Vps27-Hse1 and ESCRT-I complexes cooperate to increase efficiency of sorting ubiquitinated proteins at the endosome

Abstract

Ubiquitin (Ub) attachment to cell surface proteins causes their lysosomal degradation by incorporating them into lumenal membranes of multivesicular bodies (MVBs). Two yeast endosomal protein complexes have been proposed as Ub-sorting "receptors," the Vps27-Hse1 complex and the ESCRT-I complex. We used NMR spectroscopy and mutagenesis studies to map the Ub-binding surface for Vps27 and Vps23. Mutations in Ub that ablate only Vps27 binding or Vps23 binding blocked the ability of Ub to serve as an MVB sorting signal, supporting the idea that both the Vps27-Hse1 and ESCRT-I complexes interact with ubiquitinated cargo. Vps27 also bound Vps23 directly via two PSDP motifs present within the Vps27 COOH terminus. Loss of Vps27-Vps23 association led to less efficient sorting into the endosomal lumen. However, sorting of vacuolar proteases or the overall biogenesis of the MVB were not grossly affected. In contrast, disrupting interaction between Vps27 and Hse1 caused severe defects in carboxy peptidase Y sorting and MVB formation. These results indicate that both Ub-sorting complexes are coupled for efficient recognition of ubiquitinated cargo.

Figure 1.

NMR analysis of the Vps27 binding site on Ub. (A) The ¹⁵N ¹H NMR spectra of ¹⁵N-labeled Ub in the presence (black) and absence (red) of 100 μM GST-Vps27_1–351. (B) Enhanced view of the chemical shift spectra for the indicated residues. Scale bars indicate ppm for ¹⁵N (y axis) and ¹H (x axis). (C) Chemical shift differences for each Ub residue are plotted using the formula (0.2δN² + δH²)^1/2, where δ = the chemical shift difference in the presence and absence of GST-Vps27. Blue arrows indicate Ub residues affected by Vps27 binding that were mutated. Red arrows indicate Ub residues affected by Tsg101 binding that were mutated. (D) Structural model of Ub showing the proposed binding site for Vps27 (blue) and two residues of the Tsg101 binding site Q₆₂ E₆₄ (red). Shading of the Vps27 binding surface is weighted to the magnitude of chemical shift difference upon binding.

Figure 1.

NMR analysis of the Vps27 binding site on Ub. (A) The ¹⁵N ¹H NMR spectra of ¹⁵N-labeled Ub in the presence (black) and absence (red) of 100 μM GST-Vps27_1–351. (B) Enhanced view of the chemical shift spectra for the indicated residues. Scale bars indicate ppm for ¹⁵N (y axis) and ¹H (x axis). (C) Chemical shift differences for each Ub residue are plotted using the formula (0.2δN² + δH²)^1/2, where δ = the chemical shift difference in the presence and absence of GST-Vps27. Blue arrows indicate Ub residues affected by Vps27 binding that were mutated. Red arrows indicate Ub residues affected by Tsg101 binding that were mutated. (D) Structural model of Ub showing the proposed binding site for Vps27 (blue) and two residues of the Tsg101 binding site Q₆₂ E₆₄ (red). Shading of the Vps27 binding surface is weighted to the magnitude of chemical shift difference upon binding.

Figure 2.

Binding and sorting by mutant Ubs. Mutations were placed in Ub also containing K₂₉R, K₄₈R, and K₆₃R mutations and deleted of the COOH-terminal two glycine residues (R3ΔGG). (A) Mutant Ubs were fused in frame to an Fth1-GFP reporter protein and expressed in pep4 cells. GFP fluorescence images and differential interference contrast (DIC) images that identify the yeast vacuole are shown. Dashed lines indicate composite figure. (B) The indicated combinations of mutant Ubs were incorporated into a tandem array at the COOH terminus of Fth1-GFP. Mutations in the first and second positions are indicated, and the GFP fluorescence and DIC images of their localization are shown. (C) Binding the Vps23-UEV domain to GST-Ub. Vps23 (1–161) flanked by a V5 epitope (Vps23-UEV-V5) was expressed from the T7 promoter. Cleared lysate was incubated in GSH agarose beads with either GST, GST-Ub, or GST-Ub (R3ΔGG). Beads were washed and immunoblotted with anti-V5 antibody along with a 2% equivalent of the lysate. (D) Binding of the Vps27 and Vps23-UEV to a panel of GST-Ub mutants. GSH agarose beads bound with GST-Ub (WT) or the indicated mutants were incubated with either lysate from vps23 mutant yeast expressing Vps27-HA or bacterial lysate expressing Vps23 (1–161)-UEV-V5 domain. Beads were washed and immunoblotted with anti-HA or anti-V5 antibodies, respectively. All Ub mutants including the “wild-type” GST-Ub construct contained the R3ΔGG mutations. As a control, a truncated version of GST-Ub (Ub 1–35) is shown. Bar, 5 μm.

Figure 3.

Vps27 binds Vps23 directly via two PSDP motifs. (A) Yeast lysate containing full-length Vps23-HA was incubated with GSH-agarose bound with GST, GST-Vps27(353–622), GST-Vps27 (353–617), or GST-Vps27(353–579). Beads were washed and analyzed by immunoblotting with anti-HA antibodies or anti-clathrin antibodies together with a 10% equivalent of starting lysate. (B) Similar to A, except that bacterial lysate containing Vps23 (1–161)-UEV-V5 and Vps23 (1–186)-UEV-V5 domains was incubated with GST and GST-Vps27 fusion proteins. (C, left) Schematic of Vps27 together with the indicated NH₂-terminal VHS (Vps27, Hgs, STAM) domain, the PI3P binding FYVE domain, the region containing two UIM domains, and the clathrin box (CB). The length of the indicated Vps27 truncation mutants is graphed below. (C, right) Immunoprecipitation experiments using the indicated Vps27 truncation mutants. Yeast lysates from cells with various vps27 alleles also expressing full-length Vps23-HA or Hse1-HA were immunoprecipitated with preimmune (α-∅), anti-Vps27 NH₂ terminus (α-Vps27), or HA epitope (α-HA) antibodies. Beads were washed and analyzed by immunoblotting with monoclonal α-HA antibodies. (D) The Vps27 sequence between residues 431 and 485 is aligned with the same region from the Vps27 orthologues of other budding yeast. Identical residues are in bold. Residues within a PSDP motif selected for alanine substitution to make the vps27-ΔPSDP-1 (AAA_447–449) allele are indicated (***). A second conserved PSDP motif is found between residues 524 and 530. Residues selected for alanine substitution to make the vps27ΔPSDP-2 allele (AAA_525–527) are indicated (***). (E, left) Schematic of vps27 alleles containing a ΔPSDP-1 mutation alone or in combination with truncation at residue 524 or the ΔPSDP-2 mutation, which comprises the vps27-Δvps23 allele. (E, right) Cells expressing Vps23-HA and the indicated vps27 alleles were immunoprecipitated and immunoblotted for Vps23-HA as in C.

Figure 4.

dependent binding sites for Vps23 and Hse1. (A) The Vps27 sequence (residues 358–450) aligned with Vps27 orthologues of other budding yeast. Identical residues are in bold. Residues selected for triple alanine substitution are indicated (***). Lysates from cells expressing Hse1-HA and the indicated vps27 alleles were immunoprecipitated and immunoblotted for Hse1-HA. The ΔKIS allele (residues KIS416–418AAA) was renamed vps27-ΔHse1. (B) Lysates from vps27Δ cells with vector, VPS27 wild type, or the vps27-Δvps23 or vps27-ΔHse1 alleles in combination with Vps23-HA or Hse1-HA were immunoprecipitated with anti-Vps27, control (∅), or anti-HA polyclonal antibodies.

Figure 5.

Defects associated with loss of Vps27-Vps23 and Vps27-Hse1 association. Cells (vps27Δ::Kanr) were transformed with low copy plasmid carrying the vps27-ΔVps23 or vps27-ΔHse1 alleles or wild-type VPS27 along with plasmid expressing either (A) Fth1-GFP-Ub(Q₆₂A E₆₄A I₄₄A)-Ub(I₄₄A V₇₀A), (B) Fth1-GFP-Ub, or (C) Ste3-GFP. (D) Quantitation of CPY secretion from cells with the indicated vps27 allele (±SD). (E) Wild-type SEY6210 cells or isogenic vps27Δ cells with a stably integrated vps27-ΔVps23 allele were analyzed for sorting of GFP-Cps1. Cells were viewed by fluorescence and differential interference contrast (DIC) microscopy. Dashed lines indicate composite figure.