Folded domain charge properties influence the conformational behavior of disordered tails

Abstract

Intrinsically disordered proteins and protein regions (IDRs) make up around 30% of the human proteome where they play essential roles in dictating and regulating many core biological processes. While IDRs are often studied as isolated domains, in naturally occurring proteins most IDRs are found adjacent to folded domains, where they exist as either N- or C-terminal tails or as linkers connecting two folded domains. Prior work has shown that charge properties of IDRs can influence their conformational behavior, both in isolation and in the context of folded domains. In contrast, the converse scenario is less well-explored: how do the charge properties of folded domains influence IDR conformational behavior? To answer this question, we combined a large-scale structural bioinformatics analysis with all-atom implicit solvent simulations of both rationally designed and naturally occurring proteins. Our results reveal three key takeaways. Firstly, the relative position and accessibility of charged residues across the surface of a folded domain can dictate IDR conformational behavior, overriding expectations based on net surface charge properties. Secondly, naturally occurring proteins possess multiple charge patches that are physically accessible to local IDRs. Finally, even modest changes in the local electrostatic environment of a folded domain can substantially modulate IDR-folded domain interactions. Taken together, our results suggest that folded domain surfaces can act as local determinants of IDR conformational behavior.

Keywords: All-atom simulations; Conformational ensemble; Intrinsically disordered proteins; Sequence design.

Graphical abstract

Fig. 1

The charge sequence properties of disordered tails are similar to those of all IDRs. Distributions of relevant IDR properties. Three sets of IDRs were included in this analysis: all IDRs with or without structural data (N ​= ​34,095), all tails with or without structural data (N ​= ​11,254), and all tails with structural data (N ​= ​619). A) Violin/box plots of fraction of charged residues (FCR) across the three sets of IDRs. The horizontal black line in the box plot refers to the median, while the bottom/top of the box refers to the first/third quartile. B) Violin/box plots of net charge per residue (NCPR) across the three sets of IDRs. C) 2D density plot of fraction of positive charges and fraction of negative charges for all tails with or without structural data. Diagonal white lines separate the diagram of states into three regions: R1 (weak polyelectrolytes and polyampholytes), R2 (intermediate polyampholytes), and R3 (strong polyampholytes). D) Density plots of IDR length across the three sets of IDRs. Dashed line refers to the median IDR length for a given group.

Fig. 2

Overview of the IDR-FD systems constructed. Three variants of the sfGFP were designed (GFP⁺¹⁵, GFP⁺⁵, and GFP⁻¹⁵). For each variant, a subset of residues were mutated according to a ruleset such that each variant had a negative ‘patch’ of varying sizes emanating from the N-terminus. Each negative patch was then surrounded by a positive ‘patch’ encompassing the remainder of the protein. The black dot present on each of the GFP variants in their initial orientation corresponds to where the N-terminal tail is attached to. Black dotted lines superimposed on each GFP variant demarcate the boundary between their respective negative and positive patch. For each variant, we also report its mean electrostatic potential and patchiness value.

Fig. 3

The conformational behavior of (GSE)₁₂-GFP^Xand (GSK)₁₂-GFP^Xis dependent on the tail's net charge. In addition, the conformational behavior of (GSK)₁₂-GFP^X is dependent on surface charge properties. Each panel refers to a scaling map for a specific system: A) (GSE/K)₁₂-GFP⁺¹⁵. B) (GSE/K)₁₂-GFP⁺⁵. C) (GSE/K)₁₂-GFP⁻¹⁵. Each element on the scaling map refers to the average distance between residues i and j from all-atom implicit solvent simulations divided by the average distance between residues i and j in the corresponding excluded volume simulation. Inter-residue distances were calculated between all residues of the tail (residues 1–36) and the entire protein. To the right of each panel, we show representative snapshots of the resulting conformational ensemble.

Fig. 4

3D contour volume plots for (GSK)_n-GFP⁺¹⁵, (GSK)_n-GFP⁺⁵, and (GSK)_n-GFP⁻¹⁵at tail lengths 18, 24, 30, and 36 illustrate that conformations tend to cluster closer to the negative patch of the GFP in a manner dependent on tail length and surface charge distribution. The green surface refers to the GFP. Each colored region (excluding the green one) represents a 3D contour volume of the <x,y,z> coordinates of the terminal residue (notated as <x_t,y_t,z_t>) across all frames for each IDR-FD system. Each region was plotted to encapsulate 25% (red), 50% (orange), or 75% (yellow) of the data.

Fig. 5

3D contour volume plots for (GSE)_n-GFP⁺¹⁵, (GSE)_n-GFP⁺⁵, and (GSE)_n-GFP⁻¹⁵at tail lengths 18, 24, 30, and 36 illustrate that conformations cluster towards or near the positive patch of the GFP as tail length increases. The green surface refers to the GFP. Each colored region (excluding the green one) represents a 3D contour volume of the <x,y,z> coordinates of the terminal residue (notated as <x_t,y_t,z_t>) across all frames for each IDR-FD system. Each region was plotted to encapsulate 25% (red), 50% (orange), or 75% (yellow) of the data.

Fig. 6

The conformational ensemble of tails is influenced by the interplay of the tail's length and net charge and the FD's net charge and surface charge distribution. Internal scaling profiles for each tail ((GSE)_n or (GSK)_n) attached to each GFP variant (first, second, and third columns) and each tail in isolation (fourth column) for the full Hamiltonian simulations. The black dotted line represents the internal scaling profile for (GSE/K)₁₂ attached to each GFP variant and in isolation for the excluded volume simulations. The spatial separations for all pairs of IDR residues that are |j−i| apart in the linear sequence are calculated for each of the conformations in the relevant ensemble.The patterns of intra-IDR distances can be summarized in terms of so-called internal scaling profiles. The ensemble-averaged spatial distance for each sequence separation, denoted as $⟨⟨r_{i,j}⟩{⟩}_{|i-j|}$ is plotted against |j−i|.

Fig. 7

Framework for reasoning about how modulating specific properties of tails and folded domains can influence conformational behavior. A) Illustration of how the size of an adjoining, electrostatically complementary patch and the length of a tail can influence a tail's conformational behavior. When the length of the positive tail is small, the size of the negative patch is relatively unimportant in terms of influencing conformational behavior. However, as the length of the tail increases, the size of the negative patch influences the tail's conformation. When the size of the negative patch is small/large, the longer tail does not/does closely interact with it. B) Illustration of how the location of a distant electrostatically complementary patch and the length of a tail can influence a tail's conformational behavior. In the top left configuration, the tail is oriented towards the distant positive patch. As the length of the tail increases, this effect is magnified and the tail makes closer contact with the positive patch. However, when the location of the positive patch changes to being ‘sandwiched’ between two negatively charged regions, the tail no longer orients towards it.

Fig. 8

Patch visualization for membrane protein EPB41. A) APBS surface for the membrane protein EPB41 (PDB1gg3), with the N-terminal junction demarcated. From this surface, it is evident that there is a negative patch and positive patch, roughly equidistant from the N-terminal junction. B) This 3D information is condensed into a 2D representation by quantifying the mean electrostatic potential, relative size, and accessibility fraction of each patch as a function of distance between the N-terminal junction and the patch centroid along the surface of the protein.

Fig. 9

Patches on a protein are on average relatively large, accessible, numerous. A) Distribution of the mean, maximum, and minimum relative patch size across all 214 proteins. B) Distribution of the mean, maximum, and minimum patch accessibility fraction across all 214 proteins. C) Frequency distribution of the number of patches on a protein stratified by patch size. D) Frequency distribution of the number of patches on a protein stratified by the distance between the FD:IDR junction and patch centroid. E) Frequency distribution of the number of patches on a protein stratified by patch accessibility fraction. Vertical dotted lines correspond to the mean value for a given metric.

Fig. 10

The RBD of OC43 and CoV-2 N protein influences IDR conformational behavior and IDR-FD interactions. A) Aligned sequences of OC43 and CoV-2 N protein (top), aligned structures of OC43_RBD and CoV-2_RBD (bottom-left), and patch information for OC43_RBD and CoV-2_RBD (bottom-right). B) Internal scaling profiles for each tail (OC43_NTD or CoV-2_NTD) attached to each RBD (OC43_RBD and CoV-2_RBD) and each tail in isolation for the full Hamiltonian simulations. The black dotted line represents the internal scaling profile for the corresponding IDR-FD excluded volume simulations. C) Scaling maps for each of the four proteins; OC43_NTD-OC43_RBD (top-left), OC43_NTD-CoV-2_RBD (bottom-left), CoV-2_NTD-OC43_RBD (top-right), CoV-2_NTD-CoV-2_RBD (bottom-right). Each entry in the scaling map refers to the average distance between residues i and j in the full Hamiltonian simulation divided by the average distance between residues i and j in the corresponding excluded volume simulation. Inter-residue distances were calculated between all residues of a given tail and 11 specific residues on each RBD. Because there is not an exact one-to-one mapping between the linear sequence and their geometric location for the two RBDs, we determined an isomorphic mapping between 11 specific residues (e.g residue 92 [R] for CoV-2_RBD corresponds to residue 106 [R] for OC43_RBD, etc.). To the right of the scaling map, we show these residues in context of the electrostatic surface potential maps for OC43_RBD and CoV-2_RBD.