Toward Quantitative in vivo Label-Free Tracking of Lipid Distribution in a Zebrafish Cancer Model

Abstract

Cancer cells often adapt their lipid metabolism to accommodate the increased fatty acid demand for membrane biogenesis and energy production. Upregulation of fatty acid uptake from the environment of cancer cells has also been reported as an alternative mechanism. To investigate the role of lipids in tumor onset and progression and to identify potential diagnostic biomarkers, lipids are ideally imaged directly within the intact tumor tissue in a label-free way. In this study, we investigated lipid accumulation and distribution in living zebrafish larvae developing a tumor by means of coherent anti-Stokes Raman scattering microscopy. Quantitative textural features based on radiomics revealed higher lipid accumulation in oncogene-expressing larvae compared to healthy ones. This high lipid accumulation could reflect an altered lipid metabolism in the hyperproliferating oncogene-expressing cells.

Keywords: cancer model; coherent anti-Stokes Raman scattering; label-free microscopy; lipid metabolism; zebrafish.

Figure 1

Sample preparation for in vivo imaging. (A) Lateral view of coherent anti-Stokes Raman scattering (CARS) microscope objective showing the fish mounted in agarose gel. (B) Top view of fish mount showing the imaging region of interest (ROI) (blue dashed line). (C) Microscopic white light images of RAS- and RAS+ larvae showing the cancer phenotype (red arrows). (D) mCherry expression at location of RasG12V expression in a 96 hpf larvae. (E) Cross-section of the zebrafish larvae showing the ROI in depth (blue rectangle) with: sc, spinal cord; um, dorsal myotomes; lm, lower myotomes; nc, notochord; g, gut.

Figure 2

In vivo hyperspectral coherent anti-Stokes Raman scattering (CARS) images. (A) CARS image of healthy zebrafish larva at 120 hpf at 2,845 cm⁻¹, field of view (FOV) is 80 × 80 μm². (B) High-resolution CARS image of the green dashed square in (A) showing contrast coming from the lipid content of the muscle cells, FOV is 30 × 30 μm². (C) Quadratic power dependency of the CARS signal of the white dashed rectangle area in (A) respect to the pump power intensity. R-squared of the quadratic fit is 0.9889. (D) Spectral information provided by hyperspectral CARS allowing discrimination between the resonant signal of lipids and 1% agarose gel.

Figure 3

In vivo coherent anti-Stokes Raman scattering (CARS) images at 2,845 cm⁻¹ at three different depths for RAS- (left) and RAS+ (right) larvae. nc, notochord; um, dorsal myotome region. All images have a FOV of 100 × 100 μm².

Figure 4

In vivo confocal images of Nile Red stained RAS- (A) and RAS+ (B) 120 hpf larvae imaged with 25x, 0.95 NA water immersion objective. The contrast is based on fluorescence of Nile Red excited with 510 nm laser excitation showing lipid (triglycerides) distribution and content. All images have a field of view (FOV) of 100 × 100 μm².

Figure 5

Longitudinal in vivo coherent anti-Stokes Raman scattering (CARS) images at 2,845 cm⁻¹ for RAS- and RAS+ larvae from 72 to 120 hpf (left and right columns). All images have a field of view (FOV) of 100 × 100 μm². The single image contrast is optimized for better visualization. Center column shows the lipid distribution on the specific region of interest (ROI). Significant differences between the 3 groups with n = 3 (72, 96, and 120 hpf) are shown (^*p < 0.02, ^**p < 0.03, and ^***p < 0.04).