Structure, Dynamics and Function of biomolecules by NMR

High-Pressure NMR



Unfolding Protein with High Hydrostatic Pressure: the origin of ∆V.

The physical basis for the effects of pressure on protein structure and stability remains controversial, in contrast to a relatively clear physical understanding of temperature effects on protein structure. The fundamental observation of pressure effects has been that over most of the accessible temperature range, the application of pressure leads to the unfolding of proteins, indicating that the volume change (∆V) upon unfolding is negative, i.e., the specific molar volume of the unfolded state is smaller that that of the folded state. This observation is rather counter-intuitive, as one might expect that pressure would simply compress the folded state of the protein. While this compression does occur, pressure eventually unfolds most proteins, due to this negative volume change upon unfolding and Le Chatelier's principle, which states that that application of any perturbation will shift a system in equilibrium toward the state that alleviates the perturbation. In the case of pressure perturbation, this is the state that occupies the smallest volume, i.e., the unfolded state.



Unfolding of SNase and its cavity mutants: a fluorescence study. A) SNase 3D structure (hyper-stable variant ∆+PHS): the native cavity is colored in light blue, the mutated residues in red. B) Pressure dependence of the SNase tryptophan (Trp140) emission fluorescence. At low pressure, the fluorescence emission is intense, with a frequency located in the « blue » region of the visible light spectrum, indicative of a tryptophan residue buried in the hydrophobic core of a folded protein. At high pressure, the emission drop and shifts to « red », suggesting a solvent (water) exposed tryptophan residue, as expected in an unfolded protein C) The center of mass of the fluorescence spectra reported in B) are plotted as a function of the pressure. The 3 curves stand for 3 different samples with 3 different guanidinium chloride concentrations. Addition of this chaotropic reagent is necessary to unfold SNase in the pressure range experimentally available (≤ 300 MPa). The sigmoidal behavior of these curves suggests a two-states equilibrium between the native and the unfolded protein. Note the similar slope within the 3 curves, suggesting that addition of guanidinium chloride has little effect on ∆V. D) Measured ∆V on ∆+PHS SNase and 10 of its cavity mutants: introducing an additional cavity (or enlarging the original native cavity) brings a concomitant increase in ∆V (adapted from Roche et al., 2012a).


The origin of ∆V remains matter of debate. Several effects have been put forward: (i) a difference in compressibility between the native and the unfolded states of proteins, (ii) pressure induced changes in the structure of bulk water, (iii) density difference for water molecules close to polar or apolar amino acids, (iv) lost of dehydrated cavities inside the native structure of a protein, or a combination of these different effects. Recently, using a model protein (the Staphyloccocal Nuclease hyper-stable variant: ∆+PHS SNase) and several of its mutants, we have shown that the main contribution to ∆V was the presence of residual cavities inside the native 3D structure (Roche et al, 2012a ; Rouget et al., 2011).


Coupling High Hydrostatic Pressure to NMR.

While protein unfolding does not bring any chemical modification of the protein sequence, it considerably modifies its 3D structure. This perturbation can be measured locally, residue-per-residue, by following the evolution of the correlation peaks on HSQC experiments recorded with increasing pressures. In a two-state model of the unfolding/refolding equilibrium, the exchange between the native and unfolded forms occurs generally on a time scale slower than the time scale of the NMR measurement. Thus, we do not observe chemical shift variations (a displacement of the cross-peaks on the HSQC spectra, indicative of fast exchange) as a function of pressure, but the disappearance of correlation peaks belonging to the native form, with the concomitant appearance of new peaks (centered at 8.5 ppm on the proton chemical-shift axis) that correspond to the spectrum of the unfolded species. The weak spectral dispersion of these cross-peaks corresponds to an unfolded protein: the native 3D structure turns into a "random coil-like" structure, yielding a similar environment for all residues, hence for all amide groups. The residual dispersion comes from a weak "chemical" effect due to the different nature of the side chain between different the 20 natural amino acids used in a protein.


Pressure dependence of the ∆+PHS SNase 2D [1H-15N] HSQC spectrum.



High pressure NMR and protein unfolding: steady-state study.

Measurement of "local" thermodynamic parameters.

Contrary to fluorescence spectroscopy, for instance, which gives a "global" value for the parameters ∆G0 and ∆V, NMR yields "local" residue specific values of ∆G0 and ∆V. For our model reference protein (∆+PHS SNase), unfolding is a nearly cooperative phenomenon: most of the residues "see" a similar ∆V of ≈ 80 ml/mole, a value close to the "global" value measured with fluorescence spectroscopy. Nevertheless, in some areas of the protein, the measured residue-specific ∆V values fall below this average value (< 30 ml/mole), skewing the distribution (see Figure below), and suggesting the presence of folding intermediates, i.e. partially folded conformers having some degree of stability, in the protein energy landscape.


From left to right:

- Sigmoidal decrease of the intensity measured for 4 cross-peaks in the native 2D [1H-15N] HSQC spectrum of ∆+PHS SNase (see Fig.2 and 3) as a function of pressure. The lines come from the fit of the experimental intensities with the characteristic equation for a 2-states equilibrium (Equ. [2]).

- ∆V values obtained through the fit of the intensity decrease of the 2D [1H-15N] HSQC cross-peaks with pressure, plotted versus the protein sequence.

- Histogram of the ∆V values measured for ∆+PHS SNase. This histogram has been fitted with a Gaussian function (in red) that gives an average value for ∆V close to the one measured with fluorescence. (Adapted from Roche et al., 2012a).


Structural characterization of folding intermediates.

We used the following procedure to extract structural and energetic information about the folding intermediates from the pressure-dependent multi-dimensional NMR data. After normalizing the residue-specific denaturation (see Figure below) curves obtained from the amide cross-peak intensity decays measured on the HSQC experiments recorded at variable pressure, the value of 1 for a given cross-peak (I = IF = 1) can be associated with a probability of 1 (100%) for the corresponding residue to be in the native state, with all the native contacts present. Similarly, a residue for which, at the same pressure, the corresponding cross-peak has disappeared (I = IU = 0) from the HSQC spectrum has a probability equal to zero to be in a native state: it belongs to an unfolded state where all the native contacts are lost.

Now, we consider two residues i and j, in an intermediate situation where the probability to be in a folded states are p(i) = 0,99 et p(j) = 0,68, respectively, at a given pressure (80 MPa in the example displayed in the Figure. For two residues that are in contact in the native state, we assume the probability p(i,j) to be in contact at pressure (p) 80 MPa to be given by the product of the individual probabilities p(i,j) = p(i) x p(j) = 0,67. Pressure dependent contact maps can be constructed based on the contact map of the folded protein: a color code is used in order to report in the contact map the pressure dependent probability of contact).

3D structures can be calculated from a contact map, using the contacts as restraints in a molecular modeling program. We have "weight" the restraint list with contact probabilities calculated at a given pressure: several (usually 100) restraint lists were generated, instead of one. A contact between two residues associated to a probability of 0.8 will be randomly added to 80 lists over 100, if the probability is only 0.4, the corresponding restraints will be in 40 lists over 100, etc...




Tracking Folding intermediates with high pressure NMR.

A) Residue-specific normalized denaturation curves can be used to calculate contact probabilities between 2 residues (here between residue 38 and 117) at a given pressure (80 MPa).

B) Contact probabilities are reported with a color code (above the diagonal) in a contact map. This map will be used to define the lists of native contacts at a given pressure (80 MPa).

C) These contact lists will be translated into restraint lists for use by molecular dynamics program allowing the modeling of the corresponding conformation.

D) Conformation analysis: on the graph, the populations (Ln(P)) are plotted as a function of the residual native contacts (Q), at 5 different pressures. The red curve shows the result for a pressure of 80 MPa. It presents 3 minima: the most populated (Q=0.85) corresponds to the population of native conformers at this specific pressure; the minimum at Q=0.1 corresponds to the population of unfolded states; the minimum (shoulder) at Q=0.55 corresponds to the population of an intermediate states which structure is represented in the insert. (Adapted from Roche et al., 2012a).

Then we have modelled, classed and analyzed the 3D structures obtained from coarse grained simulations using the 100 different restraint lists. In the case of our model protein ∆+PHS SNase, this analysis allowed us to identify a population of conformers corresponding to a folding intermediate where the C-terminal α-helix is unfolded, whereas the N-terminal ß-barrel maintains its native structure. It is worth noting that this folding intermediate has been also identified through other high pressure NMR experiments, concerning the measurement of amide proton / deuteron exchange (Roche et al., 2012b).


High pressure NMR and protein unfolding: kinetics study.

A complete understanding of the protein folding / unfolding phenomenon requires, in addition to the spatial, structural description of the energy landscape, a temporal description of the sequence of events along the folding pathway followed by the protein. Such a description relies on the measurement of kinetic parameters, after perturbation of the thermodynamic equilibrium between the folded and unfolded conformers of the protein. In addition to kinetic parameters, these measurements allow characterizing the transition state associated to a first order kinetics, commonly used to describe an equilibrium reaction between two states. Thus, using pressure to unfold the protein will give access to folding (kf) and unfolding (ku) rates, as well as to the volume of the transition state ensemble (TSE): the transition state of a protein being relatively heterogeneous, it is better described as an ensemble of (close) states rather than an unique conformer (Figure below).


Protein folding and kinetics measurements. A) Unfolding reaction coordinates as a function of the free energy. The constants kf and ku stand for the folding and unfolding rates, respectively, and τ is the experimentally measured relaxation time. B) Scheme of the volumetric diagram of the folding / unfolding reaction. F, TSE and U represent the folded, transition, and unfolded state, respectively. C) Experimental volumetric diagrams obtained for ∆+PHS SNase and 10 of its "cavity" variants. ∆Veq stands for the volume difference between the folded and unfolded states as measured at thermodynamic equilibrium (steady state measurements), ∆Vf‡ is the activation volume between the folded and transition states obtained from kinetics measurements. (Adapted from Roche et al., 2013b).

The return to a new equilibrium after perturbation can be monitored by different spectroscopy techniques that give access to a "global" measurement of the kinetic parameters for the folding / unfolding reaction (Fluorescence, IR...). A "local" description of the kinetic parameters and of the transition state ensemble implies the use of a technique combining spatial resolution, allowing a precise local description of the time evolution of the structure of the protein, and a sufficient time resolution. At atmospheric pressure the return to thermodynamic equilibrium after a perturbation can be relatively fast (few ms to a few minutes, at most). NMR has high spatial resolution, but its time resolution is limited: the recording time of the 2D [1H-15N] HSQC spectra ranges from 10 to 40 minutes, depending on the sample concentration and the expected spectral resolution. Thus, such experiments can be used only in the case of proteins having extremely slow relaxation times. Practically, P-jump kinetics measurements consist in recording a series of 2D HSQC (or SOFAST-HMQC) after a pressure jump, in order to correctly sample the exponential decay of the cross-peak intensity during unfolding (in the case of a "positive" P-jump, where pressure is increased) (Figure 8), (Figure below).


High Pressure NMR and kinetics measurements. A) Series of 2D [1H-15N] HSQC spectra used to sample a positive P-jump of 20 MPa (100 to 120 Mpa) on a sample of ∆+PHS SNase. Individual measuring time for each HSQC experiment was 20 minutes, for a protein concentration of 1 mM. B) Times evolution of the amide cross-peak intensity for 4 residues. The curves were obtained by exponential fitting of the intensity values, giving the value of τ = 1/(ku+kf).

Pressure, because of the positive activation volume for folding, has the unique feature to slow down the folding reaction: a reaction completed in few seconds at atmospheric pressure will take several minutes to few hours at higher pressure, making P-jump NMR monitoring feasible. For example, ∆+PHS SNase at pressure above 1 kbar, exhibits residue-specific relaxation times greater than 10 hours! This is all the more true since methodological advances have been realized during the last decade in the field of "real-times" measurement of NMR multidimensional experiments. Now, 2D correlation experiments can be acquired in few tens of seconds (sometimes even in less than one second!) instead of few tens of minutes, with a good sensitivity and enough spectral resolution.

Analogous to the steady state measurements, combining high pressure and NMR spectroscopy in relaxation kinetics allows obtaining a local dynamic view of the folding pathway being taken by each residue, through the measurement of the kinetic parameter τ and the extraction of the activation volume for each residue. It is worth noting that when looking to the activation volumes, average values obtained for the reference protein ∆+PHS SNase and its mutants, very different results are obtained: the reference protein shows an activation volume ∆V‡f close to the equilibrium ∆V value, indicating that the molar volume of the transition state is quite close to that of the folded state, suggesting a "dehydrated" TSE where most of the native cavities are still present (Roche et al., 2013b) (Figure below).


High-pressure NMR kinetics measurements. A) Examples for the evolution of the residue-specific relaxation time τ values as a function of the final pressure for the pressure jump. Curves are obtained with the fit of the experimental values and allow determining the value of the activation volume for the folding reaction "see" by each residue. B) Comparison of ∆V at equilibrium (dark blue) and ∆Vf‡ activation volumes (light blue) as a function of the protein sequence. (Adapted from Roche et al., 2013b).




- Deciphering the mechanism of protein unfolding under high hydrostatic pressure

High-Pressure unfolding of model proteins (SNase, PP32, PASTA...) is currently under study to define the thermodynamics of the phenomenon (associated energy landscapes, cooperativity, etc...)


- High Pressure Study of Hemoproteins

. Volumetric properties underlying ligand binding in a monomeric hemoglobin: A high-pressure NMR study. Collaboration with J. Lecomte, J. Hopkins Uni. (USA). The 2/2 hemoglobin of the cyanobacterium Synechococcus sp. PCC 7002, GlbN, coordinates the heme iron with two histidines and exists either with a b heme or with a covalently attached heme. The binding of exogenous ligands displaces the distal histidine and induces a conformational rearrangement involving the reorganization of internal void volumes. We monitored the perturbation induced by pressure on the ferric bis-histidine and cyanide-bound states of GlbN using 1H-15N HSQC NMR spectroscopy. We inspected the outcome with a statistical analysis of 170 homologous 2/2 hemoglobin sequences. The high-pressure dissection combined with protein sequence analysis established locations with volumetric signatures relevant to residual communication of 2/2 hemoglobins (Dellarole et al., 2013).

. Characterization of High Hydrostatic Pressure-Induced "excited" functional states in Neuroglobin Cavity. Collaboration with B. Vallone, Uni. La Sapienza, Roma, (Italy). Neuroglobin (Ngb) is expressed in vertebrate brains and belongs to a branch of the globin family that diverged early in evolution. Ngb is involved in protection of neurons from hypoxia and is likely to be a sensor of hypoxia in cells to trigger neuroprotection. Nevertheless, its functional role in the cell is not completely understood. Contrary to other globins, Ngb is characterized by endogenous heme iron hexacoordination in the absence of external ligands. To allow binding to the sixth heme coordination site, not only must the bond with the distal His64(E7) breaks, but a conformational transition has to take place to create space to accommodate the external ligand. Determination of the structure of murine Ngb liganded to CO (NgbCO) showed that this protein adopts a "heme-sliding mechanism" to remove the blockage of the binding site. The main goal of this research is to characterize this mechanism, mostly using heteronuclear NMR relaxation and variable pressure NMR spectroscopy (work in progress).

. High-Pressure study of catch-bonds involved in cell adhesion. Collaboration F. Guerlesquin, LISM-IMM, UMR 7255, Marseille. Our objective is to use HP-NMR to investigate the conformation of galectin-3, a soluble lectin able to associate with glycans at the cell surface, induced by external forces commonly referred as "catch-bonds". Catch-bonds are force-activated bond originated from dragging forces imposed by fluid flow acting on cells or bacteria. (work in progress)