Day 3

So the mist came down this morning in Boston but the quality of the presentations at day 3 of the TPD meeting continued to shine out, and indeed the insight from many of the speakers lifted the clouds from a few thorny TPD topics.

A real highlight was the overview from Neil Bence (VP Oncology, BMS). Now BMS, especially through their historic Celgene team, have quietly established an incredibly strong pipeline in the TPD area with up to 10 TPD agents now in the clinic including improved IKZF1/3 agents (in addition to marketed IMiDs of course), other glues (aka CELMoDs) for IKZF2, CK1a, and to activate fetal hemoglobin as well as several PROTACs (aka LDDs) vs AR & BCL6. Finally, through their Orum deal, a CD33-GSPT1 degrader-antibody conjugate completes the set.

Neil reminded us how the next gen IKZF1/3 glue mezigdomide can overcome lenalidomide resistance caused by CRBN downregulation, due in part to its ability to highly potently favour the active, closed conformation of CRBN in a nice demonstration that different CRBN binders can have very different properties. The efficacy of PROTACs using these different CRBN binders in CRBN low expressing cells may need to be checked of course.

Neil then went on to describe the discovery of the AR PROTAC BMS986365 which highlighted a key issue sometimes overlooked. Despite a good Dmax of ~90%, the team had found that in some cases, the remaining 10% of undegraded AR could be potently agonized by the PROTAC leading to continued tumor proliferation. After hard work, the team figured out how to ensure that the PROTAC remained an antagonist so any undegraded protein couldn’t continue to drive proliferation but this is a salutary tale to remember that PROTACs can still mediate functional effects (desired or undesired) on any undegraded protein so you may want to check the activity of your degrader under conditions where degradation isn’t operating (eg recombinant, cell-free or under proteasomal inhibition).

We also had a Kymera double header with R&D head Juliet Williams giving a good overview of the Kymera strategy including how they “throw the kitchen sink at our favourite targets” through using a wide range of ligand ID strategies and structurally enabling all targets.

Sean Zhu gave an illuminating story on the degradation mechanism of the lead CRBN-dependent IRAK4 PROTAC KT-474 using structural analysis of the ternary complex.

Conventional TPD wisdom will tell you that high levels of ternary complex co-operativity re an important driver of degradation efficiency however, KT474, the best compound from the IRAK4 project, shows strong negative co-operativity (a=0.06). In other words, while the PROTAC binds to IRAK4 with Kd 52nM, it’s affinity for IRAK4 in a ternary complex is reduced significantly to barely better than 1uM. Furthermore, a 3.1A cryo-EM structure shows IRAK4 & CRBN protein surfaces barely touching each other in the ternary complex which is mediated by KT-474 in an extended conformation – like a wary handshake at a distance of two people who’d really don’t trust each other. So what’s going on? 

The structural biology explains things. When KT474 is in solution or bound to IRAK4 in a binary complex, the linker is wrapped up giving the molecule a globular shape which seems to be its preferred state. However, this globular conformation doesn’t allow ternary complex formation. For this, the PROTAC has to “unwind” into an extended conformation, and this costs energy so, together with the fact that there’s not really much positive interaction between the proteins in the ternary complex, forming the ternary complex is energetically less favourable than the binary – ie negative co-operativity. In this case the negative co-operativity is driven more by the higher stability of the binary complex rather than any instability (eg PPI repulsions) in the ternary per se. An important lesson to consider all parts of the equilibria involved in ternary complex formation.

So hang on, if KT474 really doesn’t like forming ternary complexes all that much, why is it a great degrader and development candidate?? Well, I’m guessing that the ternary complex that is formed must be super-efficient at allowing ubiquitin transfer so even a low ternary complex population can still be enough.

Remember, degradation efficiency is driven by a combination of ternary complex stability but also ubiquitylation competence – and we don’t always know which of these factors is rate-limiting.

Next, we return to the topic of safety of TPD agents, most notably CRBN-based ones. Jessica Sims (Genentech) reminded us that to get a full idea of potential off targets, expression proteomics is still the best approach but remember, a single cell type will not express the entire proteome so you may want to look at a small set. Genentech selected a set of 3 cell lines which express most of a set of 33 targets considered key anti-targets (GSPT1 CK1a, SALL4 etc). SALL4 is the (main but certainly not only) target which has been most strongly associated with the clinical teratogenicity noted with thalidomide but most cell lines don’t express SALL4 – you’ll need to include reproductive tissue to pick this one up.

Turning then to species selection for safety studies, always make sure that at least one will be pharmacologically active (ie will degrade your protein) – and remember that rat, mainly due to a single valine to isoleucine difference in homology, often won’t give the glue-like activity you’d see in human. Cyno monkey is best of course with 100% homology to human CRBN but using primates in safety studies can present logistic or ethical hurdles so instead the next best option may be dog (99.3% homology). Rabbit is also usually sensitive to CRBN glue mechanisms though minipig has lower homology so may not be as good.

Next, Lise Loberg (Abbvie) continued the safety theme, specifically considering potential for CRBN-binder-mediated teratogenicity. Whilst this may be viewed as a manageable issue in oncology, especially where disease severity is high, as we move to less severe or better managed conditions, concern may rapidly grow. 

Most compounds will be run through EFD (embryo-fetal developmental toxicity) studies, often run just before phase 2, to look for teratogenicity risk. These studies, which are neither cheap nor fast, require 2 species, where at least one needs to be pharmacologically-relevant. Rabbit (usually sensitive to CRBN-mediated teratogenicity) and rat (infamously, not) are often used. In principle, this should be a good test however, the dataset to establish true predictivity is limited, so all is not straightforward and may depend on drug exposure or other factors.

There are in vitro options for assessing teratogenicity risk of course, the hiPSC mesendoderm differentiation assay or the Toxys reprotracker could be used but this will remain a hot area of debate and data generation. Indeed both the IQ consortium and HESI have TPD working groups to try to plot a path through this.

Despite the ongoing discussions on safety considerations, clinical data continue to impress. Len Reyno (C4T) gave a (previously disclosed) clinical update on the BRAF V600 PROTAC CFT1946 which has now been in the clinic approaching 2 years. BRAF appears to be a tricky target to degrade and CFT1946 only reaches Dmax ~75% in cells & in vivo with perhaps the most pronounced Hook effect of any of the clinical degraders so it may be better thought of as a hybrid degrader/super inhibitor but it certainly potently shuts down the pERK signalling giving good levels of pre-clinical efficacy and so far this seems to be translating to encouraging clinical responses, even in heavily pre-treated patients (4-5 prior lines of therapy). Melanoma patients responded well with signals in other tumors also but, very encouragingly a low AE rate was seen, in particular with regard to skin effects which have blighted previous BRAF inhibitor treatments due to wt BRAF effects. A couple of patient case histories really brought home the potential impact of these medicines in patients who had failed many previous lines of therapy – something everyone in the TPD field should be proud to see. 

Following on was Paula O’Connor from Nurix with an update on the clinical progress of BTK PROTAC NX-5948. Initial safety profiles were again very encouraging, and better than many of the well-used BTK inhibitors. In CLL, strong response rates (eg 69% partial response & 23% stable disease) show wide applicability in this patient segment with responses seen as soon as treatment cycle 2 in many cases. Deep degradation (>80%) of BTK, both wt and a range of mutants, was seen at all doses (across 50mg-600mg range tested) though it appears doses of 200mg & 600mg will be progressed to maximise the potential to hit a full range of BTK mutants. Case histories again brought the story to life including one patient who was close to end stage disease after many lines of treatment but achieved what could be a complete response (to be confirmed), again showing the transformational efficacy possible with TPD agents. 

So a real treat of great TPD science and clinical impact today. It’s the final day of the conference tomorrow where focus switches to new mechanisms so expect to see some great new approaches which could be hitting the clinic soon.