Ufd2p synthesizes branched ubiquitin chains to promote the degradation of substrates modified with atypical chains

Abstract

Ubiquitination of a subset of proteins by ubiquitin chain elongation factors (E4), represented by Ufd2p in Saccharomyces cerevisiae, is a pivotal regulator for many biological processes. However, the mechanism of Ufd2p-mediated ubiquitination is largely unclear. Here, we show that Ufd2p catalyses K48-linked multi-monoubiquitination on K29-linked ubiquitin chains assembled by the ubiquitin ligase (Ufd4p), resulting in branched ubiquitin chains. This reaction depends on the interaction of K29-linked ubiquitin chains with two N-terminal loops of Ufd2p. Only following the addition of K48-linked ubiquitin to substrates modified with K29-linked ubiquitin chains, can the substrates be escorted to the proteasome for degradation. We demonstrate that this ubiquitin chain linkage switching reaction is essential for ERAD, oleic acid and acid pH resistance in yeast. Thus, our results suggest that Ufd2p functions by switching ubiquitin chain linkages to allow the degradation of proteins modified with a ubiquitin linkage, which is normally not targeted to the proteasome.

Figure 1. Ufd4p and Ufd2p synthesize K29- and K48-linked ubiquitin chains, respectively.

(a) Coomassie blue-stained gels showing the expression and purification of relevant enzymes. (b) Reconstitution of the Ufd4p-Ufd2p ubiquitination system. Either Ufd2p (Lane 2), Ufd4p (Lane 3), or both (Lane 4) were added to the Ub-V-GFP ubiquitination system in vitro, which contains E1, E2, ubiquitin, ATP and Ub-V-GFP (Lane 1). (c) Ufd4p catalyses K29-linked polyubiquitin chain assembly on Ub-V-GFP. Ufd4p-mediated polyubiquitination of a Ub-V-GFP mutant with only K29 and K48 in its ubiquitin region (Lanes 4–5). The reaction products were indistinguishable from the products of the K29-only Ub-V-GFP mutant (Lanes 1–3) in the presence of wild-type, K29R- or K29-only ubiquitin. (d) The linkage specificity of Ufd2p-mediated ubiquitination was examined using a single-round ubiquitin turnover assay. K29R ubiquitin was charged onto the active cysteine of Ubc4p, and the reaction was quenched by adding NEM/EDTA. Wild-type ubiquitin or the K6-, K11-, K27-, K29-, K33-, K48-, or K63-only ubiquitin mutants (Lanes 2–9) was added in excess (10 fold of K29R ubiquitin) to the mixtures, together with Ufd2p, allowing transfer of the charged K29R ubiquitin to either wild-type ubiquitin or its variants to form free di-ubiquitin chains (Lanes 2 and 8). (e) K29 and K48 on Ub-V-GFP are necessary and required for full ubiquitination by Ufd4p and Ufd2p. K48-only Ub-V-GFP, wild-type Ub-V-GFP, K29-only Ub-V-GFP, and K29/48-only Ub-V-GFP were tested for their ability to be ubiquitinated by the Ufd4p-Ufd2p ubiquitination system. See also Supplementary Figs 1–3.

Figure 2. Ufd2p and Ufd4p synthesize branched ubiquitin chains on the substrate.

(a) Schematic representation of the stepwise ubiquitination experiment. (b) Ufd2p promotes ubiquitination independent of Ufd4p. As illustrated in (a), either Ub-V-GFP or Ub2-GFP synthesized by Ufd4p was subjected to immunoprecipitation (IP) with the anti-GFP antibody. E1, E2, ubiquitin, ATP and Ufd2p were then added to the mixture. The reaction products were detected using the anti-GFP antibody; the Ufd2p and Ufd4p proteins were detected using the anti-Ufd2 and anti-Xpress antibodies, respectively. Input was from step I reaction products. To exclude the influence of Ufd4p in the second step, we used two controls: one contained Ufd4p alone (set a), and the other did not include Ub-V-GFP in step I (set b). (c) Ufd2p transfers one ubiquitin to the proximal ubiquitin of Ub2-GFP to form branched ubiquitin chains on the substrate. As illustrated in (a), K29R Ub, K29/48R Ub or wild-type Ub was used in step I, and wild-type Ub, K48R Ub or methylated Ub was used in step II. Input was from step I reaction products. (d) As in (c), except that the ubiquitin mutants in step I were replaced by K29R-Wt di-Ub, K29R-K48R di-Ub, K29/48R-Wt di-Ub, or K29/48R-K48R di-Ub. (e) As in (c), except that the ubiquitin mutants in step I were replaced with K29R Ub, K29R-Wt di-Ub and K29R-Wt-Wt tri-Ub. (f) As in (e), K29R-Wt-Wt tri-Ub was used in step I and wild-type Ub, K48R Ub or methylated Ub was used in step II, but the incubation time was 2 h. (g) A model for the functional role of Ufd2p and Ufd4p in the stepwise ubiquitination assay. First, Ufd4p assembles K29-linked ubiquitin chains on Ub-V-GFP; Ufd2p then catalyses multi-monoubiquitination on Ub-V-GFP modified with K29-linked ubiquitin chains to form branched ubiquitin chains. See also Supplementary Figs 4–6.

Figure 3. Branched ubiquitin chains are detected on UFD substrates in cells.

(a) Schematic representation of the method used to monitor the synthesis of branched ubiquitin chains on Ub-V-GFP. (b) In vitro validation of the method used to monitor branched ubiquitin chains on Ub-V-GFP. As illustrated in (a), Ub-V-GFP was ubiquitinated by Ufd4p or Ufd4p-Ufd2p. Reaction products were treated with the TEV enzyme and analysed by immunoblotting with the anti-GFP antibody. (c) Ub-V-GFP was modified with branched ubiquitin chains in yeast cells. Ub-V-GFP and its modified forms were expressed in a wild-type strain expressing FLAG-Ub^53TEV and Myc-Ub^64TEV/FLAG and purified under denaturing conditions in the presence or absence of the proteasome inhibitor MG132. The pull-down products were treated with the TEV enzyme and analysed by immunoblotting with anti-FLAG and anti-GFP antibodies. (d) The addition of branched ubiquitin chains to Ub-V-GFP is mainly dependent on UFD2 in vivo. Ub-V-GFP and its modified forms were purified from either a wild-type or ufd2Δ strain expressing both FLAG-Ub^53TEV and Myc-Ub^64TEV/FLAG in the presence of MG132 under denaturing conditions. The pull-down products were treated with the TEV enzyme and analysed by immunoblotting with the anti-FLAG and anti-GFP antibodies. (e) K29 and K48 on Ub-V-GFP are required for the formation of branched ubiquitin chains in yeast cells. Wild-type, K29R, K48R or K29/48R Ub-V-GFP and their modified forms were purified from a wild-type strain expressing FLAG-Ub^53TEV and Myc-Ub^64TEV/FLAG. The pull-down products were treated with the TEV enzyme and analysed by immunoblotting with the anti-FLAG and anti-GFP antibodies.

Figure 4. Ufd2p binds Ubn-GFP via two N-terminal loops.

(a) Interactions between the substrate and Ufd2p. Biotinylated Ufd2p was immobilized on streptavidin-coated beads, which were then incubated with GFP, Ub-V-GFP or Ubn-GFP that was synthesized by Ufd4p. The precipitated proteins were analysed by immunoblotting with the indicated antibodies. (b) Quantification of the relative binding affinity between Ufd2p and Ubn-GFP. Relative amounts of Ub1-4-GFP were quantified using Odyssey software and compared with the input. Error bars denote the s.e.m. of three independent replicates. * indicated P<0.05, ^** indicated P<0.01. (c) A summary of the Ufd2p variants used to map the Ubn-GFP binding motif, which include Ufd2p (amino acids 1–961), Ufd2p CF1 (amino acids 519–961), Ufd2p CF2 (amino acids 760–961), Ufd2p CF3 (amino acids 879–961), Ufd2p MFD1(amino acids Δ111-879), Ufd2p MFD2(amino acids Δ75-879), Ufd2p MFD3(amino acids Δ55-879), Ufd2p MFD4(amino acids Δ55-75&111-879), Ufd2p NHD1(amino acids Δ35-55) and Ufd2p MFD2(amino acids Δ75-111). (d) Amino acids 35–55 and 75–111 of Ufd2p are necessary to bind to Ubn-GFP. GST-Ufd2p and its variants were purified and used to pull down Ubn-GFP, which was synthesized by Ufd4p; GST was used as a control. Asterisks in the bottom panel indicate GST products cleaved from the fused proteins. (e) Ufd2p binds to Ubn-GFP via its two N-terminal loops. GST-Ufd2p and GST-Ufd2p NLM (E51A K52A L53A D54A K55A E105A N106A M109A N110A) were purified and used to pull down Ubn-GFP. Asterisks in the bottom panel indicate GST products cleaved from GST-Ufd2p NLM. (f) SPR sensorgrams for the binding of GST-Ufd2p to K29-linked Ub2-GFP. A series of two-fold GST-Ufd2p dilutions was applied to the K29-linked Ub2-GFP surface (GFP served as a control). (g) SPR sensorgrams for the binding of GST-Ufd2p NLM to K29-linked Ub2-GFP. (h) The interaction between Ufd2p and Ubn-GFP is necessary for the E4 activity of Ufd2p. Ufd2p or Ufd2p NLM was incubated with E1, E2, Ufd4p, ATP, ubiquitin and Ub-V-GFP and detected using the anti-GFP antibody. See also Supplementary Figs 7–11.

Figure 5. Substrates modified with branched ubiquitin chains are recognized by Rad23p, Dsk2p and Rpn10p.

(a) Rad23p prefers to bind to branched ubiquitin chains on Ub-V-GFP over Ub-V-GFP with K29-linked ubiquitin chains or unmodified Ub-V-GFP. Ub-V-GFP, Ub-V-GFP with K29-linked ubiquitin chains (synthesized via Ufd4p) or Ub-V-GFP with branched ubiquitin chains (synthesized via Udf4p-Ufd2p) immobilized on protein A-sepharose beads were used to immunoprecipitate GST-Rad23p. The precipitated proteins were analysed by immunoblotting with the indicated antibodies. The asterisks indicate GST products cleaved from GST-Rad23p. (b) SPR sensorgrams for the binding of GST-Rad23p to Ub-V-GFP. A series of two-fold GST-Rad23p dilutions was applied to the Ub-V-GFP surface (GFP served as a control). (c) SPR sensorgrams for the binding of GST-Rad23p to K29-linked Ub2-GFP. (d) SPR sensorgrams for the binding of GST-Rad23p to branched Ub3-GFP. (e) Proteasome adaptor proteins such as Rad23p fail to recognize Ufd2p NLM-mediated ubiquitination products. Either wild-type Ufd2p-mediated ubiquitination products or Ufd2p NLM-mediated ubiquitination products immobilized on protein A-sepharose beads were used to immunoprecipitate GST-Rad23p. Ubn-GFP modified by Ufd2p NLM could not immunoprecipitate GST-Rad23p efficiently. Asterisks indicate nonspecific background signals from the IgG chains. (f) The ubiquitin chain linkage switch is necessary for substrate degradation. Protein expression in wild-type or ufd2Δ strains expressing Ub-V-GFP under the control of a galactose-induced promoter was stopped upon transfer to 2% glucose. The ufd2Δ strains harboured empty vector, UFD2 or UFD2^NLM under the control of a galactose-induced promoter. The degradation of Ub-V-GFP over time was analysed by immunoblotting. Pgk1p served as a loading control. (g) Quantification of relative Ub-V-GFP levels in (f) using Odyssey software. Error bars denote the s.e.m. of three independent replicates. See also Supplementary Figs 12–14.

Figure 6. The branched chain forming activity of Ufd2p is required for ERAD, the OLE pathway and acid resistance in yeast.

(a) The Ubn-GFP binding motif of Ufd2p is essential to the degradation of a subset of proteins during ERAD. Protein expression in wild-type and mutant strains (ufd2Δ strain contains empty vector, UFD2, UFD2^ΔU-box or UFD2^NLM under the control of a galactose-induced promoter) expressing Myc-Hmg2p under the control of a galactose-induced promoter was stopped by transfer to 2% glucose and addition of cycloheximide. The degradation of Myc-Hmg2p over time was analysed by immunoblotting. Pgk1p served as a loading control. (b) The Ubn-GFP binding motif of Ufd2p affects the stability of Spt23p in the OLE pathway. The degradation and detection of Myc-Spt23p expression under the control of a galactose-induced promoter were analysed as described above. (c) Quantification of Myc-Hmg2p turnover. The relative amounts of Myc-Hmg2p were quantified using Odyssey software; the high molecular weight smear representing the ubiquitylated species was considered when determining the turnover rate. Error bars denote the s.e.m. of three independent replicates. (d) Quantification of Myc-Spt23p p90 turnover. Error bars denote the s.e.m. of three independent replicates. (e) The Ubn-GFP binding motif of Ufd2p is essential to the OLE pathway and acidic resistance: 10-fold serial dilutions of the indicated strains (the wild-type strain contains empty vector; ufd2Δ strains contain empty vector, UFD2, UFD2^ΔU-box or UFD2^NLM under the UFD2 promoter) were plated on media supplemented with either 0.02% oleic acid or 100 mM MES (pH 1.0) and grown at 30 °C for 2 days.

Figure 7. The functional role of Ufd2p in ubiquitination.

Model of how Ufd2p functions as a ubiquitin chain-linkage switcher to promote the degradation of proteins conjugated to non-canonically linked ubiquitin chains.