Anabella Villalobos: Pfizer challenges the dogma of brain drug design

Anabella Villalobos. Photo: Christine Fu

The number of possible small organic molecules spanning the chemical space is estimated to exceed 10^60 – a mind-boggling number. How could we navigate this enormous space efficiently to look for compounds of pharmaceutical value?

Medicinal chemists draw on the knowledge of how molecules interact with biological systems to predict their desirability as drugs. One particularly challenging arena of drug discovery is the central nervous system (CNS). This is essentially a “demilitarized zone” shielded from foreign chemicals by the blood-brain barrier (BBB), tight junctions between endothelial cells lining the capillaries. The dogma in the field has long held that high lipophilicity and low polarity are key attributes of a successful CNS-penetrant drug, allowing it to passively diffuse through the plasma membrane. This notion was challenged by Anabella Villalobos, PhD, head of Neuroscience and Antibody Directed Conjugate Medicinal Chemistry at Pfizer, in a Quadrant industry speaker seminar held on Jan 25 at UCSF Mission Bay.

Villalobos began the talk by posing a dilemma: Lipophilicity is a double-edged sword. In vivo toxicology studies have shown that more lipophilic and less polar compounds are more likely to induce toxic events, presumably due to their propensity of promiscuous off-target binding. How does one balance safety with the ability to cross BBB in neuroscience drug design?

Villalobos’ team undertook an analysis of over 200 marketed CNS drugs and Pfizer candidates. They identified optimal ranges for six fundamental physicochemical properties that impact key in vitro pharmacokinetic and safety attributes. Marketed drugs were found to be highly permeable, not a substrate of the P-glycoprotein transporter which pumps molecules back into the blood, metabolically stable, bind targets efficiently, and demonstrate low in vitro toxicity. The task of medicinal chemists is to align these favorable drug-like attributes in one molecule to maximize the probability of clinical success.

Manipulating only one or two properties in drug design is too restrictive, Villalobos argued. Instead, Pfizer scientists created a CNS multiparameter optimization (MPO) algorithm that utilizes the six physicochemical properties and calculates a desirability score for each molecule. 74% of marketed drugs exhibited high MPO scores, which correlated well with alignment of desirable attributes, validating this method. Incorporating multiple parameters is also advantageous over a hard cutoff approach by expanding the chemical space of interest to drug discovery. Currently, 44% of CNS drugs heed the “high lipophilicity, low polarity” dogma, placing them in a higher safety risk quadrant within the chemical space. In contrast, only 11% reside in the lower risk quadrant (low lipophilicity, high polarity). Villalobos made the case for a need to not only overcome safety hurdles in the higher risk space, but also move beyond dogma and explore the lower risk space, while optimizing CNS penetration. She says that MPO will be used as a prospective design tool to facilitate this new direction of neurotherapeutics Pfizer is pursuing, accelerate the identification of promising compounds, as well as mine the patent literature for competitors’ leads.

The reports are published here and here.