Idiosyncratic drug-induced liver injury (DILI) is a significant burden to health organisations and is a major contributor to cases of drug attrition. Mitochondrial dysfunction is increasingly implicated in the onset of idiosyncratic DILI; with variants in mitochondrial DNA (mtDNA) found to induce changes in mitochondrial function. We hypothesise that inter-individual variation in mtDNA underlies the idiosyncratic nature of some cases of DILI via changes in mitochondrial function which can alter susceptibility to injury and we therefore aim to develop an in vitro model to test the effect of this variation pre-clinically.

Progress prior to placement

We have performed next-generation mtDNA sequencing of 384 healthy volunteer samples and identified mitochondrial genotypes. This information has been used to inform the recall of volunteers with known mtDNA genotypes from which fresh platelets (containing mtDNA) can be isolated and fused with HepG2 rho-zero cells (liver carcinoma cell line devoid of mtDNA) to generate transmitochondrial cybrids, a personalised in vitro model.


Many standard toxicity endpoints do not identify subtle changes in mitochondrial morphology and ultrastructure as a consequence of mtDNA genotype; I planned to overcome this by using the Operetta High-Content Imaging System (PerkinElmer) available at GlaxoSmithKline, Ware.

My placement began with an Operetta training session I had arranged with a PerkinElmer software specialist. This provided me with an opportunity to formulate specific Operetta imaging and analysis protocols to best address my research hypothesis. During the placement I was able to culture transmitochondrial cybrids derived from ten genotyped individuals and use the Operetta System to identify basal differences such as variation in mitochondrial dynamics (fusion and fission of mitochondria) and interaction with other cellular compartments including lysosomes. The high throughput nature of the Operetta System also enabled the rapid assessment of cybrids under multiple conditions including treatment with five different compounds associated with clinical idiosyncratic DILI over multiple time courses. Imaging was also performed following a two hour drug-free incubation to examine for evidence of mitochondrial recovery, a significant benefit of live cell imaging.

Importantly, Operetta-associated software has enabled me to perform fast and reliable quantification of mitochondrial and cellular health via machine-learning and development of image analysis algorithms. The unbiased recognition of multiple mitochondrial phenotypes across cybrids from different genotypes and conditions has enabled the reliable quantification of genotype and compound-induced differences in mitochondrial function. The analysis protocols I have generated for this can now be used at GSK in multiple studies, providing increased consistency across assessments of mitochondrial health.

During my placement I was also able to present my PhD project at a multi-site Safety Pharmacology meeting and had ample time to discuss the future utilities of my model, as well as gain advice from toxicologists working alongside me. I have since been invited to present again at this meeting upon completion of my PhD studies and I look forward to presenting the final results of my project, particularly those from the placement.


In summary, this placement has allowed me to perform a thorough assessment of the the clinically-derived cell models that I have generated at the University of Liverpool. This has been possible not only because of the high throughput and high quality facilities available at GSK, but has also afforded the opportunity for collaboration with industry-led toxicologists, many of whom I am still in contact with. As well as benefiting my own research, the knowledge I have from this placement will be disseminated to other members of the Centre for Drug Safety Science at the University of Liverpool as part of a departmental seminar series.