By charting more than 13,000 proteins across healthy tissues and cancers, the study offers a powerful new guide to development, tumor biology, drug safety, and future therapeutic targets.
Study: Spatial distribution of the proteome in the human body and in cancers. Image Credit: cybermagician / Shutterstock
In a recent study published in the journal Nature, researchers developed one of the most comprehensive maps of protein expression to improve understanding of biological mechanisms that drive health and disease. Using data-independent acquisition mass spectrometry (DIA-MS), they measured the abundance of over 13,000 proteins across over 2,800 samples. The atlas provides in-depth characterization of proteins, key functional molecules in cells, across normal and cancerous tissues.
Diseases and cancers alter cellular function through molecular changes that affect protein activity. Identifying these molecular alterations in constituent proteins can help scientists develop more personalized treatments. Existing repositories contain protein-expression data from tumors. Some research initiatives have expanded these resources to include genetic and transcriptomic data from adjacent normal tissues. Scientists are now exploring different methods to reliably and simultaneously quantify thousands of proteins across tissue types in healthy and tumor samples.
About the study
In the present study, researchers provided a detailed quantitative map of human proteins. They performed DIA-MS to profile 13,609 proteins in 2,856 samples obtained from nine post-mortem adult donors, nine post-mortem fetal donors, eight healthy individuals, and 1,015 cancer patients.
The team used the t-SNE visualization method to investigate whether fetal, tumor, paired non-tumor, and healthy adult samples followed an ordered pattern. They then performed trajectory analysis to quantify the sequence by assigning each sample a ‘pseudotime’ score. They also noted that brain and liver tumors and paired non-tumor samples deviated from this overall pattern. They also performed unsupervised clustering to group proteins with similar expression patterns across samples.
The researchers identified proteins that were abundant in specific tissues. They then linked them to specialized functions of those tissues. The team calculated Euclidean distances to assess similarities and differences between the protein profiles. They used the Human Protein Atlas (HPA) criteria to identify proteins found in a single tissue, in groups of related tissues, those present everywhere but especially enhanced in a particular tissue, and those with unclear patterns.
The researchers then compared proteins that were especially abundant in specific tissues with known drug targets listed in DrugBank. This helped identify organs that may be prone to drug-related toxicity. They also analyzed the dataset to identify cancer-related protein changes and potential drug targets. They also analyzed drug-sensitivity data and clustered regularly interspaced short palindromic repeats (CRISPR) gene-essentiality data to assess cancer cell responsiveness to the identified drugs and gene knockouts.
Results
Researchers analyzed proteins across 58 healthy adult tissues, 251 tissue subtypes, 22 fetal tissues, and samples representing 25 different cancers. While protein profiles differed across tissue and sample types, repeated analysis of the same sample produced largely similar results. Protein profiling placed fetal tissues, tumor tissues, adjacent non-tumor tissues, and normal adult tissues along a continuum that reflected increasing tissue differentiation, although brain and liver samples were notable exceptions.
Brain tissues across fetal, tumor, paired non-tumor, and normal adult states showed relatively stable protein profiles with fewer developmental shifts. Liver tissues showed much larger changes, which the authors linked to the liver’s adaptive plasticity. While proteins involved in ribonucleic acid (RNA) processing were most abundant in fetal tissues, those linked to antibody-based immunity were abundant in adult tissues.
Tissues with specialized biological functions, such as the semicircular canals and cochlea, formed distinct molecular clusters, while functionally related tissues, including the brain, spinal cord, and peripheral nerves, clustered closely together. Across 36 tissues overlapping with HPA data, the team identified 832 proteins that were uniquely enriched in this dataset. For instance, they detected pannexin 3 (PANX3) at particularly high levels in the cochlea and confirmed its presence using synthesized PANX3 peptides.
The analysis linked 2,598 drugs to 402 proteins that were especially abundant in certain tissues. The liver contained most of these targets. Cytochrome P450 2C8 (CYP2C8), a liver-enriched enzyme involved in drug metabolism, was targeted by 302 drugs. The researchers highlighted gemfibrozil as a potent CYP2C8 inhibitor that can raise concentrations of co-administered drugs and increase the risk of adverse effects. Triclosan exposure has been associated with altered thyroid function, and it targets thyroid peroxidase, an enzyme essential for thyroid hormone production.
The team identified 8,940 differentially expressed proteins (DEPs) across 25 cancer types, of which 2,878 were specific to a cancer type. Testicular carcinomas showed increased expression of tissue-specific proteins, suggesting that these tumors may exploit intrinsic germ-cell growth programs. The researchers also screened 77 upregulated DEPs shared among tumors that were targeted by 36 biomolecular drugs, mostly receptor tyrosine kinase inhibitors, evaluated in 2,084 clinical trials. The findings suggest that Trodelvy, an antibody-drug conjugate approved for triple-negative breast cancer, could be explored for endometrial cancers because both trophoblast cell surface antigen 2 (TROP2) and DNA topoisomerase 1 (TOP1) are upregulated in the disease. They also identified potential new targets, such as TYROBP, which could be investigated as a potential shared antibody-drug conjugate target, although its myeloid expression requires toxicity evaluation.
Conclusions
The findings provide a comprehensive human protein atlas spanning healthy tissues and tumor tissues to improve understanding of tumor pathophysiology. By unraveling health- and disease-specific alterations in fetal and adult tissues, the dataset could help advance research into human development, tissue biology, and drug discovery. In future studies, researchers should include larger samples of individuals across different ages and tumor types to validate the findings.
Download your PDF copy by clicking here.
New biomarker predicts gastric cancer outcomes from CT scans