Purification and identification of endogenous polySUMO conjugates

Abstract

The small ubiquitin-like modifier (SUMO) can undergo self-modification to form polymeric chains that have been implicated in cellular processes such as meiosis, genome maintenance and stress response. Investigations into the biological role of polymeric chains have been hampered by the absence of a protocol for the purification of proteins linked to SUMO chains. In this paper, we describe a rapid affinity purification procedure for the isolation of endogenous polySUMO-modified species that generates highly purified material suitable for individual protein studies and proteomic analysis. We use this approach to identify more than 300 putative polySUMO conjugates from cultured eukaryotic cells.

Figure 1

An E3 ligase inactive RNF4 fragment that binds to polySUMO. (A) Domain layout of the SUMO-dependent ubiquitin ligase RNF4. RING and SIMs are shown. Below, an RNF4 fragment from amino acids 32–133 (RNF4wt) and a SIM-mutated fragment (RNF4mut). Mutated residues are shown in bold. (B) Sequence of the SIM peptide used to elute polySUMO conjugates comprising SIM of the PIASX protein with three amino acids added at the carboxyl-terminus. (C) Purity of the RNF4 fragments was determined by SDS–PAGE. (D) Recombinant SUMO2 chains were isolated with RNF4wt and RNF4mut, followed by elution with the SIM peptide. Samples were subjected to immunoblotting with SUMO2 antibodies. IB, immunoblotting; mut, mutant; RNF4, RING-finger 4; SDS–PAGE, SDS–polyacrylamide gel electrophoresis; SIM, small ubiquitin-like modifier interaction motif; SUMO, small ubiquitin-like modifier; wt, wild type.

Figure 2

Preservation of SUMO conjugates in nuclear and cytoplasmic fractionation. (A) HeLa cells in suspension were heat-stressed at 43°C for 1 h or control-treated at 37°C. In all, 10% of cells were lysed directly into Laemmli's sample buffer (lanes 1 and 4), with the remainder fractionated into cytoplasmic and nuclear extracts (lanes 2, 3, 5 and 6). The efficiency of fractionation and SUMO conjugates preservation was analysed by using tubulin, lamin A/C and SUMO2 antibodies (lanes 1–6). Nuclear lysates were used for purification with RNF4wt or RNF4mut crosslinked to beads. Eluates were analysed by immunoblot (lanes 7–10). Nuclear lysates represent 20% of the input for the pull-down experiments. (B) Silver-stained SDS–PAGE of the RNF4-dependent pull-downs from nuclear lysates from control and heat-stressed HeLa cells. Cyt., cytoplasmic; Nuc., nuclear; RNF4, RING-finger 4; SDS–PAGE, SDS–polyacrylamide gel electrophoresis; SUMO, small ubiquitin-like modifier.

Figure 3

Identification of 339 putative polySUMO conjugates after heat shock. (A) Left: Coomassie-stained SDS–PAGE gel showing the protein eluted from RNF4wt and RNF4mut purifications of polySUMO conjugates from HeLa cells grown under normal conditions (37°C) or heat-stressed (43°C) for 30 min. MW markers are shown and apparent MW boundaries of the nine slices are shown on the right. Right: Frequency histograms for the predicted MWs of all proteins found in each of the nine slices. Pink-shaded regions show the apparent MW regions of the slices. (B) Frequency histogram of the difference in molecular weight (ΔMW) between the apparent and predicted MWs of the 971 proteins binding to RNF4wt but not to RNF4mut. Coloured regions of the histogram indicate the ΔMW values consistent with unmodified proteins (grey) and those with single or several copies of SUMO molecules (red). (C) Comparison of the frequency of known SUMO substrates and the ψKxE/D SUMO conjugation consensus motifs in the indicated data sets. This comparison shows data for 2,663 proteins identified from a crude HeLa extract, along with the entire list of proteins identified in this study (971 proteins), as well as the subsets that have SUMO stoichiometry of 0 and 1 (‘0+1'; 632 proteins), or 2 or more (‘2+'; 339). Asterisks indicate lists filtered for redundancy. (D) Western blot analysis of RNF4 purifications, either mock or SENP1-treated, with indicated antibodies. mut, mutant; MW, molecular weight; PML, polymyelocytic leukaemia; RNF4, RING-finger 4; SDS–PAGE, SDS–polyacrylamide gel electrophoresis; SUMO, small ubiquitin-like modifier; wt, wild type.

Figure 4

Proteins involved in DNA replication and repair are polySUMOylated in response to heat stress. (A) Panther biological process gene ontology analysis of the entire data set and the ‘2+' set, and comparison with TAP–SUMO2 substrates and copurified non-substrates from Golebiowski et al (2009). (B) Same as in A, except the RNF4 purification data set is divided into six groups representing the different SUMOylation stoichiometries as indicated, and considering only the indicated gene ontology terms. (C,D) Network analysis of proteins involved in DNA repair and checkpoint control. Labels are gene names; node shapes indicate protein function: rhombus, enzyme; ellipse, transcriptional regulator; triangle, kinase; circle, other function. Lines indicate direct interactions. Nodes are coloured according to SUMOylation stoichiometry. Networks are created using ‘Ingenuity pathways analysis' (www.ingenuity.com). mRNA, messenger RNA; SUMO, small ubiquitin-like modifier.