[33], it is common practice to screen mixtures of compounds in the
same experiment. The size of the mixture varies from 4 to 10 compounds (for
1 H LO-NMR) to as high as 30 (for PO-NMR or
19 F-NMR). Where the compounds are observed directly, it is usually possible to directly distinguish ligands from nonbinding molecules based on their 1D NMR spectra if care is taken in order to
minimize spectral overlap within the mixture. In the case of
PO-NMR, deconvolution of the mixture (e.g., by subsequent analysis of subsets of molecules) or the application of orthogonal
methods must be applied to identify which specific ligand is responsible for the observed chemical shift perturbations. One potential
issue when screening mixtures of compounds is the potential for
compound–compound interactions. Although rare, these interactions do occur [15], and so ligands identified directly from mixtures
should be tested as single compounds before confirmation as a
putative hit from the NMR screen.
Having determined which set of NMR experiments are most
suitable for the target in question, several key steps need to be undertaken before the screen can begin. As discussed previously, the focus of
this chapter is on ligand-observed fragment screening by NMR
(LO-NMR) using
1
H-detected experiments since this is the most
commonly used experimental configuration. Many excellent reviews
of PO-NMR and
19
F-NMR FBS have been published, and these cover
details, which are outside of the scope of this chapter. However, some
features of NMR FBS are common to all approaches. (1) A screening
library of fragments is required; in addition to the typical principles
used to design a fragment library [34–36], these fragments need to be
compatible with the NMR screening method to be applied. The
fragments also need to be characterized in terms of solubility, stability
in both DMSO and aqueous solutions [15], and PAINS-like behavior
[37]. (2) Characterization of the protein and any competitor ligands is
vital in order to avoid erroneous results caused by unusual behavior of
these molecules. NMR screening experiments are relatively time consuming (often taking several days to run a complete screen), and so the
prior assessment and optimization of the stability of the protein (see
Chapter 1) and competitor ligands is key.
After identification of suitable conditions, the NMR screen can
then be completed using the methods discussed previously. The
screening data are then analyzed, either manually or in a semiautomated manner, and a set of preliminary fragment hits identified.
An important final step, which lies outside of the scope of this
chapter, is the validation of these initial fragment hits using orthogonal methods. Often this validation step will be performed using an
alternative biophysical method (such as SPR, MST, or X-ray crystallography). However, PO-NMR is also often used as a validation
method orthogonal to an initial screen performed using LO-NMR
or
19
F-NMR. This case will be discussed further below.
252
Ben J. Davis
same experiment. The size of the mixture varies from 4 to 10 compounds (for
1 H LO-NMR) to as high as 30 (for PO-NMR or
19 F-NMR). Where the compounds are observed directly, it is usually possible to directly distinguish ligands from nonbinding molecules based on their 1D NMR spectra if care is taken in order to
minimize spectral overlap within the mixture. In the case of
PO-NMR, deconvolution of the mixture (e.g., by subsequent analysis of subsets of molecules) or the application of orthogonal
methods must be applied to identify which specific ligand is responsible for the observed chemical shift perturbations. One potential
issue when screening mixtures of compounds is the potential for
compound–compound interactions. Although rare, these interactions do occur [15], and so ligands identified directly from mixtures
should be tested as single compounds before confirmation as a
putative hit from the NMR screen.
Having determined which set of NMR experiments are most
suitable for the target in question, several key steps need to be undertaken before the screen can begin. As discussed previously, the focus of
this chapter is on ligand-observed fragment screening by NMR
(LO-NMR) using
1
H-detected experiments since this is the most
commonly used experimental configuration. Many excellent reviews
of PO-NMR and
19
F-NMR FBS have been published, and these cover
details, which are outside of the scope of this chapter. However, some
features of NMR FBS are common to all approaches. (1) A screening
library of fragments is required; in addition to the typical principles
used to design a fragment library [34–36], these fragments need to be
compatible with the NMR screening method to be applied. The
fragments also need to be characterized in terms of solubility, stability
in both DMSO and aqueous solutions [15], and PAINS-like behavior
[37]. (2) Characterization of the protein and any competitor ligands is
vital in order to avoid erroneous results caused by unusual behavior of
these molecules. NMR screening experiments are relatively time consuming (often taking several days to run a complete screen), and so the
prior assessment and optimization of the stability of the protein (see
Chapter 1) and competitor ligands is key.
After identification of suitable conditions, the NMR screen can
then be completed using the methods discussed previously. The
screening data are then analyzed, either manually or in a semiautomated manner, and a set of preliminary fragment hits identified.
An important final step, which lies outside of the scope of this
chapter, is the validation of these initial fragment hits using orthogonal methods. Often this validation step will be performed using an
alternative biophysical method (such as SPR, MST, or X-ray crystallography). However, PO-NMR is also often used as a validation
method orthogonal to an initial screen performed using LO-NMR
or
19
F-NMR. This case will be discussed further below.
252
Ben J. Davis