Credit: S.Harris/Springer Nature Limited

With the rise of targeted protein degraders over the past decade, early adopters promised that these small molecules would be able to unlock previously intractable targets. A first wave of molecular glue and heterobifunctional degraders mostly focused on well-validated targets. A second surge is now pushing into more novel target space.

“We're on the cusp of a revolution,” says Neil Bence, head of oncology discovery at Bristol Myers Squibb (BMS), which is using both molecular glues and ligand-directed degraders to breakdown novel targets in cancer and other indications.

Traditionally hard-to-drug targets — including transcription factors, GTPases and guanine nucleoside exchange factor (GEFs) — are increasingly within reach, shows the growing degrader pipeline (Table 1).

Table 1 | Degraders move into novel target space

Target

Target properties

Molecule (degrader type)

Company

Indication

Status

Newly prosecuted targets

GSPT1

GTPase, translation termination factor

BMS-986497 (antibody–glue conjugate);

MRT-2359 (glue);

CC-90009 (glue)

BMS/Orum;

Monte Rosa;

BMS

Haematological malignancies;

MYC-driven cancer

Phase I;

Phase I/II;

Discontinued

VAV1

GEF, scaffold protein

MRT-6160 (glue)

Monte Rosa

Autoimmunity

Phase I

Not disclosed

Transcription factor

HbF-activating CELMoD (glue)

BMS

Sickle cell disease

Phase I

WIZ

Transcription factor

NA (glue)

Novartis

Sickle cell disease

Preclinical

BCL6

Transcription factor

ARV-393 (heterobifunctional);

BMS-986458 (heterobifunctional)

Arvinas;

BMS

B-cell malignancies

Phase I;

Phase I

STAT6

Transcription factor

KT-621 (heterobifunctional)

Kymera

Allergic diseases

Phase I in 2024

IKZF2

Transcription factor

Helios CELMoD (glue);

PLX-4545;

DKY709 (glue)

BMS;

Plexium;

Novartis

Cancer

Phase I;

Phase I;

Discontinued

HuR (ELAVL1)

mRNA stability regulator, RBP

NA (glue)

Degron

Cancer

Preclinical

Previously prosecuted targets, without approvals

IRAK4

Kinase, scaffold protein

KT-474 (heterobifunctional)

Kymera/Sanofi

AD and HS

Phase II

LRRK2

Kinase, scaffold protein

ARV-102 (heterobifunctional)

Arvinas

Parkinson’s disease

Phase I

STAT3

Transcription factor

KT-333 (heterobifunctional)

Kymera

Cancer

Phase I

MDM2

E3 ligase

KT-253 (heterobifunctional)

Kymera

Cancer

Phase I

RBM39

Splicing factor, RBP

NA (glue)

Seed

Cancer

Phase I in 2025

NEK7

Kinase

MRT-8102 (glue);

NA (glue)

Monte Rosa;

Novartis

Inflammation

Preclinical;

Preclinical

This is enabling molecular glue degraders — small molecules that reshape an E3 ligase to make it tag targets with ubiquitin, shunting problem proteins to the cell’s proteasomal recycling system — to expand beyond their oncology origins. BMS is testing a transcription-factor-degrading glue for sickle cell disease, while Monte Rosa has advanced its VAV1-targeted GEF degrader into the clinic for autoimmune diseases.

Heterobifunctional degraders — larger dumbbell-like molecules that bind a target of interest with one end and an E3 ligase with the other — are making headway in novel target space too. Kymera is advancing a first-in-class degrader against the immune-mediating transcription factor STAT6, for example, while both BMS and Arvinas are taking on the oncogenic transcription factor BCL6.

Zoran Rankovic, director of the Centre for Protein Degradation at the Institute of Cancer Research, is buoyed by this progress. Degraders against previously drugged targets could be a boon to patients, he explains, if they can outperform approved inhibitors. But most of the human proteome is still undrugged, and the bigger opportunity for degraders is to push these boundaries.

The field has a way to go, he adds. Glue degrader discovery remains limited as yet mostly to serendipitously identified targets, and heterobifunctional degraders remain constrained by ligandability issues and rational-design limitations. But researchers are making progress across the entirety of the degrader modality.

“This is a hype that actually lives up to its promise,” says Rankovic.

Old glues, new clues

Interest in targeted protein degraders has exploded in the past 10 years, and dozens of companies are now operating in this space. While heterobifunctional drug discovery companies were faster out of the gate, the ranks of the glue degrader biotechs are growing too — fuelled especially by the field’s understanding of how the FDA-approved myeloma drug lenalidomide and related immunomodulatory drugs (IMiDs) bind and reshape the E3 ligase cereblon to ubiquitinate the transcription factors IKZF1 and IKZF3. Other small-molecule glues might be able to reshape cereblon to take on other targets too, researchers quickly realized.

The first programmes to advance into the clinic, however, took on targets that were also degraded by lenalidomide. Celgene, now part of BMS, for example, worked quickly with its lenalidomide analogues to discover and optimize CC-92480, now mezigdomide, to breakdown IKZF1 and IKZF3. That drug is now in phase III development for myeloma. The kinase CK1α was another low-hanging fruit that is degraded by lenalidomide.

A further stepping stone was GSPT1, a GTPase that researchers pulled down during an immunoprecipitation assay of cereblon and a lenalidomide analogue. GSPT1 helps the protein-making machinery to disengage from completed proteins, and its blockade kills cells — especially fast-growing cancerous ones — creating oncology applications for the previously undrugged GTPase target. BMS first advanced its GSPT1 degrader CC-90009 into the clinic in 2016, but has since terminated that glue for undisclosed reasons.

“GSPT1 degradation shuts down global protein translation, and there are a number of adverse events that are likely to be associated with that,” cautions Ian Churcher, a consultant with Janus Drug Discovery and a former degrader developer at both Amphista and GSK. “It's all about therapeutic index.”

At BMS, that now means using an antibody–glue conjugate to better deliver the degrader to cancer cells. Its BMS-986497, acquired from Orum Therapeutics, consists of a GSPT1-degrading glue tacked on to a CD33-targeted antibody, to home in on malignant B cells.

“To improve both the efficacy and tolerability of GSPT1 degradation, an antibody–conjugate approach would be ideal,” says Bence. “We're excited to see how this type of approach performs. It’s a really exciting time right now for degrader–antibody conjugates.”

BMS has also moved a glue degrader forward against another transcription factor for sickle cell disease, but as yet has not disclosed its target. “Stay tuned,” says Bence.

A cereblon-based glue degrader that targets the transcription factor WIZ can boost fetal haemoglobin levels in mice and primates, Novartis reported this year, showcasing one way a glue could be useful in sickle cell disease.

Target hopping

Monte Rosa was another early mover against GSPT1, developing MRT-2359. Clinical data as yet shows that this glue has a viable therapeutic index and a tolerable safety profile in patients with MYC-driven solid tumours.

As the company has built out its degrader platform, however, more distant target opportunities have emerged. The first of these is VAV1. “It is a target that I'm pretty sure every immunology group in big pharma at some point in the last 10 to 20 years has looked at,” says Monte Rosa CEO Markus Warmuth. “This seemed like the perfect thing to work on.”

VAV1 is a GEF that helps to unload GDP from the GTPase RAC. A growing body of data links the GEF to B- and T-cell signalling, both by enabling RAC signalling and by acting as a scaffold for other signalling proteins. The immunosuppressant azathioprine, approved decades ago for rheumatoid arthritis, locks RAC into an inactive state — achieving a similar effect as VAV1 inhibition. But although researchers have speculated that VAV1 inhibition could bring unique benefits, the protein lacks a clearly defined binding pocket and acts via protein–protein interactions, making it a med chem nightmare.

Monte Rosa — like others in the glue space — often uses unbiased screening to find out what its library of compounds can degrade, in the hopes that a therapeutically relevant protein will pop up. “The proteomics approach is really great in identifying new target opportunities,” adds Sharon Townson, the company’s CSO. But the company also screens its library against targets of interest, guided especially by its understanding of the types of surfaces that its glues can bind to. The VAV1 programme stemmed from a combination of these approaches, she adds.

“When we first saw the proteomics hit, it was barely statistically significant,” says Townson. But the team’s structural biologists had predicted that a domain of VAV1 had surface mimicry with GSPT1, she says, so they persevered. “We found this really great starting chemistry from a library screen. It was almost like lead-like chemistry from the beginning.”

The resulting VAV1 degrader MRT-6160 has preclinical efficacy in animal models of various autoimmune diseases, adds Townson, with comparable effects in some cases to standard-of-care JAK inhibitors and sphingosine 1-phosphate receptor modulators.

Monte Rosa just started a phase I trial of MRT-6160 in healthy individuals, and has future trials planned in autoimmune diseases.

“This programme seems quite unique in the target landscape,” says Tudor Oprea, CEO of Expert Systems, which uses AI tools to de-risk preclinical drug discovery. “At the end of the day, VAV1 is part of a pathway that's been explored by other programmes though, and only time will tell which node will actually turn out to be the right choice to drug.”

VAV1 exemplifies, moreover, how target hopping can expand glue degrader frontiers. The transcription factors IKZF1 and IKZF3 are canonical cereblon neosubstrates, and rely on a G-loop degron to engage with cereblon. Hundreds of transcription factors have G-loop-containing C2H2 zinc fingers — including both IKZF2 and WIZ — providing a rich vein of potential targets, notes Rankovic. But cereblon neosubstrates in non-transcription factor proteins, like GSPT1 and CK1α, carry G-loop degrons that put them within reach of the IMiDs too. And although VAV1 doesn’t have a G-loop degron, it looks enough like GSPT1 to be in scope.

“When we identify novel binding surfaces we can ask ‘what else looks like that?’,” says Townson. “It’s not a perfect process yet, but we are as a company identifying a lot of novel binding modes, which then helps us reach further into target space.”

More than 1,600 proteins, from many target classes, might as a result already be degradable by cereblon-binding glues, Townson and colleagues recently predicted in a preprint.

Serendipity now

And yet, a lot of luck is still involved in expanding the molecular glue degrader space. “What's mostly happening is that people are doing unbiased screens, and are finding novel proteins and then saying, ‘Ah, interesting. Let's go after it’,” says Kymera Therapeutics CEO Nello Mainolfi, who works on both glues and heterobifunctional degraders. “We’re doing that too with glues, and it’s not a bad strategy.”

This mindset lends itself to the exploration of novel target space, adds Rankovic, especially given that most proteins remain undrugged. “Right now, I think molecular glues are giving us a better chance of identifying novel targets,” he says.

But it is a far cry from a target-first approach to glue discovery. “There are a number of companies that claim that they're doing amazing stuff with rationally designed glues, but we've still got to see the data really,” says Churcher. “Truly applying glues across the whole proteome is going to be a long journey.”

One way to move in that direction could be to recruit other E3 ligases. The body uses over 600 E3 ligases to recycle proteins, but the drug discovery community mainly works with cereblon. “There’s no other ligase that is so reprogrammable to such a variety of different proteins,” says Warmuth, explaining why Monte Rosa remains committed to the cereblon-based degraders.

SEED Therapeutics is amongst those who are nevertheless working to let other E3 ligases shine. Its lead programme harnesses the DCAF15 ligase to degrade the splicing factor RBM39. This programme builds on over 25 years of research on aryl sulfonamide small molecules, adds SEED president and CSO James Tonra. In 1999, Eisai reported that its indisulam stalls cell cycle progression in cancer cells — prompting a failed attempt to develop the drug as a chemotherapy candidate. In 2017, researchers reported that this class of drug in fact acts by remodelling DCAF15 to ubiquitinate RBM39, a protein that regulates the splicing of mRNA precursors.

Armed with a better understanding of RBM39 biology, SEED is set to advance an optimized RBM39 degrader into the clinic next year.

“There’s a big opportunity for RBM39 degraders in the clinic for new indications, in everything from neuroblastoma to liver cancer,” says Tonra.

SEED’s broader approach, he adds, is to first hunt for weak interactions between a target of choice and an E3 ligase, and then to find small molecules that can strengthen that interaction. “Those weak interactions are something that we just didn't pay attention to before, because they weren't functional,” says Tonra. “For molecular glues, they are the key to success.”

SEED has 9 preclinical programmes in its public pipeline, he adds, and none of these rely on cereblon or VHL, another well-studied E3 ligase. Only the RBM39 programme uses DCAF15. “If there's already an interaction that has some ubiquitination, you can strengthen it and increase the ubiquitination,” he adds. “The data is making us believe.”

Rankovic sees merit in this sort of approach. As further proof of potential, he points to a recent preprint that identified a KRAS degrader that acts by gluing the oncogene to its native E3 ligase Nedd-1. “We will see more and more approaches like this,” says Rankovic. “Ultimately, every protein in the human genome could be accessible to this kind of approach, because most proteins are substrates for some kind of a degradation or recycling process.”

Protracted progress

Heterobifunctional degraders are also moving into new target space.

An early promise of these drugs was that they could be used to destroy nearly any target that a medicinal chemist could bind a ligand to. But most of the first programmes into the clinic took on drugged targets. The first proteolysis-targeting chimera (PROTAC) into the clinic targeted the androgen receptor (AR) for prostate cancer, and at least 5 other AR-directed degraders have followed it into human trials.

This trend of focusing on validated targets rather than novel ones may have been driven by the trade-off between ‘technology risk’ and ‘biology risk’, says Rankovic. Heterobifunctional molecules are big and bulky, and PROTAC pioneers wanted to prove that these ungainly molecules were nevertheless viable and safe drugs. “Thankfully, what we've seen is that with some of the clinical compounds we’ve had a difficulty in identifying dose-limiting toxicities. This is really fascinating and very promising,” he adds.

But efficacy data has been less convincing.

In the case of prostate cancer, approved AR antagonists like enzalutamide lose activity after cancer cells upregulate androgen or AR expression. A protein-depleting drug with catalytic activity, researchers hypothesized, should be able to silence AR even in those contexts. But drug developers are still hunting for this efficacy benefit. (Arvinas has discontinued work on its once-lead AR degrader bavdegalutamide, and earlier this year Novartis acquired rights to the firm’s better backup AR degrader, ARV-766.)

“The early data just hasn't really played out in the way that the field wanted, and it has taken a bit of the wind out of the field’s sails,” says Mainolfi. “People have started questioning whether protein degradation really is a transformative technology. My answer has always been that it depends on the target you go after.”

At Kymera, that means taking on more biological risk. “Our goal has always been to go after targets that have not been drugged before by marketed products, where protein degradation would provide a technical solution to unlocking new target biology,” explains Mainolfi.

This can be done while also mitigating biological risk, he adds. Kymera’s first programme into the clinic targeted IRAK4, a kinase in the well-understood and already drugged inflammatory IL-1 pathway. Drug developers have suspected that IRAK4-targeted drugs could have utility too, but several small-molecule kinase inhibitors have failed to prove this potential — possibly because the protein also serves as a scaffold for a complex that mediates inflammatory signalling. Kymera’s IRAK4 degrader KT-474 is now in a phase IIb trial, testing whether blockade of IRAK4’s kinase activity and scaffolding biology will yield an approvable drug.

Kymera’s next clinical entrant, a STAT6 degrader called KT-621, ups the novelty factor again. “STAT6 has been the holy grail of immunology for the past 10 years,” says Mainolfi. “But nobody thought this target could be drugged.”

The STAT6 transcription factor sits downstream of the IL-4Rα receptor, modulating allergic immunity signalling. Sanofi and Regeneron’s IL-4Rα-targeted antibody dupilumab (Dupixent) has proven the power of modulating this pathway in eczema, asthma and COPD, and the injectable antibody is on track to become the world’s best-selling immunology drug. But a STAT6 degrader could offer the same efficacy profile, says Mainolfi, with the added benefit of being available as a pill.

“This could be the first immunology programme where an oral drug doesn't have to compromise on efficacy,” he adds.

KT-621 is set to enter a phase I trial in healthy volunteers later this year.

Heterobifunctional heterodoxy

As yet, Kymera has disclosed few details about the makeup of KT-621. “It's a picomolar degrader that is exceptionally selective and more potent than a monoclonal biologic,” says Mainolfi. Once the company discloses the drug’s structure, a key question will be how and where it binds with the STAT6 transcription factor.

Transcription factors are hard to drug because they don’t have deep active pockets, acting instead through protein–protein and protein–DNA interactions to control gene expression. In the early days of the heterobifunctional gold rush, chemists hoped that these degraders might fare well against these types of targets. Even a weak binder of a shallow groove should suffice to trigger ternary complex formation and ubiquitination, the theory went. “But that hope has not panned out,” says Oprea.

Degrader developers, instead, still need good binders as starting points for the rational design campaigns. All too often, that means known inhibitors. “In the early days of PROTACs, you had to start with an inhibitor of a target,” recalls Churcher. “And there's still an awful lot of companies using existing inhibitors to make PROTACs,” he adds.

After all, novel ligands are hard to come by. In 2016, recalls Mainolfi, Kymera had its eye on around 30 unliganded transcription factors as possible degrader fodder. Traditional drug discovery techniques had already failed against these targets, so Kymera turned to a DNA-encoded library partner company to prosecute this wish list. They didn’t get a single hit. “It seemed like this was impossible. To be honest, I thought we’d have to give up,” he says.

Instead, Kymera built up internal screening capabilities. “We're not doing a carpet-bombing approach. We select the targets we're going after, and we screen them to find chemistry,” says Mainolfi. “We have succeeded in all the transcription factor screening that we've undertaken in the past 2 to 3 years,” he adds.

Kymera has learned during this process to prioritize neutral, or silent, ligands to bind their targets of interest. “We have many examples within the company where we're using totally allosteric, non-active binders to proteins,” he adds. “I feel strongly the field will move in that direction.”

On the BCL6 front, meanwhile, researchers have reported various ligands and small molecule inhibitors for this much-watched oncogenic transcription factor. But so far none of these have been publicly advanced into the clinic. Arvinas and BMS’s BCL6 degraders, consequently, provide the first therapeutic test of this target.

Another way to grow the PROTAC target space, says Churcher, could be to embrace larger target-binding warheads. Heterobifunctional degraders were once spurned by medicinal chemists as oversized and unsightly dumbbells, but even gawkier three-pronged trivalent and cross-shaped tetravalent degraders are now popping up in the literature. Degraders with molecular weights of over 2,000 kDa have been shown to work well in cells, offering up “sensible in vivo pharmacokinetics,” he adds.

Companies who take this approach may have to give up on oral bioavailability, he adds, but an injected development candidate is better than no candidate at all.

“If you relax your view of what a good binder needs to look like, you will find some great development candidates,” says Churcher. “If you want to talk about novel targets, I wouldn't count out PROTACs yet.”