Trapping single proteins in self-assembled molecular cages let researchers cool a 22.5-kilodalton enzyme below 0°C and use NMR to resolve two conformations around its active site that average into one signal at room temperature. The cages prevented aggregation in an acetone–water mixture, giving spectra down to −30°C, yet let ligands enter, revealing at −20°C which normally invisible sugar hydroxy groups contact the enzyme lysozyme. The approach exposes hidden protein motions and ligand contacts, and could extend subzero NMR to proteins up to 35 kilodaltons.
By enclosing single protein molecules in self-assembled molecular cages, researchers could cool an enzyme below 0°C. Nuclear magnetic resonance (NMR) spectroscopy then resolved two distinct conformations around its active site, which appear as one averaged signal at room temperature. Reporting in Chemistry – An Asian Journal, the team recorded multidimensional NMR spectra of the 22.5-kilodalton cutinase-like enzyme (CLE) in an acetone–water mixture. They detected signals down to −30°C. Earlier subzero NMR studies had used only proteins of 10–15 kilodaltons. At −20°C, the same strategy also showed which hydroxy (OH) groups of a sugar contact the enzyme lysozyme. These contacts match hydrogen bonds seen in known crystal structures.
Cooling a protein slows internal motions that are otherwise too fast for NMR to resolve. However, standard solution NMR in water is limited to temperatures above 0°C, and the alternatives, supercooled water and high pressure, need specialized setups. Water–organic solvent mixtures stay liquid at lower temperatures, but most proteins unfold or aggregate in them. Reverse micelles can protect proteins, but they also cut them off from molecules that would bind to them.
The team attached each protein by one end inside a hollow sphere that self-assembles from 12 palladium ions and 24 organic linkers. Keeping proteins apart prevents aggregation, and the cage's large openings still let binding molecules diffuse in. Measurements used 70% acetone–water, which freezes only at −34°C and is less viscous than cold water. At −30°C, a 20-kilodalton protein is calculated to tumble in 14 nanoseconds there, compared with 112 nanoseconds in supercooled water, which limits signal broadening. The researchers followed caged CLE by NMR. They also used saturation transfer difference (STD) NMR, which flags the ligand atoms touching a protein, to study how a three-unit sugar, (GlcNAc)3, binds to caged lysozyme.
In 70% acetone, free CLE immediately lost its activity and precipitated. Caged CLE kept its native structure and activity and gave well-resolved spectra from 0°C to −20°C. At −5°C, signals from a subset of amino acids split in two, and more split at −10°C. These residues clustered mainly around the catalytic site and in nearby strands and loops. The paired signals had similar intensities, which indicates two similarly populated states and suggests concerted changes around the active site. Because the native fold is preserved, these states give access to otherwise hidden active-site motions, although their populations and exchange rates may differ from those in living systems. Below about −15°C, CLE gradually cold-denatured and lost its hydrophobic-core signals by −25°C, but it returned to its native structure on rewarming. Cooling also slows the exchange of OH protons with water, and the sugar's normally invisible OH signals appeared at −10°C and below. At −20°C, STD NMR showed contacts at the sugar's 6-OH, 6′-OH, 3′-OH and 4″-OH groups, matching known hydrogen bonds. Uncaged lysozyme gave no such signals, which the authors attribute to unfolding and aggregation. Caged lysozyme stayed highly active, indicating a folded catalytic core, although partial cold unfolding cannot be fully excluded.
The method directly detected OH-group contacts in protein–sugar binding without isotope labeling or specialized pulse sequences, while leaving the protein accessible to its ligand. Because these cages can hold proteins of up to 35 kilodaltons, the authors suggest the approach could widen the range of proteins open to subzero NMR. The STD data identify the sugar's contact points. To also locate the binding site on the protein, the authors propose combining the method with multidimensional NMR of isotope-labeled proteins. They also name the side-chain NH groups of lysine and arginine as promising future targets.