Field of Science

Protein-protein interactions: Can't live without 'em, can't easily drug 'em

The many varieties of protein-protein interactions
Here's a pretty good survey of efforts to classify and drug protein-protein interaction (PPI) targets by a group from Cambridge University in Nature Review Drug Discovery. Most drug developers have been aware of the looming mountain of these ubiquitous interactions - there's at least 300,000 of them at last count and most likely many more - and have been trying to attack them in one way or another for more than a decade. There's also no doubt that many PPIs are involved in crucial ways in disease like cancer and inflammation. By any token PPIs are important.

As the review indicates though, attacking these interactions has been daunting to say the least. They present several challenges that seem to ask for both new scientific and institutional models of strategy. From the scientific standpoint PPIs present a nightmarish panoply of difficulties: proteins that change conformation when they comes together, water molecules that may or may not be involved in key interactions, the universal challenges of designing 'beyond rule of 5' drugs for such targets and the challenges of developing highly sensitive new biophysical techniques to detect ligand binding to begin with.

Consider protein flexibility, a factor which often is the nemesis of even 'regular', non PPI projects. Protein flexibility not only make crystallization hard and its results dicey, but it decidedly thwarts computational predictions, especially if the conformational changes are large. PPIs however regularly present cases in which the conformations of the unbound proteins are different from the bound ones, so at the very minimum you need crystal or NMR structures of both bound and unbound forms. This is even harder if one of the partners is a peptide, in which case it's likely going to undergo even more conformational changes when it binds to a protein target. The hardest case is two peptides such as Myc and Max, both of which are unstable and disordered by themselves, which stabilize only when they bind to each other. That's quite certainly an interaction forged in the fires of Mount Doom; good luck getting any kind of concrete structural data on that gem.

The screening challenges involved in studying PPIs are as if not harder than the structural challenges. NMR is probably the only technique that can reliably detect weak binding between proteins and ligands in as much of a 'natural' state as possible. although it presents its own difficulties like protein size and other technical challenges. SPR and FRET can help, but you are really testing the limits of these binding assays in such cases. Finding reliable, more or less universal methods that combine high throughput with good sensitivity has been an elusive goal, and most data we have on this front is anecdotal.

Finally, the medicinal chemistry challenges can never be underestimated, and that's where the institutional challenges also come in. In many PPI projects you are basically trying to approximate a giant block of protein or peptide by a small molecule with good pharmacological properties. In most cases this small molecule is likely going to fall outside the not-so-hallowed-anymore Lipinski Rule of 5 space. I should know something about this since I have worked in the area myself and can attest to the challenges of modeling large, greasy, floppy compounds. These molecules can cause havoc on multiple levels: by aggregating among themselves, by sticking non-specifically to other proteins and by causing weird conformational changes that can only be 'predicted' when observed (remember that old adage about the moon...). Not only do you need to discover new ways of discovering large PPI inhibitors that can make it across cell membranes and are not chewed up the moment they get into the body, but you also need new management structures that encourage the exploration of such target space (especially in applications like CNS drug discovery: in oncology, as happens depressingly often, you can get away with almost anything). If there's one thing harder than science, it's psychology.

In spite of these hurdles which are reflected in the sparse number of bonafide drugs that have emerged from PPI campaigns, the review talks about a number of successful pre-clinical PPI projects involving new modalities like stapled peptides, macrocycles and fragment-based screening that at the very least shine light on the unique properties of two or more proteins coming together. One of the more promising strategies is to find an amino acid residue in one partner of a PPI that acts like an 'anchor' and provides binding energy. There have been some successful efforts to approximate this anchor residue with a small molecule, although it's worth remembering that nature designed the rest of the native protein or peptide for a reason. 

Another point from the review which seems to me like it should be highlighted is the importance of academia in discovering more novel features of PPIs and their inhibitors. In a new field even the basics are not as well known, it seems logical to devote as many efforts to discovering the science as to applying it. At the very least we can bang our collective heads against the giant PPI wall and hope for some cracks to emerge.

1 comment:

  1. I feel like what's said in your blog post has been the state of the field for the past half-decade, if not longer. i.e. the field feels stagnant. Have there been any developments? New drugs? Computational predictions seem to remain insufficiently useful.


Markup Key:
- <b>bold</b> = bold
- <i>italic</i> = italic
- <a href="">FoS</a> = FoS