• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. Sep 11, 2007; 104(37): 14658–14663.
Published online Sep 5, 2007. doi:  10.1073/pnas.0703594104
PMCID: PMC1976189

Monitoring of lipid storage in Caenorhabditis elegans using coherent anti-Stokes Raman scattering (CARS) microscopy


Better understanding of the fundamental mechanisms behind metabolic diseases requires methods to monitor lipid stores on single-cell level in vivo. We have used Caenorhabditis elegans as a model organism to demonstrate the limitations of fluorescence microscopy for imaging of lipids compared with coherent anti-Stokes Raman scattering (CARS) microscopy, the latter allowing chemically specific and label-free imaging in living organisms. CARS microscopy was used to quantitatively monitor the impact of genetic variations in metabolic pathways on lipid storage in 60 specimens of C. elegans. We found that the feeding-defective mutant pha-3 contained a lipid volume fraction one-third of that found in control worms. In contrast, mutants (daf-2, daf-4 dauer) with deficiencies in the insulin and transforming growth factors (IGF and TGF-β) signaling pathways had lipid volume fractions that were 1.4 and 2 times larger than controls, respectively. This was observed as an accumulation of small-sized lipid droplets in the hypodermal cells, hosting as much as 40% of the total lipid volume in contrast to the 9% for the wild-type larvae. Spectral CARS microscopy measurements indicated that this is accompanied by a shift in the ordering of the lipids from gel to liquid phase. We conclude that the degree of hypodermal lipid storage and the lipid phase can be used as a marker of lipid metabolism shift. This study shows that CARS microscopy has the potential to become a sensitive and important tool for studies of lipid storage mechanisms, improving our understanding of phenomena underlying metabolic disorders.

Keywords: lipid metabolism, nonlinear microscopy, obesity

The world-wide increase in obesity has intensified the need for better understanding of metabolism regulation and mammalian lipid storage. Imbalance in this complex network of processes is thought to be the origin of many widespread diseases such as type-2 diabetes, cardiovascular diseases, and several types of cancers (1). Most studies on energy storage have been based on biochemical analysis of mammalian cell culture systems (2). However, lipid storage is genetically and physiologically regulated, which necessitates systemic studies only possible in intact living animals. Even an organism as distant from mammals as the nematode Caenorhabditis elegans can provide insights not attainable by studying, e.g., isolated mammalian adipocytes. A majority of the ≈400 genes identified to affect lipid storage in C. elegans have human homologs (3), making C. elegans a highly relevant model organism in this respect. Its lipid deposits, primarily located in the intestinal cells, have been studied microscopically by staining fixed nematodes (4) or, more recently, by adding a fluorescent dye to the diet (3). Unfortunately, these techniques are associated with severe limitations. Fixation and staining procedures are highly invasive and do not allow any in vivo studies under natural conditions. The fluorescent-dye diet method is biased by differences in food uptake efficiency, metabolic rate, distribution properties of the dye and the physical properties of the surrounding medium, as will be revealed in this study. In principle, the lipid molecules could also be specifically tagged by fluorescent marker molecules (5). However, the tags affect the intrinsic properties of the lipids due to their bulkiness and impact on the hydrophilic/hydrophobic balance (6), which in turn may change the cellular lipid trafficking and metabolism. For these reasons it is desirable to develop a microscopy technique that utilizes inherent properties of the sample for image contrast. The category of nonlinear microscopy techniques fulfills this requirement, one of which, third harmonic generation (THG), has been used for the visualization of lipid droplets in cells, plant seeds, and tissue (7). However, THG microscopy brings forward all interfaces between cellular structures with different nonlinear properties. Plasma and nuclear membranes emerge in THG microscopy images (8), as well as membrane-based organelles such as mitochondria (9). It consequently presents an unspecific image of the cellular morphology, without any information on the molecular composition (10). Here, we show that imaging of lipids instead can be accomplished with high chemical selectivity by means of another nonlinear microscopy technique using the coherent anti-Stokes Raman scattering (CARS) process. Thus, CARS microscopy is able to provide insights into fat metabolism and storage with molecular specificity, giving it a strong potential to become a vital instrument within metabolism research.

In CARS microscopy, molecules are visualized without the need for labeling by probing their characteristic vibrational properties (11). Additional virtues are high intrinsic 3D resolution, deep-penetration depth in tissues, and low average excitation powers. The CARS process is induced by simultaneously illuminating the specimen with two laser beams at frequencies ω1 (pump/probe beam) and ω2 (Stokes beam), separated by a vibrational frequency of the target molecule Ω = ω1−ω2. The nonlinear interaction between the laser light and the molecule is then resonantly enhanced, giving rise to a strong CARS signal at the frequency ω1 + Ω (12). For selective imaging of lipids, the vibration of the carbon–hydrogen bond at 2,845 cm−1 can be probed, here with excitation wavelengths of 817 nm (pump/probe) and 1,064 nm (Stokes). The CARS process is schematically illustrated in the energy level diagram inserted in supporting information (SI) Fig. 5. For imaging, the tightly focused laser beams are scanned over the specimen as for confocal microscopy, and the generated, blue-shifted CARS photons are detected at each point. Concentration differences of the target molecule result in intensity variations of the CARS signal, forming the molecule-specific contrast in the image. The prospects of the technique have been shown for single adipocytes (13), axons (14), and even components in skin tissue of living mice (15).

After initial proof-of-principle measurements on C. elegans (16), the unique capabilities of CARS microscopy are here fully used for quantitative imaging of lipids in populations of different strains in vivo, allowing holistic understanding of lipid storage mechanisms. Specifically, we examined the influence of environmental conditions and genetically regulated signaling pathways on how energy is stored or alternatively used for reproduction and cell maintenance. In C. elegans, this is partly regulated by chemosensory systems responding to environmental changes such as food availability, population density, and temperature. Under favorable conditions, growth to fast-metabolizing adults is the norm, whereas harsh environmental conditions trigger arrest as lipid-storing, slow-metabolizing, and long-lived dauer larvae (17). The two most important chemosensory systems regulating the metabolism are the insulin/insulin-like growth factor (4) and TGF-β pathways (18). These organism-wide systems form together with the local cellular energy status and nutrient availability input signals to the target of rapamycin (TOR) signaling pathway (19). The TOR regulatory network promotes anabolic cell growth by up-regulating mRNA transcription at favorable conditions (20) and autophagy under reduced nutrient conditions (21).

Although the lipid droplet was long considered a mere compartment for storage of excess fatty acids, it is now recognized as a fundamental and dynamic organelle, actively controlling cellular lipid homeostasis (22, 23). Hence, there is a need for methods to characterize and monitor the dynamic behavior of this cell organelle. In this work, we establish CARS microscopy for this purpose by drawing attention to its benefits relative to the limitations of fluorescence microscopy. The capability of CARS microscopy in providing chemically specific image contrast is explored by spectral measurements. This allows us to gain new insights into how the phase of lipids changes when the metabolic route of lipid storage is promoted. Furthermore, we have visualized and characterized lipid droplets during postembryonic growth in C. elegans involving four larval stages (L1–L4) before adult size is reached. This was conducted for wild type (N2) as well as the daf-2 and -4 mutants with defective insulin growth factor (4) and TGF-β (24) pathways, respectively, representing growth during both normal (N2) and unfavorable (daf-2, daf-4 dauer) conditions. In addition, a feeding-deficient strain (pha-3) (25) was studied, resulting in starvation solely sensed on a cellular level in contrast to the organism-wide growth factor signaling pathways. Using CARS microscopy we were able to monitor the lipid volume fraction in unlabeled intestinal and hypodermal cells, as well as the size and 3D distributions of the lipid stores.

Results and Discussion

CARS vs. Fluorescence Microscopy.

Fluorescence microscopy is the technique commonly used for visualization of lipids on a cellular level. In living C. elegans nematodes, the Nile red-diet method is considered as the most useful protocol (3). This motivates a comparison between the fluorescence and CARS microscopy approaches. For wild-type nematodes with normal feeding, metabolism, and lipid-storage mechanisms, the lipid droplets in the CARS and fluorescence images match well. Only a minor underestimation is made in the evaluation of the Nile red fluorescence image (11% vs. 15% in the CARS microscopy image). The feeding deficiency of the pha-3 mutant does not prevent the fluorescent dye from entering the intestine, and a corresponding degree of colocalization between the lipid droplets in the two images can be discerned (data not shown). However, for the dauer-arrested daf-4 mutants, a clear difference is observed. Although good correspondence is obtained for the intestinal cells, the lipid droplets in the hypodermal cells are only vaguely visible in the fluorescence images compared with the CARS images, as exemplified in Fig. 1. One explanation for this difference between fluorescence and CARS microscopy could be that the low metabolic rate of the dauer larvae reduces the exchange of metabolites from the intestine to the more peripheral hypodermal cells. An alternative interpretation is that differences in the environment of the fluorophores, i.e., of the lipids, account for the failure of Nile red to fluoresce in the hypodermal collection. The fluorescence yield of Nile red has been shown to depend highly on the phase of the surrounding medium, liquid phase being less favorable than the more ordered gel phase (26). This explanation is supported by the spectral CARS microscopy measurements discussed below. The consequence is that the hypodermal energy storage pool might be underestimated or neglected in studies that rely on staining. Whereas the CARS microscopy data indicate a lipid volume fraction of 26 ± 6% of the dauer-arrested larvae, the same analysis based on the corresponding fluorescence data results in a significantly lower value of 14 ± 3%, i.e., an underestimation with a factor of 1.9. It proves that the fluorescence-labeling technique must be used with caution for the visualization of lipid stores, because signal generation is influenced by variations in uptake/distribution of the dye as well as by local variations in the physical properties of the lipids. These uncertainties are completely avoided in CARS microscopy and provide a strong argument to promote the implementation of this technique for studies of such complex processes as lipid storage in living cells and organisms.

Fig. 1.
Comparison of CARS and fluorescence microscopy. (A) Autoscaled CARS and (B) two-photon fluorescence images (80 × 80 μm2, 20-s integration time) of a Nile red-stained daf-4 mutant (C. elegans) arrested in its dauer stage for 3 weeks. The ...

Imaging with Chemical Selectivity.

CARS microscopy is capable of differentiating between different organelles and tissue structures based on their characteristic and inherent vibrational properties. The selective imaging of lipids, permitted by their characteristic abundance in CH2 groups, allows us to specifically visualize intracellular lipid stores consisting of a high-density core of neutral lipids encircled by a monolayer of phospholipids, glycolipids and/or sterols (27). The selectivity achieved is demonstrated by the two CARS microscopy images of lipid droplets in the intestinal cells of a wild-type L4 nematode shown in Fig. 2. Whereas structures can only vaguely be discerned off resonance (e.g., 2,790 cm−1), the lipid droplets are clearly visualized at the resonance of the symmetric CH2 vibration (≈2,845 cm−1). Normalized CARS spectra of lipid droplets for the different C. elegans strains and surrounding tissue matrix are shown in Fig. 2 as the average CARS signal vs. the probed vibrational frequency (Raman shift). In contrast to the flat tissue matrix spectrum, the lipid droplets exhibit a strong resonance at the symmetric CH2 vibration, thus forming optimal image contrast. The spectra of the wild-type nematode also exhibit a peak at the weaker asymmetric CH2 vibration (≈2,880 cm−1), higher for the late- (L4) than the early-stage (L1) larva, which probably is a consequence of the difference in accumulated lipid stores. In the spectrum of the feeding-deficient pha-3 mutant, the CH2 peaks are replaced by spectral profiles of dispersive shape, typically obtained at low concentrations of the probed molecules (28) and in good accordance with the depleted lipid stores of the pha-3 mutant. With the increased accumulation of lipid stores typical for the daf-2 and -4 dauer mutants, a prominent asymmetric CH2 peak (≈2,880 cm−1) could be expected. In contrast, it is found to be insignificant relative to the symmetric CH2 vibration (≈2,845 cm−1) for these two species. In Raman spectroscopy, the ratio between the asymmetric and symmetric CH2 vibrations is used to determine the ordering of polymethylene chains (29) and phospholipids (30). A higher ratio is obtained for the gel phase, characterized by highly ordered methyl chains, compared with the liquid phase with less-ordered chains. This has also been confirmed for lipid droplets monitored by CARS microspectroscopy (31). In addition, a slight frequency shift of the symmetric CH2 peak can be noted, also characteristic of lipids in liquid phase (29). Thus, our data strongly suggest that the shift in metabolism to increased lipid storage observed for the stressed daf-2 and -4 (dauer) mutants is accompanied by a shift in the ordering of the lipids from gel (high ordering) to liquid (low ordering) phase. This is further supported by the significantly reduced fluorescence of Nile red observed in the hypodermal region of the labeled daf-4 (dauer) mutants, which is known to occur when Nile red is surrounded by less-ordered methyl chains (26). This illustrates the kind of insights that CARS microscopy offers via its capability of chemical imaging, and from this we conclude that the technique can be used for the visualization of lipid droplets in a living organism, characterized by a complex environment of other organic molecules, in contrast to, e.g., THG microscopy.

Fig. 2.
CARS spectra of C. elegans. Series of CARS images were collected of N2 L4, N2 L1, pha-3, daf-2, and dauer daf-4 nematodes in vivo simultaneously with a reference image of glass (no resonant features) for normalization while tuning the molecular vibration ...

Lipid Storage During Genetically Controlled Metabolism: pha-3, daf-2, and daf-4 Dauer.

Metabolic routes of specific interest, such as increased lipid storage in conjunction with environmental stress, can be studied by means of mutants genetically determined to follow these pathways. The ability to study lipid stores with a method not affected by differences in feeding behavior and metabolic rates offers possibilities to gain insights into the regulation mechanisms of cellular lipid storage. In the CARS microscopy images of Fig. 3 the results of genetically controlled metabolic shifts are observed for different mutants in terms of the degree and character of their lipid stores. The feeding-deficient pha-3 mutant allows us to study a condition of local nutritional restriction, achieved without involving the complex global signaling system responding to lack of nutrients in the environment (32). This results in depleted lipid stores, as illustrated by the CARS microscopy image in Fig. 3B. Significantly lower CARS signals are generated here from lipid droplets, compared with the wild type (Fig. 3A), indicating a lower concentration of lipids. The droplets are also of significantly smaller size (SI Fig. 7), resulting in a lipid volume fraction of 5.4 ± 5%, approximately three times lower than the 17 ± 5% of the wild type (diagram in Fig. 3F).

Fig. 3.
Quantitative analysis of CARS microscopy images of C. elegans. Representative images (42 × 42 μm2) of different mutants at the L4 larval stage are shown (intensity profiles are shown in SI Fig. 6). Compared with the wild-type (A), the ...

The daf-2 and -4 mutants allow us to indirectly study the influence of food deprivation, increasing population density and unfavorable environmental conditions, because their signaling system acts accordingly, despite beneficial environmental conditions: instead of growing to fast metabolizing adults they arrest as nonfeeding dauer larvae after a fat-accumulating L2 predauer condition (17). The daf-2 mutation studied here does not cause a complete blocking of the normal insulin-signaling pathway, which results in an intermediate phenotype with maintained reproductive growth as well as characteristic dauer-like features, such as lipid accumulation and extended life span (4). Such a phenotype is exemplified by the L4 daf-2 mutant (Fig. 3C). It clearly exhibits a higher density of lipid stores than the wild type, which is a consequence of the down-regulated glycolysis enzymes and up-regulated lipid synthesis enzymes (33). This change in metabolism seems to trigger the formation of new small-sized lipid droplets (SI Fig. 7), also in the hypodermal cells, which allows increased lipid storage. Still, the lipid accumulation is not as prominent as for the daf-4 mutant (Fig. 3D) that exhibits complete dauer morphogenesis, including arrest of cell division and feeding. In addition to the overall higher fraction of lipid droplets, the dauer larva also displays a characteristic dense collection of smaller lipid stores in the hypodermis. Again, this is confirmed by the analysis of the droplet size distributions (SI Fig. 7). Compared with the L3 wild type, the dauer larva exhibits a large pool of small-sized droplets primarily corresponding to the hypodermal collection. In quantitative terms, the daf-2 mutants hold a lipid volume fraction that is 1.4 times higher (23 ± 6%) than the wild type, and the recently arrested daf-4 dauer larvae as much as two times more (35 ± 3%). After 3 weeks of dauer arrest, the larvae still host a lipid volume fraction of 1.6 times (27 ± 5%) that of the wild type. Thus, the lipid stores are mobilized at a fairly low rate, assuring long-term survival.

The fraction of lipid stores in the hypodermal cells relative to the total amount in the larva is an interesting parameter, because it accompanies metabolic changes involving increased lipid storage. The wild-type and the pha-3 mutant exhibit a low fraction of the total lipid stores in the hypodermal cells (9 ± 3% and 13 ± 5%, respectively), whereas the daf-2 strain contains a somewhat larger fraction, although with a high degree of variability (18 ± 11%). The dauer arrested larvae store initially as much as 40 ± 4% of their lipid resources in the hypodermal cells, a number which increases with time (47 ± 11%, after 3 weeks of arrest). This increase could mean that the energy mobilization occurs to a higher degree in the intestinal than in the hypodermal cells. A study involving a larger population of long-term dauer larvae is needed to conclude this.

Lipid Storage During Larval Development.

During larval development, there is a delicate balance in the use of energy resources for either growth or storage and the ability to visualize these processes provides improved insight how this is maintained. Fig. 4A shows a representation of the energy reservoir for a C. elegans larva of the feeding-defective pha-3 mutant at the start of its life. A regular and aligned structure can be seen along the ventral cord, most likely yolk granules within the intestinal cells. The corresponding volume image, formed from the stack of 2D CARS images, is shown in Fig. 4B. In L1 larvae, the early yolk granules are complemented by an accumulated collection of more homogeneously distributed lipid droplets in the intestinal cells (Fig. 4 C and D). The dark areas lacking lipid droplets are most likely the intestinal nuclei. In later larval stages, the increasing volume of the intestinal cells allows a more substantial accumulation of lipid stores, as shown for the L4 larva in Fig. 4 E and F. This is supported by SI Fig. 7, showing that lipids are accumulated in larger formations with increasing developmental stage. The inability of the feeding-deficient pha-3 mutant to attain a normal lipid volume fraction during its first developmental stages results in reduced growth and consequently in the formation of a small-sized adult (SI Fig. 8). In contrast, for the daf-2 and -4 (dauer) mutants, the increase in lipid volume fraction during stages L1–L3 is significantly higher than for the wild-type larva. This is in turn accomplished by a reduced growth rate (SI Fig. 8). In all, this illustrates the close connections between the metabolism regulation and developmental growth, where the amount of stored lipids seems to play an important role.

Fig. 4.
CARS microscopy volume images. Normalized CARS volume images (B, D, F, and H) were reconstructed from z-stacks, represented by one of the optical sections shown in A, C, E, and G (A, 58 × 58 μm2; C, 26 × 26 μm2; E, 45 × ...


After several years with focus on technical development, this work now establishes CARS microscopy for full-scale biological studies, here of lipid storage in a living organism, the nematode C. elegans. The major benefit of imaging without possible influence of a reporter molecule has been demonstrated by comparative fluorescence and CARS microscopy measurements. The volume images illustrate the strength of CARS microscopy in terms of 3D sectioning capacity for unbiased and detailed information on the morphology and overall geometrical arrangement of lipid stores in a living organism. The lack of nutrients at the cellular level results in small depleted lipid droplets in the intestinal cells and a reduced lipid volume fraction. In conjunction with a genetically controlled shift in the metabolism favoring lipid synthesis and down-regulation of the glycolysis, as for the daf-2 and -4 dauer larvae, the lipid volume fraction is instead significantly increased and characterized by large intestinal droplets as well as a dense population of small hypodermal droplets. CARS microspectroscopy measurements indicate that this is accompanied by a shift in lipid ordering from gel (ordered structure) to liquid (less-ordered) phase. Thus, the monitoring of changes in lipid phase, as well as quantitative analysis of the hypodermal population of lipid stores, reveals shifts in cell metabolism. CARS microscopy provides a unique means for such studies, allowing improved insights into how the delicate balance of metabolic processes is maintained or shifted. Studies of this nature are crucial for a fundamental understanding of obesity-related diseases.

Materials and Methods

Sample Preparation.

The Bristol variety of C. elegans, the N2 strain (wild-type parent), the pha-3 strain (ad607) from linkage group (LG) III, and the daf-2 (e1370) and daf-4 (cb1364) strains from LG IV were obtained from the Caenorhabditis Genetics Center. The nematodes were cultured on agar plates seeded with Escherichia coli at 20°C and handled as described by Sulston and Hodgkin (34). To induce environmentally triggered lipid accumulation, daf-2 larvae were kept at the nonpermissive temperature of 25°C overnight before examination. Full dauer phenotype was attained by exposing the daf-4 strain to 25°C for >24 h before sample preparation. Besides daf-4 dauer larvae, samples of the N2, pha-3, and daf-2 strains at the preadult stages L1–L4 and as 1-day-old adult were selected for CARS microscopy. In all, 60 larvae were characterized. The correct developmental stage was established by examining the gonad and vulva by using differential interference contrast (Zeiss Axioplan; Zeiss, Oberkochen, Germany) microscopy at ×400 and ×1,000 magnification. Two to three larvae at the same developmental stage were placed in a droplet of Levamisol (100 mM) on a 0.17-mm-thick microscopy cover glass covered with a thin agarose film. A second cover glass was then gently applied on the sample.

The CARS Microscope.

The CARS microscope (see schematic outline in SI Fig. 5) consists of an inverted microscope (Nikon, Eclipse TE2000-E; Nikon, Kawasaki, Kanagawa, Japan) and a near-infrared laser system generating three completely synchronized beams. One beam is directly guided into the microscope from a high-power picosecond Nd:Vanadate laser at 1,064 nm, whereas the main part of its output is used to pump two optical parametric oscillators (OPO) with individually variable output wavelengths in the ranges 785–845 and 852–920 nm, respectively. Various combinations of these two OPO beams and the fundamental pump beam allow us to probe any molecular vibration by means of the CARS scheme (SI Fig. 5) in the important range of ≈0–3,400 cm−1. Before the excitation beams enter the microscope, they are temporally and spatially overlapped by means of two delay lines and dichroic mirrors. For 3D scanning, the microscope is equipped with a laser-scanning head and a motorized focusing stage with a high numerical aperture objective (Nikon Plan Apochromat, ×100, oil, N.A., 1.4). The generated CARS signal is collected in the forward direction through a second objective (Nikon Plan Fluor, ×40, oil, N.A. 1.3). The CARS photons are detected by a photomultiplier (PMC 100-20; Hamamatsu, Hamamatsu City, Japan) equipped with suitable band-pass filters and subsequently counted by a time-correlated single-photon counting device (SPCM-830; Becker & Hickl, Berlin, Germany).

Comparative Fluorescence Measurements.

Two-photon fluorescence images were collected on C. elegans nematodes, cultured for 24–48 h on the fluorescent dye Nile red [1 ng/ml 5H-benzo[α]phenoxazine-5-one-9-diethylamino (35)]. Only the 1,064-nm beam was used for excitation. By unblocking the 817 nanometer beam, a CARS image, in addition to the two-photon fluorescence from the Nile red, was collected. Two-photon/CARS twin images were collected for each of the wild-type (L4), pha-3 (L4), daf-2 (L4), and dauer-arrested daf-4 strains.

Imaging with Chemical Selectivity, a CARS Spectrum.

Spectral CARS microscopy measurements were carried out with the purpose of ascertaining the chemical selectivity of the image contrast. CARS spectra were formed for each of the strains studied (N2 L1, N2 L4, pha-3, daf-2, and daf-4 dauer) by collecting 21–41 CARS microscopy images for vibrational frequencies in the range of 2,773–2,955 cm−1, the C-H bond region. This was accomplished by tuning the wavelength of the probe/pump beam between 810 and 822.2 nm in steps of ≈0.3 nm, while keeping the Stokes beam constant at 1,064 nm. The integration time was 60 s for each image, and the excitation powers at the sample were 10–38 mW for each of the Stokes and pump/probe beams. For each image, a corresponding nonresonant image of the cover glass was collected and used for normalization, allowing variations in excitation powers and beam overlap during the scan to be taken into account. The normalized mean intensity of the lipid droplets, identified by automatic thresholding (details below), was plotted vs. the vibrational frequency, forming CARS spectra.

Imaging of Lipid Stores.

Sixty z-stacks of CARS microscopy images were collected, each consisting of 19–100 slices covering a depth of 9.9–40 μm and an area of 20.5 × 20.5 μm to 100 × 100 μm in the xy plane (256 × 256 pixels), depending on the size of the larva studied. A central 3D region just anterior to the vulva was depicted, as illustrated in Fig. 3E. In all, 16 z-stacks were collected on wild-type larvae (N2), 18 stacks on pha-3, 17 stacks on daf-2, and 9 on dauer-arrested daf-4 mutants (4 immediately after entrance and 5 after 3 weeks). It resulted in two to five stacks per larval stage (L1, L2, L3, L4, and 1-day adult) within each category. In addition, two z-stacks were collected on entire larvae immediately before hatching, as exemplified by Fig. 4 A and B. Typical excitation powers at the sample were 7.5 + 15 mW (Stokes + pump/probe beams), and the integration time was 20 s.

Each stack of CARS images was quantitatively evaluated by means of ImageJ (36), with respect to (i) the volume fraction of lipid deposits, (ii) the relative fraction present in the hypodermal cells, (iii) the body size in terms of the body radius, and (iv) the radius and geometrical distribution of the lipid droplets. The lipid droplets were automatically identified through adaptive thresholding, and the ability of this routine for quantitative analysis of CARS microscopy images was evaluated from images of polystyrene spheres with well known diameters. Applying the evaluation routine to these data resulted in estimated diameters of 0.94 and 2.5 μm for spheres having specified mean diameters of 1.072 and 2.836 μm, respectively. Hence, the diameters were underestimated by typically ≈12%. The volume fraction of lipid stores in the nematodes was evaluated by summing up the voxels defined as of lipid droplets in a well defined ellipsoidal volume of interest for each z-stack, as shown in Fig. 3E. Care was taken that the ellipsoid included as much as possible of the larva body, i.e., the circumference touched the body wall. The volume of thresholded voxels was divided by the total volume of the ellipsoid, resulting in the volume fraction of lipid stores. The 3D distribution of lipid stores was visualized with the plug-in routine VolumeJ (37). Further quantitative parameters were evaluated from the central cross-section image of the larva in each z-stack. The radius of the larva body was quantified as an average distance from the intestinal lumen to the outer boundary of the body. This was done so as not to include irrelevant variations in the intestinal lumen due to differences in feeding. The droplets were categorized as either intestinal or hypodermal, depending on their localization. The total droplet area in the hypodermal cells was computed and related to that in the entire cross-section of the larva, forming the hypodermal fraction of lipid stores. The size of the lipid droplets was monitored with larval development for each strain and quantified as an equivalent radius calculated from the droplet area obtained with the particle analyzer in ImageJ.

Supplementary Material

Supporting Figures:


Madeleine Åkeson is gratefully acknowledged for her assistance in the spectral CARS microscopy measurements. We thank the C. elegans Genetics Center (funded by the National Institutes of Health Center for Research Resources), in particular Theresa Stiernagle. We appreciate financial support from the following Swedish organizations: Vetenskapsrådet, Cancerfonden, Carl Tryggers Stiftelse, Fredrik och Thurings Stiftelse, as well as from the Deutsche Akademie der Naturforscher Leopoldina (T.H.).


coherent anti-Stokes Raman scattering
third harmonic generation.


The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

This article contains supporting information online at www.pnas.org/cgi/content/full/0703594104/DC1.


1. Spiegelman BM, Flier JS. Cell. 2001;104:531–543. [PubMed]
2. Rosen ED, Spiegelman BM. Annu Rev Cell Dev Biol. 2000;16:145–171. [PubMed]
3. Ashrafi K, Chang FY, Watts JL, Fraser AG, Kamath RS, Ahringer J, Ruvkun G. Nature. 2003;421:268–272. [PubMed]
4. Kimura KD, Tissenbaum HA, Liu Y, Ruvkun G. Science. 1997;277:942–946. [PubMed]
5. Maier O, Oberle V, Hoekstra D. Chem Phys Lipids. 2002;116:3–18. [PubMed]
6. Kaiser RD, London E. Biochim Biophys Acta. 1998;1375:13–22. [PubMed]
7. Débarre D, Supatto W, Pena AM, Fabre A, Tordjmann T, Combettes L, Schanne-Klein MC, Beaurepaire E. Nat Methods. 2006;3:47–53. [PubMed]
8. Chu S-W, Chen SY, Tsai TH, Liu TM, Lin CY, Tsai HJ, Sun CK. Opt Expr. 2003;11:3093–3099. [PubMed]
9. Barzda V. Opt Expr. 2005;13:8263–8276. [PubMed]
10. Débarre D, Beaurepaire E. Biophys J. 2007;92:603–612. [PMC free article] [PubMed]
11. Cheng J-X, Xie XS. J Phys Chem B. 2004;108:827–840.
12. Shen YR. The Principles of Nonlinear Optics. New York: Wiley; 1984.
13. Nan X, Cheng J-X, Xie XS. J Lipid Res. 2003;44:2202–2208. [PubMed]
14. Wang HF, Fu Y, Zickmund P, Shi RY, Cheng JX. Biophys J. 2005;89:581–591. [PMC free article] [PubMed]
15. Evans CL, Potma EO, Puoris'haag M, Cote D, Lin CP, Xie XS. Proc Natl Acad Sci USA. 2005;102:16807–16812. [PMC free article] [PubMed]
16. Burkacky O, Zumbusch A, Brackmann C, Enejder A. Opt Lett. 2006;31:3656–3658. [PubMed]
17. Riddle DL, Albert PS. In: C. elegans II. Riddle DL, Blumenthal T, Meyer BJ, Priess JR, editors. New York: Cold Spring Harbor Lab Press; 1997. pp. 739–768.
18. Savage-Dunn C. Cytokine Growth Factor Rev. 2001;12:305–312. [PubMed]
19. Jacinto E, Hall MN. Nat Rev Mol Cell Biol. 2003;4:117–126. [PubMed]
20. Long XM, Spycher C, Han ZS, Rose AM, Muller F, Avruch J. Curr Biol. 2002;12:1448–1461. [PubMed]
21. Melendez A, Talloczy Z, Seaman M, Eskelinen EL, Hall DH, Levine B. Science. 2003;301:1387–1391. [PubMed]
22. Zweytick D, Athenstaedt K, Daum G. Biochim Biophys Acta. 2000;1469:101–120. [PubMed]
23. Martin S, Parton RG. Nat Rev Mol Cell Biol. 2006;7:373–378. [PubMed]
24. Estevez M, Attisano L, Wrana JL, Albert PS, Massague J, Riddle DL. Nature. 1993;365:644–649. [PubMed]
25. Avery L. Genetics. 1993;133:897–917. [PMC free article] [PubMed]
26. Hou YW, Bardo AM, Martinez C, Higgins DA. J Phys Chem B. 2000;104:212–219.
27. Tauchi-Sato K, Ozeki S, Houjou T, Taguchi R, Fujimoto T. J Biol Chem. 2002;277:44507–44512. [PubMed]
28. Eesley GL. Coherent Raman Spectroscopy. Oxford: Pergamon; 1981.
29. Snyder RG, Hsu SL, Krimm S. Spectrochim Acta A. 1978;34:395–406.
30. Gaber BP, Peticolas WL. Biochim Biophys Acta. 1977;465:260–274. [PubMed]
31. Wurpel GWH, Schins JM, Müller M. J Phys Chem B. 2004;108:3400–3403.
32. van Gilst MR, Hadjivassiliou H, Yamamoto KR. Proc Natl Acad Sci USA. 2005;102:13496–13501. [PMC free article] [PubMed]
33. Wadsworth WG, Riddle DL. Dev Biol. 1989;132:167–173. [PubMed]
34. Sulston J, Hodgkin J. In: The Nematode Caenorhabditis elegans. Wood WB, editor. New York: Cold Spring Harbor Lab Press; 1988. pp. 587–606.
35. Greenspan P, Mayer EP, Fowler SD. J Cell Biol. 1985;100:965–973. [PMC free article] [PubMed]
36. Abramoff MD, Magelhaes PJ, Ram SJ. Biophot Int. 2004;11:36–42.
37. Abramoff MD, Viergever MA. IEEE Trans Med Imag. 2002;21:296–304. [PubMed]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...