Thermodynamic stability of hnRNP A1 low complexity domain revealed by high-pressure NMR

Abstract

We have investigated the pressure- and temperature-induced conformational changes associated with the low complexity domain of hnRNP A1, an RNA-binding protein able to phase separate in response to cellular stress. Solution NMR spectra of the hnRNP A1 low-complexity domain fused with protein-G B1 domain were collected from 1 to 2,500 bar and from 268 K to 290 K. While the GB1 domain shows the typical pressure-induced and cold temperature-induced unfolding expected for small globular domains, the low-complexity domain of hnRNP A1 exhibits unusual pressure and temperature dependences. We observed that the low-complexity domain is pressure sensitive, undergoing a major conformational transition within the prescribed pressure range. Remarkably, this transition has the inverse temperature dependence of a typical folding-unfolding transition. Our results suggest the presence of a low-lying extended, and fully solvated state(s) of the low-complexity domain that may play a role in phase separation. This study highlights the exquisite sensitivity of solution NMR spectroscopy to observe subtle conformational changes and illustrates how pressure perturbation can be used to determine the properties of metastable conformational ensembles.

Introduction

High-pressure NMR spectroscopy has emerged as a powerful technique to characterize the stability of globular proteins,^1–4 their mechanisms of folding,^5–7 and presence of low-lying intermediate states.^8–10 According to Le Chatelier’s principle an increase in pressure will shift the thermodynamic equilibrium toward states of lower molar volume. Since the application of pressure leads (over most of the accessible temperature range) to protein unfolding, it signifies that the volume change upon unfolding is negative (ΔV_F→U <0), i.e. the molar volume of the unfolded state is smaller than that of the folded state.¹¹ The magnitude of ΔV_F→U values measured for globular proteins typically lies around 50 to 100 ml/mol, which represent only 0.5% to 2% of the protein’s molar volume.¹² The physical origin of ΔV_F→U has been the subject of much debate but strong experimental and computational evidences point toward an imperfect balance between negative contributions (i.e. elimination of cavities and void volume^13–14) and positive contributions (i.e. volume changes due to exposure of hydrophobic side chain upon unfolding¹⁵). The mechanisms of pressure unfolding have also been thoroughly investigated by molecular simulations and experimental approaches. It is generally believed that under the influence of high pressure, water molecules penetrate into internal cavities of the protein core and destabilize hydrophobic interactions.^16–18 The volume difference upon unfolding is known to be strongly temperature dependent. Indeed, because the thermal expansivity of the unfolded states is larger than that of the folded state, ΔV_F→U is more negative at lower temperatures and it’s magnitude decreases as temperature increases. ΔV_F→U can even become positive at high enough temperatures.^19–21

Only a handful of studies have examined the effects of pressure on protein unfolded states and intrinsically disordered proteins and peptides.^22–24 In the case of α-synuclein, no major effect was observed besides non-specific pressure-induced chemical shift changes and a small decrease in ³J_HNHα couplings.²³ Separately, it has been reported that folding of small helical motifs can be promoted under high-pressure conditions due to preferential hydration of helical structure.^25–27 These evidences suggest that unfolded protein chains may not remain entirely featureless at high-pressure.

Low-complexity (LC) domains found in RNA-binding proteins associated with liquid-liquid phase separation form a specific class of intrinsically disordered domains. LC domains have distinct amino acid compositions; they are enriched in polar amino acids (especially Ser and Gly), and feature conserved patterns of aromatic residues.²⁸ In many cases, these disordered LC domains are necessary and sufficient for driving phase separation.^29,30 Liquid condensates are believed to be the result of multivalent weak interactions between multiple interacting motifs in LC domains, including electrostatic, cation-π and π-π interactions.^31–35 Yet, the relative contribution of these interactions to phase separation is largely unknown. The effects of pressure on phase separated systems have recently been investigated revealing a strong pressure dependence that can lead to complete dissolution of liquid condensates over a few hundred bars.^36–38

In the present study we investigated the effects of pressure on the LC domain of the isoform A of heterogeneous nuclear ribonucleoprotein A1 (hnRNP A1-A). hnRNP A1-A is a 34 kDa protein composed of 320 amino acids and three separate protein domains, two RNA recognition motifs (RRMs) and a disordered C-terminal domain (LCD_A1) rich in glycine and arginine (Fig. 1A). The two RRMs form the nucleic acid binding domain, referred to as Unwinding Protein-1 (UP1), and work in tandem to bind RNA and DNA substrates.³⁹ The C-terminal domain engages in protein interactions that are necessary for the protein to carry out its nucleic acid processing activities.⁴⁰ Many of these processing activities are regulated by post-translational modification of LCD_A1, such as methylation, phosphorylation, and SUMOylation.⁴¹ LCD_A1 also contains a 38 amino acid nuclear localization signal (NLS) responsible for shuttling the protein between the nucleus and the cytoplasm, with the shuttling activity being mediated between interactions with Transportin 1 and Transportin 2.^42,43 The NLS mediates protein interactions, primarily through aromatic residues present in this region.⁴⁴ Given the multitude of protein-protein and protein-RNA interactions the LCD_A1 forms, it is imperative to examine the biophysical properties of the domain as well as the nature of the chemical interactions it is capable of engaging in with other proteins and nucleic acids.

Figure 1.

Schematic representation of the sequence of (A) hnRNP A1-A with the N-terminal Unwinding Protein-1 domain (UP1) encompassing two RNA Recognition Motifs (RRM1 and RRM2) and the C-terminal Low-Complexity Domain (LCD_A1) encompassing the glycine-rich region (Gly-rich), prion-like domain (prLD), M9 nuclear localization signal (M9 NLS), and phosphopeptide (F Peptide). (B) GB1-LCD_A1 construct composed of GB1 domain fused with hnRNP A1-A Low-Complexity Domain. This construct encompasses an N-terminal His-tag and a TEV recognition site between the GB1 domain and the LCD_A1 domain. (C) Isolated GB1 domain obtained after TEV cleavage of GB1-LCD_A1.

hnRNP A1 has shown the ability to form liquid condensates in response to cellular stress and other stimuli.²⁹ These stress granules are formed from pools of stalled, untranslated mRNAs, with assembly often occurring through interactions of the LC domains of various RNA binding proteins.⁴⁵ Although RNA binding to the UP1 domain is known to promote granule assembly, the isolated LC domain has the ability to self-assemble into granules.^29,46,47 The arginines present in the RGG box within the Gly-rich region were shown to be necessary for stress granule assembly in hnRNP A1.⁴⁸ The details of this mechanism has been investigated for hnRNP A2, where it was found that arginine methylation of the RGG boxes reduces phase separation by disrupting interactions between the charged arginines and the aromatic residues in the region preceding the prion-like domain.⁴⁹

The LCD_A1 domain used in the present study encompasses amino acids 197–320 of hnRNP A1-A (Fig. 1A). We examined the behavior of the LCD_A1 under pressure using solution NMR spectroscopy in solution phase (no liquid condensate) with the aim of characterizing low-lying conformational states that may play a role in triggering phase separation. For this purpose, we used a chimeric construct consisting of protein G B1 domain (GB1) fused at its C-termini to LCD_A1 (Fig. 1B). The GB1 domain aided protein purification and sample stability. Its presence, along with the sample buffer conditions, prevents phase separation even at the high sample concentration required for NMR experiments and low temperature range sampled here (268 K to 290 K). Therefore, fusion of LCD_A1 with a soluble domain such as GB1 allows the characterization of the conformational ensemble of LCD_A1 in fully homogeneous solution phase without formation of liquid condensates. This construct also offers the advantage of presenting within the same chain two structurally different domains: a well folded globular domain (GB1) and a low complexity disordered domain (LCD_A1). By designing this construct, we intended to use GB1 domain as an internal control to help interpret the effects of pressure measured on LCD_A1.

Materials and Methods

Sample preparation –

Constructs for the isolated GB1 domain and GB1-LCD_A1 fusion protein were cloned from gBlock gene fragments (IDT) into pMCSG plasmid.⁵⁰ The following sequence was used for the GB1-LCD_A1 construct: MHHHHHHSSGVDLQYKLALNGKTLKGETTTEAVDAATAEKVFKQYANDNGVDGEWTYDDATKTFTVTEGTENLYFQSNIMRSGSGNFGGGRGGGFGGNDNFGRGGNFSGRGGFGGSRGGGGYGGSGDGYNGFGNDGSNFGGGGSYNDFGNYNNQSSNFGPMKGGNFGGRSSGPYGGGGQYFAKPRNQGGYGGSSSSSSYGSGRRF (the TEV recognition site separating the GB1 sequence to the LCD_A1 sequence is underlined). Overexpression was carried out in BL21 (DE3) cells (NEB) in minimal media supplanted with ¹⁵NH₄Cl. Cells were induced at OD₆₀₀ ~1.0 with 0.2 mM IPTG. Cells were harvested after overnight expression. For GB1-LCD_A1, cell pellet was resuspended in lysis buffer (20 mM Na₂HPO₄ pH 7=5, 1 M NaCl, 20mM imidazole, 0.25 mM EDTA, 1 mM PMSF, 10% Glycerol, protease inhibitor cocktail tablet). Cells were lysed by sonication, cell debris was spun down, and lysate was applied to a HiTrap Chelating HP column (GE Healthcare) charged with NiSO₄. Protein was washed with lysis buffer, then eluted with elution buffer (20 mM Na₂HPO₄ pH=7.5, 1 M NaCl, 250mM imidazole, 0.25 mM EDTA, 1 mM PMSF, 10% Glycerol). Eluted protein was further purified by FPLC gel filtration with a Sephacryl S-100 column in gel filtration buffer (100 mM HEPES pH=7.5, 350 mM NaCl, 0.5 mM EDTA, 10% glycerol). Fractions containing protein were pooled and washed into NMR Buffer through Amicon centrifugal filtration. Purification of GB1 from GB1-peptide was conducted the same as GB1-LCD_A1, except before the FPLC gel filtration, the peptide was cleaved off with TEV protease and separated from GB1 by another run through the HiTrap Chelating HP column.

NMR –

¹H-¹⁵N TROSY-HSQC spectra were recorded with uniformly ¹⁵N labeled samples on a Bruker 700 MHz Avance II spectrometer, equipped with a z-shielded gradient triple resonance cryoprobe. A total of 100 × 1048 complex points were collected, for acquisition times of 104 and 121 ms in the ¹⁵N and ¹H dimensions, respectively, using an interscan delay of 1.5 s. ¹H-¹³C HSQC spectra were recorded with triple labeled ²H-¹⁵N-¹³C samples for a total of 128 × 512 complex points on a Bruker 800 MHz spectrometers equipped with a z-shielded gradient triple resonance cryoprobe. resonance cryoprobe. All the experiments were recorded using 150 μM protein in buffer 100 mM HEPES pH=7.5, 350 mM NaCl, 0.1 mM leupeptin, 0.5 mM EDTA, 10% ²H₂O, 10% glycerol. A commercial ceramic high-pressure NMR cell and an automatic pump system (Daedalus Innovations, Philadelphia, PA) were used to vary the pressure in the 1 bar to 2.5 kbar range. The spectra were processed using NMRPipe⁵¹ and displayed with SPARKY.⁵²

SAXS-

GB1-LCD_A1 for SAXS was prepared the same as NMR, but overexpressed in TB broth instead of minimal media. SEC-SAXS experiments were performed on a 650 μM GB1-LCD_A1 sample at BioCAT (beamline 18-ID, Advanced Photon Source). Data was collected as previously described.⁵³ Molecular weight calculations and data analysis were performed in PRIMUS⁵⁴ from the ATSAS suite of programs.⁵⁵

Results

Pressure-induced unfolding of isolated GB1 domain.

We first studied the stability of the isolated GB1 domain under pressure at both 290 K and 277 K (Fig. 1C). The intensity of amide crosspeaks was monitored from series of ¹⁵N-¹H heteronuclear single-quantum coherence (HSQC) spectra collected every 500 bar (Fig. S1 A and B). The intensity profiles measured for each residue followed a sigmoidal curve, which makes reasonable the assumption of a two-state equilibrium between folded (F) and unfolded (U) states (Fig. S1 C and D). Assuming a negligible difference in compressibility between F and U, one can express the change of the equilibrium constant K_eq as a function of pressure as:

$${\text{K}}_{\text{eq}}=[\text{U}]/[\text{F}]=\exp \left(-\mathrm{\Delta }{\text{G}}_{\text{F}\to \text{U}}/\text{RT}\right)\text{where}\mathrm{\Delta }{\text{G}}_{\text{F}\to \text{U}}=\mathrm{\Delta }{\text{G}}^0{\text{F}}_{\to \text{U}}+\mathrm{\Delta }{\text{V}}_{\text{F}\to \text{U}}(\text{p}-{\text{p}}_0)$$ (1)

Here ΔG_F→U and ΔG⁰_F→U are the Gibbs-free energy changes from F to U at pressures p and p₀ (= 0.1 MPa), respectively; ΔV_F→U is the molar volume change between F and U; R is the gas constant, and T is the absolute temperature.

Using NMR spectroscopy observables such as amide crosspeak intensities, the equilibrium constant can be written as:

$$K_{eq}=\frac{[U]}{[F]}=\frac{I_F-I}{I-I_{\mathit{II}}}$$ (2)

where, I_F represents the maximal intensity of the folded crosspeaks, I_U the intensity of the same crosspeaks when the protein is fully unfolded (I_U usually converges to 0) and I the intensity of the crosspeak at a given pressure-temperature condition. Combining equation (1) with equation (2) gives:

$$I=\frac{I_F+I_U{e}^{-\left[\mathrm{\Delta }{G}_{F\to U}^0+\mathrm{\Delta }{V}_{F\to U}\left(p-p_0\right)\right]/RT}}{1+e^{-\left[\mathrm{\Delta }{G}_{F\to U^{+}}^0+\mathrm{\Delta }{V}_{F\to U}\left(p-p_0\right)\right]/RT}}$$ (3)

We noticed that the peak widths of the folded-state amide resonances did not broaden as a function of pressure, which means that native crosspeak intensity can be reliably used as a proxy to monitor the relative folded population at a residue level. Nevertheless, it should be acknowledged that these residue-level ΔV_F→U values must be considered as apparent and not as pure thermodynamic parameters.⁵⁶

The average ΔV_F→U measured for the isolated GB1 domain at 290 K is −34.1 ± 1.2 ml/mol (Table 1 and Fig. S1 E), which is similar to volume differences measured for other small globular protein domains such as ubiquitin, CI2, and CspB.¹¹ We found that the average ΔV_F→U values were slightly more negative at 277 K compared to 290 K (−37.4 ± 1.6 vs. −34.1 ± 1.2 ml/mol) (Table 1). This indicates, as expected for globular proteins, that the unfolded states of GB1 has a larger thermal expansivity than that of its folded states.

Table 1.

Thermodynamic parameters measured for the isolated GB1 domain, GB1 domain in GB1-LCD_A1 construct (GB1-LCD_A1), and LCD_A1 domain in the GB1-LCD_A1 construct (GB1-LCD_A1). Measured ΔV values correspond to the volume change upon unfolding in the case of GB1 (ΔV_F→U) and to the volume change associated with LP-to-HP transition in the case of LCD_A1 (ΔV_LP→HP). Values in the table represents the average and standard deviation calculated over all residue-specific ΔV values.

	ΔV (ml/mol)
GB1 (290 K)	−34.1 ± 1.2
GB1 (227 K)	−37.4 ± 1.6
GB1−LCDA1 (290 K)	−32.9 ± 1.1
GB1−LCDA1 (277 K)	−37.5 ± 1.7
GB1-LCDA1 (290 K)	−43.6 ± 2.9
GB1-LCDA1 (277 K)	−32.2 ± 1.2

GB1-LCD_A1 is monomeric in solution and does not phase separate at low temperature or high pressure.

We then determined the oligomeric state of the chimeric protein GB1-LCD_A1 in conditions similar to that of the NMR experiments. Figure S2 shows the profile of GB1-LCD_A1 eluting from a Superdex 200 10/300 GL gel filtration column with an injected sample concentration of 150 μM. The protein sample and column were kept at 277 K. A single peak is observed with an elution profile similar to isolated GB1, as the presence of LCD_A1 does not alter its SEC profile. No peak corresponding to higher molecular weight oligomers was observed confirming that GB1-LCD_A1 is predominantly monomeric in the NMR experiments conditions. Further evidence for the monomeric state of the construct was found by small-angle x-ray scattering (SAXS) experiments run at 298K and 650 μM. The molecular weight of GB1-LCD_A1 was calculated using a five different concentration-independent determination Bayesian method.⁵⁷ The results ranged from 16–23 kDa, in-line with the expected weight of 21 kDa for the monomeric protein (Table 2). Visual analysis of the NMR spectra also indicates that GB1-LCD_A1 does not form liquid condensate over the range of temperature and pressure sampled here. Liquid-liquid phase separation leads to drastic broadening of peak linewidths which was observed in none of the 2D NMR spectra collected from 268 K to 290 K and 1 bar to 2.5 kbar.

Table 2.

Concentration-independent molecular weight determination for GB1-LCD_A1 calculated using the PRIMUS program.

Method	Results
MMQp	qmax[A−1]=0.26626 MW=15294 Da
MoW	qmax[A−1]=0.30025 V[A3]=22104 MW=18237 Da
Volume of Correlation	qmax[A−1]=0.30025 VC=234 MW=16918 Da
Size & Shape	MW=22974 Da
Bayesian Inference	MW=18050 Da MW Probability=43.6% Credibility Interval=15800–19000 Da Interval Probability=96.83%

GB1-LCD_A1 ensemble is compact and disordered in solution.

Kratky plots are commonly used for determining the degree of structural disorder of protein ensembles in solution.⁵⁸ The Kratky plot generated for GB1-LCD_A1 shows an initial parabolic peak is followed by a plateaued baseline (Fig. 2A) as expected for a multi-domain protein combining the features of folded and disordered domains.^58,59 Guinier analysis of the SAXS data revealed a radius of gyration, R_g of 26.2 ± 0.1 Å (Fig. 2B), which shows that the chimeric protein is highly compact in solution (i.e. the theoretical R_g of GB1-LCD_A1 in a fully extended state is ~46.5 Å).^60,61

Figure 2.

Small-angle X-ray scattering (SAXS) profile of GB1-LCD_A1 collected at a protein sample concentration of 650 μM. (A) Analysis of the Kratky plot shows that GB1-LCD_A1 is a multi-domain protein composed of both globular and disordered domains. (B) Pair-wise distance distribution function P(r). A radius of gyration estimated of 26.2 ± 0.1 Å was calculated for GB1-LCD_A1 in solution.

Pressure-induced unfolding of GB1 in GB1-LCD_A1.

To probe the effect of pressure on the GB1-LCD_A1, we first examined the stability under pressure of the GB1 domain in the chimeric construct and compared it with the isolated GB1 domain described above. Comparison of the spectra recorded for isolated GB1 domain and GB1-LCD_A1, show an almost perfect overlap of GB1 domain crosspeaks in both constructs (Fig. S3). In addition, ΔV_F→U values measured for GB1 crosspeaks in the GB1-LCD_A1 construct were remarkably similar to those measured for the isolated GB1 domain at both 290 K and 277 K (Fig. 3 and Table 1). These results demonstrate that GB1 structure and folding/unfolding thermodynamics remain unchanged when fused to LCD_A1 domain.

Figure 3.

Pressure-induced unfolding of GB1 domain in GB1-LCD_A1 construct. ¹H-¹⁵N HSQC spectra of GB1-LCD_A1 were collected at 290 K (A) and 277 K (B) at pressure varying from 1 bar to 2,500 bar. Intensity profiles of individual GB1 crosspeaks in GB1-LCD_A1 were measured as a function of pressure at 290 K (C) and 277 K (D) and fitted to equation (3) (solid line) to obtain residue-specific ΔV_F→U values. Comparison of ΔV_F→U values as a function of GB1 sequence at 290 K (E) and 277 K (F). ΔV_F→U values measured for GB1 in GB1-LCD_A1 construct are compared here with those measured for the isolated GB1 domain (see Fig. S1).

Pressure-induced transition of LCD_A1 in GB1-LCD_A1.

While the GB1-LCD_A1 construct presents the advantage of preventing phase separation at low temperature and high protein concentration, which is essential for this study, it seemingly causes significant exchange-induced linewidth broadening of a large number of LCD_A1 crosspeaks. As apparent from the 2D ¹H-¹⁵N spectrum of GB1-LCD_A1 collected in standard conditions (i.e. 290 K, 1 bar), less than twenty crosspeaks can be confidently assigned to the LCD_A1 domain (Fig. S3). Similarly, comparison of ¹H-¹³C HSQCs spectra of GB1 and GB1-LCD_A1 shows that only about twelve crosspeaks can be attributed to the LCD_A1 (Fig. S4). These experiments suggest that conformational exchange over unfavorable time scale rather than extensive solvent exchange is the main factor limiting the number of LCD_A1 resonances that can be detected. Since residue-specific chemical shift assignment is impossible in such conditions, we chose 11 non-overlapping crosspeaks in the 2D ¹H-¹⁵N spectrum (indicated with arrows in Fig. S3) as internal probes to characterize the overall pressure-induced transitions experienced by LCD_A1.

Remarkably, when monitoring the intensity of the selected LCD_A1 crosspeaks over the 1–2.5 kbar pressure range, we also observed a significant pressure-induced loss of intensity that can be fitted to the same two-state model described by equations (1–3) (Fig. 4 A and B). Since the pressure-induced transition observed for LCD_A1 crosspeaks is manifestly not a simple “folded-to-unfolded” transition, we will define the corresponding volume changes as ΔV_LP→HP, where “LP” corresponds to main state populated by LCD_A1 at atmospheric pressure and “HP” the state promoted in high-pressure conditions. ΔV_LP→HP values measured at 290 K were large and negative with an average value of −43.6 ± 2.9 ml/mol (Table 1 and Fig. 4C). We found that the magnitude of ΔV_LP→HP values significantly decreased at 277 K with an average of −32.2 ± 1.2 ml/mol (Table 1 and Fig. 4C). This result indicates that the “HP” state of LCD_A1 has a smaller thermal expansivity than its “LP” state, strongly suggesting that the LP↔HP transition is indeed fundamentally different from a simple Folded↔Unfolded transition such as that observed in the GB1 domain. Indeed, in the case of GB1 unfolding transition, the thermal expansivity of the unfolded states is larger than that of the folded states (as expected for any globular protein) (Table 1).

Figure 4.

Pressure-induced conformational changes of LCD_A1 in GB1-LCD_A1. Individual LCD_A1 crosspeak intensities were monitored as a function of pressure at 290 K (A) and 277 K (B) and fitted to equation (3) (solid line) to obtain individual ΔV_LP→HP values. (C) Comparison of ΔV_LP→HP values measured for 11 LCD_A1 crosspeaks in GB1-LCD_A1 at 290 K (black line and dots) and 277 K (cyan line and dots).

In order to map the pressure-temperature landscape of GB1-LCD_A1 construct we recorded additional spectra at extreme low temperature and extreme high-pressure. The relative population of the folded state, in the GB1 domain, and “LP” state in the LCD_A1 were calculated for each condition based on average crosspeak intensities (Fig. 5). By mapping the relative population of each state over a broad range of pressure and temperature conditions, one can clearly observe that the folded population of GB1 domain decreases as pressure increases, and decreases even further with lower temperatures (green circle in Fig. 5). On the other hand, the main state populated by LCD_A1 (“LP” state) in standard conditions also decreases as pressure increases but its relative population increases at lower temperatures (blue rectangle in Fig. 5).

Figure 5.

Pressure-temperature mapping of the relative folded population of GB1 in GB1-LCD_A1 (green circle) and relative LP population of LCD_A1 in GB1-LCD_A1. The theoretical maxima (100%) were calculated for both GB1 folded state LCD_A1 LP state from the fit of the pressure denaturation profiles (see Fig. 3 C and D for GB1 and Fig. 3 A and B for LCD_A1) using equation (3) (i.e. IF in equation (3)). Numbers shown here correspond to the average of the relative population calculated over all individual crosspeaks.

Discussion.

The use of GB1 domain as a fusion tag to overcome solubility and sample stability issues has become a standard tool for biomolecular NMR studies.⁶² Fusion of GB1 with a low complexity domain such as LCD_A1 presents the additional advantage of preventing phase separation. The high salt buffer conditions of the experiments also prevented phase separation. The structure and thermodynamics of the GB1 has been extensively studied by NMR spectroscopy,^63,64 and the influence of pressure on GB1 native chemical shifts⁶⁵ as well as on its equilibrium and kinetic unfolding at acidic pH have been previously described.⁶⁶ Here we found that GB1 domain, either isolated or as part of the GB1-LCD_A1 construct, exhibits the typical pressure-temperature dependence that has been observed for many other small globular proteins: i) negative ΔV_F→U values (i.e. GB1 unfolds as pressure increases); ii) a decrease in ΔV_F→U magnitude as temperature increases (i.e. the thermal expansivity of GB1 unfolded states is larger than the expansivity of its folded states); and iii) the folded population of GB1 decreases at lower temperature (i.e. indicative of cold denaturation).¹¹

In contrast, LCD_A1 exhibits unique pressure-temperature dependence properties, that are distinct from both well-structured proteins (e.g. GB1 domain) and intrinsically disordered proteins (e.g. α-synuclein²³). Indeed, the results presented here demonstrate that LCD_A1 undergoes conformational changes as pressure increases (called here “LP↔HP” transition), characterized by negative ΔV_F→U values. Our data also show that the “HP” state has a smaller thermal expansivity than that of the “LP” state. In addition, the “LP” state is stabilized at lower temperatures. To reconcile all above observations, we propose a model in which the “LP↔HP” transition corresponds to transient formation and stabilization of structural motifs with concomitant exposure of polar and charged residues to solvent (Fig. 6). In this model, we hypothesize that LCD_A1 adopts a structurally disordered but compact conformation in standard conditions (i.e. atmospheric pressure) stabilized by a mesh-like network of transient polar and electrostatic interactions. This model conforms with recent experiments showing the LC domain of hnRNP A2 to be largely disordered, yet compact.⁴⁹ Exposure of polar and charged side chain to solvent is accompanied by a significant negative volume change as the density of water molecules is higher around polar and charged moieties. The sum of negative contribution resulting from exposure of these side chains would explain the large and negative ΔV_LP→HP values measured for LCD_A1, as pressure disrupts the network of transient interactions present in the LP state. Since pressure can promote the formation of small helical motifs,^25–27 we also hypothesize that the HP state is not entirely disordered but rather encompasses short structural motifs. The higher degree of secondary structure in the HP state compared to the LP state would explain the negative change in thermal expansivity measured upon LP-to-HP transition. Cold denaturation of these motifs would explain why the LP state is favored at low temperature (Fig. 5). Similar cold denaturation is expected for short β-strand motifs that have been identified recently in the low complexity domains of hnRNPA1 and hnRNPA2.^67,68 It should be noted that the significant increase in water density from 290 K to 268 K could also play a role in modulating the free-energy of LCD_A1 sub-ensembles at subfreezing temperatures. While we recognize that assessing the nature of these structural motifs is important, the small number of LCD_A1 crosspeaks observable with the present construct prevents us to fully assign the backbone chemical shifts of the low complexity domain with triple resonance NMR experiments.

Figure 6.

Model of the conformational changes experienced by LCD_A1 under high-pressure and low temperature conditions. Our data suggest that in standard conditions (atmospheric pressure), LCD_A1 adopts a compact conformation stabilized by a mesh-like network of polar and electrostatic interactions (LP state). High-pressure will promote the exposure of polar and charged side chains to solvent, therefore breaking the network of interactions present in the LP state. We hypothesize that the HP state also contains structural motifs that are denaturated at very low temperature.

Phase separation is generally believed to be triggered by multivalent weak interactions including electrostatic, cation-π, and π-π interactions.^31–35 A recent study on hnRNP A1 has demonstrated that valence of aromatic residues plays a major role in determining the temperature dependence of chain compaction.⁶⁹ In the case of TDP-43, phase separation has been shown to be mediated by an α-helix in its LC domain.^70,71 The data presented here suggest that a similar mechanism may take place for hnRNP A1. Indeed, the low-lying HP state identified in our study may contain short helical motifs able to mediate interactions between LC domains. While this mechanism is certainly not the sole thermodynamic factor driving hnRNP A1 phase separation, it may constitute a non-negligible contribution to the formation of liquid condensates.

Conclusion

We have shown that the low-complexity domain of hnRNP A1 exhibits unique temperature and pressure dependences. We measured volume differences associated with pressure-induced conformational changes, characterizing the transition between a disordered, compact state and a fully solvated, extended state. In view of these results, we hypothesize that the extended state of LCD_A1 contains helical motifs that may play a role in driving the formation of liquid condensates. This study illustrates how high-pressure NMR can be used to describe low lying populations in a structurally disordered ensemble.