Pressure and Chemical Unfolding of an α-Helical Bundle Protein: The GH2 Domain of the Protein Adaptor GIPC1

When combined with NMR spectroscopy, high hydrostatic pressure is an alternative perturbation method used to destabilize globular proteins that has proven to be particularly well suited for exploring the unfolding energy landscape of small single-domain proteins. To date, investigations of the unfolding landscape of all-β or mixed-α/β protein scaffolds are well documented, whereas such data are lacking for all-α protein domains. Here we report the NMR study of the unfolding pathways of GIPC1-GH2, a small α-helical bundle domain made of four antiparallel α-helices. High-pressure perturbation was combined with NMR spectroscopy to unravel the unfolding landscape at three different temperatures. The results were compared to those obtained from classical chemical denaturation. Whatever the perturbation used, the loss of secondary and tertiary contacts within the protein scaffold is almost simultaneous. The unfolding transition appeared very cooperative when using high pressure at high temperature, as was the case for chemical denaturation, whereas it was found more progressive at low temperature, suggesting the existence of a complex folding pathway.


Introduction
Although small single-domain proteins are generally found to exhibit highly cooperative two-state unfolding transitions [1,2], the thousands of interactions that stabilize their 3D structure are unlikely to form simultaneously and folding intermediates should exist along their folding pathway. Nevertheless, especially in the case of small, fast-folding single-domain proteins, folding intermediates are generally low populated at equilibrium, and cannot be easily identified when using classical thermal or chemical perturbation in association with methods applied to a single probe in the 3D structure of the protein (for instance, intrinsic fluorescence of a tryptophan residue) or with methods giving global structural information (for instance, molar ellipticity in the case of circular dichroism (CD) study).
Multidimensional NMR spectroscopy is a particularly powerful tool to obtain highresolution structural information about protein folding events because an abundance of site-specific probes can be studied simultaneously in a single spectrum. In the recent past, NMR combined with high-pressure perturbation has emerged as a powerful tool to explore in detail the folding landscape of small proteins, at a quasi-atomic resolution [3][4][5][6]. Indeed, contrary to chemical or thermal denaturation, which acts globally and depends on exposed surface area in the unfolded state, pressure denaturation depends on the elimination of the solvent-excluded internal voids, due to imperfect protein packing, by water penetration inside the core of the protein [7][8][9]. Thus, because the distribution of solvent-excluded

NMR Resonance Assignments and Solution Structure of GIPC1-GH2
Proton, nitrogen, and carbon NMR resonances of GIPC1-GH2, renumbered 1-79 for simplicity, have been assigned and its solution structure solved using essentially [ 1 H, 15 N, 13 C] triple-resonance and [ 1 H, 15 N] double-resonance 3D NMR spectroscopy (see Section 4) with the classical sequential assignment strategy. 1 H and 15 N resonances have been assigned for all amide groups of nonproline (76) residues (Figure 1), and Cα, Cβ, C' resonances for 96.2% residues. Resonance assignments have been deposited at the BMRB data bank (BMRB code 34609).
NOEs were measured on a 3D [ 1 H, 15 N] NOESY-HSQC experiment and on a 2D [ 1 H, 1 H] NOESY spectrum recorded in a deuterated buffer. Dihedral restraints (ϕ, φ, and χ 1 ) were obtained from TALOS-N [17] analysis of backbone atom chemical shifts. After conversion into distance restraints, these data sets were used with CYANA [18] to build the 3D structure of GIPC1-GH2. H-bond restraints were also used for the structure modeling. Usually, these restraints are deduced from hydrogen/deuteron (H/D) exchange experiments, yielding amide protons potentially involved in H-bonds as donor atoms. In the case of GIPC1-GH2, H/D exchange rates were unusually fast, even at low temperature (5 • C): all amide proton resonances disappeared during the few minutes needed for the setting of the experiments, preventing the use of this approach. Instead, we used CLEANEX-PM [19,20] experiments to determine which amide protons are solvent-exposed in the 3D structure. In these experiments, the water resonance was selectively excited, and water magnetization transferred to solvent-exposed amide protons with an appropriate spin-locking module. This sequence was incorporated in a conventional HSQC scheme to resolve amide peaks along the 15 N indirect dimension. Thus, amide corresponding cross-peaks which were not present in this experiment (Supplementary Materials, Figure S2) and which belonged to residues exhibiting ϕ, ψ values characteristic of secondary structure elements (α-helices, in the present case), as determined from TALOS analysis, were considered as involved in a regular H-bond, and the corresponding distance restraints were used for structure modeling. NOEs were measured on a 3D [ 1 H, 15 N] NOESY-HSQC experiment and on a 2D [ 1 H, 1 H] NOESY spectrum recorded in a deuterated buffer. Dihedral restraints (φ, ϕ, and χ1) were obtained from TALOS-N [17] analysis of backbone atom chemical shifts. After conversion into distance restraints, these data sets were used with CYANA [18] to build the 3D structure of GIPC1-GH2. H-bond restraints were also used for the structure modeling. Usually, these restraints are deduced from hydrogen/deuteron (H/D) exchange experiments, yielding amide protons potentially involved in H-bonds as donor atoms. In the case of GIPC1-GH2, H/D exchange rates were unusually fast, even at low temperature (5 °C): all amide proton resonances disappeared during the few minutes needed for the setting of the experiments, preventing the use of this approach. Instead, we used CLEANEX-PM [19,20] experiments to determine which amide protons are solvent-exposed in the 3D structure. In these experiments, the water resonance was selectively excited, and water magnetization transferred to solvent-exposed amide protons with an appropriate spinlocking module. This sequence was incorporated in a conventional HSQC scheme to resolve amide peaks along the 15 N indirect dimension. Thus, amide corresponding crosspeaks which were not present in this experiment (Supplementary Materials, Figure S2) and which belonged to residues exhibiting φ, ψ values characteristic of secondary structure elements (α-helices, in the present case), as determined from TALOS analysis, were considered as involved in a regular H-bond, and the corresponding distance restraints were used for structure modeling.
A final set of 1219 restraints was used, and the pairwise rmsd calculated for backbone heavy atoms between the 20 best refined structures was 0.47 Å (residues 5-76) ( Figure 2). Most of the residues (95.6%) fall in the most favored region of the Ramachandran plot, with no residue in the generously allowed or disallowed regions, highlighting the high A final set of 1219 restraints was used, and the pairwise rmsd calculated for backbone heavy atoms between the 20 best refined structures was 0.47 Å (residues 5-76) ( Figure 2). Most of the residues (95.6%) fall in the most favored region of the Ramachandran plot, with no residue in the generously allowed or disallowed regions, highlighting the high quality of our model (see structural statistics, Supplementary Materials, Table S1). The solution structure of GIPC1-GH2 consists of four antiparallel amphipathic helices arranged in an α-helical bundle. It is virtually identical to the X-ray structure adopted by this domain in the full-length protein, as shown by their superimposition displayed in Figure 2. An rmsd value of 0.85 Å was measured between them for all backbone heavy atoms (residues 5-73). This value drops to 0.68 Å when considering only backbone heavy atoms involved in helices. The bundle can be divided in two subdomains, each one consisting of two antiparallel helices stapled against each other: the N-(helix I) and C-terminal (helix IV) helices form the first subdomain, while the α-hairpin made by helix II and III forms the second subdomain. These two subdomains make an angle of approximately 50 • in the 3D structure of GIPC1-GH2. The structure coordinates of the NMR structure of GIPC1-GH2 have been deposited at the Protein Data Bank (PDB code 7NRN).
The intrinsic dynamics of GIPC1-GH2 were also investigated by measuring heteronuclear 15 N T 1 , T 2 relaxation times and [ 1 H, 15 N] heteronuclear nOes, and converting these parameters into J(0), J(ω N ), and <J((ω H )> spectral densities through Solomon equations [21] (see Section 4 and Supplementary Materials, Figure S3). Spectral densities were then fitted with model-free Lipari-Szabo equations [22] to extract the global (τ c ) and internal (τ e ) correlation times of the molecule, and generalized order parameters S 2 for each residue ( Figure 3). A τ c value of 6.09 ns was obtained from the fit of the relaxation parameters, in good agreement with the expected correlation time of this small protein at 20 • C. atoms involved in helices. The bundle can be divided in two subdomains, each one consisting of two antiparallel helices stapled against each other: the N-(helix I) and C-terminal (helix IV) helices form the first subdomain, while the α-hairpin made by helix II and III forms the second subdomain. These two subdomains make an angle of approximately 50° in the 3D structure of GIPC1-GH2. The structure coordinates of the NMR structure of GIPC1-GH2 have been deposited at the Protein Data Bank (PDB code 7NRN). The intrinsic dynamics of GIPC1-GH2 were also investigated by measuring heteronuclear 15 N T1, T2 relaxation times and [ 1 H, 15 N] heteronuclear nOes, and converting these parameters into J(0), J(ωN), and <J((ωH)> spectral densities through Solomon equations [21] (see Section 4 and Supplementary Materials, Figure S3). Spectral densities were then fitted with model-free Lipari-Szabo equations [22] to extract the global (τc) and internal (τe) correlation times of the molecule, and generalized order parameters S 2 for each residue (Figure 3). A τc value of 6.09 ns was obtained from the fit of the relaxation parameters, in good agreement with the expected correlation time of this small protein at 20 °C.  The regular Lipari-Szabo model was used for most of the residues involved in αhelices, yielding S 2 values close to 0.85, confirming that these secondary structure elements are well defined. The extended Lipari-Szabo model [23] was needed to fit the Nand C-terminal residues, suggesting the existence of more complex motions in these flexible regions, as usually observed. Interestingly, adding exchange contribution (Rex) to J(0) was mandatory to fit some residues located in the loops connecting the four helices, but also in helix IV and, to a lesser extent, helix II, suggesting that these two helices are prone to conformational exchange. The regular Lipari-Szabo model was used for most of the residues involved in αhelices, yielding S 2 values close to 0.85, confirming that these secondary structure elements are well defined. The extended Lipari-Szabo model [23] was needed to fit the N-and C-terminal residues, suggesting the existence of more complex motions in these flexible regions, as usually observed. Interestingly, adding exchange contribution (R ex ) to J(0) was mandatory to fit some residues located in the loops connecting the four helices, but also in helix IV and, to a lesser extent, helix II, suggesting that these two helices are prone to conformational exchange. 15 N] HSQC spectra of 15 N uniformly labeled GIPC1-GH2 were recorded at variable pressures (1 to 2500 bar) and at 10, 20, and 30 • C ( Figure 4). As usually found, the intensity of each native state peak decreases as a function of pressure, while the intensity of peaks corresponding to the unfolded state, centered around 8.5 ppm in the proton dimension, increases concomitantly. This supports a slow equilibrium on the NMR timescale for each residue between the native and unfolded state, and a two-state transition for each residue between their native/unfolded states during the unfolding process. Thus, even though the global protein unfolding does not likely conform to a two-state transition, locally this simple model can be used to interpret the loss of intensity for each native state cross-peak [6].  A total of 44 residues (58% of the nonproline residues) gave overlapping cross-peaks neither in the folded state nor in between the folded and unfolded states at any of the temperatures used for the study. These residues displayed cross-peaks of sufficient intensity at atmospheric pressure to be accurately fitted to the two-state pressure-induced unfolding model described in the Section 4 (Equation (4)), Figure 4, giving a substantial  15 N] HSQC at 1, 1000, and 2500 bar are displayed from left to right. The rightmost panels report overlays of four (residues K9, T32, L50, and V67) residue-specific experimental denaturation curves obtained from the fits of the pressure-dependent sigmoidal decrease of the corresponding residue cross-peak intensities in the HSQC spectra with Equation (4).
A total of 44 residues (58% of the nonproline residues) gave overlapping cross-peaks neither in the folded state nor in between the folded and unfolded states at any of the temperatures used for the study. These residues displayed cross-peaks of sufficient intensity at atmospheric pressure to be accurately fitted to the two-state pressure-induced unfolding model described in the Section 4 (Equation (4)), Figure 4, giving a substantial number of local probes for the description of the GIPC1-GH2 folding pathway. At the residue level, the two-state model was adequate to fit all individual unfolding curves, but yielded significantly different values for apparent free energy ∆G 0 u and apparent volume change ∆V 0 u ( Figure 5) of unfolding, suggesting a substantial deviation from a two-state behavior for the global unfolding of the protein, whatever the temperature of the study. The asymmetric distributions observed for apparent ∆G 0 u and apparent ∆V 0 u strongly support this assumption and suggest that partial unfolding of the molecule should appear when increasing the pressure (Supplementary Materials, Figure S4).
GIPC1-GH2 displayed a weak stability that appeared to be maximum at 20 • C, with an average value for the apparent free energy of unfolding <∆G 0 u > of 1293 ± 62 cal/mol, and significantly decreased at higher (<∆G 0 u > = 925 ± 57 cal/mol at 30 • C) or lower ((<∆G 0 u > = 903 ± 49 cal/mol at 10 • C) temperatures. Also, a linear decrease with temperature was observed for the average values (in magnitude mode) of apparent ∆V 0 u (<∆V 0 u > = −61 ± 4 mL/mol, −49 ± 7 mL/mol, and −38 ± 8 mL/mol at 10, 20, and 30 • C, respectively). The temperature-dependent decrease in ∆V 0 u is a well-known effect due to the difference in thermal expansion between the folded and unfolded states [24]. mL/mol, −49 ± 7 mL/mol, and −38 ± 8 mL/mol at 10, 20, and 30 °C, respectively). The temperature-dependent decrease in ∆ is a well-known effect due to the difference in thermal expansion between the folded and unfolded states [24]. Average values of ∆ calculated for each helix indicate differences in local stability (Supplementary Materials, Figure S5). Interestingly, this local stability depends on the temperature. At 10 °C and 20 °C, helix I and II appear slightly more stable than helix III and IV, helix II being the most stable at 20 °C whereas it has a similar stability as helix I at 10 °C. At 30 °C, the highest values of ∆ are found in helix I and III. Importantly, the Average values of ∆G 0 u calculated for each helix indicate differences in local stability (Supplementary Materials, Figure S5). Interestingly, this local stability depends on the temperature. At 10 • C and 20 • C, helix I and II appear slightly more stable than helix III and IV, helix II being the most stable at 20 • C whereas it has a similar stability as helix I at 10 • C. At 30 • C, the highest values of ∆G 0 u are found in helix I and III. Importantly, the local thermal stability measured for each helix follows the thermal stability of the whole domain, with a maximum observed at 20 • C. Likewise, a similar decrease with temperature is observed for average ∆V 0 u values calculated for each helix and for the average value calculated over the whole structure, but without any significant variations between the four helices.
Information brought by normalized residue-specific denaturation curves has been used to track and to characterize possible intermediates in the folding pathway of GIPC1-GH2 [6,8]. Thus, at a given pressure, the value of 1 measured for a given cross-peak (I = I f = 1; Equation (4)) is associated with a probability P i of 1 (100%) to find the corresponding residue "i" in the native state, while a residue "j" for which the corresponding cross-peak has disappeared (I = I U = 0; Equation (4)) from the HSQC spectrum has a probability P j equal to zero to be in a native state. Since these probabilities are related to the "native fraction" for a given residue, they are called fractional probabilities.
Given a pressure where these two residues i and j are in an intermediate situation (0 < P i and P j < 1), and if these two residues are in contact in the native state (at atmospheric pressure), their fractional probability P ij to be in contact at this pressure is given by the geometric mean of the two individual probabilities: P ij = P i × P j [12] (Figure 6).
At 20 • C, the temperature where GIPC1-GH2 exhibits the highest stability, the pressure dependence of the contact maps shows that helix III and IV are the first regions affected by an increase of pressure: tertiary contacts between these two helices are already significantly weakened at 500 bar (P ij ≤ 50%), as well as secondary contacts characteristic of the helical structure ( Figure 6). This partial unfolding concerns mainly these two helices up to 900 bar. Above this pressure, we observed a loss of contacts between helix IV and helix I, while helix I and II remain unaffected until 1100 bar. The unfolding of these two helices, as well as the loss of tertiary contacts between them, is observed at higher pressure (1300 bar, not shown). An identical scenario is observed at 10 • C, but with a shift to lower pressures, consistent with the lower stability of the protein at this temperature. At 700 bar, helix III and IV are unfolded (P ij ≤ 50%), and local unfolding already concerns helix I and II, while all the contacts are lost at 900 bar. While a similar stability of the protein is observed at 30 • C, a rather different scenario is observed for unfolding. The structure remains stable until 900 bar, with little loss of tertiary contacts observed. At 1100 bar, a sharp unfolding transition is observed, that concerns both the tertiary contacts between the four helices and the secondary contacts in all the helices, simultaneously. Finally, we observed a global unfolding of the molecule at 1300 bar.
has disappeared (I = IU = 0; Equation (4)) from the HSQC spectrum has a probability equal to zero to be in a native state. Since these probabilities are related to the "native fraction" for a given residue, they are called fractional probabilities.
Given a pressure where these two residues i and j are in an intermediate situation (0 < and < 1), and if these two residues are in contact in the native state (at atmospheric pressure), their fractional probability Pij to be in contact at this pressure is given by the geometric mean of the two individual probabilities: Pij = × [12] ( Figure 6).

Chemical Denaturation
2D [ 1 H, 15 N] HSQC spectra of 15 N uniformly labeled GIPC1-GH2 were recorded at 20 • C, the temperature where the protein exhibits the highest stability, and at increasing urea concentrations.
A total of 49 residues (64% of the nonproline residues) gave overlapping cross-peaks neither in the folded state nor in between the folded and unfolded states, and these residues can be accurately fitted to the two-state pressure-induced unfolding model described in the Section 4 (Equation (5), Figure 7). As observed for pressure denaturation, the intensity of each native state peak decreases as a function of urea concentration in the NMR sample, while the intensity of peaks corresponding to the unfolded state increases concomitantly, supporting a slow equilibrium on the NMR timescale for each residue between the native and unfolded state during the unfolding process. As reported above for pressure denaturation, a two-state transition model has been used to interpret the loss of intensity for each native state cross-peak (see Section 4). The residue-specific values obtained for the apparent free energy ∆G 0 u of unfolding and for the apparent m-values are displayed in Figure 8.    As previously observed, GIPC1-GH2 displays a weak stability at 20 °C, with an average free energy value for unfolding <∆ > of 1484 ± 45 cal/mol, a value close to that measured from pressure denaturation curves at the same temperature. But contrary to what we observed with pressure denaturation, we did not observe significant variations in between the different helices when looking at the average values of ∆ calculated for each helix (Supplementary Materials, Figure S6).
As in the case of pressure denaturation, we built fractional contact maps from probabilities of contact calculated from fractional probabilities of individual residues extracted from the normalized residue-specific chemical denaturation curves obtained at 20 °C (Figure 9). As previously observed, GIPC1-GH2 displays a weak stability at 20 • C, with an average free energy value for unfolding <∆G 0 u > of 1484 ± 45 cal/mol, a value close to that measured from pressure denaturation curves at the same temperature. But contrary to what we observed with pressure denaturation, we did not observe significant variations in between the different helices when looking at the average values of ∆G 0 u calculated for each helix (Supplementary Materials, Figure S6).
As in the case of pressure denaturation, we built fractional contact maps from probabilities of contact calculated from fractional probabilities of individual residues extracted from the normalized residue-specific chemical denaturation curves obtained at 20 • C (Figure 9).   Interestingly, if some contacts are lost between the C-terminal end of helix IV and the N-terminal end of helix III at low urea concentrations ([urea] ≤ 2.4-2.6 M), a sharp unfolding transition takes place between 2.6 and 2.8 M urea that concerns both secondary and tertiary contacts in the four helices. In this view, the folding scenario looks like what was observed for pressure denaturation at 30 • C, where an increased unfolding cooperativity was observed when compared to 10 • C or 20 • C.

Discussion
The structure in solution of the GH2 domain of GIPC1 shows that this domain keeps its α-helical bundle fold outside the full-length protein context, even though it exhibits a rather low stability, probably due to the loss of intra-and intermolecular interactions that occur within the dimeric structure of the full-length protein. Indeed, this low stability is supported by the fast exchange rates exhibited by the amide protons that cannot be measured by regular H/D exchange NMR experiments. Nevertheless, CLEANEX-PM experiments showed that the amides involved in the H-bonds stabilizing the helical structures are not solvent-exposed, contrary to those located in the loops linking the different helices and in the flexible N-and C-terminal ends of the domain, supporting the idea that they are involved in the regular H-bonds expected in those regular elements of secondary structure. In addition, S 2 values close to 1, as obtained from the 15 N relaxation study, indicate that the helices are well structured, even though significant exchange contributions (R ex 2 ) can be observed, especially in helix IV. This suggests that this helix is prone to conformational exchange. Notably, GIPC1-GH2 was found to be stable only in a limited range of temperature. The stability was found to be maximum at 20 • C, whereas it decreases significantly when increasing or decreasing the temperature. As evaluated from the ratio of the intensity of native/unfolded amide cross-peaks of representative residues measured on HSQC spectra, the fraction of native protein is about 70% at 40 • C and 60% at 0 • C (Supplementary Materials, Figure S7), indicating that the protein is sensitive both to thermal and to cold denaturation. GIPC1-GH2 is also very sensitive to high hydrostatic pressure, since it unfolds at 20 • C in the 1-2500 bar range without adding any sub-denaturant concentration of chaotropic reagents, as was usually observed for high-pressure denaturation of all-β or mixed-α/β structures in our previous study: sub-denaturant concentration ranging from 0.5 M [25] to about 2 M [8,10] of guanidinium chloride was used to tune the stability of these proteins into the pressure range allowed by the experimental set-up.
Looking closely to the unfolding pathways of GIPC1-GH2 under high hydrostatic pressure, we observed unexpected results. Indeed, since high-pressure denaturation is closely related to the presence of dehydrated internal solvent-excluded voids in the structure, we expected a two-step process starting with the loss of tertiary contacts within the 3D structure, followed by the loss of the secondary contacts yielding helices unfolding. This is because helices are well-packed secondary structure elements, without significant voids inside, while packing defaults are expected within the tertiary structure of the αhelical bundle. In fact, whatever the temperature of the study, local or global unfolding entails the quasi-simultaneous loss of the tertiary and secondary contacts in the concerned areas, meaning that helices are not stable outside the 3D scaffold context. Depending on the temperature used for the high-pressure denaturation study, we observed significant differences in the unfolding process. At 20 • C, the temperature where GIPC1-GH2 exhibits the highest stability, a partial unfolding of the molecule occurs first at helix III and helix IV, while helix I and helix II remain intact until approximately 1300 bar (note that for clarity of the discussion, we consider that a contact between two residues i and j is lost when P ij ≤ 50%). The same scenario is observed at 10 • C, shifted at lower pressure: unfolding of helix III and IV starts below 500 bar instead of 900 bar at 20 • C, while helix I and II start to unfold at 700 bar. A different scenario is observed at 30 • C, where the four helices unfold almost simultaneously, with the onset of unfolding around 1000 bar. Interestingly, globally the protein appears to be more stable at high temperature (30 • C) than at low temperature: at 10 • C, the protein appears completely unfolded (P ij ≤ 50%) at 900 bar while some residual structures are still present at 1100 bar and 30 • C.
This scenario described for high pressure denaturation at high temperature is very similar to what is observed for chemical denaturation of GIPC1-GH2 at 20 • C. We did not observe significant partial unfolding of any helix upon increase in urea concentration but rather a global unfolding of the molecule between 2.6 and 2.8 M urea. This is probably due to the different rules underlying high-pressure and chemical unfolding. As reported above, pressure unfolding is linked to the presence of solvent-excluded voids inside the 3D structure of the protein, and hence depends on the structure of the native state of the protein. On the contrary, the chemical denaturation process is driven by the increase of solvent-accessible area of the unfolded state with respect to the folded state, and is more dependent on the size of the protein [26][27][28]. This probably explains the difference that we observed in the folding pathway between high-pressure denaturation of GIPC1-GH2 at 20 • C or 10 • C, and its chemical denaturation at 20 • C. Note that thermal denaturation is also related to the solvent-accessible area of the unfolded state: this could explain the similar scenario observed for high-pressure denaturation at high temperature (30 • C) and for chemical denaturation. Indeed, at 30 • C, the stability of GIPC1-GH2 is decreased, and thermal denaturation probably competes with high-pressure denaturation, sweeping away the partial unfolding occurring at lower temperature.

Protein Expression and Purification
The construct GIPC1-GH2 domain (residues 255-333) was subcloned in pProEXHTB, allowing the expression of a 6xHis-TEV fusion protein, and was transformed into E. coli BL21-Gold (DE3) (Stratagene, Amsterdam, The Netherlands). Uniform 15 N or 15 N/ 13 C labeling was obtained by growing cells in minimal M9 medium containing 15 NH4Cl and/or 15 NH4Cl/ 13 C-u-labeled glucose as the sole nitrogen or carbon sources (Cortecnet). Protein was expressed overnight at 20 • C after induction with 0.2 mM IPTG. Cells were collected by centrifugation and suspended in lysis buffer comprising 20 mM Tris-HCl buffered at pH 7.5 and containing 150 mM NaCl, 2 mM imidazole, and a cOmplete™ EDTA-free tablet (Roche). Cells were lysed by sonication (1 s bursts for 4 min, at 30% amplitude with a large probe, Branson). Cell debris and insoluble materials were removed by centrifugation (Beckman Coulter Avanti J-20 XP centrifuge equipped with a 25.50 rotor, set at 20,000 rpm, at 6 • C). The supernatant was loaded through a benchtop peristaltic pump (Cytiva) onto a cOmplete™ His-Tag Purification Column (Roche, Basel, Switzerland) equilibrated with lysis buffer. After elution with lysis buffer supplemented with 200 mM imidazole, fractions containing the protein were dialyzed with homemade recombinant His tagged rTEV protease (mixed at 100:1 ratio) overnight at 4 • C in 20 mM Tris-HCl buffered at pH 7.5, 150 mM NaCl, and 1 mM imidazole. Cleavage was checked with SDS-PAGE and loaded again into a cOmplete™ His-Tag Purification Column equilibrated with the same buffer used for dialysis in order to remove the protease and the cleaved 6xHis tag. The GipC1-GH2 domain was finally injected through an AKTA system into a Superdex S75 16/60 (GE Healthcare) column, equilibrated with 20 mM Tris-HCl buffered at pH 7.2, 150 mM NaCl. The fractions containing the pure protein were pooled, concentrated to about 1 mM (protein concentration) (Vivaspin 15R, Sartorius). PMSF and EDTA were added (0.1 mM) to the samples that were then flash-frozen in liquid N2 and stored at −80 • C until NMR analysis.

NMR Assignments and Solution Structure
Protein samples were dissolved in 200 µL of aqueous buffer containing 20 mM Tris-HCl pH 7.2, 150 mM NaCl, and 0.1 mM PMSF and EDTA (5% D 2 O for the lock) at a concentration of about 1 mM. Experiments were recorded at 20 • C on a Bruker AVANCE III 800 MHz (Bruker Biospin, Wissenbourg, France) equipped with a 5 mm Z-gradient TCI cryogenic probe head. 1 H chemical shifts were directly referenced to the methyl resonance of DSS, while 13 C and 15 N chemical shifts were referenced indirectly to the 13 C/ 1 H and 15 N/ 1 H absolute frequency ratios. All NMR experiments were processed with Gifa [29].
The final list of restraints, from which values that were redundant with the covalent geometry were eliminated, was used for structure modeling. A total of 200 three-dimensional structures were generated using the torsion angle dynamics protocol of CYANA 3 from 1219 NOEs, 88 hydrogen bonds, and 164 angular restraints. The 20 best structures (based on the final target penalty function values) were minimized with CNS 1.2 according to the RECOORD procedure [31] and analyzed with PROCHECK [32]. The rmsds were calculated with MOLMOL [33]. Models are displayed with PyMOL [34]. All statistics are given in Supplementary Materials, Table S1.

Relaxation Studies
Relaxation rate constant measurements were performed on a 1 mM 15 N-labeled protein sample, at 14.1 T (600 MHz), using a Bruker AVANCE III spectrometer equipped with a 5 mm Z-gradient TXI probe head. The pulse sequences used to determine heteronuclear 15 N R 1 , R 2 relaxation rates, and 15 N{ 1 H}NOE values were similar to those described [35][36][37], and experimental parameters and processing were previously reported in detail for other proteins studied in the laboratory [38][39][40]. The 15 N longitudinal relaxation rates R 1 were obtained from nine standard inversion-recovery experiments, with relaxation delays ranging from 18 ms to 1206 ms. The 15 N transverse relaxation rates R 2 were obtained from eight standard CPMG experiments, with relaxation delays ranging from 16 ms to 128 ms. Heteronuclear 15 N{ 1 H} NOE were determined from the ratio of two experiments, with and without saturation.
The model-free approach of Lipari and Szabo [22] was then used to further describe the mobility in terms of specific types of motion. This formalism makes the assumption that overall and internal motions contribute independently to the reorientation time correlation function of 15 N-1 H vectors and that internal motions occur on a much faster time scale than the global rotation of the molecule. For a protein with isotropic tumbling protein, one obtains: where τ is the harmonics of the overall and the internal (fast) correlation time which pertains to each residue: τ −1 = τ −1 c + τ −1 f . Fast internal motions are characterized by the square of a generalized order parameter S 2 , which describes the relative amplitude of internal motions and ranges from 0 to 1, and an internal correlation time τ f for the internal motions.
For some of the residues, the simple form of equation (1) turns out to be insufficient to fit the whole set of experimental data. This occurs for residues where observed J(0) values are higher than expected, due to exchange contributions R ex 2 . In this case, the expression for the observed spectral density at 0 frequency is: where λ is a scale factor. This occurs also when residues exhibit internal motions in a time window close to 1 ns. In this case, the expression for the spectral density function is extended to [23]: where S 2 f and S 2 s are the square of the partial order parameters for fast (picosecond time scale) and slow (τ s , nanosecond time scale) internal motions, respectively. The square of the generalized order parameter S 2 , defined as S 2 f S 2 s , is a measure of the total amplitude of the internal motions. Note that S 2 equals S 2 f in Equations (1) and (2). Equation (3) assumes that the contribution of the fastest motion to the spectral density function is negligible.
The values of the motional parameters of the individual residues can be derived from the fit of experimental J(0), J(60 MHz), and <J(600 MHz)> using Equations (1)- (3) implemented in the software DYNAMOF [45]. An iterative nonlinear least-squares algorithm [46] was employed to further minimize the error function. The "right" model was selected from χ 2 analysis.

Protein Unfolding
2D [ 1 H, 15 N] HSQC were recorded on a Bruker AVANCE III 600 MHz spectrometer, at 3 temperatures (10, 20, and 30 • C) and 15 different hydrostatic pressures (1,50, 100, 300, 500, 700, 900, 1100, 1300, 1500, 1700, 1900, 2100, 2300, and 2500 bar) for pressure denaturation, and at 20 • C and 13 different urea concentrations (0, 0.1, 0.2, 0.375, 0.75, 1.125, 1.5, 1.875, 2.25, 3, 3.75, 5, and 6 M) for chemical denaturation. Samples with about 1 mM concentration of 15 N-labeled proteins were used on conventional 3 mm NMR tubes (200 µL of sample volume) for chemical denaturation, or in 5 mm o.d. ceramic tubes (330 µL of sample volume) from Daedelus Innovations (Aston, PA, USA) for pressure denaturation. Hydrostatic pressure was applied to the sample directly within the magnet using the Xtreme Syringe Pump also from Daedelus Innovations. Samples with different urea concentrations were prepared about 10 h before recording the NMR experiments used for chemical denaturation studies, although each pressure jump was separated by a 2-h relaxation time, to allow the protein to reach full equilibrium. In the case of pressure denaturation, relaxation times for the folding/unfolding reactions were estimated from a series of 1D NMR experiments recorded after 200 bar P-Jump, following the increase of the resonance band corresponding to the methyl groups in the unfolded state of the protein.
The cross-peak intensities for the folded species were measured at each pressure or each urea concentration, then fitted with a two-state model: in the case of chemical denaturation. In these equations, I is the cross-peak intensity measured at a given pressure or at a given urea concentration, and I f and I u correspond to the cross-peak intensities in the folded state (1 bar or 0 M urea) and in the unfolded state (2500 bar or 6 M urea), respectively. ∆G 0 f stands for the residue-specific apparent free energy at atmospheric pressure or at 0 M urea. ∆V 0 f corresponds to the residue-specific apparent volume of folding for pressure denaturation, while m is related to the steepness of the unfolding transition for chemical denaturation.
Native contact maps were obtained by using software CMView (http://www.bioinfor matics.org/cmview/; accessed on 25 March 2020) with a threshold of 9 Å around the Cα of each residue, using the best structure obtained for GIPC1-GH2 among the 20 refined ones.

Conclusions
We demonstrate that combining NMR spectroscopy, which can bring information at an atomic resolution, with a mild and reversible method, such as high hydrostatic pressure, for protein unfolding can bring unprecedented details on the folding landscape of a protein.
Here, we applied high-pressure NMR spectroscopy to the study of the folding/unfolding pathways of an α-helical bundle. Indeed, whereas similar studies have been widely applied to all-β or mixed-α/β 3D scaffolds, they are lacking for all-α helical structures. Unexpectedly, we found that the secondary and tertiary structures unfold simultaneously, although partial unfolding can occur. Importantly, this partial unfolding cannot be revealed with chemical denaturation, confirming the superiority of high pressure for exploring the folding landscape of a protein. Of course, whether these results obtained for GIPC1-GH2 can be generalized to other comparable α-helical bundles or whether more will have to be done remains an open question.
Supplementary Materials: The following are available online at https://www.mdpi.com/artic le/10.3390/ijms22073597/s1: Table S1: NMR restraints and refinement statistics for GIPC1-GH2 solution structures, Figure S1: Crystal structure of the full-length GIPC1 protein, Figure S2: Phase modulated CLEANEX-PM experiment recorded on 15N-labeled GIPC1-GH2, Figure S3: Relaxation measurements performed on 15N-labeled GIPC1-GH2., Figure S4: Distribution of the values of the thermodynamic parameters ∆G 0 and ∆V 0 , Figure S5: Local pressure stability of GIPC1-GH2, Figure S6: Local chemical stability of GIPC1-GH2. Figure S7: Estimation of the folded/unfolded fraction of GIPC1-GH2. He also wrote analysis software for the analysis of the relaxation data and HP results. He finally participated in the coordination of the project with V.R., C.R. supervised the project, conceived experiments, participated in the interpretation, and wrote the article with P.B. All authors have read and agreed to the published version of the manuscript.