Unlocking the Undruggable: How Biotechnology Is Rewriting Drug Discovery

Mar 4, 2024·
Yassir Boulaamane
Yassir Boulaamane
· 3 min read

“Undruggable” usually means “no useful pocket for a small molecule.” KRAS, BCL-2, transcription factors, and intrinsically disordered proteins sat in that category for a long time. A few approved drugs show the label was about the modality, not about the biology being unreachable.

The figures and examples below are the ones in the DrugBank overview this note follows. They are not a new census.

Why a target resists a pocket, and which modalities answer itFour reasons a target is called undruggable: conserved sites, disorder, flat protein-protein interfaces, and polar surfaces. Four modalities named in the note: PROTACs, multispecifics, nucleic acids, and peptidomimetics.The obstacle and the modalityConserved active siteSelectivity is the problemDisordered proteinNo single bound poseFlat PPI surfaceNo obvious cavityPolar, solvent-exposedFights drug-like rulesPROTACDegrade instead of inhibitMultispecificMore than one siteNucleic acidRNA, antisense, siRNA, editingPeptidomimeticBind the flat surface
Figure 1. The top row is why a pocket-shaped small molecule fails. The bottom row is the set of modalities named in this note. They are not assigned one-to-one. Venetoclax is the BCL-2 protein–protein example, and sotorasib is the KRAS G12C pocket that turned out to exist.

How the pocket became the default

Discovery started with phenotypic screens: an extract, a phenotype. Target-based design then went after receptors, channels, and enzymes that already had a cavity. Biologics widened that, and still reached only about 2% of the human proteome. Transcription factors, flexible proteins, and broad protein–protein interfaces stayed outside.

Imatinib and the kinase inhibitors that followed are the success of the pocket strategy. They are also the demonstration of its edge.

Why the difficult targets are difficult

  • Conserved active sites make selectivity hard.
  • Intrinsically disordered proteins do not offer one binding pose.
  • Protein–protein interfaces are flat.
  • Polar, solvent-exposed surfaces fight the usual drug-like property ranges.

Venetoclax, a BCL-2 inhibitor, and sotorasib, a KRAS G12C inhibitor, are the two cases this overview uses to show that those constraints can be worked around.

Repurposing

Repurposing skips part of the early safety and manufacturing work. In the overview, it accounts for one-third of recent FDA approvals and about 25% of pharmaceutical revenue. Sildenafil, propranolol, and aspirin are the classic re-readings of an existing drug. Current practice adds transcriptomic matching, electronic health records, and network pharmacology.

The limits are the ones stated there: weaker intellectual-property protection, and data you cannot always get. The openings it names are rare disease, pandemics, and more personalized use.

Modalities that do not need a classical pocket

  • PROTACs hand the target to the cell’s degradation machinery.
  • Multispecifics, small molecules or antibodies, hit more than one function.
  • Nucleic-acid therapeutics act on expression: RNA, antisense, siRNA, gene editing.
  • Peptidomimetics copy a protein surface so they can bind a flat interface.

The same overview points at four results. Venetoclax got onto the BCL-2 interface. Sotorasib used a KRAS pocket that had not been the starting assumption. HIF-2α inhibitors reached a transcription factor. Peptilogics’ PLG0206 is the computational peptide example.

The development path is still the bottleneck

Approval rates in the 2000s are given as about 12% from Phase I to approval. The failure modes named are lack of efficacy, toxicity, and poor pharmacology. Outsourcing and academic partnerships do not remove that.

The process changes listed are adaptive trials, biomarker-based recruitment, AI-guided study planning, real-world evidence, and network pharmacology for multi-target questions. Orphan, fast-track, and breakthrough designations are the regulatory routes named alongside them.

What “druggable” is doing in this sentence

The word expands when the toolbox expands: repurposing, PROTACs, multispecifics, nucleic acids, designed peptides, high-resolution structures, and large-scale computational screening. A protein can still be a bad target. It is less often a bad target only because a small molecule cannot find a hole.

Source: DrugBank. Unlocking Undruggable Targets: Biotechnology’s Role in Expanding Drug Discovery. 2024.

Biotechnology Undruggable Targets Therapeutic Modalities
Yassir Boulaamane
Authors
Yassir Boulaamane
Postdoctoral Researcher