This is part 4 of our 5-part series on the Drug Discovery for Neurodegenerative Disease conference. See parts 1, 2, 3, and 5.
12 March 2007. When it comes to treating AD and other brain diseases, there is always something standing between the therapeutic pill and the cells needing help. That something is the blood-brain barrier (BBB), and it is a killing field of many a promising treatment. Two lively talks on medicinal chemistry drove home just how the need for brain penetration complicates compound design for AD. Then, researchers offered some novel ideas about what the BBB is (and is not), and what drug developers might do about it.
It’s the Compounds, Dummy
Medicinal chemist Christopher Lipinski, formerly of Pfizer, is now a scientific advisor to Melior Discovery in Exton, Pennsylvania. He sees chemical space, that is, the universe of possible small-molecule chemical structures, as mostly barren of candidate therapeutics. Lipinski is known for his “rule of 5. ” Derived from an analysis of existing drugs, this set of criteria tries to define which physico-chemical properties a drug must have to be orally bioavailable—that is, suitable to be taken as pills. (For a comprehensive review, see Lipinski and Hopkins, 2004). Only a small fraction of possible compounds fit the rule of five, and of those only a subset satisfies the more stringent criteria defining CNS penetrance.
Because of the scarcity of compounds that exhibit favorable properties, commercial libraries are mostly useless for CNS drug discovery, Lipinski said. They can yield valuable research and probe compounds, because drug discovery is different from chemical biology. There, the criteria for tool-like compounds are more relaxed, and the compounds themselves can be cheaper. But for real medicines, Lipinski stressed, permeability, potency, and solubility must all coexist in one compound. These properties must be worked in at the beginning, making good compounds expensive. As an example, Lipinski said the library the NIH is putting together is tool-quality, whereas to amass good compounds for serious drug discovery would cost ten to 20 times more per compound. Moreover, a standard pharma library of compounds for high-throughput screening runs to 500,000 compounds. With such high-cost libraries, target selection for drug development requires an increased measure of stringency. The developers want biologists to find relevant targets, but they also want to know if the target is druggable, in other words if there is a good chance of finding drug-like compounds that will affect the target?
Camille Wermuth of France’s University Louis Pasteur in Strasbourg, and president/CSO of Prestwick Chemical Company added his perspective on the selection and refinement of drugs for the brain. He ties the massive investment in high-throughput screening (HTS) for identifying hits and leads directly to decreased productivity and a higher failure rate of compounds, either because of poor pharmacological properties (called ADME for absorption, distribution, metabolism and excretion) or toxicity. Wermuth recommends that besides HTS, researchers also employ other methods to identify promising compounds, such as feedback from clinical experience and building off existing drugs. Companies would do well to spend more time optimizing structures for better ADME and toxicology, and to use chemistry to aid formulation, he said.
The discussion brought up the question of how large useful chemical space is in terms of compound numbers, and how an academic lab can acquire such compounds. Lipinski said the number of compounds that have been synthesized barely scratches the surface compared to the number of theoretically possible compounds. To make real inroads into chemical space, Wermuth suggested using libraries that consist of marketed drugs, which would represent 1,000-2,500 good compounds.
Believing in the BBB
The caveat to Lipinski’s rules for predicting compound behavior and CNS penetrations is that they only apply to small molecules, and only those that are not transported across the BBB. The rule of five does not hold for natural products, or substances that bind p-glycoprotein pumps or other transporters. What are the rules for peptides, antibodies, or RNAi? William Banks, of St. Louis University, Missouri, opened his talk with the disclaimer that there are no absolutes. When it comes to the BBB, he said, “Every rule has a major exception, every exception has a caveat, and every caveat has an aside.” Researchers can treat the BBB as a black box, or they can try to understand its properties. Either way, data trumps theory, so researchers should not make assumptions for any compound but test it anyway, Banks noted.
What exactly is the BBB? It is the special structure and function of the blood vessels in the brain that limits passage of substances from the blood to the CNS. Circulating drugs enter the CNS by diffusion through cells if they are lipid-soluble, or by active transport. Whether a compound accumulates in the CNS depends on its rates of efflux or degradation at the BBB. A medicine can be effective even if little of it gets in; in the case of morphine, less than 1 percent of the total given dose enters the brain yet its potency ensures that is enough.
Banks showed data on CNS entry of several potential therapies. Small, lipid-soluble peptides that interfere with beta-sheet formation cross the BBB by transmembrane diffusion (Permanne et al., 2002). Grehlin, (a feeding hormone and neurotrophic factor) is subject to saturable transport, but its fate differs across the CNS. It preferentially enters the hippocampus, where it has been shown to enhance synaptic density and promote LTP, and to have beneficial effects on learning and memory in a mouse model of AD (see ARF related news story). With antibodies, Banks’ data shows slow uptake, which seems to occur via an extracellular pathway. Continuous efflux further slows down CNS accumulation of these large proteins (Banks et al., 2002). Even antisense oligonucleotides are transported across the BBB slowly and accumulate in the brain after intravenous injection, despite the literature saying otherwise, Banks noted. Because antisense oligonucleotides are so stable, they can accumulate to therapeutic levels. Antisense to the Aβ peptide, given intravenously to the SAMP8 mouse model, decreased Aβ levels by 50 percent and reversed learning and memory impairments (Banks et al., 2001). Banks said he has started a company based on this approach.
In the discussion, participants’ opinions ranged from “the BBB is a misnomer—everything in the blood gets into the brain” to “we tried for years and couldn’t get any of those neurotrophic factors across.” One caveat in penetration studies done in rodents is the potential for species differences in brain penetration. More important, perhaps, is the idea that the BBB itself changes with aging and disease. The idea that alterations in the BBB may play a role in AD will be an important area to pursue, both for understanding the disease, and for creating new therapies. For more on circumventing the BBB, see part 5 of this series.—Pat McCaffrey.