The generation of HepG2 transmitochondrial cybrids to reveal the role of mitochondrial genotype in idiosyncratic drug-induced liver injury
Curation statements for this article:-
Curated by eLife
eLife assessment
This paper is of potential interest to scientists within the field of drug-induced liver injury. The concept of the study is interesting by generating mitochondrial genotype-specific liver cell lines to evaluate idiosyncratic hepatotoxicity. While the proof-of-concept is clearly presented, the current data do not yet allow to draw broad conclusions about the significance of the study in terms of drug effects.
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
- Pharmacology and Toxicology (eLife)
Abstract
Evidence supports an important link between mitochondrial DNA (mtDNA) variation and adverse drug reactions such as idiosyncratic drug-induced liver injury (iDILI). Here, we describe the generation of HepG2-derived transmitochondrial cybrids, to investigate the impact of mtDNA variation on mitochondrial (dys)function and susceptibility to iDILI. This study created 10 cybrid cell lines, each containing distinct mitochondrial genotypes of haplogroup H or haplogroup J backgrounds.
Methods:
HepG2 cells were depleted of mtDNA to make rho zero cells, before the introduction of known mitochondrial genotypes using platelets from healthy volunteers (n=10), thus generating 10 transmitochondrial cybrid cell lines. The mitochondrial function of each was assessed at basal state and following treatment with compounds associated with iDILI; flutamide, 2-hydroxyflutamide, and tolcapone, and their less toxic counterparts bicalutamide and entacapone utilizing ATP assays and extracellular flux analysis.
Results:
Whilst only slight variations in basal mitochondrial function were observed between haplogroups H and J, haplogroup-specific responses were observed to the mitotoxic drugs. Haplogroup J showed increased susceptibility to inhibition by flutamide, 2-hydroxyflutamide, and tolcapone, via effects on selected mitochondrial complexes (I and II), and an uncoupling of the respiratory chain.
Conclusions:
This study demonstrates that HepG2 transmitochondrial cybrids can be created to contain the mitochondrial genotype of any individual of interest. This provides a practical and reproducible system to investigate the cellular consequences of variation in the mitochondrial genome, against a constant nuclear background. Additionally, the results show that inter-individual variation in mitochondrial haplogroup may be a factor in determining sensitivity to mitochondrial toxicants.
Funding:
This work was supported by the Centre for Drug Safety Science supported by the Medical Research Council, United Kingdom (Grant Number G0700654); and GlaxoSmithKline as part of an MRC-CASE studentship (grant number MR/L006758/1).
Article activity feed
-
-
Author Response
Reviewer #1 (Public Review):
Idiosyncratic drug-induced liver injury is a disease that appears to be linked to mitochondrial DNA (mtDNA), but there is a lack of model cell lines for the study of this link. To help address this problem, the authors developed ten cybrid HepG2 cell lines that have had their mitochondrial DNA replaced with the mitochondrial DNA of ten human donors. Analysis of single nucleotide polymorphisms in all of the patients' mtDNA allowed the authors to assign the donors to two haplogroups (H and J) with five patients each. The authors also present the results of several assays (e.g. oxygen consumption, ATP production) performed on all ten cell lines in the absence and presence of five clinically-relevant drugs (or drug metabolites). Significant attention was paid to differences observed between …
Author Response
Reviewer #1 (Public Review):
Idiosyncratic drug-induced liver injury is a disease that appears to be linked to mitochondrial DNA (mtDNA), but there is a lack of model cell lines for the study of this link. To help address this problem, the authors developed ten cybrid HepG2 cell lines that have had their mitochondrial DNA replaced with the mitochondrial DNA of ten human donors. Analysis of single nucleotide polymorphisms in all of the patients' mtDNA allowed the authors to assign the donors to two haplogroups (H and J) with five patients each. The authors also present the results of several assays (e.g. oxygen consumption, ATP production) performed on all ten cell lines in the absence and presence of five clinically-relevant drugs (or drug metabolites). Significant attention was paid to differences observed between the cell lines in the H and J haplogroups. The work is methodologically and scientifically rigorous, ethically conducted, and objectively presented according to the appropriate community standards.
While I feel that the manuscript will be useful to the research field and is an important step towards improving patient outcomes, I feel that the work lacks a broad interest. Much of the paper is spent discussing small and/or statistically insignificant differences between haplogroups H and J. While some interesting interpretations and suggestions are presented in the discussion, the authors didn't perform follow-up experiments to try to nail down any particular mechanistic insights that would be useful to the broader community. I also didn't feel a strong sense that the paper produced any specific suggestions for how clinical outcomes could be improved. Accordingly, any clear insights that would be interesting to a broad scientific community would probably require follow-up studies.
Again, we strongly believe that the subject is of broad interest to researchers in both academia and the pharmaceutical industry. Evidence of the level of interest in this subject can be quantified by the access metrics of the 3 publications we have recently published on this topic (Biochem Soc Trans, 2020, PMID: 32453388; Arch Toxicol, 2021, PMID: 33585966; Front Genetics, 2021, PMID: 34484295), which have been accessed >6000 times.
The structure of the paper is also not friendly to a broad audience; the results are presented without interspersed commentary that could help the reader understand the meaning or utility of the results as they are being presented. Accordingly, I often felt unsure about how the results being presented were relevant to solving the broader problem established nicely in the introduction.
We thank the reviewer for this comment and have revised the manuscript to now contain a combined results and discussion section.
Finally, it wasn't clear that the generated cell lines were made available for anyone to purchase through a cell bank (perhaps the authors did do this, but I don't recall seeing a mention of it). As these cell lines appear to be the primary output of this work, it seems important to better highlight the extent to which they are being made accessible to the scientific community.
The cells are currently in the process of being deposited under licence with XimBio. This will allow other researchers to easily access them. They are also available upon request from me. This has been conveyed in the revised manuscript (pg 18, lines 1-2).
Reviewer #2 (Public Review):
In this work, Ball et al. investigated the possibility to generate a novel set of HepG2 liver cell lines to generate "mitochondrial DNA-personalized" models as novel tools to study idiosyncratic drug-induced liver injury related to mitochondrial variation. This work represents the generation of a comprehensive collection of n=10 HepG2 lines, half reflecting haplogroup H and half reflecting haplogroup J. The authors then assessed their impact on basic mitochondrial function in liver cells. Interestingly, they find a greater respiratory complex activity driven by complex I and II of the haplogroup J lines relative to haplogroup H. Finally, the authors make an attempt at using this novel set of lines to probe the consequential effects of mitochondrial genotype on drug-induced liver toxicity. This work provides an interesting proof-of-concept study and is a starting point towards studying and predicting idiosyncratic drug-induced liver injury in a personalized manner. This technique may be broadly extrapolated to other commonly used liver cell models within the toxicology field.
Strengths:
- This work presents an exciting initiative to study interindividual variability in idiosyncratic drug-induced liver injury focusing on mitochondrial haplotypes. In further follow-ups, this work could be extended to also represent other different haplogroups to establish a thorough "biobank". The established lines allow for future in-depth characterization and testing of many putative hepatotoxic compounds through a variety of toxicity measures that could shed further light on the impact of mitochondrial DNA variation on (idiosyncratic) drug-induced liver injury.
- This technique may be broadly extrapolated to other commonly used liver cell lines within the toxicology field (e.g. HepaRG cells or iPSC-derived cells) that are potentially also more metabolically competent. A short discussion on this could be added to the current manuscript.
We thank the reviewers for this comment, which we agree with. We have now incorporated this into the conclusion (pg 18, lines 23 - 27).
Weaknesses:
- The major weakness of the current manuscript is the rather large variation across sample measurements regarding the proof-of-concept experiments to study drug effects (fig. 3-6). This makes much of the data rather hard to interpret and to infer conclusions. As an example, proton leak (fig. 3f/4f) seems to 2-fold increase in the J group even under basal conditions (0 uM flutamide/metabolite), while this is not observed in fig. 2a and this effect seems to be also absent under 0 uM tolcapone (fig. 5f). Unfortunately, the current data do not allow us to draw confident conclusions about whether the tested drugs have effects on the mitochondrial respiration of the different haplogroups. This may well be linked to the methods used for measuring mitochondrial activity, but since this is the predominant method needed in the current paper, either increasing the number of experiments (across more lines) or identifying a more rigorous methodological manner to obtain consistencies of experiments would help the authors to make more confident claims about their data.
The reviewers have noted the inherent variability in the respiratory measurements from plate to plate. To counter this, experiments were designed so that for each cybrid cell line the control and treated cells were always positioned on the same plate. However, we believe that the reporting of such data, and their limitations, is a fundamental aspect of unbiased science reporting feeding into the principles of data reproducibility. In this resubmission, we have updated the methodology of our data analysis, which better accounts for this variability. The new figures plot each cybrid as a distinct point to easily visualise the variation across haplogroups dependent upon each cybrid within the group. We have included this limitation in the conclusion (pg 18, lines 15 – 19).
- The data on the effects of inhibition of complex I/II activity are not sufficiently convincing to support the claim that haplogroup J is more susceptible to flutamide/metabolite (fig. 6). Both seem to respond rather identical to flutamide or its metabolite, i.e. at higher concentrations complex I/II activity decreases, but with the sole difference that the haplogroups represent different basal activity (not influenced by the drug). Estimating fold changes, for example, for both haplogroups, complex I and II activity decreases ca. 2-fold at the highest concentration of the metabolite (fig. 6c-d), therefore concluding that there is no difference between haplogroup susceptibility unlike the authors claim. It is furthermore unclear what the statistical significance currently represents: it should represent whether at different/increasing concentrations the activity of the complexes significantly differs vs. the previous/basal conditions from the same haplogroup. If it represents (which it seems to be) the significance of the haplogroup J vs. the haplogroup H, it is non-informative as it is obvious that haplogroup J presents with a higher baseline.
Thank you for this comment, we agree with the shortcomings of statistical analysis in fig 6 and have reanalysed the dataset using a more appropriate statistical methodology, see response 2.2.
- It would help to mention how many lines per haplogroup H/J were used in the analyses across all figures. This should be clarified, as the error bars for most experiments are rather high and therefore statistical significance is lacking, making data interpretation complex. It could be helpful if the authors present at least for some analyses single plots of data obtained across different lines from the same haplogroup to evaluate the consistency of the effects of the genotypes as supplementary figures. If only 1-2 lines were used per group, it would help to perform additional experiments to assess consistencies across groups.
We apologise that the number of lines per haplogroup that were employed in the analyses is unclear. In every case, we included 5 cybrid lines per haplogroup. We have further clarified this point in the methods and results. Furthermore, in the new figures, each cybrid is now represented as a single data point.
-
eLife assessment
This paper is of potential interest to scientists within the field of drug-induced liver injury. The concept of the study is interesting by generating mitochondrial genotype-specific liver cell lines to evaluate idiosyncratic hepatotoxicity. While the proof-of-concept is clearly presented, the current data do not yet allow to draw broad conclusions about the significance of the study in terms of drug effects.
-
Reviewer #1 (Public Review):
Idiosyncratic drug-induced liver injury is a disease that appears to be linked to mitochondrial DNA (mtDNA), but there is a lack of model cell lines for the study of this link. To help address this problem, the authors developed ten cybrid HepG2 cell lines that have had their mitochondrial DNA replaced with the mitochondrial DNA of ten human donors. Analysis of single nucleotide polymorphisms in all of the patients' mtDNA allowed the authors to assign the donors to two haplogroups (H and J) with five patients each. The authors also present the results of several assays (e.g. oxygen consumption, ATP production) performed on all ten cell lines in the absence and presence of five clinically-relevant drugs (or drug metabolites). Significant attention was paid to differences observed between the cell lines in the …
Reviewer #1 (Public Review):
Idiosyncratic drug-induced liver injury is a disease that appears to be linked to mitochondrial DNA (mtDNA), but there is a lack of model cell lines for the study of this link. To help address this problem, the authors developed ten cybrid HepG2 cell lines that have had their mitochondrial DNA replaced with the mitochondrial DNA of ten human donors. Analysis of single nucleotide polymorphisms in all of the patients' mtDNA allowed the authors to assign the donors to two haplogroups (H and J) with five patients each. The authors also present the results of several assays (e.g. oxygen consumption, ATP production) performed on all ten cell lines in the absence and presence of five clinically-relevant drugs (or drug metabolites). Significant attention was paid to differences observed between the cell lines in the H and J haplogroups. The work is methodologically and scientifically rigorous, ethically conducted, and objectively presented according to the appropriate community standards.
While I feel that the manuscript will be useful to the research field and is an important step towards improving patient outcomes, I feel that the work lacks a broad interest. Much of the paper is spent discussing small and/or statistically insignificant differences between haplogroups H and J. While some interesting interpretations and suggestions are presented in the discussion, the authors didn't perform follow-up experiments to try to nail down any particular mechanistic insights that would be useful to the broader community. I also didn't feel a strong sense that the paper produced any specific suggestions for how clinical outcomes could be improved. Accordingly, any clear insights that would be interesting to a broad scientific community would probably require follow-up studies. The structure of the paper is also not friendly to a broad audience; the results are presented without interspersed commentary that could help the reader understand the meaning or utility of the results as they are being presented. Accordingly, I often felt unsure about how the results being presented were relevant to solving the broader problem established nicely in the introduction. Finally, it wasn't clear that the generated cell lines were made available for anyone to purchase through a cell bank (perhaps the authors did do this, but I don't recall seeing a mention of it). As these cell lines appear to be the primary output of this work, it seems important to better highlight the extent to which they are being made accessible to the scientific community.
-
Reviewer #2 (Public Review):
In this work, Ball et al. investigated the possibility to generate a novel set of HepG2 liver cell lines to generate "mitochondrial DNA-personalized" models as novel tools to study idiosyncratic drug-induced liver injury related to mitochondrial variation. This work represents the generation of a comprehensive collection of n=10 HepG2 lines, half reflecting haplogroup H and half reflecting haplogroup J. The authors then assessed their impact on basic mitochondrial function in liver cells. Interestingly, they find a greater respiratory complex activity driven by complex I and II of the haplogroup J lines relative to haplogroup H. Finally, the authors make an attempt at using this novel set of lines to probe the consequential effects of mitochondrial genotype on drug-induced liver toxicity. This work provides …
Reviewer #2 (Public Review):
In this work, Ball et al. investigated the possibility to generate a novel set of HepG2 liver cell lines to generate "mitochondrial DNA-personalized" models as novel tools to study idiosyncratic drug-induced liver injury related to mitochondrial variation. This work represents the generation of a comprehensive collection of n=10 HepG2 lines, half reflecting haplogroup H and half reflecting haplogroup J. The authors then assessed their impact on basic mitochondrial function in liver cells. Interestingly, they find a greater respiratory complex activity driven by complex I and II of the haplogroup J lines relative to haplogroup H. Finally, the authors make an attempt at using this novel set of lines to probe the consequential effects of mitochondrial genotype on drug-induced liver toxicity. This work provides an interesting proof-of-concept study and is a starting point towards studying and predicting idiosyncratic drug-induced liver injury in a personalized manner. This technique may be broadly extrapolated to other commonly used liver cell models within the toxicology field.
Strengths:
This work presents an exciting initiative to study interindividual variability in idiosyncratic drug-induced liver injury focusing on mitochondrial haplotypes. In further follow-ups, this work could be extended to also represent other different haplogroups to establish a thorough "biobank". The established lines allow for future in-depth characterization and testing of many putative hepatotoxic compounds through a variety of toxicity measures that could shed further light on the impact of mitochondrial DNA variation on (idiosyncratic) drug-induced liver injury.
This technique may be broadly extrapolated to other commonly used liver cell lines within the toxicology field (e.g. HepaRG cells or iPSC-derived cells) that are potentially also more metabolically competent. A short discussion on this could be added to the current manuscript.
Weaknesses:
The major weakness of the current manuscript is the rather large variation across sample measurements regarding the proof-of-concept experiments to study drug effects (fig. 3-6). This makes much of the data rather hard to interpret and to infer conclusions. As an example, proton leak (fig. 3f/4f) seems to 2-fold increase in the J group even under basal conditions (0 uM flutamide/metabolite), while this is not observed in fig. 2a and this effect seems to be also absent under 0 uM tolcapone (fig. 5f). Unfortunately, the current data do not allow to draw confident conclusions about whether the tested drugs have effects on the mitochondrial respiration of the different haplogroups. This may well be linked to the methods used for measuring mitochondrial activity, but since this is the predominant method needed in the current paper, either increasing the number of experiments (across more lines) or identifying a more rigorous methodological manner to obtain consistencies of experiments would help the authors to make more confident claims about their data.
The data on the effects of inhibition of complex I/II activity are not sufficiently convincing to support the claim that haplogroup J is more susceptible to flutamide/metabolite (fig. 6). Both seem to respond rather identical to flutamide or its metabolite, i.e. at higher concentrations complex I/II activity decreases, but with the sole difference that the haplogroups represent different basal activity (not influenced by the drug). Estimating fold changes, for example, for both haplogroups, complex I and II activity decreases ca. 2-fold at the highest concentration of the metabolite (fig. 6c-d), therefore concluding that there is no difference between haplogroup susceptibility unlike the authors claim. It is furthermore unclear what the statistical significance currently represents: it should represent whether at different/increasing concentrations the activity of the complexes significantly differs vs. the previous/basal conditions from the same haplogroup. If it represents (which it seems to be) the significance of the haplogroup J vs. the haplogroup H, it is non-informative as it is obvious that haplogroup J presents with a higher baseline.
It would help to mention how many lines per haplogroup H/J were used in the analyses across all figures. This should be clarified, as the error bars for most experiments are rather high and therefore statistical significance is lacking, making data interpretation complex. It could be helpful if the authors present at least for some analyses single plots of data obtained across different lines from the same haplogroup to evaluate the consistency of the effects of the genotypes as supplementary figures. If only 1-2 lines were used per group, it would help to perform additional experiments to assess consistencies across groups.
-
