In praise of weakness

It's not about strength but about adaptability

Biology has always had to do a tricky dance. On one hand, interactions between biomolecules need to be strong enough to be productive - enzymes need to hold substrates long enough to stabilize their transition states, antibodies need to bind antigens tight enough, protein-protein interactions need to last long enough to trigger downstream responses. On the other hand, molecules need to let go of each other soon enough so that they can participate in the next round of binding and interactions. In many ways this is very much like a contra dance where you hold a partner long enough to orchestrate a few steps, then swap partners to do the same thing again.

It was understanding this balance between tight and loose binding that made the discovery of the hydrogen bond by Pauling and others in the mid-twentieth century so groundbreaking. Compared to covalent and electrostatic interactions, hydrogen bonds are weak - 2-5 kcal/mol - but it’s precisely because of their weakness that they can achieve two critical goals: let go on time, and combine together synergistically to provide very strong binding - what’s called positive cooperativity - as is the case in DNA. Being loose rather than tight also achieves another goal, that of giving molecules the flexibility to move around so that they can explore productive conformations. In his seminal book, “The Nature of the Chemical Bond”, Pauling predicted that the importance of the hydrogen bond for understanding the structures of biological molecules would be greater than any other.

It is this tight interplay between binding and flexibility that’s very much visible when you trawl through the fertile field of PROTACs, molecular glues and other three-body systems. To recapitulate, all these systems consists of a ligand first binding to a protein, with this binary assembly then binding to and inhibiting or degrading another protein as part of a ternary complex. There are thus multiple events here which orchestrate the final act: ligand binding to the first protein (competing with ligand binding to the second protein) and the binding of the binary complex to the other protein. For PROTACs in particular there are three processes: binary ligand binding, ternary complex formation and degradation. Not surprisingly, several papers have tried to disentangle the complicated kinetics and thermodynamics of this process, both experimentally and theoretically.

The fascinating observation that has emerged from PROTAC field in particular in the last few years is that tight binary binding is not just unnecessary for ternary complex formation but might even hinder it. Too tight binding locks an E3 ligase into a conformation that prevents it from exploring conformations enabling it to latch on to the protein target and degrade it. Studies have found that each of these processes depends on the other in non-linear ways: for instance, binary complex formation is necessary but not sufficient for ternary complex formation, and ternary complex formation is necessary but not sufficient for degradation. Most importantly from the viewpoint of practical drug design, moderate or weak affinity ligands can be more effective in ternary complex formation and degradation than high affinity ligands because they allow enough conformational flexibility in both proteins to explore productive degradation-capable orientations.

Here are a few examples, starting with my favorite.

In a broad-spectrum kinase-targeting PROTAC study (using VHL as the E3 ligase), researchers observed a striking example of a weak-binding target being degraded better than a strong-binding target​. The PROTAC was designed with a promiscuous kinase inhibitor warhead that targeted multiple kinases. Remarkably, the PROTAC bound p38α kinase with only moderate affinity (Kd ≈ 11 μM), yet induced robust p38α degradation with DC50 ≈ 210 nM and rapid degradation kinetics. Within 12 hours of treatment, p38α levels dropped significantly (>90% degraded) and the protein’s half-life was reduced from hours to minutes, confirming efficient ubiquitination​.

By contrast, the same PROTAC bound Axl kinase with very high affinity (Kd ~26 nM) but failed to induce any degradation of Axl, even at 10 µM concentration​. In other words, a >400-fold stronger binder (Axl) was non-degradable, whereas a weak binder (p38α) was efficiently degraded. The authors attributed these striking differences to the formation of a viable ternary complex for one kinase vs the other; essentially, the Axl ligand could not reposition the protein into a conformation that allowed ubiquitination. A few other kinases obeyed this trend as well. The critical observation here is that you are getting degradation selectivity through the use of weaker ligands, a much sought-after goal.

The second example is from the AR (androgen receptor) field, which is a popular PROTAC area. In this study, the authors explored AR degraders that recruit the VHL E3 ligase, and in doing so they explicitly compared weaker vs. stronger VHL ligands on otherwise similar PROTACs. They used the same AR ligand with different VHL recruiters. One of the ligands had an IC50 ~ 190nM for VHL while the other bound VHL with much weaker low micromolar affinity. Surprisingly, the PROTAC with the weaker VHL ligand achieved greater AR degradation efficacy than those with higher-affinity VHL ligands​. It showed sub-nanomolar degradation potency, with DC50 ~0.5 nM in an AR-driven cell line. This was ~1.7-fold more potent than the other ligand (DC50 ~0.86 nM) despite the latter’s much tighter VHL engagement​. While both degraders are extremely potent, the key point is that reducing the E3 ligand affinity led to a measurable increase in degradation potency.

Structural modeling indicated that the shorter linker and weaker VHL ligand promoted an arrangement that maximized AR–VHL protein contacts​. In essence, the PROTAC was better able to adopt the optimal conformation for ubiquitination when the VHL arm was not locked in too tightly. The authors concluded that low-affinity E3 ligands can still induce highly efficient ternary complex formation and degradation, provided the induced PPIs are favorable.

The third case involves one of the earliest degraders targeting the BET bromodomain, MZ1. MZ1 was reported in 2015 and notably achieved potent BRD4 degradation (DC50 in the low nanomolar range) despite using a relatively weak VHL ligand (based on hydroxyproline, Kd ≈ 3–5 μM for VHL)​. The BET-targeting warhead in MZ1 is JQ1 (a pan-BET inhibitor with Kd ~50 nM for BRD4). Thus, neither binding arm of MZ1 is exceptionally high affinity on its own – the strength lies in the ternary complex.

In 2018, Testa et al. introduced a fluoro-substituted VHL ligand variant and showed that PROTACs can tolerate even weaker E3 affinity while maintaining degradation efficacy. A JQ1-based PROTAC was synthesized with a VHL ligand epimer that bound VHL more weakly (μM Kd) than the original ligand​. In cellular assays, compound 15a selectively degraded BRD4 with DC50 ~1–3 nM, and also degraded BRD2/3 at slightly higher concentrations​. This was on par with, or better than, MZ1’s performance (for reference, MZ1’s BRD4 DC50 is ~2–20 nM depending on cell line​. Thus, even though the VHL binder had weak, micromolar affinity, it drove extremely potent BET protein degradation​. The success of the ligand was again attributed to strong positive cooperativity – the formation of a high-affinity ternary complex that stabilizes the target–E3 interaction far beyond what the binary affinities would predict.

This case reinforced that binary affinity requirements for PROTACs can be lenient: a weak affinity ligand is sufficient if it can induce the right protein–protein contacts. As the authors noted, this “expanded the chemical space of TPD toward low-affinity molecules”​. It also suggested that traditional SAR (which optimizes for highest affinity) might not apply straightforwardly to PROTACs – instead, one must consider ternary complex SAR.

So what we are dealing with here is a kind of “Goldilocks principle” which is going to take us a long way - bind tightly but not too tightly. Ligands for ternary complexes have to navigate at least three processes - binary binding, ternary binding and ubiquitination or degradation. They of course have to bind to the individual target, but as the studies above show, moderate or weak binding may not just be sufficient but necessary. What is critical is firmly but gently setting up the protein targets close enough to each other so that they can find the right orientation for positive cooperativity and further enzymatic action like degradation.

It’s always very pleasing to realize that principles in science extend across domains of increasingly emergent complexity. From the finding in the mid-twentieth century that weak hydrogen bonds provide the foundation of molecular interactions in biology to the finding in the twenty-first century that the same balance between weakness and strength reigns supreme in drug design for ternary complexes, we have slowly but surely cemented the principle, as Darwin realized, that life is not about strength or weakness but about change and adaptability. It’s a good thing to remember in general.