It was recently discovered1, 2 that proteins covalently 'tagged' with small, synthetic, hydrophobic molecules are degraded by the cell's quality-control machinery. Writing in Chemistry & Biology, Long et al.3 now report that non-covalent binding of such molecules also marks proteins for degradation. This finding could open up a wide range of proteins as targets for drug-discovery programmes.
The dearth of newly approved drugs in the past decade reflects the challenges faced by the pharmaceutical industry. Although advances in genomics have identified many proteins that are implicated in disease, many of these proteins — especially those that are not enzymes — are not currently viable drug targets. In fact, it has been estimated that only about 15% of the human proteome is 'druggable' with small molecules4.
Many attractive drug targets have therefore been dubbed 'undruggable'. For instance, there are roughly 1,400 human transcription factors — proteins that regulate messenger RNA synthesis from DNA, but which lack enzymatic activity. These proteins remain largely undruggable, despite the fact that aberrant expression of some of them is known to cause cancer. One possible solution to this challenge has been the development of small interfering RNAs (siRNAs), which intervene in gene expression by binding to mRNA. However, delivering siRNAs to their targets in vivo has been a difficult hurdle to overcome, and so small molecules that can affect the function of undruggable proteins are needed.
Another emerging approach is to destroy, rather than inhibit, target proteins in cells. Normal protein turnover in cells is mainly mediated by the ubiquitin–proteasome system (UPS), which tags unwanted or misfolded proteins with chains of the ubiquitin protein. Once ubiquitinated, the marked proteins are recognized by the proteasome, a large, barrel-like molecular machine that cleaves proteins into small peptides. Efficient removal of unwanted proteins is key to cell survival, as evidenced by the development of proteasome inhibitors as effective antitumour agents5.
Several strategies have been reported that co-opt the UPS for targeted protein degradation. One of these uses 'proteolysis-targeting chimaeric molecules' to bring the protein of interest close to a ubiquitin ligase (an enzyme that mediates the ubiquitination of a target protein), thus bringing about protein ubiquitination and subsequent degradation6.
An alternative approach is to mimic a misfolded protein state using small molecules. Normally, the 'greasy' (hydrophobic) side chains of polypeptides are buried in the interior of a globular protein, with the hydrophilic amino-acid residues lying at the surface. Even a small increase in surface hydrophobicity can make a protein unstable. For instance, the deletion of a single amino acid from the CFTR protein is the main cause of cystic fibrosis. The deletion results in the exposure of hydrophobic patches on the surface of CFTR, leading to misfolding and subsequent degradation of the protein (Fig. 1).
We have recently shown1, 2 that the covalent attachment of a synthetic hydrophobic group (such as adamantane, a bulky hydrocarbon) to the surface of proteins attracts chaperone proteins whose job it is to help refold misfolded proteins, or, if they cannot be refolded, to target them for degradation by the proteasome. But most drugs bind to proteins through non-covalent interactions, and it was unclear whether non-covalently bound molecules could also trigger this sequence of events.
Long et al.3 have settled this concern. They investigated the biological effect of attaching a hydrophobic group (Boc3Arg, a modified arginine amino acid) to trimethoprim (TMP), a ligand molecule that binds non-covalently to the dihydrofolate reductase (DHFR) enzyme from the bacterium Escherichia coli. The authors observed that TMP–Boc3Arg induces 30–80% DHFR degradation in mammalian cells, depending on the rate of DHFR synthesis. This effect could be blocked either by TMP, which competes with TMP–Boc3Arg for binding to DHFR, or by inhibitors of proteasome activity.
The authors also demonstrated that the glutathione S-transferase (GST) enzyme is degraded when treated with a compound in which Boc3Arg is attached to ethacrynic acid (EA), a GST inhibitor that becomes covalently bound to the enzyme's active site. This demonstrates that the degradation effect of Boc3Arg occurs for at least two enzymes. Long et al. went on to make a fusion protein in which DHFR is attached to GST, and then treated cells producing the protein with either TMP–Boc3Arg or EA–Boc3Arg. They observed that DHFR–GST was degraded more efficiently by EA–Boc3Arg, which binds covalently to the protein, than by TMP–Boc3Arg, which binds non-covalently. This suggests that the covalent attachment of hydrophobic tags to enzymes is the more effective strategy for protein degradation.
As TMP is a high-affinity inhibitor of E. coli DHFR, further studies are needed to determine whether a small molecule that is both a protein inhibitor and a degradation signal is more effective in abrogating protein function than a simple inhibitor. As pointed out by the authors, the case of botulinum toxin illustrates the advantage of the degradation approach. The most potent form of this toxin, which causes muscle paralysis, has a half-life in the body of about 3 months. Although an inhibitor of the toxin would be able to suppress toxicity in the short term, elimination of the toxin is obviously a preferable therapeutic approach.
However, the Boc3Arg moiety is large (almost 500 daltons in mass), and large molecules often have poor pharmacokinetic properties that limit their use as drugs. So, appending it to an existing inhibitor could potentially worsen that inhibitor's pharmacokinetic properties. Curiously, even though TMP has high affinity for E. coli DHFR and is thought to have excellent cell permeability, Long et al. needed to use a high concentration of TMP–Boc3Arg to observe protein degradation. This suggests that TMP–Boc3Arg has difficulty permeating cells.
Other non-covalent ligand–protein systems need to be tested to establish the minimum ligand–protein affinity necessary to initiate protein degradation. Meanwhile, it is intriguing to speculate about how the modularity of this protein-degradation strategy might be used for drug discovery. One could envisage a streamlined process in which ligands for an undruggable protein are identified, appended with hydrophobic moieties (such as adamantane or Boc3Arg) and tested for their ability to degrade the target protein. Finding high-affinity ligands for undruggable proteins will certainly be a challenge, but methods are becoming available to facilitate this.
For instance, chemical libraries in which each compound is attached to a unique DNA 'barcode' can be tested for protein binding, and the chemical entities that have the highest binding affinities subsequently identified using the barcodes7. This method would allow the rapid screening of up to 109 compounds, whereas the largest screens currently used assay only about 106 compounds8. A combination of such high-throughput screening methods with the hydrophobic tagging approach could make today's undruggable proteins attractive biological targets in the search for compounds that ameliorate human disease.
- Nature Chem. Biol. 7, 538–543 (2011). et al.
- ChemBioChem 13, 538–541 (2012). et al.
- Chem. Biol. 19, 629–637 (2012). , &
- Drug Discov. Today 10, 1607–1610 (2005). &
- 441–449 (Lippincott Williams & Wilkins, 2011). , , & in Cancer: Principles & Practice of Oncology (eds DeVita, V. T. Jr, Lawrence, T. S. & Rosenberg, S. A.)
- J. Am. Chem. Soc. 126, 3748–3754 (2004). et al.
- Chem. Soc. Rev. 40, 5707–5717 (2011). , &
- Nature Chem. Biol. 5, 647–654 (2009). et al.