As noted in our poll last year,
ligand-detected NMR ranks among the most popular fragment-finding approaches.
The various methods are able to detect even weak binders, so determining
affinities is important to effectively prioritize hits. This, however, can be time-consuming.
In a recent J. Am. Chem. Soc. paper, Ridvan Nepravishta, Dušan Uhrín,
and collaborators at CRUK Scotland Institute, University of Edinburgh, and Universidad
de Sevilla present a clever way to speed up the process.
Normally, NMR spectra of small
molecules show multiple spectral lines, with each line corresponding to a
different atom or atoms (typically protons). Indeed, depending on the details,
the signal from a single proton might be split into multiple peaks. All these
signals are great for understanding the details of individual atoms, but the
more lines there are, the lower the signal to noise ratio. For maximum sensitivity
it would be nice to combine all the lines from all the atoms in a given
molecule into a single, intense singlet. This is exactly what the researchers
have done.
The approach is called Sensitive,
Homogeneous And Resolved PEaks in Real time, or SHARPER. For the NMR aficionados
out there, “when placed before the acquisition of the NMR signal, a train of
spin-echoes in the form of the Carr-Purcell-Meiboom-Gill (CPMG) pulse sequence
suppresses evolution due to chemical shifts and J couplings…. All these
attributes of the CPMG pulse sequence are maintained when the spin-echo train
is employed during the acquisition of the NMR signal. However, this time, the
outcome is not a regular spectrum, but under certain conditions, a single
spectral line formed as a sum of Lorentzian lines of contributing spins.”
The researchers initially applied
SHARPER to two commonly used ligand-detected methods: 1H STD, which
we wrote about here, and 1H CPMG, which we wrote about here. The
first test system was human serum albumin (HSA) binding to naproxen. Keeping
protein concentration constant at 9 µM and varying ligand concentration gave
similar KD values (210-280 µM) for standard STD, STD SHARPER, and
CPMG SHARPER (conventional CPMG failed due to insensitivity at lower ligand
concentrations). These values are an order of magnitude higher than those
reported using SPR and ITC (25 and 10 µM, respectively) because of the high
protein and ligand concentrations needed for conventional NMR approaches; when
the SHARPER experiments were rerun at 1 µM HSA, the KD values were 39
µM. Several other HSA ligands also gave good agreement with the literature.
Next, the researchers applied STD
SHARPER to the anti-cancer target fascin, which we wrote about in 2019. An examination
of 11 ligands from that study gave good agreement with the published dissociation
constants. Importantly, SHARPER was faster than conventional approaches, with
15 KD determinations per day instead of four.
Not content with this four-fold
improvement in throughput, the researchers developed a new experiment based on
line broadening called 1H LB SHARPER. This allows the determination
of 48 dissociation constants per day, and the results for HSA and fascin agreed
with the other methods.
One of the most time-consuming
aspects of most NMR-based affinity measurements is preparing and analyzing
samples at multiple ligand concentrations, so the researchers turned to machine
learning to choose which ligand concentrations would be most informative and
choose just two of them rather than the six or more commonly used. This worked
too, thereby potentially increasing throughput to 144 dissociation constants
per day.
The researchers suggest that SHARPER
could also be applied to some of the other recent NMR techniques we’ve
discussed, such as PEARLScreeen and photo-CIDNP. Although I always emphasize
that I’m no NMR spectroscopist, this strikes me as a neat, practical approach.
What do you think?