Interaction networks of SIM-binding groove mutants reveal alternate modes of SUMO binding and profound impact on SUMO conjugation

Abstract

The best-characterized mode of noncovalent SUMO interaction involves binding of a SUMO-interaction motif (SIM) to a conserved binding groove in SUMO. Our knowledge on other types of SUMO interactions is still limited. Using SIM-binding groove SUMO2/3 mutants and mass spectrometry, we identified proteins that bind to SUMO in an alternate manner. Domain-enrichment analysis characterized a group of WD40 repeat domain-containing proteins as SIM-independent SUMO interactors, and we validated direct binding of SEC13 and SEH1L to SUMO in vitro. Using AlphaFold-3 modeling and in vitro mutational analysis, we identified residues in the WD40 domain of SEC13 and SUMO2/3's C terminus involved in the interaction. Furthermore, SIM-binding groove mutants failed to interact with SUMO E3 ligases belonging to the PIAS family, RANBP2, ZNF451, and TOPORS, leading to loss of covalent conjugation to most of SUMO target proteins. Together, our dataset serves as a unique resource and offers valuable insights on the intricacies of the SUMO interactome and SUMO targets.

Fig. 1.. A noncovalent SUMO interactomics screen to identify SIM-independent SUMO binders.

(A) Cartoon illustrating the typical class I SUMO-SIM interaction and the concept of a SIM-binding groove SUMO mutant to capture alternate modes of interaction. The box depicts the SUMO variants used in the screen and corresponding sequences around the mutation site. (B) Cartoon illustrating the experimental setup of the screen. Empty nickel–nitrilotriacetic acid (Ni-NTA) beads and beads coated with either SUMO2/3 wild-type (WT), QFI, or IKR were incubated with HeLa lysates, and noncovalent SUMO interactors were purified and identified by mass spectrometry (MS). (C) Immunoblotting (IB) of a representative replicate of the screen, n = 4 (top: short exposure; bottom: long exposure). RNF4 was used as a control for binding to SUMO2/3 WT, but not the SIM-binding groove mutants. (C) Duplicated in the replicate 1 panel in fig. S1B.

Fig. 2.. Identification of exclusive SUMO2/3 WT and SIM-independent SUMO2/3 binders.

Volcano plots depicting the interactors for (A) SUMO2 WT, QFI, and IKR and (B) SUMO3 WT, QFI, and IKR compared to control beads. Proteins are represented by dots. Red dots represent highlighted examples of exclusive SUMO2/3 WT binders; green dots represent proteins previously validated in literature to bind SUMO in a SIM-dependent manner; blue dots represent highlighted examples of common SUMO2/3 WT and IKR/QFI binders; purple dot represents positive control RNF4. Significance was determined with a Student’s t test, false discovery rate (FDR) = 0.05 and s0 = 0.1, n = 4 independent experiments. FC, fold change. (C) Venn diagrams for the identified SUMO2 (left) and SUMO3 interactors (right). The exclusive SUMO2/3 WT binders are highlighted in red, and the common SUMO2/3 WT/QFI/IKR binders are highlighted in blue.

Fig. 3.. Analysis of exclusive SUMO2/3 WT binders.

(A) STRING network and MCODE clustering analysis of the exclusive SUMO2/3 WT binders. MCODE scores and associated biological processes are included for each cluster. Circles represent proteins identified as both SUMO2 and SUMO3 binder; diamonds represent proteins only identified as SUMO2 binder. Black border indicates proteins that we also identified as SUMO2 WT binders in a previous interactomics screen. Node color and size indicate the log₂ fold change and −log P value, respectively, of the enriched proteins compared to the control beads. rRNA, ribosomal RNA. (B) Gene Ontology (GO) enrichment analysis of the exclusive SUMO2/3 WT binders for biological process and molecular function. Bars indicate the −log FDR of the enrichment compared to the reference dataset. ncRNA, noncoding RNA.

Fig. 4.. Analysis of the common SUMO2/3 WT and SIM-binding groove mutant binders.

(A) STRING network and MCODE clustering analysis of the common SUMO2/3 WT/QFI/IKR binders. MCODE scores and associated biological processes are included for each cluster. The top six MCODE clusters are shown. Circles represent proteins identified as both SUMO2 and SUMO3 binder; diamonds represent proteins only identified as SUMO2 binder. Black border indicates proteins that we also identified as SUMO2 WT binders in a previous interactomics screen. Node color and size indicate the log₂ fold change and −log P value, respectively, of the enriched proteins compared to the control beads. (B) GO enrichment analysis of the common SUMO2/3 binders for biological process and molecular function. Bars indicate the −log FDR of the enrichment compared to the reference dataset. ATP, adenosine 5′-triphosphate; ADP, adenosine 5′-diphosphate.

Fig. 5.. Noncovalent SUMO binding of the Ub/Ubl machinery.

(A) Volcano plots from Fig. 2A depicting the interactors for SUMO2 WT, QFI, and IKR compared to control beads, respectively. Proteins are represented by dots. Red dots represent highlighted examples of the Ub/Ubl machinery exclusively binding to SUMO2 WT; blue dots represent highlighted examples of the Ub/Ubl machinery binding to SUMO2 WT/QFI/IKR; yellow dots represent highlighted examples of the Ub/Ubl machinery binding to SUMO2 WT and IKR or QFI. (B) STRING network of the identified Ub/Ubl proteins. Circles represent proteins identified as both SUMO2 and SUMO3 binder; diamonds represent proteins only identified as SUMO2 binder. Black border indicates proteins that we also identified as SUMO2 WT binders in a previous interactomics screen. Node color indicates whether proteins were identified as exclusive SUMO2/3 WT binder (red), SUMO2/3 WT/QFI/IKR binder (blue), or SUMO2/3 and IKR or QFI binder (yellow).

Fig. 6.. Domain enrichment analysis of SIM-dependent and SIM-independent SUMO2/3 binders.

(A) Domain enrichment analysis using Pfam (left) and InterPro databases (right). Bars indicate the −log FDR of the enrichment compared to the reference dataset. GTPase, guanosine triphosphatase. (B) STRING network of zinc finger proteins binding to SUMO2/3. Circles represent proteins identified as both SUMO2 and SUMO3 binder; diamonds represent proteins only identified as SUMO2 binder. Black border indicates proteins that we also identified as SUMO2 WT binders in a previous interactomics screen. Node color indicates whether proteins were identified as exclusive SUMO2/3 WT binder (red), SUMO2/3 WT/QFI/IKR binder (blue), or SUMO2/3 and IKR or QFI binder (yellow). (C) STRING network of WD40 repeat domain proteins. Circles represent proteins identified as both SUMO2 and SUMO3 binder; diamonds represent proteins only identified as SUMO2 binder. Black border indicates proteins that we also identified as SUMO2 WT binders in a previous interactomics screen. Node color and size indicate the log₂ fold change and −log P value, respectively, of the enriched proteins compared to the control beads.

Fig. 7.. WD40 repeat domain proteins bind directly to SUMO2/3 in a SIM-independent manner.

(A) Immunoblot verification of a representative replicate of the screen to confirm binding of the WD40 repeat domain proteins SEC13, SEH1L, and TAF5 to both SUMO2/3 WT and the SIM-binding groove mutants. A short (top) and a long (bottom) exposure are shown. (B) Cartoon depicting an in vitro glutathione S-transferase (GST)–pull-down assay with recombinant GST-SEC13. Elutions were analyzed by immunoblotting. GST only was used as a control for a specific binding. GST and GST-SEC13 were visualized by Ponceau S staining. Immunoblotting for SUMO2/3 confirmed binding of His¹⁰-SUMO2 to GST-SEC13. n = 3 independent experiments. (C) Similar as in (B), but for GST-SEH1L. GST and GST-SEH1L were visualized by Ponceau S staining and immunoblotting for GST. Immunoblotting for SUMO2/3 confirmed binding of His¹⁰-SUMO2 to GST-SEH1L. n = 3 independent experiments. (D) AlphaFold-3 prediction modeling of the SUMO-SEC13 interaction. Key residues involved in the interaction are highlighted. (E) Cartoon depicting an in vitro GST–pull-down assay with recombinant GST-SEC13 and His¹⁰-SUMO2 WT versus ∆D84-G92. Elutions were analyzed by SEC13 and SUMO2/3 immunoblotting. The amount of bound SUMO2 was quantified with SUMO2 WT set to 1.0 and plotted in a bar graph. Bar height indicates the mean. Dots represent replicate data. n = 2 independent experiments. (F) Cartoon depicting an in vitro GST–pull-down assay with recombinant GST-SEC13 WT versus P59A-Y79A-I151A-R216A-Q236A mutant and His¹⁰-SUMO2 WT. Elutions were analyzed by SEC13 and SUMO2/3 immunoblotting. The amount of bound SUMO2 was normalized to the amount of SEC13 present on the beads, quantified with SEC13 WT set to 1.0, and plotted in a bar graph. Bar height indicates the mean. Dots represent replicate data. n = 3 independent experiments.

Fig. 8.. A covalent SUMOylation screen to explore SIM-dependent SUMOylation.

(A) Cartoon illustrating a proposed model for the SIM dependency of covalent SUMOylation through the SIM-mediated recruitment of SUMO thioester-charged UBC9 to substrates, thereby facilitating lysine modification (top). Cartoon illustrating a proposed model for the SIM dependency of covalent SUMOylation through SUMO E3-dependent SUMOylation (bottom). RANBP2 contains a SIM that interacts with the donor SUMO (SUMO^D) and anchors it in the correct orientation for discharge. ZNF451 contains two SIMs in tandem, one which anchors the donor SUMO and another which binds the backside SUMO (SUMO^B) of UBC9. Members of the PIAS family also contain at least one SIM-like motif that positions the donor SUMO in the correct conformation. (B) Cartoon illustrating the experimental setup of the screen. HeLa cells expressing His¹⁰-SUMO2 WT or His¹⁰-SUMO2 IKR were lysed, and SUMOylated proteins were enriched by means of Ni-NTA pulldown, eluted, and analyzed by MS. (C) SUMO2/3 immunoblotting of a representative replicate of the screen, n = 4 independent experiments. Total lysates and His¹⁰–pull-down elutions were analyzed. Equal loading of total lysates was verified by Ponceau S staining. (C) Duplicated in the replicate 1 panel in fig. S9B. Volcano plots depicting the enriched proteins for HeLa His¹⁰-SUMO2 WT (D) and HeLa His¹⁰-SUMO2 IKR (E) compared to parental HeLa and for HeLa His10-SUMO2 WT compared to HeLa His¹⁰-SUMO2 IKR (F). Red dots represent highlighted examples of proteins exclusively SUMOylated in HeLa His¹⁰-SUMO2 WT; green dots represent proteins previously validated in literature; blue dots represent highlighted examples of proteins SUMOylated in both HeLa His¹⁰-SUMO2 WT and IKR. (G) Venn diagram for the identified SUMOylated proteins. Proteins exclusively SUMOylated by His¹⁰-SUMO2 WT are highlighted in red, and proteins SUMOylated by both His¹⁰-SUMO2 WT and IKR are highlighted in blue.

Fig. 9.. Proteins exclusively SUMOylated by SUMO2/3 WT.

STRING network of proteins exclusively SUMOylated in HeLa His¹⁰-SUMO2 WT. Node color and size indicate the log₂ fold change and −log P value, respectively, of the enriched proteins compared to parental HeLa.

Fig. 10.. Overlap between covalent SUMOylation and noncovalent SUMO binding.

(A) STRING network of proteins SUMOylated in both HeLa His¹⁰-SUMO2 WT and HeLa His¹⁰-SUMO2 IKR. Node color and size indicate the log₂ fold change and −log P value, respectively, of the enriched proteins compared to parental HeLa. (B) Venn diagram of the SUMO2 binders identified in the noncovalent screen and the proteins SUMOylated by SUMO2 as identified in the covalent screen. (C) Venn diagram of the proteins exclusively SUMOylated by His¹⁰-SUMO2 WT and proteins exclusively binding to His¹⁰-SUMO2 WT (top). Overlapping proteins are highlighted in red and listed in the adjacent box. Venn diagram of the proteins SUMOylated by both His¹⁰-SUMO2 WT and IKR, and proteins binding to both His¹⁰-SUMO2 WT and IKR (bottom). Overlapping proteins are highlighted in blue and listed in the adjacent box.