Non-Invasive Biomarker Can Track and Verify Senolytic Drug Efficacy

Cellular senescence is a complex cellular process that involves the irreversible arrest of the cell cycle; in other words, cells permanently stop dividing. Senescent cells are not "dead"; their complex secretome – the senescence-associated secretory phenotype, or SASP – is implicated in biological processes such as inflammation and immune responses. An increasing amount of research is demonstrating that the pathophysiology of many age-related diseases are linked to cellular senescence. Consequently, a growing body of work is aimed at developing senolytic drugs, these pharmacological agents are capable of selectively eliminating senescent cells.

The lack of a biomarker to measure and track the efficacy of senolytic drugs has proven a hindrance to the research field thus far, says Professor Judith Campisi. The Campisi laboratory at the Buck Institute for Research on Aging studies the regulation and characteristics of cell states, with a particular emphasis on cellular senescence.

In a new study published in Cell Metabolism, Campisi and colleagues share their discovery of a novel, non-invasive biomarker test that can be utilized to measure and track the performance of senolytic drugs. The biomarker, a unique signaling lipid metabolite, is released when senescent cells are forced to die. Consequently, it can be collected and tested via non-invasive mechanisms such as urine collection or blood sampling. 

Technology Networks interviewed Campisi and Christopher Wiley, lead scientist on this project and researcher at the Jean Mayer USDA Human Nutrition Research Center on Aging, to learn more about this research and its clinical significance. 

Molly Campbell (MC): Why it is important to be able to measure and track the performance of senolytic drugs?

Judith Campisi (JC): Most senolytic drugs are first tested in mice. Humans live 30-35 times longer than mice (roughly three vs 100 years). So, if a senolytic takes several weeks or months to improve a pathology in mice, it could take several years to improve that pathology in humans. But how would we know whether the drug even killed senescent cells in living humans? A biomarker that is easily detectable in plasma or urine would allow rapid assessment of whether a senolytic was even active in humans.

Christopher Wiley (CW): Given the number of diseases that senolytic drugs could target, and the growing number of senolytic drugs, we really need a non-invasive standard candle for senolysis – something people can look at and agree that senolysis is taking place in people. While we are not there yet (we have not tested in people), our results are promising.

MC: There are a growing number of age-related diseases that are linked to cellular senescence. Can you provide examples of these disease areas?

JC: The list is long! It includes Alzheimer’s disease and other forms of cognitive decline, several aspects of cardiovascular disease, cancer metastasis, cataracts, diabetic complications, lung fibrosis, osteoarthritis, osteoporosis, several side effects of certain drugs used to treat cancer and HIV-AIDS, sarcopenia -- and others.

CW: Add to that Parkinson’s disease, some forms of diabetes and insulin resistance, fatty liver disease, hair loss, intervertebral disc degeneration and retinopathy. The list has started to grow so rapidly that it is becoming difficult to keep up with the rapid pace of new diseases believed to be driven by senescence.

MC: What key methods did you adopt in this study that enabled the discovery of the biomarker?

JC: Our methods include human cells in culture, transgenic mouse models, whole transcriptome sequencing and mass spectrometry.

CW: We utilized a mass spectrometry platform to measure lipids in senescent cells. We initially detected a number of prostaglandins, so we expanded our search to all known prostaglandins and putative prostaglandins. Dihomo-15d-PGJ2 was a hypothetical lipid at the time – and has since only been reported as part of a list of lipids in one additional manuscript, with no validation. We therefore validated it using a synthesized dihomo-15d-PGJ2 standard.

Most prostaglandins are thought to be secreted factors, but we only detected dihomo-15d-PGJ2 inside of senescent cells, but not in the culture media. This was the key observation that suggested to us that it might be useful as a biomarker of senolysis. Since dihomo-15d-PGJ2 is small (about 344 Da), we postulated that it would be likely to escape from senescent cells when they died. We tested this by giving control and senescent cells a very high dose of a senolytic – high enough that it also killed the control cells. Only the senescent cells plus the senolytic resulted in release of dihomo-15d-PGJ2. Importantly, we reproduced this in a mouse model of chemotherapy-induced senescence, showing that the biomarker is specific for senolysis in mammals.

There is also a commercially available ELISA for 15d-PGJ2, but we found that it does not distinguish between 15d-PGJ2 (a related prostaglandin with 2 fewer carbons that is better-studied) and dihomo-15d-PGJ2. Still, it worked very well for detection of senolysis in our models and could theoretically be used in lieu of mass spectrometry for a quicker test in the clinic.

Laura Lansdowne (LL): Please can you talk to us about the biomarker dihomo-15d-PGJ2? How did you demonstrate that it is unique to senescent cells?

CW: Dihomo-15d-PGJ2 accumulates inside of senescent cells to micromolar levels – which is extremely abundant for an oxylipin. It is virtually unstudied and was essentially a hypothetical metabolite until we detected it inside senescent cells. While it may occur in other biological contexts, it was so abundant in senescent cells that we hypothesized that it would be selective for senolysis, when a bolus of the lipid would be released as senescent cells die.

Once we had a dihomo-15d-PGJ2 standard, we assayed multiple cell types, including immune cells and activated platelets. Macrophages did have low levels of the lipid, but they did not increase during M1 or M2 polarization/activation and never reached levels observed inside of senescent cells. In mice, we only observed dihomo-15d-PGJ2 when mice with large numbers of senescent cells (we showed this for the chemotherapeutic drug doxorubicin in the paper) were given a senolytic.

Most importantly, dihomo-15d-PGJ2 was not detectable in any condition we studied (e.g. chemotherapy, aging, antiretroviral therapy) unless we treated with a senolytic. While we exclusively used ABT-263 in the manuscript, we have since validated using transgenic models of senolysis and unpublished senolytics discovered at Buck – and we only detect dihomo-15d-PGJ2 during senolysis.

LL: Why have lipid components of the SASP been vastly understudied?

JC: Enormous progress has been made in characterizing the transcriptomes and proteomes of cells. So many recent studies have focused on these cellular components. However, there is now rising interest in cellular metabolites, including lipids.

CW: The Campisi lab initially characterized the SASP using protein biomarkers, and lipids in general tend to be understudied. Part of this is likely due to fewer tools in the space, and the available tools like mass spectrometry tend to be more expensive. By comparison, transcriptomics is pretty standard and relatively cheap – and many RNAs encode proteins. Most antibodies are raised to proteins, so the tools for studying proteins are more accessible.

By comparison, RNAs do not code for lipids, and very few lipids can be assayed by antibodies – so you really need to use tools like mass spectrometry, which are more expensive and require additional expertise to study. Development of new technologies, such as metabolomics and lipidomics, is changing this, so I expect we will see similar discoveries in the future.

MC: Why is a biomarker that can be measured via non-invasive testing advantageous?

JC: If you, or your physician, wanted to know if a drug or intervention was working, would you rather urinate in a cup or have a foreign object inserted into your body?

CW: With senolytics in clinical trials for a large number of conditions, there are two key factors that these trials need to establish:

1. That the drugs are effective in treating the disease.

2. That senolysis is occurring.

Currently, the only way to establish that senolysis is occurring is to extract tissue before and after treatment and test the tissue for markers of senescence. For example, fat and skin biopsies were taken before and after senolytic treatment in one of the first trials that establish that a senolytic works in humans. This biomarker allowed us to detect senolysis in blood and urine from senolytic-treated mice as it was occurring. So if you want to detect senolysis without conducting biopsies, this is the way to go. 

LL: What are your next steps in this space, and what clinical impact do you envision this work may have?

JC: We are focusing now on better understanding what these senescence-associated lipids do, both inside and outside the senescent cell. We hope to determine whether and how senescence-associated lipids interact with – or affect the function of – other cellular components (e.g., proteins, small metabolites).

CW: First, after we found this marker, we went back and looked at our untargeted lipidomic profiling, and there were 4-5 other candidate lipids that might also be useful as biomarkers. We are looking at dihomo-15d-PGJ2 as proof-of-principle for discovering other biomarkers.

Next, we want to validate our biomarkers in humans and non-human primates. This is the most important step in moving toward a commonly-used diagnostic test for senolysis in patients.

For clinical impact, we envision two major uses for this biomarker:

1. Showing that the drug is working when people are being treated with senolytic for diseases driven by senescence.

2. As a companion assay for clinical trials using new senolytics. Our hope is that we will be able to have dihomo-15d-PGJ2 detection become the standard for senolytic testing so as to avoid those invasive biopsies and tissue samples, and instead use biological fluids. We showed blood and urine in the paper, but one could imagine tears, cerebrospinal fluid, interstitial fluid, etc. as potentially assayable.

For example, if you wanted to treat Alzheimer’s’ disease with senolytics, you are not going biopsy a person’s brain to assay for senescent cell removal, but you might conduct a spinal tap and assay for dihomo-15d-PGJ2.

                  

Judith Campisi and Christopher Wiley were speaking to Molly Campbell, Science Writer, and Laura Lansdowne, Managing Editor for Technology Networks0