Sunrise over Boston and the start of the biggest TPD meeting of the year. It’s not often that organisers of big international conferences give you carte blanche to organise a day of exceptional talks but I was given that opportunity to invite some of the best and most insightful scientists in the TPD field in today’s “New Frontiers in TPD & Induced-Proximity” opening day at the Hanson Wade 7th TPD & Induced-Proximity conference in Boston. And what a day it was – here’s a few of my personal highlights and thoughts on where the latest TPD innovations can take the field.

Who else better to open proceedings than Stuart Schreiber (ex Harvard, Broad and now ArenaBioWorks) who, with his work on FK506, cyclophilins and other glues, laid a lot of the foundations for modern TPD work as far back as the late 1980s. Stuart reminded us that when a small molecule binds to a protein, it can be thought of as producing a “neo-protein” which may change its interactome completely, especially if the small molecule can mimic a post-translational modification (more on that later). In drug discovery, we have a lot of small molecules which bind to proteins so it may be worth looking to see what they do in addition to what we designed them for (say, inhibiting the protein function) – if we can do this at scale, many pleasant surprised await I’m sure. We often say how difficult it can be to discover glues but it’s likely we have lots already but just don’t know it. 

Amongst a blizzard of vignettes from Stuart on more recent work, we were reminded that (FK506-based) glues can target proteins with virtually no tertiary structure at all (eg CEP250) - This is truly off in the deep blue water of low ligandability and may get us to reappraise what could be truly undruggable at all.

Also, by running DEL screens using PROTAC libraries on a given target in the presence and absence of effector (eg VHL), highly co-operative, almost glue-like (a>1000) PROTACs can be found straight out of a screen.

Finally, bifunctionals can be targeted to cells which selectively overexpress one or both of their binding partner proteins (cf RIPTACs). This characteristic, akin to ADC-type targeting (without the antibody and using intracellular “antigens”), could yet turn out to be one of the most powerful aspects of bifunctional drug design.

Continuing the theme of getting bifunctionals to do neat things, Jian Jin (Mount Sinai, NY) related a succession of stories including his design of “bridged” PROTACs (which could also be thought of as “collateral degraders”) where the protein directly recruited to the E3 by the PROTAC doesn’t actually get degraded but instead a binding partner does. Eg a VHL PROTAC with a short linker to CDK4/6 ligand palbociclib allows degradation of CDK4/6 partner cyclin D1 more efficiently than CDK4/6 itself. By playing with the linker, the ratio of degradation of the directly bound protein (CDK4/6) to the indirect target (Cyclin D1) can be varied. 

This is a neat approach (cf cyclin K degraders like CR8 which bridge from DDB1 to cyclin K via CDK12). Is your favourite target undruggable? – No problem, as long as it binds to a ligandable target, you can use that to bridge across to get ubiquitylation and degradation of the pesky undruggable protein.

Jian also described a foray into DUBTACs to stabilize, rather than degrade, proteins via deubiquitinase recruitment. Although active site, non-covalent DUB (USP7) inhibitors were used, efficient deubiquitylation was still seen even though the DUB is functionally inhibited when engaged in the ternary complex. This shows the power of “catch and release”-type bifunctional mechanisms where the ternary complex itself is functionally inactive but instead acts to maintain a local high concentration of free (ie uninhibited) DUB which can then deubiquitylate the target which is still nearby (ie bound to the other end of the DUBTAC). NB this neat trick may not apply to covalent active site inhibitors – the “catch” may work but the “release” may not.)

It's always a delight to see the latest work from Christina Woo (Harvard) in understanding why cereblon is a good E3 ligase for TPD. Christina reprised her now well-known story that CRBN may have evolved to recognise a C-terminal degron resulting from chain cleavage following cyclization of asparagine or glutamine residues. This may explain why if you use a CRBN-binding PROTAC or glue to present an IMiD motif on the surface of your protein of interest, CRBN quickly takes an interest and mistakes your PROTAC-bound target (ie neo-protein) for one of these aberrant proteins bearing this “eat me” post-translational modification. The same may be true for VHL which is extremely good at recognising proteins (eg HIF1a) bearing specific hydroxyproline motifs – something that is mimicked by a VHL PROTAC-bound neo-protein. Christina also went on to describe some newer work with GSK which has identified a new allosteric binding site of CRBN which appears to modulate its function – watch this space to see how that observation can be harnessed further.

After their respective talks, I also had the pleasure of being joined on stage by both Alessio Ciulli (CeTPD, U Dundee, flying the flag for VHL) & Eric Fischer (Dana Farber & Harvard Med, tour de force on CRBN) for a panel discussion where we tackled topics thrown up by the audience including glue design (increasingly possible for CRBN though still some very subtle selectivity observations which may be difficult to predict), use of machine learning for glue discovery (probably not enough available good data in a controlled, standard format to yet be useful enough to train models), potential clinical teratogenicity risk for CRBN PROTACs (data suggests can get away from SALL4-based effects but will this be enough to persuade regulators that risk is reduced rather than assume guilt by association with anything which looks similar enough to thalidomide?) and use of degrader antibody conjugates (many more TPD companies are now going down this route - could be powerful way to target degraders to desired cells to potentiate catalytic efficacy and reduce further any off-tissue degradation tox potential – certainly a lot easier than finding the mythical functionally broad-acting, tissue selective E3 ligases).

Collectively, all the talks today (and those where I’ve run out of space to describe from Kylie Walters [NIH, effects of proteasomal Rpn13 binders] & Vivek Vishnudas [BPGBio – E2 recruiters]) showed the enormous diversity of insight and pharmacological tools and tricks we now have available courtesy of TPD mechanisms. Drug discoverers and chemical biologists are very lucky to be able to call on this arsenal of tools and I look forward to seeing more great applications in a range of TPD areas over the next few days.