Dynamic undocking, a structure guided tool for virtual drug discovery
Drug discovery, the procedure at the heart of all big pharmaceutical companies today, is a long, risky and laborious endeavor. With average development times from 11–15 years, the fate of a company’s future very much lies on the commercial success of few compounds per decade. To add insult to injury, drug discovery costs have exploded in the last decade as well as successful candidate compounds are ever harder to find. All these issues cause drug costs to skyrocket, which underlines and exacerbates ever increasing inequalities in medical care.
One of the major problems in drug discovery is the narrowing down of prominent candidate compounds, which includes screening for millions of chemical substances to find the ones that show an effect.
To understand what makes a compound effective, one has to zoom in on proteins of interest at the level of basic molecular interaction.
Let’s say one wants to control inflammatory responses with a drug; one way to do so is influencing the behavior of core proteins called glucocorticoid receptors (GCRs) that are master regulators of immune response genes. GCRs are usually bound by scaffolding proteins and inactive. However, once their designated ligand (a hormone called cortisone) shows up, it has a higher binding preference to the GCR and thus displaces all other scaffolding proteins. Once the GCR is “free” of the old interaction partners, it can finally conduct business with other relevant interaction partners, thus modulate the immune response.
Taking a reductionist approach, one could say that the effectiveness of any drug can be predicted by how strong it binds to its designated protein target. This statement especially holds true for drugs that inhibit the function of enzymes by obstructing their catalytic center.
To estimate the strength of binding, scientists consider the binding constants of ligand-protein bonds, which gives a numerical value to the energy of a chemical bond. Fundamentally, the binding constant describes the thermodynamic stability (how likely does it fall apart) of the ligand-protein complex.
Up until recently, the assessment of binding constants was the only parameter drug discovery could use to narrow down compounds, after that it would still come down to doing the experiments for thousands of candidate compounds.
In a recent paper published in the journal Nature Chemistry, scientists from Spain, France and the UK propose a new parameter, a property they call Wqb, the work needed to break a quasi-bond.
Rather than thermodynamic stability, their method tries to estimate the structural stability of ligand-protein complexes.
Many ligand-protein complexes are subjected to strict geometric constraints, usually but not exclusively imposed by hydrogen bonds because of their sharp distance requirements as well as angular dependencies. In their studies, the authors investigated whether the work (Wqb) needed to displace these rigid hydrogen bonds can be used as predictors for ligand binding.
In a computational method the authors termed dynamic undocking (DUck), the Wqb is calculated by estimating the energetic value of only one key native hydrogen contact.
This magnitude solely indicates if the interaction under investigation gives rise to a (local) minimum in the free-energy landscape and estimates the depth of the said minimum. -Sergio Ruiz-Carmona et al., Nature Chemistry, 2016
More importantly, in their studies the authors could show that Wqb is not correlated with the binding constant, thus making DUck a truly independent and complementary way of estimating ligand-protein interaction stability. While DUck is not perfect, since it underestimates the contributions of alternative events increasing structural stability as well as the fact that some ligands achieve potency without robust anchoring interaction, it can be used for virtual screening. In their study, the authors assessed on the example of HSP90 interactors how well DUck would improve target identification in a real drug discovery screening. To their surprise, DUck was able to improve the hit rate from 4.4% to a staggering 38%, almost an order of magnitude gain in efficiency.
Hydrogen-bonding groups in the active site are privileged structures to fix the ligand in place, particularly when they act as binding hot spots and can form water-shielded hydrogen bonds. The work needed to break such interactions (Wqb) is very useful to detect true ligands, even though it is a non-equilibrium property that is not expected to correlate with ΔGbind. This intriguing fact may reflect the nature of proteins, which have been designed to bind their natural ligands not only with high affinity and selectivity, but also to form structurally stable complexes. -Sergio Ruiz-Carmona et al., Nature Chemistry, 2016
In summary, DUck virtual screening provides a new method to estimate suitable drug targets by using a novel parameter not correlated with thermodynamic, but structural stability. One can easily imagine that DUck’s initial reliance on only once key hydrogen bond can be expanded, thus increasing its potency for virtual screening even further.
For the global picture, DUck is a complementary method for the drug-discovery community which will help speed up the long risky way of drug development, and thus hopefully incentivize cheaper prices and better drugs in the long run.
This story is part of advances in biological sciences, a science communication plattform that aims to explain ground-breaking science in the field of biology, medicine, biotechnology, neuroscience and genetics to literally everyone. Scientific understanding has too much barriers, let’s break them down!
You can also help us to improve by giving feedback. Your voice matters.
