Regulation of the endosomal SNX27-retromer by OTULIN

Abstract

OTULIN (OTU Deubiquitinase With Linear Linkage Specificity) specifically hydrolyzes methionine1 (Met1)-linked ubiquitin chains conjugated by LUBAC (linear ubiquitin chain assembly complex). Here we report on the mass spectrometric identification of the OTULIN interactor SNX27 (sorting nexin 27), an adaptor of the endosomal retromer complex responsible for protein recycling to the cell surface. The C-terminal PDZ-binding motif (PDZbm) in OTULIN associates with the cargo-binding site in the PDZ domain of SNX27. By solving the structure of the OTU domain in complex with the PDZ domain, we demonstrate that a second interface contributes to the selective, high affinity interaction of OTULIN and SNX27. SNX27 does not affect OTULIN catalytic activity, OTULIN-LUBAC binding or Met1-linked ubiquitin chain homeostasis. However, via association, OTULIN antagonizes SNX27-dependent cargo loading, binding of SNX27 to the VPS26A-retromer subunit and endosome-to-plasma membrane trafficking. Thus, we define an additional, non-catalytic function of OTULIN in the regulation of SNX27-retromer assembly and recycling to the cell surface.

Fig. 1

Identification of SNX27 as an OTULIN interactor. a Selective protein enrichment after OTULIN ABP-PD. Volcano plot illustrating the enrichment of proteins identified by LC-MS/MS after PD in the presence or absence of OTULIN ABP. Curves depict significant enrichment or depletion, respectively. b Selective depletion of proteins from OTULIN ABP by PR-619 treatment. Volcano plot demonstrates loss of protein binding in samples treated with PR-619 before OTULIN ABP incubation and PD. Curves depict significant enrichment or depletion, respectively. c OTULIN-ABP PD was performed under conditions used for LC-MS/MS and binding of HOIP and SNX27 was analyzed by WB. d OTULIN ABP-PD was performed from extracts of parental or OTULIN KO Jurkat T cells. Co-precipitation of HOIP and SNX27 was analyzed by WB. e OTULIN ABP-PD was performed from extracts of parental or HOIP KO Jurkat T cells. Co-precipitation of HOIP and SNX27 was analyzed by WB. f Enrichment of proteins by PD of SF-OTULIN. Volcano plot depicts significant enrichment of proteins after Strep-PD from extracts of OTULIN-deficient Jurkat T cells expressing SF-OTULIN or mock control. Curves depict significant enrichment or depletion, respectively. g OTULIN-SNX27 interaction in reconstituted OTULIN KO Jurkat T cells was analyzed by WB after Strep-PD. h OTULIN-IP and SNX27-IP from extracts of Jurkat T cells. Rabbit or mouse IgG antibodies were used as isotype controls and protein binding was analyzed by WB. i OTULIN-IP from extracts of WT and OTULIN KO MEFs. Co-immunoprecipitation of SNX27 was analyzed by WB. j ABP-PD (left) and OTULIN-IP (right) from murine primary CD4 T cell extracts. Rabbit IgG antibody was used as an isotype control. Interaction of SNX27 to murine OTULIN-ABP complex and OTULIN was assessed by WB. Source data are provided as a Source Data file

Fig. 2

High affinity binding between OTULIN and SNX27. a Schematic representation of human OTULIN and SNX27 domain structure. b Sequence alignment and conservation of C-terminal PDZbm in OTULIN orthologues from tetrapod animals. c, d HEK293 cells were co-transfected with Flag-OTULIN and GFP-SNX27 constructs as indicated. Co-IPs were performed using anti-Flag antibody and tested for OTULIN and SNX27 interaction by WB. e ITC data for OTULIN PDZbm peptide (green) or OTULINcat (black) titrated against SNX27 PDZ domain. f ITC data for OTULINcat ΔETSL (red) and the reciprocal titration of OTULIN into SNX27 (purple). g Table summarizing the ITC data with mean parameters and the standard deviation derived from at least two independent experiments. Source data are provided as a Source Data file

Fig. 3

Structure of the OTULIN-SNX27 complex revealing the canonical PDZ-PDZbm interface. a Structure of OTULIN (blue cartoon) bound to SNX27 (brown surface). The OTULIN catalytic center and also the OTULIN PDZbm are highlighted (dotted lines). b Surface representation of OTULIN colored based upon conservation calculated through the Consurf webserver. The binding site for Met1 diubiquitin is shown as cartoon from superimposition of the OTULINcat Met1 diubiquitin structure (PDB: 3ZNZ). c Close-up view of the interaction between OTULIN PDZbm and SNX27_PDZ rotated 90° from enclosed box in a. A weighted 2Fo-Fc map is shown for residues from the OTULIN PDZbm contoured at 1.1σ. Hydrogen bonds are denoted by dotted lines. d Additional residues from OTULIN PDZbm engage with SNX27. A canonical PDZbm cargo of SNX27 (DGKζ; PDB: 5ELQ) (green) is superimposed onto the OTULIN-SNX27 structure. Residues at each position are shown and numbered according to convention. e Mutation of the canonical binding site disrupts the interaction between OTULIN and SNX27. ITC data for either a single point mutation within the OTULIN PDZbm (black) or SNX27 PDZ (red) are shown

Fig. 4

Secondary interface mediates high affinity interaction and selectivity between OTULIN and SNX27. a Surface representation of OTULIN colored based upon conservation calculated through the Consurf webserver. Box encloses the second SNX27 interface. b Close-up stereo view of the second interface of OTULIN from a. The unique β3–β4 hairpin insertion within the SNX27 PDZ domain is shown in orange. Hydrogen bonds between residues are shown with dotted lines. c Sequence alignment and conservation of second SNX27 binding surface in OTULIN orthologues from tetrapod animals. Conserved polar and electrostatic amino acids involved in binding to SNX27 are depicted in red. d Summary of SNX27–OTULIN affinities (K_D) determined by ITC for mutations in the secondary interface. e HEK293 cells were co-transfected with HA-OTULIN and GFP-SNX27 WT or mutants that carry missense mutations in the canonical cargo binding site (H114A) or in the second OTULIN interface (G64E/R100E). Binding was analyzed after GFP-Traps by WB. f HEK293 cells were co-transfected with GFP-SNX27 and HA-OTULIN constructs. OTULIN mutants either lack the C-terminal PDZbm (ΔETSL) or carry missense mutations in the second SNX27 interface (E85R/D87R). SNX27 binding was analyzed after HA-IP by WB. g Superimposition of structurally related PDZ domains onto SNX27–OTULIN structure. All closely related PDZ domains lack the SNX27 β3-β4 hairpin insertion. PDZ domains of SHANK1, NHERF1, and LNX2 contain a larger insertion that may clash with the second interface on OTULIN (left). PDZs of RhoGEF and Syntenin1 contain a short loop at the equivalent position that would not clash with the second interface of OTULIN (right). Source data are provided as a Source Data file

Fig. 5

OTULIN competes with VPS26 for SNX27 binding. a SNX27 utilizes residues from the β3–β4 insertion to engage with the retromer complex. Structure of SNX27 bound to VPS26A (green surface) (PDB: 4P2A). The canonical PDZbm of OTULIN is shown as a blue cartoon. Rotation of the complex by 45° with superimposition of the OTULIN-SNX27 and VPS26A-SNX27 structures onto SNX27. b Close-up view of a, with residues from the SNX27 β3–β4 hairpin that bind to VPS26A shown. c Close-up view of a revealing extensive clashes between OTULIN and VPS26A through overlapping binding sites on the β3–β4 hairpin. d Topology diagram of SNX27 showing the canonical PDZbm binding site and the VPS26A and OTULIN binding sites on the β3–β4 hairpin. e ITC data for SNX27 PDZ domain titrated against VPS26A. f HEK293 cells were co-transfected with GFP-SNX27, Flag-VPS26A, and HA-OTULIN constructs as indicated. GFP-SNX27 was precipitated using GFP-Traps and tested for simultaneous interactions with VPS26A and OTULIN by WB. g HEK293 cells were co-transfected with GFP-SNX27 and HA-OTULIN constructs as indicated. Following GFP-Trap precipitation, binding of GFP-SNX27 to HA-OTULIN and effects on interaction of GFP-SNX27 with the endogenous retromer subunits VPS26 and VPS35 were monitored by WB. Source data are provided as a Source Data file

Fig. 6

Binding of OTULIN antagonizes SNX27 association with early endosomes. a Co-elution of endogenous OTULIN and SNX27 was analyzed by size exclusion chromatography. Upper panels: cell lysates of parental Jurkat T cells were fractionated using a Superdex 200 column and elution profiles of endogenous proteins (OTULIN, SNX27, HOIP, VPS35, and VPS26) were determined by WB. Lower panels: determination of elution profiles of endogenous SNX27 and OTULIN in OTULIN and SNX27 KO Jurkat T cells, respectively. Peak elution of molecular weight standards is depicted at the top. b HEK293 cells virally transduced with GFP-SNX27 (green) were stained for early endosomes using anti-EEA1 antibody (red) and co-localization was analyzed by confocal microscopy. c HEK293 cells were co-transduced with GFP-SNX27 (green) and RFP-OTULIN WT, ΔETSL or C129A (red). Localization of proteins was analyzed by confocal fluorescence microscopy. d HEK293 cells virally transduced with RFP-OTULIN WT, ΔETSL or C129A (gray) were stained for endogenous SNX27 (green) and EEA1 (red) and localization was analyzed by confocal fluorescence microscopy. e WT or OTULIN KO HEK293 cells were stained for endogenous SNX27 (green) and EEA1 (red) and localization was analyzed by confocal fluorescence microscopy. Co-localization in d and e was quantified by determining Pearson’s correlation using at least 12 pictures and imaging more than 100 cells for each condition. Graphs depict the mean ± SD. Two-tailed p-values: ns not significant, *p ≤ 0.05, ***p ≤ 0.001 by unpaired t-test. Scale bars: 10 µm. Source data are provided as a Source Data file

Fig. 7

OTULIN counteracts SNX27 cargo loading. a Formation of a ternary HOIP-OTULIN-SNX27 complex was analyzed after co-expression of GFP-SNX27, HA-HOIP, and Flag-OTULIN in HEK293 cells. Ternary complex via Flag-OTULIN was verified using OTULIN Y56F (PIM mutation) and ΔETSL (PDZbm deletion) mutants. GFP-SNX27 was precipitated by GFP-Traps and association of OTULIN and HOIP was analyzed by WB. b SNX27 deficiency does not affect accumulation of Met1-ubiquitin chains. Extracts of WT, OTULIN-deficient, or SNX27-deficient HEK293 cells were analyzed for the abundance of Met1-ubiquitin chains and the expression of HOIP, OTULIN, and SNX27 by WB. c GFP or GFP-SNX27 constructs (WT or H114A) were expressed in HEK293 cells and cargo loading was analyzed by GFP-Traps by WB. d GFP-SNX27 WT was expressed in HEK293 cells alone or in the presence of HA-OTULIN WT, ΔETSL (PDZbm deletion), C129A (catalytically inactive), or C129A/L259E (catalytically inactive ubiquitin-binding mutant) and cargo loading was analyzed by GFP-Traps from cell lysates by WB. e Cargo-loading to endogenous SNX27 was assessed after anti-SNX27-IP in parental HEK293 cells as well as SNX27 KO or OTULIN KO cells. Binding of SNX27 to cargos and OTULIN was analyzed by WB. f Cargo-loading onto endogenous SNX27 was assessed by WB after anti-SNX27-IP in OTULIN KO HEK293 cells reconstituted with mock, OTULIN WT or OTULIN ΔETSL. Binding of each cargo to GFP-SNX27 (d) or SNX27 (e, f) was quantified from four independent experiments (see Supplementary Fig. 7a, b, e). A cumulative score for relative binding of all cargos to SNX27 was calculated and depicted as mean ± SD below the WB. Two-tailed p-values: ns not significant, ***p ≤ 0.001 by unpaired t-test. Source data are provided as a Source Data file

Fig. 8

OTULIN binding to SNX27 limits recycling of GLUT1 and SLC1A4 to the cell surface. a GLUT1 surface staining in mock, OTULIN WT and OTULIN ΔETSL transfected U2OS cells was performed using GLUT1.RBD.GFP reagent on living cells. GLUT1 surface expression was analyzed in untransfected (ΔCD2-low) and in transfected (ΔCD2-high) cells by FACS (left). Relative GLUT1 surface expression was determined as depicted in the histogram and changes in median fluorescence intensity (MFI) were normalized to mock. Graphs represent the mean ± SD of three independent experiments. Two-tailed p-values: ns not significant, ***p ≤ 0.001 by unpaired t-test. b U2OS cells lentivirally transduced with GFP-SLC1A4 (green) together with RFP, RFP-OTULIN, or RFP-OTULIN ΔETSL (gray) were stained for endogenous LAMP1 (red) and analyzed for SLC1A4/LAMP1 co-localization by confocal microscopy. Scale bars: 10 µm. c Co-localization was quantified by determination of Pearson’s correlation analyzing at least 14 random pictures and imaging of more than 100 cells for each condition. Graphs depict the mean ± SD. Two-tailed p-values: ns not significant, ***p ≤ 0.001 by unpaired t-test. d Knock-down in U2OS cells after siRNA transfection was verified by WB. e GLUT1 surface staining in U2OS cells after siRNA transfection as indicates was performed using GLUT1.RBD.GFP reagent on living cells. GLUT1 surface expression was analyzed by FACS. f Changes in median fluorescence intensity (MFI) were normalized to siRNA control to calculate relative GLUT1 surface expression. Graphs represent the mean ± SD of three independent experiments. Two-tailed p-values: ***p ≤ 0.001 by unpaired t-test. g Knock-down in HeLa cells after siRNA transfection was verified by WB. h HeLa cells transfected with siRNA as indicated and stained for endogenous GLUT1 (red) and the lysosomal marker LAMP1 (green). Co-localization of GLUT1 and LAMP1 was analyzed by confocal microscopy. Scale bars: 10 μm. i Co-localization was quantified by determination of Pearson’s correlation analyzing at least eight random pictures and imaging of more than 100 cells for each condition. Graphs depict the mean ± SD. Two-tailed p-values: ns not significant, ***p ≤ 0.001 by unpaired t-test. j Schematic model for the dual function of OTULIN. Through LUBAC binding and catalytic activity OTULIN controls Met1-ubiquitin chain homeostasis. By binding to SNX27 OTULIN counteracts cargo recruitment and retromer assembly to antagonize endosome-to-plasma membrane trafficking of internalized cargos. Source data are provided as a Source Data file