September 15, 2016 (Vol. 36, No. 16)

Novel Two-Step Assay Integrates Toxicity Testing in 3D Liver Models and Assessment of OCR

Mitochondrial impairment is a major mechanism of action (MoA) causing drug-induced liver injury (DILI), as evidenced by drugs that have received black box warnings,1 such as Amiodarone (used to treat ventricular tachycardia or ventricular fibrillation), Perhexiline (prophylactic antianginal agent), and Tamoxifen (used to treat some types of breast cancer).

This is not surprising, as mitochondria play a central role in the energy metabolism of cells, involved to various extents in oxidative phosphorylation, fatty acid beta-oxidation, heme synthesis, and urea generation.2 Mitochondrial toxicants can inhibit respiratory complexes of the electron transport chain (ETC), inhibit or uncouple oxidative phosphorylation, induce mitochondrial oxidative stress, and inhibit DNA replication, transcription or translation.

Mitochondrial injury is also implicated in several diseases including Parkinson’s, Alzheimer’s, cancer, as well as cardiac, cholesterol, and lipid disorders, making the assessment of potential mitochondrial liabilities a standard element of drug development.3

In the case of mitotoxicity in liver cells, the hepatocellular carcinoma cell line HepG2 is commonly used to assess mitochondrial function using glucose and galactose selective media conditions (Glu/Gal assay).4

This assay capitalizes on the preferential use of glycolysis as a primary energy source in high glucose media, versus oxidative phosphorylation in galactose conditioned media, which sensitizes HepG2 cells to mitochondrial impairment by drugs. However, this preference (and hence sensitivity) observed in proliferating HepG2 cells, is not seen in primary human hepatocytes, the preferred source of cells for in vitro DILI testing due to their more robust CYP expression and metabolic capacity. Thus, there is a need for advanced mitotoxicity assays that can be used with primary human hepatocytes.

Label-free continuous monitoring of oxygen consumption rates (OCRs) in response to drugs using the Agilent XFe96 Analyzer enables a direct correlation to mitochondrial activity conducive for use with monolayer culture of cells and tissues from various organs such as liver (including primary human hepatocytes), cardiac, brain, and tumor.5

While the platform provides higher throughput and in-depth information regarding which complex of the ETC is affected, assessment of long-term compound exposure (>72 hours) is usually limited, as cells either overgrow (e.g., tumor cells) or quickly lose metabolic capacity and cell polarization (as with primary hepatocytes grown in 2D).

Combining the use of 3D models composed of primary human cells with OCR assessment on the XFe96 platform would be desirable, as 3D microtissues display more organotypic structure and function. 3D microtissues also maintain a differentiated state and stable metabolic activity in culture for several weeks, allowing evaluation of long-term exposure to drugs.

Here we describe the integration of an advanced 3D in vitro human liver model (3D InSight™ Human Liver Microtissues, InSphero) with the Agilent XFe96 Analyzer to identify potential mitochondrial liabilities following 48-hour exposure to drugs. We present a 13-compound validation set, and introduce a classification method to categorize potential mitochondrial hazards associated with test compounds.

Enhanced Spare Respiratory Capacity of 3D Liver Microtissues

3D liver microtissues enable the use of more physiologically relevant in vitro liver models for mitotoxicity testing. Liver microtissues composed of primary human hepatocytes in co-culture with Kupffer cells are stable for over four weeks in culture, as verified by ATP content (viability), albumin secretion, and CYP3A4 activity.6

3D liver microtissues also retain glycogen storage capacity and form functional bile canaliculi networks. The in vitro longevity of 3D liver microtissues allows mitotoxicity assessment to be divided into two distinct steps: (1) extended culture and flexible drug exposure times (up to 14 days) with repeat dosing, and (2) discrete transfer of the spheroids to the Agilent platform for OCR analysis (Figure 1A).

Initially, the mitochondrial respiration (OCR) and spare respiratory capacity (SRC—measured as the difference between baseline and maximum OCR following treatment with the uncoupler FCCP), were compared in untreated primary human hepatocytes grown in monolayers (2D) or as microtissues (3D).

Significant differences were observed in the maximal respiration rate. 3D human liver microtissues showed a fourfold higher SRC than the corresponding 2D culture, in which an increase in SRC was barely detectable (Figure 1B). The increased SRC in 3D culture is important because it defines the extra capacity that is available in cells to produce energy in response to increased stress or work, and as such is associated with cellular survival.7,8

Primary human hepatocytes exhibit higher reserve capacity in 3D liver microtissues than in monolayer culture, suggesting that 2D hepatocytes might have undergone a metabolic change to a more oxidative phenotype. 

In normal differentiated cells, oxidative phosphorylation is the prime mechanism to generate cellular energy, whereas in proliferating cells (e.g., cancer cells), aerobic glycosylation is the major energy source (i.e., the Warburg effect).9 Cell shape and tension are also known to impact differentiation and proliferation.10 Thus, one might postulate that by losing their polarized, differentiated structure, 2D hepatocytes enter the initial stages of becoming proliferative and change their metabolism accordingly.

Figure 1. The two-step mitotoxicity assay workflow (A) consists of drug exposure (typically 2 to 14 days) followed by OCR analysis. Bioenergetic profiles (B) of 2D and 3D cultured primary human hepatocytes (untreated), displaying approximately fourfold greater spare respiratory capacity in 3D microtissues.

Validation and Classification of Mitochondrial Liabilities

The high level of SRC observed in 3D InSight Human Liver Microtissues makes them an ideal model system for use as a primary readout to assess mitochondrial function, because a drop in cellular energy reserves is the first direct consequence of mitochondrial impairment. Furthermore, assessing dose response of the SRC in combination with tissue viability enables discrimination whether mitochondrial impairment is the primary toxicological mechanism or a secondary effect.

Resulting IC50 values for reducing the SRC (IC50SRC) were compared to ATP-based IC50 values (IC50ATP) reflecting cell viability (CellTiter-Glo® assay, Promega). The ratio of IC50SRC to IC50ATP was used to determine the mitochondrial liability, and classify drugs in three categories (Figure 2): 

1. No hepatotoxicity or mitochondrial liability: no impact on cellular viability and SRC was observed

2. Hepatotoxicity with no mitochondrial liability: decrease in cellular viability, but equivalent or lower impact on SRC

Figure 2. Schematic representing classification of compound mitotoxicity using ATP cell viability assay (CV) and spare respiratory capacity (SRC).

3. Mitochondrial liability with or without hepatotoxicity: obtained IC50SRC values were less than cellular viability

A panel of 13 drugs of different classes were tested using this method to assess their mitochondrial liabilities. Positive controls were chosen based on clinical data and known association with mitochondrial toxicity; negative controls were drugs not previously associated with mitochondrial toxicity.

3D InSight Human Liver Microtissues were exposed to serially diluted drugs for 48 hours, and the IC50SRC (Agilent XFe96) and IC50ATP (cell viability, CellTiter-Glo) were determined.

Results are summarized in Table 1. These data indicate that 3D human liver microtissues detected the tested mitochondrial toxicants correctly, with the exception of Fialuridine, a drug that is known to affect metabolism only after longer drug exposures.11 Additionally, the classification method allowed categorization of drugs into different risk categories. With this data set, the sensitivity for detecting mitochondrial liability was 88% (7 out of 8 positive drugs correctly predicted) with a specificity of 100% (5 out of 5 negative drugs correctly predicted). The method thus enables the investigation of specific disease states and the mechanism behind cellular bioenergetics effects.

Table 1. Summary of test compound mitotoxicity classification results.


Investigating mitochondrial impairment in a metabolically relevant environment requires in vitro liver models that reflect in vivo function, long-term stability and functionality, and compatibility with state-of-the art analytic instruments. The 3D InSight Mitotchondrial Toxicity Assay described herein combines 3D human liver microtissues with the Agilent XFe96 Analyzer to provide a novel platform for assessment of mitochondrial liabilities.

1. Nadanaciva S, Will Y. Curr Pharm Des. 2011; 17(20):2100-12.
2. Dykens JA and Will Y. Drug Discov Today, 2007; 12(17-18): 777-85.
3. Kaplowitz N. Nat Rev Drug Discov. 2005; 4(6):489-99.
4. Wills LP, et al. Toxicol Appl Pharmacol. 2013; 272(2):490-502.
5. Chan K, et al. Expert Opin Drug Metab Toxicol. 2005; 1(4):655-69.
6. Takahashi Y, et al. Biosci Rep. 2015. 35(3).
7. Van der Windt GJ, et al. Immunity. 2012; 36(1):68-78.
8. Vander Heiden MG, et al. Science. 2009; 324(5930):1029–1033.
9. McBeath R, et al. Dev Cell. 2004; 6(4):483-95.
10. Bell CC, et al. Sci Rep. 2016; 6:25187.

Simon Messner is senior product manager ([email protected]), Randy Strube is global marketing director, Katrin Roessger is application scientist, and Jens M. Kelm is chief technology officer at InSphero.

Previous articleDiscovery of Rapidly Adapting Antibody Opens Door for Universal Flu Vaccines
Next articleGenomic Signatures Clearly Dictate Schizophrenic Drug Responses