Methodological advancements in organ-specific ectopic lipid quantitative characterization: Effects of high fat diet on muscle and liver intracellular lipids

Objective Ectopic lipid accumulation is a hallmark of metabolic diseases, linking obesity to non-alcoholic fatty liver disease, insulin resistance and diabetes. The use of zebrafish as a model of obesity and diabetes is raising due to the conserved properties of fat metabolism between humans and zebrafish, the homologous genes regulating lipid uptake and transport, the implementation of the ‘3R’s principle and their cost-effectiveness. To date, a method allowing the conservation of lipid droplets (LDs) and organs in zebrafish larvae to image ectopic lipids is not available. Our objectives were to develop a novel methodology to quantitatively evaluate organ-specific LDs, in skeletal muscle and liver, in response to a nutritional perturbation. Methods We developed a novel embedding and cryosectioning protocol allowing the conservation of LDs and organs in zebrafish larvae. To establish the quantitative measures, we used a three-arm parallel nutritional intervention design. Zebrafish larvae were fed a control diet containing 14% of nutritional fat or two high fat diets (HFDs) containing 25 and 36% of dietary fats. In muscle and liver, LDs were characterized using immunofluorescence confocal microscopy. In liver, intrahepatocellular lipids were discriminated from intrasinusoid lipids. To complete liver characteristics, fibrosis was identified with Masson’s Trichrome staining. Finally, to confirm the conservation and effect of HFD, molecular players of fat metabolism were evaluated by RT-qPCR. Results The cryosections obtained after setting up the embedding and cryopreservation method were of high quality, preserving tissue morphology and allowing the visualization of ectopic lipids. Both HFDs were obesogenic, without modifying larvae survival or development. Neutral lipid content increased with time and augmented dietary fat. Intramuscular LD volume density increased and was explained by an increase in LDs size but not in numbers. Intrahepatocellular LD volume density increased and was explained by an increased number of LDs, not by their increased size. Sinusoid area and lipid content were both increased. Hepatic fibrosis appeared with both HFDs. We observed alterations in the expression of genes associated with LD coating proteins, LD dynamics, lipogenesis, lipolysis and fatty acid oxidation. Conclusions In this study, we propose a reproducible and fast method to image zebrafish larvae without losing LD quality and organ morphology. We demonstrate the impact of HFD on LD characteristics in liver and skeletal muscle accompanied by alterations of key players of fat metabolism. Our observations confirm the evolutionarily conserved mechanisms in lipid metabolism and reveal organ specific adaptations. The methodological advancements proposed in this work open the doors to study organelle adaptations in obesity and diabetes related research such as lipotoxicity, organelle contacts and specific lipid depositions.


INTRODUCTION
Excessive fat intake leads to increased lipid storage and adipose tissue overload, consequently causing ectopic lipid depositions and obesity [1e3]. Ectopic lipids are stored in dynamic cytosolic lipid droplets (LDs). In skeletal muscle, ectopic lipid stores are called intramyocellular lipids (IMCL) [2]. IMCL increase along with body fat mass during the development of obesity [4,5]. In the absence of physical activity, IMCL are involved in lipotoxic effects which are associated with altered mitochondrial fatty acid oxidation, apoptosis, inflammation and impaired muscle fiber contractility [6,7]. In the liver, ectopic lipid depositions are called intrahepatocellular lipids (IHCL) [2]. Under physiological conditions, only small amounts of IHCL are stored while most lipids are distributed in the blood stream according to metabolic demand. Hepatic LD accumulation is the hallmark of nonalcoholic fatty liver disease (NAFLD) [8]. In order to investigate human metabolic diseases [9], ectopic lipid depositions in liver and skeletal muscle have been studied in humans and animal models, as well as in vitro in organ specific cells [9e16]. The use of zebrafish as a model to investigate metabolic diseases, particularly obesity and diabetes, is attractive due to the functional conservation of lipid and glucose metabolism as well as the presence of the main organs playing key roles in metabolism (pancreas, muscle, liver, etc) [9]. Compared to other models, zebrafish have similar white adipose tissue depots, such as visceral and subcutaneous adipose tissue [17]. Because zebrafish are ectotherms, they lack brown adipose tissue [18]. While previous studies visualized hepatic steatosis [6,19], to our knowledge, the development of subcellular lipid depositions in zebrafish larvae have not been reported. Our aim was to develop the quantitative tools necessary to assess the effect of nutritional perturbations on IMCL and IHCL in zebrafish larvae. Organ specific characterization of LDs requires adjusted protocols for embedding, sectioning and visualization [20] depending on samples and organisms. While cryosection preparation has been optimized for portions of adult zebrafish [21], larvae require full body embedding for cryosection due to their small size and heterogeneous texture. Paraffin and resin embedding techniques described previously are either unsuitable to study lipids due to alcohol dehydration [22] or expensive and time consuming [23]. Here, we custom-made a cryosection and imaging procedure to visualize and characterize tissue specific LDs in zebrafish larvae.

Zebrafish husbandry
Zebrafish (Danio rerio, Oregon AB) were housed at the Zebrafish facility of the Faculty of Biology and Medicine, maintained at 28.5 C on a 14 h light and 10 h dark cycle. Embryos were staged according to days post fertilization (dpf) as previously described [24]. Animal experimentation and husbandry were approved by the Service de la consommation et des affaires vétérinaires (SCAV) of the Canton of Vaud.

Custom mold
To control larvae position in the histology blocks, we designed 3 custom molds (modified from Sabaliauskas et al. [25]) to create imprints adapted for each larval age ( Figure 1A). The molds were 3D printed by a professional service (FabArt3D, https://www.fabart3d.ch/, Montreux, Switzerland) with digital light processing resin (Prima-Creator, Malmö, Sweden). Details of our custom molds are available in the supplementary data.

Embedding and cryosections
Embedding medium was prepared by boiling 10 g Type B bovine skin gelatin (Sigma, Kawasaki, Japan), 5 g sucrose (Sigma) and 1.5 g agarose (Roche, Basel, Switzerland) in 100 ml of phosphate-buffered saline (PBS, Dr. G. Bichsel AG, Interlaken, Switzerland) (modified from Nelson et al. [21]). The medium was poured warm into petri dishes with the molds placed on top and removed once the medium cooled down enough. Nine 21 dpf larvae per condition were added one by one into the imprints oriented with the head on the deeper side ( Figure 1B). To mark the region of interest, red waxed dental floss (Johnson & Johnson, New Brunswick, New Jersey) ( Figure 1A and B) was inserted into the medium beside each larva, which were then covered by the liquid medium once it cooled below 50 C. Once solid, each block was shaped distinctly and put in 0.9 M sucrose overnight at 4 C ( Figure 1C). The blocks were placed in isopentane (SigmaeAldrich, Saint-Louis, USA) cooled with liquid nitrogen for 2e3 min, and then placed in liquid nitrogen. Each block was mounted on a small piece of cork ( Figure 1D) with optimal cutting temperature compound (OCT) (Thermo Fisher Scientific, Waltham, USA). Serial sections of 10 and 30 mm were performed in a cryostat (CM3050S, Leica Microsystems, Wetzlar, Germany) at À18 C and placed on precleaned glass slides (Thermo Fisher Scientific). Sections were performed for the entire volume where the red string was visible in the blocks. Timings for each procedure are detailed in the supplemental Table 1.

Abbreviations acsl3
Acyl-CoA synthetase 3 atgl Adipose triglyceride lipase BSA Bovine serum albumin cidec Cell death inducing DFFA like effector C cpt1 Carnitine O-palmitoyltransferase 1 DAG Diacylglycerol dgat2 Diglyceride acyltransferase 2 Dpf Days post fertilization ef1a Alpha subunit of the elongation factor-1 complex ER Endoplasmic reticulum fabp11a Fatty acid binding protein 11a faf2 Fas associated factor family member 2 gpat4 Glycerol - To allow the implementation and confirmation of the usefulness of our methodological advancement, we used a three parallel arms interventional design (Figure 2A). One cohort of 750 larvae was divided at 5 dpf in three groups of 250 larvae, each group housed in one tank. One group was fed a control diet (14% of fat), one group was fed a HFD with 25% of fat and one group was fed with a HFD with 36% of fat. To ensure reproducibility, three successive cohorts were used as three independent repeats.
All diets contained Zebrafeed <100 mm (Sparos, Olhão, Portugal), a nutritionally balanced microdiet for zebrafish, either alone or complemented with egg yolk (Eigelbpulver, Bouhwouis Enthoven, Bodenhaltung, Austria) to reach the desired level of nutritional fat (Table 1A). All groups were fed 4 meals per day with the same nutritional mass, which was adapted to larval age (Table 1A). Calories and protein content of the three different diets are shown in Red concentration of 0.5 mg/ml [27]. Larvae were then placed in the dark for 30 min, washed twice with housing water, sacrificed and fixed    Fig. 2). The volume of each LD was then acquired.
2.10. Liver LD and sinusoid lipids quantification Liver images from nine 21 dpf larvae per condition were deconvolved as described above. To quantify LDs in the parenchyma, LDs were reconstructed using Imaris 9.8 in twelve equal volumes (15 Â 15 Â 10.5 mm 3 ) blindly selected and processed as described above. The volume of each LD was acquired.
Sinusoids were defined using the combination of a thick WGA staining to identify the endothelial glycocalyx layer [28,29] and confirmed with actin staining [30]. The Z-stack was used to exclude hepatocytes surrounded by immune cells or stellate cells, ascertaining the absence of subcellular compartments. One 60 Â 60 mm 2 region of interest (ROI) was selected in each blinded liver image. Using Fiji, sinusoid surface was determined and measured. A background of 20 was removed from the Bodipy channel to prevent possible deconvolution artifacts. Machine-learning based segmentation was performed using Ilastik [31]. Bodipy positive area were quantified inside the sinusoids on the obtained binary images (322 Â 322 pixels) using Fiji's analyze particle function. Given the high density of particles in the intrasinusoid space, we chose to measure surface instead of particle size to take into account that chylomicrons were possibly on top of each other or coalescent.

Liver fibrosis
Slides containing 10 mm sections from nine 21 dpf larvae per condition were dried overnight at À20 C, and stained with modified Masson's Trichrome without hematoxylin to prevent excessive darkness. Images were acquired with 10X to 40X objectives using an ECLIPSE 90i microscope (Nikon) with a Nikon digital sight camera. For quantification, 40X images were blinded and liver tissue was selected. Macros in Fiji were used to measure liver and fibrotic areas. The percentage of fibrotic tissue was computed as the ratio between the fibrotic area and the parenchyma.
2.12. RNA extraction 25 snap frozen larvae per time point per group per cohort were pooled for RNA extraction adapted from guidelines provided by SigmaeAldrich for TRI Reagent RNA extraction. In brief, larvae were lysed using a mechanical pestle (Kimble Chase, Vineland, USA) in TriReagent (Sigma). RNA was isolated with chloroform (SigmaeAldrich), precipitated with isopropanol and resuspended. Final RNA concentration was measured using a ND-1000 Spectrophotometer (Thermo Fisher Scientific).

RT-qPCR
One mg RNA per pool was retrotranscribed using GoScriptÔ Reaction Buffer (Random Primers) and GoScriptÔ Enzyme Mix (Promega, Madison, USA). Retrotranscription products were mixed with Power SYBR Green PCR Master Mix (Thermo Fisher Scientific) and with 600 nM of forward and reverse primers (Supplemental Table 2). Samples were analyzed with ViiA 7 Real-Time PCR System (Thermo Fisher Scientific) or QuantStudio 5 (Thermo Fisher Scientific). Relative expression of mRNA was estimated using the 2 ÀDDCT method using ef1a as reference.
2.14. Statistical analyses Data are presented as mean and standard error of the mean (SEM). Normality was checked with the ShapiroeWilk test and equality of variance was assessed with the Bartlett's test and the Browne Forsythe test. If assumptions were met, data were analyzed with one-way ANOVA and multiple comparisons were performed using Tukey HSD. If assumptions were not met, data were analyzed with the KruskalleWallis test and multiple comparisons were performed using the Dunn's test. Data of the length, survival rate and ORO absorbance were analyzed with a two-way ANOVA with a factor of time (3 levels: 15, 18 and 21 dpf) and of diet (3 levels). All statistical analyses were performed using Prism (GraphPad, version 9.0.0, San Diego, USA) with the significance set at p < 0.05.  Figure 1E and F, the cryosections obtained after setting up the embedding and cryopreservation method (Figure 1AeD) were of high quality, preserving tissue morphology and allowing the visualization of ectopic lipids. To investigate IMCL, we used skeletal muscle transversal sections at the topological marker of the junction between mid-intestine and intestinal bulb as indicated by the pink dashed line in Figure 1F. To investigate IHCL, we used the end of the esophagus as topological marker for liver transversal sections as indicated by the blue dashed line in Figure 1F.

Organ landmarks As shown in
3.2. High fat diet increased neutral lipid content and adipocyte number in zebrafish larvae As previously described, larvae survival decreased over time ( Figure 2B). Our overall mean survival rate at 21 dpf was of 69.46 AE 0.74%, in the range of reported survival rates [32]. Larvae length increased throughout development ( Figure 2C) with a significant simple main effect of time (F(2, 317) ¼ 56.91, p < 0.0001). Survival and length were not significantly different across diet groups ( Figure 2B and C). Larvae neutral lipid content increased with time and dietary fat ( Figure 2D and E). The two-way ANOVA revealed a statistically significant interaction (F(4, 310) ¼ 2.971, p ¼ 0.0197). Post hoc comparisons of simple effects of diet (F(2, 310) ¼ 69.65, p < 0.0001) are represented in Figure 2E. To determine the optimal time point to evaluate the effect of HFD on ectopic lipids, we compared 18 and 21 dpf neutral lipid content to 15 dpf, which is the known age when LDs begin to expand beyond the visceral cavity [33,34]. We chose 21 dpf as our time point of interest given that neutral lipid content was significantly higher with both HFDs compared to 15 dpf (p < 0.0001). This was not the case when comparing 18 dpf to 15 dpf. At 21 dpf, the pattern of adipocyte distribution confirmed the effect of the two HFDs ( Figure 2F and G). 3.5. Lipid droplet associated genes were differentially affected by high fat diet Whole larvae transcriptional changes were studied for targeted genes associated with LD coating proteins, LD dynamics, adipogenesis, lipogenesis, lipolysis and fatty acid oxidation (Figure 5AeG). At 21 dpf, plin2 and plin3 were increased by HFD ( Figure 5B). Cidec, cpt1a and cpt1b, fasn and fabp11a also responded to HFD (Figure 5CeG).

DISCUSSION
Excess fat and caloric intake are cornerstone of the lipotoxicity theory linking obesity to insulin resistance through the accumulation of ectopic lipids, particularly in skeletal muscle and liver. The use of zebrafish as a model to study obesity, lipid metabolism and diabetes is rising [9,35,36]. Our main aim was to develop a detailed protocol to characterize quantitatively intracytosolic ectopic lipids, in liver and skeletal muscle of zebrafish larvae after a nutritional perturbation.
Here, we propose five methodological novelties. First, in order to image whole larvae, we optimized cryosection and cryopreservation methods allowing the conservation of lipid droplets. These protocols can also be of use for many other outcomes and models with small fragile organs. The 3D printed molds allow proper positioning of animals, are adapted for age and length, and provide reproducible outputs. The second innovative aspect is that we present multiple ectopic lipid readouts that are all quantitative. Thirdly, we imaged ectopic lipids in multiple organs in one single study and with the exact same treatment. Fourthly, we differentiate liver parenchyma ectopic lipids from lipids in the sinusoids. This, to our knowledge, has not been done previously in zebrafish, mice or human although lipids in sinusoids are not ectopic lipids but circulating lipids. Finally, we describe a complete step by step protocol investigating the effects of high fat diet on zebrafish larvae, which could also be applied to many other treatments such as high glucose/high fructose diets or drug treatments. To investigate specific ectopic lipids of our two organs of interest, we developed a zebrafish larvae sectioning method. Paraffin embedding, the current gold standard for zebrafish larvae sectioning, requires ethanol dehydration that causes the loss of lipids from tissues and induces morphological changes [22]. In comparison, our method allows complete tissue conservation of small larvae and does not require dehydration, preventing subsequent deformation of the tissues and loss of lipids. Gelatin embedding provides samples useable for both traditional histology techniques, like Masson's Trichrome, or immunohistochemical staining. We believe that this methodological development will serve multiple other inquiries, such as studying other organs or organelles of interest. While most of overfeeding and HFD studies in zebrafish explore adiposity or global hepatic lipids [10,37], our method allowed to quantify ectopic lipids distribution and volume, rather than fluorescence intensity or area as described previously [10,19,38e41]. We tested two HFDs (25 and 36% of fat) to validate their suitability for quantitative ectopic lipids characterization. The increase of neutral lipid content at all time points (15, 18 and 21 dpf), measured by single larvae extraction of Oil Red O to reduce the bias caused by measuring fish color, confirmed the obesogenic effect of both HFDs without impairing larvae development and survival. Similarly to other studies supplementing zebrafish larvae diet with different proportions of egg yolk [42,43], the number of larvae showing adipocytes increased with nutritional fat. In skeletal muscle, 15 days of HFD feeding caused an increase in skeletal muscle LD size but not LD count. The increased LD size following HFD is in accordance with previous studies in rodents [40] and human volunteers [44e46] pointing to the negative association of  larger LDs, but not IMCL per se, to insulin sensitivity. From a mechanistic point of view, the observed LD expansion could be due to an increased TAG synthesis, to ripening or to LD coalescence, with the creation of expanding pores between two LDs [47]. Further, LD expansion also affects LD surface tension, impacting LD monolayer curvature and its components [48], including LD coating proteins and access to the lipolysis/lipophagy machineries. In liver, HFD increased the number of LDs in the hepatic parenchyma, caused larger sinusoid areas with higher lipid content in their lumen and provoked hepatic fibrosis. Importantly, we separated IHCL, i.e. ectopic lipids in the hepatic parenchyma, from lipid micelles in the sinusoids. In the hepatic parenchyma, the average number of LDs and their volume density increased as described in other models [49], but not LD size. Mechanistically, the potential source of fats contributing to fatty liver could come from the esterification of dietary fatty acids provided in excess through the uptake of intestinally derived chylomicron remnants, from lipogenesis derived fats stored in adipose tissue and/or from liver de novo lipogenesis [50]. In the sinusoids, increased lipids reflect the increased presence of lipoproteins due to their role as lipid transporters and their structure similar to those of LDs [8]. In hepatic injury due to excess lipid exposure, hepatic stellate cells surrounding the sinusoids produce collagen and extracellular matrix. In our larvae, we observed an enlargement of sinusoids accompanied by fibrosis in the HFD groups, confirming the development of hepatic injury. Given the difficulty of isolating organ-specific mRNA from zebrafish larvae due to their small size and fast degradation, we opted to measure whole body gene expression. This is an important limitation as these results should not be extrapolated to organ-specific gene expression in other models. Similarly to another study executed in zebrafish larvae [43], we did not observe changes in adipogenesis markers with HFD. The fact that zebrafish body composition is mostly composed of fat free mass [51] may cover adipose tissue specific pathways modulations. The upregulation of fabp11a, zebrafish ortholog of human FABP4 [10], implies that lipid transport pathways are affected by HFD. The increased expression of cidec, which has been demonstrated to form a pore-like structure between LDs [52] also speaks in favor of inter-LD lipid transfer. Further, Cidec has been shown to control HFD induced fatty liver [52] and its increased expression supports the development of steatosis in larvae. Supporting the hypothesis of chylomicron remnants in liver sinusoids, we observed an upregulation of lpl. Plin2 and 3 expressions increased with HFD, reflecting the higher volume of LDs and the modulation of LD access to lipases [19,53]. We witnessed altered fatty acid oxidation, characterized by higher cpt1a and cpt1b mRNA expression. The downregulation of fasn suggests the suppression of de novo lipogenesis in response to excess dietary fats. In our results, we observe differences between the two HFDs in causing ectopic lipids depositions. The 25% fat diet induced significant increases of average LD volume and LD volume density in muscle. These modifications were not observed with the 36% fat diet. In liver, both HFDs increased significantly LD count and LD volume density in hepatocytes. While only the 25% fat diet increased intrasinusoid lipid area density, both diets induce similar fibrosis. These organ-specific observations are in contrast with whole larvae outcomes at the same time point. Indeed, overall ORO absorbance at 21 dpf increased similarly with both HFDs. Significant changes in whole larvae mRNA levels were mostly observed after the 36% HFD or similarly with the two HFDs. Only cpt1a increased in an incremental pattern between the 25% and 36% nutritional fat. These observations support the hypothesis that fat metabolism may be enhanced in the 36% fat fed animals, thus using more intramyocellular lipids at a given time point.
This may be in part secondary to the relative difference in dietary proteins. To note that all of our diets had a protein content above the minimal juveniles requirements estimated by Fernandes et al. [54] although complete nutritional requirements of both adult and larval zebrafish are not yet determined [32,55]. With a thorough organ-specific LD characterization, our study demonstrates the impact of HFD on ectopic lipid depositions in liver and skeletal muscle, as well as the changes of key players in fat metabolism in zebrafish larvae. We propose a reproducible and fast method for imaging zebrafish larvae without sacrificing LD quality or organ morphology. Our findings confirm the evolutionarily conserved mechanisms in lipid metabolism, reveal organ specific adaptations and open the door to studying organelle modifications in obesity and diabetes related research such as lipotoxicity, organelle contacts and specific lipid depositions in this cost-effective animal model.

FUNDING
This project was supported by the Swiss National Science Foundation grant number 310030_188789 (to FA).

CONFLICT OF INTEREST
None declared.

DATA AVAILABILITY
Data will be made available on request.