Determination of intracellular protein-ligand binding affinity by competition binding in-cell NMR

Abstract

Structure-based drug development suffers from high attrition rates due to the poor activity of lead compounds in cellular and animal models caused by low cell penetrance, off-target binding or changes in the conformation of the target protein in the cellular environment. The latter two effects cause a change in the apparent binding affinity of the compound, which is indirectly assessed by cellular activity assays. To date, direct measurement of the intracellular binding affinity remains a challenging task. In this work, in-cell NMR spectroscopy was applied to measure intracellular dissociation constants in the nanomolar range by means of protein-observed competition binding experiments. Competition binding curves relative to a reference compound could be retrieved either from a series of independent cell samples or from a single real-time NMR bioreactor run. The method was validated using a set of sulfonamide-based inhibitors of human carbonic anhydrase II with known activity in the subnanomolar to submicromolar range. The intracellular affinities were similar to those obtained in vitro, indicating that these compounds selectively bind to the intracellular target. In principle, the approach can be applied to any soluble intracellular target that gives rise to measurable chemical shift changes upon ligand binding.

Keywords: bioreactor; carbonic anhydrase inhibitors; competition binding; in-cell NMR; real-time NMR.

Figure 1

Chemical structures of the sulfonamide-derived compounds analyzed in this study. The K _i and K _d values previously reported in vitro for CA II are shown (see Table 3 ▸).

Figure 2

Overlay of ¹H–¹⁵N NMR spectra of (a) pure [¹⁵N]-labeled CA II and (b) cells expressing [¹⁵N]-His-labeled CA II either in the absence of ligands (black) or bound to AAZ (red), MZA (green), SLC (blue) and ETZ (magenta). The signals of the free protein are labeled with the corresponding residue number and atom type (Shimahara et al., 2007; Vasa et al., 2019 ▸).

Figure 3

Closed-tube in-cell ¹H–¹⁵N NMR spectra of cells treated with (a) 50 µM AAZ + 100 µM MZA, (b) 50 µM MZA + 100 µM SLC and (c) 50 µM AAZ + 100 µM SLC. Signals arising from CA II-bound AAZ, MZA and SLC are indicated with blue, black and red arrows, respectively. Only the least overlapped signals (marked with asterisks) were integrated for nonlinear fitting (see Supplementary Fig. S3).

Figure 4

Bioreactor in-cell NMR of cells treated with constant 10 µM MZA and increasing amounts of SLC (bioreactor run 1 in Table 1 ▸) and subsequent data analysis. (a–c) Representative ¹H–¹⁵N NMR spectra at different time points and concentrations of SLC. Signals arising from SLC are marked with red arrows. (d) NMR spectra of the pure components, i.e. CA II–MZA (black) and CA II–SLC (red), reconstructed by MCR-ALS. (e) Concentration profiles of CA II–MZA (black squares) and CA II–SLC (red circles) obtained by MCR-ALS. (f) Bound fractions obtained from the plateau values after each step of the run plotted as a function of SLC concentration. Binding curves from nonlinear fitting are shown as dashed lines.

Figure 5

Bioreactor in-cell NMR of cells treated with constant 10 µM MZA and increasing amounts of ETZ (bioreactor run 2 in Table 1 ▸) and subsequent data analysis. (a–c) Representative ¹H–¹⁵N NMR spectra at different time points and concentrations of ETZ. Signals arising from ETZ are marked with magenta arrows. (d) NMR spectra of the pure components, i.e. CA II–MZA (black) and CA II–ETZ (magenta), reconstructed by MCR-ALS. (e) Concentration profiles of CA II–MZA (black squares) and CA II–ETZ (magenta circles) obtained by MCR-ALS. (f) Bound fractions obtained from the plateau values after each step of the run plotted as a function of ETZ concentration. Binding curves from nonlinear fitting are shown as dashed lines.

Figure 6

Bioreactor in-cell NMR of cells treated with constant 10 µM MZA and increasing amounts of ETZ (bioreactor run 3 in Table 1 ▸) and subsequent data analysis. (a–c) Representative ¹H–¹⁵N NMR spectra at different time points and concentrations of ETZ. Signals arising from ETZ are marked with magenta arrows. (d) NMR spectra of the pure components, i.e. CA II–MZA (black) and CA II–ETZ (magenta), reconstructed by MCR-ALS. (e) Concentration profiles of CA II–MZA (black squares) and CA II–ETZ (magenta circles) obtained by MCR–ALS. (f) Bound fractions obtained from the plateau values after each step of the run plotted as a function of ETZ concentration. Binding curves from nonlinear fitting are shown as dashed lines.