And that’s a wrap. The Boston TPD conference ended with a “Next Generation” day of a wide range of newer approaches to induced-proximity including some non-degrader plays. These strategies were all earlier in development but what they lacked in terms of clinical data, they more than made up for with excitement and potential.

After a major focus on small molecule degraders in the previous few days, a number of today’s talks focused instead of using protein-based agents for target degradation. This is certainly a very sensible thing to do…if you can deliver your therapeutic agent to where it needs to get to. The target scope of many of these biologic approaches can certainly be very different to small molecule PROTACs & glues but a few other issues start to creep in giving us a nice chance to balance risk and opportunity across a portfolio of TPD targets.

Work on targeting extracellular and cell surface proteins for degradation is a few years behind their intracellular-degrading cousins (as I wrote about here) but there’s an increasing number of groups trying this now.

Maureen Spit (Laigo Bio) showed some encouraging data with SureTACs (Surface removal targeting chimeras) which are heterobispecific antibodies which simultaneously bind membrane-associated E3 ligases (eg RNF43/128/130/167 etc) and an extracellular epitope on the protein of interest. There may be up to 60 E3s which hang out in the membrane and many have different expression profiles allowing you to introduce some level of tissue biasing. Targets with no blocking antibodies are the place you’d think his approach would be most useful and these can be found using single chain/nanobody-based motifs. There’s mix and match needed to find the best E3-POI combo but that can easily be screened for and Maureen showed removal of PD-L1 as a test case using RNF128 (1nM SureTAC internalised all PD-L1 in 3h). Some nice confocal movies showed PD-L1 being hauled off to the lysosome then disappearing which is satisfying to see. The in vivo half-life is likely to be monoclonal-like and efficacy studies are running now. A potential wrinkle is that the E3 is often co-degraded so catalysis is not assured (though Maureen did caution that the mechanism can be complex…) which led on to Natalie Nairn (Cyclera) who described the use of related bispecifics which piggyback the transferrin receptor to get into cells (and into the CNS via transcytosis) using what Cyclera call CYpHER (Catalytic pH dependent endolysosomal delivery with recycling). The name says it all with the target protein dissociating from the CYpHER as the pH drops during the journey to the lysosome while the CYpHER, still bound to the transferrin receptor is recycled to the cell surface (control studies confirm apparent recycling to give effective catalytic effect). The transferrin receptor looks like a useful shuttle to use as its expressed on all tumors and the blood brain barrier and the bispecifics tolerate a range of linkers with lengths from 40A to 270A. Using probe target EGFR, most receptor is internalised in 30mins though the lysosome struggles to keep up and takes 24h to clear only half of the internalised protein. This may not be an issue as the CNS targets which could be a good fit for this may not need complete ablation, rather just partial reduction/normalisation.

Next up was a pair of talks that didn’t try to piggyback receptor-mediated routes to the lysosome but instead aimed to get protein-based degraders into cells by brute force using cationic lipid nanoparticles. There’s been a lot of work in this area which works ok getting to the liver but other tissues can be more of a challenge. Jaehyun Choi (Genexine, recently acquired EPDBio) delivered mRNA-encoding “bioPROTACs” which are essentially truncated E3 ligases with their substrate-binding domain removed and replaced with a bespoke POI-binding domain (eg nanobody). Genexine have a panel of 26 E3 ligases they use in their bioPROTAC fusions – interestingly CRBN was described as a so-so ligase with many others (not disclosed unfortunately) working much better. A STAT3 degrader gave 0.1nM DC50 with the hope that it may be safer than the small molecule PROTACs though delivery to target tissue may be non-trivial. Some unexpected cytotoxicity on nanoparticule delivery was also noted as well as some autoubiquitylation of the bioPROTAC with the potential to trigger self-destruct mode so there it’s not all plain sailing.

Andrew Tsourkas (UPenn) is taking a more direct approach to bioPROTACs (in this case antibodies fused to the E3 ligase IPAh9.8), namely delivering the protein in lipid nanoparticles directly instead of mRNA. Getting proteins to cross membranes has been one of the long-standing problems in drug development but Andrew took an interesting approach. Recognising that neutral proteins can’t easily be loaded into cationic nanoparticles (like negatively charged oligonucleotides can), he fused anionic polypeptide domains onto the antibody heavy chain and showed the resultant bioPROTACs can be loaded into nanoparticles and escape into the cytosol, degrading targets like KRAS. The potency will likely need to be improved (from current ~20nM DC50) but smaller binders (darpins, nanobodies) could also be used to cram more bioPROTAC into the nanoparticles. The protein based bioPROTAC seemed to be a faster degrader than its mRNA cousin though proteomics showed both approaches had their fair share of off-target degradation to manage and reduce.

A pair of talks then looked at non-degrading proximity-inducers. Kanak Raina (Halda) reprised the RIPTAC (Regulated Induced Proximity Targeting Chimera) concept. A powerful way to use a tumor marker (eg AR for prostate tumors) as an “intracellular antigen” to enable targeting of a cytotoxin (eg PLK1 inhibitor, BRD4 inhibitor etc) using co-operative bifunctionals which look a lot like PROTACs sans the E3 ligase part. As the most common mechanism of AR resistance is AR overexpression, this actually could make RIPTACs work better than worse as the higher AR levels suck the RIPTAC into prostate tumor cells even more effectively. 

The RIPTACs give selective cytotoxicity to tumor cells (eg AR +ve) through a combination of three factors: i) co-operative complexes mean the RIPTAC has higher affinity to the ternary complex when both proteins (eg AR & BRD4) are present in high levels; ii) this effect directly leads to increased intracellular drug concentration, effectively concentrating the drug in desired cells (cf “intracellular ADC”) and iii) a less well-defined gain of function toxicity which may be due to the likely induced transcriptional chaos caused by recruiting a transcriptional regulator like BRD4 to AR-dependent gene loci. IND entry for the lead RIPTAC is on track for mid 2025 and I’ll be watching with interest – this approach is amongst the best of the non-degrading induced-proximity strategies I’ve seen.

Also on the non-degrading front, Rick Ewing (Rapafusyn) described how libraires of rapamycin analogs termed Rapaglues (essentially a FKBP12 binding motif with a loop of amino acids which can give cell-penetrant macrocycles) can glue FKBP12 to specific proteins to inhibit their function. The Rapaglues can be built into DEL libraries (they have 8 billion Rapaglues) as a suitable exit vector fortunately exists – using DEL for finding glues can be tricky if the glue molecule needs to be linked to DNA as there’s often no room for the exit vector if the glue is buried at the PPI interface. The Rapaglue macrocycles take full advantage of cyclosporine-like conformational chameleonicity to try to get into orally available space with some subtle changes (eg single atom lactone to lactam switch) giving huge conformational diversity.

For a completely different approach for degrading secreted or intramembranous proteins, Pat Sharp (Gate Bio) described their efforts to use small molecule modulators of the Sec61 translocon (the pore which allows proteins to move from the ribosome into the ER & secretory pathway). Through discovering binders which subtly change the shape of the translocon channel, they can persuade the channel to prevent certain substrates (with particular signal peptides) from transiting. There’s over 4000 proteins which use the Sec61 translocon so selectivity will of course be key but the Gate team seem to be doing a good job teasing out a small subset of proteins for selective blocking. As selectively blocking or gating the channel will prevent the nascent protein supply from making it into the secretory pathway, the drugs can act sub-stoichiometrically, potentially matching the PROTAC or glue catalytic advantage.

Finally, I must mention the Phoremost work - as Ben told me I had to ;-)  Ben Cross described the GlueSeeker approach where small loops (eg 3-10 amino acids) can be engineered into E3 ligases then screened to look for glue-like activity where the new loop creates just the right neosurface to bind to a substrate of choice and define a small pocket from which small molecules can subsequently be designed to mimic the loop insert.

The approach seems to work well with a small loop on CRBN creating a glue interaction with STAT3 using an interface which looks very different to the now well-described G-loop-type mode. If this approach can be used to persuade CRBN to glue to substrates which do not have the beta-hairpin G loop or related motifs, it could expand the CRBN repertoire even further. Phoremost is now applying the approach to non-CRBN E3 ligases as well of course.

So that’s the end of the conference, and what a great meeting it’s been. I’m off for a lay down but will reflect on everything I’ve seen this week and be back posting again next week.

Thanks also everyone in Boston who came up to me and said how much they enjoy these posts. All opinions are personal and I’m not saying all of my analysis & conjecture will be 100% correct but I hope it helps make the induced-proximity community better appreciate the complex science we all do and think about it in new ways. If you have other thoughts or disagree with my perspective in any way, please add a comment and start a conversation!