Of the 75%+ of human proteins not currently targetable by drugs, some are known cancer drivers. Others may hold the key to arresting as-yet-incurable diseases. Could small molecule protein degraders break open the door to this novel target space and catalyse a new era of drug development?
We can talk about AI’s potentially transformative role in drug discovery all we like, but one major obstacle stands in its way: unlocking the undruggable human proteome. Only 20-25% of human proteins can be drugged by small molecules at present; we are all operating in black and white, when the true picture is in colour.
However, the targetable space could be vastly expanded by the arrival of new experimental therapies – namely gene- and RNA-targeting therapies and the newest addition, small molecule protein degraders. AI and predictive algorithms could play a major role in assigning relevant targets to those new therapies, including both infamous undruggable cancer drivers, such as Myc and Ras, and as-yet-unexplored targets.
Proteins are among the cell’s key functional units. They exercise and modulate cellular processes, from energy production to movement, from inter- and intracellular communication to cell division. The misregulation of protein abundance or activity is at the root of many diseases – and this, unsurprisingly, makes them attractive drug targets. In the case of several cancers, for example, the EGFR kinase becomes overactive – and a group of small molecule inhibitors such as Erlotinib (Tarceva) successfully target this kinase.
However, due to their structural features, only 20−25% of human proteins can currently be exploited as drug targets by small molecule drugs. And even the small molecule drugs that can target this accessible pool have certain limitations.
First of all, inhibiting a specific catalytic domain – as most of these drugs do – means ignoring the non-catalytic functions many proteins have. They may act as molecular scaffolds that bring together other signalling molecules, or they may have more than one catalytic centre (and, therefore, variable functions).1
Other problems lie in the a) specificity b) completeness and c) duration of the inhibitory effect exerted by small molecule inhibitors:
a) Small molecules that target a protein’s catalytic centre, such as kinase inhibitors, are often promiscuous, hitting multiple targets. This makes it difficult to assign a drug’s properties – good and bad – to the presence of a single factor, such as overexpression of a specific kinase in cancer.
b) Blocking a protein’s catalytic centre is based on occupancy-driven pharmacology, which means that high systemic drug concentrations might be needed for full effect, increasing the risk for off-target toxicities.
c) At the same time, effects are often not durable over longer periods of time, as resistance against inhibitors arises via multiple routes, including mutations in the catalytic centre.1,3 This is a well-known problem for kinase inhibitors in cancer.
How to overcome these problems? Instead of fishing in a small pond of protein targets that can be addressed with classical small molecule inhibitors, wouldn’t it be better to have the whole sea of drug targets accessible? Instead of merely inhibiting a single function of a given target protein, wouldn’t it be better to just get rid of the protein altogether?
Hijacking the protein garbage disposal system
To reduce the levels of malignant proteins, the system can be attacked on multiple levels. On the highest level, we have DNA, the original blueprint of the protein building plan, which is transcribed into the intermediary molecule RNA. This transcription process determines how much of a protein will ultimately be produced. Changes in protein-coding DNA can be achieved by gene therapy or CRISPR-CAS-9-based approaches.
One level down is messenger RNA (mRNA), which is translated into protein, and whose abundance and stability regulates how much of a given protein will be made. Reduction of mRNA levels can be achieved via different knockdown approaches, such as RNA interference (RNAi).
Lastly, the protein itself is subject to a number of so-called post-translational modifications – additions of small tags, such as glycosylations or phosphorylations, that impact its localisation, activity, function and turnover. The latter is regulated by the cell’s built-in degradation machinery, a sort of protein garbage disposal system, which can be exploited in a process known as ‘chemical knockdown’.
In the natural process, proteins are labeled with a ‘to-be-disposed’ tag (the small protein ubiquitin) which marks them for degradation by the proteasome. The tagging of proteins for this degradation system relies on the action of ubiquitination enzymes – with the class of E3 ubiquitin ligases (of which around 600 exist in human cells) conferring specificity to the system.2
In a chemical knockdown, small molecule degraders bring an E3 ligase and a target protein together, inducing the protein’s ubiquitination and degradation.
This new class of compounds could vastly expand the pool of druggable proteins, as they do not require the presence of a catalytic domain, and address many of the challenges of classical protein inhibitors outlined above. For example, its event-based mode of action – a single interaction can target a protein for degradation – gives it an advantage over occupancy-based pharmacology.1 Plus, targeting a protein with small molecule degraders can effectively inhibit both its catalytic and scaffolding functions, as shown in a preclinical study for focal adhesion kinase (FAK), a key player in cancer metastasis.4
Unlike a CRISPR-CAS-9- or RNAi-based reduction in target protein levels, a chemical knockdown with small protein degraders can utilise the benefits of small molecules. These include the possibility for oral dosing and systemic delivery without viral vectors or nanopolymers. Like RNAi approaches, chemical knockdowns are reversible, which might, in some respects, be preferable to the one-shot, long-lasting gene therapy approach. Nevertheless, due to their properties and size, small molecule protein degraders come with their own set of pharmacokinetics/ pharmacodynamics (PK/PD) challenges, and the system will need to be fine-tuned to accommodate them.1,3
Small molecule protein inhibitors – a drug developer’s dream come true?
Interestingly, while the concept of small molecule protein degraders is comparably new, one drug exhibiting these properties has been hiding in plain sight. Thalidomide’s past is chequered: prescribed as the morning sickness drug Contergan in the late 1950s, it caused serious tetraplegic birth defects in more than 10,000 children, and the death of 2,000 more before it was taken off the market in 1961. Celgene then relaunched Thalidomide – first as a treatment for leprosy, then as a blockbuster treatment for multiple myeloma, before the drug’s major target was discovered. Thalidomide and other members of the IMiD (Immunomodulatory imide drugs) class act as a ‘molecular glue’ for cereblon (CRBN), a substrate receptor of cullin-RING ligase 4, and induce degradation of oncogenic transcription factors.5
In the past five years the field of small protein degraders has seen a veritable gold rush6, marked by a number of high-priced deals between the biotechs spearheading the degradation space and big pharma players. Arvinas, which went public for $120m in 2018, has partnerships with Genentech and Pfizer, and recently signed a $115m cash/equity and joint venture deal with Bayer, for both biotech and agrotech.7 And in July 2020 Kymera inked a collaboration with Sanofi focused on its Interleukin-1 receptor-associated kinase 4 (IRAK-4) protein degrader for $150m in cash, with a further $2bn dependent upon milestones being reached.8
Dubbed PROTACs (proteolysis-targeting chimeras), LYTACs (lysosome-targeting chimeras, investigated by Stanford spinout Lycia Therapeutics9), SNIPERs (Specific and Nongenetic IAP-dependent Protein ERasers), or molecular glues (investigated by Monte Rosa Therapeutics10), small molecule protein degraders come in different flavours. All bring their target proteins into close proximity with some part of the cellular protein degradation machinery, although their modes in doing so differ.3
While PROTACs are heterobifunctional small molecules, which connect target-binding warheads to E3 ligase-binders via flexible linkers and depend on the presence of ligandable sites on the target protein, molecular glues can help degrade unligandable proteins by sticking them directly to the ligase.11 The LYTAC system aims to use the cellular endosomal/lysosomal system to target extracellular, as well as intracellular, proteins.9
Arvinas’s PROTAC degraders of two well-established nuclear targets, the Androgen receptor (AR – a key driver in prostate cancer) and the Estrogen receptor (ER – a key driver in breast cancer), are among the furthest advanced small protein degraders.12 In preclinical models, the AR-targeting PROTAC could overcome resistance against a commonly used AR-targeting prostate cancer drug, Enzalutamide.13 Initial clinical results of Arvinas’s AR PROTAC ARV110 have, however, been underwhelming: only one of 28 (albeit heavily pretreated and castration-resistant) patients showed a confirmed response.12 The company’s ER PROTAC ARV-471 produced a 90% ER reduction in breast cancer patients, but (so far) only one confirmed response out of 15 evaluable patients.12
This is a very early snapshot of data for the drug class. More relevant than PROTACs’ ability to address well-established targets in difficult-to-treat patient populations will be the question of whether PROTACs and other small molecule protein degraders can deliver on the promise of drugging the undruggable proteome.
A new realm of drug targets
Despite the promise of targeting as-yet-undruggable targets across different indications, the field is currently focused on the druggable proteome and is heavily invested in the oncology space.14,15 Pioneers wishing to step out of these shallow waters, however, could head in one of two directions.
One path leads towards the “holy-grail” – the undruggable targets, notorious cancer drivers, that have evaded pharmacological inhibition for decades. These include oncogenic transcription factors, such as Myc or the small GTPase Ras16, both of which are being tested at the preclinical stage by Arvinas and other players.3,14 The other points toward a host of alternative targets that might be addressed with small molecule degraders. These range from proteins involved in brain diseases, such as Tau17, to non-human pathogenic proteins used for fighting infections.1
Which of those targets would be most interesting to explore? And which will lend themselves to the PROTAC technology?
The molecular properties of the cellular degradation machinery determine which types of proteins and PROTACs can efficiently form a ternary complex – containing the target protein of interest, the small molecule and the E3 ligase – and thus induce efficient target protein degradation.18 Most current (semi-) computational approaches focus on predicting optimised efficacy of ternary complex formation and ubiquitination17-19, as well as identifying target proteins/small molecules that are suitable for the interaction with a small number of ubiquitously expressed E3 ligases, or have molecular glue-like properties.9,19,20
One of the key challenges in the field is identifying new targets for degradation, as well as novel E3 ligases that could interact with small molecule degraders of those targets working specifically in diseased cells or tissues. More than that, predicting the properties of the respective system, including potential off-target effects and PK properties, will help to identify promising candidate molecules to move into the clinic.15
Where does AI fit in?
Computational methods can help to predict new targets by identifying proteins that:
- have the suitable structural properties and (re)synthesis rates to be targetable by small molecule degraders18
- play a relevant role in a given disease context
- cannot be addressed by inhibitors that target catalytic centres
Although chemical knockdown might not be a suitable option for some of those targets, alteration of the genetic or RNA levels by gene therapy or CRISPR-CAS-9, as well as RNAi, might be an option to complement the small molecule based knockdowns. Combined, these methods might provide a much fuller spectrum of opportunity for drug developers, opening the door to a new era of drug discovery.
The race is on to find new targetable areas in the currently undruggable human genome and proteome – and bring fresh hope to patients with (currently) incurable diseases.
- Short Review article describing the underlying biology and technology, key benefits and major challenges ahead for small protein degraders, published in 2020.
- Nature News Feed article describing the history of PROTACs, from early research efforts by Crews’ group to newer developments. Published in 2019.
- Review article discussing currently pursued PROTAC targets across different therapeutic areas.
- Research article, which discusses addressing non-kinase functions of the focal adhesion kinase (FAK) in preclinical experiments. Published by Craig Crews’ group in 2018.
- Review article by Hirosih Handa’s group from Tokyo Medical University, who discovered that Thalidomide targets cereblon, discussing the cereblon-targeting functions of the IMiDs drug class.
- Opinion piece by Raymond Deshaies, one of the founding fathers of the field, marking the beginning of the PROTAC “goldrush”. Published in 2015.
- Description of the 2019 deal between Bayer and PROTAC pioneer Arvinas for both R&D and agrotech partnership.
- Article discussing the partnership between protein degrader-focussed biotech Kymera and Sanofi for their Ph I-ready IRAK-4 inhibitor in July 2020.
- Endpoints article discussing Lycia’s approach to protein degraders exploiting the endosomal and lysosomal machinery.
- Endpoints article discussing the molecular glue approach of Swiss biotech Monte Rosa and its selection of a first target to advance into the clinic.
- Research article from the group of Georg Winter and collaborators, discussing a multi-omics, chemical screening approach for targeted discovery of molecular glue compounds.
- Investor presentation from Arvinas (Dec 2020) presenting approach and initial clinical data on their AR and ER PROTACs.
- Research article on preclinical data of an androgen receptor targeting PROTAC, compared to the approved drug Enzalutamide, which also targets the androgen receptor and is a mainstay treatment in prostate cancer.
- Review article published in 2019, discussing different targets that have been addressed using PROTACs by industry and academic groups.
- Short perspective by Mariell Pettersson and Craig Crews, discussing current state and future trends for PROTACs
- Viewpoint article in which four scientists discuss undruggable targets in cancer and highlight the most crucial advances and challenges for the field.
- Expert opinion on PROTACS published in 2019, discussing in detail the potential of PROTACs in targeting Tau.
- Review article, published in 2019, which discusses the molecular properties of the ternary complex formation.
- Research paper describing computational methods for predicting ternary complex formation and target protein degradation efficiency.
- Research paper describing PRosettaC a computational tool for predicting ternary complex formation and target protein degradation efficiency.
- Research paper published in Science 2018, describing a screening approach to identify zink finger transcription factors that can be modulated by Thalidomide-analogue-mediated interactions with CRBN, combined with computational zinc finger docking and biochemical analysis.
- Research paper published in Nature 2020, describing an approach that relies on database mining to identify correlations between the cytotoxicity of clinical and preclinical small molecules and the expression levels of E3 ligase components across hundreds of human cancer cell lines.