3.5.4
Relaxation-Filtered 1D
For slowly tumbling molecules, magnetization relaxes rapidly via
the T2 or T1ρ routes, while for rapidly tumbling molecules, these
relaxation rates are significantly slower. This differential relaxation
is used in the relaxation-filtered 1D to selectively attenuate the
signal of molecules bound to a slowly tumbling protein compared
to those molecules that are free in solution. Since the magnetization does not recover after dissociation, the relaxation-filtered 1D
therefore reflects the unbound population of the ligand. After
displacement by a potent competitor molecule, the unbound population of the compound is increased and the signal in the
relaxation-filtered 1D increases as a result.
3.5.5 Combined Data
Analysis
Since each experiment contains data reflecting different aspects of
the sample and of the interactions between protein and compounds, it is useful to analyze all acquired data in order to identify
any potential ligands. We have found a simple empirical grouping
system to be useful when analyzing binding data from multiple
LO-NMR experiments, where “class 1” refers to a compound
showing binding and displacement in all three acquired experiments, “class 2” refers to a compound showing binding and displacement in any two of the three experiments, and “class 3” refers
to a compound showing binding and displacement in only one of
the three experiments (Fig. 4).
However, when considering this combined analysis of all
acquired data, it is important to recognize that many phenomena
can give rise to false-positive or negative results, and as such consistent behavior across multiple experiments is an indicator of
increased confidence rather than a prerequisite for classification of
a compound as a putative ligand. Ligands with high confidence
levels should be prioritized for subsequent validation steps, but if
resources allow, then all putative ligands should be characterized in
order to find as complete a set as possible of fragments that bind to
the protein target.
3.6 Analysis
Software
Several software packages are available that assist in analyzing the
data from a LO-NMR FBS (see Subheading 2.5). These software
tools apply the analysis principles discussed above in an automated
or semiautomated manner to assist the user in analyzing the large
amount of data generated during a screening campaign. The reader
is advised to contact the software vendors directly for more details
on the availability and use of these tools.
3.7 Singleton
Validation
Compounds identified as putative ligands from the analysis of
samples containing mixtures of compounds should be verified as
“singletons.” A sample is prepared containing the compound of
interest along with protein and buffer as determined previously.
LO-NMR experiments are then acquired as described above, competitor molecule added, and the set of LO-NMR experiments
Fragment Screening by NMR
263
Relaxation-Filtered 1D
For slowly tumbling molecules, magnetization relaxes rapidly via
the T2 or T1ρ routes, while for rapidly tumbling molecules, these
relaxation rates are significantly slower. This differential relaxation
is used in the relaxation-filtered 1D to selectively attenuate the
signal of molecules bound to a slowly tumbling protein compared
to those molecules that are free in solution. Since the magnetization does not recover after dissociation, the relaxation-filtered 1D
therefore reflects the unbound population of the ligand. After
displacement by a potent competitor molecule, the unbound population of the compound is increased and the signal in the
relaxation-filtered 1D increases as a result.
3.5.5 Combined Data
Analysis
Since each experiment contains data reflecting different aspects of
the sample and of the interactions between protein and compounds, it is useful to analyze all acquired data in order to identify
any potential ligands. We have found a simple empirical grouping
system to be useful when analyzing binding data from multiple
LO-NMR experiments, where “class 1” refers to a compound
showing binding and displacement in all three acquired experiments, “class 2” refers to a compound showing binding and displacement in any two of the three experiments, and “class 3” refers
to a compound showing binding and displacement in only one of
the three experiments (Fig. 4).
However, when considering this combined analysis of all
acquired data, it is important to recognize that many phenomena
can give rise to false-positive or negative results, and as such consistent behavior across multiple experiments is an indicator of
increased confidence rather than a prerequisite for classification of
a compound as a putative ligand. Ligands with high confidence
levels should be prioritized for subsequent validation steps, but if
resources allow, then all putative ligands should be characterized in
order to find as complete a set as possible of fragments that bind to
the protein target.
3.6 Analysis
Software
Several software packages are available that assist in analyzing the
data from a LO-NMR FBS (see Subheading 2.5). These software
tools apply the analysis principles discussed above in an automated
or semiautomated manner to assist the user in analyzing the large
amount of data generated during a screening campaign. The reader
is advised to contact the software vendors directly for more details
on the availability and use of these tools.
3.7 Singleton
Validation
Compounds identified as putative ligands from the analysis of
samples containing mixtures of compounds should be verified as
“singletons.” A sample is prepared containing the compound of
interest along with protein and buffer as determined previously.
LO-NMR experiments are then acquired as described above, competitor molecule added, and the set of LO-NMR experiments
Fragment Screening by NMR
263