The protein quality control machinery regulates its misassembled proteasome subunits

Abstract

Cellular toxicity introduced by protein misfolding threatens cell fitness and viability. Failure to eliminate these polypeptides is associated with various aggregation diseases. In eukaryotes, the ubiquitin proteasome system (UPS) plays a vital role in protein quality control (PQC), by selectively targeting misfolded proteins for degradation. While the assembly of the proteasome can be naturally impaired by many factors, the regulatory pathways that mediate the sorting and elimination of misassembled proteasomal subunits are poorly understood. Here, we reveal how the dysfunctional proteasome is controlled by the PQC machinery. We found that among the multilayered quality control mechanisms, UPS mediated degradation of its own misassembled subunits is the favored pathway. We also demonstrated that the Hsp42 chaperone mediates an alternative pathway, the accumulation of these subunits in cytoprotective compartments. Thus, we show that proteasome homeostasis is controlled through probing the level of proteasome assembly, and the interplay between UPS mediated degradation or their sorting into distinct cellular compartments.

Fig 1. (A) (i) The expression of GFP-RPN5 through a galactose inducible promoter can fully support cell growth.

Cells containing N-terminal fusions of GFP to Rpn5 and Rpn5ΔC, both expressed from a galactose-inducible promoter (GAL1-GFP-Rpn5 (SB147), and GAL1-GFP-Rpn5ΔC (SB148) respectively), were streaked on glucose (promoter shut-off), and galactose (induced expression) containing medium (YPD and YP-GAL respectively) at the semi-permissive temperature (30°C). (ii) Truncation at the C-teminus domain (CTD) domain of Rpn5 (Rpn5ΔC) leads to its nuclear mislocalization. Representative images of the GFP-Rpn5 nuclear localization (top), vs. the cytoscolic inclusions detected in GFP-Rpn5ΔC (bottom). The cells that were described in (i) were grown in 2% galactose-containing medium to logarithmic phase at the semi-permissive temperature (30°C). Cells were visualized by differential interference contrast (DIC) and GFP fluorescence. DAPI was used for nuclear staining. Unless otherwise stated, for all the fluorescent microscopy experiments, pictures represent high resolution (63x) images of a single plane chosen from z series images extending above and below the entire cell. The images shown were selected from at least 200 cells that were visualized in each experiment. Bars 5 μm. (B) The truncation of 45 amino acids at the C-terminus of Rpn5 (rpn5Δ45) results in cell death. To examine how different deletions at the CTD of Rpn5 affect cell growth, we created and sporulated heterozygous diploid strains for either wt, RPN5 (YSB243X4741), rpn5ΔC (34 amino acid truncation) (YSB688X4741), and rpn5Δ45 (45 aa truncation at its CTD) (YSB655). Tetrad dissection and incubation at 30°C showed that the presence of rpn5Δ45 in haploids caused cell death, as indicated by failure of spores harboring the hygromycinB (HygB) marker to germinate, when compared to the RPN5 control, and rpn5ΔC. HygB was used as a marker for wt and the truncated forms of RPN5. (C) The truncated human ortholog of RPN5 (PSMD12) fails to complement the temperature sensitivity of RPN5ΔC. 10-fold serial dilutions of the indicated strains were spotted on SD-uracil medium. Cells were incubated at the semi-permissive (30°C) and non-permissive (37°C) temperatures. All the strains harbor the temperature sensitive allele of RPN5 (rpn5ΔC) in the presence of a plasmid expressing either a wt copy of RPN5 (positive control, LSB2), an empty pRS316 plasmid (negative control, LSB3), a full copy of PSMD12 (LSB1) or PSMD12 truncated at its C-terminus (PSMD12ΔC, YSB106). (D, E) Truncation of a fragment larger than 20aa from the CTD of Rpn5 impairs its interaction with other proteasomal lid subunits. (D) Protein complementation assay [20,43]. The indicated strains are diploids heterozygote for various truncations (indicated at the x-axis; YSB756, YSB758, YSB861, YSB863, YSB741, YSB688, YSB243) at the C-terminus of RPN5 fused to mDHFR-F3 (Rpn5ΔX-F3), and one of the proteasomal lid subunits (indicated at the y-axis; YSB671, YSB244, YSB677, YSB245, YSB672, YSB689, YSB670) fused to mDHFR-F1,2 (RpnX-F1, 2). Single colonies were spotted on a rich medium (SD-complete) and allowed to grow for 2 days at 30°C. These colonies were then spotted back to rich medium, and methotrexate (MTX) containing media (30°C). Growth on MTX indicates a positive physical interaction. CDC19-F1,2/CLN3-F3 (YSB59) and CDC19-F1,2/MCK1-F3 (YSB60) diploids were used as negative (Neg) and positive (Pos) controls, respectively. Since it was previously shown that the proteasome lid subunit Rpn10 doesn’t interact with Rpn5, the fusion of Rpn10 to F1,2 (Rpn10-F1,2) was used as an additional negative control. (E) Schematic representation of the different constructs in which one half of the mDHFR fragment (mDHFR-F3) was fused by homologous recombination to the C-terminus of the wt Rpn5 (RPN5-wt-F3), and by truncating various aa lengths at its CTD (Rpn5Δ5-F3 to Rpn5Δ34-F3 (termed Rpn5ΔC-F3). The DHFR-F3 fragment is linked to HygB. (F) Correlation between the extent of truncation at the CTD of RPN5 and cell viability. We used a similar experimental setup as in C, this time using the same strains indicated along the x-axis of D, grown at semi-permissive (30°C), and restrictive temperature (34°C). (G) Lid misassembly in rpn5ΔC cells is associated with proteasome dysfunction. GFP-RPN5 (RPN5) (SB147), and GFP-rpn5ΔC (rpn5ΔC) (SB148) cells were grown at the semi-permissive (30°C), and samples were transferred to the restrictive temperature (34°C) for 3 hrs. Total protein extracts from both temperatures were subjected to immunoblotting (IB) with anti ubiquitin antibody (α-Ub). Ponceau staining of the blotted protein extracts is shown for loading control.

Fig 2. (A) The RP subunit Rpn11-RFP co-localizes with GFP-Rpn5ΔC. Logarithmically growing cells containing Rpn11 fused to RFP (Rpn11-RFP) and GFP-Rpn5ΔC (YSB1090) where grown in galactose containing medium at 30°C and shifted to 34°C for 3 hrs.

Cells were visualized by DIC, GFP, and mCherry. (B) Misassembled lid in rpn5ΔC is not associated with the proteasome. Rapidly lysed whole cell extracts from wt (endogenous levels of RPN5) (YSB219), and cells over expressing GFP-RPN5, or GFP-rpn5ΔC (SB147, SB148) thorough a galactose inducible promoter (GFP-RPN5, and GFP-rpn5ΔC respectively) were resolved by nondenaturing PAGE, and proteasome visualized by in-gel peptidase activity. wt proteasomes are found as a mixture of RP2CP and RPCP. Proteasomes in rpn5ΔC mutants migrate faster, pointing to a structural defect. Cells were grown at the semi-permissive (30°C), and samples were then transferred to the restrictive temperature (34°C) for 2 hrs. The GFP-rpn5ΔC/Δhsp42 samples are discussed below. (C,D) Rpn5ΔC cytosolic aggregates co-localize with the IPOD. Logarithmically growing cells co-expressing GFP-Rpn5ΔC, together with the IPOD markers Hsp104-TFP (YSB747) (C) and Rnq1-mCherry (YSB1004) (D), were grown in rich galactose containing medium at 30°C. Hsp104-TFP, and Rnq1-mCherry co-localized with GFP-Rpn5ΔC throughout the experiment. (E) GFP-Rpn5ΔC large cytosolic aggregates remain sequestered to the mother cells. Samples in logarithmic growth at the semi-permissive temperature (30°C) were stained by calcofluor. The localization of GFP-Rpn5ΔC (SB148) signal was scored as the percentage of daughter cells issued from IPOD containing mother cells with (puncta), or without (no puncta) cytosolic puncta. A minimum of 100 cells was counted (n >100); error bars show the standard deviation between two independent experiments.

Fig 3. (A) Complementation of rpn5ΔC by the wt copy of RPN5. 10-fold serial dilutions of diploid strains that are homozygote (ΔC/ΔC, YSB906), or heterozygote (wt/ΔC, YSB908) for the temperature sensitive allele of RPN5 (rpn5ΔC).

A wt diploid (wt/wt, BY4743) was used as a positive control. Cells were spotted on rich medium, and incubated at the semi-permissive (30°C), and restrictive (34°C) temperatures. (B) The wt copy of RPN5 in the GFP-RPN5ΔC/RPN5-RFP heterozygous diploid localizes to the nucleus. Logarithmically growing GFP-Rpn5ΔC/Rpn5-RFP heterozygous diploid cells (YSB1056) were grown in galactose containing medium at 30°C. Cells were visualized by DIC, GFP, DAPI and mCherry. (C) Complementation of the temperature sensitive phenotype of rpn5ΔC by the wt copy of RPN5 is associated with the elimination of Rpn5ΔC cytosolic aggregates. Similar to B, but using GFP-rpn5ΔC/RPN5 heterozygotes diploids (YSB908), and GFPΔC/GFP-rpn5ΔC homozygous diploids (YSB906) as a control. The graph shows quantitation of the percentage of cells with (blue), or without (w/o) (red) GFP-Rpn5 cytosolic puncta. (D) Degradation of GFP-Rpn5ΔC cytosolic aggregates is mediated by the proteasome. The presence of GFP-Rpn5ΔC signal was visualized in GFP-rpn5ΔC/RPN5 heterozygotes diploids (YSB908), in logarithmically growing cells (t-0), and at the indicated time points at 30°C, after the addition of the proteasome inhibitor, MG132, or DMSO (control). The presence or absence of cytosolic puncta was quantitated at the 210 min time point.

Fig 4. (A-C) Hsp42 and Hsp26 co-localize with misassembled proteasome lid.

(A,B) Logarithmically growing cells co-expressing GFP-Rpn5ΔC, together with Hsp42-TFP (YSB726) (A), or Hsp26-TFP (YSB903) (B), were grown in galactose containing media at 30°C. (C) Logarithmically growing cells co-expressing rpn5ΔC, Hsp42-TFP and Rpn11-GFP were grown in galactose containing media at 30°C and shifted to 34°C for 3 hrs (YSB577). (D) Co-localization of Hsp42 with the IPOD marker Hsp104 in an rpn5Δc background. Similar to C, but using rpn5ΔC cells co-expressing Hsp42-TFP and Hsp104-GFP (YSB792). (E-F) HSP42 is essential for the formation of the cytosolic aggregates by misassembled proteasome lid. (E) Similar to A, but in HSP42 wt control (YSB1044) (top), and Δhsp42 cells (YSB1045) (bottom). The localization of GFP-Rpn5ΔC signal was scored as the percentage of cells showing nuclear localization, or cytosolic puncta (blue and red bars respectively). An mCherry fusion with NIC96, a component of the nuclear pore complex, was used as a nuclear marker. Unless otherwise stated, for each of the graphs, minimum of 200 cells was counted (n >200); error bars show the standard deviation between two independent experiments. (F) Similar to A,B, but in Δhsp42 (YSB1002) and HSP42 (SB163) control cells grown at 34°C. DAPI was used for nuclear staining. (G) The nuclear relocalization of GFP-Rpn5ΔC in Δhsp42 cells is associated with growth restoration. 10-fold serial dilutions of the indicated strains (SB147, SB148, YSB868) were spotted on SC media supplemented with galactose (SC-GAL). Cells were incubated at the semi-permissive (30°C) and restrictive (34°C) temperatures. The wt, and strains harboring rpn5ΔC, were used as positive and negative controls, respectively. (H) The nuclear re-localization of GFP-Rpn5ΔC in Δhsp42 is associated with proteasome reassembly. Similar to 2B, but in cells over producing GFP-Rpn5ΔC, and deleted in the indicated chaperones (YSB219, SB148, YSB868, YSB954, YSB1174 respectively). Cells were grown at the semi-permissive temperature of 30°C, and shifted for 2 hrs to the restrictive temperature (34°C).

Fig 5. Models: UPS assembly can be naturally impaired by many factors, and thus, there is competition between assembly, degradation and aggregation of proteasome subunits.

When the proteasome lids are partially misassembled, as demonstrated in rpn5ΔC mutant at the semi-permissive temperature (i), the misassembled subunits are targeted to UPS-mediated degradation by the assembled 26S proteasomes which are still available. This degradation can take place in the nucleus, as recently shown for misfolded proteins [44]. The employment of rpn5ΔC at the restrictive temperature (ii), shifts the balance to slower assembly, which has a dual effect: It increases the amount of the unassembled subunit, and at the same time decreases its degradation, because there is less proteasomes available. Hence, missassemled proteasome subunits are aggregated in the IPOD, a process that depends mostly on HSP42.