Introduction
Mitochondria play critical roles in all cell types. They are involved in processes such as calcium storage and signaling, cell proliferation, apoptosis, steroid synthesis, immunity, energy balance, and heat production. Inherent to these vital functions, mitochondria represent frequent targets for toxicity and disease. A plethora of toxicants are indeed known to affect mitochondria. Furthermore, >200 clinical pathologies have been associated with impairment of mitochondrial functionality, including cancer, inflammatory disorders, neurodegenerative diseases, and diabetes. Both structural defects and functional deterioration may underlie mitochondrial toxicity, and as such form key events in various adverse outcome pathways (AOPs). Not surprisingly, mitochondrial toxicity is increasingly gaining attention by toxicologists and, therefore, was the central theme of this expert roundtable discussion.
Dr. Mathieu Vinken: Thank you all for joining our panel today. My name is Mathieu Vinken and I am the European Editor of Applied In Vitro Toxicology. The idea of today's panel is to discuss the importance of mitochondrial toxicity.
Now, before getting into the discussion, I welcome you all to briefly introduce yourselves.
Dr. Yvonne Will: My name is Yvonne Will. I am currently based in Connecticut. I spent the past 16 years at Pfizer pioneering mitochondrial toxicity testing in industry.
Dr. Hartmut Jaeschke: This is Hartmut Jaeschke with University of Kansas Medical Center in Kansas City. My interest is in drug hepatotoxicity and especially acetaminophen toxicity, and I have done a lot of work on mitochondria over the past 20 years.
Dr. Ken Wallace: This is Ken Wallace with University of Minnesota, and my interest is in mitochondrial toxicity as well, focusing somewhat on Adriamycin and the heart, but also looking at the molecular biology of mitochondrial biogenesis and turnover.
Dr. Philip Hewitt: This is Phil Hewitt. I am at Merck in Germany. I am head of the investigative toxicity group here, so my expertise is with in vitro and organ toxicity. I also have some experience with mitotoxicity testing.
Dr. Hilmi Orhan: This is Hilmi Orhan with Ege University, Faculty of Pharmacy. I am interested in mitochondria both as target for and as perpetrator of various drug-induced adverse reactions. Specifically, we work on enzyme content and potential metabolic capacity of this organelle toward xenobiotics.
Dr. Mathieu Vinken: Thank you. My main expertise is in the field of liver-based in vitro modeling and also mechanistic modeling, so I am really interested in AOPs. Mitochondrial toxicity is a key event that pops up in many of these AOPs. This is one of the points that we will touch upon during the roundtable discussion.
So, the first question for our panel is: How indicative is mitochondrial toxicity for organ-specific toxicity? It seems to manifest as a generic event in many toxicities, so how specific would that event be?
Dr. Philip Hewitt: I would say, with my limited experience and also working in industry, within the investigative toxicity groups here in Europe, that mitotoxicity is not very specific for target organs.
I get the impression that there are known links to certain target organs, such as liver, kidney, or heart, mainly because they are the most studied and perhaps we will find more target organs coming up as we study them in the future. So, I think it is not very specific, apart from when you start looking at metabolism, maybe the liver is the main target only because of exposure and the metabolism the liver causes in the toxicity in mitochondria. I put that out there first as my generic response, before anyone jumps in with very specific cases, which are linking the mitochondria to a specific organ toxicity.
Dr. Yvonne Will: I agree with Phil. The organ toxicities studied the most, at least in large pharma, were due to attrition in the clinic due to liver and cardiac toxicity.
There were a few exceptions to the rule, such as cerivastatin, which was attrited due to muscle toxicity. So, I would agree with Phil that the most studied are those that really attributed to severe toxicity or death. But, if you look more into the academic field, you do find that it is applicable to any organ toxicity, even reproductive toxicity (sperm).
Dr. Ken Wallace: My impressions are that interfering with mitochondrial function is a universal event. It has the potential of occurring in almost any tissue and a major factor that determines organ-specific toxicity first is pharmacokinetics, including the metabolism, as you already mentioned. If it is not delivered to that tissue for pharmacokinetic reasons, through transporters or perfusion or metabolism, then it will not have a prominent effect in that tissue, so that would be the first thing, pharmacokinetic/absorption, distribution, metabolism, excretion (ADME) type of considerations.
The other thing is you will see adverse effects, but only those that you are looking for. So, in the heart, you might see it as contractility or rate, cardiac function. In the liver, if you have a model system that depends highly on cholesterol synthesis or steroid hormone synthesis, etc., you will see it there. But, if your experimental model system does not have a prominent or a very high activity in those metabolic pathways, you may not see it. Basically, you are only going to see it depending on the experimental model and what you are looking for.
Dr. Mathieu Vinken: That is a really good point, Ken. I agree, and we will be coming back to this when we talk about in vitro toxicity. Regarding the first question, on the organ-specific features, any comments left from Hilmi or Hartmut, for example?
Dr. Hilmi Orhan: I think somewhat differently from Phil and Yvonne, probably because they both have extensive industrial experience in drug development also regarding mitochondria, while I am in academia. In my opinion, although examples are few/scarce, mitochondria may have a major role in organ-specific toxic effects of chemicals. This is especially the case for tissues that are relatively more in need of oxygen and energy (adenosine triphosphate [ATP]) compared with other tissues. These are the nervous system and heart, may be muscle tissue as well. Although it would be highly speculative without experimental support, we still do not know whether mitochondria from different tissues/organs respond in a similar manner or differently to the same toxicants.
Dr. Hartmut Jaeschke: The only thing I would add is it depends on the cell as to how many mitochondria are available and how relevant the mitochondria really are for the function. For example, the liver has a lot of mitochondria and can be exposed to many of the chemicals in high concentration. That is why it is important. Other organs may not quite be as exposed as well as they do not have as much mitochondria. They are not as relevant, depending on the mitochondrial function directly.
One thing we have to be careful to argue is mitochondrial toxicity being the primary cause of cell dysfunction, versus an indirect effect. If it is an indirect effect, I think mostly every cell type would probably be affected by that. If you start killing the cell, mitochondria will go down, too. The question is—is that a primary cause of it?
I think in testing and also in the pathophysiology, whether you want to target mitochondria or protect mitochondria against an insult, it is a big question if that insult really causes cell toxicity by initially targeting mitochondria or as an indirect effect. That will definitely be a question when you measure or try to analyze these things in vivo, what is the cause versus an effect.
Dr. Philip Hewitt: I think you have hit the nail on the head there, as to how you actually define that and how you actually study that. And if you look at that question—is mitotoxicity indicative of organ toxicity—I would probably change my answer to say “yes.” It is not organ specific, there are too many other factors involved, like you described, and the different mechanisms where you might be hitting the mitochondria, which is then maybe driving the organ-specific toxicity.
Dr. Ken Wallace: I would like to agree with what Hartmut said—is the mitochondrial dysfunction a cause or consequence—it is important to distinguish between these two.
Dr. Mathieu Vinken: Exactly. And this brings us to the second question, which more or less relates to this. How would you define or how would you consider mitotoxicity within the concept of AOPs? Is mitotoxicity a mode of action or a mechanism of action? Would you consider mitotoxicity as a mechanism of action or an AOP on its own?
Dr. Hartmut Jaeschke: I do not want to halt the discussion too much but I think it is a key event in an AOP. I do not think you can just put mitochondria out there by itself. It has to be considered in the context of the pathophysiology, and it comes back to the cause or effect question. My vote would be it is one of the key events, or it can be a key event, by itself. However, mitotoxicity is not an AOP by itself.
Dr. Hilmi Orhan: When thinking about the universal apoptotic cell death mode where mitochondria play a central role, or acetaminophen-induced programmed necrosis in hepatocytes, in which many steps were enlightened by Hartmut's group,
1
for instance, mitochondria seem to be a significant mediator. In this sense, I also would agree with Hartmut that mitochondrial dysfunction is a key event by itself, but not an entire AOP, as subsequent events are necessary for the ultimate toxic end point.
Dr. Philip Hewitt: How I see it is I completely agree that there are so many different mechanisms and events that lead up to that or what would be the consequence of mitotoxicity, suggesting that it is not an AOP in itself. It is involved in so many different things that it is a very important key event in many AOPs probably.
Dr. Yvonne Will: Yes, in my laboratory, and in other laboratories as well, we have studied hundreds of drugs for their mechanisms of toxicity, but over time we came to the conclusion that the majority of drugs have multifactorial toxicity. The only drug I can think of being close to a pure mitochondrial toxicant, where you could pinpoint the mode of action, would be phenformin.
But, when you look at drugs, for example, inducing liver injury and there is enough out there in the literature now showing that most of them either have a bile salt export pump (BSEP) inhibition, metabolite formation, endoplasmic reticulum (ER) stress, etc., I have to agree that, 10 years ago, I would have said that mitotox is driving the organ toxicity, but today I know that in most instances it is just part of the overall mechanisms contributing to organ toxicity.
Dr. Mathieu Vinken: The reason why I brought this question up is because people in the AOP field really would like to make or establish an AOP on mitochondrial toxicity. I fully agree with you that it is rather like a key event in different AOPs than an AOP as such. According to you, would it make sense to make an AOP on mitochondrial toxicity?
Dr. Ken Wallace: I have a fundamental concern with this question. First of all, I agree that if mitochondrial dysfunction is part of an AOP, it is a key event. Certainly, it is not the end result. And the reason is that you may have mitochondrial dysfunction measured by whatever means you wish to measure it, reactive oxygen species production, decreased ATP, and mitochondrial biogenesis, whatever measure you choose to use and label it as a mitochondrial toxicity. It may or may not result in an adverse outcome.
Dr. Mathieu Vinken: Correct.
Dr. Ken Wallace: My belief is that mitochondria are highly resilient and they are able to withstand a variety of different changes in environment. They adapt to this and so many of the changes that we report as mitochondrial toxicity are, in my opinion, compensatory to avoid an adverse outcome. So, if I had my way, I would discourage all of us from using the word mitotoxicity and use “altered mitochondrial function” or “mitochondrial dysfunction.” The term “mitotoxicity” itself implies adversity and, in my mind, it does not necessarily result in an adverse outcome.
Dr. Yvonne Will: I really like that, Ken. I think that is important for us, in this introductory chapter, to talk about. I am smiling as I listen to everybody because, for me and all of us, we have done this for a long time and we had a lot of learning from thinking about toxicity to admitting that it is most likely multifactorial. I liked when Ken talked about the dysfunction, the adaptation, and the lack of true toxicity. This is wonderful that we can talk about the learnings we have had in the past 10 years and think about how we can be doing some of this more systematically.
Dr. Hartmut Jaeschke: I would fully agree with that. The mitotoxicity also implies somewhere that the chemical hits the mitochondria directly and, in many cases, mitochondria are a secondary target during the process of toxicity.
The issue with mitotoxicity, as Ken pointed out, implies something different than what the actual involvement is of mitochondria in cell injury or cell death in addition to the adaptation issue. So, that is in addition to the question of adaptation. I fully agree with Ken; biogenesis, mitophagy, and other things, the cell really tries to respond to an insult and depending on, of course, if it is a chronic insult or very acute, it can or cannot successfully mitigate some of the dysfunction.
During cell death mechanism, the mitochondria can be involved very central to it but it may not be the direct hit. So, mitotoxicity implies that and I think what Ken suggested, mitochondrial dysfunction, or something like that, is probably the better term here to be used.
Dr. Hilmi Orhan: It is indeed a very good point that “mitotoxicity” may not be the right term, thanks Ken. Adaptation of cells to a new order in the presence of/dictated by a chemical or its reactive metabolite is often an overlooked phenomenon in toxicology and related disciplines. Also, as already has been stated, mitochondria are organelles that have relatively higher adaptation/defense capability toward chemical insults.
Dr. Mathieu Vinken: I fully agree with that. Thank you, all. Maybe we can move to the next set of questions, which is more about the in vitro testing of mitochondrial toxicity. How relevant is the testing of mitotoxicity in vitro for the in vivo situation? What are the most important biomarkers? What are the prominent techniques and in vitro models that are currently used?
Dr. Yvonne Will: I can share what we have learned. There was a 2019 article that we published (Rana et al.
2
) and it shows that you really cannot predict organ toxicity from in vitro testing.
And, something you will see very soon is a special issue (of Chemical Research in Toxicology) on liver injury that Eric Blomme from AbbVie is the editor of, and there are numerous articles in there where people have developed what they call a risk matrix, which is combinations of the mechanisms that could lead to a liver injury. And mitochondria are certainly part of the assessment and contribute to the risk.
If you assign arbitrary severity scores, so you say anything <30 times Cmax (cyclosporine peak concentration) for instance, gets a score of 1 and everything more than it gets a score of 5, you will get a predictivity toward adverse outcomes or toxicity. However, what we do know is that a simple in vitro assessment alone will not tell you which organ will be affected.
So, unless you have other very organ-specific contributing factors, such as inhibition of BSEP for a liver and you include that measurement, you will not even get close to predicting where it might happen.
Dr. Mathieu Vinken: If I understand correctly, basically this would mean that the selection of the biomarkers and the in vitro models is to be seen on a case-by-case basis?
Dr. Yvonne Will: Yes, an in vitro model alone will not tell you what organ toxicity you will experience.
We have made progress, especially in the liver space by knowing the additional factors and using computational and modeling approaches to actually build the relationship from in vitro to in vivo outcomes. And, like we said earlier, it is, at this point, a contributing factor, not per se the driving factor.
Dr. Philip Hewitt: You could also say that you cannot predict the organ that will be affected as an in vitro mitotox assay per se will also not predict organ toxicity, or potentially additive mechanisms—in addition a positive in vitro assay does not guarantee a positive in vivo assay.
Dr. Yvonne Will: That is correct, Phil, yes. And that is why there are a couple of articles coming out very soon at least showing that progress has been made with liver, where once you do know the additional mechanisms, you can establish an in vitro to in vivo correlation.
Dr. Philip Hewitt: But that does not mean the basic mitochondrial dysfunction screening in vitro assay is not worthwhile, especially in the pharmaceutical industry when you do it early enough and to screen out those compounds that have very potent effects on the mitochondria. Why would you continue with those generally? So, I think it is not like you should be picking and choosing assays. Some of these screening assays still have a lot of utility.
Dr. Yvonne Will: Oh, absolutely. You brought up a really good point, Phil. Within industry, we had a lot of learnings of how physical–chemical property space can drive certain toxicities just as it can drive positive ADME properties. So, yes, there is definitely a need, even in the absence of any kind of predictivity, to try to make sure you have better chemical matter. There is no doubt about it.
Ten years ago, we screened thousands of compounds, and then we learned. We learned about physical–chemical property space, we learned some of the structural alerts. Now, this has shifted to more of an in silico approach. So, some companies that have done a lot of the work can apply more of the in silico approach and then can go in with more targeted testing.
Dr. Mathieu Vinken: Thank you. Hartmut, you wanted to add on that?
Dr. Hartmut Jaeschke: No, I want to ask a question first. When you talk of in vitro assays, are you talking about using isolated mitochondria and exposing a drug to that, or do you mean cell lines or primary cells are being exposed to the drug? That would be a fundamental difference.
Dr. Yvonne Will: Yes, correct. I can only speak for pharma, and every pharma company has a bit of its own testing paradigm. Some companies do isolated mitochondria, knowing the caveats, and then go in to, for example, the glucose/galactose model system or primary hepatocyte. Mathieu and Phil probably know under Horizon 2020, there was a big initiative in Europe, where they validated a particular glucose/galactose cell model for the prediction of liver injury (MIP-DILI). So, every company has a little bit of their own flavor of what they do, knowing what the caveats are of the cell system, as well as the isolated system.
Dr. Hartmut Jaeschke: From my perspective, and it is more from an academic perspective, but what I see also in the literature is that no, you can really bark up the wrong tree by exposing isolated mitochondria to any chemical. That can lead you in a completely wrong direction. I would definitely propose using isolated cells as a minimum to do that. Isolated mitochondria are, to some degree, very questionable. I do not know what Ken's perspective is?
Dr. Ken Wallace: I see this as basically two different questions we are asking. One is more on the industrial side or regulatory side. From pharma—if I may take the liberty even though I am in academia—I would imagine that you set up a mitochondrial screen and the objective there is to screen out any chemical that has a “potential” effect against the mitochondria, so you are very willing to withstand false positives.
What you do is you take a sensitive model system, and it might be isolated mitochondria in this case, and you screen for a potential effect that the chemical “may” have on mitochondrial function. Then you compare the concentrations required with that that might be achieved clinically.
Using isolated mitochondria, however, will not reveal any effect that your chemical may have on transporter systems or the citric acid cycle or fatty acid transporters or metabolism, so it is limited in that sense. The question for the clinical application or clinical trials is a different question and as Hartmut, you were getting toward, is to look for relevance.
You try to pick a model system that may be most relevant for extrapolation to the in vivo model. Take a primary hepatocyte, for example, and it has all these components that are active, including drug metabolism, drug transporters, and intermediary metabolic pathways and such. There, you would avoid the isolated mitochondria and you would use a model system that is closest to in vivo that you can. But to me, it is a totally different question if you are asking for clinical screening.
Dr. Hartmut Jaeschke: I recognize that exposing isolated mitochondria to a compound may reveal direct mitochondrial toxicity, which may be useful information under certain circumstances. However, I would still emphasize a cell-based screening approach to mitochondrial toxicity testing as this occurs under more relevant conditions. If there is a positive signal for a compound, these experiments can easily be repeated with primary human cells, which would help with the mechanistic understanding of the toxicity and would allow extrapolation to the clinic. The main concern with exposing isolated mitochondria to drugs is that too many effects are being generated that may lack pathophysiological relevance.
Dr. Ken Wallace: Right, I do not mean to discount that. Again, I am in academe so I may be speaking out of my field, but I would say the in vivo cell model would be like a stage 2 process.
Dr. Yvonne Will: That is correct, Ken. There is a tier 2. What industry does, they know what their issues have been. They know what drugs have failed in animals or in humans and so they have put in place what they saw as the most relevant.
And from there, if there are hits in whatever these assays are for that particular company, there is a tier 2, which looks more academic, very much similar to what you guys are trying to do. Can we have a better model? Can we go in with better biomarkers? Can we do gene expression? What is the best model system, absolutely? Just not when you have hundreds or thousands of compounds.
Dr. Hilmi Orhan: The general/fundamental problem associated with in vitro testing of biological functional structures is also the case for testing mitochondrial dysfunction; to what degree a particular in vitro system represents what in fact happens under in vivo conditions? It is also crucially different testing mitochondrial dysfunction in cell line or in isolated mitochondria, as Hartmut pointed out.
I can give an example for a potential inconsistency between in vitro and in vivo situations, an interesting article has been published recently stating that the average local temperature of mitochondria is ∼47°C.
3
If this is correct, it may have a significant effect on the predictions made based on in vitro testing of mitochondria, as almost all in vitro tests are conducted at the standard temperature of 37°C.
Dr. Mathieu Vinken: So, we have been talking about the industrial relevance of all of this, but what about regulatory relevance to testing mitochondrial toxicity in vitro? Do you think that would be of interest to regulators?
Dr. Yvonne Will: Some of us have made that our lifetime dream and hope that that was the case. It has happened when you look at the transporters. There is an understanding how inhibition of the liver transporters can lead to adverse events and there are at least some guidance documents from the FDA.
Again, the learnings we had in the past 10 years, and mostly in the liver injury space, mitotox has been, with the exception of phenformin, mostly a contributing factor. As such, I do not think that you have a case to push it as a regulatory mandate. I think the most you can get is an acceptance that it can be a significantly contributing factor and if you have more than three or four mechanisms of toxicity, you would have to show certain biomarker requirements to monitor it in the clinic.
Dr. Philip Hewitt: I personally am in the early discovery, not really going into clinic, but I would say the question is wrong. It has of course a relevance to the regulatory environment. I mean, if you understand the mechanism of your drug as you go into clinic and it may involve mitochondrial dysfunction, and you have data to show it, you should submit it. And the regulatory agencies will highly appreciate it, but it should not be regulated by the agencies to say you need to provide this information.
Whether the assays themselves or the data you are generating need to be qualified or validated is up for debate. Normally, from my very limited experience, they are quite open for these types of assays that are not fully validated.
Dr. Ken Wallace: I agree with what has been said. Any type of mitochondrial evidence you have would be supplementary to support observations that are made in a more robust model system.
Dr. Hartmut Jaeschke: I would agree with that.
Dr. Mathieu Vinken: Okay, thank you. What do you all think the future will hold for this field? What are the main obstacles? What should be prioritized from the technical or conceptual point of view?
Dr. Yvonne Will: For pharma, it is to get from hazard to risk prediction. So far, we have mostly only been able to show mitotox as a hazard and not as the risk. In the liver, there has been progress made to get to more of a risk assessment, but it is based on retrospective analysis of compounds we do not make anymore. Compounds with many structural alerts, metabolite formation, BSEP inhibition, and ER stress, you name it, they have it all.
Generally speaking, pharma does not make these types of promiscuous compounds anymore. People ask for the relevance to the current portfolio, but I would say this is really the biggest issue. You cannot get uptake if you do not know what the risk is.
A good example would be the human Ether-à-go-go-Related Gene (hERG). And even that has been revisited in the CiPA initiative but, for the longest time, there was an acceptance that a certain potency in hERG would lead to adverse events, and people believed that and people adopted it and it became regulatory mandate. That is what I know where pharma is struggling, that there is no risk assessment.
Dr. Philip Hewitt: I can only agree. At Merck, this is where I see the main problem, the acceptance of the assays. We need to generate our own positive experiences with these early assays and with the mechanistic assays, like Seahorse. Then you can start persuading teams to use it and then, hopefully, we have some nice publications coming out on how to perform risk assessments.
But that is, to me, the key hurdle at the moment. We know we need it. I think pharma has realized that this is an important key mechanism of toxicity, but how we do it and how we present it to regulatory agencies is still unclear. How you assess that in the clinic as well, I also have no experience how you would do that.
In the future, we will definitely perform more in vitro testing and I am hoping to also use primary cells in the future, rather than cell lines. Maybe the 3D models, which can be accessed through Seahorse—I do not know whether you do that at Pfizer, Yvonne, whether you use spheroids. All these things need to come in and be tested and where models such as Zebrafish assays fit is unclear. We have no experience whether that is better for predicting. Do you have experience, Yvonne?
Dr. Yvonne Will: No, and I think one of the handicaps in all of this is that with the exception of liver, we do not actually have enough compounds to start modeling. You need certain numbers and certain outcomes to actually start modeling risk and we do not have that. So we would love to do this for other organ toxicities, but how many compounds do we really have even in the large companies such as Merck or Pfizer. There have been efforts at times under Innovative Medicines Initiative (IMI) and others to try to push forward joint studies trying to get to this risk prediction, but it has not really come to fruition, I would say.
Dr. Mathieu Vinken: Thank you for that. Ken, any ideas or suggestions for the future of mitochondrial toxicity testing?
Dr. Ken Wallace: I think that Yvonne summarized my major concern, and it goes back to using the term mitotox, distinguishing between risk and hazard. There is value in doing both types of toxicity testing but I favor encouraging our colleagues to be very discreet in defining whether it is a hazard identification or a risk assessment based on the model systems that they are using.
Dr. Hartmut Jaeschke: I agree, but in contrast, I come from a slightly different angle compared with industry. If you want to look at the pathophysiology, you normally start with a disease or an injury or toxicity, and then you go into the mechanism. From that, mechanistically, based on the importance of mitochondrial injury in the pathophysiology overall, it is very important.
This is a different kind of approach compared with what industry is trying to do. You have a compound, you have no idea how safe it is and you want to test whether there is any issue when you go into people with that. And so, the testing and the way you approach it is I think completely different.
Dr. Yvonne Will: Yes and no, Hartmut, because that is how it started, right? Drugs attrited and people started trying to figure out why and that is how mitochondria were identified as well as transporters and reactive metabolites. Where it started was the acceptance that, yes, every drug we ever attrited would flag in these effects. So, I think both exist. It is just the realization of it in industry that spurred an in vitro testing cascade.
There is something else I would like to point out. What is very important is to continue the partnership between industry and academia; it is because of people such as Ken, Hartmut, and Hilmi who are helping us to understand the nuances, and all the different mechanisms and the adaptations. That was not even something we talked about until recently.
I think it would be nice to continue to encourage the relationship between academia and pharma because without academia, there is no way that pharma will put any more money into basic research. Those times are over. The partnership needs to be there and that is why efforts such as Horizon 2020 or IMI are beautiful, because both partners can be involved, and only if you have that partnership, you are going to advance the field.
Dr. Mathieu Vinken: I fully agree. As an academic, I can only echo this. This kind of collaboration with industry is really valuable, especially in such a challenging field.
