Fig 1.
Fentanyl derivatization and the μOR.
(A) Commonly-modified positions around the 4-anilidopiperidine core of fentanyl. (B) The seven-helix transmembrane domain of the μOR in complex with agonist BU72; the binding site is represented by the transparent surface.
Fig 2.
Molecular docking poses and scores.
(A) Docking score distribution from the 10 docking simulations of fentanyl (purple), carfentanil (green) and N-methyl fentanyl (blue). (B) Superimposition of the ten lowest energy fentanyl poses in the μOR binding pocket. (C) The left and right panels present the primary (7) and secondary (3) carfentanil poses from the 10 docking simulations, respectively. (D) The left and right panels present the primary (6) and secondary poses (4) of N-methyl fentanyl from the ten docking simulations, respectively. The six primary N-methyl fentanyl poses are virtually identical. In panels B-D, the negatively charged sidechain of Asp147 that forms the salt bridge with the positively charged amine of each opioid is highlighted by the red mesh surface, and the aromatic sidechain of His297 is represented by the blue mesh surface.
Table 1.
Experimentally determined and predicted binding affinity ranges of the 23 opioids docked at the μOR.
The 8 fentanyl derivatives and 7 fentanyl congeners are listed first and ordered by increasing binding affinity, and the 8 morphine derivatives are listed last. The green labels indicate that the model correctly predicted the binding concentration regime, yellow indicates that the model predicted the compound to bind stronger than the measured Ki, and red indicates that the model predicted the compound to bind less strongly than the measured Ki.
Fig 3.
Summary of the fentanyl analog chemical features assessed by molecular docking.
Fig 4.
Fentanyl analog binding prediction and classification.
Scatterplot of the experimentally determined binding affinity, Ki, with the ADS from the molecular docking procedure of the 8 fentanyl analogs (shown as diamonds). The grey, shaded regions display the sub-nM, 1–100 nM, and greater than 100 nM binding concentration regimes used to classify the predicted binding strength of new drugs. N-methyl fentanyl (blue), fentanyl (purple), carfentanil (green) and lofentanil (red) are presented with colored diamonds to demonstrate the method’s ability to separate fentanyl analogs into the proper binding concentration regimes. The remaining black diamonds represent the 4 other fentanyl analogs that were docked and scored.
Fig 5.
Fentanyl congener binding prediction and classification.
Scatterplot of the experimentally determined binding affinity, Ki, with the average docking score from the molecular docking procedure of the 7 fentanyl congeners (shown as circles). (+)-Tramadol (blue), meperidine (purple), propoxyphene (green) and methadone (red) are highlighted and demonstrate the method’s ability to predict the correct binding concentration regimes of fentanyl congeners. The remaining black circles represent the three fentanyl congeners that were docked and scored. The black diamonds represent the fentanyl analogs from Fig 4.
Fig 6.
(A) Distribution of the decoy binding scores. The vertical red and blue lines indicate the docking score of fentanyl and carfentanil, respectively. (B) Distribution of the Tanimoto index of the fentanyl decoys with respect to fentanyl. The vertical blue line indicates the Tanimoto index of carfentanil (Tc = 0.76) with respect to fentanyl. (C) Correlation of the fentanyl derivative molecular weight with respect to the ADS. (D) Correlation of the fentanyl derivative molecular weight with respect to the experimentally determined binding affinity (Ki). (E) Correlation of the fentanyl decoy molecular weight with respect to the ADS. In panels C-E, the red and blue diamonds indicate fentanyl and carfentanil, respectively.
Fig 7.
(A) Chemical structure of the designer opioid Fu-F. (B) Binding concentration prediction of Fu-F (red). (C-D) Primary and secondary binding poses from the 10 independent docking simulations of Fu-F. In panels C and D, the key salt bridge (Asp147) and aromatic stacking interactions (His297) with Fu-F are highlighted by the red and blue mesh surfaces, respectively.
Fig 8.
Morphine analogs: Binding prediction and classification.
Scatterplot of the experimentally determined binding affinity, Ki, with the ADS from the molecular docking procedure of the 8 morphine analogs (shown as squares). Pentazocine (purple) and buprenorphine (blue) are the most structurally dissimilar to morphine. The remaining 6 drugs are structurally similar to morphine; 3 are predicted correctly (green) and 3 are predicted incorrectly (red). The black diamonds and circles represent the fentanyl analogs and fentanyl congeners from Fig 5. For clarity, correctly and incorrectly predicted morphine analogs are on the left and right side of the vertical black line, respectively.