PROTACs: a hard road, but one worth taking?
There has been a lot of interest in the potential of PROTACs – protein-targeting chimeras – as drugs in the past few years. While some are now in clinical development, it remains early days for this therapeutic modality. But interest is growing, and there are various scenarios where a PROTAC might actually be preferable to a more traditional small molecule inhibitor.
In brief, PROTACs are hetero-bifunctional molecules comprising an E3 ligase binder and a ligand that targets the protein of interest (POI), connected by a linker. In contrast to biologics, they are relatively small molecules, with molecular weight ranging from about 700 to 1200. Once they enter a cell, the target protein, the cellular ubiquitinylation system and the PROTAC form a ternary complex, and the recruited E3 ligase polyubiquitinates any surface-exposed lysines on the protein. This targets them for destruction by the proteasome and, once they have been eliminated, the PROTAC is released to complex with another POI molecule.
The potential is enormous. Unlike most drug targets, non-functional or weakly binding ligands can be used, as they don’t have to be antagonists or inhibitors. They merely need to bind to the protein in a manner that allows the recruitment of the E3 ligase. And whilst 20% of the proteome has some form of catalytic or receptor-based activity, PROTACs could give access to at least some of the remaining 80% of drug targets we currently cannot exploit.
As the E3 ligase needs to sit in a specific geometry to work, less selective ligands could be used if they orientate correctly with the protein of interest but not off-target proteins. Studies deriving PROTACs from non-selective ligands demonstrate that it is possible to achieve very specific degradation of a single protein even if the isolated protein ligand itself hits multiple targets and, as such, off-target effects might also be reduced or avoided.
But the design of PROTACS that can induce these productive, degrading complexes of the POI and the ubiquitinylation machinery remains somewhat empirical, and relies very much on a brute force, trial and error approach at present. Whilst computational simulations suggest that there likely isn’t just a single productive complex structure, and that many different orientations of the POI and the E3 ligase can lead to degradation, they also suggest significantly larger ensembles of unproductive conformations can occur. The protein dynamics of these systems, and ability of any specific PROTAC to sample all these different conformations is unclear. Given this, predicting the productivity of a potential PROTAC up front through rational design appears to be some way off yet.
Do the benefits outweigh the extra effort?
It’s often overlooked that the benefits of PROTACs can extend beyond simple removal of the biological target. For example, further selectivity may be attained through differential tissue expression of the chosen E3 ligase. In some cancers, for example, certain E3 ligases are over-expressed compared to normal tissue. Or perhaps the E3 ligase may be present in the diseased organ, but not in those where the POI ligand alone causes off-target toxicity. Tolerability and therapeutic index might, therefore, be better, and there is even the opportunity for prolonged pharmacology with sub-stoichiometric concentrations of PROTAC, which may give the desired pharmacodynamic effect if the rate of degradation is much faster than the rate of resynthesis.
This potential for sub-stoichiometric binding, and the ability of one PROTAC molecule to degrade many molecules of the POI, opens the opportunity to use POI binding ligands of more modest potency. And it is often overlooked that if the POI binding ligand has inhibitory properties alone, this activity is likely to be retained in the final degrader chimera and may still demonstrate all the off-target effects of the parent compound at stoichiometric doses. Here, then, the potential of PROTACs to be effective at lower doses offers an additional therapeutic benefit.
So, why not deliver PROTACs for every target?
The number of E3 ligases presents both a challenge, and an opportunity. At present, we have good recruiter ligands for just a small handful of the E3 ligase space. So whilst efforts are underway to map the tissue and disease-specific distributions and lineages of the ligase families, exploiting that knowledge may remain challenging in the absence of a systematic exploration of ligase binding chemotypes and their selective recruitment. Thankfully, efforts are underway to address these challenges, with teams like EUbOPEN beginning to advance a more holistic understanding of this area, and perhaps allow a more design-led approach to novel therapeutic PROTACs.
Whilst all drug discovery projects take time, there are additional challenges involved in developing a PROTAC on top of those facing a small molecule project. As a PROTAC falls outside the rule-of-5 space, absorption may be an issue, and the three-part structure adds to the complexity. There is, as yet, no simple alternative to making and testing molecules to make advances. Those in silico techniques that are routinely, and successfully, applied to small molecules are not sufficiently mature to apply to all PROTAC molecules.
Assessing PK/PD and efficacy relationships is harder still. Additional factors to consider include the concentration of the E3 ligase in the tissue, the resynthesis time of the protein, and even whether there is cooperativity in the ternary complex which allows the degradation to occur at all.
The key question – PROTAC or small molecule?
With all this in mind, when is the right time to switch to a PROTAC approach? And in which instances might the obvious choice be to start with a PROTAC?
Sometimes a PROTAC might be a better choice than a small molecule inhibitor at the outset, particularly if experiments using gene editing tools such as shRNA or CRISPR knockouts have been used to validate the approach. Often, small molecule catalytic inhibitors simply do not recapitulate the observed phenotype, and perhaps here a PROTAC may be a better therapeutic option compared to a traditional small molecule inhibitor approach?
The recent history of drug discovery is littered with many molecules which are potent binders but do not show the required in vitro or in vivo pharmacology, and perhaps these suggest both the start points, and the applications, for repurposing into therapeutically useful chimeras. Or perhaps the protein’s resynthesis time is found to be a few days rather than a few hours, and protein ablation may allow intermittent dosing, vs continual exposure for catalytic inhibition.
Protein degraders can also be more effective than protein inhibitors in the right context, such as the greater success with oestrogen receptor degraders than oestrogen receptor antagonists in treating breast cancer. And perhaps any disease characterised by protein build-up might prove promising for a PROTAC approach, as destroying these proteins may have a clinical effect. And with an appropriate ligand, perhaps PROTACs might differentiate between a mutant variant and a healthy protein and restore normal physiology. Could beta-haemoglobin be selectively degraded to treat thalassaemia or sickle-cell disease? Or might it even be possible to degrade the mutant calcium pumps in cystic fibrosis, or recognise and selectively degrade malfunctional splice variants whilst leaving the normal protein intact?
And what about neurodegeneration? There are indications that a PROTAC might be more effective in the treatment of cognitive decline, for example, or other diseases of the central nervous system that result from the accumulation of proteins. While there is the additional issue there of getting the PROTAC into the CNS, there are encouraging reports from Arvinas that this is possible, despite their size.
Where are we now, and where are we going?
Much work remains to be done. To date, the field remains very empirical, with little substitute for physical testing of the different options in the absence of computational predictions. There needs to be a better way of optimising the combination of components that make up a PROTAC, and predicting which are likely to succeed.
And the key challenge of converting an early PROTAC probe molecule into an in vivo relevant tool or a development candidate suitable for human dosing, remains significant, and largely based on brute force and trial and error. Cell permeability has traditionally been a significant challenge for PROTACs. Application to extracellular targets, effectively mimicking the action of therapeutic antibodies, has been little explored to see if such binding triggers the endocytosis and selective proteasomal degradation of these extracellular components.
Yet despite the challenges, PROTACs have moved quickly. They have evolved from a scientific curiosity at the turn of the century to the first in vivo validation in 2015, and three agents are already in the clinic. The first PROTAC, an androgen receptor degrader from Arvinas, entered clinical trials in 2019 to treat prostate cancer, and is now in Phase 2. Another Arvinas project, an oestrogen receptor degrader, entered the clinic later the same year, and is also now in Phase 2 for ER-positive, HER-negative breast cancer. A third from Kymera, this time targeting IRAK4 in inflammatory disease, entered Phase 1 in March 2021.
Many more are in the earlier stages of development, and other, related approaches such as lysosome-targeting LYTACs (and even selective RiboTACs, designed to remove specific RNA messages) are following on behind. The emerging area of molecular glues deserves a discussion in its own right, but whilst showing promise and perhaps being easier to optimise, these agents are most often discovered by chance, and the mechanism deconvoluted later. It may be that many of these glues are already available in our screening collections but we’re simply unaware of their existence, as we have yet to develop the tools to systematically identify and prioritise them as we now do routinely for small molecule inhibitors.
In summary, the key choice is when and when not to use PROTACS as a discovery and therapeutic tool; to be sure of the modality up front in order to justify the significant investment in optimisation and the pursuit of oral bioavailability. And also to understand which of the hundreds of E3 ligases to recruit to best deliver the required pharmacology. It’s important to be sure that PROTACs are indeed the right approach, rather than adding additional complexity to a target equally susceptible to small molecule inhibition. It will be fascinating to see how the lessons learnt from the current generation of PROTACs can be applied to that next generation of effective therapeutics.