the latter harbors the catalytic site where both the cofactor FAD and the substrate
bind [127]. The AOD of LSD1 shares a significant sequence similarity with other
FAD-dependent amine oxidases showing 26% homology with PAO (plant amine
oxidase) and 20% homology with MAO-A and MAO-B enzymes [128]. The
FAD-binding pocket of these FAD-dependent amine oxidases shares a high
sequence similarity, whereas the substrate binding pocket, accommodating the
natural substrates, shows substantial differences [129]. Importantly, the substrate
peptide of LSD1 is embedded in a deep and negatively charged pocket which is
much more spacious and open as compared to other FAD-dependent amine oxidases
[127, 128]. Moreover, the other amine oxidases characteristically possess an aromatic cage to recognize the positively charged ammonium group through cation-π
interactions, which is missing in LSD1 [127, 128]. In LSD1, only one aromatic
amino acid residue (Tyr761) is preserved, whereas the second one is replaced by
Thr810.
Currently, more than 50 crystal structures of LSD1 are available in the Protein
Data Bank (PDB), which have been used for virtual screening and structure-based
optimization studies of LSD1 inhibitors. However, docking and VS of LSD1 small
molecule inhibitors face two major challenges: the huge size of the substrate binding
pocket (more than 1,700 Å
3 ) and the fact that many known LSD1 inhibitors are
irreversible and covalently bind to FAD.
Several VS approaches utilizing diverse techniques have been successfully
applied to find novel LSD1 inhibitors [130–132]. In 2013, Schmidt et al. reported
on a substructure-based screening combined with molecular docking to the LSD1
substrate binding pocket to identify novel propargylamines as irreversible LSD1
inhibitors [130]. The Enamine compound collection (~750,000 compounds) was
first screened for compounds bearing an N-propargylamine warhead, known from
irreversible amine oxidase inhibitors such as pargyline, and the retrieved hits were
subsequently docked into LSD1 substrate pocket. This led to the identification of the
3-aryloxy-2-hydroxypropargylamine derivative T5342129 (30, Fig. 7) as an LSD1
inhibitor (IC 50 of 44 μM and 34 μM in a hydrogen peroxide-dependent assay and a
formaldehyde dehydrogenase (FDH) assay, respectively). An in vitro dilution assay
confirmed that T5342129 is an irreversible LSD1 inhibitor.
Another VS approach was described by Zhou et al. who used a combination of
pharmacophore-based VS and docking to identify novel LSD1 inhibitor chemotypes
[132]. A common ligand-based pharmacophore model was generated on the basis of
22 known LSD1 inhibitors, although these inhibitors encompassed different
chemotypes, including irreversible tranylcypromine and propargylamine inhibitors
and FAD-competitive and FAD-noncompetitive inhibitors. The resulting
pharmacophore was subsequently screened against the SPECS database (about
172,000 compounds), and the resulting hits (about 7,000 molecules) were further
filtered according to their drug-likeness and predicted ADMET properties. Finally,
950 compounds were docked to the LSD1 substrate binding pocket. Among the nine
selected compounds, XZ09 (31, Fig. 7) was the most potent, showing an IC 50 value
of 2.41 μM and selectivity for LSD1 over MAO-A/MAO-B. However, an
470
D. Robaa et al.
bind [127]. The AOD of LSD1 shares a significant sequence similarity with other
FAD-dependent amine oxidases showing 26% homology with PAO (plant amine
oxidase) and 20% homology with MAO-A and MAO-B enzymes [128]. The
FAD-binding pocket of these FAD-dependent amine oxidases shares a high
sequence similarity, whereas the substrate binding pocket, accommodating the
natural substrates, shows substantial differences [129]. Importantly, the substrate
peptide of LSD1 is embedded in a deep and negatively charged pocket which is
much more spacious and open as compared to other FAD-dependent amine oxidases
[127, 128]. Moreover, the other amine oxidases characteristically possess an aromatic cage to recognize the positively charged ammonium group through cation-π
interactions, which is missing in LSD1 [127, 128]. In LSD1, only one aromatic
amino acid residue (Tyr761) is preserved, whereas the second one is replaced by
Thr810.
Currently, more than 50 crystal structures of LSD1 are available in the Protein
Data Bank (PDB), which have been used for virtual screening and structure-based
optimization studies of LSD1 inhibitors. However, docking and VS of LSD1 small
molecule inhibitors face two major challenges: the huge size of the substrate binding
pocket (more than 1,700 Å
3 ) and the fact that many known LSD1 inhibitors are
irreversible and covalently bind to FAD.
Several VS approaches utilizing diverse techniques have been successfully
applied to find novel LSD1 inhibitors [130–132]. In 2013, Schmidt et al. reported
on a substructure-based screening combined with molecular docking to the LSD1
substrate binding pocket to identify novel propargylamines as irreversible LSD1
inhibitors [130]. The Enamine compound collection (~750,000 compounds) was
first screened for compounds bearing an N-propargylamine warhead, known from
irreversible amine oxidase inhibitors such as pargyline, and the retrieved hits were
subsequently docked into LSD1 substrate pocket. This led to the identification of the
3-aryloxy-2-hydroxypropargylamine derivative T5342129 (30, Fig. 7) as an LSD1
inhibitor (IC 50 of 44 μM and 34 μM in a hydrogen peroxide-dependent assay and a
formaldehyde dehydrogenase (FDH) assay, respectively). An in vitro dilution assay
confirmed that T5342129 is an irreversible LSD1 inhibitor.
Another VS approach was described by Zhou et al. who used a combination of
pharmacophore-based VS and docking to identify novel LSD1 inhibitor chemotypes
[132]. A common ligand-based pharmacophore model was generated on the basis of
22 known LSD1 inhibitors, although these inhibitors encompassed different
chemotypes, including irreversible tranylcypromine and propargylamine inhibitors
and FAD-competitive and FAD-noncompetitive inhibitors. The resulting
pharmacophore was subsequently screened against the SPECS database (about
172,000 compounds), and the resulting hits (about 7,000 molecules) were further
filtered according to their drug-likeness and predicted ADMET properties. Finally,
950 compounds were docked to the LSD1 substrate binding pocket. Among the nine
selected compounds, XZ09 (31, Fig. 7) was the most potent, showing an IC 50 value
of 2.41 μM and selectivity for LSD1 over MAO-A/MAO-B. However, an
470
D. Robaa et al.