sample tubes. If the protein absolutely requires the presence of
glycerol or detergent, it is advised to consider screening using
19 F LO-NMR or
15 N–
1
H HSQC PO-NMR rather than
1 H
LO-NMR.
5. Typical initial screening conditions are 500 μM fragment and
10 μM protein in a suitable buffer containing 90% H 2 O/10%
D 2 O. This molar ratio typically gives an observable signal in
STD, water-LOGSY, and relaxation-filtered 1D experiments. If
relatively hydrophobic fragments are to be screened, a lower
concentration may be required owing to solubility limits. Protein concentration should be increased in situations where
lower affinity interactions are expected. 10% D 2 O is used to
ensure reliable locking under automation when mixtures of
fragments are used since this can result in high concentrations
of d6-DMSO present in the sample that could otherwise confuse the lock system.
6. The most widely used experiment for PO-NMR is the
15
N–
1 H
HSQC spectrum, which is rapid, sensitive and requires only
inexpensive
15
N labeling of the protein. Numerous variants of
the
15
N–
1 H HSQC exist; the two most applicable for FBS
validation are the
15
N–
1 H SOFAST-HMQC (which is rapid
but has a narrow excitation range than the HSQC) and the
15 N–
1 H TROSY (which is relatively slow and insensitive but
allows spectra to be acquired on significantly larger proteins
than the HSQC experiment). The
13 C–
1 H HSQC or HMQC
spectra are also a robust option for validation of putative
ligands since, in general,
13
C chemical shift perturbations are
less prone to long-range effects (and so are more localized
around the ligand-binding site) than is the case for
15 N. However, this typically requires relatively expensive
13 C
labeling of the protein, which can apply practical limitations on
the number of samples that can be screened using this
approach. It is also worth considering the use of simple
1 H
spectra—either 1D or 2D homonuclear TOCSY spectra. In
favorable circumstances (notably proteins of less than 20 kDa
where good chemical shift dispersion is observed and the inherent linewidth of the protein is not excessive), these can be
highly informative and provide robust validation of a putative
ligand without the requirement for specific isotopic labeling.
References
1. Erlanson DA (2012) Introduction to
fragment-based drug discovery. Top Curr
Chem 317:1–32. https://doi.org/10.1007/
128_2011_180
2. Erlanson DA, Fesik SW, Hubbard RE,
Jahnke W, Jhoti H (2016) Twenty years on:
the impact of fragments on drug discovery.
Nat Rev Drug Discov 15(9):605–619.
https://doi.org/10.1038/nrd.2016.109
3. Fischer M, Hubbard RE (2009) Fragmentbased ligand discovery. Mol Interv 9
(1):22–30. https://doi.org/10.1124/mi.9.
1.7
268
Ben J. Davis
glycerol or detergent, it is advised to consider screening using
19 F LO-NMR or
15 N–
1
H HSQC PO-NMR rather than
1 H
LO-NMR.
5. Typical initial screening conditions are 500 μM fragment and
10 μM protein in a suitable buffer containing 90% H 2 O/10%
D 2 O. This molar ratio typically gives an observable signal in
STD, water-LOGSY, and relaxation-filtered 1D experiments. If
relatively hydrophobic fragments are to be screened, a lower
concentration may be required owing to solubility limits. Protein concentration should be increased in situations where
lower affinity interactions are expected. 10% D 2 O is used to
ensure reliable locking under automation when mixtures of
fragments are used since this can result in high concentrations
of d6-DMSO present in the sample that could otherwise confuse the lock system.
6. The most widely used experiment for PO-NMR is the
15
N–
1 H
HSQC spectrum, which is rapid, sensitive and requires only
inexpensive
15
N labeling of the protein. Numerous variants of
the
15
N–
1 H HSQC exist; the two most applicable for FBS
validation are the
15
N–
1 H SOFAST-HMQC (which is rapid
but has a narrow excitation range than the HSQC) and the
15 N–
1 H TROSY (which is relatively slow and insensitive but
allows spectra to be acquired on significantly larger proteins
than the HSQC experiment). The
13 C–
1 H HSQC or HMQC
spectra are also a robust option for validation of putative
ligands since, in general,
13
C chemical shift perturbations are
less prone to long-range effects (and so are more localized
around the ligand-binding site) than is the case for
15 N. However, this typically requires relatively expensive
13 C
labeling of the protein, which can apply practical limitations on
the number of samples that can be screened using this
approach. It is also worth considering the use of simple
1 H
spectra—either 1D or 2D homonuclear TOCSY spectra. In
favorable circumstances (notably proteins of less than 20 kDa
where good chemical shift dispersion is observed and the inherent linewidth of the protein is not excessive), these can be
highly informative and provide robust validation of a putative
ligand without the requirement for specific isotopic labeling.
References
1. Erlanson DA (2012) Introduction to
fragment-based drug discovery. Top Curr
Chem 317:1–32. https://doi.org/10.1007/
128_2011_180
2. Erlanson DA, Fesik SW, Hubbard RE,
Jahnke W, Jhoti H (2016) Twenty years on:
the impact of fragments on drug discovery.
Nat Rev Drug Discov 15(9):605–619.
https://doi.org/10.1038/nrd.2016.109
3. Fischer M, Hubbard RE (2009) Fragmentbased ligand discovery. Mol Interv 9
(1):22–30. https://doi.org/10.1124/mi.9.
1.7
268
Ben J. Davis