Time for Proteomics To Shine
Complete the form below to unlock access to ALL audio articles.
This article includes research findings that are yet to be peer-reviewed. Results are therefore regarded as preliminary and should be interpreted as such. Find out about the role of the peer review process in research here. For further information, please contact the cited source.
Over the last 20 years, advances in next-generation sequencing have transformed genomics research, enabling scientists to sequence a human genome in a day, for less than $1000. Now, technological advances are edging ever closer to achieving the same for proteomics, with the $1000 proteome in sight. Studying the proteome can reveal valuable functional information unattainable from genomics, providing rich insights into health and disease. Several recent studies have highlighted the value proteomics can bring to drug discovery and our understanding of diseases such as cancer. Improvements in the capabilities of proteomic technologies are putting proteomics in the spotlight and opening up a myriad of applications across the life sciences.
Technology Networks had the pleasure of interviewing Dr. Oliver Rinner, CEO at Biognosys, to learn more about some of the recent developments in proteomics, and discuss what the future could hold for the field. Rinner also highlighted the role that advances in mass spectrometry (MS), including Limited Proteolysis Mass Spectrometry (LiP-MS) technology, are playing in driving progress in proteomics.
Anna MacDonald (AM): In recent opinion articles, you comment that “After being in the shadow of genomics for the past decade, proteomics is now ready to shine,” and that “progress in MS technology is driving a rapid scale-up in capability.” Why has more focus not been given to proteomics up to now, and what has spurred this shift to move beyond the genome?
Oliver Rinner (OR): I was part of a genome annotation project myself as a student, and I still remember the excitement of discovering and annotating new genes every day. The ability to read the code of life while at the same time having tools that develop with breathtaking pace in scale and cost-efficacy are probably the reason why we almost forgot about protein science for a while.
But we had to learn that the connection between genotype and phenotype is far from straightforward. Knowing that a gene is transcribed does not tell us how the resulting protein is expressed or organized into a functional state that ultimately drives a phenotype. The proteotype, or the organization and functional relationships of all proteins in a specific tissue in a particular state, connects the gap between genotype and phenotype. Without protein-level data, this critical link is missed.
Proteomics is the technology that aims to characterize the proteotype on a large scale. Until recently, proteomics has been viewed as a pricey and complex technology that can't compete with genomics' high throughput and low cost. However, recent advances in large-scale mass spectrometry proteomics, pioneered by our team at Biognosys, are changing the picture. The $1000 proteome – or 1 cent per data point – is now a reality, delivering an unprecedented depth and information content of quantitative proteomic data.
AM: There are several technologies available for large-scale proteomics. Why do you believe mass spectrometry is the most scalable?
OR: MS is a physical technology that generates data by breaking proteins into smaller peptide fragments and then identifying and quantifying them according to their mass and charge. Alternative technologies – affinity-based methods in particular – by contrast rely on detecting specific proteins with a panel of antibodies or aptamers.
The fundamental difference in analytical outcome between these approaches is that the mass spectrometer can be run in a true discovery mode and provide peptide- or even amino acid-level resolution, which affinity-based methods cannot. True discovery means that MS does not rely on affinity recognition, availability of antibodies or aptamers, and conformational modifications and aggregation of proteins do not interfere with the analysis. Therefore, it can also be hypothesis-free because the technique allows to explore the entire proteome and detect the mutations (e.g., different proteoforms such as post-translational modifications (PTMs)) and perturbations of the proteotype.
It is also more scalable because when using affinity proteomics, detecting more proteins per sample requires more antibodies/aptamers to be added to the detection panel, increasing the cost. As a result, as the number of data points captured per experiment increases, the cost per run rises proportionally. On the other hand, mass spectrometry instrumentation and workflows are continuously improving without substantial cost increases, which means that the cost per data point keeps going down in a similar way like the cost per base pair in next-generation sequencing continues to fall.
AM: Are there any recently published Biognosys studies that you would like to highlight?
OR: Contributing high-quality science to advance the proteomics field has always been at the heart of Biognosys' mission. In 2021 alone, 500 publications mentioning Biognosys technology and tools came out, adding up to 2,000 publications since our inception.
The first publication I want to put in the spotlight is “Mechanistic Insights into a CDK9 Inhibitor Via Orthogonal Proteomics Methods” recently published on ACS Chemical Biology. The study has been conducted in collaboration with AstraZeneca and Pelago Bioscience AG and highlights how orthogonal proteomics approaches can be applied to profile the selectivity of new compounds. Our TrueTarget™ platform, powered by our proprietary LiP-MS technology, was used to screen the entire proteome and identify potential targets via structural alterations. In addition to target identification, by exploiting the technology's peptide-level resolution, a unique feature of the approach, we could also identify the putative binding site of the CDK inhibitor. In this way, LiP-MS enabled the analyzed compound's target identification, binding affinity estimation and binding site localization.
Also, our publication on bioRxiv “Biomarker Candidates for Tumors Identified from Deep-Profiled Plasma Stem Predominantly from the Low Abundant Area” on our ultra-deep plasma proteomics workflow and its utility for oncology biomarker discovery is a highlight. The publication presents a novel automated analytical approach for deep plasma profiling and applies it to a 180-sample cohort of human plasma from lung, breast, colorectal, pancreatic and prostate cancer. It resulted in the identification of 190 targets for FDA-approved drugs targets, 66% of which were detected in the lower intensity range with the addition of many novel possible targets. Moreover, the final dataset recapitulates biological features of intra-patient heterogeneity. A comprehensive case study of the publication can be found on our website and the workflow is available via our TrueDiscovery™ platform.
AM: What other progress in the field has interested you?
OR: Of course, the year's climax was the publication of AlphaFold structure prediction that gives us access to the static protein structures. Moreover, the publication of our scientific advisor, Prof. Paola Picotti, “Dynamic 3D Proteomes Reveal Protein Functional Alterations at High Resolution in Situ,” published in Cell, shows how a dynamic view on protein structure could change the way we think about signaling cascade. The team investigated the role of protein structural alterations as a first-line cellular response integrating peptide-level LiP-MS data with orthogonal information such as phosphorylation, demonstrating the potential of integrating structural and abundance-based proteomics to achieve deeper insights into biological processes.
Another exciting new way to look at spatiotemporal changes in the phosphoproteome was published in Nature Communications by the group of Jesper Olsen. They showed that it is possible to see signaling events with an organelle resolution.
The last study that I want to highlight is “Immunotherapy-Chemo Combination Therapy Can Benefit Metastatic Pancreatic Cancer Patients.” The abstract was published at ASCO 21. Our collaborators from the Parker Instititure for Cancer Immunotherapy (PICI) underlined the PRINCE trial data, including our proteomics finding. The trial investigated how to activate the immune system to eliminate pancreatic tumors using chemotherapy combinations and/or an experimental antibody that targets the CD40 protein and activates immune cells. Notably, the trial did generate state-of-the-art biomarker data showing that the two immunotherapies, PD-1 and CD40, each had their expected effects on the immune system when individually combined with chemotherapy.
AM: You highlight that when studying proteins, it is important to gain action shots to fully understand the underlying biology. How does LiP-MS achieve this, and what benefits does it offer over other proteomics approaches?
OR: LiP-MS detects structural or surface accessibility changes, not static structures, like an event camera. Moreover, it does so with a resolution of a few amino acids, which allows the interpretation of the changes mechanistically.
AM: Biognosys has plans to overlay LiP-MS data onto predicted AlphaFold structures. What possibilities will this open in drug discovery?
OR: The ultimate determinant of protein function is its structure. The unique possibility of using our mass spectrometry-based technology for structural proteomics will change the way scientists look at the data. For example, if you open a classical biochemistry textbook, you will see that all fundamental mechanisms are explained with reference to protein structure.
We now launched the Biognosys 3D Protein Explorer, a tool that lets researchers visualize the proteomic signatures identified with our next-generation Plasma Biomarker Discovery solution in our latest pan-cancer study. We mapped each protein onto DeepMind's AlphaFold2 protein structures and UniProt's topological domains. We will keep expanding this portal to allow researchers to have a 3D view of the proteome and gain deep biological insights for drug discovery.
AM: Aside from drug discovery, what other applications can LiP-MS be used for to provide proteome-wide structural and functional information?
OR: We expect that structural biomarkers could become very interesting in the future. Many disease-relevant processes do not lead to changes in protein expression but changes in folding or protein interactions. Such changes may be caused by modifications, protein cleavage, or protein aggregation.
AM: What challenges in structural proteomics remain? How are you working to address these?
OR: Interpreting structural changes in a functional context remains challenging. In many cases, we see changes in binding pockets that are obviously relevant. The structure-function relationships outside compound binding sites are less clear and may be caused by changes in protein complexes or by modifications. The most important factor in overcoming these challenges is the method's sensitivity. The more signals we see, the better we can cover the whole protein sequence and pinpoint sites on the proteins where changes happen.
AM: What future advances do you envision in the field? How close are we to proteomics realizing its potential?
OR: The potential of proteomics is unlimited because, in contrast to genome sequencing, where the sequence analysis and its mutations are logical endpoints, there is no such endpoint in proteomics. After all, the functional proteome plays on several analytical dimensions. The deeper we go, the more we can learn about the function of protein modifications, different proteoforms, and protein-protein interactions. Future advances will come from improved sample preparation, chromatography and instrumentation. But the perhaps biggest untapped potential lies in the data analysis. We use only a fraction of the information hidden in the peptide or even peptide-fragment level data that we record. I expect deep learning technology to significantly contribute to uncovering functional insights from these data.
Oliver Rinner was speaking to Anna MacDonald, Science Writer for Technology Networks.