Opto-Lipidomics of Tissues

Abstract

Lipid metabolism and signaling play pivotal functions in biology and disease development. Despite this, currently available optical techniques are limited in their ability to directly visualize the lipidome in tissues. In this study, opto-lipidomics, a new approach to optical molecular tissue imaging is introduced. The capability of vibrational Raman spectroscopy is expanded to identify individual lipids in complex tissue matrices through correlation with desorption electrospray ionization (DESI) - mass spectrometry (MS) imaging in an integrated instrument. A computational pipeline of inter-modality analysis is established to infer lipidomic information from optical vibrational spectra. Opto-lipidomic imaging of transient cerebral ischemia-reperfusion injury in a murine model of ischemic stroke demonstrates the visualization and identification of lipids in disease with high molecular specificity using Raman scattered light. Furthermore, opto-lipidomics in a handheld fiber-optic Raman probe is deployed and demonstrates real-time classification of bulk brain tissues based on specific lipid abundances. Opto-lipidomics opens a host of new opportunities to study lipid biomarkers for diagnostics, prognostics, and novel therapeutic targets.

Keywords: Raman spectroscopy; lipidomics; mass spectrometry.

Figure 1

Integrated Raman and DESI‐MS imaging system. a) Schematics of the integrated Raman and DESI‐MS imaging system. The electro‐sprayer pneumatically focuses electrically charged MeOH solvent on the tissue sample using pressurized nitrogen. Desorbed molecules are then passed through ambient air into the mass spectrometer inlet. For Raman spectroscopy, a 785 nm laser light is delivered through an objective lens and focused onto the sample at a 45‐degree angle. Raman scattered light is passed back through the objective and separated from the laser path by a dichroic mirror and long pass filter before it is fiber‐coupled into a near‐infrared (NIR) Raman spectrometer. b) Photograph showing the integrated Raman/DESI‐MS imaging system using a commercial Raman probe (InPhotonics). c) Integrated Raman/DESI‐MS imaging workflow for tissue analysis including cryosectioning, correlative imaging, and analysis. d) Computational framework for analysis: Data processing and correlation analysis pipeline for Raman and DESI‐MS data. The Raman and DESI‐MS data are preprocessed separately before they are co‐registered. Heterospectral analysis is performed to ensure accurate co‐registration and can be used to confirm their correlation. Lastly, a regression model can be constructed to predict relative m/z abundances from the Raman spectra.

Figure 2

Co‐registered Raman/DESI‐MS imaging of healthy mouse brain tissue. Co‐registration of the Raman and DESI‐MS hyperspectral images both sampled at 50 µm resolution. a) DESI‐MS image of PI 38:4, ST 24:1, and PS 40:6. b) Mean DESI‐MS spectrum of the entire brain tissue. c) Raman images generated from peaks at 1060, 1445, and 1650 cm⁻¹. White triangles in DESI‐MS and Raman images indicate markers of similar anatomical location. d) Mean Raman spectrum ± 1 standard deviation (SD) of all tissue spectra from the brain.

Figure 3

Opto‐lipidomic imaging of healthy mouse brain tissue. a) 2D heterospectral correlation between Raman and DESI‐MS for healthy mouse brain tissue. b) Images of a coronal mouse brain section showing target DESI‐MS and Raman predicted (opto‐lipidomics) images: ST 24:1, PS 40:6, and PI 38:4. c) Representative mean spectra of region of interest (ROI) associated with the three lipids using DESI‐MSI and Raman predicted (opto‐lipidomics). Also shown are the residual spectra. Image quality metrics comparing Raman predicted versus DESI‐MS imaging: d) Structural similarity index (SSIM), e) Mean squared error (MSE), and f) peak signal to noise ratio (PSNR), for PS 40:6, PI 38:4, and ST 24:1 (n = 3) (error bars: mean ± standard deviation (SD)).

Figure 4

Opto‐lipidomics in a mouse model of transient cerebral ischemia‐reperfusion injury. a) Images of a coronal mouse brain section post ischemia‐reperfusion showing DESI‐MSI target and Raman predicted (opto‐lipidomic) images associated with the brain (FA 18:0, FA 22:6, Cer 36:1; O2, PS 40:6, and ST 24:1. b) Merged Raman predicted (opto‐lipidomic) distribution of FA 22:6, PS 40:6, and ST 24:1. c) Representative mean predicted MS spectra for the stroked and contralateral control side. Also shown is the residual. d) DESI‐MS target peak intensity values in stroked and control side ROI for FA 18:0, FA 22:6, Cer 36:1; O2, PS 40:6, and ST 24:1. e) Raman predicted (opto‐lipidomic) peak intensity values in stroked and normal side ROI for FA 18:0, FA 22:6, Cer 36:1;O2, PS 40:6, and ST 24:1. f) DESI‐MS target and Raman predicted peak residual values between stroked and normal side ROI for FA 18:0, FA 22:6, Cer 36:1;O2, PS 40:6, and ST 24:1. (Data denote mean ± standard deviation (SD), ^* = Significant difference between stroked and control side, p < 0.05).

Figure 5

Handheld real‐time opto‐lipidomics of bulk tissues. a) Schematic of coronal cut mouse brain with measurement position represented in blue (grey matter) and red (white matter). b) Image of handheld Raman probe and coronal cut mouse brain. c) Representative mean Raman spectra for the grey matter, white matter as well as the residual difference spectrum. d) Representative mean predicted DESI‐MS spectra for white matter and grey matter. Also shown is the residual. e) Scatter plot depicting the predicted content of PS 36:1, PS 40:6, PI 38:4, against ST 24:1, showing separation of white and grey matter measurements (n = 20).