Uncovering chemical biodegradation in the environment - Kathrin Fenner, Eawag
In this episode I speak with Prof. Kathrin Fenner, one of the leading academic scientists in the study of chemical biodegradation and environmental persistence.
Kathrin and I have a far-reaching conversation covering her background and research interests, challenges and developments in chemical biodegradation research, and implications for policy and society.
We have learned an enormous amount about the fate and persistence of chemicals in recent years, and the work of Kathrin and her group has been at the heart of this.
Listen to the episode to find out about biodegradation of chemicals in different environmental settings, what factors contribute to persistence, understanding transformation products and pathways, predicting biodegradation for different chemicals and scenarios, and how we can utilise this knowledge toward safer and more sustainable use of chemicals.
Works cited:
Innovate beyond PFAS | Science
Designing Small Molecules for Biodegradability | Chemical Reviews
Discover enviPath.org
Prefer to read? Here is a transcript:
Chris:
Hello everyone, and welcome to the Chemical Journeys Podcast. Today I'm speaking with Professor Kathrin Fenner, Senior Scientist and Research Group Leader in the Department of Environmental Chemistry at Eawag and Professor in the Department of Chemistry at the University of Zurich. Kathrin, thanks for joining me.
Kathrin: 00:50
Yeah, thanks a lot for having me. Looking forward to our conversation.
Chris: 00:55
Me too. I must say I'm really excited to be speaking to you today on this podcast because for people who may not know, your research is really focused on the topic of environmental persistence of chemicals, which the listeners will know is really one of the main themes of this podcast. And if you don't mind me saying, you have been a really key figure in this area for quite some years now and have contributed enormously to both our understanding of how chemicals degrade in the environment and also towards developing tools and practices so that we can continue to improve and make better use of this knowledge. So before I start, I just wanted to acknowledge that and to really thank you for what you've achieved in your career so far.
Kathrin: 01:38
Thanks a lot for the flowers, Chris. I guess that's what I've been trying to achieve, and I'm really glad to see that it's kind of heard and taken up by people. Thanks a lot.
Chris: 01:49
No problem. And with that out of the way, it would be great if we could kick off by you just telling us a bit about yourself. So something about your your background and your research interests.
Kathrin: 02:01
Right, yeah. So I'm a chemist by training, but I think to understand a little bit kind of where I'm coming from, it's also good to know, you know, how I grew up. So I grew up in something you would probably call like an environmentalist household. My dad was one of the first to install solar collectors in our village. We had an organic vegetable garden and a pond and frogs and chicken as well that we were growing. So I think I really, as a child, just absorbed the value of nature surrounding us. But then at the same time in high school, I was really interested in natural sciences and mathematics and also really the beauty of chemistry. You know, chemistry really talked to me, I would say. It included some of the physics and mathematics that I liked, but it was much more tangible. So I went for that. But then already during the studies, it was funny, you know, whenever I got to do an independent project, it would be something environmentally relevant. Like I would measure the oxygen levels in our ponds, and I did a project on acid rain, and then a master thesis on photodissociation of chlorine dioxide clusters, which was relevant in the ozone depletion context. But then when I wanted to go with this for a PhD thesis, I realized that environmental chemistry is really not a thing in chemistry in Switzerland. So my professor at the time, you know, although he kind of I think really loved me as a student, he was like, no, no, no, I don't want to go into this. I had to look for other places. And then I was lucky that ETH a few years ago had founded this chair of environmental safety and technology by also this was Professor Hungerbühler, and this chair was sponsored by chemical industry after the Schweizerhalle accident. And it was all about chemical risk assessment, life cycle assessment in chemical industry. And this was really interesting to me, and so I kind of found my new home there. And the other lucky thing about this was that Martin Scheringer, who some might know, right, from being, especially most recently, one of the most outspoken PFAS experts as well, he was my PhD supervisor, and I was his first PhD student. So we really kicked it off together. And these were extremely formative years for me, because Martin at the time was really thinking very deeply about persistence and long-range transport potential, and he really derived this from first principles of intergenerational justice, interregional justice. And I had a liking also for humanities and social sciences, so kind of trying to really think about these concepts and then how we can translate this into models to capture this quantitatively together with Martin was just wonderful. And then this is also kind of what I worked on in my PhD, you know, it's like how to assess overall persistence in the environment and what is the role of transformation products. And this is kind of still what I'm doing today, if you want. So I want to find out how we can best assess persistence and how we can include transformation products into these considerations, also how we can predict this really from structure, ideally. How can we predict persistence and transformation products from structure? And now, increasingly, also kind of taking this knowledge and shaping it into reasonably easy tools that could be used at the early stages of chemical design to already consider at least persistence at those early stages of research and development. So, yeah, that's more or less my my journey and and what I do.
Chris: 06:22
No, that's a really nice introduction. There's a lot in there that would be great to sort of dive into with you. I guess the first thing you mentioned is is the beauty of chemistry. And I and I think that you know, especially in this area that we're working, environmental chemistry and and and biodegradation and trying to understand the environment. I think that it's it is really quite enchanting in terms of its complexity, its its depth, and and and the mystery of it. Even though we are, you know, intended to be rational scientists, you can't help but get sort of drawn in by that.
Kathrin: 07:00
Yeah, yeah. I mean, I remember when I read my first paper about structure activity relationships, you know, this was, I don't know, in the early 2000s or something like that, or even still in the 90s. And that really struck me that you can look at the structure of a chemical and learn a lot from it and predict things. And this fascination is still there all the way to today, where we're trying to use more modern, like machine learning-based approaches and so on to tackle this, but it's really this fundamental structure, in our case, biodegradability relationship, which is challenging, and maybe we'll come to it in a bit more detail.
Chris: 07:46
But yeah, yeah, and then the other piece is around seeing that connection between our so you know, in some ways, narrow fields of chemicals in the environment and how that connects to the broader society and the the the human picture of where we are on this on this planet and the economy and everything else. It comes comes out of all the discussions, especially when we go to a place like SETAC and start speaking with people and and seeing how you know a lot of these relationships of chemicals to to health can can can be really impactful.
Kathrin: 08:27
Yeah, and I think this is where persistence has such a key role, you know, especially if we think about this more intergenerational justice aspects, and obviously PFAS bring this home very well, right? I mean, these chemicals are really, it looks like they're gonna be around for a very long time, and that our children will inherit them, so to say, from us, and also the costs of removing them are potentially enormous. So it just shows that one of the key or maybe most urgent tasks for us is to make sure that we recognize these kinds of chemicals early enough before we get into these kind of lock-in situations as as we have it now with PFAS. And I think this is where some of the research that's been going on in the field also on developing maybe more efficient biodegradation assays that can, for instance, recognize just that, you know, when when do we have these like extreme persistence cases as they're called nowadays? That is is a really important task, in my opinion. Yeah.
Chris: 09:47
Yes, yeah, and and we can I guess we can we can dive into this area of persistence of chemicals and the the the factors that influence it, the processes that we're that we're trying to understand, because there is an immense complexity there, and that is a challenge that you've highlighted already. That you know, amongst that we need to get useful, meaningful information that we can base decisions on amongst that sea of other information, and you know, make good decisions with the chemicals that we use in daily life without getting bogged down in the kind of the the the fascinating curiosity of every aspect or paralysis by analysis, yes, or you know decisions being influenced by by other factors, you know, different interests as well. But maybe maybe we before we before we dive into some of the more recent science, I I wanted to ask you because I know I'm aware of your earlier work in the in on your PhD, and uh you did a lot of work with multimedia fate modeling at that time, but you seem to have moved much more into an experimental space now. And do you have any broad reflections about that journey?
Kathrin: 11:08
Good question. I think one thing that we all experience as modelers is that we're always looking for experimental data, and so that was also a major part of my PhD, you know, that I was collecting data on the compounds I wanted to use as case studies for my PhD. So half-lives in different environmental media, but also partition coefficients and so on and so forth, which is what you feed into those models, and then you realize A, that data is rather scarce, and B, I did a lot of sensitivity analysis at the time, and this sensitivity analysis was clearly telling me that the one thing that is most important in my models are those half-lives that were hard to find. And I think out of that, it there is some logic, you know, related to saying, okay, let's let's try and measure some more of these half-lives. And then another part of it really was the advances in mass spectrometry. So high-resolution mass spectrometry became really a thing kind of shortly after I'd moved to Eawag in the 2008-2010, around that time. And this really also opened up a lot of opportunity to do experiments more efficiently because the analysis became more efficient, and also look at transformation products a lot. And I got also really fascinated by these opportunities. So I think those two things came together and made me move more back towards experimental work as well.
Chris: 12:54
Yeah, and maybe we can jump to that then in terms of going about the task of measuring degradation as it as it happens in the environment. Maybe you could talk us through a bit of what that entails and what the the challenges are uh around that. And I guess the first thing that jumps out to me is the variability that we observe in how chemicals degrade.
Kathrin: 13:23
Yeah. Yes, I think I mean that definitely is the one or the biggest challenge, perhaps, right? That if we if we take a chemical and measure its half-life, even in different soils, so we stay within one compartment, so to say, we will get potentially large differences in the answers up to an order of magnitude, or even more, as you well know, right? Um, and then if we go to another compartment, sediment activated sludge things will start looking even more different. And this variability is a real thing, that's exactly what it looks like in the environment as well, but it makes it very challenging on uh on a on the regulatory side of things or just in general on the decision side of things, because we want to usually compare things to a threshold or something like that. So this is true, but I think over the many many years of research that we've now been doing in the field, two things have come out quite clearly for me, if I may briefly go there. Despite all of this variability, structure matters. So when we look across many sets of different uh half-lives, you know, for the same chemical or for a bunch of chemicals, then we do find some logic. So the relative ranking is usually not all that different. We recently compared half-lives in soil and half-lives in water sediment systems for a lot of pesticides, and and there is a correlation very clearly. So this says despite all of this variability, there are structures that make a chemical more persistent, and there are structures that make a chemical more degradable, I would argue, overall, right? These are general statements, but I think they're true in that general sense. And if you then look at these structures and what makes them more persistent and less persistent, it's also not breaking news. It's more or less what Bob Boethling has already been describing in the 90s when they developed BIOWIN, right? It's the branched structures, the halogenated structures, the ether structures, and anything that nature doesn't really know so well are harder to degrade than other things that they know from that are known from biomolecule structures are tend to be easier to degrade. So I think that is still true. And the other thing that I've also learned is that there are certain factors that influence the half-lives that we know about and that we should take into consideration. It's very clear that the amount or concentration of biomass matters. It's also very clear that uh the bioavailability matters or the bioavailable fraction matters. And this has been known for 20, 30 years. We can find it in the textbooks. And I'm I'm sometimes just a bit frustrated how these uh simple things are not taken into account when we when we look at different experimental outcomes. Because I think a lot of the things that look random or disparate, when we do take these aspects into account, they they actually start falling into place. And and this is, I think, what we should do much more. And and the the third aspect that is also obvious, I think, is that there is such a thing as adaptation to high concentrations or higher concentrations of a substance. We can discuss what that means exactly, adaptation. We have quite some evidence that makes me think I know what's going on. But these are three aspects that we can be rational about, that we can make sure we also describe with the parameters that we report for the experiments, and that really help us to make sense out of what looks like house numbers, right?
Chris: 18:06
Yes, yes, and uh yeah, you've done some work in various areas related to these topics. I guess the thing that's coming to mind is normalizing the degradation rates to biomass concentrations to get a the second order rate constant, which seemed very fruitful at the time. And I and I guess a lot of this stuff is sort of in the literature and has been for a number of years. Um, but seemingly there's a challenge to get things implemented into our decision-making procedures because we rely on you know OECD test guidelines, for instance, which have long, long time horizons for being refined and updated, and there's a political process there. Um and yeah, this uh all these things take a a lot of time. I guess the other thing that came to mind is there's a that there's this variation, and so with that, as you say, when you s when you have enough data you start to see that things do rank, uh, you know, or or we can start to get make more sense of that, but getting that data is still a challenge from a resource point of view in terms of the the the costs and the effort of generating this data.
Kathrin: 19:22
Yeah, definitely. I mean the way we generate it currently seems not ideal to to get at these larger volumes of data that we would ideally have. I think there is a good case to be made that we can already work with the data that we have using transparent schemes. And I think your recent publication there on weight of evidence schemes and the kind of a clear procedure how we deal with this data that we already have is is a very important aspect there. And there we probably cannot treat it like quantitatively so much, but if we record all the metadata on the soils, on on the initial biomass that we find even in these regulatory studies, I think it does it does help in this weight of evidence, like high quality, transparent weight of evidence procedures. So I think that's important. If we talk about kind of more data and faster data generation, yeah, as I said, the simulation studies usually are just too long, too time consuming, too expensive, so to say. And on the other side, we have the ready biodegradability type studies, or we see OECD 301 that they fulfill a certain job, so to say, but in in many ways are too conservative to get good persistence estimates, right? And so that's why I think it is important to think about alternatives, which is also something that ECHA, for instance, has promoted in their kind of key areas of research report. They talk about middle ground tests. But I think the the challenge there now for us as a community is really to think about what that could be and what the purpose of that would be exactly. Because it's easy to say middle ground or whatever test, and then everybody has some ideas what that could be and why. And I think kind of finding that middle ground that makes sense and and for what purpose is something we now really have to have a discussion on, right? And yes, we have some suggestions in that realm, but I'm not saying this is one and only reasonable suggestion, yeah.
Chris: 21:58
Yes, yes. I mean Perhaps briefly on the screening tests. I mean, one thing that I've been thinking about with this, because I've been following a lot of the regulatory decisions and also the guidance developments, and the sense that I get is that regulators are really moving away from the conditions of screening tests and are less convinced that the screening tests give them useful information for their assessment of persistence, which ultimately is to measure a half-life, and they measure a half-life, you know, typically using a simulation test in water, water sediment systems or soil, but with the idea that that gives them an idea of the rate of degradation in the real environment. And I think you know, further to that, they are when they're thinking about that, they're thinking about diffuse low concentrations of a chemical in the environment, you know, where perhaps you know you're not seeing this stimulation of the microbial communities to degrade or to utilize the chemical, but that it's degrading kind of in concept, you know, with other things. And I think that's a a problem. I see because from an industry point of view, it seems that there's still very much when they think about biodegradation testing and persistence, they're still thinking very much in the case of screening tests, and there's very little data out there from from a simulation test point of view. So I this is a bit of a tangential point, but I wondered if you had any reflections on that.
Kathrin: 23:44
I think what I really like about the screening tests is the endpoint they're measuring. So they're they're looking for a complete mineralization if you want so. And this in principle would allow dealing with this whole problem of transformation products to some extent, right? If a chemical mineralizes to large extents, one could hope that persistent transformation products would be less of an issue. It has been shown that that is not necessarily the case. You might still have some small persistent transformation products forming, so I'm really not sure it solves that issue. But that's what I really like about it in principle, because with all the simulation studies, we tend to look at mostly primary half-lives, and this is where we might lose sight of the transformation products, and that's an issue I see with the simulation studies. But definitely with the screening tests, this high substrate to biomass ratio, I think that has been shown to be a real problem and not very realistic. And the problem with it is really that it causes these adaptation phenomena, which might really mean that the test is deviating from realistic conditions. And it also is the reason, I think, for a lot of the variability we see in the test results for screening tests, and that also doesn't make them more valid, right? So they they really do have a lot of issues. My dream would be something like a screening test in terms of endpoints, you know, where we would really capture perhaps mineralization as an endpoint, but have a more realistic substrate to biomass ratio, which is very challenging because then you get into other practical problems. So I'm I'm not sure this is a realistic dream, but that would be a nice middle ground test for me.
Chris: 26:01
And I do feel that there's almost kind of like two scenarios, largely that we're thinking about here. We're thinking about those low diffuse concentrations, which might be where there's a residue of chemicals in the environment or from widespread dispersive emissions or aerial deposition, and there's an interest to understand the fate of chemicals in that context, and then there's also the concentrations at a point of emission, so where chemicals are being applied to land or they're coming out of a wastewater treatment plant, and it seems that we at the moment we kind of blinkered in our persistence framework in that we're only looking at the first thing and we'd lost sight of the or we've not lost sight, but we don't really take into account that the degradation at higher concentrations, which in some ways accounts for the ability for the environment to react to elevated concentrations and bring them down and bring the risks down. I think this comes a bit to light for me as well when we talk about PFAS, because PFAS obviously are very resistant to degradation, and one of the big concerns around TFA, for example, is that its concentrations are getting quite quite high now in the environment. And if it was possible for microorganisms to adapt to TFA and bring those concentrations down, we I don't think we would be as as concerned about it. So I my feeling is that there are different scenarios which both have value or both are both relevant depending on the context. And yeah, uh I I mean we have a framework and it's what we use to make decisions today, and I hope in the future we can continue to discuss and and refine the framework, but you know, that's just some some reflections there.
Kathrin: 27:53
Yeah. Um as I said before, I think our most urgent task, so to say, would be to flag things like TFA or PFAS in general and have a have a test that really flags those very persistent chemicals quite easily. And this is maybe brings me to another middle level or middle ground test, is I think also a bit of a no-brainer, in my opinion. I mean, if you want to test for these really non-degradable chemicals, then you throw in enough biomass and you make sure your chemical is reasonably bioavailable. And the best kind of approximation of that, in my opinion, right now, is something like an activated sludge test. In principle, right? High biomass, but at the same time also a high liquid to solids ratio, not a lot of that organic matter as we have it in soils and sediments. Most of the organic matter is actually live biomass. And that I think could give you that differentiation into those that are really very hard to degrade, and others that degrade under certain circumstances, as you just described. And those kind of chemicals there in the middle, they're obviously the most challenging. I fully agree that we can create conditions where some chemicals degrade actually quite well. And we've just seen it in a very recent study that just got published where we compared degradation across different wastewater treatment plant technologies for a large set of nearly 150 chemicals. And there we see that certain quite persistent chemicals like Dichlofenac, for instance, that's a painkiller, can get degraded in MBBR, so moving bad bioreactors, which are kind of biofilm-based systems, because the biomass to degrade them can obviously grow in those systems. And we mostly see these kinds of phenomena for highly concentrated substances, so that are used a lot and present in high concentrations in the influent. So, yes, diclofenac does get degraded under certain circumstances in the wastewater treatment plant, but it doesn't in the downstream river. And actually, we also could show that that's work that is not published yet, that a lot of these compounds that get degraded in wastewater treatment plants because of these kinds of adaptation phenomena don't get degraded in the downstream or upstream rivers, which also shows exactly what you said before. There might be situations in the environment where they actually don't get degraded because the concentrations aren't high enough for the communities to adapt. And this really, yes, it depends on the exposure scenario, whether or not those chemicals do get degraded. And for me, this is a very tricky question to answer, what that then means in terms of regulatory persistence assessment, which which I mean, in a way we want to protect the most vulnerable environments, I would say, like for instance, like groundwater resources, and there we often have quite dilute concentrations, and so I think we should have then also the more conservative test situation, you know, to reflect these diffuse low concentration situations. But yeah, these are all real phenomena, and how to exactly deal with this in a regulatory context is is really challenging, I find, and I don't have the answer to be honest.
Chris: 32:23
Yeah, no, I fully appreciate that, and yeah, it's it's there's always a kind of what you could do kind of question, and then what what you should do kind of question. And yeah, I guess with with PBTs, for example, there there's this issue of the compound magnifying in organisms and in the food chain, and even their you know the low concentrations can can still lead to issues, and then that's and that's documented, and then there's also this uncertainty of whether chemicals are going to have specific biological activity, which I think some of uh publications that you've been associated with has kind of drawn out this this this point. We have intentionally biologically active chemicals and non-intentionally biologically active chemicals, but we still have an uncertainty over the latter as to whether they truly are not biologically active. Um so it is there's still a a real there's a lot of question marks over the chemicals that we that we have and and yeah, I I we we also have we have a modern situation that you know wasn't really by design, but it's what we have, right? And so that that's why there's a lot of debate raging about what to do about that that situation. Yeah. But maybe perhaps we again we're kind of meandering through through our topics, but uh going back to your um the work you mentioned there on on miniaturization or or high throughput testing, do you want to expand a little bit on on what has been done there?
Kathrin: 33:60
Yes, there is work out there, right, on miniaturizing more the screening tests, and I guess in my team we've worked more on miniaturizing something like a simulation test, but we actually have done this based on activated sludge. And this is because we've shown previously that the half-lives that we get from these activated sludge assays are actually also predictive to some extent of the half-lives we measure in soil or in water sediment studies. I think this is important to realize, you know, that it's not just activated sludge, it does show us some of this general biodegradability capacity of chemical structures, and that's why we like to run it, but we like to use it because it has this high biomass and this high bioavailability compared to the other systems, and so it allows for really short test durations. So we usually run these tests for like three days. And now we've also been able to show that if we make this smaller, right, to use like two milliliter or one milliliter volumes, we get very comparable results to the large volumes. And again, I think this is also due to the high biomass concentration. So we don't seem to get any of these lottery effects of subsampling certain parts of the community. And right now we're trying to also see to what extent this can be implemented on pipetting robots, because the whole idea would really be to make this a tool that fits much better into a R&D pipeline in industry where you have small amounts of your substances typically, and and work a lot with like high throughput, so pipetting robots, high throughput assays. And yeah, this is this is ongoing, I would say. Maybe one point to mention also is that we work with rather large mixtures of compounds there, and we've also done work to show that you know, even if we work with a hundred compounds, these mixtures don't create any kind of negative or mixture type effects, but we work at pretty low concentrations, right? So in the low microgram per liter range, and I think that's really key. Then the other thing we're also working on there, which I find very important, is that we try to couple this to bioassays. This is together with my colleague Beate Escher at UFZ Leipzig.
Chris: 36:56
Yes.
Kathrin: 36:56
Because of this transformation product issue, yes, right? Because we mostly look at parent compound disappearance, and we can never be sure enough that that there aren't any problematic transformation products formed. And so by coupling degradation and kind of testing this degradation mixture at different stages with a battery of bioassays, we can at the same time see whether the parent is degrading, but also whether nothing more hazardous or persistent is being formed. And so this I find really exciting. We still have to demonstrate it. We're really in the in the beginning of this experimental work to couple these two assays, which is not all that easy. And now we want to demonstrate it for different compounds. And I've I find that very exciting, yeah.
Chris: 37:57
Yeah, yeah, I really look forward to seeing any results that you publish in that area. And I guess again, it kind of goes back to um looking for highly recalcitrant things versus transient things, right? And perhaps our frameworks are kind of optimized towards measuring rates, and but it could be potentially more important to understand is there something that is essentially non-degradable that has this potential to stick around and cause adverse effects.
Kathrin: 38:27
Exactly. It sounds yeah, it's really those two things. That's what, at least in theory, such a coupled essay could really capture. Yes.
Chris: 38:35
It sounds like you are moving very much towards more intrinsic approaches to assess chemicals. Um having worked for a long time on trying to measure degradation in different settings in the environment or in different environmental compartments. And uh I know you've done a lot of work on the different simulation tests and how to perhaps make those more refined or more robust. And I guess that the other thing that that kind of overhangs a lot of this is that we can never be fully sure that what we measure in a lab is really what is going on in the environment, and there are often a lot of reasons why what we're measuring in the lab is quite divergent from what's going on in the environment. Do you have any reflections around that?
Kathrin: 39:22
This is also one thing we have tried to look into a bit more. The most important piece of work there is the work of Carolin Seller, who did 309 type studies in the lab, and then we compared that to the fate of different substances in the Rhine catchment. So we measured the substances in the Rhine catchment, their fate, and then modeled the Rhine catchment, which also allowed us to derive then degradation rate constants in the catchment. And then you can start comparing lab and field studies. And this looked reasonable, the the numbers that we got in the lab, so to say, or yeah, the rate constants and those in the field correlated to some extent, but we also realized to get kind of the absolute numbers right is is very difficult. So they did actually differ by an order of magnitude or so. And I think this uh shows that there might be an important role for something like degradation benchmarks, so kind of reference chemicals that we know something about their their fate really in the field, and then also use them in in laboratory experiments to kind of relate those results to these reference benchmarks, because that direct translation is is challenging and yeah.
Chris: 41:06
Yes, yes, it's really tricky, and it's also very difficult to find a nice field study, you know, where the conditions are just right and you can account for all the kind of inputs and outflows of chemicals and water to it's very difficult to really measure what's going on.
Kathrin: 41:26
It's it's actually extremely hard to find just like a stretch of river, for instance, where you could really observe degradation because of all these inputs, and and the rivers flow fast usually, right? So there isn't much degradation going on. That's why we then ended up kind of modeling the whole rain catchment because it's only at this like very large level that you can even observe degradation in the Rhine. I I just had one more thought. Sometimes feel that this whole discussion of how well our laboratory experiments really predict the half-lives that we observe in the environment is a bit misleading. Um of course, we would want that our laboratory experiments are a reasonable representation of the conditions in the environment, and that's where I, for instance, criticize the screening tests as not being a very reasonable representation. So that's that's okay. But then I think we should kind of maybe not get stuck with trying to achieve this direct agreement between a half-life in the lab and a half-life in the field. I think we'll not get there. No again, what's important is that we get the relative behavior right.
Chris: 42:53
Yes.
Kathrin: 42:54
And I think there the evidence is still hard to find because, as you said, it's extremely difficult to observe degradation in the field. But yeah, from our data at least, we have the impression we're not doing that badly now for degradation in the river. Degradation in the soil is a whole different box of worms with NER formation and really kind of strange long-term behavior of some of the substances. And yeah, that is a is as I said, it's a whole other box of worms that we don't do much work on.
Chris: 43:33
Yeah, because you can find residues in the soils from decades of certain compounds, right? And you wouldn't necessarily expect them to be there. Yeah. And you can extract them so you don't you know they're definitely there, but but there remains a question of whether they're causing an impact to receptors in that sort of situation. But that goes back to then. How are you basing your decision making about whether things are acceptable or not, right? And and and also if we're confident that we can be sure that things don't pose a risk in in in the long run as well. So yeah, I mean that's one thing I've been thinking about also with this podcast is especially with the sustainability discussions that have been happening more broadly, it seems to have brought people around to rethinking a lot of the paradigms that we used to to make decisions about chemicals in the past, the risk assessment paradigm and and persistence really finds itself at the heart of it, I think, because um you know when people look at issues like microplastics in the environment and they see that the environment's filling up with microplastics, they they tend to not they tend to have an aversion to that, even irrespective of whether, you know, the the the final clarity on whether there's a the whether whether there are serious environmental or health impacts from that contamination.
Kathrin: 45:07
Yeah, but I I also think these examples uh just bring home that that persistence is is kind of a key factor in all of this. And it's also so this safe and sustainable by design, right, really tries to bring together these two aspects. And this is also quite interesting because the two communities, the sustainability community and hazard risk assessment community, when it comes to chemicals, have in the past not had so many like touching points. Yeah, but I think this is changing quite rapidly. So, for instance, we're currently trying to develop a framework for this safe by design within a project, also with Beate Escher and the computational chemists from ETH Zurich. But now we're already starting to talk to our colleagues that work more on LCA and bio-based raw materials, and how we could also bring these aspects, you know, kind of how much CO2, for instance, is being produced while a chemical is being produced, so to say, how that could be brought into this framework. So this is another really exciting development, in my opinion, that these two communities are coming more closely together again.
Chris: 46:41
I agree. I think we can't think about these things as single issues, we have to try to take everything into account and weigh it. And the sense I get is that that has created a lot of friction and complexity to making progress, and maybe if you're able to kind of go sim, you know, simplistic or you know pragmatic in your screening approaches, then you can avoid some of that. But the sense I get is that the more we try to add to this, it the more unmanageable it becomes. And yeah, we really are meandering a lot of this conversation, but on on the uh experimental data side of things, thank you first of all for mentioning the persistence assessment tool. And I know that you have also done quite a bit of work in this space to try to bring some consistency and accessibility to and and collating and curating the data that are out there. Maybe you could talk to us a little bit about that.
Kathrin: 47:44
Yeah, we have spent quite some time and money over the last years to just curate biotransformation data that's out there, and we really just started with curating the OECD 307, so soil degradation data from the EFSA substance dossiers, simply because that data is out there. Everybody can look at it, but it's in PDFs, and you cannot do any analysis kind of across the substances, across the data in this format. So our work was in that sense very trivial. We just transferred this PDF information into an electronically accessible format, and these are the kind of data sets that we now provide through envipath.org, which is our kind of online biotransformation database and prediction system. And now we've started also sourcing more data from the public literature and also defining a bit of a reporting standard that others could potentially follow, because it's still a huge problem in our field that this information is not provided electronically. Like even when we publish papers or when others in the community publish papers, there is often not even kind of an Excel file that goes with it that has the rate constants and the structures and the metadata information, but it's somewhere again in PDF tables and stuff like that. So, yeah, we call that BART bioassessment reporting tool, not PAT, but very similar, I think, in many ways. It kind of defines the kinetic information that should be reported, um, the models that have been used for these kinetic fits, and then also the metadata. In our case, also the pathway information if we have reactions. And we've at least convinced ES&T or the ES&T group of journals now to also mention this in their author guidelines. Studies that report degradation information should use either BART or even simpler formats that are also provided by others. It's just a recommendation, it's not a must-have. Actually, different from like biology, for instance, or microbiology, where you have to deposit all your sequence information with metadata and in electronic format, we're a bit lagging behind. But yes, so that's something I think we all kind of try to work towards, and that I find very important because then we can start learning by comparing across data sets, across compounds, and we start seeing that things are maybe more consistent than we've been thinking.
Chris: 50:54
Yeah, I know it's really valuable work, and I think that it's something that I've been thinking about on the industrial chemical side, maybe not so much with biodegradation data, but with other data that's now being generated and uploaded in REACH dossiers, it seems to me that we must be getting to a place where we can start to better utilize the data that we've already generated, the knowledge that we have on certain chemistries to to learn about other chemistries, you know, avoid the use of animal testing as much as we can. So I hope that and especially with the advent of AI and machine learning, I think this should get even this should accelerate even more. So I hope that we can get to a place where we where we start to do utilise that much more and and you know perhaps change a mindset from where we were in 2010, where it was kind of like tick all the boxes, generate every single test for every single substance. Having said that, on the biodegradation and persistence assessment side, we we're still very much lacking the data, I think you you would agree, and there's also such a diversity of different structural, you know, structural complexity that can have a bearing on this that we're not quite there yet, and that we would we need to start leveraging some of these screening tools, perhaps that can give us the useful information that you've already talked about. And then that may then help us to get onto the side of predicting these properties as well. And and perhaps we could go on to a little bit of of the work you're doing in prediction.
Kathrin: 52:35
Yes. Yeah, so we've obviously one of the goals of collecting all of this data in electronically accessible form was to use it then to develop models that would predict some of these endpoints from structure. And we've explored this especially for two endpoints currently. One is Half-Lives in Soil from OECD 307 studies, and the other one is removal in the wastewater treatment plant. And this was actually based on monitoring data, so where you monitor the influent and the effluent of wastewater treatment plants, and with current kind of again high-resolution mass spectrometry-based screening approaches, you can get data for up to a thousand substances. This was actually work done together with Michael McLachlan from Stockholm University. And so in both cases, we ended up with data sets that were around a thousand compounds, a bit less in the case of the soil half-lives. And thought we could now develop nice models with this. And yes, this was quite a sobering experience because we did not get models that you would call well performing. You know, our our R squares were on the order of 0.3, 0.35, and I feel we really tried pretty much everything in terms of state of the art, um, machine learning, and descriptors, and data curation, and whatnot to improve this, and it was not possible. And yeah, I really took two learnings from this. First of all, I don't believe any model that's published in literature that claims to have a R-square of 0.8 and more for half-life prediction. I simply don't. And um, secondly, yes, we need to do better. And I think uh generating more data on more compounds is key because we could really show that, well, in the case of the pesticide half-lives, it we really only cover the narrow part of the chemical space anyway, with these substances. But uh in general, we just don't have a dense enough coverage of that that structural diversity as as you explained a minute ago. This is this structural diversity is really huge. If you combine different functional groups, you very quickly get a kind of a combinatorial explosion, and and even a thousand substances are are far from being able to cover that well. So I think that's one part. I still have some hope in trying other things also on the data that we have, like combining data sets across environments, or using some of these foundational models from chemistry and can retrain them on our smaller data sets. Or for some specific assays, we've also seen that quantum chemical descriptors might be very powerful descriptors, so kind of using smarter descriptors, I think, is still a possibility. So I think we should do both, and that's why it's also important to make these data sets public because I'm I'm sure there are people out there that are much better at using state-of-the-art AI tools than we are, because we're not specialists in this. So that's why I think we really try to make these data sets public through enviPath, but also through GitHub, through yeah, other means, so that people can do better than we did so far with these data.
Chris: 56:49
I find, going back to what we said earlier about the beauty and the fascination of this field, it's also very humbling, right? Because I think more than many others, getting to a true understanding of things is so elusive at times. But yeah, credit you to continue the efforts that you're making in this space. And yeah, you've mentioned enviPath a couple of times. Um, and yeah, I'm really glad to be also working with the team at enviPath to help them with sort of engaging with different potential users of the of the platform and also giving them some insights towards features that they could develop. And I know you you're still very much active also in working with that team. So it sounds like it sounds very exciting in terms of where enviPath could go in the future, especially as we seem to be on the the cusp of this acceleration of data generation and utilization.
Kathrin: 57:53
Yeah, yeah, first of all, it's great to have you on the team, Chris. This is so good because I think your insight into what industry might be interested in, and also regulators, and your domain knowledge as well. It will really yeah, it's really good to have you on board. Yeah, I also think that there are really new opportunities now opening up. One thing we haven't talked about at all yet, but it's definitely coming, is using LLMs to extract data. So far, these were the students, you know, extracting the data. And this can be accelerated and has already been demonstrated in other areas. That's something we're definitely trying out at the moment. We cannot lag behind there, and this will speed up, hopefully, this data collection, at least of the data that is out there. And then my colleagues at enviPath have also been playing around with then using LLMs directly for reaction prediction. This is definitely something we have to keep trying out and seeing how well it does. My very personal opinion is that at the moment we are still in the data poor regime when it comes to transformation reactions and the rule-based system that we've been working with so far. So enviPath is based on a set of around 300 transformation rules that are somehow an extract of all our biochemical understanding, but then also the data we have in the database is for me still at the moment the way to go. We then have kind of machine learning layer on top of that to prioritize the different rules against each other, and that can be kind of continuously refined as new data is coming in. We've now also developed an algorithm that can more automatically extract rules from new data or adjust rules as new data is added to become, for instance, more general or more specific. And so, yeah, I think this combination of getting access to more data through LLM, hopefully help of LLM in data curation, and then this automated rule extraction and training this probability layer should really move us ahead and while at the same time also checking kind of the capacity of LLMs more broadly to directly predict from structure again.
Chris: 01:00:59
Yeah. That sounds really interesting. Yeah, look forward to hearing more about how we can utilize LLMs in this space. And of course, if if anybody wants to know more about enviPath, we'll include a link in the in the show notes and feel free to reach out if you'd like to learn more about that platform. Kathrin, I think we've gone really round the houses with our discussion. I knew it would be fascinating, and I really appreciate the opportunity to talk with you today. We kind of covered, I was going to ask you, you know, what's your outlook for persistence assessments moving forward? I think it's been a very future-focused conversation. So, I'll only invite you to add if there is anything you'd like to add. I also normally ask my guests, you know, what's a big goal that they're working towards in the months ahead. And I think you've you've given us plenty of examples there that you're working towards. But again, uh I don't know if you want to add anything to either of those two things.
Kathrin: 01:02:01
No, as you said, I think we've already covered a lot of ground and also where we're heading. Maybe just to reiterate one thing and and bring in one last aspect that we haven't talked about. So I think now we need to assess persistence at scale. And this is also a perspective that we have submitted to ES&T that talks more about this, where you're also a co-author, right?
Chris: 01:02:30
Yes, of course.
Kathrin: 01:02:30
So and yeah, I think we need this for the R&D pipelines. We need this to generate more data for in silico model development and also to assess our existing chemicals that we have. And there I think this screening for extreme persistence is very key. One thing we haven't talked about at all is how we can also use this understanding in engineering approaches, bio-based, cost-effective engineering approaches. Because as you pointed out, I think if we get the conditions right and the concentrations high enough, then we can train biomass to degrade chemicals. I'm not sure about PFAS, but many of the others, right? And so this is also something I think we should do as a community that we use this understanding to collaborate with engineers and explore kind of bio-based solutions further, because in developed countries like Switzerland we can afford some of these more costly and energy-intense solutions like ozonation, activated carbon, but in in other places this might not be the way to go. And so this is still also very fascinating to me. Can we enrich substances to get to high enough concentrations? Can we make sure that the competent degraders can grow comfortably in a biofilm or something to really increase these degradation efficiencies?
Chris: 01:04:06
Yes, yes, of course. There's a whole world out there of treatment and remediation that we didn't really touch on in this discussion at all, but I know you've been quite involved in that space as well, and I think that also feeds into the adaptation discussion that we touched on. So yes, if there are opportunities in that space, we should we should stay open to that for managing the impacts of chemicals that we're using in commerce. A lot of chemicals, the pharmaceuticals, for example, we're not really in a place where persistence is being decisive at present as to whether we will continue to use those kinds of substances, but they are out there in the environment and they are biologically active.
Kathrin: 01:04:51
Yes.
Chris: 01:04:52
I do also ask, if you don't mind, I normally ask if you had one piece of advice to either yourself or somebody else starting out in their career today, what would it be?
Kathrin: 01:05:02
I think what worked out well for me at the end was to not get too stuck on just one career path. I was always considering, you know, switching into regulation, switching into industry. I was even interviewing for jobs while I was a postdoc or even later when I was already a group leader. I think kind of keep the eyes and the options open really helped me to be relaxed. And then at the end, I kind of was really fortunate to now go along the way I was able to go on. But this was not to be expected. There was a lot of luck involved, and so kind of staying a bit open, I think, helped me on the way.
Chris: 01:05:54
Yeah, it sort of found you in the end, but it was a perfect fit in the end, yeah. No, that's great advice. Thank you, Kathrin. And thank you to everyone who's been listening. If you've enjoyed this, be sure to go back and listen to the other episodes in the podcast where you'll find several discussions on the topic of persistence and the emerging related policy. I'd also appreciate it if you would share the podcast or if you have the opportunity to review or rate the podcast. That would really help me to get the word out to more people. Last but not least, I'd like to mention that I'd be delivering a training course on persistence assessments at the upcoming SETAC conference in Maastricht. So feel free to get in contact with me if you'd like any more information about that. But with that, Kathrin, I yeah, as I said, it's been a real honour and privilege to speak with you today. And congratulations again on the work that you've done to date, and I'm sure there's much more to come. So yeah, thank you very much.
Kathrin: 01:06:57
Yes, thanks a lot, Chris, also from my side.
