Abstract

Confocal Raman spectral imaging (RSI) enables high-content, label-free visualization of a wide range of molecules in biological specimens without sample preparation. However, reliable quantification of the deconvoluted spectra is needed. Here we develop an integrated bioanalytical methodology, qRamanomics, to qualify RSI as a tissue phantom calibrated tool for quantitative spatial chemotype of major classes of biomolecules. Next, we apply qRamanomics to fixed 3D liver organoids generated from stem-cell-derived or primary hepatocytes to assess specimen variation and maturity. We then demonstrate the utility of qRamanomics for identifying biomolecular response signatures from a panel of liver-altering drugs, probing drug-induced compositional changes in 3D organoids followed by in situ monitoring of drug metabolism and accumulation. Quantitative chemometric phenotyping constitutes an important step in developing quantitative label-free interrogation of 3D biological specimens.

Introduction

There is a significant need for reliable human organ representations (termed organoids) that provide bio-relevant model systems with foreseeable utility in disease modelling, drug discovery, and personalized drug testing. Induced pluripotent stem cell (iPSC) technology enables in vitro development of human organoids that show features of the organs they represent. However, organoids typically lack the structure and functional maturity of their human counterparts and show significant intra- and inter-batch variations. Hence, there is a need for advancing multi-factorial organoid characterization, including their response to therapeutic interventions, by applying high-content and high-resolution imaging tools.

Conventional fluorescence-based confocal imaging and in situ hybridization are the methods of choice for visualizing localization and dynamics of biomolecules at cellular and subcellular levels, based on specific labels. However, while researchers have a broad selection of probes to mark specific proteins and nucleic acid sequences, other classes of biomolecules including carbohydrates, lipids, and metabolites are in general more difficult to visualize. Moreover, because of a so-called colour barrier, only a limited number of targets can be investigated simultaneously in a specimen by most label-dependent techniques.2 In addition, in situ quantification of biomolecules is challenging and can only be done indirectly and in relative units (e.g., mean fluorescence intensity)

Methods

This study did not generate new unique reagents. A unique code was generated, as detailed below.
Cryopreserved primary human hepatocytes (Gibco, catalogue no. HMCPSQ, lot HU8339-A, female (referred to in the study as PHH_1), Lonza, catalogue no. HUCPG, lot HUM180201A, male (referred to in the study as PHH_2) and Gibco, catalogue no. HMCPMS, lot HU8287, female (referred to in the study as PHH_3)) were thawed in the Hepatocytes thaw media (Gibco, catalogue no. CM7500) according to the manufacturer’s protocol. Moreover, one sample of PHH (donor 73 years old, male) was obtained from the Department of Clinical Science, Intervention and Technology (CLINTEC), Division of Transplantation Surgery, Karolinska Institutet (Stockholm, Sweden). The regional committee for medical and health research ethics in Norway approved the use of human material (REK 50786). Uniform PHH spheroids were created by aggregation in ultra-low attachment micro-wells (Corning, catalogue no. 4440) or in house-made agarose microwells - a format in which PHH showed stable functionality over at least 7 days as described before22 (Figure S2G). Briefly, cells were plated into microwells at the concentration of 1000 viable cells per microwell and were centrifuged at 100 g for 2 min. For the first 3 days PHH were cultured in the Williams E medium (Thermo Fisher Scientific, catalogue no. A1217601) supplemented with 7 % FBS (Thermo Fisher Scientific, catalogue no. 41400045), 2 mM L-glutamine (Thermo Fisher Scientific, catalogue no. 35050038), 10 μg/ml insulin, 5.5 μg/ml transferrin, 6.7 ng/ml sodium selenite (Thermo Fisher Scientific, catalogue no. 41400045) and 0.1 μM dexamethasone (Sigma Aldrich, catalogue no. D4902). From day 4 onwards, the FBS concentration gradually decreased to 1 % (v/v). Spheroids were maintained in serum-free media from day 7 for up to 2 weeks.

Discussion

The role of localized concentrations and biomolecular density at subcellular levels is gaining attention as a relevant phenotypic parameter for cell-type-specific functions49,50 and can be indicative, for example, of the maturation state of cells and their response to xenobiotic exposures 49. In this study, we developed a robust data-processing and calibration pipeline, qRamanomics, that allows label-free, multi-factorial, chemometric phenotyping of 3D biospecimens using RSI. Unlike conventional fluorescence imaging tools that are mostly optimized for proteins, the qRamanomics framework described here enables direct quantitative in situ classification of a broad range of biomolecules, including different classes of lipids, Cyt c, proteins, nucleic acids, glycogen, vitamins, and selected xenobiotics. The technology, therefore, promises multiple practical applications in organoid research and development.

Results

Development of a platform for quantitative chemometric phenotyping of 3D biospecimens: Ramanomics To develop a platform for quantitative chemometric phenotyping of hepatic organoids by confocal RSI (Figure 1A), we designed a 3D tissue phantom calibration technology. This calibration methodology enables direct simultaneous measurement of the absolute local concentrations of the most abundant biomolecular components and sequestered xenobiotics in organoids.