July 22, 2025 | In a proof-of-concept study, a platform integrating nanomedicine with artificial intelligence (AI) and “actual causality” analysis succeeded in pinpointing a pair of proteins thought to play key roles in the development of metastatic prostate cancer and atherosclerosis. The technology is the basis of a biotech company launching today that will use the same approach to identify potential biomarkers and drugs for treating a wide range of catastrophic diseases, including cardiovascular and neurodegenerative diseases as well as different types of cancers, according to Morteza Mahmoudi, Ph.D., associate professor in the department of radiology and the precision health program in the Michigan State University College of Human Medicine.
The startup—XProteome—aims to “significantly reduce the cost of clinical trials and drug design and accelerate the development of diagnostic techniques and drugs so patients can get the benefit of those things as early and as cheaply as possible,” he says. The nanomedicine concept being employed here is known as the “protein corona,” the layer of surface-bound proteins on a nanoparticle that gives it a distinctive biological identity.
“How cells respond to nanoparticles depends on the decoration of that protein corona in terms of the conformation, and the amount and type of proteins,” explains Mahmoudi, noting that his team was the first to introduce the concept of personalized protein coronas more than a decade ago (Biomaterials Science, DOI: 10.1039/C4BM00131A). The real novelty with the latest study is the combination of the three R&D tactics, since the AI tool (LASSO regression technique used for many years in classification models) and causality analysis method (commonly employed in cybersecurity applications) have also both been around a while.
Using 35 human plasma samples, researchers first used LASSO—short for Least Absolute Shrinkage and Selection Operator—to identify 22 proteins strongly associated with either metastatic prostate cancer or atherosclerosis. But these were only correlations, which can oftentimes be misleading, he notes.
So, they next applied actual causality analysis to narrow the list of 22 down to two potential cardio-oncology biomarkers, CCT7 (Chaperonin Containing TCP1 Subunit 7) and TBXAS1 (Thromboxane-A Synthase 1). TBXAS1 emerged as the primary causal factor. Study results were published recently in Chemical Engineering Journal (DOI: 10.1016/j.cej.2025.161134).
In another study that published only days later, researchers in the UK showed how aspirin could help prevent cancer from metastasizing, the reasoning being that it acted against TBXAS1 (Nature, DOI: 10.1038/s41586-025-08626-7).
The biological clues under scrutiny are rare types of proteins and other biomolecules, such as lipids and metabolites, secreted into the bloodstream by disease-affected cells, says Mahmoudi.
In proteomics, mass spectrometry is the leading technique for protein identification and quantification. However, it often struggles to identify low-abundance plasma proteins because of the overwhelming presence of highly abundant proteins, he says. Due to the large dynamic concentration range of proteins in a sample and the limited detection range of mass spectrometers, “most of the low-abundance proteins [roughly one percent of the total protein mass] go to noise.”
Researchers in the protein corona field have been actively working to address the shortcomings of mass spectrometry-based protein analysis for close to two decades now, notably by characterizing the protein coating that forms around nanoparticles used as therapeutic drug agents. They’re seeking to address one of the major obstacles in nanomedicine—how to replicate success in the lab to the in vivo condition where cells encounter a different, protein-coated version of nanoparticles, Mahmoudi says.
By virtue of their exposure to biological fluid, in this case blood, the targeted nanoparticles never find the receptors whereby they would attach to the targeted cells to deliver their therapeutic cargo. This is primarily because the formation of a protein corona covers the nanoparticles' targeting elements. “My lab revealed that if you take identical nanoparticles and incubate them with human plasma from different people and different diseases, they create different protein coronas on the surface,” says Mahmoudi, getting back to the discovery of personalized, disease-specific protein coronas and their potential predictive value.
Mahmoudi built on his background in materials science and biomedical engineering with a Ph.D. in nanomedicine in 2009, when the field was in its infancy and there was a “big gap” between the impacted disciplines such as mass spectrometry technology, proteomics, and cardiovascular medicine. He consequently traveled extensively to acquire the necessary knowledge. These included educational trips to University College Dublin, the EPFL (public research university in Lausanne, Switzerland), the University of Illinois-Champagne, and Stanford School of Medicine before he became an assistant professor at Harvard University and subsequently accepted his current position at Michigan State University.
On a personal note, Mahmoudi says he grew up in war-torn Iran during the country’s armed conflict with Iraq in the 1980s. “My childhood was unfortunately occupied with harrowing events, [notably] air strikes and missile attacks,” he shares, but his mother succeeded in mitigating the potential trauma and encouraged him to devote his life to “making the world a better place for everyone in my capacity.”
That tale was told in 2018 when Mahmoudi was a biomedical investigator at Brigham Research Institute and won a BRIght Futures Prize. The competition’s $100,000 award supported development of a skin patch, made from multifunctional nanofibers mimicking the skin’s characteristics, to heal chronic wounds.
As Mahmoudi’s group and researchers elsewhere have demonstrated, protein coronas have “limited affinity to high-abundance proteins, as compared to many of the low-abundance proteins” so their proportion on the surface of nanoparticles is much less than the 99% in human plasma, he says. Depending on the chemical properties of the nanoparticles, the abundant proteins claim perhaps 5%, 10%, or 20% of the overall protein layer, giving the low-abundance proteins a new opportunity to be detected via mass spectrometry. They serve as a kind of “enrichment tool to improve the depth of the plasma protein coverage.”
The strategy commonly employed is “bottom-up proteomics,” whereby enzymes are used to cut down proteins in the protein corona to the peptide level before analyzing them with mass spectrometry, Mahmoudi says. The other major mass spectrometry strategy is “top-down proteomics,” which is when intact proteins are detached from the surface of the nanoparticles and analyzed in their entirety.
More accurate results are theoretically possible with top-down proteomics, but because it is the newer approach, the technique and its associated software that play critical roles in determining and interpreting depth of coverage are still immature, he says, as is covered in greater detail in a group perspective that published recently in Nature Protocols (DOI: 10.1038/s41596-025-01204-1). Bottom-up proteomics was therefore used for the proof-of-concept study.
Mahmoudi and his colleagues introduced top-down proteomics in the field of protein corona in a study that published last fall where the approach identified about 900 proteoforms, 48 of which provided biological clues missed by bottom-up proteomics (ACS Nano, DOI: 10.1021/acsnano.4c04675). In it, they propose using both strategies to obtain more comprehensive information about the protein corona.
Given the promising results from the proof-of-concept study, Mahmoudi and his team are now attempting to validate the CCT7 and TBXAS1 biomarkers and potentially identify additional biological signals on a cohort of over 1,000 plasma samples taken from healthy patients who went on to develop cardiovascular disease, prostate cancer (metastatic or otherwise), or both. Identifying the culprit proteins common to cardiovascular disease and cancer metastasis could help cardio-oncology researchers better understand the relationship between the often-co-occurring conditions.
It was only a few decades ago that researchers first hypothesized that a dependency exists between the two diseases, he adds. “But there is still a debate about whether they are interdependent or one of them is caused by the other.”
One of the overarching goals here is early disease detection to prevent complications, particularly in patients with comorbidities. It’s bad enough with either condition alone, says Mahmoudi.
When caught at an early stage, the five-year survival rate for many cancers is as high as 95%, but at later stages sometimes not even 5%, he points out. For breast and prostate cancers, most of the mortality is a result of cancer spreading from its original site to another part of the body. The same early-treatment advantage applies to cardiovascular diseases such as atherosclerosis, enabling available treatment strategies to prevent otherwise significant and inevitable damage to heart tissue.
Atherosclerosis and prostate cancer are now serving as model diseases. But the integrative platform is intended for broader use, including different types of heart disease and multiple cancers, says Mahmoudi.