The complexity of ‘omics offers challenges and opportunities. That’s particularly true of proteomics and metabolomics. About 20,000 proteins make up the encoded human proteome, and post-translational modifications add many more. The human metabolome—the intermediates or end products of reactions—consists of more than 100,000 small molecules.
The makeup of any area of ’omics can vary in time and tissue. The metabolome in particular is very dynamic, because metabolites get created, degraded, absorbed, modified, transported and more.
Although the depth and dynamics of proteomics and metabolomics make them complicated to study, these biological entities offer endless opportunities to better understand basic biology and to improve the accuracy of precision medicine.
Proteomics offers many tools for improving healthcare. “The biggest opportunities are using proteomics from plasma and tissue biopsies to discover biomarkers that can be related to the early detection of disease, disease severity, and to direct personalized therapies,” says Gary Kruppa, Ph.D., vice president of proteomics at Bruker Daltonics.
Those opportunities, however, come side by side with challenges. Although the amount of a particular protein in plasma from person to person is remarkably similar, wide variations from the biological norm can occur as the result of a disease state. “The dynamic range of the proteome in plasma spans 11 orders of magnitude and in tissue more than 5,” Kruppa explains. “This makes it difficult to detect low-abundance proteins that may be the most clinically relevant biomarkers.”
Even when scientists identify a potential biomarker from the tens of thousands of forms of proteins, much more work remains before that biomarker can be used clinically. “Hundreds or thousands of clinical research samples must be run to validate the clinical utility of a biomarker for early detection of disease, disease severity or as a biomarker for personalized therapy,” Kruppa says. “To characterize the proteome in depth, strategies that use methods like depletion to eliminate high-abundance proteins and extensive fractionation have been developed, but it can then take several hours to a day to analyze a single patient sample, making that incompatible with the number of samples that need to be run for validation.”
Running so many samples creates a challenge in the data analysis. For this, Kruppa says that his company is developing “a high speed, GPU-based search engine that analyzes the results in real time—identifying and quantifying proteins as the experiment is done.”
New methods of measurement
Scientists often use mass spectrometry (MS) to discover biomarkers. To work with enough proteins, scientists need more throughput from MS-based methods. These methods should also be easier to run for this technology to be used in the clinical setting to improve precision healthcare. Those technological advances already exist—to some degree.
“Mass spectrometry continues to get faster and more sensitive,” says Kruppa. As an example, he notes that Bruker’s MS-based technology “can now routinely detect and quantify several hundred proteins using a method that takes about 10 minutes per sample, meaning more than 100 samples can be run per day, making clinical research cohorts of thousands of samples practical for the first time.” Plus, the technology works with untreated plasma.
Scientists also continue to develop new MS-based methods. For instance, chemist Richard Smith, Ph.D., director of proteome research at the Pacific Northwest National Laboratory, developed Structures for Lossless Ion Manipulations (SLIM), which can increase the throughput, decrease the required sample size, and improve the sensitivity of MS. MOBILion Systems has now used this technology to create its SLIM-based high-resolution ion mobility mass spectrometry (HRIM-MS) platform. “Ion-mobility separation has been available for decades,” says MOBILion Systems CEO Melissa Sherman, Ph.D., “but, until recently, there were always challenges around sensitivity, resolution, and throughput.”
To get the most complete analysis of a sample, scientists need to separate the components. Sherman says that “the heart of the SLIM system is a serpentine ion path that enables us to separate molecules that have very minor differences, making the analysis very sensitive and providing high resolution power.” Plus, doing that in a gas phase, instead of liquid, makes this method faster than MS-based systems that separate samples with liquid chromatography (LC). As an example, Sherman says that a peptide map that takes 60–90 minutes to make with LC-MS can be completed in five minutes with SLIM-based MS and still provides “greater post-translational modification elucidation.”
Probing proteins more deeply
In some cases, very specific proteins—often at very low levels—provide clues about a patient’s health. These proteins can even provide information about a patient’s future health. At the Mayo Clinic, for example, clinical biochemist David Murray, M.D., Ph.D., and his colleagues developed MASS-FIX, which uses matrix-assisted laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF MS) to detect monoclonal proteins. Plasma cells produce these antibodies, which appear in the blood or urine of people with multiple myeloma or plasma-cell disorders, such as smoldering multiple myeloma or light-chain amyloidosis.
So, MASS-FIX “is used to screen patients for plasma-cell disorders and monitor patients being treated for multiple myeloma,” Murray explains. “This screening allows for detection of smaller levels of the disease at earlier stages.”
This is the first big advance in analyzing multiple myeloma samples in decades. In fact, Murray says, “MASS-FIX marks the first major breakthrough in multiple myeloma screening since gel electrophoresis was developed in 1967.” The key to MASS-FIX comes from analyzing monoclonal proteins. This approach, Murray says, “overcomes electrophoresis’s limitations in detection and provides the most accurate understanding of a patient’s monoclonal proteins.” An oncologist could use that information to assess a patient’s risk of progressing to multiple myeloma or light-chain amyloidosis. “This level of insight is not possible with traditional testing methods” Murray notes. “Results from this approach allow providers to put patients on a path to individualized treatment.”
A patient’s path to improved health could also arise from protein sequencing. That’s the latest objective of Jonathan Rothberg, Ph.D., who is a chemical engineer, serial biotechnology entrepreneur and a crucial pioneer in high-throughput gene sequencing. In 2020, Rothberg’s Quantum-Si introduced a next-generation protein-sequencing platform. This platform consists of instruments for sample preparation, protein sequencing and data analysis. Like the transition of DNA sequencing from analog to digital, Rothberg’s technology pushes proteomics into an electronic chip-based world. Only time will tell if the transition will provide as much improvement in proteomic data as it did with genomics, but the technology looks promising.
Metabolomics already informs clinical approaches, but that impact is really in its infancy. As one team of scientists from international locations—North America to the Middle East—reported in a paper published in 2016 entitled “Metabolomics enables precision medicine: ‘A White Paper, Community Perspective,’” metabolomics “has the potential to have a profound impact upon medical practice.” This research team added that “a person’s metabolic state provides a close representation of that individual’s overall health status.”
A collection of factors—genes, diet, environment and more—generate a person’s metabolome. By measuring the metabolome, a clinician obtains data on a patient’s biochemical state that cannot be acquired in other ways, such as genome sequencing.
Metabolomics proves especially useful in addressing heterogeneous conditions, such as autism spectrum disorder (ASD). Elizabeth Donley, CEO of Stemina Biomarker Discovery, and her colleagues ran what she describes as the “largest clinical study of the metabolism of children with autism.” This research team turned to the metabolome because genes only explain about 20% of autism, Donley says.
With 90 plasma samples from young children, 18–48 months old, the team analyzed the samples with several forms of LC-MS. After collecting the data on the metabolites, the analysis requires sophisticated tools. Scientists at Stemina use proprietary software and a laboratory information management (LIMS) system, plus tools for big-data analysis. The researchers also make use of publicly available resources, including the Kyoto Encyclopedia of Genes and Genomes (KEGG) and the Human Metabolome Database (HMDB). “There’s lots of noise in the system,” Donley says. “Our developers mask some of that to, for example, eliminate things that are contaminants, like plasticizers.” Then, machine-learning tools identify the important biomarkers associated with a particular disorder like autism, she says, “and links things in a systems-biology approach to look at pathways.”
Using quantities of metabolites or ratios of related metabolites in the modeling, Donley and her colleagues determined 34 metabotypes, which they defined as subpopulations “of individuals with a shared metabolic characteristic or phenotype that can be distinguished from the larger population.” Analysis showed that this method found more than 50% of the cases of ASD in this group. Perhaps most importantly, this tool can predict the risk of an autism diagnosis as early as 18 months, which is much faster than the average of 4 years in the U.S. The combination of subtyping ASD based on the underlying biology of the child and diagnosing it sooner could lead to better treatments for specific children.
Beyond ASD, Donley sees other applications of her company’s metabolomics tools in other neurological disorders. A study of bipolar disorder, for example, is underway.
Donley believes that many brain disorders consist of subtypes. Similar to the evolution in thinking about cancer diagnosis and treatment—50 years ago it was thought of as one disease and now we know that it’s many—she sees “subsets of disorders under broad names like anxiety and depression, and the underlying biology in these disorders is not all the same.”
Adding other ’omics
With the breadth of proteomics and metabolomics, scientists face hundreds of thousands of entities to test, but some companies create technologies to analyze even more areas of ’omics in combination.
“Having a clear, multi-dimensional view of ’omics information—from genomics and transcriptomics, through proteomics, cytomics and metabolomics—allows the testing and confirmation of hypotheses of disease stage and severity and which, if any, pharmaceutical intervention could impact the disease,” explains Marilyn Matz, CEO and cofounder of Paradigm4. “With multi-omics, we are on the way to reach what for many is the ideal state of precision medicine, where relevant and reliable data informs clinical decisions, allowing each patient to be managed in the most appropriate way.”
To improve the process of working with data from different ’omics, Paradigm4 developed an app that can “integrate the multi-dimensional ’omics datasets that are now prevalent, and interrogate them in a rapid, scalable and cost-effective way,” explains Zachary Pitluk, Ph.D., vice president of life sciences and healthcare at Paradigm4.
This app works with Paradigm4’s computational database engine. Pitluk says that this combination of technologies “streamlines cross-study, scalable analyses and machine-learning across heterogeneous datasets.”
To expand the range of scientists that can use this kind of analysis, Paradigm4 keeps it simple, because a user can interact with the app through what Pitluk calls “multi-omics aware graphic user interfaces and low-code solutions.” He adds that this “enables researchers to focus on framing the right questions using familiar vocabulary and ontologies, without getting bogged down in data wrangling, data modelling, or scalable elastic computing.”
Advanced analysis of multi-omics could be applied to many areas of healthcare, and scientists have been pursuing this for some time. In 2015, for example, a team of scientists from the bioinformatics and high-throughput analysis laboratory at the Seattle Children’s Research Institute explored the use of multi-omics approaches to ASD. As the scientists noted: “A broad shift toward molecular subtyping of disease using genetic and omics data has yielded successful results in cancer and other complex diseases.” For ASD, these researchers combined data on gene sequencing, gene and protein expression, psychometric testing and imaging to develop molecular subtypes. “The future of molecular subtyping for ASD and other complex diseases calls for an integrated resource to identify disease mechanisms, classify new patients, and inform effective treatment options,” the scientists wrote. “This in turn will empower and accelerate precision medicine and personalized healthcare.”
Making methods more accessible could be the key goal of today’s tools for proteomic, metabolomic and multi-omic analysis. In essence, these tools promise to democratize work in these fields, so that researchers and clinicians can assess a patient’s condition more completely. Hopefully, that ability will also spawn therapeutics that improve the treatment of some conditions and bring novel care to previously untreatable diseases and disorders. Ultimately, combining as much of this knowledge as possible will drive the biggest advances in healthcare.