Synthetic and computer-assisted analyses of the pharmacophore for the benzodiazepine receptor inverse agonist site

Article abstract of DOI:10.1021/jm00171a007

The structural requirements for ligand binding to the benzodiazepine receptor (BzR) inverse agonist site were probed through the synthesis and in vitro evaluation of 3-substituted β-carbolines 6, 7, 11, 12, γ-carboline 13, and diindoles 18-21, 23-25, 27, 28, and 34. On the basis of the apparent binding affinities of these and other analogues, a hydrogen bond acceptor site (A₂) on the receptor is proposed to interact with the N(9) hydrogen atom of the β-carbolines or the N(7) hydrogen nuclei of the diindoles. Likewise, a proposed hydrogen bond donating site (H₁) interacts with the N(2) nitrogen atom of the β-carbolines or the N(5) nitrogen atom of the diindoles. It appears that interaction with both sites is a prerequisite for high affinity since analogues which have either one or both of these positions blocked exhibit substantial reduction in affinity. Moreover, H₁ appears to be capable of engaging in a three-centered hydrogen bond with appropriately functionalized ligands, which explains the increase in potency observed in the following series of 3-substituted β-carbolines: the n-butyl (12, IC₅₀ = 245 nM), n-propoxy (9, IC₅₀ = = 11 nM), and propyl ketone (11, IC₅₀ = 2.8 nM) congeners. In addition to H₁ and A₂, there appears to be a relatively narrow hydrophobic pocket in the binding cleft that can accommodate substituents at the 3-position of the β-carbolines which have chain lengths ≤ C₅. There is a 1 order of magnitude decrease in affinity between n-propoxy analogue 9 (IC₅₀ = 11 nM, chain length = 4) and n-butoxy derivative 7 (IC₅₀ = 98 nM, chain length = 5). Furthermore, α- and γ-branching [e.g. ethoxycarbonyl (2), IC₅₀ = 5 nM and tert-butoxycarbonyl (31) IC₅₀ = 10 nM] but not β- and δ-branching [e.g. isopropoxy (6), IC₅₀ = 500 nM and (neopentyloxy)carbonyl (48), IC₅₀ = 750 nM] at position 3 are tolerated. Occupation of this hydrophic pocket is clearly important for high affinity as evidenced by the relatively low affinity of 30, a β-carboline which possesses a hydrogen atom at the 3-position. This same hydrophobic pocket is partially filled by the D and E rings of the diindoles, which accounts for the high affinity of several members of this series. An excluded volume analysis using selected 3-substituted β-carbolines and ring-E substituted pyridodiindoles is consistent with the presence of this hydrophobic pocket. A model which distinguishes inverse agonists from antagonists is also proposed based in part on experimental findings which demonstrate that 3-n-propoxy-β-carboline (9) is an antagonist at the BzR with low efficacy and is devoid of proconvulsant activity at the highest dose tested (40 mg/kg). β-Carbolines which possess substituents of shorter length which are constrained to be in the plane of the aromatic ring tend to display inverse agonist activity while β-carbolines with longer substituents which can access regions of space above and below the plane of the aromatic rings are likely to have antagonist activity. Lastly, results from a 3D QSAR analysis of 37 test compounds (cross validated r² = 0.59) correlate well with and strongly support the previously proposed model of the pharmacophore for the benzodiazepine inverse agonist receptor site. The 3D QSAR electrostatic map is consistent with the existence of hydrogen bonding sites H₁ and A₂. Moreover, the steric map supports the existence of a hydrophobic binding pocket and is in qualitative agreement with the receptor essential volume obtained from an excluded volume analysis.