Drugging the Undruggable
How Science Is Conquering Cancer’s Toughest Targets
For decades, some of the most powerful drivers of cancer have remained beyond reach. We have long recognized the roles that key proteins play in fueling tumor growth, yet efforts to target them with drugs have repeatedly failed. These proteins lack the deep binding pockets that conventional small-molecule inhibitors require, operate through intricate networks of protein-protein interactions, or reside in cellular compartments inaccessible to traditional biologics. Some, rather than being overactive, are missing or mutated, making it far more difficult to restore their function than to simply inhibit them. Until recently, the field called these proteins undruggable—a term that captured not just our frustrations but also the limits of what technology could achieve at the time. When I was in medical school, the idea of undruggable targets was nearly an immutable law of nature, shaping the way drug discovery was approached. Pharmaceutical research largely focused on well-defined enzymes with clear binding pockets, while proteins lacking these characteristics were often deemed impossible to target. This belief influenced investments, research priorities, and even the way students like myself were trained to think about therapeutic possibilities. We were taught that certain cancer-driving proteins—those without enzymatic active sites or those buried in complex cellular machinery—are beyond our therapeutic reach. The idea that we might one day successfully drug these targets seemed as unlikely as rewriting the fundamental rules of biochemistry.
Today, that definition is rapidly changing. Advances in molecular biology, structural chemistry, and computational drug design have led to a new wave of strategies that are rewriting the rules of drug discovery. Instead of trying to force conventional inhibitors onto these unconventional proteins, we are now developing methods to eliminate them entirely, alter their interactions, or exploit vulnerabilities that had gone unnoticed. The idea that certain targets are simply impossible to drug is being replaced by an expanding toolkit of approaches that challenge what was once considered out of reach.
Why Some Cancer Targets Were Left Behind
The success of precision oncology has been built on the ability to design drugs that precisely shut down cancer-driving proteins. Many of the early victories in targeted therapy focused on enzymes such as kinases, which have well-defined pockets where small molecules can bind and inhibit function. This led to landmark treatments, such as kinase inhibitors that transformed leukemia and lung cancer from deadly diseases into somewhat manageable conditions.
Protein kinases regulate cellular processes and are key drug targets for diseases like cancer and autoimmune disorders. Inhibitors like Imatinib, Erlotinib, Sorafenib, Tofacitinib, and Trametinib disrupt abnormal signaling pathways, offering effective treatment strategies.
But not all cancer drivers fit neatly into this paradigm. KRAS, a small GTPase frequently mutated in pancreatic, lung, and colorectal cancers, has long been considered undruggable because its smooth surface lacks the deep binding pockets that traditional inhibitors require. MYC, a transcription factor involved in cell proliferation, has an intrinsically disordered structure, making it difficult to target directly. Beta-catenin, a central player in the Wnt signaling pathway, has a broad, shallow protein-protein interaction interface, which presents significant challenges for small-molecule drug design. Some, such as the tumor suppressor p53, are frequently mutated in cancer, meaning that reactivating their function is more complex than simply inhibiting an overactive protein.
Recognizing these challenges, researchers developed alternative strategies to target elusive cancer-driving proteins indirectly. Rather than attempting to bind directly to these structurally complex molecules, efforts shifted toward disrupting the broader signaling networks that sustain them. By inhibiting upstream regulators, blocking downstream effectors, or exploiting vulnerabilities unique to cancer cells, this approach aimed to dismantle oncogenic drivers by cutting off their functional support systems. While promising in theory, it has faced significant hurdles, as cancer cells rapidly evolve mechanisms to bypass these interventions. Attempts to disrupt oncogenic signaling pathways often failed to produce lasting responses, as tumors adapted by rerouting their internal circuitry.
For example, farnesyltransferase inhibitors were designed to prevent KRAS from localizing to the cell membrane, but cancer cells circumvented this by using alternative lipid modifications to activate KRAS. Similarly, indirect strategies to suppress MYC, such as bromodomain inhibitors, showed promise in preclinical models but faced challenges in clinical settings, limiting their effectiveness. Over time, the belief that these proteins were simply beyond the reach of medicine became deeply ingrained in drug discovery, leaving some of the most critical oncogenic drivers untouched for decades.
Yet, the landscape is changing. With new technologies, deeper biological insights, and novel therapeutic strategies emerging, we are getting closer to drugging the undruggable. In many ways, we are just getting started.
Some indirect approaches aimed at undruggable targets have seen success. For example, inhibitors targeting BCL-2, a key anti-apoptotic protein, have proven effective in certain cancers where MYC overexpression creates a reliance on BCL-2 for survival. This strategy, rather than directly inhibiting MYC, exploits a vulnerability in its downstream effects.
Progress in Drugging the Undruggable
Recent progress in targeted protein degradation, structure-based drug design, and synthetic lethality are offering new ways to engage these difficult targets. Some of the most promising breakthroughs focus not on inhibiting these proteins but on removing them entirely. PROTACs (proteolysis-targeting chimeras), for example, have been developed to degrade previously intractable targets such as androgen and estrogen receptors in hormone-driven cancers. We are now applying this technology to more challenging proteins, including MYC and beta-catenin, offering a potential pathway to eliminating them from cancer cells.
Building on the success of targeted protein degradation, another major breakthrough comes from a deeper understanding of the dynamic nature of protein structures. This shift in focus has led to the discovery of previously unrecognized allosteric sites and transient binding pockets that can be exploited for drug development. The discovery of a previously unrecognized allosteric pocket on KRAS-G12C led to the development of the first-ever direct KRAS inhibitors, sotorasib and adagrasib, which irreversibly bind to the mutant form of KRAS and prevent it from signaling. This represents a shift in thinking about how "undruggable" targets can be approached—not by directly inhibiting their primary function, but by locking them in an inactive state.
An allosteric site is a distinct region on a protein where molecules bind to regulate its activity without directly interacting with the active site. This binding induces conformational changes that can enhance or inhibit the protein’s function, allowing for precise control of biological pathways. For example, ATP acts as an allosteric inhibitor of phosphofructokinase, regulating glycolysis based on cellular energy levels. Similarly, allosteric kinase inhibitors modulate enzyme activity by targeting these sites, providing a mechanism for selective drug action in cancer and other diseases.
Synthetic lethality is also providing a new way to exploit vulnerabilities in cancers driven by difficult targets and we are now applying this principle to cancers driven by targets such as KRAS and MYC mutations, identifying co-dependent pathways that can be disrupted to selectively kill cancer cells while sparing normal tissues. RNA-based therapies are also expanding the druggable landscape by allowing interventions at the genetic level rather than targeting proteins directly. For example, antisense oligonucleotides and small interfering RNAs have been developed to suppress the production of disease-driving proteins, while emerging mRNA-based approaches aim to restore functional tumor suppressors like p53 in cancers where they are lost or mutated.
Synthetic lethality occurs when the simultaneous loss of function in two genes leads to cell death, while a defect in only one of them is survivable. This concept is exploited in cancer therapy to target tumor cells with specific genetic mutations. For example, PARP inhibitors like Olaparib exploit synthetic lethality in BRCA1/2-mutated cancers, as BRCA-deficient cells rely on PARP for DNA repair. Another example is Wee1 kinase inhibitors, which target cancers with TP53 mutations, making them unable to manage DNA damage effectively, leading to cell death.
Antisense oligonucleotides (ASOs) are short, single-stranded DNA or RNA molecules that bind to complementary mRNA to block translation or induce degradation, while small interfering RNAs (siRNAs) are double-stranded RNA molecules that guide RNA-induced silencing complexes (RISC) to degrade specific mRNA, effectively silencing gene expression.
In recent years, computational drug discovery, fueled by artificial intelligence, has been used to accelerate the identification of entirely new ways to engage difficult targets. A notable example is the discovery of novel inhibitors for DDR1 kinase, an otherwise challenging target, through AI-driven molecular screening. This approach has enabled the identification of promising drug candidates much faster than traditional methods, demonstrating AI’s potential to unlock new therapeutic opportunities. Machine learning models trained on massive datasets of molecular interactions are now predicting which drug-like compounds might bind to previously unrecognized sites, allowing researchers to scan billions of potential molecules and prioritize candidates that traditional methods might overlook.
Despite this promise, first-generation AI-powered drug discovery efforts have struggled to translate computational insights into clinically successful drugs, due to challenges such as poor training data quality, model interpretability, and the unpredictable complexity of biological systems. Many early efforts focused on narrow aspects of drug discovery, such as virtual screening, without fully integrating experimental validation and downstream development. However, a new generation of AI-driven efforts is emerging, leveraging end-to-end, high-throughput automation and self-improving models that integrate computational predictions with real-world experimental feedback. These next-generation platforms combine AI with robotics, bioinformatics, and large-scale biological data to accelerate not just target identification, but also lead optimization and preclinical validation, increasing the likelihood of successful clinical translation. This shift suggests that AI’s role in drug discovery is maturing, with the potential to redefine what is considered druggable and expand the boundaries of therapeutic innovation.
The Next Frontier in Cancer Drug Discovery
The notion that certain cancer proteins are undruggable is rapidly fading, as advances in targeted protein degradation, synthetic lethality, RNA therapeutics, and AI-driven drug discovery prove that the limits of discovery are not fixed but constantly evolving. What was once deemed impossible is now yielding promising therapeutic modalities and the pace of progress is accelerating.
Back when I was in medical school, we were taught that some targets were simply beyond our reach—an immutable reality we had to accept, a limitation we were trained to work around rather than overcome. But scientific progress thrives on challenging assumptions. Today, many of those same targets are being successfully drugged, not because biology has changed, but because our understanding of it has deepened. This evolution underscores a fundamental truth: science is not just the pursuit of knowledge, but the relentless drive to transform the impossible into the achievable, one single experiment at a time.