To develop a strategy for quantitative chemometric unmixing of Raman spectra from organoids, we designed a 3D tissue phantom calibration technology which enables direct simultaneous measurement of absolute local concentrations of the most abundant biomolecular components and sequestered xenobiotics. The poor solubility of lipids in aqueous media was resolved by employing formulated saturated (DPPC) and monounsaturated (POPC) synthetic high-density lipoprotein nanodiscs (sHDLs) of ~10 nm particle diameter. This allowed dissolution of lipids up to 80 mg/mL and facilitated miscibility with other biomolecules at varying concentrations in PBS. The calibration was repeated across concentration ranges for protein, DPPC, POPC, DNA, and glycogen. The complete unit-scaled reference spectral model accurately identified and deconvoluted single-component tissue phantom data with minimal errors using linear combination modelling approaches. Bayesian model fitting enabled accuracy and precision confidence assessments across measured concentration ranges and specificity in complex multi-component mixtures for model validation. The fundamental linear theory presented herein establishes a robust framework for quantitative analysis of 3D biospecimens.
Primary human hepatocyte spheroids (3D PHH) and iPSC-derived hepatocyte-like cell containing organoids (3D iHLC) were generated using modified differentiation protocols to ensure functional maturation. We employed the Raman methodology for high-content quantitative comparison of 3D PHH and 3D iHLC specimens to determine quantitative and qualitative differences between both. UltraRamanomics revealed similar patterns of glycogen accumulation between 3D iHLC and 3D PHH, an important hallmark of hepatocytes functionality and essential for the maintenance of glucose homeostasis. No significant differences were observed in total lipids content. UltraRamanomics revealed lower protein concentration, a marker for functional activity, per iHLC organoids as compared to the benchmark PHH spheroids. Lower protein concentrations were subsequently confirmed by UV-spectroscopy and proteomics.
We also applied the platform for quantitative chemometric phenotyping of 3D PHHs and 3D iHLCs upon exposure to a panel of drugs with reported impact on hepatocytes. 3D PHHs and 3D iHLCs were exposed to 10 µM amiodarone, nilotinib, fluticasone-propionate, ketoconazole, neratinib, and methadone for 48 h prior to analysis. UltraRamanomics elucidated drug-specific compositional changes induced in 3D PHH and 3D iHLC. We observed a drug-specific modulation of lipid content in all treatment groups, with marked increases in the amiodarone- and nilotinib-treated groups. Elevated levels of glycogen, as well as changes in cyt c levels and its co-localization with DNA, proteins, and lipids were seen in drug-treated groups. Among all tested drugs, amiodarone treatment induced the most significant changes in glycogen (increase) and cyt c concentrations (decrease). To the best of our knowledge, amiodarone-induced glycogenosis has not yet been reported.
Beyond biomolecular phenotyping upon drug exposure, UltraRamanomics enables direct measurement of contextual drug and metabolite accumulation within cells. We report the first spectroscopic evidence of xenobiotic deposits of amiodarone and fluticasone detectable by Raman within the 3D biospecimens along with evidence of nilotinib and neratinib accumulation, while no deposits were detected in the ketoconazole or methadone treatment groups. 3D images were generated for to visualize the distribution of drug/metabolite deposits within the spheroids and organoids. To the best of our knowledge, this has not been reported previously.
