Improving the resolution of multiomic studies with proteomics gives deeper insights into disease biology
Background
Scientists are increasingly taking a “systems biology” approach to complex biological questions, with studies that provide a bigger picture by looking at the multiple molecular and cellular aspects involved in the disease or process under investigation. In the wake of the past few decades of intense genomics research, additional approaches that combine multiple molecular and cellular assays with clinical and phenotypic data are adding vital insights into the dynamic biology underlying human health and disease.
With their close proximity to phenotype, proteins are a crucial class of biomolecules in medical research, especially for complex, multi-factorial diseases. Recent advances in proteomics technologies have overcome the previous technical limitations that slowed the adoption of protein biomarker analysis into large omics projects. Olink’s Proximity Extension Assay (PEA) technology enables thousands of proteins to be measured simultaneously with exceptional specificity and sensitivity, using just a few µL of plasma, serum or other biological sample types. Here we present examples of how Olink is being used alongside other approaches such as genomics and transcriptomics to gain important new insights across a broad range of disease areas, and guide more effective drug development.
Cardiovascular disease
Proteomics provides important insights into major cardiovascular health issues like coronary artery disease, atrial fibrillation, and heart failure. These studies are advancing understanding of the underlying pathways and risk factors involved and how they link to associated comorbidities. The following example shows how the combination of proteomics and genomics (“proteogenomics”) was used to identify proteins with causal roles in ischemic heart disease (IHD), and which could provide robust new targets for drug development.
A population-scale study used GWAS, and proteomics data derived from the China Kadoorie Biobank (CKB) to identify novel biomarkers with causal roles in IHD, using the Olink® Explore 1536 high-throughput platform to measure the levels of almost 1,500 proteins in ~4,000 individuals. Sequence variants associated with protein expression levels that were located proximally to the protein-coding gene (cis-protein quantitative trait loci, pQTLs) were identified for 212 proteins that showed significant association with IHD incidence. Since protein level-disease associations alone could reflect either cause or effect, cis-pQTLs are unique tools to assess the relationship between genotype, protein expression and phenotype. Using Mendelian Randomization (MR), the authors could demonstrate that 13 of the 212 IHD-associated proteins had likely causal roles in disease, making them potential drug targets of interest.
“The present study identified 13 potential novel protein targets for drug treatment of IHD that had not been previously discovered using large-scale genomic data alone”
For independent validation and to take account of genetic diversity, the results from the Chinese population were compared to data from the UK Biobank Pharma Proteomics Project (UKB-PPP), where Olink Explore was used to measure plasma proteomics in over 53,000 participants. In the validation set, 307 of the 361 IHD-associated proteins identified in the CKB cohort had cis-pQTLs and MR showed that 16 of these had causal associations. Crucially, four proteins (FURIN, ASGR1, MMP3 and F2R) showed causal associations in both the Chinese and European cohorts. Of these, F2R & ASGR1 are currently in clinical trials for cardiovascular conditions, MMP3 is in clinical trials for non-CVD indications (providing drug expansion/repurposing possibilities), while FURIN represents an entirely novel potential target for IHD.
Mazidi M, Wright N, Yao P, et al. Plasma Proteomics to Identify Drug Targets for Ischemic Heart Disease. (2023) Journal of the American College of Cardiology, DOI: 10.1016/j.jacc.2023.09.804
Immunological & inflammatory diseases
Sensitive, accurate measurement of the plethora of immune response proteins that interact with each other, and multiple cell types is essential to better understand the complexity of immune system responses and the changes underlying autoimmune and inflammatory diseases.
In a study from Professor Emma Guttman-Yassky’s lab, multiple Olink Target 96 panels were used alongside mass cytometry and transcriptomics in a phase II clinical trial for the JAK inhibitor, ritlecitinib in the treatment of the autoimmune depigmenting disorder, nonsegmental vitiligo (NSV). The multiomics approach was used to identify invaluable new biomarkers of drug response. RNAseq of lesional skin biopsies showed drug-dependent changes in the expression of genes involved in immune response, T-cell activation, and antigen receptor–mediated signaling. The overall pattern trended towards those seen in non-lesional skin after treatment.
“This study also suggests the need for systemic treatment approaches in patients with substantial NSV involvement, as changes in immune signaling are not localized only to the skin”
In parallel, patients given the drug showed a significant decrease in T-cell activation, NK cell activation, and cytotoxic protein markers over the course of treatment. Clear dose-dependent changes in serum proteins were also seen in relation to clinical response to ritlecitinib, with a reduction in pro-inflammatory and immune-response markers correlating with positive patient responses. The parallel changes seen in lesional skin and blood indicate a systemic disease in patients with severe NSV and demonstrate the value of combining tissue-specific transcriptomics with serum proteomics.
Guttman-Yassky E, Del Duca E, Da Rosa JC, et al. Improvements in immune/melanocyte biomarkers with JAK3/TEC family kinase inhibitor ritlecitinib in vitiligo. (2023) The Journal of Allergy and Clinical Immunology, DOI: 10.1016/j.jaci.2023.09.021
Oncology
Adding a much-needed layer of resolution to genomic data, proteomic studies are identifying new diagnostic, prognostic and long-range risk biomarkers for multiple cancers and providing invaluable insights into the complexities of immuno-oncology.
As highlighted previously in this blog, combining genomics and proteomics provides a unique and powerful tool to address the causal role of proteins in disease and identify novel drug targets. In this study, the Olink Explore 3072 platform was used to measure >2,900 circulating proteins in almost 600 women from the KARMA prospective breast cancer cohort. Associations of protein levels were seen for multiple clinical traits, including novel findings and corroborations with previously established associations. Proteogenomic analysis revealed 812 cis-pQTLs associated with 737 different proteins and when these findings were compared to a previous study that looked at just 92 proteins, the overall correlation coefficient between effect sizes for the 33 overlapping variants was an impressive 0.91. Interestingly, when the findings were compared to a proteogenomic study using a non-Olink, aptamer-based proteomics platform, 229/603 cis-pQTLs from the 596 overlapping proteins were not identified in the previous study (possibly reflecting differences in specificity between these high-multiplex technologies).
“This suggests that these five proteins play an etiological or causal role in breast cancer, providing a basis for further functional evaluation of their potential as drug targets”
Mendelian Randomization analysis of the KARMA dataset then provided strong evidence of causality in breast cancer risk for 7 proteins, five of which (CD160, DNPH1, LAYN, LRRC37A2 and TLR1) could be replicated with data from previous case/control GWAS studies from the UK Biobank and FinnGen. While LAYN is a registered drug target for hyaluronic acid, none of the remaining four proteins are known drug targets. The findings therefore open up the possibilities for both drug development based on novel targets, and repurposing opportunities for existing therapeutics.
Mälarstig A, Grassmann F, Dahl L, et al. Evaluation of circulating plasma proteins in breast cancer using Mendelian randomisation. (2023) Nature Communications, DOI: 10.1038/s41467-023-43485-8
Neurology
While many proteomic and multiomic studies use plasma or serum as a convenient, minimally invasive sample-type with which to investigate the proteome, some aspects of human disease biology may require a different approach. The flexibility of Olink’s PEA technology to provide high-quality, multiplexed protein measurements across a broad range of sample matrices can be of great benefit in such cases. Neurological diseases are a prime example of this, where cerebrospinal fluid (CSF) may be a more biologically relevant source of protein content than blood. As demonstrated by almost 100 peer-reviewed scientific articles to date, Olink has been widely used to investigate CSF proteomics in key neurological conditions such as Alzheimer’s disease and dementia, traumatic brain injury, multiple sclerosis, and chronic pain.
A proteogenomics investigation from Hansson and colleagues measured 398 proteins in CSF samples from 1,591 participants from the BioFINDER longitudinal prospective study for neurodegenerative diseases. A combined CSF proteomics and GWAS analysis identified 176 pQTLs for 145 CSF proteins (117 cis and 59 trans), most of which were entirely novel findings. A meta-analysis of available transcriptomics databases showed that 53 of the cis-pQTLs identified showed previous evidence of association with gene expression in brain tissues. Mendelian Randomization suggested causal roles for several proteins in neurological diseases, such as ApoE, CD33, and GRN in Alzheimer’s Disease, MMP-10 in preclinical Alzheimer’s disease, SIGLEC9 in amyotrophic lateral sclerosis, and CD38, GPNMB, & ADAM15 in Parkinson’s Disease. These could provide novel therapeutic pathways to be explored for several important diseases.
“The main novelties included the use of highly specific proximity extension assays in a large cohort, and the integration of brain volumetrics to account for confounding factors. New possible treatment targets for several neurological diseases were nominated”
Another interesting observation was that the genetic region GMNC-OSTN was important for several of the trans-pQTLs identified, and there were previous links made between genetic variants in this region and brain ventricle volume. Since ventricle volume could affect the concentration of some proteins in CSF, available magnetic resonance imaging (MRI) data for a subset of participants was utilized. This supported the notion that ventricle volume could be a confounder for some CSF pQTLs, which could be an important factor for future CSF studies.
Hansson O, Kumar A, Janelidze S, et al. The genetic regulation of protein expression in cerebrospinal fluid. (2022) EMBO Molecular Medicine, DOI: 10.15252/emmm.202216359
