Proteogenomics: When Genomics Is Not Enough

When studying diseases and treatments, multi-omics analysis is justified solely on the basis of “more information is better.” The central dogma of molecular biology provides an even more compelling case since not all DNA is transcribed and not all RNA is translated. Proteogenomics (PG), particularly the single-cell variety, seeks to understand these relationships and exploit them therapeutically.

In a recent review of the role of PG in precision medicine, Henry Rodriguez Ph.D., Director, Office of Cancer Clinical Proteomics Research at the National Cancer Institute, related how, at one time, genomics was considered sufficient for delivering precision treatments. DNA- and RNA-based methods alone, however, were limited in many instances by the inability to associate genotypes to phenotypes.

“Fundamental biology states that the flow of information in our cells is DNA to RNA to protein—with proteins doing most of the work,” Rodriguez explains. So, when trying to understand the biology of a disease, why would anyone study one-third or two-thirds of the formula, or each item (DNA, RNA, protein) in isolation? Modern analytic methods allow investigators to look beyond DNA and RNA to view cancer with new clarity.

Not by genomics alone

“Precision oncology could not exist without genomics. However, there is an acknowledgement that genomics alone does not always provide sufficient insights into the molecular classification of cancer, Rodriguez adds. “Genomics and proteomics must be seen as partners, as in proteogenomics, which I believe holds the promise to reveal new molecular patterns of cancer biology, with the potential to inform new approaches to cancer diagnostics and therapeutics.”

Relating genotype to phenotype begs the question, however, on the correlation of normal or abnormal genetics (i.e., mutations) to normal or abnormal phenotypes (diseases). We are tempted by arguments that mutated genes will result in an abnormal protein and hence, in disease. Rodriguez warns against such simplistic thinking.

“The root causes of cancer are more complex. Protein post-translational modifications add more complexity to cancer biology by increasing the functional diversity of proteins. We should not view cancer on the two-dimensional plane of ‘normal gene equals no disease, while mutated gene equals disease,’ but rather through a multidimensional scale studied in unison. For years, different groups have discovered more and more mutations involved in cancers, but unfortunately, those mutations are found in many different genes. Scientists can’t make sense of it all.”

Simply put, genomics alone do not always provide the necessary or sufficient clues as to disease origins and potential treatments. The advantage of proteogenomics, according to Rodriguez, is the ability to exploit simultaneous signals at the DNA, RNA, and protein levels regardless of whether genetic alterations result in abnormal protein function or not.

“From my perspective,” Rodriguez tells Biocompare, “a ‘discovery to translational to clinical’ program should incorporate proteogenomics directly into the fabric of clinical research and patient care, along with interdisciplinary and international collaboration, patient involvement, and making research data available to the broader community as a global public good. For data-driven precision oncology to become more precise, proteogenomics needs to be fully integrated.”

The major question here is “at what cost?” Personalized oncology is already here, albeit in primitive form. The price tag for such treatments has already surpassed, by a large multiple, the ability of individuals to pay. Health plans, as well, should at some point begin asking questions about cost-benefit, quality of life for those undergoing treatment, and the allocation of increasingly precious healthcare resources—discussions which today are either absent or lost amid the excitement of scientific discoveries.

“This is a topic with many complexities, no easy answers, and ultimately requiring public and private parties working together to achieve the best outcomes for patients with cancer,” says Dr. Rodriguez.

The case for single-cell

Readers are by now aware of the rationale for single-cell analysis, particularly single-cell multi-omics, which was named “Method of the Year'' by Nature Methods in 2019. Single-cell PG involves the simultaneous characterization of the genome and proteome in individual cells, which allows the cell-by-cell correlation of genotype (including at the epigenomic-level) with phenotype.

The TotalSeq workflow from BioLegend, which is an extension to well-established single-cell DNA or RNA-seq methods, is illustrative. A typical RNA-seq experiment involves sample preparation (focused mainly on obtaining single cells and quality control for cell viability), labeling cells for surface markers of interest using antibody-oligonucleotide conjugate reagents, and finally initiating the single-cell RNA experiment as usual in the presence of oligo-coated gel beads within oil droplets.

Cell lysis releases RNA molecules, which bind quantitatively to specific oligo probes on the gel beads. Simultaneously, the oligomeric tags from the antibody reagents (the proteomic component of the experiment) also bind to the gel beads, which are uniquely identifiable through molecular barcoding.

“After template extension, PCR amplification, and purification, you are left with a sequencing-ready library for the mRNA, and another one for the antibody-derived tags (ADTs),” says Dipesh Risal, Ph.D., Sr. Product Marketing Manager—Proteogenomics, at BioLegend. “TotalSeq adds the dimension of proteomics in single-cell RNAseq experiments so that when you sequence, you can say, ‘this RNA and this protein came from this cell.’”

Data from ADTs look similar to single-cell RNA data users might be familiar with. Aligning the reads allows for the quantification, at the single-cell level, of the number of reads per antibody/epitope identified.

“Protein signatures generally correlate with the RNA, but we have seen cases where the protein is upregulated while the RNA is down-regulated, and vice versa, indicating a complex regulatory mechanism,” Risal explains. “The protein signature has a wider dynamic range than RNA, and it also compensates for the well-known RNA dropout phenomenon, in which some RNA signatures are low-to-non-existent due to capture and amplification issues. We have found that adding the proteomic signal to a dimension-reduction step allows for a better resolution of cellular subtypes, because we are adding data orthogonal to the RNA signature for resolving those subtypes. Adding the proteomics component to RNA-seq therefore adds a lot of power to single-cell experiments.”

Whether proteogenomics is of the bulk or single-cell variety, or includes DNA (as the BioLegend workflow does), molecular signatures represent only the diagnostic side of personalized oncology. The other ingredient, if these efforts are to have independent significance, must come in the form of therapies. That is, new therapies that specifically target those pathways and mechanisms uncovered by PG.