Structure‐based drug development is often hampered by the lack of in vivo activity of promising compounds screened in vitro, due to low membrane permeability or poor intracellular binding selectivity. Herein, we show that ligand screening can be performed in living human cells by “intracellular protein‐observed” NMR spectroscopy, without requiring enzymatic activity measurements or other cellular assays. Quantitative binding information is obtained by fast, inexpensive 1H NMR experiments, providing intracellular dose‐ and time‐dependent ligand binding curves, from which kinetic and thermodynamic parameters linked to cell permeability and binding affinity and selectivity are obtained. The approach was applied to carbonic anhydrase and, in principle, can be extended to any NMR‐observable intracellular target. The results obtained are directly related to the potency of candidate drugs, that is, the required dose. The application of this approach at an early stage of the drug design pipeline could greatly increase the low success rate of modern drug development.

In-cell NMR spectroscopy provides precious structural and functional information on biological macromolecules in their native cellular environment at atomic resolution. However, the intrinsic low sensitivity of NMR imposes a big limitation in the applicability of the methodology. In this respect, the recently developed commercial 1.2 GHz NMR spectrometer is expected to introduce significant benefits. However, cell samples may suffer from detrimental effects at ultrahigh fields, that must be carefully evaluated. Here we show the first in-cell NMR spectra recorded at 1.2 GHz on human cells, and we compare resolution and sensitivity against those obtained at 900 and 950 MHz. To evaluate the effects of different spin relaxation rates, SOFAST-HMQC and BEST-TROSY spectra were recorded on intracellular α-synuclein and carbonic anhydrase. Major improvements are observed at 1.2 GHz when analyzing unfolded proteins, such as α-synuclein, while the TROSY scheme improves the resolution for both globular and unfolded proteins.

Candidate drugs rationally designed in vitro often fail due to low efficacy in vivo caused by low tissue availability or because of unwanted side effects. To overcome the limitations of in vitro rational drug design, the binding of candidate drugs to their target needs to be evaluated in the cellular context. Here, we applied in-cell NMR to investigate the binding of a set of approved drugs to the isoform II of carbonic anhydrase (CA) in living human cells. Some compounds were originally developed toward other targets and were later found to inhibit CAs. We observed strikingly different dose- and time-dependent binding, wherein some drugs exhibited a more complex behavior than others. Specifically, some compounds were shown to gradually unbind from intracellular CA II, even in the presence of free compound in the external medium, therefore preventing the quantitative formation of a stable protein–ligand complex. Such observations could be correlated to the known off-target binding activity of these compounds, suggesting that this approach could provide information on the pharmacokinetic profiles of lead candidates at the early stages of multitarget drug design.