hydrophobic core. The dye binds to the core and increases in fluorescence monitored
as a function of temperature. The midpoint of the sigmoidal curve is the thermal
melting temperature (T m ) of the protein. Ligand binding will stabilize or destabilize
the protein, changing the protein T m (Fig. 16). Rank ordering bromodomain binders
based on ΔT m has been done by creating a correlation curve of ΔT m vs K d of known
binders [9, 71, 106]. Although the correlation is only moderate with higher T m
indicative of higher affinity, this has been shown an as effective way to rank order
hit compounds for BRD4. This method may not work as well for bromodomains
with few known binding ligands. DSF uses small amounts of unlabeled protein, is
quick, and thus has been used as a high-throughput screening platform for
bromodomains [107–109]. Additionally, DSF has been used as a means to assess
inhibitor isoform selectivity of inhibitors [105, 110–112]. An inhibitor that is
selective would only perturb the T m of the protein it targets. For example, the
assessment of BET selectivity of (+)-JQ1 was carried out by doing DSF experiments
with 36 bromodomains. Only the eight BET family bromodomains had a ΔT m > 7
C,
indicating binding [9]. One challenge with DSF is how to categorize/characterize
compounds that destabilize the protein (negative ΔT m ). These compounds may still
bind in other assays. There is no consensus on how to effectively describe these
compounds. As a new advance, a variant of the DSF assay has been demonstrated for
bromodomains to show target engagement in cells called a “cellular engagement
thermal shift assay” or CETSA [113, 114].
4.3 Competition-Based Assays for Inhibitor Discovery
4.3.1 AlphaScreen High-Throughput Competition Assays
AlphaScreen is a competition-based assay which is used as an initial screening
platform and as a follow-up assay to determine IC 50 values of compounds
Fig. 16 (a) Example of DSF raw data of protein (black) and protein stabilized by ligand (red). (b)
The first derivative of the raw data is taken to determine the ΔT m (apex of the curve)
318
W. C. K. Pomerantz et al.
as a function of temperature. The midpoint of the sigmoidal curve is the thermal
melting temperature (T m ) of the protein. Ligand binding will stabilize or destabilize
the protein, changing the protein T m (Fig. 16). Rank ordering bromodomain binders
based on ΔT m has been done by creating a correlation curve of ΔT m vs K d of known
binders [9, 71, 106]. Although the correlation is only moderate with higher T m
indicative of higher affinity, this has been shown an as effective way to rank order
hit compounds for BRD4. This method may not work as well for bromodomains
with few known binding ligands. DSF uses small amounts of unlabeled protein, is
quick, and thus has been used as a high-throughput screening platform for
bromodomains [107–109]. Additionally, DSF has been used as a means to assess
inhibitor isoform selectivity of inhibitors [105, 110–112]. An inhibitor that is
selective would only perturb the T m of the protein it targets. For example, the
assessment of BET selectivity of (+)-JQ1 was carried out by doing DSF experiments
with 36 bromodomains. Only the eight BET family bromodomains had a ΔT m > 7
C,
indicating binding [9]. One challenge with DSF is how to categorize/characterize
compounds that destabilize the protein (negative ΔT m ). These compounds may still
bind in other assays. There is no consensus on how to effectively describe these
compounds. As a new advance, a variant of the DSF assay has been demonstrated for
bromodomains to show target engagement in cells called a “cellular engagement
thermal shift assay” or CETSA [113, 114].
4.3 Competition-Based Assays for Inhibitor Discovery
4.3.1 AlphaScreen High-Throughput Competition Assays
AlphaScreen is a competition-based assay which is used as an initial screening
platform and as a follow-up assay to determine IC 50 values of compounds
Fig. 16 (a) Example of DSF raw data of protein (black) and protein stabilized by ligand (red). (b)
The first derivative of the raw data is taken to determine the ΔT m (apex of the curve)
318
W. C. K. Pomerantz et al.