy and COPAS settings were kept identical to allow a direct comparison between groups.

Oil Red O staining
After 15 min of 4% paraformaldehyde fi xation followed by three freeze/thaw cycles in liquid nitrogen, worms were washed with PBS and dehydrated in 60% isopropanol. Afterward, worms were stained in fi ltered Oil Red O staining solution (60% Oil Red O stock solution [5 mg/ml isopropanol]/40% water). Worms were washed with PBS and observed with an Axio Observer D1 inverted microscope (Zeiss). Images were captured with an Axio-CamICc1 camera.

Flow cytometry and biochemical assays
Synchronized worms were cultivated until they reached adulthood. Synchronized adult nematodes (300-600) were counted for each biological replicate by using the COPAS Biosort system. The worms were resuspended in 70 µl of lysis buffer (50 mM Tris, pH 7.5; 150 mM NaCl; 1 mM EDTA, pH 8.0; 0.5% NP-40; protease inhibitors) and sonicated using a Sonopuls HD 2070 (Bandelin) with an MS72 sonotrode at 30% power for 2 × 8 s. Insoluble material was removed by centrifugation at 21,000 g for 20 min. We assayed triacylglycerol (TAG) content spectrophotometrically, using an enzymatic assay kit (Analyticon) with a TAG standard (Biovision) according to the manufacturer's instructions. As shown in supplementary Fig. III, the TAG assay kit is suitable for determination of TAG in C. elegans . The protein content was determined using a bicinchoninic acid reagent kit (Pierce) with ( 11 ) or stimulated Raman scattering (SRS) signals ( 12 ) but with the lysosome-related organelle (LRO)-specifi c markers, LysoTracker and glo-1 ( 10,13,14 ). So far, fi xativebased Oil Red O and Nile red staining have been characterized as appropriate methods to stain and quantify the main fat stores in C. elegans ( 10,15 ).
In order to characterize metabolic C. elegans mutants regarding body composition in this study, we employed the fl uorescent neutral lipid dye BODIPY 493/503. As shown by fl ow cytometry with MA-10 Leydig tumor cells, BODIPY 493/503 is more specifi c than Nile red for staining lipid droplets ( 16 ). In addition to cell culture, the dye has been used in a wide range of organisms and tissues including yeast ( 17,18 ), Drosophila ( 19,20 ), and rat soleus muscle fi bers ( 21 ) and as vital dye in staining of LRO markers in C. elegans ( 13 ). Here, we have developed a protocol for staining and quantifi cation of fi xed worms, using BODIPY 493/503. Furthermore, body composition analysis and a novel quantitative BODIPY 493/503-based fl ow cytometry method allowing parallel measurement of fat content and body volume revealed specifi c fat proportions of dietary restriction, TGF-␤ , and germline mutants at single-worm level. Additionally, we show that vital staining of worms with BODIPY 493/503 using a short incubation time leads to a very effi cient staining of lipid droplets. This method may be suitable for screening assays.

Strains and C. elegans maintenance
Bristol N2 was used as the wild-type strain. The glp-1 ( e2141 ), eat-2 ( ad465 ), and daf-7 ( e1372 ) mutant strains were used. The Glp-1 mutant was a gift from Ralf Schnabel (Technical University of Braunschweig, Germany). Other strains were obtained from the Caenorhabditis Genetic Center. Nematodes were cultured on Escherichia coli OP50 lawns on NGM agar plates, as described previously ( 22 ), at 15°C, 20°C, or 25°C. For all experiments, strains were synchronized by hypochlorite treatment of gravid adults.

BODIPY 493/503 staining and quantifi cation
The BODIPY 493/503 stock solution (1 mg/ml) was prepared in DMSO. A fresh solution of BODIPY 493/503 diluted in M9 buffer at the indicated concentrations was used for staining. For fi xative staining, freshly harvested nematodes were washed in M9 buffer, incubated in 4% paraformaldehyde solution for 15 min, which was followed by three freeze/thaw cycles in liquid nitrogen. After paraformaldehyde solution was removed by three washing steps with M9, worms were incubated in a volume of 500 µl of 1 µg/ml BODIPY 493/503 (Invitrogen) in M9 for 1 h at room temperature. Nematodes were washed three times with M9 buffer and were used for microscopy or fl ow cytometry analysis. For vital staining, freshly harvested nematodes were incubated in M9/BODIPY 493/503 (6.7 µg/ml) solution for 20 min. An incubation of 20 min generated a suffi cient BODIPY 493/503 signal and prevented starvation of the worms and staining of LROs. Afterward, worms were washed three times with M9 and used immediately for further experiments. BODIPY 493/503 fl uorescence was visualized using an Axio Observer D1 inverted microscope (Zeiss) using fi lter 38HE (excitation, bandpass [BP] 470/40; beam splitter FT 495; emission BP 525/50). Images were captured with an Axiocam MRm camera. Quantifi cation of fi xative by an air-cooled argon-ion laser (series 77; Lasos, Jena, Germany). The exciting laser radiation is coupled to a Zeiss microscope through a wavelength-specifi c single-mode optical fi ber. The incident laser beam is collimated via an achromatic lens and passes through a holographic BP fi lter before it is focused onto the sample through the microscope objective. A Nikon Fluor (60×/1.00 NA, WD = 2.0 mm) water immersion objective was used in the studies reported here. The sample is located on a piezo-electrically driven microscope scanning stage with an x-y resolution of approximately 3 nm and a repeatability of ±5 nm, and a z resolution of approximately 0.3 nm and ±2 nm repeatability. The sample is scanned through the laser focus in a raster pattern at a constant stage speed of fractions of a micrometer per second. The continuous motion prevents sample degradation at the focal point of the laser beam. Spectra are collected at a 0.5 m grid, with a dwell time of 0.25 s. Preceding fl uorescent images were recorded at approximately 0.1 mW and an illumination time of 0.1 s. Raman images were reconstructed using a spectral decomposition algorithm based on vertex component analysis ( 26 ). Fluorescence images are based on fl uorescence intensities.

A BODIPY 493/503-based protocol stains major fat stores in C. elegans and preserves native lipid droplet morphology
In order to characterize fat stores in C. elegans , we developed a gentle staining procedure with short fi xation times and without dehydration steps, using the lipid droplet-specifi c fl uorophore BODIPY 493/503. As shown in Fig. 1A and B , the BODIPY 493/503-based protocol preserves worm anatomy and stains spherical structures in the intestine and hypodermis. These compartments are described as the major fat stores of C. elegans wild-type N2 ( 2 ). Droplets in oocytes and embryos are also stained by BODIPY 493/503. The BODIPY 493/503-labeled intracellular vesicles are fi ne and distinctly distributed ( Fig.  1C, D ). Most of the droplets are <3 µm in diameter. To provide evidence that the BODIPY 493/503-labeled vesicles are lipid droplets, we tested our protocol with Lipid-Tox deep red and Nile red and performed colocalization studies. As shown in Fig. 2 , BODIPY 493/503 stains the same compartments and droplets that the two other neutral lipid dyes do. To measure the extent of overlap between the BODIPY 493/503 stain and that of the two other dyes, we calculated the Pearson correlation coefficient ( r ). The value of this coeffi cient ranges from Ϫ 1 to 1; the higher the value of r , the stronger the correlation between the red and green images. The mean Pearson correlation coeffi cients ± standard deviations (SD) for BODIPY 493/503 with LipidTox, r = 0.927 ± 0.0056, and Nile red, r = 0.934 ± 0.027 (3-5 stacks), reported a high degree of colocalization of BODIPY 493/503 with LipidTox deep red and Nile red. We further proved our BODIPY 493/503 staining procedure with CARS microscopy, a nonlabeling method for visualizing lipid droplets in C. elegans in vivo without fi xation steps ( 11,27 ) or after fi xation with paraformaldehyde ( 28 ). As shown in Fig. 3 , CARS microscopy of vital worms shows intestinal and hypodermal lipid droplets similar in size, morphology, and distribution to droplets in images obtained by fi xative BSA as standard. We performed two to three independent experiments for each genotype and condition with two replicates, respectively. We measured total TAG and protein contents for each homogenate in duplicate in two independent measurements, respectively. To account for differences in total TAG content per worm, we used protein and body volumes as covariates in the analysis of the data.

CARS microscopy
Worms were washed in M9 buffer, resuspended in 2% NaN 3 and placed on glass slide. A cover glass was then gently applied to the sample. CARS images shown in Fig. 3 were recorded using a CARS microscope that was custom built by A. Zumbusch and M. Winterhalder at the University of Konstanz, based on a Leica TCS SP5 multiphoton microscope. As excitation source for the pump beam, the output of an optical parametric amplifi er (APE, Berlin, Germany) was used, whereas the output of a Ti:sapphire laser (Coherent Mira, Santa Clara, CA) served as the Stokes beam. Both sources were electronically synchronized (coherent Syn-chroLock, Santa Clara, CA). Images were recorded at a laser frequency difference of 2,845 cm Ϫ 1 , corresponding to an aliphatic CH vibrational resonance.
The setup used for CARS microscopy shown in Fig. 4, 5, and 10 was described previously ( 25 ). For the experiments, narrowband (bandwidth, 0.5-1 nm) ps pulses were used to match the C-H stretching region at about 2,900 cm Ϫ 1 . The Stokes pulses were provided by a Ti:sapphire laser centered at 831.7 nm. In addition, this laser served as a pump source for an optical parametric oscillator (OPO), which was tuned to 670 nm for resonant and 678 nm for vibrationally nonresonant excitation. These last settings were used as controls to ensure that image contrast stemmed from a CARS process. After the pump and Stokes pulses were overlapped both temporally and spatially by a delay stage and a longpass fi lter, the two laser beams were directed into a commercial laser scanning microscope (510 Meta; Zeiss). The two beams were focused onto the object plane by a 63×/NA 1.40 oil immersion objective, which was used for CARS and fl uorescence measurements as well. The power of the Stokes pulse was about 12 mW, and the power of the pump pulse was about 16 mW at the sample. The anti-Stokes signal at 561 nm was detected in forward direction and spectrally fi ltered from the incoming beams by the combination of shortpass and bandpass fi lters and detected by a photomultiplier. For every sample position, a depth profi le (z-stack) was measured, recording 30 focal plane images separated by 1 µm each. The image size was 512 × 512 pixels, and the image acquisition time per image plane was about 6.3 s

Two-photon fl uorescence
The two-photon fl uorescent signal was recorded simultaneously with the CARS images. The signal, which was generated by the excitation from the OPO wavelength at 670 nm, was recorded in refl ection mode by using an internal BP 435-485 fi lter.

One-photon fl uorescence
For one-photon excitation, an argon-ion laser from the laser scanning microscope was used, which provided 488 nm (10 µW at the sample) for exciting the fl uorescence of the BODIPY 493/503 dye. The fl uorescence signal was fi ltered by a band-path fi lter and detected in refl ection. The z-stacks with the same experimental settings were recorded.

Raman data acquisition and data analysis
Raman spectra were acquired using a WITec, GmbH (Ulm, Germany) CRM 2000 model confocal Raman microscope. Excitation (approximately 5 mW at 488 nm at the sample) is provided

Quantifi cation of BODIPY 493/503-based microscopy data agreed with biochemical analysis and revealed low fat content of eat-2 mutants and high fat levels of daf-7 and glp-1 mutants
The fat content of many C. elegans mutants is still under discussion because the well-established vital dyes Nile red and C1-C12-BODIPY-labeled fatty acids are not accurate proxies of fat stores in the worm ( 10 ). Here we selected the eat-2 ( ad465 ), daf-7 ( e1372 ), and glp-1 ( e2141 ) strains as principal fat storage mutants in order to quantify their fat content with BODIPY 493/503. The eat-2 ( ad465 ) mutant, a genetic model of dietary restriction, lacks a functional BODIPY 493/503 staining. Furthermore, colocalization experiments revealed that BODIPY 493/503 stains the same vesicles that are detected with CARS microscopy ( Fig. 4 ) . As a second method that allows a label-free visualization of lipids, proteins, and other metabolites, we applied Raman microspectroscopy of fi xed BODIPY 493/503-stained animals. As shown in Fig. 5 , the BODIPY 493/503-labeled structures overlap with lipid signals of Raman spectra, and similar signal intensities are observed. Thus, our data demonstrate that the BODIPY 493/503stained organelles in fi xed animals are lipid droplets with a well-preserved morphology.  worms ( Fig. 6 ). To further verify BODIPY 493/503-based quantifi cations, TAG contents of the eat-2 , daf-7 , and glp-1 mutants were determined using an enzymatic assay from a distinct number of worms, which were obtained using a fl ow cytometry-based object parametric analysis and sorting system (COPAS) ( Table 1 ). As shown in Table 1 and Fig. 7A , the mean TAG level ± standard error of the mean (SEM) of the eat-2 mutant (8.3 ± 0.7 ng/worm) is about 4-fold lower than that of N2 (37.2 ± 3.0 ng/worm), whereas the daf-7 mutant accumulated 2-fold more TAG (73.3 ± 10.1 ng/worm). TAG levels of the glp-1 (45.9 ± 2.3 ng/ worm) and N2 (53.8 ± 5.9 ng/worm) strains were similar at permissive temperature (15°C). At restrictive temperature (25°C), the glp-1 mutant showed a 27.5% increase in TAG (72.0 ± 10.2 ng) compared with that of N2 (56.5 ± 6.4 ng). Thus, biochemical measurement of TAG content per worm correlates well with the BODIPY 493/503 microscopy quantifi cation data of metabolic C. elegans mutants.

Body composition analysis and BODIPY 493/503-based fl ow cytometry revealed specifi c fat proportions of the eat-2 , daf-7 , and glp-1 metabolic mutants at the singleworm level
Rather than the absolute fat content of an organism, the fat-to-fat-free ratio is one of the most relevant physiological parameters used to evaluate body composition and adiposity ( 35 ). To calculate this ratio in the eat-2 , daf-7 , and glp-1 mutants, we also determined the protein fraction. A part of the protein fraction contributes to the metabolic active body compartment of the nematodes. As shown in Fig. 7B , the protein content per worm of the eat-2 (96.2 ± 7.4 ng/ worm) mutant is 2.4-fold lower and that of the daf-7 (292.7 ± 6.6 ng/ worm) mutant is 30% higher than in the wild type (227 ± 7.1 ng/worm). The glp-1 and N2 strains differ slightly in protein content at permissive as well as at restrictive temperatures. The TAG and protein values were used in order to calculate fat-to-fat-free ratios. Based on these values, the eat-2 mutant could be described as a lean mutant, whereas the daf-7 and glp-1 mutants are obese ( Fig. 7C ). Another central parameter for evaluating leanness or adiposity is the proportion of protein or fat in relationship to the body volume. Volumes ( V ) of the eat-2 , daf-7 , and glp-1 mutants were determined using a worm-adapted zylinder volume formula that includes worm area ( A ) and perimeter ( P ) ( 36,37 ). pharyngeal nicotinic acetylcholine receptor subunit ( 29 ), resulting in low pumping rate ( 30 ). The TGF-␤ daf-7 ( e1372 ) mutant has ‫ف‬ 3-fold higher de novo fat synthesis rate than the wild type ( 31 ). The glp-1 ( e2141 ) mutant is a germline defective mutant with an early interruption of mitosis of germ cells at restrictive temperatures ( 32,33 ). It was previously described as a low-fat mutant based on vital Nile red staining ( 34 ). A more recent work using fi xativebased Oil Red O stain showed that the glp-1 mutant accumulates more fat than the wild type ( 10 ).
We stained fi xed N2, eat-2 , daf-7 , and glp-1 nematodes with BODIPY 493/503 and quantifi ed the resulting fl uorescence signals of worm images. Figure 6 shows representative images obtained from BODIPY 493/503 staining, and Table 1 shows quantitative data. For comparison, Oil Red O staining was also performed ( Fig. 6 ). The eat-2 mutant shows 57% of BODIPY 493/503 signal per worm in comparison to that of N2. The daf-7 mutants accumulated twice as much BODIPY 493/503 as the wild type. The N2 and glp-1 strains did not differ in staining intensity when grown at the permissive temperature (15°C). However, at restrictive temperature (25°C), when germ cell development is inhibited, the glp-1 mutant showed a 1.5-fold higher BODIPY 493/503 intensity per worm than the wild type. Our BODIPY 493/503 data are in accordance with those of Oil Red O staining, although this procedure caused remarkably high (about one-third) shrinkage of   To measure the fat content per volume at the singleworm level in large C. elegans populations, we developed a high-throughput method based on BODIPY 493/503 and the fl ow cytometry sorting system, COPAS. In principal, the method simultaneously quantifi es the integrated fl uorescence intensity of BODIPY 493/503 (green fl uorescence), the length of the worms (TOF), and the body volumedependent decrease in laser light (extinction, 488 nm) of single worms. Because fi xative BODIPY 493/503 protocol may infl uence the volume of the worms, we checked this putative confounder. There was no signifi cant change in worm volume when vital and fi xed animals were compared ( Fig. 8A ). The extinction signals of worms obtained by The body volumes of the different strains vary by a factor of 3 ( Fig. 7D ). The eat-2 mutant shows a markedly lower body volume (1.08 ± 0.14 nl/worm) than the N2 (2.7 ± 0.08 nl/worm). The volume of the daf-7 mutant is 1.3-fold higher than the wild type. The body volume of the glp-1 and N2 strains are almost identical at permissive temperature (15°C). At restrictive temperature, the glp-1 mutant shows a 20% reduction in body volume in comparison to that of N2. Interestingly, the TAG content per volume ( Fig. 7F ) is specifi c for each strain and represents leanness of the eat-2 mutant (64% of N2) and fatness of the daf-7 (150%) and glp-1 (163%) mutants. In contrast, the protein-to-volume ratios representing the proportions of fat-free mass in the worm are constant among strains ( Fig. 7E ). Taken together, the fat-to-fat-free mass and the fat-to-volume ratio   Table 1 . red is, as also reported by Brooks et al. ( 15 ), a useful quantitative marker for worm TAG content, demonstrated here with fl ow cytometry using the COPAS Biosort instrument.

A novel vital staining protocol with BODIPY 493/503 enables the labeling of lipid droplets in vital worms
In addition to fi xative staining of worms with BODIPY 493/503, we developed a new vital staining protocol. In contrast to feeding of worms at NGM plates with fl uorescent dyes over a long time period, worms were stained in M9/BODIPY 493/503 solution for 20 min. For visualization of BODIPY 493/503 fl uorescence in vitally stained worms, microscopy settings were used that did not allow the detection of autofl uorescence signals in the green emission spectrum in unstained worms. Confocal threedimensional (3D) projection images ( Fig. 9 ) demonstrate that vesicles in intestine and hypodermal cells in the head region, the main body hypodermal syncytium hyp7, and the tail region, as well as in gonads and oocytes but not in embryos are stained. To verify that the stained compartments are lipid droplets and not LROs, we performed colocalization studies. First, we determined whether the vital BODIPY 493/503 staining before fi xation is in accordance with LipidTox deep red staining after fi xation. As shown in Fig. 10A , the vitally BODIPY 493/503-stained structures in head and proximal intestine colocalize with the postfi xation LipidTox-stained lipid droplets. Second, we show that the vitally stained organelles do not overlap with LysoTracker red, a marker for acidic autofl uorescent LROs ( 13,38 ) ( Fig. 10B ). Third, the BODIPY 493/503 signals are in agreement with the CARS signals and do not colocalize with the autofl uorescent signals of the LROs ( Fig. 10C ,  and supplementary Fig. IIA). Of note, BODIPY 493/503feeding worms on NGM plates for 18 h causes dye accumulation in the autofl uorescent LROs (supplementary Fig. IIB). Thus, our data indicate convincingly that BODIPY 493/503 labels lipid droplets in vital worms when applied to the worm for a short time period. In Fig. 10D the fl ow-cytometric analysis of BODIPY 493/503 fl uorescence and extinction of vitally stained wild-type N2 and eat-2 , daf-7 and glp-1 (25°C) COPAS correlated well ( r 2 = 0.81) with their volumes ( Fig. 8C ) and were not infl uenced by fi xation ( Fig. 8B ). Thus, the COPAS-derived extinction signals of BODIPY 493/503-fi xed worms can be used as an accurate proxy for the volume of the worms. Next, we measured BODIPY 493/503 fl uorescence emission of fat storage mutants by using COPAS. The resulting averaged BODIPY 493/503 signals per worm of the eat-2 , daf-7 , and glp-2 mutants are in agreement with fat contents obtained by BODIPY 493/503 microscopy and biochemical measurement ( Table 1 ). Finally, BODIPY 493/503 fl uorescence and extinction values of single worms were plotted. As shown in Fig. 8D, E , populations of the eat-2 , daf-7 , and N2 strains show distinct distributions and differences in fat contents and body volumes. To illustrate this for the glp-1 mutant in more detail, N2 worms and glp-1 mutants showed almost identical BODIPY 493/503 signals, body lengths ( Fig. 8F ), and body volumedependent extinction signals ( Fig. 8H ), indicating similar body compositions for both strains at permissive temperature. At restrictive temperature (25°C), the glp-1 mutants showed higher fat content but similar body lengths compared to those of N2 ( Fig. 8G ). The body volume-dependent extinction signals of the glp-1 strain were slightly lower than those of N2 ( Fig. 8I ) and are in agreement with direct body volume measurements using image analysis ( Fig. 7D ). As a fi nal result, the glp-1 mutant showed a 1.6-fold higher fatto-volume ratio than the N2 strain at restrictive temperature, indicating fatness of the glp-1 mutant. Thus, data for BODIPY 493/503-based fl ow cytometry represents the proportion of fat at the single-worm level in large C. elegans populations.
Furthermore, we stained the eat-2 , daf-7 , and glp-2 mutants with Nile Red after fi xation and measured the red fl uorescence and extinction by fl ow cytometry. As shown in supplementary Fig. I, where red fl uorescence signal and extinction are plotted, the relative fl uorescence data provide quantitative information about fat content of the mutants in relation to that of the wild type N2. For example, the glp-1 mutant shows a 1.8-fold higher fat-to-volume ratio than the N2. Therefore, fi xative Nile a Biochemical, TAG, measurement was made for each homogenate in duplicates in two independent measurements, respectively. We performed two to three independent experiments for each genotype and condition with two replicates, respectively. b BODIPY 493/503 mean intensity (FU, fl uorescence units) obtained by microscopy was measured for 20-30 worms from two to three independent experiments. Representative BODIPY 493/503 phenotypes are shown in Fig. 4 . c BODIPY 493/503 mean intensity (FU, fl uorescence units) obtained by COPAS fl ow cytometry was measured for 250-2,000 nematodes in each of two to three independent experiments. Data shown are means ± SEM of two to three independent experiments. agreement with the observation of Zhang et al. ( 14 ), who found that fi xative Nile red shows only the lipid dropletspecifi c emission spectrum, in contrast to that of the vital staining method with Nile red ( 14 ). Fixative Oil Red O and Nile red staining were established as a suitable and quantitative method to stain the main fat stores in C. elegans ( 10,15 ). In comparison to the Oil Red O procedure, our BODIPY 493/503 protocol easily detects small lipid droplets because of high sensitivity and resolution of the fl uorophore. Particularly in regions with high lipid droplet density, visualization of single droplets is feasible with BODIPY 493/503 stain. Furthermore, the BODIPY 493/503 staining procedure preserves the morphology of lipid droplets because the protocol omits any dehydration steps using isopropanol or other alcohols. It has been reported that isopropanol leads to fusion of small adjacent lipid droplets ( 40 ). Thus, the BODIPY 493/503 staining procedure seems to be a reliable method to explore lipid droplet mutant worms is shown. The resulting fl uorescence distribution patterns of the strains are very similar to those obtained by our fi xative BODIPY 493/503 procedure ( Fig. 8 ).

DISCUSSION
We characterized the body composition of C. elegans fat storage mutants by using novel BODIPY 493/503-based microscopy and fl ow cytometry protocols. Comparison of images obtained with our protocol and label-free imaging in vivo using CARS microscopy revealed similar intestinal and hypodermal fat distributions and lipid droplet morphology of the N2 worms. The size of the lipid vesicles (<3 µm in diameter) stained with BODIPY 493/503 fi ts well with that recently reported ( 39 ). Colocalization of BODIPY 493/503-stained structures with the label-free CARS and Raman lipid signals also proves that BODIPY 493/503 exclusively stains lipid droplets. This fi nding is in Data are presented as means ± SEM from two to three independent experiments. Signifi cant differences were analyzed using ANOVA followed by unpaired t -test of mutants with the wild-type under identical temperature conditions; and wild-type worms under different conditions; *, P < 0.05; **, P < 0.005; ***, P < 0.001. Bars indicate SEM. fi xative LipidTox ( 10 ) but do with LRO markers ( 10,11,13 ). In contrast, our BODIPY 493/503 signals in vital worms colocalized with fi xative LipidTox signals as well as with CARS signals and do not colocalize with the autofl uorescent signals of the LRO markers. The feeding of dyes over a longer period, as used in common vital staining protocols, may lead to metabolism of lipid dyes like morphology in C . elegans and may be useful for colocalization studies using immunofl uorescence or mCherry reporter.
We developed an easy protocol for vital staining with BODIPY 493/503 that enables the labeling of lipid droplets in vital worms. Common vital fat staining procedures in C. elegans showed that the dyes do not colocalize with correlate well with those found by Brooks et al. ( 15 ). We transformed the data from Brooks et al. to TAG per phospholipids to make them comparable with our data. After this transformation, we found that the eat-2 mutant contains 60% and the daf-7 mutant contains 169% TAG/phospholipid of the wild-type strain, which are consistent with our own data. The glp-1 mutant (25°C) contains 163% TAG/volume of N2. This is lower than the TAG data (297%) reported by O'Rourke et al. ( 10 ) but correlates very well with their Oil Red O staining results (147%) and our own BODIPY 493/503 quantifi cation data (155%) for the glp-1 mutant. Because the data obtained by BODIPY 493/503 fl ow cytometry represent fat content per volume at a single-worm level, this high-throughput fat assessment method will allow an effi cient and accurate screening of C. elegans fat storage mutants. Such a screening may be confounded by the number of eggs and embryos stored in the uterus of the hermaphrodites. For example, the eat-2 and glp-1 mutants (restricted temperature) have fewer or no eggs in comparison to that of N2 worms, whereas the daf-7 mutants retain more eggs and early-stage embryos. These differences infl uence triglyceride as well body volume measurement. Thus, we assumed that the resulting fat content per volume is not substantially infl uenced by different numbers of eggs. To address our hypothesis, the fat and protein content of the eggs and embryos should be determined in future studies.
In contrast to the TAG to volume ratio that seems to be specifi c for each mutant, protein-to-volume ratios refl ecting the proportion of fat-free mass are very similar when comparing different conditions and strains. In support of this, Jonassen et al. ( 43 ) showed that wet weight-to-protein ratio (11.8) is constant between larvae and adults . This ratio is in good agreement with our protein-to-volume ratio (12.5), assuming a worm density of ‫ف‬ 1 g/cm Ϫ 3 . Furthermore, xenobiotics that need to be degraded or stored in lysosomelike organelles. This is not the case in our vital staining procedure because we used short-term incubation of worms with BODIPY 493/503. Our method is in agreement with staining methods of fat in cell culture, where the staining occurs also within minutes ( 41 ). For future studies, our vital staining protocol in combination with the fl ow cytometry approach could provide an important advantage in screening C. elegans mutants with perturbed fat stores.
Different methods are employed to study lipid content and composition of C. elegans ( 1 ). Our quantifi cation results of fat content per worm, using an enzymatic assay and BODIPY 493/503-based microscopy and fl ow cytometry, revealed a good agreement among these approaches. The fat contents of fat storage obtained from the eat-2 , daf-7 , and glp-1 mutants are also in accordance to other lipid biochemistry methods and reports that employed Nile red ( 15 ) or Oil Red O ( 10 ) as fi xative dyes. In contrast to vital staining protocols, postmortem analysis of C. elegans fat content provides accurate results with different dyes and biochemical methods. A detailed analysis of reported fat contents in C. elegans revealed different reference values such as per worm phospholipid content or protein content. This is of importance with respect to data comparison as well as body composition analysis and assessment. The C. elegans wild type displays a TAG fraction of 40%-55% of total lipids ( 7,13,15 ). Thus, TAGs can be represented as TAG per phospholipid ( 31,42 ). The resulting TAG data are referred to cell membranes that are proxy for the volume of an organism. Here, we used the volume of the worm as a direct reference value to describe the proportion of body fat in fat storage mutants. We found that the eat-2 and daf-7 mutants contain 64% and 150% TAG per volume (w/v) of the N2, respectively. These data to-protein ratio of 0.2. This is in contrast to that of other species. For example, rats have a TAG-to-protein ratio of 0.8 ( 48 ). However, the low fat content and high protein level of C. elegans refl ects the low fat-to-protein ratio of E. coli , the food source of the worms ( 15 ). Fatty acid absorption from diet contributes essentially to saturated fatty acids palmitic (C16:0) and sterinic acid (C18:0) content in the nematode, because de novo synthesis is very low (5% and 7%, respectively), whereas the de novo synthesis of MUFAs and PUFAs is up to 20% ( 42 ). Brooks et al. ( 42 ) showed that there is an inverse correlation of C17iso fatty acid and TAG. C17iso fatty acid is de novo synthesized from the branched chain amino acid leucine. A selective protein deprivation as it is given in pep-2 mutants that lack intestinal peptide transporter activity ( 49 ) results in a decrease in C17iso fatty acid and increased TAG storage, whereas an overall dietary restriction (e.g., in the eat-2 worm) does not change the C17iso fatty acid concentration but reduces the content of TAGs in total lipid ( 1 ) or volume (our data). Also, in mammals, high-protein diets result in a shift to a higher ratio between lean and fat mass ( 50 ) and a reduction in food energy intake ( 51 ). Hereby, particularly leucine stimulates protein synthesis in muscle ( 52 ) and mediates hormonal hypothalamic responses by activating mTOR signaling ( 53,54 ). In Drosophila , the fat-to-fat Houthoofd et al. ( 44 ) found similar protein-to-volume ratios for the N2 wild type and the eat-2 (ad1113) mutant. Protein is also commonly used as a reference value for TAG. Our data for N2 worms (0.16-0.23 µg of TAG/µg of protein) are supported by other enzymatically determined TAG levels: 0.2 µg of TAG/µg of protein ( 45 ) and 0.12 µg of TAG/µg of protein ( 46 ). TLC/GC measurements revealed a ratio of 0.15 ( 7 ). Brooks et al. reported lower ratios ( ‫ف‬ 0.06-0.07), and Srinivasan et al. reported 10-fold higher values ( 47 ). Differences in ratios may be due to different extraction and analysis methods and may be dependent on the developmental stage of nematodes or composition of E. coli food source.
Based on the literature and our own data, we estimated the average body composition of adult wild-type N2 worms. The C. elegans wild type consists of ‫ف‬ 16 ng of TAG/nl, which is about 1.6 w/v % of TAG (w/v %). Assuming a phospholipid fraction of 45%-60% of total lipid, the worm contains 3-4 w/v % lipids. The proportion of protein ( ‫ف‬ 78 ng/ nl) is about 7.8 w/v %. Assuming a water content of about 80%-85% ( 43 ), there is a difference of 5 to 10 w/v % that may be carbohydrates and insoluble matter, such as components of cuticula. Taken together, the C. elegans wild type contains about 5-fold less TAG (1.6 w/v %) than protein (7.8 w/v %), resulting in a TAG- free ratio is highly variable depending on macronutrient intake ( 55 ). An intriguing question is whether the body composition in C. elegans is variable under different dietary protocols such as reduction of protein content in combination with high carbohydrate or high fat diets.
In conclusion, we developed BODIPY 493/503-based approaches in order to characterize fat storage and body composition of C. elegans mutants. The procedures have a number of advantages. First, the worm anatomy and the native morphology of lipid droplets are preserved. Second, the fl uorescent dye allows an accurate and very sensitive quantifi cation of body fat. Third, fl ow cytometry using BODIPY 493/503-stained worms displays fat quantity per volume at a single-worm level, allowing an accurate screening of fat storage mutants. Using these methods, we characterized the body composition of C. elegans and found that the fat content per volume is specifi c for dietary restriction, TGF-␤ , and germline mutants. In contrast, the proportion of fat-free mass is constant among these strains. Together, BODIPY 493/503-based and fl ow cytometry approaches allow sensitive and accurate assessment of adiposity in large C. elegans populations at a single-worm level.