Ubiquitin-independent proteasomal degradation driven by C-degron pathways

Abstract

Although most eukaryotic proteins are targeted for proteasomal degradation by ubiquitination, a subset have been demonstrated to undergo ubiquitin-independent proteasomal degradation (UbInPD). However, little is known about the molecular mechanisms driving UbInPD and the degrons involved. Utilizing the GPS-peptidome approach, a systematic method for degron discovery, we found thousands of sequences that promote UbInPD; thus, UbInPD is more prevalent than currently appreciated. Furthermore, mutagenesis experiments revealed specific C-terminal degrons required for UbInPD. Stability profiling of a genome-wide collection of human open reading frames identified 69 full-length proteins subject to UbInPD. These included REC8 and CDCA4, proteins which control proliferation and survival, as well as mislocalized secretory proteins, suggesting that UbInPD performs both regulatory and protein quality control functions. In the context of full-length proteins, C termini also play a role in promoting UbInPD. Finally, we found that Ubiquilin family proteins mediate the proteasomal targeting of a subset of UbInPD substrates.

Keywords: C-degron; Global Protein Stability (GPS) technology; Ubiquilin; Ubiquitin; degron; proteasome; ubiquitin-independent degradation.

Graphical abstract

Figure 1

GFP-peptides are degraded by the proteasome independent of ubiquitination (A) Ubiquitination is attenuated in E1-inhibited cells. HEK293T cells were treated with 0.5 or 1 μM MLN7243 for 7 h, and the abundance of ubiquitin conjugates assessed by western blot (WB) using the FK2 anti-ubiquitin antibody. Ponceau staining served as a loading control. (B) Cells expressing GPS constructs in which GFP was fused to the last 23 residues of CPS1, SREBF2, OR4C13, or MAGEA6, or last 37 residues of mouse ODC (mODC), were treated with vehicle control (DMSO), 1 μM MLN7243 or bortezomib for 7 h and analyzed by flow cytometry. The GFP/dsRed ratio represents the stability of the indicated GFP-fusion proteins. Stabilization of the target protein by the indicated inhibitors is indicated by a sharp peak to the right side of each panel. (C and D) HEK293T cells were treated with 1 μM MLN7243, 10 μM bortezomib, or 1 μM MLN7243 plus 10 μM bortezomib for 6 h. Ubiquitin-conjugated substrates abundance was assessed by WB using the VU-0101 anti-ubiquitin antibody (C), and the stability of the GFP-fusion proteins was assessed by flow cytometry (D).

Figure 2

GPS-peptidome screen identifies UbInPD substrates (A) For each of the indicated GFP-C23mers, the GPS screen profiles (representing the distribution of sequencing reads across the bins in control treated cells (DMSO) versus MLN7243 or bortezomib treatments) are shown on the left, and individual validations as analyzed by flow cytometry following treatment with the indicated inhibitors for 7 h are shown on the right. (B and C) U2OS cells (B) or mIMCD-3 cells (C) expressing GPS-C23mers of the indicated UbInPD substrates were treated with 1 μM MLN7243 or bortezomib for 7 h and analyzed by flow cytometry.

Figure 3

Specific C-terminal motifs are enriched among UbInPD substrates (A) Heatmaps showing the relative depletion (blue) or enrichment (red) of each amino acid across all positions of the 23mer peptide, comparing ubiquitin-independent substrates with ubiquitin-dependent substrates. (B) Boxplots showing the distribution of PSI scores for all UbInPD peptides harboring the indicated C-terminal motifs in untreated, bortezomib (Bort) or MLN7243 treated cells. Motifs were grouped as follows: Ala-end (A-1, A-2, and A-3), Cys-end (C-1, C-2, and C-3), and Val-end (V-1, V-2, V-3, and V-4). ^∗p < 1e⁻⁶⁸ and ^∗∗p < 1e⁻⁸ denote statistical significance comparing untreated with bortezomib or MLN7243, respectively (t test, see STAR Methods). (C) Boxplots showing the distribution of PSI scores for all peptides harboring the indicated classes of motif internally within the 23mer peptide (gray boxes) or specifically at the C terminus (colored boxes). (D) Scanning mutagenesis of UbInPD substrates. For each of the indicated genes, three independent mutagenesis experiments were performed: mutagenesis of single residues (top), mutagenesis of three consecutive residues (middle), and deletion of three consecutive residues (bottom). In each case, darker colors represent a greater degree of stabilization conferred by the mutation/deletion. Name of the gene is indicated, and the critical C-terminal residues regulating stability are shown in brackets. A universal scale of stabilization is shown at the bottom. (E) Saturation mutagenesis was performed for two UbInPD substrates. Each residue across the 23mer peptide was mutated to all other possible amino acids, and the stability of the resulting GFP-peptide fusions was measured by means of FACS and Illumina sequencing. In each case, the darker the color, the greater the degree of stabilization as compared with that of the wild-type sequences.

Figure 4

UbInPD substrate turnover is ubiquitination-independent (A–C) Ubiquitin conjugates are not detected on UbInPD substrates immunoprecipitated from cells using anti-GFP beads. The ubiquitination status of GFP-fused mODC and the C23mer from CAP2 are shown in (A), C23mers from ETV1 and JUND in (B), and C23mers from SHROOM and FBXO6 in (C). LSS represents a ubiquitin-dependent substrate that is heavily ubiquitinated when purified from cells. Immunoblot analysis was performed on both total cell extracts (TCEs) and immunoprecipitated (IP) GFP using anti-HA antibody to detect HA-ubiquitin conjugates. Anti-GFP antibody was used to show equal GFP pull downs across samples. (D) HEK293T cells expressing GFP-C23mers of UbInPD substrates HARS, JUND, FBXO6, and the ubiquitin-dependent substrate CDK5R1 were transfected for 24 h with HA-tagged wild-type (WT) or lysine-less ubiquitin (K0). Cell extracts analyzed using antibodies against GFP, HA, and TSPYL1. Vinculin served as a loading control. (E) HEK293T cells were transduced with HA-tagged lysine-less BirA alone (K₀BirA) or fused to the C23mer UbInPD substrates FBXO6 or ETV1. Following 6 h treatment with MLN7243 (MLN) or bortezomib (Bort) cells were harvested for western blot. HA-K₀BirA abundance was assessed by immunoblot with HA antibody while vinculin antibody was used as a loading control. The protein levels of TSPYL1, a ubiquitin-dependent substrate were monitored with endogenous TSPYL1 antibody. FBXO6 and ETV1 C23mers do not encode lysine.

Figure 5

UbInPD substrates are proteolyzed by both 20S and 26S proteasomes in vitro but are mostly degraded by 26S proteasomes in vivo (A–F) UbInPD substrates are degraded by purified proteasomes in vitro. Recombinant GFP (A and B), GFP-ETV1 (C and D), or GFP-FBXO6 (E and F) were incubated with purified 20S or 26S proteasomes for 3–9 h followed by analysis using Coomassie stain and western blot with anti-GFP antibody. Bortezomib (+Bort) was added to show that degradation is dependent on the catalytic activity of the proteasomes. Quantification of the degradation (based on three independent experiments) is presented in (B), (D), and (F). (G) UbInPD substrates are degraded by 26S proteasomes in vivo. Cells expressing inducible shRNAs against either luciferase (sh-control) or PSMD1 (sh-PSMD1) were transduced with the indicated GPS-C23mers, and their stability was analyzed by flow cytometry following 3 days of doxycycline (+DOX) treatment.

Figure 6

GPS-ORFeome screen identifies full-length UbInPD substrates (A) The nine indicated UbInPD candidate full-length proteins were expressed as GFP-fusions in HEK293T cells and their stability was analyzed by flow cytometry following 7 h treatment with 1 μM MLN7243 or bortezomib. (B) HEK293T cells expressing the indicated UbInPD substrates were transfected with HA-tagged wild-type (WT) or lysine-less ubiquitin (K0). 24 h post-transfection cells were harvested, and cell extracts analyzed by immunoblot. (C) Five candidate UbInPD substrates were expressed as N-terminal HA epitope tag in HEK293T cells, and protein abundance assessed by immunoblot using anti-HA antibody following 6 h treatment with 1 μM MLN7243 or bortezomib (Bort).

Figure 7

Stability of full-length UbInPD substrates is regulated by Ubiquilins (A) Stability of the indicated UbInPD full-length ORFs as analyzed by flow cytometry in control KO (cells transduced with AAVSI-targeting single guide RNA [sgRNA]) or penta KO cells lacking RAD23A, RAD23B, and Ubiquilin 1, 2, and 4 in the presence or absence of 1 μM MLN7243. (B and C) Stability analysis of the indicated GPS-ORFs in control KO cells or double KO cells lacking RAD23A and RAD23B (B), or triple KO cells lacking Ubiquilin 1, 2, and 4 (C) in the presence or absence of 1 μM MLN7243. (D and E) The indicated GPS-ORFs were expressed in control KO or Ubiquilins triple KO cells that were transduced with either wild-type (WT) UBQLN1, or the indicated deletion mutants. 48 h post-transduction, stability analysis was performed by flow cytometry (D). Protein abundance of the transduced complementary DNAs (cDNAs) was assessed by immunoblot using HA antibody (E). (F) HA-tagged UBQLN1 interacts with UbInPD GPS-ORFs in vivo. HEK293T cells stably expressing HA-tagged UBQLN1 with or without the indicated UbInPD GPS-ORFs substrates were generated. The GFP-fusions were immunoprecipitated from cells treated with 1 μM MLN7243 + 1 μM bortezomib followed by western blot analysis. GFP and HA antibodies are used to detect by immunoblot the GFP-fusions, and HA-tagged UBQLN1, respectively, in immunoprecipitation (IP) (left) and total cell extract (TCE) (right).