Exercise and Prediabetes

Welcome to the Proteomics in Proximity podcast, where your co-hosts, Dale Yuzuki, Cindy Lawley, and Sarantis Chlamydas from Olink Proteomics talk about the intersection of proteomics with genomics for drug target discovery, the application of proteomics to reveal disease biomarkers, and current trends in using proteomics to unlock biological mechanisms. Here we have your hosts, Dale, Cindy, and Sarantis. Hey there, welcome to Proteomics in Proximity where, today, we're talking about a wonderful paper in Cell Reports Medicine with Diaz- Canestro and Aimin Xu out of Aimin Xu's lab at Hong Kong University. We're talking about the characterization, their characterization of inflammatory and cardiometabolic proteins, in particular in response to chronic exercise. And this is a cohort of 36 overweight and obese men with prediabetes. So the team looked at how does response to exercise work? And they used a machine learning algorithm to actually characterize response to exercise. So, this is a very, very different paper than we've talked about before. And I think this is because we're seeing proteomics being leveraged in these areas that we haven't seen before. So very exciting! And you think about the global burden of obesity, being overweight, and diabetes. I mean, it's a huge, huge problem, right? When you look at the obesity rates in developed countries, it's just increasing at these breathtaking rates and diabetes, right? I mean, how many people that we know are pre-diabetic or on diabetes medication or what have you? Sarantis, what can you tell us about this particular group of 36 men? First of all, I will start with the limitations of the study and something that the authors have discussed already, and we are discussing before this meeting, right? It's a small cohort. There are 36 individuals, but also there are males, there are no females. And not very diverse in ethnicity as well. And they pointed out that probably in future studies, they will follow up just to make it more strong, the evidence of these biomarkers for prediction. And I think that's really important to keep in mind when you design a study. Even in their age range, right? They commented on how women, the age range was like 20 to 60, and longitudinally over the course of twelve weeks. They're talking about menstrual cycles, and they're also talking about hormones and how they're varying in women. So therefore, you can say that for this first study, men made a simpler population. Just removing a few variables, right? On the small numbers. And then the other one, which I was thinking about, which is how difficult it is to get obese men to volunteer for something like this. So this group of men were, on average, 40 years old with a full BMI average of 30. So we have obese men, and they were going through high-intensity interval training, which was over the twelve-week period. And, Cindy, how often did they do this? Yeah, they're doing it three times a week. So I think that this is 70 minutes per session. Right. So they had the structure of the exercise set up in several stations to keep it novel. And they maintained the exercise to improve and to change as they got used to the exercise. So they tried to maintain engagement. But like you mentioned, it's hard to keep people exercising over a twelve-week period that frequently, right? Especially if they're going from zero to three times a week. Zero to three times a week, 70 minutes per session. It's too much. It was a ten- minute warm up. It was these stations for about 15 minutes per station. So four different stations and then a cool-down period. And you think, okay, how do we get enough people? But they did this. They found people with an average age 40, average BMI of 30 willing to go through twelve weeks of this. But then they went ahead and, Sarantis, they took a look at their blood samples at zero, four weeks, and twelve weeks. Unfortunately, the test itself, to take the fasting insulin and glucose tolerance tests, are not easy tests, actually. Because they really need a lot of preparation and it's not a straightforward study then and it has a lot of difficulties. Yeah, so the glucose tolerance, Sarantis, what can you tell me about that? I think most of us go through this kind of diabetes screening, but from an insulin resistance point-of-view, what does glucose tolerance show? The insulin resistance actually is a condition of diabetic patients, actually. And there are a lot of actual causes of this disease and most likely it is gene related. But also the environment can play a role. Epigenetics also could play a role on this disease. And at the end, or an outcome, there is a tolerance of the genes and there is resistance of the cells to the insulin. And for this, we have accumulation of glucose in the blood. In simple words, at the end, we just see a lot of glucose being in the circulating blood. And tests also, like C-peptide test or OGTs or insulin fasting and glucose fasting tests, help people to understand the condition of the disease. So by pre-diabetic, meaning their glucose levels weren't so high, but they were approaching that threshold. With the glucose tolerance test, literally they're coming in in a fasted state and they're drinking straight glucose and then measuring their glucose levels. Or maybe they're also measuring insulin levels. But certainly seeing whether your glucose levels shooting straight up and not coming down over a period of time. And if they're not coming down because your body is not releasing what it needs to move that sugar from your bloodstream into your cells, then you are portraying, you're demonstrating some insulin resistance. That's my understanding of it. The remarkable thing about the study is the intervention was the exercise, right? The intervention was these overweight individuals coming in three times a week to a center where they were monitored, told what to do, walk through the steps three times a week for twelve weeks. Must not have been pleasant. And yet, at the same time, they had had no intervention with regard to diet. Basically said you eat the way you normally eat. And through the exercise, I think one of the interesting things about the cohort, in addition to the biology, which we'll get to in just a minute, was they lost weight. I mean, their average BMI went from 30.05 to that time frame, not adjusting for any kind of diet. And it made me think, hey, there's hope. There's hope for the rest of us, as long as we do high- intensity training. I think the paper also points out it was not moderate or low intensity. It had to be, had to be intense. Then they went ahead and took a look at blood at baseline, at four weeks, and at twelve weeks. What did they use to take a look at the protein levels? They use an Olink platform. Yeah, I can answer that. So this group decided to use two Explore panels. So, just as a reminder to our audience, Olink came out with a qPCR readout in the founding of the technology to measure proteins called the proximity extension assay, which is the namesake of our podcast. Then in 2020, Olink expanded its product portfolio to have an NGS, or a next generation sequencing, readout. And the advantage of that is that, in a run of a sequencing instrument, you can get more molecules read out at one time. And so, with our NGS readout, that's our Explore technology, we're able to, today, measure about 3000 proteins in each of using an Illumina NovaSeq instrument, just as an example. Here, they used not the full 3000, they used, what was it? Do I remember that? Right. So they decided, kind of like you were saying earlier, Sarantis, focusing, it in to see, in these extreme cases of men, trying to reduce your variability, reduce your number of variables that you control, and see what the signals are. They also focused in on cardiometabolic and inflammation proteins. We roughly categorized them in these buckets, right? And they stuck with those two sets of proteins. We have a full 3000 and, in fact, you mentioned their future suggestions for where to go in the future, and they suggested expanding it beyond that, because we actually have two cardiometabolic panels and two inflammation panels in our Explore 3072 platform. And so it would be really interesting to see this study expanded. And I think, just from seeing the patterns within just these the machine learning algorithm, we should talk a little bit about that. This machine learning algorithm that was able to integrate protein data and understand, sort of, and predict someone's response to exercise, whether they're going to be what they called a responder or non- responder. Sounds really compelling around precision or individualized training, or individualized prevention of diabetes. And when you talk about responder versus non- responder, you're actually saying that there was a subset of men who did not respond biologically to the sort of pre- diabetes down to sort of a normal level, yet there was a whole other group going through the same exercise regimen that responded. And that's really interesting, the findings. Yeah, and the parameter they used, and this is not something I'm familiar with, so I'm not going to portray that I'm an expert at all, but a clinical parameter called H-O-M-A-I-R. So it's an insulin resistance metric: HOMA-IR. And they had this criterion that it needed to reduce by two-fold in order for them to be categorized as responders. And so, yeah exactly, Dale And regardless, one of the figures that was really interesting was the trajectory of the proteins. So we're looking at 688 proteins from an Explore 384 panel - oh, I'm sorry - two Explore 384 panels. Cindy, correct me if I'm wrong ... You don't need the NovaSeq capacity? That is true. You can do individual 384 panels in each run of a NextSeq. That's right. So they did two of these panels on a NextSeq. I have to mention something for the assays that's really clear and really nice because they orthogonally validated these assays with MSD's [Mesoscale Diagnostics platform]. And they see, in 19 individuals, the really nice correlation with MSD. That's another way that shows specificity and how specific are the assays in our detected proteins, right? Because they can validate it with other orthogonal technologies and have really great data from your data. And to be clear, MSD only had, I think, an overlap of 15 inflammatory proteins. Okay. So they only looked at 15 proteins, a very small subset. But nonetheless, across the 19 individuals, they found a very nice correlation to the vast majority of those proteins. Eleven out of those of 15 were moderately correlated. So you just say, hey, that's a very small subset of only But good enough, right? Yeah. And I'll also emphasize these proteins that they were identifying are low abundant proteins. These are ones where Olink has really shined a light because of the ability for us to come in and hook out these low abundant proteins. These are certainly able to be seen through a mass spec approach, but it's harder to do that because of high abundant proteins being so strong in a mass spec readout. I think you'd have to do some sort of subtraction to remove that component. And so what we're finding is that Olink has a nice way to reveal some of these low abundant signatures that we just weren't easily able to see using existing methods. And I think you point out something really important, Sarantis: that there just aren't that many low abundant assays out there that are commercially available. So it's exciting to me to see. And to look at some of these results now over time, they clustered the response over week zero, week four, and week twelve in Figure 2, which the more I think about it, the more interesting those figures become. Why? Because, for example, cluster one was steadily increasing over time. It starts out at a very low level, goes to an intermediate level at four weeks, and then high at 12 weeks. And those proteins were what? EPO and something they call myokines, which I'd never heard of before. They are a subset of cytokines that are released by the muscle, is how they defined it. But I agree, it's an interesting term. Actually, in the abstract, they call it exerkines, although I haven't heard of exerkines. The myokines are a subset of the exerkines. Exactly. That's great. And IL-6 is there. And IL-6 produced by muscle having a different function than IL-6 produced by the immune system. It's like, wow. Fascinating, right? And didn't we talk about that, or didn't Tthe two of you talked about that with Katerina? She had an example of IL-6 as an assay, which was compelling and very specific and actually, there was a corroborative assay where Olink was able to show these IL-6 signatures that other assays weren't. Not to just detract from the main message, but I'm thinking, okay, how can you determine that this IL-6 molecule, which is identical to the molecule of IL-6 produced by the immune system, how can you tell it's from the muscle? Right, because it's in circulation. But then it was well, no, this is in response to exercise. Right? Yeah. Well, I think the bottom line is this is a signature you're seeing in response to an exercise intervention. And so we're speculating that that's due to exercise. But I think that's a pretty good bet, as you say. Yeah. And then there was another cluster that was also interesting, cluster three, where you had protein expression levels starting at a high level and then steadily decreasing. Okay, so this is another really simple example. But not necessarily the same proteins, right? No. I think that's important to point out. So in these they're looking at pro-apoptotic proteins, meaning these are proteins that encourage cell death. And, Sarantis, you want to comment on that? Why, as a result of exercise, you have the ones that encourage apoptosis decreasing? That's a great point. I mean, I can speculate. I'm not an expert of apoptosis. I think it has to do probably with different signaling pathways so that they're regulating different gene expressions that are related, and auto metabolism, or could be due to regeneration. Or something for cell tissue regeneration. But that's the only some speculation that I have. I mean, these are obese individuals. The body mass index is really high, and you can say they actually have a higher level of apoptotic activity. And to be clear, apoptosis is programmed cell death. This is sort of cells being pruned from your organism as a whole. And you just say, well, maybe this is a function of inflammatory processes. Or homeostasis, Dale. They could be that you shift the homeostasis somehow and could be different pathways that may affect apoptosis. But yeah, it's certainly an interesting finding, that's true. And then to move onto cluster number four, this is appetite stimulating that bounces. It starts at a certain level, goes really low, and then goes up to normal again. And this is appetite stimulating. And I thought ... Well, it's appetite stimulating hormones that are decreased over time. So the idea might be that you're reducing appetite stimulation. That's really interesting. I agree. Yeah. They basically get less hungry, which explains why that, even though they're not modulating their diet consciously, their body is saying, hey, you don't need to eat so much. And probably this is interesting, there is a crosstalk, probably, with neuro genes in the brain. Probably there's some, if you do some CSF [cerebrospinal fluid] proteomics, you may see some hormones released that are changing in the brain. Probably, there's for sure, there's a crosstalk with the brain. But remember, these returned to baseline after four weeks. So these dropped within four weeks, but by twelve weeks, they were back to baseline in this cluster four. And I think there's been some strong evidence that appetite increases when you take on an exercise program. I don't think it's known whether it's a biological increase or a psychological increase that we tend to eat more because we think that we're exercising. But it's really an interesting finding. And then the proapoptotic, like you say, Dale, these functions of cell turnover, it makes sense to me if you're breaking down muscle and then reforming muscle, which I would expect them to be doing. They saw that pattern in cluster one and cluster six in particular, which I thought was really compelling. And related to appetite was cluster five. And that was where it dropped and then stayed relatively low. And among those proteins was Leptin. And, Cindy, what can you tell me about Leptin? Isn't that involved with appetite as well? Instead of appetite stimulating isn't Leptin suppressing? Yeah, that's my understanding of Leptin. But you know, that's in isolation. So what is Leptin doing when it's interacting with all these other proteins? Right? I also remembered in that one, what I'd written down was MSTN, right? So another protein within that cluster, which I think had to do with musculogenesis, but I'd have to go pull it up again. And then going back to cluster two, which I skipped at the beginning on purpose. And this is anti- inflammatory proteins, right? Like, was it IL-10, where it starts low, goes a little lower, and then spikes up after twelve weeks. And what does that imply? Where this high level inflammation that people who are obese generally have after twelve weeks, we see a marked improvement in terms of this high level of inflammatory activity, which I thought was really interesting and informative the very fact that the body is a system. All these particular clustered changes that we're monitoring over time as a direct result of intervention. So I think the high level message is keep exercising. Well, exercise has so many benefits other than just weight. Right? I mean, it's got clear benefits to your brain, to depression. I mean, it's one of the most important interventions that you can do in psychological, many psychiatric cohorts. So it's the most powerful intervention we all have at our fingertips, right? Yes. And this one is just a really interesting experiment where they're able to just change one variable, which is the amount of exercise. Although from zero to three times a week, I might want to suggest maybe start with one time a week, 30 minutes. What's interesting, too, because they had to go to a particular place right there in Hong Kong, an exercise physiology center, where they had monitors and all this, and encouragement and coaches, they basically say they were encouraged. They use that language, right? To do this, to do this, do this. And they don't give any more details as far as the compliance. What did they use? I don't know. What kind of rewards? I don't know. How much do they pay these volunteers? I don't know. Well, I think the really interesting take-home message is, to dance through several figures on volcano plots and significantly up and down regulated proteins, was Figure six about the differential changes being able to distinguish between responders and non- responders. And that has a potential clinical kind of implications. Sarantis, you want to comment on Figure 6? Yeah, I think not just in this figure, but among these 23 proteins that significantly change in responders versus non-responders, I think they also mentioned that there are proteins like TFF-2 that regulate mucosal gastro- intestinal immunity. And this could be a direction through the interconnection with gut macrobiome. And they don't have really direct data to show this, but I understand the fact that identifying regulating genes like TFF-2, you may indirectly influence the expression of immune response genes in mucosa. And with that reason, you can regulate, for example, gut microbiome. That's really important also to your metabolism. It's really important when you have to deal with exercise and diabetes. And you think of the gut microbiome as a part of your system, right? Yeah. Here it is, we're intervening with exercise. An extra organ. Exactly. We're intervening with exercise. And the gut microbiome is changing, and they're unpicking some of the biology using this TFF-2. And then they use several other molecules, which I thought was pretty deep. I'm not an expert on the microbiome, but they talked about a prior study where they looked at exercise intervention. This is so important. I'm so glad you brought this up, Sarantis. This same team out of Dr. Zhu's lab found a role for gut microbiota in conferring the metabolic benefits of exercise. In other words, mediating they saw a signature in responders of exercise that was specific in this gut microbiota, which would suggest and they talk about this in this paper that there might be an opportunity in responders versus non- responders to do an intervention with those that you predict to be non- responders. Do an intervention, see if you can't nudge the microbiome in a direction that might make the body more responsive to the metabolic benefits of exercise, which I thought was super compelling. I mean, it's not easy to nudge the microbiome, but it is possible. And so bringing these two bodies of work together in a preliminary way and think about what they might do in the future, I think is super fun, super exciting. Yeah, it's a system. The particulars on the mechanism gets super complicated in terms of exactly how the effects of the immune system in the gut microbiome, exactly how that mechanism works. I'm sure that's sort of an area of active interest. But to think that to be able to get a say, all right, you group here with these of benefiting from high-intensity exercise. And then, I guess, thinking about it now ... well, there are other benefits anyway. But then the endpoint was pre- diabetes. The whole endpoint was, how do we lower that risk of diabetes? Such a big health care burden, right? Yeah. Any concluding comments from either of you? Sarantis? Cindy? Sarantis, I'll let you go first if you had closing ideas. I think, first of all, the use of a few biomarkers like machine learning algorithm, it's again and again coming to our attention. And I think it will really help a lot of diagnoses for a lot of diseases from now on. And, of course, the exercise is, at the end, the best medicine for a lot of things, right? There you go. That should be our final statement right there. I would just double-click on that. Exercise is the best medicine. This particular paper in Cell Reports Medicine was published just in February 2023. The title is, "A machine learning algorithm integrating baseline serum proteomic signatures predicts exercise responsiveness in overweight males with prediabetes." Well, thank you for joining us today. See you next time. Thank you for listening to the Proteomics in Proximity podcast brought to you by Olink Proteomics. To contact the hosts or for further information, simply email info@olink.com