Skip to main content

Chemical Biology

What Those Degraders Are Actually Doing

Since targeted protein degradation is such a hot topic these days, this paper (which adds to the results obtained by this one) should get some interest. It’s a report of a detailed look at the kinetic behavior of several bifunctional degraders – and there’s a lot of kinetic behavior to look at. That’s because you’re looking at more than one potential binary complex (the bifunctional molecule with either its protein target or the E3 ligase) and a ternary complex as well (all three), with variable on and off rates for all of these. The tool used in this paper, surface plasmon resonance (SPR) is ideal for this sort of thing, since you can see all those rate constants directly in the data.

The authors (from Dundee and Boehringer Ingleheim) picked four different bifunctional molecules (two different chemical targeting “head” groups and different linkers to a ligand for the VHL ligase enzyme) that had been shown already to have cellular activity in degrading various bromodomain (BRD) proteins. It can take some work to get such SPR assays set up – one partner of the system has to be attached to the surface of a chip for the technique to work, and it’s not always obvious which one that should be. You are in theory going to get better signal/noise if the smaller partner is attached (and thus has a bigger change in molecular weight when the binding event occurs), but you have to try it and see, and there are of course different options for immobilizing the species of interest, so the whole process can be pretty empirical.

Once tuned up with the VHL enzyme immobilized, though, the system delivered some interesting readouts. The degrader MZ1 forms a well-characterized complex with the BD2 bromodomain of the BRD4 protein. It shows, in fact, pretty selective degradation of that protein, despite the fact that its “warhead” has nearly equivalent binding to several other BRD subtypes. It turns out that its ternary complex with that BD2 part forms with a relatively fast on-rate and a slow off-rate, and high cooperativity (that is, formation of the binary complex favors the formation of the ternary one). Trying this with the BD2 domain of the other BRD subtypes showed that that BRD4 complex is indeed the longest-lasting (130 second half-life), while BRD2 had a 70-second half-life and BRD3 only a six-second one. (Similar experiments with the BD1 binding sites of any of these, though, showed that all the complexes  dissociated very quickly (less than one second). The X-ray structures suggested that a Glu-for-Gly switch accounted for the change in BD2/BRD3, and sure enough, this mutation when introduced into BRD4 caused to half-life to crater, and when reversed in BRD3 gave it a far longer one.

Moving away from the MZ1 degrader, two others (with variations in linker length, etc.) showed rather different profiles. They had very strong binding constants for their BRD targets, but negative cooperativity in the ternary experiments: that is, they form binary complexes very well, but probably only flicker in and out of the ternary complex state where they bring the BRD protein and the VHL ligase in contact with each other. The results in cells line up with this mental picture, suggesting that the lifetime of the ternary complex is the driver in these cases for successful degradation. The longer the ligase and the target protein are in proximity, the more the latter presumably has a chance to get ubiquitinated and marked for destruction.

That’s what the earlier paper I referenced above (from a group at Promega) pointed to: that the rate of degradation depends (quite reasonably) on how much the target protein gets ubiquitinated, and that in turn depends on the rate of successful ternary complex formation. There are several ways that such complexes can be a success, depending on how the various rate constants balance out, but this current SPR work suggests that having good binding to your target protein is only the beginning of the story. That binary complex in turn can have a whole range of different behavior when it encounters the third protein (the ligase), and you’d have to assume that the exact same situation holds for cases where the bifunctional degrader/ligase pair is the first thing that forms (indeed, this new paper has experiments where the E3 ligase is immobilized on the SPR chip and ones where the target protein is immobilized, although it seems that in most cases one of those modes will work a lot better than the other – such is the way of SPR assays).

That makes you wonder, though, if cooperativity is a key requirement. This paper from last year suggests that it doesn’t have to be. That one is on a different target protein (BTK) and uses a different ligase (cereblon instead of VHL), and the authors found, probably a bit to their surprise that the formation of the ternary complex in that case showed basically zero cooperativity. As they point out, the ternary complex is very likely to still be a key intermediate, but it’s possible that entropy/enthalpy compensation makes its formation less favorable (although still enough overall to be effective). It’s worth noting that cereblon does a fine job of degrading the BTK protein, but VHL is notably ineffective (even though it seems at first glance like it should work). Add to these the complications of protein expression, concentration, and localization in different cell types, and it’s no wonder that the protein degradation field is keeping everyone busy!

7 comments on “What Those Degraders Are Actually Doing”

  1. Martin (the other one) says:

    I hate to be the one saying that, but… I think you forgot the “a” in “cooperativity” throughout the post…

    1. Derek Lowe says:

      You are that guy, but I’m glad you’re here! Fixed.

      1. rhodium says:

        Reminds me of my first independent paper. A referee pointed out I had misspelled homogeneous throughout. Sometimes referees are worthwhile.

    2. Eugene says:

      And I thought it was an alternate spelling (British vs US). Google search thinks you are mis-spelling cooperativity but if you force a search on “coopertivity” there seems to be a lot of usage of that spelling.

      1. Peter S. Shenkin says:

        “Coopertivity” describes how people work together to make barrels.

  2. CB says:

    I have been following Ciulli’s lab research with great interest. I have often wondered if ubiquitination rate vary for each ternary complex Ub-E2-CRL-E3~>(target protein) whether the protein is the endogenous substrate or forced with PROTAC or glue. Presumably this is a reason why there are 600+ E3 ligases out there of 3 supertypes to tune the ubiquitin transfer for each protein(s) it interacts with. For instance, SPR coopertivity is low with CRBN but appears to be a measurable with VHL, however, both will form a ternary structure just the same. It’s like a quick kiss versus a hug, but the deed is done.

    As much as I would love to have some high throughput assay to screen PROTACs prior to a Western blot, you can’t make that jump until you know that you are in fact degrading your target. MZ1 is well studied now so this is all hindsight. However, this work is still important to fully characterize any PROTAC after it has been identified it actually works in a relevant cell. I have to keep telling colleagues not to get ahead of themselves with fancy biophysical assays or over-designed linkers because you don’t know what will work with your target.

  3. John bradley says:

    I totally agree with CB. Biophysical assays provide valuable information, but even a ternary complex is an approximation to the endogenous cullin complexes.

    Screen PROTACs in a cellular context preferably against the endogenous (not ectopically expresses Target). High content imaging ?

Comments are closed.