Functional in vitro diversity of an intrinsically disordered plant protein during freeze-thawing is encoded by its structural plasticity

Abstract

Intrinsically disordered late embryogenesis abundant (LEA) proteins play a central role in the tolerance of plants and other organisms to dehydration brought upon, for example, by freezing temperatures, high salt concentration, drought or desiccation, and many LEA proteins have been found to stabilize dehydration-sensitive cellular structures. Their conformational ensembles are highly sensitive to the environment, allowing them to undergo conformational changes and adopt ordered secondary and quaternary structures and to participate in formation of membraneless organelles. In an interdisciplinary approach, we discovered how the functional diversity of the Arabidopsis thaliana LEA protein COR15A found in vitro is encoded in its structural repertoire, with the stabilization of membranes being achieved at the level of secondary structure and the stabilization of enzymes accomplished by the formation of oligomeric complexes. We provide molecular details on intra- and inter-monomeric helix-helix interactions, demonstrate how oligomerization is driven by an α-helical molecular recognition feature (α-MoRF) and provide a rationale that the formation of noncanonical, loosely packed, right-handed coiled-coils might be a recurring theme for homo- and hetero-oligomerization of LEA proteins.

Keywords: freezing tolerance; functional plasticity; intrinsically disordered protein; late embryogenesis abundant protein; self‐assembly.

FIGURE 1

Molecular crowding induced conformational changes of COR15A were studied in increasing concentrations of the proteinaceous crowder BSA and measured by SANS contrast matching in 40% D₂O. (a) Exemplary SANS data of COR15A without and with 15% v/v of the proteinaceous crowder BSA. The symbols represent the experimental data in log–log representation with solid lines depicting fits using generalized Gauss functions. (b) Radii of gyration (R _G) calculated from Guinier plots and (c) power law scaling coefficients d as derived from the fits in (a) in increasing concentrations of BSA. Error bars depict SD from three measurements. The full set of data in included in Figures S1–S3.

FIGURE 2

COR15A is associated in oligomeric complexes in planta. (a) Visual maps of exemplary rBiFC expression cassettes carrying the full‐length COR15A gene sequence in expression sites 1 and 2, and for a negative control, in which both sites encode the chloroplast signal peptide of COR15A (COR15A‐SP). (b) Representative confocal images of transiently transfected tobacco leaves expressing rBiFC constructs from (a) (Chl, Chlorophyll). (c) Quantification of normalized mean fluorescence from confocal imaging of rBiFC constructs reporting on COR15A homo‐ and COR15A/COR15B hetero‐assembly. We included two COR15A mutants with increased helicity (4GtoA and G66A) and in addition, prepared COR15A‐SP and COR15B‐SP as negative controls. Significant differences to COR15A calculated by unpaired t‐test or one‐way ANOVA are indicated (****p ≤ 0.0001, *p ≤ 0.01, ns = not significant, nd = not determined). (d) The upper panels show eluates after Co‐IP, which were resolved by SDS‐PAGE, and immunodetection was performed using n = 2 independent biological replicates. Vertical labels correspond to the gene sequence inserted in site 1 and horizontal ones to the one inserted in site 2. The asterisk (*) indicates free reconstituted YFP (28 kDa). Black arrow heads indicate dimeric (~65 kDa) and oligomeric COR15A and COR15A/B complexes (>170 kDa). Lower panels in (d) depict expression controls of both putative interaction partners in the raw protein lysates immunodetected with anti‐HA (upper) and anti‐Myc (lower) antibody. The respective controls for 4GtoA and G66A are shown in Figure S4.

FIGURE 3

Molecular docking followed by MD simulation of the COR15A dimer reveals intra‐ and intermolecular contacts. Structure predictions of the COR15A monomer (a) and dimer (b). The right panel in (a) and (b) shows the models in a projection along the helix axes. The COR15A monomer model depicts an amphipathic helix–loop–helix structure, with the apolar faces of both helices interacting via hydrophobic residues (a). Molecular docking of the COR15A dimer results in a four‐helix bundle (b). The hydrophobic core is flanked by hydrophilic and charged amino acids. Hydrophobic (red) and charged (blue) residues are highlighted. (c) Illustrative representation of the antiparallel orientation of the two monomers within the dimer with cylindrical shapes referring to the helical regions H1 and H2, and lines to the disordered linkers. The two COR15A monomers in the dimer oppose each other, rotated by 180° such that the loops of the monomers are located on different ends of the dimer. Thus, the N‐terminal helix of the first monomer (M1‐H1) faces the N‐terminal helix of the second monomer (M2‐H1) and vice versa for the C‐terminal helices (M1‐H2 and M2‐H2). (d–i) indicate the stability of the COR15A dimer in terms of changes in RMSD as a measure of the structural distance between backbone atoms (d), intramolecular (e) and protein‐water (f) H‐bonds and secondary structure (g–i) during 120 ns MD simulations. Error bars depict the standard deviation (SD) of 10 simulation replicates. (h) shows unraveling of helical structure resolved per time and residue. Ratio of helicity is represented by color gradient and the first monomer (M1) comprises residues 1 to 89 and second monomer (M2) residues 90–178. (i) Contact maps of the COR15A dimer during MD simulation. Values on the abscissa and ordinate of the contact maps represent the residue number of the COR15A dimer. Contact probabilities were calculated from 10 simulation replicates. (j) shows the most frequently established contacts, derived from all contacts with a contact probability of ≥0.75 from (i). Contacts between the two monomers in the dimer (intermolecular) are shown in blue and contacts within one monomer (intramolecular) are shown in red. (k) shows a ribbon representation of the COR15A dimers after 120 ns simulations, in which the two residues with the most contacts, F21 and V22, are indicated in shades of blue and red, respectively. The insets depict magnified image sections from (k), with the upper inset showing contact amino acids of F21 and the lower one showing contact amino acids of V22.

FIGURE 4

FV:AA mutation suppresses oligomerization in vitro and in planta. (a) Schematic representation of COR15A WT and mutants investigated in this study (sequence details are provided in Figure S7). The FV:AA substitution is indicated by blue and red spheres and the 4GtoA substitution (referring to the substitution of glycine at positions 52, 66, 79, and 82 with alanine) by gray and yellow spheres. The two helical domains H1 and H2 are schematically depicted in light blue and light red, respectively. Apparent relative masses (M _rel = M/M _monomer) (b) and hydrodynamic radii R _S (c) of COR15A and its mutants were measured as a function of protein concentration by simultaneous SLS/ DLS experiments and extrapolated to infinite dilution in order to eliminate the influence of intermolecular repulsion or attraction. The resulting M _rel and R _S are plotted as a function of glycerol concentration. (d) The radius of gyration (R _G) calculated from SAXS in buffer (filled bars) and in 70% glycerol (shaded bars). Error bars depict SD from at least three measurements. Data on COR15A in (b–d) were taken from (Shou et al., 2019). rBiFC (e) and Co‐IP (f) of COR15A_FV:AA and 4GtoA_FV:AA to address homo‐oligomerization and COR15A_FV:AA/COR15B and the swapped construct COR15B/COR15A_FV:AA to address hetero‐oligomerization in transiently transformed tobacco leaves. For controls as introduced earlier, compare Figure 2. In (e), asterisks indicate significant differences calculated using one‐way ANOVA statistical analysis (****p ≤ 0.0001, ***0.0001 ≤ p ≤ 0.001, *p ≤ 0.5, ns = not significant). The asterisk in (f) represents free reconstituted YFP. In (f), the anti‐Myc antibody was used to pull‐down protein complexes, which were immunodetected using an anti‐HA antibody. Vertical labels correspond to the gene sequence inserted in site 1 and horizontal ones to the one inserted in site 2. Black arrow heads indicate dimeric (65 kDa) and oligomeric COR15A and COR15A/COR15B complexes (>170 kDa).

FIGURE 5

Impact of COR15A mutations on overall disorder, coil–helix transition, intermolecular contacts and structural rigidity. Far‐UV CD spectra of COR15A and its mutants were measured in the fully hydrated state (a) and with increasing concentrations of the cosolutes glycerol and TFE (Figure S9). Fraction of α‐helicity, estimated from the θ _MRW at 222 nm, of the four proteins as a function of glycerol and TFE concentration are show in (b) and (c), respectively. The datasets for COR15A and 4GtoA in the presence of TFE and for COR15A in the presence of glycerol have been published previously (Sowemimo et al., 2019). Contact maps (d) and correlation of Cα carbons (e) during 120 ns MD simulation of the monomers of COR15A and its mutants. Values on the abscissa and ordinate of the contact maps represent the residue numbers. Contact probabilities and correlations were calculated from 10 simulation replicates for each protein.

FIGURE 6

Impact of mutations on COR15A functionality in vivo and in vitro. (a) The freezing tolerance of not cold acclimated Arabidopsis wild type (Col‐0) and COR15A overexpression lines. Ox_COR15A refers to plants overexpressing COR15A and ox_COR15A_FV:AA to plants overexpressing COR15A_FV:AA. COR15A expression levels of all plant lines are summarized in Figure S10. Freezing tolerance was determined from electrolyte leakage measurements and is indicated as LT₅₀. Error bars represent SEM from at least five biological replicates. Asterisks indicate the p‐value determined by unpaired Student's t‐test <0.01. (b–d) refer to the stabilization of biological targets by COR15A during freeze–thawing in vitro. (b) and (c) CF leakage from ICMM liposomes after a freeze–thaw cycle to −20°C. ICMMs were frozen in increasing concentrations of COR15A and its mutants at 12 different liposome surface occupancies. Raw data (Figure S11) were fitted to a dose response model yielding the minimum CF leakage at high liposome surface occupancies as an adaptable parameter. The regression curves are plotted as a function of ICMM surface occupancy in order to account for size differences between the COR15A variants (b). The horizontal bold line shows CF leakage of ICMMs frozen in the absence of (wo) protein ± SEM (gray shaded area). (c) Minimum CF‐leakage values at full surface occupancy derived from nonlinear regression analysis expressing the liposome stabilization capacity with error bars referring to ± SEM and asterisks indicate the p‐value determined by unpaired student's t‐test <0.05. (d) The effect of COR15A and its mutants on the activity of LDH after zero to five freeze–thaw cycles in liquid nitrogen as a function of LDH surface occupancy. LDH activity is plotted normalized to unfrozen controls in the absence of any protectant.

FIGURE 7

Schematic summary of the impact of COR15A mutations on helicity, oligomeric state and helix–helix interaction of COR15A under conditions promoting folding and their impact on functionality. (a) Protein monomers are represented by gray boxes. H1 and H2 are shown along the helix axes and are depicted as gray circles. Distances between boxes reflect the dimeric or monomeric character of the respective COR15A variant and distances between circles an interaction between H1 and H2. Blue circles indicate amino acid 21, either occupied by an F or an A and red circles amino acid 22, either occupied by a V or an A, both situated on H1. Colored arrows indicate contacts established via these residues. Green circles indicate residues on H2, originally G, but replaced by A in some of the mutants. μH depicts the hydrophobic moment vector calculated from helical wheel projections using the respective amino acid sequences as input (Figure S13) and gray vertical lines indicate the limit between the hydrophobic and the hydrophilic faces of the amphipathic helices. The substitution of four conserved G by A resulted in the oligomeric COR15A variant 4GtoA, characterized by superior α‐helicity in vitro (Sowemimo et al., 2019). Substitution of F21 and V22 with alanine in COR15A_FV:AA severely destabilized the oligomeric state in vitro and in planta and largely suppressed the coil–helix transition. In the double mutant 4GtoA_FV:AA self‐assembly was suppressed in vitro, but not in planta, indicating that monomer–monomer interaction was stabilized by the cellular compared to the in vitro environment. In addition, the coil–helix transition was not impacted in the double mutant in vitro. We rationalized this behavior by interactions between the two transiently α‐helical halves of native COR15A conferring mutual stabilization, as V22, situated on H1, is a major inter‐monomeric contact between H1 and H2. It is oriented in the center of the hydrophobic face, which represents the interface for intramolecular helix–helix interaction and is mainly in contact with V62 and G66. This interaction is destabilized by the V22A substitution in COR15A_FV:AA, but not in 4GtoA_FV:AA, where the G66A substitution introduced a more hydrophobic residue and thus rescued H1–H2 interaction and consequently the coil–helix transition. Looking into inter‐monomeric helix–helix interaction, H2 does not represent an interface for self‐assembly. Intermolecular interactions are exclusively driven by H1, primarily by F21, which is oriented perpendicular to μH in the boundary of the hydrophobic face and establishes helix–helix interaction with A14, K17, and A18 of H1 of the second monomer and vice versa, thus stabilizing the COR15A and 4GtoA oligomers. Interestingly, suppression of oligomer formation is not a consequence of the reduced folding propensity of COR15A_FV:AA, as also 4GtoA_FV:AA did not self‐assemble in vitro, while its coil–helix transition was not impaired. (b) Summary of the impact of the mutations on protein structure and the assessed in vitro functions. Colors indicate the interconnection of structure and function. Light, full and dark red colors reflect (i) low, intermediate and high helicity and (ii) low, intermediate and high membrane stabilization capacity of the four COR15A variants and both patterns are similar. Likewise, dark and light blue colors indicate (i) dimeric and monomeric and (ii) high and low molecular shielding efficiency and again, both patterns are similar.