Affinity screening using competitive binding with fluorine-19 hyperpolarized ligands

Abstract

Fluorine-19 NMR and hyperpolarization form a powerful combination for drug screening. Under a competitive equilibrium with a selected fluorinated reporter ligand, the dissociation constant (K(D)) of other ligands of interest is measurable using a single-scan Carr-Purcell-Meiboom-Gill (CPMG) experiment, without the need for a titration. This method is demonstrated by characterizing the binding of three ligands with different affinities for the serine protease trypsin. Monte Carlo simulations show that the highest accuracy is obtained when about one-half of the bound reporter ligand is displaced in the binding competition. Such conditions can be achieved over a wide range of affinities, allowing for rapid screening of non-fluorinated compounds when a single fluorinated ligand for the binding pocket of interest is known.

Keywords: NMR spectroscopy; drug discovery; hyperpolarization; protein-ligand interaction.

Figure 1

Structure of reporter ligand 4-(trifluoromethyl)benzene-1-carboximidamide (TFBC). b) Single-scan CPMG spin-echo intensities, averaged 128 times, of 100 μM TFBC without hyperpolarization, obtained in absence (○) and presence (◇) of 1.8 μM trypsin, and in the presence of both 1.8 μM trypsin and 40 μM benzamidine (□). Each point shown is the average of 581 successive data points, which were measured at time intervals of 420 μs. R₂ relaxation rates were obtained from fit to a single exponential; $R_{2,r}^{\left(f\right)}$ = 0.60 s⁻¹ (○), $R_{2,r}^{\left(nc\right)}$ = 3.00 s⁻¹ (◇), $R_{2,r}^{\left(c\right)}$ = 1.61 s⁻¹ (□).

Figure 2

CPMG spin-echo intensities of 1 μM hyperpolarized TFBC ($p_{b,r}^{\left(nc\right)}$ = 0.7 %). The data points for reference experiments with TFBC in the absence (○) and presence (◇) of 1 μM trypsin are only included in the first graph, but the fit curves are kept in all panels. The data points of hyperpolarized TFBC in the competition experiments are shown in all graphs (□). Each point shown is the average of 465 successive data points measured at time intervals of 420 μs. NMR probe background has been subtracted and data before 42.4 ms discarded.

Figure 3

a) Dependence of log(K_D,c) on α, for [[R]_tot, [P]_tot, [C]_tot] = [1, 1, 1] μM (bottom curve), [1, 1, 35] μM (middle curve) and [1, 1, 350] μM (top curve) and K_D,r = 142 μM. Points indicate the DNP-NMR measurements from Table 1. Shaded areas contain 80 % of K_D,c values obtained from the Monte Carlo simulations. b) Logarithmic histograms of K_D,c from the Monte Carlo simulations, for selected values of α on the middle curve in (a).