In this session, Gordan Lauc (Genos Glyco & GlycanAge) and Vadim Gladyshev (Harvard), gave their point of view on biomarkers & aging clocks development. Gordan Lauc went through the interesting glycomic data they recently and insights it generated about aging and menopause, exceptional predictive capability of glycans for hypertension, and much more. Vadim Gladyshev then went through the approach to molecular signatures and biomarkers that they are employing to find and test interventions to extend lifespan. Part of it is also a new epigenetic clock called scAge functioning on a single cell basis. This clock enabled them to find when aging actually begins during embryonic development.


Presentation: Gordan Lauc

  • Glycans are the ultimate layer of cellular complexity and communication. And at the same time the most neglected molecules of cellular communication. This became obvious when the director of NIH wrote about the S glycoprotein without a single glycan depicted on its structure. Glycans are not just decorations, they actively participate, and are a functional part of the molecule. And a majority of proteins are glycosylated. So to have a complete picture of biology, we have to include glycans in what we are doing.
  • This is finally entering mainstream, even the Office for strategic coordination of NIH launched The Common Fund programme for glycoscience aimed to develop analytical technology for glycans in the US.
  • One of the extremely important molecules where glycans are important are the immunoglobulins. Glycans regulate the effective function of IgG, we could have either pro-inflammatory functions of IgG which we see in old people or anti-inflammatory functions of IgG which we see in younger people. So they are balancing inflammation and therefore immunoaging on many levels.
  • How can they affect so many different things and functions? Because of alternative glycosylation, so attaching different glycan to a specific glycosylation site. In a similar way we think about coding mutations. Coding mutations can change the structure of a protein, a bad mutation can kill you. And the same thing can happen to a glycan, if you have wrong glycan on a wrong glycosylation site, the protein can be completely nonfunctional. The key difference is that mutations are inherited in mendelian ways, while glycans are inherited as a complex trait.
  • Glycans integrate genetic, epigenetic and environmental components into creating a functional molecule that has a specific function
  • We started high throughput glycomics over a decade ago, we analyzed over 150,000 samples from some of the best genotype and phenotype cohorts in the world with collaboration with leading researchers and institutions, which enabled us to generate a lot of knowledge and data about glycans.
  • This was of course all enabled by multiple research grants and incentives.
  • Now we’re trying to translate this research into products which can be used – like GlycanAge. The first test we translated is a biological age test.
  • GlycanAge integrates genetic, epigenetic and environmental factors into a simple index that can be easily quantified.
  • Everything we do, we try hard to be based on a lot of research, using our history of over 200 papers on the topic, to show that we have really a lot of good evidence published in high impact journals supporting the claims we are making.
  • What can we do to move this clock backwards?
  • We have a lot of anecdotal evidence that some people have way better GlycanAge than others. And we even see patterns, for example for all patients in the same clinic in New York undergoing the same recommendations, they are 20-40 years younger measured by GlycanAge compared to chronological age. But this is not science, all anecdotes, so we are trying to figure out which of all the things they are doing are actually working for GlycanAge.
  • The first proper study that I really liked (placebo controlled, randomised trial) was an estrogen study, where we had 36 young women with chemically induced menopause, and there was an estrogen recovery group and also a placebo group that did not receive estrogen. And the placebo group aged on average 9 years compared to the estrogen group that did not age at all.
  • If you are wondering how this study could pass an IRB, we don’t know because it happened over 10 years ago for a completely different reason, and we just got left over samples for analysis. It was done in Colorado. But we are aware of the potential problems with chemical induced menopause data and we are doing a better study now with natural menopause and hormonal replacement therapy.
  • Women who were younger on their GlycanAge index before the chemical intervention actually aged much more rapidly than women who were already older based on GlycanAge. So it affected the biologically younger women much more, and we are now looking into the molecular pathways to understand how the estrogen signaling towards the IgG glycome.
  • Another extremely interesting story we have is based on this older paper showing that IgG glycans are causative of hypertension in mice. I was confused because we can see a similar situation in humans. We see a strong correlation between cardiovascular risk score and glycans. So we have kind of indirect confirmation of this completely unexpected finding. We then teamed up for a larger study which was published in Circulation 2 years ago, in which we were able to prevent hypertension in mice by feeding mice N-Acetylmannosamine, which is a supplement of sialic acid – a sugar that makes glycans look younger, which actually worked in mice. The mice who were fed with it did become obese but did not develop hypertension.
  • We then did a large human cohort, where we observed the correlation between hyposialyation (having older IgG glycome) and blood pressure in humans. And using that data we approached the EPIC cohort data (collected 30 years ago and followed up for a decade after), and its german subset of 27,000 people in it. In that sample 508 had a cardiovascular event.
  • And we looked at the glycans in the baseline, and glycans were as predictive as AHA score, and in women, just one glycan was more predictive than the AHA score. So it does seem that these IgG glycans contribute to future development of cardiovascular diseases.
  • So let’s try to fix it. What can we do? We know that obesity is associated with bad GlycanAge. So we looked into a cohort of patients with bariatric surgery. And indeed, undergoing bariatric surgery is very effective in rejuvenating your IgG glycome. Of course bariatric surgery is radical and not everybody wants to do it.
  • But even a regular weight loss seems to help. We were tracking 2000 twins and we observed that the twins who were gaining weight were aging much faster than the twins who were losing weight.
  • In another study (this one is unpublished), we have shown that there is no magic diet – no diet works for everybody. We had 5 different diets, 1000 people in each and in all diets there were people who improved and people who got worse. We have to choose what’s good for us specifically.
  • And the same goes for exercise. At the beginning we were thinking that with sufficient exercise people would get better, but we found out that overtraining is bad. When people overtrain too much, their GlycanAge gets worse. What seems to work is HIIT for immediate decrease of GlycanAge.
  • Glycans are functional effectors that regulate inflammation and associate with numerous diseases, and in many of them they change up to decades before the actual onset. We just published a huge review in the Comprehensive Glycoscience encyclopedia – it’s freely available here.
  • These things are extremely complicated, there are over 40 genes in the network regulating IgG glycosylation, many of them are known factors or inflammatory and auto-immune conditions.
  • When we look at people with different diseases, they actually make people look similarly older as old people when we look at them with the GlycanAge index.
  • The idea of this whole story is not to wait for people to become ill and get to hospital and develop into a chronic disease. But rather find the proinflammatory biomarker through screening (GlycanAge), and with some kind of intervention (lifestyle or pharmacological), we can fix things before they become too big of a problem to handle.
  • The way we see GlycanAge here is that it can motivate people to live healthier. We all know what is generally healthy, but we don’t do it because it is difficult and the reward comes only after several decades.
  • But if we have objective insight into what is going on today and see in a few weeks or months, what is the result of those interventions, it could help to motivate people and help intervene earlier.
  • GlycanAge is now available globally.


What is one thing that this group can do for you and what would help?

  • You can buy the test, which is available either through web shop, or a network of providers and partners.
  • We are also highly interested in collaborating with people either doing research or doing commercial partnership to expand our coverage (right now we are mainly in the UK, but are looking to expand into other countries, especially the US), or to just use it in their clinics or other anti-aging efforts.


Estrogen data: does that mean that hormones are not beneficial in women with already old glycan age? If so, could you use it for triaging in deciding who should take HRT?

  • So far we haven’t done many studies on actual menopause, but it seems that most women react to estrogen by reduction in GlycanAge. We are doing a study in one clinic in the UK, and are interested in more. We’ve seen drops in GlycanAge by a decade or more in a few years, so we are interested in improving glycanage.


I am involved in a project that involves Adaptive Immune Receptor Repertoire (AIRR) sequencing of autoimmune disease patients with antigen-stimulation of primary cells. Do you think IG glycation can be used together with immune repertoires?

  • It is challenging since it is not possible to analyze glycans on a single cell level yet, but we’re getting there. One of the key questions is when is the glycosylation defined. But it is not that much related to aging. It is a relevant immune question though.


What is the glycosylation status of therapeutic antibodies?

  • Some research was blocked earlier by NIH, now it is a big focus in biopharma. But we still don’t know which are the proper ones. Just the blocking antibody. It is a focus of the pharma industry.


In the long term, what are your thoughts on what will be the sampling frequency of tracking? Once a year, once a month?

  • What we do at the moment is for people with good glycanage, we recommend 1 year, for people doing interventions roughly 2 months, because they change in around a month or two usually. With some more acute infections like covid, it can change more quickly. We have to accumulate much more data before we can confidently say.

Presentation: Vadim Gladyshev

  • The reason I want to focus in this talk on biomarkers and signatures is because both are needed to test interventions in an unbiased way.
  • What do I mean by signatures? Signatures are molecular patterns, in our case particularly gene expression patterns, that indicate the potential of organisms to live longer. They are not comparing the young vs old, they are comparing the potential.
  • And then we combine all those signatures to find molecular patterns so we could indicate how we could alter the system, so the biological system lives longer.
  • In this example we analyze mammals, with a huge variety of maximum lifespans.
  • When we do gene expression analysis, we just look at how nature alters lifespan. And what we find is that central metabolism is downregulated in liver (more) and kidney (less) with lifespan. In the brain those are almost unaffected though. But there are some other factors that are upregulated in different issues. So it illustrates that it is very difficult to reproduce this experimentally, because we need to upregulate and downregulate not only in gene-specific manner, but also in organ-specific manner. But at the same time it tells us the direction, so if we can move the system in that direction, we expect the system to live longer.
  • Also when we compare different interventions in mice (this matrix shows similarity in molecular patterns), we see that many interventions change the system very similarly, similar pathways are altered. But there are also some other interventions which also extend lifespan that do it differently (for example Rapamycin is not similar to any other interventions). So there are apparently many ways to alter the system and expand lifespan. This is also needed for us to understand how we could combine different interventions so we could achieve maximum effect.
  • These patterns can be naturally used to identify new interventions. When we screened public gene expression data about interventions with our signatures, we saw that some interventions came up as really good candidates with high potential for lifespan extension (like hypoxia). We’ve also done some chemical screening for interventions that could be tested on mice in our lab.
  • Now a bit about biomarkers. Longevity signatures are needed to identify treatments in an unbiased way, but we need to test them rapidly, for which we need robust biomarkers (clocks). We focus mainly on mice clocks, because we are able to test interventions very quickly in them compared to humans. Our idea is to first to really understand in mice what is the best approach to extend lifespan, how can we combine interventions, and the most fundamental level of research and then choose just the best candidates and test them in humans.
  • At this point epigenetic clocks are still the main focus since they are most robust and accurate. These are examples of how we apply those clocks to measure interventions.
  • However we should keep in mind that the basic element of life is the cell, so it would be really important for us to see how cells age and identify the biological age of individual cells. If you have a control and an intervention group, and a particular subset of cells got 100x older but all other cells stayed the same, it would show in the normal bulk clock as overall aging, even though the bulk is just affected by a certain type of cells aging rapidly. We really need to understand how cells age, whether they all age or whether just a subset of them age. Some might even get younger. And see what is the difference between different populations of cells, and between different interventions. This is completely unknown, so we wanted to know that, and recently invested in the development of single cell clocks. There are many challenges of course. For example when we use the bulk clock based on bulk sequencing data, we have many reads. We can quantify the methylation fraction on each CpG site. But if we do single cell, it is completely different, since the coverage of reads are completely different. In bulk we rely on methylation fraction, in single cells we rely on binary data, many times data that are not read in each cell, so we cannot apply the same approach.
  • So we developed a completely different approach and different type of clock we call scAge – flexible and scalable probabilistic framework for epigenetic age predictions at single-cell resolution.
  • This is an example application of this clock. You can see scDNAm age in different cells (each dot) in different chronological ages of the mice. Cells mostly cluster around the chronological age, but you do see some variation, some cells are hugely increased in age, so we think those are senescent cells. So in general we see that individual cells age, it is not just a subset or a type of cells that age, but actually all cells in a tissue. The advantage of this method is that we can rely on a very small number of reads (literally around 15k-20k single reads) to identify the biological age of this.
  • Example of dataset for mouse muscle stem cells, what they found applying bulk clock is that wild muscle ages appropriately, but muscle stem cells seem to stay almost the same age in the young and old mouse. Which is strange, they should be aging, but based on this data they do not. So we looked at it through analysis of single cells, and we see the same thing. We see that muscle cells age a bit, but the difference is very small and the biological age is really low, so it confirms the initial hypothesis.
  • The picture that emerges from this work is that there are cells with different aging trajectories – some cells age more rapidly, and some less. In the end when we apply bulk clocks, it doesn’t capture this heterogeneity in biological age across different tissues. Our clock shows a lot of promise in helping with that, but needs to be developed more, it’s just the first version.
  • I just wanted to show one example which for me is the most exciting. Using this tool we identified a rejuvenation event during the embryonic development phase, followed by aging. You can actually see that biological age decreases during the development and then at a certain point aging starts. People in the field thought that biological age needs to be lowest at the complete beginning, but it seems not to be the case. It gets rejuvenated from the beginning so the biological age decreases for a while, and then reaches the lowest point (ground zero) approximately at the time of gastrulation, and from that point aging picks up and the biological age increases – we think life begins at this lowest point. We see this with different clocks as well. Of course our idea is to find the molecular mechanism of rejuvenation in the embryonic development, and see the molecular pattern, so we could try to induce such a pattern in the adult state in humans as well and rejuvenate them. This could hopefully be a different pattern than rejuvenation by Yamanaka factors for example, or maybe we could find common features among these two types of rejuvenation processes which would help us to develop treatments.
  • The single cell clock was developed by Alex Trapp, and this rejuvenation during embryonic development was investigated by Csaba Kerepesi. And the work is funded by NIH.


If you had the resources you want, what would you want to research, what would you need?

  • I would want smart people to join us, there are other ways to support other labs as well. In terms of science that could be done, in a long term view I see potential in rejuvenation technologies. The question is how it will be done – maybe Yamanaka based reprogramming, tissue replacement or cells, or embryonic rejuvenation. In the short term it would be exciting to prove that interventions work. I am a fan of TAME because it could prove the hypothesis and bring more people into the field.


In terms of markers of age, epigenetic markers are very useful for measuring normal aging. My question is that correlation is not causation, and addressing correlations might not have the desired effect. We know that many changes are compensatory and intervening in compensations might be actually harmful. I am not convinced that in many of the measurements of aging we are actually measuring that and not just a good correlative marker on normal aging.

  • This was the point that I skipped. I sometimes give an example – build a clock based on the use of glasses, hearing aid, etc. We can use that to predict age, but they are not causal because it is measuring compensatory and adaptive changes. For rejuvenation we would need different types of clocks. We need new types of clocks for that.



Seminar summary by Bolek Kerous.