Discovering how individual protein levels vary among people with diabetes could revolutionise treatment options. Although most diabetes medications target cellular proteins, scientists are only beginning to understand the significant role these proteins play in disease manifestations. New research examines protein types and quantities to predict disease pathways and customise treatments, potentially offering more precise drug interventions. This approach, once primarily used in cancer treatment, may soon provide groundbreaking insights into diabetes care.
Most drugs approved to treat diabetes act on proteins, the biological machinery of the cell. But we know remarkably little about how the levels of these proteins might differ from individual to individual and what these differences might mean, according to Atul Shahaji Deshmukh, an associate professor at the University of Copenhagen who studies the relationship between proteins and metabolic conditions.
Diabetes is a multi-organ system disease that can manifest in very different ways. Some people with diabetes face retinopathy, which degrades vision and can lead to blindness; for others, the kidney is hit hard and the eyes are left unscathed. Deshmukh, who studies the proteome – the balance and relative proportions of proteins in our bodies – thinks that these tiny molecular machines may hold the key to why people with diabetes experience such divergent disease pathways.
Deshmukh sees proteomics as a new frontier in precision diabetes medicine. Protein censuses could help to identify the complications for which a person newly diagnosed with diabetes has the highest risk and enable doctors to tailor their treatment accordingly.
A new article by Deshmukh and colleagues, published in February in Diabetologia, lays out the promise and challenges of incorporating proteomics into precision diabetes medicine.
“This technology is constantly evolving,” Deshmukh says, and has real capacity to improve patient outcomes.
From cancer to diabetes
Studying proteomics is something like being a census-taker, explains lead author Nigel Kurgan, a postdoctoral researcher at the University of Copenhagen.
Using blood or tissue, proteomics researchers measure the types of proteins in the sample and how many are present, most often through mass spectrometry, a technique that helps scientists to identify proteins by their weight. “We can quantify thousands of proteins simultaneously,” Deshmukh says. That volume is essential, because although “we have an estimated 20,000 genes, proteins could take millions of different forms,” he adds.
But why bother with a protein headcount at all when our genetic code can be extracted from a single cell? Kurgan says that while DNA is stable, proteins are dynamic – the types and proportions of proteins present appear to be responsive to changes, such as treatment, diet or exercise.
Proteomics has proven its clinical value in cancer treatments, Kurgan says. Various patterns of proteins have been associated with early-stage development of specific types of cancer, resistance to certain types of chemotherapy and even a patient’s prospects for survival.
Researchers hope that learning more about how protein signatures are associated with various subcategories of diabetes will better enable clinicians to predict what path the disease might take, personalise treatment strategies and, thanks to the dynamic nature of proteins, even assess how well interventions are working.
A needle in a haystack
But since tissues differ vastly in protein composition, where in the body should diabetes researchers conduct these censuses? All body samples are not created equal from a proteomics perspective, Kurgan emphasises.
Blood has distinct advantages – it is easy to access, is often already collected during hospitalisation or diabetes treatment and is stored in biobanks around the world.
But looking for informative proteins in a blood sample is like looking for a needle in a haystack, says co-author Jeppe Kjærgaard Larsen, a PhD student studying muscle metabolism.
The 20 most abundant proteins comprise about 99% of the proteins in blood plasma – albumin alone accounts for 60% – so proteomics researchers must laboriously sift through that last 1%. “Current methods for reliably detecting low-abundance proteins in plasma are very challenging, and these are the ones we use for biomarker studies,” Larsen explains.
And ultimately, “the proteins you identify in the plasma have leaked from the organs and tissues” you are really trying to study, Kurgan says. So why not take samples from the organs themselves?
Although tissue samples are a richer source for proteomics researchers (and with exponentially less albumin to wade through), tissue sampling is invasive, painful for patients and difficult to perform at large scale.
“I think one reason why the cancer field is ahead in proteomics is because they always take tumour biopsies,” Deshmukh says, giving them a uniquely deep bench of tissue samples to study. Cancer researchers also have the opportunity to compare healthy tissue with tumour tissue for a single patient, meaning that differences in protein levels are easier to pin down. Meanwhile, researchers in diabetes proteomics need vast sample sizes to detect similar trends.
What is next for proteomics in precision diabetes medicine?
Deshmukh emphasises that for proteomics to make advances in diabetes urgently requires larger and more diverse samples – which means “democratising” the technology, he says.
Currently, “access to this technology is really limited to a certain group of people,” namely wealthy research institutions in the West, Deshmukh explains. People of European ancestry are vastly over-represented in existing proteomics data sets, which leaves researchers in the dark about the variation that might exist in the non-white majority of the global population.
And proteomics researchers studying diabetes need to look beyond just blood samples and begin developing tissue biobanks, the study authors agree. Today, “I think it is probably only Scandinavia where people tend to collect adipose biopsies and muscle biopsies,” Deshmukh says. As protein identification technologies become more sensitive, low amount of samples – like those from small needle biopsies, which are less invasive and less painful – are becoming viable for proteomics research.
While Deshmukh thinks that mass spectrometry, the system he and his team primarily use, should remain the workhorse of proteomics in the near term, Kurgan says that emerging technologies could make detecting low-concentration proteins easier. Affinity-based approaches use specially designed capturing molecules to target and count only the proteins of interest rather than wading through and identifying the whole field, he explains.
Deshmukh underscores that, for diabetes research, proteomics is not an either/or proposition in competition with genetics, transcriptomics and metabolomics. Patients are best served by a multiomics approach, with experts in these fields all having a seat at the table for planning treatments. “The end goal is to give better treatments and prognoses to people with diabetes,” Kurgan agrees.
