HBAs and efflux

Not a really a new topic... There was a nice review by Lilly scientist a while back, which highlighted the impact of Hydrogen Bond Acceptor strength on efflux (BMCL2012). If you are a fan of DrugHunter (if not, you are missing out!), there was also a really nice talk last year by Patrick Schnider where he highlighted the impact of HBA strength on efflux in Roche's vasopressin 1a receptor antagonist program (available on YouTube:link and corresponding article JMedChem2020). A few years ago, I had posted a few examples of this as part of a broader look HBA strength (link)....However, 2 fairly recent examples made me want to revisit the topic.

When it comes to (P-gp mediated) efflux, we are quite obsessed by carefully limiting the number of available hydrogen bond donors we put in a molecule. It turns out that hydrogen bond acceptors can be a pain too. The influence is less strong (and less documented) than for HBDs but there are examples where simply tweaking the strength of HBAs has an impact on P-gp recognition. This has been suggested to explain the propensity of carbamates to be less effluxed than amides, as showcased by Pfizer scientists ( BMCL2008, JMedChem2011) and AstraZeneca scientists (JMedChem20218 ). If you are interested in learning a bit more about carbamates, you can also check out this pretty old (but still relevant) post (link).

The strength of HBAs can be measured and is called pKBHX or pKβ. It can also be predicted. If you are not familiar with this, I would highly encourage you to take a look at the publications from University de Nantes scientists in collaboration with Roche scientists (JMedChem2009 , JChemInfModel2016) and of course Peter Kenny's seminal articles on predicting these (JMedChem2016, JMedChem2008).

The game-changer for me this past year has been Rowan scientific's platform (https://www.rowansci.com) which allows easy and quick access to this kind of prediction (Wagen, C. 2025, ChemRxiv (link)). All the values below have been generated using this platform.

The first example comes from Amgen scientists and their quest for Glycine receptor potentiators (JMedChem2017). The following table only shows a small fractions of the compounds shared in the article but it highlights, in my view, how pKBHX can help rationalise P-gp SAR on heteroaromatics. It is not a perfect correlation but would certainly suggest limiting your pKBHX to values around or below 1.0.

The second example comes from University College London’s drug discovery unit and their Notum inhibitor program (BMCL2020). The authors shared quite a lot of MDR1 (P-gp) efflux data which I've plotted below. The markers are coloured by subseries.

A couple of comments: a well placed CF3 substituent can really drive pKBHX and efflux down (more on this in the AZ example below). Differences in isomers of heterocyles (green vs pink) can have a huge impact (see all the literature on oxadiazole isomers (MedChemComm2012)). Finally, I had never really looked into imidazolidinones (yellow) as piperazinone replacements (cyan), but they certainly fit the bill in this program and ended up on the lead compound. Something to keep in mind...

I believe these are two nice examples of how using pKBHX could help you design out a P-gp liability (prior to synthesis). If you need more convincing, take a look at the must-read AZ example below (BMCL2016). I remember reading about it internally back in 2012 when I was working at the Reims AZ site and being blown away by those results. It took just the right number of fluorines on this carbon to maintain potency and mitigate efflux. ....Talk about precision MedChem design....(By the way, although the values are different from the ones calculated in the AZ paper - the important thing is that ranking of the pKBHX is the same)

Just keep this in mind when you start adding amides or heterocycles to your brain penetrant molecules...

#MedChem #DrugDesign #pKBHX #Efflux