A manually curated PML-NB network. The central node is the scaffold protein PML. PML-NB components are connected to PML. Black edges represent protein components found in any of the databases (MINT, BIOGRID, IntAct, NPD, HRPD), pink edges were exclusively found in literature. We also represent the interactions between PML-NB components which are depicted as blue edges. For each node we provided the canonical name. Boxed components are those exclusively found in ALT cells.

Schematic representation of the SUMOylation pathway. Before conjugation, immature SUMO is processed by SUMO specific proteases (SENPs) to expose two C-terminal glycine residues required for conjugation. After processing, mature SUMO is activated in an ATP dependent manner by E1 enzymes AOS1 and UBA2. Subsequently, SUMO is transferred to E2 conjugation enzyme UBC9 and finally conjugated to a target protein by an E3 ligase. SUMO conjugation can be reversed by protease activity of SENPs thereby releasing free SUMO and target. K= Lysine residue.

Functional distribution of the PML-NB components. For each functional category the number of nuclear, cytoplasmatic and shuttle proteins are plotted. The categories were based on the keywords in UniProt Ontologies. Several related categories were merged to facilitate interpretation, such as 'intracellular signaling' (merge of Notch signaling, Wnt signaling, signal transduction and intracellular signaling ontologies), 'differentation and growth' (merge of growth, differentiation, muscle development, spermatogenesis and neurogenesis), 'PTM (chemical)' (acetylaction and phosphorylation), 'RNA processes' (RNA splicing, catabolism and processing), nuclear structure (chromatin and centrosome) and miscellaneous (lipid metabolism, oncogene and phagocytosis).

A. Receiver Operator Curve (ROC) of our in-house prediction algorithm with Area Under the Curve (AUC) value of 0,91. B. Overlap between literature, our predictions and the predictions performed by Yang et al.. Due to the different approach there are some unique prediction data for both our prediction method and the predictions performed by Yang et al.. C. Predicted percentages by our in-house algorithm (FPR =1%). Genome-wide we predict about one fourth of the proteins to be SUMOylated. Membrane and extracellular proteins are less likely to be SUMOylation targets then cytoplasmatic or nuclear proteins. Within the PML-NBs we see a clear enrichment of SUMOylatable proteins.

Representation of the SUMOylation pathway in relation to the PML-NB. Indicated are the proteins which have been linked to PML (blue, linked with PML). SUMO is synthesized as an immature protein which needs processing by SENPs. SENP1, SENP2 and SENP5 exert processing activity and have been linked to the PML-NBs implying that maturation may take place at the NBs. After maturation, the SUMO protein is activated by two E2 enzymes of which UBA2 was co-localized with the PML-NBs. Also the only conjugating UBC9 can be found in the PML-NBs and molecules which influence UBC9 such as p14ARF and MEL-18 have been targeted to the PML-NBs. For final conjugation of SUMO to a target protein, an E3 ligase is needed and apart from PML which potentially contains E3 activity itself, four other ligases (TOPORS, RNF4, RANBP2, PIASy) have been targeted to the NBs. We propose that the target protein, the E3 ligase and UBC9-SUMO meet at the nuclear bodies for the completion of the SUMO pathway. Finally, SUMOylation can be reversed by SENPs. Only de-conjugation SENP3 and SENP7 have not been linked to the PML-NBs thus far.