Role of protein conformation and weak interactions on γ-gliadin liquid-liquid phase separation

Wheat storage proteins, gliadins, were found to form in vitro condensates in 55% ethanol/water mixture by decreasing temperature. The possible role of this liquid-liquid phase separation (LLPS) process on the in vivo gliadins storage is elusive and remains to be explored. Here we use γ-gliadin as a model of wheat proteins to probe gliadins behavior in conditions near physiological conditions. Bioinformatic analyses suggest that γ-gliadin is a hybrid protein with N-terminal domain predicted to be disordered and C-terminal domain predicted to be ordered. Spectroscopic data highlight the disordered nature of γ-gliadin. We developed an in vitro approach consisting to first solubilize γ-gliadin in 55% ethanol (v/v) and to progressively decrease ethanol ratio in favor of increased aqueous solution. Our results show the ability of γ-gliadin to self-assemble into dynamic droplets through LLPS, with saturation concentrations ranging from 25.9 µM ± 0.85 µM (35% ethanol (v/v)) to 3.8 µM ± 0.1 µM (0% ethanol (v/v)). We demonstrate the importance of the predicted ordered C-terminal domain of γ-gliadin in the LLPS by highlighting the protein condensates transition from a liquid to a solid state under reducing conditions. We demonstrate by increasing ionic strength the role displayed by electrostatic interactions in the phase separation. We also show the importance of hydrogen bonds in this process. Finally, we discuss the importance of gliadins condensates in their accumulation and storage in the wheat seed.

(Top) IUPred plot predicts the intrinsic disorder of γ-gliadins. Residues with a score above 0.5 are predicted disordered (grey area), and residues with a score below 0.5 are predicted to be ordered (white area). (Bottom) Kyte & Doolittle plot estimates hydropathy scores of γ-gliadins residues. Residues with positive scores are predicted hydrophobic while residues with negative scores are predicted hydrophilic. B. Diagram of states for IDPs from CIDER tool. All γ-gliadin accessions are located in the region 1 (black points overlapped) corresponding to weak polyampholytes or weak polyelectrolytes that form globule or tadpole-like conformations (light green). C. Sequence properties of γ-gliadin accessions. K: charge patterning parameter; FCR, fraction of charged residues; NCPR, net charge per residue which is the difference between the fractions of positively charged and negatively charged residues.

2-Determination of saturation concentrations by fluorescence method
To determine saturation concentrations by fluorescence intensity, a calibration curve of labelled γ44 (fluorescence intensity vs γ44-TRITC concentration) was established at 55% ethanol to ensure complete solubilization of the protein (Figure S1, left). After inducing liquid-liquid phase separation, fluorescence intensities of dilute phases were measured and concentrations were determined according to the calibration curve. Labelling with TRITC that could affect saturation concentrations of γ44 was also tested (Figure S1, right). Results show limited effects of the labelling on the saturation concentrations. Saturating concentrations determined by fluorescence measurements are overall consistent with the established phase boundary of the γ44-gliadin phase diagram excepted at 30 % and 35 % ethanol (v/v). At these ethanol proportions, samples are highly turbid which may explain the overestimation of C sat values. All measurements were performed in triplicate and data were expressed as the mean ± standard deviation (SD).

3-Dynamic behaviour of γ44 liquid-like droplets
To determine whether γ44 droplets are in equilibrium with the diluted phase, we monitored exchanges assays under confocal microscopy at 20% ethanol (v/v). γ44 was previously labelled using two different fluorescent dyes: TRITC (red) or FITC (green). Dynamic exchanges were followed during 30 min at room temperature. The whole microscopic images are shown and demonstrate the progressive inclusion of γ44-FITC (green) into previously formed γ44-TRITC (red) droplets ( Figure S2). Fluorescence intensity and exchange rate vary from one droplet to another, some saturated signals are observed in the early stage of the assay (10 min) before becoming blurred (30 min).

4-Analysis of sulfhydryl groups of non-reduced and reduced γ44-gliadin
The free thiol content of non-reduced and reduced (10 mM DTT) γ44-gliadin was quantified using the DTNB (5,5'-dithio-bis-(2-nitrobenzoic acid)) assay at 55% ethanol (pH 8.0). The results show a free thiol content about 300 times higher in presence of 10 mM DTT. In other words, about 75% of the protein disulfide bonds are reduced. These observations are expected and confirm the presence of disulfide bounds in γ-gliadin protein.

γ44-gliadin Non-reduced Reduced
Mole of free thiol per mole of protein 0.1 ± 0.002 28.8 ± 0.021 Table 1 : DTNB analysis of free thiols in non-reduced and reduced γ-gliadin. All data are expressed as the mean ± standard deviation (SD).