Natural Products as Drug Starting Points

(Frank E. Koehn, Pfizer Inc.)

The DNP describes ~250,000 distinct compounds derived from microbes, plants and other organisms.

Major impact on drug discovery

  • From 1940-2006, 46% of NCEs are “natural product derived”.
  • From 1970-2010, 31 approved NCEs came directly from natural products.
    • 13 are compliant with the Rule of 5.
    • LogP and hydrogen bond donors are the key physiochemical factors, enabling bioavailability in large NP's.
    • The compounds are mostly from soil microorganisms and are predominantly polyketides, peptides and terpenoids.

Unique challenges with natural products:

  • Chemical accessibility and synthetic manipulations can be difficult with complex NP structures.
  • The isolation of active compounds from NP extracts is usually slow and resource-intensive.
  • Pure NP libraries are difficult to enable (e.g. very expensive to make a high-quality NP library).

Although the overall “chemical space” is extremely large, not all of the space is biologically relevant.

  • For effective drug screening, the focus should be on the biologically-relevant part of the space.
  • NPs are privileged (biased to occupy biologically-relevant chemical space).

Phenotypic screening is better suited for NPs, instead of single target biochemical HTS.

  • Screen privileged NP library with cell cultures → phenotypic response → target identification and mechanism
  • Exposes the library to the largest number of targets possible (whole cells)
  • Need to carefully consider the screening strategy to get past the “black box” of target identification

NP libraries:

  • Crude extract – may contain compounds that interfere with HTS assays
  • Pre-fractionated – Can help identify specific compounds that have activity
  • Pure compound – best approach but resource-intensive to design and characterize; provides most rapid route to hit identification

Lipinksi’s Rule of 5 (Ro5) has guided the design of compounds into privileged ADME space. This approach is excellent for many targets but not protein-protein interactions.

  • The overlap between components of the “druggable genome” and disease-modifying genes is limited; this is typically the space targeted by small molecule drugs (See FIG. 1).
  • NPs can increase targetable space with beyond Ro5 compounds and may be better for tougher disease targets.
  • See TABLE 1 of examples of orally active beyond Ro5 natural product drugs.

Polyketide synthase (PKS) engineering

  • Total synthesis of NP-based compounds is possible but can be very challenging.
  • Biosynthesis allows chemical elaborations that are not possible by semi-synthetic methods.
  • Mutasynthesis to produce altered polyketide starting units and/or altered PKS cassette modules allows modification of the structure.
  • EXAMPLE: Rapamycin, a polyketide, is assembled by a large biosynthetic pathway; a starting unit is built up with additional 2-3 carbon fragments by PKS modules. By mutasynthesis, an alternate starting unit can substituted in, resulting in novel analogues. See REFS (Gregory 2005 and Gregory 2006).

Natural products, by virtue their origin, are within or at least proximal to biologically relevant chemical space.

  • Protein structures are conserved far more than protein sequences (only about 1000 protein folds are populated in nature).
  • NPs are privileged for binding cellular protein targets because they are synthesized by proteins and are naturally designed to bind conserved protein structures.
  • NPs can lead to unanticipated drug targets and mechanisms.
    • EXAMPLE: The mechanism of rapamycin was discovered after the phenotypic outcome. Rapamycin recruits a protein (FKBP12) in order to inhibit another protein (mTOR). While rapamycin is not BBB-penetrant, modification creates an immunosuppressive neurotrophic agent that does access the CNS.
    • Biosynthetic principles can be used to produce natural Bro5 compounds for difficult targets.
Neurotrophic Factor Mimetics: Target Validation to Lead Optimization >