On the Prevalence and Roles of Proteins Undergoing Liquid-Liquid Phase Separation in the Biogenesis of PML-Bodies

Abstract

The formation and function of membrane-less organelles (MLOs) is one of the main driving forces in the molecular life of the cell. These processes are based on the separation of biopolymers into phases regulated by multiple specific and nonspecific inter- and intramolecular interactions. Among the realm of MLOs, a special place is taken by the promyelocytic leukemia nuclear bodies (PML-NBs or PML bodies), which are the intranuclear compartments involved in the regulation of cellular metabolism, transcription, the maintenance of genome stability, responses to viral infection, apoptosis, and tumor suppression. According to the accepted models, specific interactions, such as SUMO/SIM, the formation of disulfide bonds, etc., play a decisive role in the biogenesis of PML bodies. In this work, a number of bioinformatics approaches were used to study proteins found in the proteome of PML bodies for their tendency for spontaneous liquid-liquid phase separation (LLPS), which is usually caused by weak nonspecific interactions. A total of 205 proteins found in PML bodies have been identified. It has been suggested that UBC9, P53, HIPK2, and SUMO1 can be considered as the scaffold proteins of PML bodies. It was shown that more than half of the proteins in the analyzed proteome are capable of spontaneous LLPS, with 85% of the analyzed proteins being intrinsically disordered proteins (IDPs) and the remaining 15% being proteins with intrinsically disordered protein regions (IDPRs). About 44% of all proteins analyzed in this study contain SUMO binding sites and can potentially be SUMOylated. These data suggest that weak nonspecific interactions play a significantly larger role in the formation and biogenesis of PML bodies than previously expected.

Keywords: PML bodies; SUMOylation; intrinsically disordered proteins; intrinsically disordered regions; liquid–liquid phase separation; membrane-less organelles; post-translational modifications; protein–protein interactions.

Figure 1

PML protein domain organization scheme. All PML isoforms have the same N-terminal part that contains RING (R), B-Box1 (B1), B-Box2 (B2), and coiled-coil (CC) domains. The majority of PML isoforms contain a nuclear localization signal (NLS).

Figure 2

Scheme illustrating the design of our study.

Figure 3

Venn diagram of three sets of proteins potentially included in PML bodies. Blue circle represents the set based on the analysis of the PML interactome in the BIOGRID database. Green circle represents the set according to the analysis of the GO:0016605 and SL-0465 databases. Yellow circle represents the set collected during the analysis of literary sources.

Figure 4

Analysis of the interactome of PML bodies’ scaffold proteins. (A) Venn diagram represents the intersection of the interactomes of four PML bodies’ scaffold proteins (PML, DAXX, SP100, and CREBB(CBP)). Data were obtained using BioGRID v4.4. (B) Interaction map for eight potential PML body envelope proteins based on the statistical representation in the BIOGRID (evidence score) database. Yellow shows results with weak statistics (data from one experiment), blue shows data with strong statistics (>10 experiments). The heat map was constructed using CytoScape [221]. (C) Table of results for predicting the propensity of detected proteins to undergo spontaneous phase separation using FuzDrop and PSPrediction scores.

Figure 5

Analysis of global protein disorder within the PML-body proteome. (A) The output of PONDR^® VSL2, where the PONDR^® VSL2 score denotes the mean disorder score (MDS) for a given protein. This score is derived by summing up the per-residue disorder scores and dividing this sum by the number of residues in a query sequence. The PONDR^® VSL2 (%) corresponds to the percentage of predicted intrinsically disordered residues (PPIDR), indicating the proportion of residues with disorder scores above 0.5. Color-coded blocks depict distinct regions within the MDS-PPIDR phase space based on the degree of protein (dis)order: mostly disordered proteins are found in the red colored block, moderately disordered proteins are within the pink and light-pink blocks, and predominantly ordered proteins are positioned in the blue and light-blue blocks. Dark-colored background areas (blue, pink, and red) indicate concordance between the two parameters; i.e., MDS and PPIDR, while light-blue and light-pink colors signify areas where only one criterion aligns. The delineation of colored regions is contingent upon the arbitrary yet accepted cutoffs for PPDR (x-axis) and MDS (y-axis). (B) The cumulative distribution function and charge–hydropathy plot (ΔCDF-ΔCH) for the nuclear PML-body proteome comprising 205 proteins. The Y-coordinate is determined by the distance of each protein from the boundary in the CH plot, while the X-coordinate is calculated as the average distance of the protein’s CDF curve from the CDF boundary. The protein classification is based on the quadrant in which it is positioned. In Quadrant 1 (Q1), 70 proteins (34.1%) are projected to be ordered according to CDF and exhibit compact characteristics in the CH plot. In Quadrant 2 (Q2), 88 proteins (42.9%) are anticipated to be ordered/compact based on the CH plot but display disorder as per the CDF plot. Quadrant 3 (Q3) encompasses 44 proteins (21.5%) predicted to be disordered based on the CH plot and CDF. Quadrant 4 (Q4) includes 3 proteins (1.5%) anticipated to be disordered according to the CH plot but ordered according to the CDF analysis.

Figure 6

Proteome analysis of PML bodies. (A) Correlation of the results of two predictors (FuzDrop and PSPredictor) and the results in the form of a pie diagram. (B) Proportion of proteins potentially capable (SUMO) and incapable (NoSUMO) of SUMOylation in the analyzed set of proteins. (C) Share of LLPS drivers/clients. (D) Proportion of disordered proteins (IDPs) and proteins with intrinsically disordered protein regions (IDPRs). (E) Proportion of proteins potentially carrying a negative charge, a weak positive charge, and a positive charge. Proportion of proteins potentially capable (SUMO) and incapable (NoSUMO) of SUMOylation for LLPS (F), IDP (G) and weakly positive (H) drivers.

Figure 7

Result of GO analysis for proteins potentially prone to LLPS. Biological process and molecular functions are shown.