How Protein Mapping Could Transform Target Discovery in Precision Oncology

The efficacy of most oncologic therapeutics is determined by their interaction with a target, typically a protein within the tumor. The therapeutic either binds that protein and allows a payload, such as a chemotherapy, to enter the tumor cell, or inhibits a fundamental process essential to the tumor's survival.

The real challenge is identifying the best proteins to target for next-generation therapeutics. An ideal target is one that is highly expressed in cancerous tissue but present at very low levels in normal tissues. This differential is what gives rise to the therapeutic index: the dosing window within which a drug can selectively affect the tumor without harming healthy tissue.

To date, identification of such differentially expressed targets has been heavily dependent on genomics measures: DNA and, increasingly, RNA through bulk RNA sequencing, single-cell RNA sequencing, and spatial transcriptomics.

The challenge is that this data is largely inferring protein levels. We know that protein abundance and activity can shift independently of RNA due to post‑transcriptional regulation, alternative splicing, protein degradation and localization, and post‑translational modifications—meaning that in reality, RNA and protein levels are often poorly correlated.

RNA may suggest that a particular gene product represents a great target, but when we look at actual protein levels, that is not the case. The reverse is also true: RNA can miss a substantial number of targets where the differential expression appears minimal at the transcript level but is actually quite significant at the protein level.

What has now changed is our ability, for the first time on both a technological and scientific level, to assay proteins at scale directly in human tumors and tissues—whether fresh-frozen or FFPE samples—on the order of 10,000–12,000 proteins per sample.

We can also localize them to specific regions of the tumor cell, such as the cell surface, and identify proteoforms that contain therapeutically relevant epitopes. This enables us to identify new targets and assess their validity in a more direct and comprehensive manner, improving target selection up front, which in turn improves the overall probability of success in drug development.

It fundamentally depends on the underlying biological question being asked. There is no single “best” tool for all experiments, and I believe a large portion of experimental failures stem from an improperly scoped study rather than technical limitations.

That is why we focus on the question to be answered first and develop a bespoke experimental plan around that—curating methods to align with the specific objective to maximize the likelihood of generating meaningful, interpretable results.

You’ll see that in how we have built our Tumor Protein Mapping Platform: it can address each of these layers—surface, signaling, immune, and resistance biology—individually or in combination, depending on what the program requires.

In cases where we are trying to define tumor biology at scale, for instance, in building a tumor cell atlas and then using it to prioritize ideal targets as well as the ideal patient population for enrollment, having broad characterization is essential.

That means measuring not only the global proteome in its entirety but also understanding signaling cascades, immune context, resistance mechanisms, and cell surface protein expression together. This integrated view is what our Tumor Protein Mapping Platform provides, enabling comprehensive target prioritization.

Oftentimes, however, the question is much more discrete. In the case of an antibody-drug conjugate (ADC) or T-cell engager, the central question may be: what specific proteins are present at the cell surface that represent ideal targets for my therapeutic?

In the case of a kinase inhibitor or tyrosine kinase receptor inhibitor, the key question may be understanding which signaling cascades are turned on or off in a specific tumor type that you would like to antagonize.

In the case of an immuno-oncology-based therapeutic, understanding the immune context becomes paramount. And certainly, for all of these therapeutics, when patients develop resistance, understanding those specific resistance mechanisms is critically important.

The biological context and the question at hand are what determine the best assay and analytical approach.

Historically, many cancer therapies have been associated with significant toxicity because they broadly target fast-dividing cells, which include cancer cells but also normal cells like skin and bone marrow.

This has driven the shift toward precision medicine, with ADCs, T-cell engagers, and RLTs emerging as modalities that can selectively target tumor cells to minimize normal tissue drug exposure and off-target effects.

These therapies all fundamentally work by binding a protein at the tumor cell surface. In the case of an ADC or RLT, that binding enables a therapeutic payload to enter the cell. T-cell engagers are designed to bind both a cell surface target and a trigger molecule on T cells, bringing them into close proximity so the T cell can recognize and attack the tumor cell.

The challenge with these modalities is that, given their potent nature, you need to identify a target that is selectively expressed on tumor cells but not on normal tissue cells. That selectivity is what allows us to deploy these very precise but very potent precision-based therapeutics to selectively harm a tumor cell without in any way causing adverse events in normal tissue—but is hindered because our knowledge of the cell surface proteome is still limited.

To date, target identification for these modalities has been done through inference from genomics, whether DNA or RNA. As we discussed earlier, DNA and RNA do not directly correlate to protein levels.

The potential for on-target, off-tumor effects can therefore still be high, as a target that appears selectively expressed on tumor cells at the transcript level may in fact be present at significant protein levels on normal tissue cells, or conversely, a target assumed to be highly expressed on the tumor cell surface based on RNA data may not be adequately present at the protein level to support therapeutic engagement.

Through our mass spectrometry-based SurfaceSeek™ workflow, we can directly measure proteins, localize them to the cell surface, and understand the specific isoforms of the protein to identify ideal protein targets that are selectively expressed at the cell surface in a tumor and not on normal tissues.

In particular, we enrich proteins that have undergone N-linked glycosylation as part of maturation, enabling preferential identification of proteins that have completed trafficking and are exposed on the cell surface. That level of specificity, down to the isoform and the subcellular localization, is what makes direct measurement a critical advancement for these modalities.

This is a practical question more than anything else. Being able to understand tumor biology and inform patient enrollment and selection is dependent upon having large amounts of data across diverse sets of individuals and populations, or in this case, across diverse tumors.

This breadth of data is needed to best classify targets, understand their distribution across tumor types and patient subsets, and subsequently enroll patients from a particular subset. This is the very basis of precision-based oncologic therapeutics.

The challenge from a practical perspective is that finding large banks of fresh-frozen samples can be quite difficult. It is expensive to maintain these tumor samples. Keeping them under frozen conditions and in large cryopreservation systems is not always practical or feasible at the scale needed.

Conversely, FFPE samples are easy to store and are collected in almost every clinical and pathology setting—but formalin fixation and the paraffin embedding process introduce extensive protein cross-linking, which has historically made these samples inaccessible to deep proteomic analysis.

By developing proteomic methods that overcome these fixation effects and recover high-quality protein measurements from FFPE tissue, even from FFPE blocks stored for multiple decades, we unlock access to a vast untapped biorepository comprising millions of archived samples that exist worldwide across biobanks and hospital systems.

The ability to assay specific proteins—whether by identifying cell surface-localized proteins in tumor versus normal tissue through our SurfaceSeek workflow or directly measuring resistance mechanisms with our ResistanceSeek™ approach—is what allows biopharma teams to be more honed and more deliberate in how they apply their therapeutics.

This spans the full arc of drug development. In early phases, it provides a means to confidently select the ideal target to progress forward, based on how its biological and functional characteristics align with your selected modality.

Once targets and therapeutics have been developed, these workflows can be used to identify the ideal patient population, based on pathway activation and immune context, in order to enrich for responders.

And throughout development, it provides a way to characterize resistance mechanisms that may limit durability of response—turning what has historically been a retrospective explanation for therapeutic failure into a functional state that can be quantified, compared, and acted upon.

These measurements are what ultimately allow biopharma teams to close the loop between target, therapeutic, patient, and resistance biology. Rather than relying on genomic inference at each of these decision points, they can now ground their choices in direct protein-level evidence. That continuity of direct measurement across the full development process is what positions programs for durable clinical efficacy.

The reality is quite different from what our standard scientific teachings would suggest. When we think of the standard biological tenets, we think of DNA giving rise to RNA, giving rise to proteins. This concept is so central to biology that we call it the central dogma. Under that model, we think of a linear system in which there should be very strong correlation: when an RNA species is high, the corresponding protein should be high as well.

However, when we look across large tumor atlases in which both proteomics and matched RNA sequencing have been performed, we find that biologically, this is not the case. There is quite a bit of stochastic regulation that occurs between RNA and protein, and between DNA and protein. There are layers of regulation that control protein levels, and those layers are simply not captured in DNA or RNA measurements.

We can take this one step further. Not only is protein abundance poorly correlated to RNA, but there are specific protein isoforms, or proteoforms, that carry critical biological information: post-translational modifications, splice variants, translational mismatch variants, cleavage variants, and localization variants where a protein is now at a different place in the cell, such as the cell surface versus elsewhere. This entire layer of information is not captured at all on an RNA level.

Being able to directly make the protein measurement, both in initial assessment of targets and in patient stratification, is ultimately what is going to determine the success of a particular therapeutic, both in initial response and in understanding resistance mechanisms.

I believe this is going to be transformative. The last 20 years of oncology-based therapeutics have been built around the use of DNA and RNA and inference of protein levels. Under that model, I suspect the majority of ideal protein targets have actually been missed.

Approaches like our Tumor Protein Mapping Platform, which enable protein-based measurements at scale—whether around global proteomics measurements, cell surface localization, specific protein variants, or characterization of immune proteins and resistance mechanisms—will serve as a massive unlock. It will enable us to develop better therapeutics and deploy them in a very precise manner to benefit patients without causing adverse events.

The shift from inference to direct measurement represents a fundamental change in how we identify, validate, and ultimately deliver cancer therapeutics.

The introduction to this interview includes text that has been created with the assistance of generative AI and has undergone editorial review before publishing. Technology Networks' AI policy can be found here.