Ubiquitin-recognition protein Ufd1 couples the endoplasmic reticulum (ER) stress response to cell cycle control

Abstract

The ubiquitin-recognition protein Ufd1 facilitates clearance of misfolded proteins through the endoplasmic reticulum (ER)-associated degradation (ERAD) pathway. Here we report that prolonged ER stress represses Ufd1 expression to trigger cell cycle delay, which contributes to ERAD. Remarkably, down-regulation of Ufd1 enhances ubiquitination and destabilization of Skp2 mediated by the anaphase-promoting complex or cyclosome bound to Cdh1 (APC/C^Cdh1), resulting in accumulation of the cyclin-dependent kinase inhibitor p27 and a concomitant cell cycle delay during the G1 phase that enables more efficient clearance of misfolded proteins. Mechanistically, nuclear Ufd1 recruits the deubiquitinating enzyme USP13 to counteract APC/C^Cdh1-mediated ubiquitination of Skp2. Our data identify a coordinated cell cycle response to prolonged ER stress through regulation of the Cdh1-Skp2-p27 axis by Ufd1 and USP13.

Regulated protein degradation by the ubiquitin-proteasome system (UPS) plays a central role in diverse cellular processes. An effort to dissect this proteolytic system in Saccharomyces cerevisiae uncovered Ufd1 in a genetic screen for mutations in the ubiquitin fusion degradation (UFD) pathway responsible for the degradation of a synthetic substrate N-terminally fused to one ubiquitin moiety (1). Functional and structural evidence suggests that Ufd1 acts as an ubiquitin-recognition protein with putative monoubiquitin and polyubiquitin binding sites (2).

Ufd1 is best characterized as an adaptor protein that, together with Npl4, confers AAA-ATPase Cdc48/p97/VCP–specific activity in endoplasmic reticulum (ER)-associated degradation (ERAD) (3), a process that functions constitutively to export misfolded proteins from the ER to the cytosol for UPS-dependent degradation. ERAD is part of the unfolded protein response (UPR), which is critical for restoring homeostasis in the ER when its function is perturbed, such as by the accumulation of misfolded proteins. Analyses of ERAD performed primarily in yeast suggest a role for the Ufd1-Npl4-p97 complex in recognizing and extracting polyubiquitinated misfolded proteins from the ER of misfolded proteins. Intriguingly, in mammalian cells, Ufd1 also directly enhances the activity of gp78, an ubiquitin ligase involved in ERAD, independently of p97 and Npl4 (4). Studies in yeast have reported impaired degradation of the ER proteins HMG-CoA reductase, H-2Kb (MHC class I heavy chain), and CPY* (an aberrant form of carboxypeptidase Y) in the Ufd1 mutant, suggesting an essential role for Ufd1 in promoting ERAD (3, 5). However, RNAi-mediated depletion of Ufd1 in mammalian cells has yielded opposite results. While some studies reported impaired ERAD in Ufd1-depleted cells (6, 7), others showed accelerated degradation of the classical ERAD substrates, such as cholera toxin and T-cell receptors (8, 9). Although these seemingly contradictory observations might stem from the use of distinct cellular systems, they point to greater complexity in the regulation and function of Ufd1 that impinge on the ER stress response.

Furthermore, In Xenopus laevis egg extracts, Ufd1-Npl4-p97 promote chromatin decondensation (10) and regulate spindle disassembly (11).

In the present study, we examined the function of Ufd1 in maintaining steady-state levels of Skp2 (the F-box adaptor of the E3 ubiquitin ligase SCF^Skp2) in mammalian cells by regulating its ubiquitination. We report that Ufd1 acts as a scaffold for Skp2 and the deubiquitinating enzyme (DUB) USP13 to antagonize anaphase-promoting complex or cyclosome bound to Cdh1 (APC/C^Cdh1)-mediated ubiquitination of Skp2. Our findings also show that prolonged ER stress down-regulates Ufd1 levels, triggering Skp2 destabilization and accumulation of p27, which contribute to delayed progression through G1. Our data also demonstrate facilitated degradation of misfolded proteins in G1-arrested cells, suggesting that the link between Ufd1 and cell cycle control serves to optimize ERAD.

Results

Prolonged Treatment with Tunicamycin Down-Regulates Ufd1 Expression.

To investigate the role of Ufd1 in the UPR of mammalian cells, we first examined the expression of endogenous Ufd1 protein in HeLa cells treated with tunicamycin (TM), a glycosylation inhibitor. To our surprise, we found decreased Ufd1 expression at 20 h after treatment with 0.5 μg/mL of TM, concomitant with the accumulation of GRP78/BiP, a known marker of the UPR (Fig. 1A). Down-regulation of Ufd1 in mammalian cells exposed to prolonged TM treatment may accommodate a novel function of Ufd1 in the UPR, distinct from its direct role in the retrotranslocation of misfolded proteins.

Fig. 1.

Prolonged ER stress down-regulates Ufd1, and depletion of Ufd1 by RNAi affects cell cycle progression. (A) Protein levels of endogenous Ufd1 were compared in HeLa cells treated with DMSO (−) or 0.5 μg/mL of TM (+) for 20 h by immunoblot analysis. GRP78 was used as a marker of UPR, and β-actin served as a loading control. (B) Immunoblot analysis of cell cycle markers in control and Ufd1-KD cells released from G1/S arrest after double-thymidine block. Control refers to cells infected with an empty shRNA plasmid. Levels of Ufd1 were quantified and are shown in arbitrary units (AU). (C) (Upper) Comparison of CDK1 and CDK2 kinase activity in control and Ufd1-KD cells released from G1/S arrest. CDK2 kinase reactions from control and Ufd1-KD cells were run in separate gels but were processed identically otherwise. Same for CDK1 kinase reactions. (Lower) Protein levels of Skp2 in control and Ufd1-KD cells. (D) Analysis of cell cycle markers in control and Ufd1-KD cells expressing FLAG-Ufd1. The melanoma cell line CHL1 expressing high levels of p27 was used solely as a positive control for p27 immunodetection.

Ufd1 Regulates the Cdh1-Skp2-p27 Axis.

To understand the significance of ER stress-dependent down-regulation of Ufd1, we first examined the effect of Ufd1 down-regulation under nonstressed conditions by depleting Ufd1 from HeLa cells. Knockdown of Ufd1 (~75% reduction) was achieved in a stable Ufd1-deficient cell line (Ufd1-KD) and was validated at both the mRNA and protein levels (Fig. S1 A and B; drug-selected pool 2).

Given the link between Ufd1 and the cell cycle, we asked whether Ufd1 knockdown would affect cell cycle progression. By examining the levels of major cell cycle markers in Ufd1-KD cells synchronized in G1/S by double-thymidine block and release, we observed an overall down-regulation of Skp2 accompanied by an up-regulation of p27, a major substrate of SCF^Skp2 (12) (Fig. 1B). Accordingly, we found that although the order of activation of cyclin-dependent kinase (CDK) 1 and CDK2 was preserved in Ufd1-KD cells, their activation was delayed (Fig. 1C). Expression of exogenous Ufd1 in Ufd1-KD cells restored Skp2 and p27 levels to those seen in control cells (Fig. 1D). The effect of Ufd1 on Skp2 was confirmed in an independently established stable Ufd1-KD clone, as well as in cells with transient knockdown of Ufd1 by siRNA (Fig. S2A).

Down-regulation of Skp2 in Ufd1-KD cells can occur at the level of transcription, translation, or protein stability. Whereas qRT-PCR analysis revealed a twofold increase in Skp2 transcription in Ufd1-KD cells compared with controls (Fig. 2A), kinetics performed with cycloheximide (CHX) demonstrated reduced protein stability of endogenous Skp2 in Ufd1-KD cells (Fig. S2B). Using X. laevis egg extracts supplemented with exogenous Ufd1 in the presence of recombinant Cdh1, the adaptor of the APC/C ubiquitin ligase responsible for Skp2 degradation in the cell cycle (13–17), we confirmed that Ufd1 is a negative regulator of Skp2 degradation (Fig. 2B). Accordingly, the stability of endogenous Skp2 on exit from mitosis, when APC/C^Cdh1 is active in the cell cycle, was increased in cells that overexpressed Ufd1 (Fig. 2C).

Fig. 2.

Ufd1 interferes with the ubiquitination of Skp2 in vivo. (A) Quantification of Skp2 mRNA levels in control and Ufd1-KD cells by SYBR-green qRT-PCR. (B) Skp2 degradation assays in interphase Xenopus egg extracts supplemented with recombinant Cdh1, in the presence of the indicated amount of bacterially purified recombinant GST-Ufd1. Samples were obtained at the indicated times after ³⁵S-Skp2 was added to the extract. (C) (Upper) Half-life analysis of endogenous Skp2 protein with CHX treatment in empty vector- or FLAG-Ufd1–transfected HeLa cells released from nocodazole (noco) arrest. Total lysates were immunoblotted for Skp2 or FLAG. Quantification is shown below the immunoblots. (D) In vivo ubiquitination of Skp2. myc-Skp2 and HA-Ub were expressed in control and Ufd1-KD cells. Immunoprecipitates obtained with c-myc antibodies were immunoblotted for either ubiquitin (Ub) or c-myc to detect polyubiquitinated Skp2, denoted by (Ub)_n-Skp2. Control refers to cells infected with empty shRNA plasmid. (E) In vivo ubiquitination of Skp2. myc-Skp2 and HA-Ub were coexpressed with either empty vector or FLAG-Ufd1. Immunoprecipitates obtained with c-myc antibodies were immunoblotted for Ub. (Bottom) Expression of FLAG-Ufd1 in total cell extracts.

Given that ubiquitination of Skp2 is required for its proteasomal degradation, its down-regulation in Ufd1-KD cells might result from enhanced activation of APC/C^Cdh1. To test this, we compared the levels of Skp2 ubiquitination in control and Ufd1-KD cells. We found enhanced ubiquitination of Skp2 in Ufd1-KD cells (Fig. 2D), which could be rescued by reexpression of the full-length Ufd1 (Fig. 2E) or depletion of Cdh1 from Ufd1-KD cells (Fig. S2C). These findings suggest that Ufd1 stabilizes Skp2 by antagonizing the ubiquitination of Skp2 by APC/C^Cdh1.

Ufd1 Acts as a Scaffold for Skp2–USP13 Interaction.

We explored whether Ufd1 protects Skp2 from ubiquitin-dependent degradation through recruitment of a DUB that counteracts Skp2 ubiquitination. Consistent with data from a recent proteomic survey of interaction partners of human DUBs (18), we observed robust in vivo binding between Ufd1 and USP13 (Fig. 3A, lane 2). Using a series of Ufd1 truncation mutants, we found that amino acids 261–280 of Ufd1 were required for the in vivo binding of Ufd1 to USP13 (Fig. 3A, lanes 2–8), which was further confirmed by the inability of the Ufd1 mutant containing the internal deletion Δ261–280 to interact with USP13 in vivo (Fig. S3A). Notably, the N241-Ufd1 mutant containing the p97 binding site (19) interacts with p97 but not with USP13, indicating that Ufd1 has distinct binding sites for both proteins (Fig. 3A, lane 5). We then sought to determine whether interaction with USP13 enables Ufd1 to protect Skp2 from ubiquitination by testing the ability of the mutant Ufd1-Δ261–280 to rescue the low protein expression and enhanced ubiquitination of Skp2 in Ufd1-KD cells. Ufd1-Δ261–280 was unable to rescue the reduced levels of endogenous Skp2 (Fig. 3B) or the enhanced ubiquitination of Skp2 in Ufd1-KD cells (Fig. 3C), indicating that Ufd1–USP13 interaction is important in controlling Skp2 stability.

Fig. 3.

Ufd1 interferes with the ubiquitination of Skp2 by recruiting the deubiquitinating enzyme USP13. (A) Mapping of the Ufd1 binding site to USP13 in vivo. USP13 was expressed with empty vector (lane 1) or with the indicated FLAG-Ufd1 C-terminal truncation mutants in HEK-293T cells (lanes 2–8). Ufd1-N215 denotes an Ufd1 truncation mutant containing the N-terminal amino acids 1–215, and so on. FLAG-Ufd1 immunoprecipitates were immunoblotted for USP13, p97, and FLAG. (B) Western blots of endogenous Skp2 in control (cont) and Ufd1-KD cells transfected with empty vector, FLAG-Ufd1, or mutant FLAG-Ufd1Δ261–280. (C) In vivo ubiquitination of Skp2 in control and Ufd1-KD cells overexpressing myc-Skp2 and HA-Ub with empty vector, FLAG-Ufd1, or FLAG-Ufd1Δ261–280. c-myc immunoprecipitates were immunoblotted for Ub. Levels of ubiquitination were quantified and are shown as AUs. (D) Mapping of the Ufd1 binding site to Skp2 in vivo. myc-Skp2 was expressed in HEK-293T cells with empty vector (lane 1) or the indicated FLAG-Ufd1 C-terminal truncation mutants (lanes 2–8). FLAG immunoprecipitates were immunoblotted for Skp2 and FLAG. (E) Coimmunoprecipitation of USP13 and Skp2 in vivo. USP13 was expressed with empty vector (lane 1) or together with myc-Skp2 (lanes 2 and 3). c-myc immunoprecipitates were immunoblotted for USP13 or Skp2. (F) Semiendogenous coimmunoprecipitation. HeLa cells were transfected with FLAG-Skp2, and FLAG immunoprecipitates were immunoblotted for endogenous USP13 and Ufd1. *Short exposure; **long exposure. (G) Evidence of ternary complex formation among USP13, Ufd1, and Skp2 by sequential immunoprecipitations. HeLa cells were transfected with FLAG-Ufd1, myc-Skp2, and USP13. FLAG immunoprecipitates were eluted with FLAG peptide. Eluates were then used for c-myc immunoprecipitation. Then 10% of FLAG-IP, 10% of eluate, and all of myc-IP were immunoblotted for USP13, Skp2, and Ufd1. (H) (Left) Immunoblots of endogenous Ufd1 in control and Ufd1-KD cells. (Middle) USP13 and myc-Skp2 were coexpressed in control cells (lanes 1 and 2) or Ufd1-KD cells (lane 3). Control cells were transfected with either half (*) or an equal amount (**) of the DNA used for transfection in Ufd1-KD cells. c-myc immunoprecipitates were immunoblotted for USP13. (Right) Control and Ufd1-KD cells were transfected with USP13 and myc-Skp2 together with empty vector (ev), FLAG-Ufd1, or the mutant FLAG-Ufd1Δ261–280. c-myc immunoprecipitates were immunoblotted for USP13 and Skp2.

If Ufd1 facilitates interaction between USP13 and Skp2, then it should be able to bind Skp2. Indeed, such an interaction was observed in vivo (Fig. 3D, lane 2). In our effort to map the domain of Ufd1 required for the binding of Ufd1 to Skp2, we again identified amino acids 261–280, which also are required for the binding of Ufd1 to USP13 (Fig. 3D, lanes 2–8). A schematic summarizing the ability of the Ufd1 variants to bind USP13, Skp2, and p97 is presented in Fig. S3B. In addition, Ufd1 interacts with free Skp2 proteins (i.e., those not in complex with Skp1-Cul1), as demonstrated by a lack of endogenous Cul1 in complex with Ufd1-Skp2 in vivo (Fig. S3C, Left) and the binding of Ufd1 with equal affinity to both Skp2 and the Skp2ΔF-box mutant, which cannot be incorporated into SCF complexes (12) (Fig. S3C, Right). In support of our hypothesis that Ufd1-dependent regulation of Skp2 is mediated by USP13, we also observed in vivo interaction between USP13 and Skp2 (Fig. 3E) and among these three proteins (Fig. 3 F and G), suggesting the formation of a functional ternary complex in cells. Indeed, we confirmed the existence of a complex containing Ufd1, Skp2, and USP13 in vivo by sequential immunoprecipitations (Fig. 3G).

Next, to assess the role of Ufd1 as a mediator of Skp2–USP13 interaction, we compared the binding of USP13 to Skp2 in control and Ufd1-KD cells. A decreased amount of Skp2-bound USP13 was found in Ufd1-KD cells (Fig. 3H, Middle). Moreover, although exogenous Ufd1 can restore the interaction between Skp2 and USP13 in Ufd1-KD cells, the ability of the mutant Ufd1-Δ261–280 to do so was attenuated (Fig. 3H, Right). Taken together, our data suggest that Ufd1 acts as a scaffolding protein that enables a functional interaction between USP13 and Skp2.

USP13 Deubiquitinates Skp2.

Although USP13 shares sequence and structural homology with the ubiquitin protease USP5, it has been reported only as an ISG15-reactive protease (20). To directly assess whether USP13 has DUB activity, we performed deubiquitination assays using K48-linked di-ubiquitins as substrates. Immunopurified USP13 (Fig. S4A) elicited a steady cleavage of K48-linked di-ubiquitins (Fig. S4B).

We next evaluated the effect of USP13 on Skp2. We found that USP13 overexpression increased the level of endogenous Skp2 protein (Fig. 4A). In agreement, USP13 knockdown reduced the level of endogenous Skp2 and consequently increased p27 expression (Fig. S4C and Fig. 4B), supporting a positive role of USP13 in the regulation of Skp2 protein levels. To directly test whether USP13 exhibits DUB activity toward Skp2, we incubated FLAG-USP13 with ubiquitinated-Skp2, both immunopurified from cells. The amount of ubiquitin conjugates on Skp2 was reduced in the presence of USP13 (Fig. 4C). Taken together, our data show that USP13 has DUB activity toward both K48-linked di-ubiquitin and ubiquitinated Skp2.

Fig. 4.

USP13 controls Skp2 levels via deubiquitination. (A) Immunoblots of endogenous Skp2 and exogenous USP13 in HeLa cells transfected with empty vector or untagged USP13. (B) Knockdown of USP13 in HeLa cells infected with lentivirus packaged with either empty vector (mock) or one of three individual USP13-specific shRNA vectors, as indicated. Immunoblots of endogenous USP13, Skp2, and p27 are shown. (C) In vitro deubiquitination of Skp2 by USP13 as described in Materials and Methods. Immunopurified FLAG-USP13 and immunoprecipitated myc-Skp2 were obtained from HeLa cells. Reactions were immunoblotted for Ub and Skp2.

ER Stress Regulates the Ufd1-Skp2-p27 Axis and G1 Cell Cycle Progression.

The observation that Ufd1 knockdown induced Skp2 down-regulation and consequently p27 up-regulation under nonstressed conditions (Fig. 1 B and D) led us to examine whether ER stress-dependent Ufd1 down-regulation (Fig. 1A) would result in similar biochemical changes modulating cell cycle progression. Indeed, we observed that TM triggered Ufd1 down-regulation in a dose-dependent manner that was correlated with Skp2 clearance and p27 accumulation after 20 h of treatment (Fig. 5A). ER stress-dependent regulation of Ufd1-Skp2-p27 was observed in other cell lines as well, including HeLa-S3, HEK-293T, and HFF-1 (Fig. S5A), although the degree of response differed among different cell lines subjected to the same TM treatment. Thapsigargin, an inhibitor of ER Ca²⁺ ATPase that also induces ER stress, elicited similar responses (Fig. S5B). Remarkably, the extent of the G1 delay was correlated with the dose- and time-dependent regulation of Ufd1, Skp2, and p27 levels after TM treatment (Fig. 5B and Fig. S5C).

Fig. 5.

ER stress regulates the Ufd1-Cdh1-Skp2-p27 axis to delay progression through G1. (A) HeLa cells were treated with DMSO or increasing concentrations of TM (0.05, 0.1, 0.25, and 0.5 μg/mL) for 20 h. Lysates were immunoblotted for Ufd1, Skp2, p27, and GRP78 (a marker of UPR). (B) Cells were treated with DMSO or with 0.5 μg/mL or 2.5 μg/mL of TM over a 24-h course. At the indicated time points, samples were collected for immunoblot and FACS analyses. Fig. S5C shows the corresponding quantification of G1 delay based on FACS analysis. (C) (Left) HeLa cells were treated with DMSO or with 2.5 μg/mL of TM alone or together with MG-132 (2 or 5 μM) for 12 h, and then collected for immunoblot analysis for the indicated proteins. (Right) HeLa cells transfected with control or Cdh1-specific shRNA for 24 h were treated with DMSO or with 2.5 μg/mL of TM alone or together with MG-132 (5, 10, or 15 μM) for 12 h. Total cell extracts were immunoblotted for the indicated proteins. (D) HeLa cells transfected with empty vector or FLAG-Ufd1 were treated with DMSO or TM (1 μg/mL for 8 h). (Upper) Western blot of Ufd1. (Bottom) Increase in G1 at 8 h after the addition of TM compared with DMSO treatment. Fig. S6C shows FACS histograms. (E) HeLa cells were transfected with either control or two different p27-specific shRNAs, as indicated, for 24 h. (Upper) Western blot analysis of p27 in control and p27-KD cells treated with TM (2.5 μg/mL for 8 h). (Bottom) Percent increase in G1 at 4 h and 8 h after TM addition relative to 0 h. The percentage of cells in G1 at 0 h is set as the baseline, with a 0% increase in G1. (F) Half-life analyses of GFP-CFTR and NHK proteins with CHX treatment in double thymidine-arrested (G1/S) or nocodazole-arrested (G2/M) HeLa cells. Total lysates were immunoblotted for GFP or α1-antitrypsin/NHK. Fig. S8A verifies cell cycle synchronization by FACS, and Fig. S8B quantifies protein half-lives. (G) Half-life of NHK protein in double thymidine-arrested G1/S HeLa cells or cells released from double-thymine arrest for 7 h (approximately corresponding to G2/M). Total lysates were immunoblotted for α1-antitrypsin/NHK. Fig S8C presents verification of cell cycle synchronization by FACS, and Fig. S8D shows quantification of NHK half-life. (H) In the proposed model, down-regulation of Ufd1 after prolonged ER stress reduces recruitment of USP13 to Skp2, thereby resulting in ubiquitin-dependent degradation of Skp2 by APC/C^Cdh1. Consequently, levels of p27 increase, contributing to G1 arrest that supports degradation of misfolded proteins.

ER stress-dependent Ufd1 down-regulation is a response to prolonged UPR activation (Figs. 1A and 5 A and B). Notably, changes in the expression of Ufd1 and Skp2 occur mainly in the nuclear fraction (Fig. S5D), suggesting that the pool of Ufd1 contributing to cell cycle control is distinct from that involved in the retrotranslocation of misfolded proteins in the cytosol. Consistent with this idea are observations that both USP13 (20) and Skp2 (Fig. S5D) are localized mainly in the nucleus.

Cdh1-Skp2-p27 Axis Contributes to ER-Induced G1 Delay.

Under nonstressed conditions, Ufd1 knockdown destabilized the Skp2 protein. To examine whether Skp2 also is regulated at the level of protein stability under ER stress conditions that down-regulate Ufd1, we first treated cells with TM and the proteasome inhibitor MG-132. The addition of MG-132 prevented TM-induced down-regulation of Skp2, indicating that ER stress accelerates proteasomal degradation of Skp2 (Fig. 5C). Interestingly, treatment with a low dose of MG-132 (2 μM) for 12 h stabilized Skp2 and abolished p27 accumulation after TM treatment, suggesting that ER stress-dependent p27 accumulation is a consequence of Skp2 destabilization, and that the coregulation of Skp2 and p27 in response to TM occurs at the level of protein stability (Fig. 5C, Left, lane 3 and Fig. S6A; Fig. 5C, Right, lanes 3 and 7).

We observed that although MG-132 attenuated TM-induced Skp2 down-regulation, the effect was nevertheless incomplete. Indeed, analysis of Skp2 and p27 transcript levels after TM treatment revealed a 30% decrease in Skp2 mRNA and a 1.6-fold increase in p27 mRNA (Fig. S6B). These data suggest ER stress-dependent transcriptional regulation of Skp2 and p27. In contrast, Ufd1 mRNA levels were comparable before and after TM treatment (Fig. S6B). Further, TM-dependent Ufd1 down-regulation was not restored by MG-132 (Fig. 5C, Left), suggesting translation or proteasome-independent degradation.

Next, to determine whether APC/C^Cdh1 targets Skp2 for ubiquitin-mediated degradation in response to ER stress, we knocked-down Cdh1 by RNAi, followed by treatment with TM. Notably, Cdh1-RNAi alone attenuated TM-induced proteasomal degradation of Skp2 and diminished TM-induced up-regulation of p27 (Fig. S6A, lane 10), demonstrating that APC/C^Cdh1 targets Skp2 for degradation during ER stress, which in turn leads to p27 accumulation. Cdh1 knockdown incompletely prevented TM-induced Skp2 down-regulation (Fig. 5C and Fig. S6A) because TM treatment also transcriptionally repressed Skp2 (Fig. S6B). Compared with the addition of MG-132 alone, the combination of Cdh1-RNAi and MG-132 did not further stabilize Skp2 under ER stress, indicating that APC/C^Cdh1 is the primary E3 ligase responsible for targeting Skp2 for degradation under such conditions (Fig. 5C, Right, lanes 11–15). Taken together, our data show that ER stress directly regulates the Cdh1-Skp2-p27 axis through Ufd1 to delay cell cycle progression. Finally, the ability of Ufd1 overexpression or p27 knockdown to overcome TM-induced G1 delay, although only partial, further supports a role of these proteins in regulating the UPR-induced cell cycle response (Fig. 5 D and E and Fig. S6C).

Because it was previously reported that cyclin D1 down-regulation also can mediate TM-dependent G1 delay in NIH 3T3 cells (21, 22), we sought to analyze the contribution of cyclin D1 down-regulation and p27 up-regulation to ER stress-induced cell cycle arrest. Interestingly in HeLa cells, overexpression of cyclin D1 could not overcome TM-induced G1 delay (Fig. S7). Of note, Skp2 expression was still reduced regardless of cyclin D1 overexpression, indicating that Skp2 and cyclin D1 down-regulation after TM treatment are independent. This points to the existence of complementary mechanisms in ER stress-dependent cell cycle arrest, which may be cell type–specific. Our data suggest that in the absence of cyclin D1–driven arrest, Ufd1-Skp2-p27 contributes to cell cycle delay.

G1 Cell Cycle Phase Facilitates Clearance of ERAD Substrates.

Given the link between ER stress and cell cycle, we next asked whether ER stress-induced G1 delay affects ERAD. Toward this end, we examined the efficiency of ERAD in different phases of the cell cycle by comparing the rate of degradation of two classic ERAD substrates—cystic fibrosis transmembrane conductance regulator (CFTR) and ER luminal protein null Hong-Kong α1-antitrypsin (NHK)—in G1/S- and G2/M-arrested cells. Notably, accelerated degradation of both proteins was observed in G1/S-arrested cells compared with the G2/M-synchronized cells (Fig. 5F and Fig. S8 A and B). The impaired clearance of ERAD substrates seen in G2/M-arrested cells did not result from nocodazole-induced toxicity, as the degradation of NHK was still slower in the G2/M population obtained after a 7-h release from G1 arrest (Fig. 5G and Fig. S8 C and D). These data suggest that ER stress-induced G1 delay serves to facilitate degradation of misfolded proteins.

This hypothesis prompted us to examine the cell cycle status of fibroblasts bearing two alleles of the mutant CFTR gene (CFTR-ΔF508) isolated from a cystic fibrosis–affected individual. If G1 phase of the cell cycle is more conducive to ERAD, then the constitutive expression of endogenous mutant CFTR targeted to ERAD in these cells may render them more prone to G1 delay. We found that these cells not only had a slightly higher basal G1 population compared with normal human fibroblasts (Fig. S9A), but also had a greater increase in G1 when treated with TM (Fig. S9B). These observations further support the physiological relevance of ER stress-induced cell cycle arrest.

Discussion

The present study provides insight into cell cycle control in response to ER stress. We have demonstrated that ER stress-dependent control of Ufd1 modulates Skp2 protein expression and consequently p27, to delay progression through G1 and facilitate the clearance of misfolded proteins. Mechanistically, Ufd1 promotes the interaction between USP13 and Skp2 to maintain steady-state Skp2 levels. Fig. 5H presents a proposed model of this process.

In support of our findings, TM treatment has been reported to trigger G1 cell cycle arrest by inhibiting cyclin D1 translation through the activation of PKR-like endoplasmic reticulum kinase (PERK) (22). Our findings reveal a cyclin D1-independent mechanism for attaining G1 delay after protein misfolding through regulation of the Cdh1-Skp2-p27 axis at the level of protein stability. We have begun to examine the efficiency of ERAD in the context of the cell cycle, a currently underexplored area. Our data raise the intriguing possibility that cell cycle arrest is not simply a passive response that provides cells time to restore homeostasis, but creates conditions conducive to the degradation of misfolded proteins. Considering that CFTR and NHK are characterized ERAD substrates subjected to degradation by the ER resident E3 ligases gp78, RNF5/RMA1, and Hrd1 (23, 24), the cell cycle–dependent degradation of ERAD substrates we observed may be the result of cell cycle–regulated expression or activity of these ligases and other ERAD components.

We also have presented an initial characterization of USP13 as a di-ubiquitin able to process K48-linked ubiquitins and antagonize APC/C^Cdh1-mediated ubiquitination of Skp2 in an Ufd1-dependent manner. Our mapping of Ufd1-USP13 interaction sites shows that Ufd1 binds p97 and USP13 through distinct domains. Our observation that Ufd1 binds Skp2 proteins not in complex with Skp1-Cul1 suggests that Ufd1 positively regulates the abundance of free Skp2 proteins to control the formation of SCF^Skp2 complexes.

Based on our findings, we propose a model in which the functions of Ufd1 are regulated temporally and spatially under conditions activating the UPR. In the immediate phase, Ufd1 in cooperation with p97 and Npl4 in the cytosol contribute to the retrotranslocation of misfolded proteins from the ER, whereas in a delayed response, down-regulation of nuclear Ufd1 mediates G1 cell cycle delay. Both responses serve to clear misfolded proteins. Such temporal and spatial modulation of Ufd1 expression shown here could not have been detected in previous yeast studies using loss-of-function mutants that yielded all-or-none phenotypes (3, 5).

The identification of Ufd1-dependent recruitment of USP13 to control Skp2 stability and G1 delay provides mechanistic insights into cell cycle regulation under ER stress while also pointing to the implications of a cell cycle–dependent control of ERAD.

Materials and Methods

Protocols for the immunoblotting, immunoprecipitations, cell cycle synchronization, siRNA/shRNA knockdown, qRT-PCR, assays for measuring kinase, degradation, ubiquitination and deubiquitination, cell fractionation and the details of cell lines and expression plasmids used in this study can be found in SI Materials and Methods.