Calreticulin and Arginylated Calreticulin Have Different Susceptibilities to Proteasomal Degradation*

Background: Calreticulin (CRT) retrotranslocated from the endoplasmic reticulum to the cytoplasm is post-translationally arginylated. Results: Arginylated CRT (R-CRT) has a longer half-life than non-arginylated CRT and displays different ubiquitin dependence. Conclusion: CRT and R-CRT differ in their susceptibility to proteasomal degradation. Significance: Arginylation is not required for proteasomal degradation of CRT, although R-CRT displays ubiquitin modification. Post-translational arginylation has been suggested to target proteins for proteasomal degradation. The degradation mechanism for arginylated calreticulin (R-CRT) localized in the cytoplasm is unknown. To evaluate the effect of arginylation on CRT stability, we examined the metabolic fates and degradation mechanisms of cytoplasmic CRT and R-CRT in NIH 3T3 and CHO cells. Both CRT isoforms were found to be proteasomal substrates, but the half-life of R-CRT (2 h) was longer than that of cytoplasmic CRT (0.7 h). Arginylation was not required for proteasomal degradation of CRT, although R-CRT displays ubiquitin modification. A CRT mutant incapable of dimerization showed reduced metabolic stability of R-CRT, indicating that R-CRT dimerization may protect it from proteasomal degradation. Our findings, taken together, demonstrate a novel function of arginylation: increasing the half-life of CRT in cytoplasm.

Calreticulin (CRT) is localized mainly in the endoplasmic reticulum (ER) and has multiple functions within and outside the ER (14 -16). Our group was the first to demonstrate the retrotranslocation of CRT from the ER (ER-CRT) to the cytoplasm under conditions that promote reduction of cytoplasmic Ca 2ϩ levels (7,12). In the cytoplasm, CRT is post-translationally arginylated in vivo by Ate1 (7,12). Arginylated CRT (R-CRT) is not detectable in the ER lumen (7), consistently with the cytosolic and nuclear localization of Ate1 (17). We showed that arginylation of CRT is essential for its association with stress granules (SGs) and promotes its dimerization (7,13). Apoptosis induction is associated with an increased R-CRT level at the cell surface (11). Surface exposure of CRT on tumor cells has been suggested to promote their uptake by phagocytes (18). The effects of CRT arginylation on its cytoplasmic functions and final destination remain to be elucidated.
According to the "N-end rule," which relates the half-life of a protein to the identity of its N-terminal residue, N-terminal Asp and Glu are secondary destabilizing residues that function through their arginylation to yield the primary destabilizing residue Arg (19). Arginylation in turn promotes protein recognition by specific E3 ubiquitin (Ub)-protein ligases that polyubiquitinate proteins on internal Lys residues (20). Proteasomal degradation depends in most cases on Ub conjugation (ubiquitination). However, a significant subset of proteins displays ubiquitination-independent turnover (21), which is not surprising in view of the co-existence of several types of proteasomal complexes in eukaryotic cells (22). The possible role of CRT as a substrate for proteasomal degradation is controversial (23)(24)(25). The degradation mechanism of R-CRT remains unknown.
To assess the effect of CRT arginylation on its stability, we studied the degradation of cytoplasmic CRT (cyt-CRT) and R-CRT in fibroblasts and CHO cells, including the roles of proteasomes, Ub modification, and dimer formation in this process. We found that arginylated and non-arginylated isoforms * This work was supported by grants from the Agencia Nacional de Promoción Científica y Technologica (PICT 2012(PICT -1979, Consejo Nacional de Investigaciones Científicas y Técnicas, and La Secretaría de Ciencia y Tecnología, Universidad Nacional de Córdoba. The authors declare that they have no conflicts of interest with the contents of this article. 1 Recipient of a fellowship from Consejo Nacional de Investigaciones Científicas y Técnicas. 2  of CRT are proteasomal substrates that follow different degradation pathways: one dependent on and the other independent of Ub modification. Biochemical analysis showed that CRT arginylation leads to ubiquitination of R-CRT isoforms but reduces turnover rate. Our finding that arginylation stabilizes CRT is in contrast to the traditional view of arginylation as a destabilizing factor.
Plasmids-For cloning of pEGFP-CRT (with signal peptide) we digested pEYFP-CRT plasmid (12) with EcoRI and KpnI endonucleases and cloned the fragment containing the coding sequence of human full-length CRT into pEGFP-N1 expression vector (Clontech Laboratories; Palo Alto, CA). To express mature CRT (without signal peptide) in cytoplasm, we adapted a strategy described previously for GFP (26). In brief, DNAs encoding Ub fused to mature human CRT (Ub-CRT) or R-CRT (Ub-R-CRT) were cloned in pEGFP-N1 expression vector. The Ub moiety was cleaved in cytoplasm by Ub-C-terminal hydrolases to release mature CRT with exposed N-terminal Glu or Arg, as confirmed by Western blotting (Fig. 3). For cloning of pEGFP-Ub-CRT and pEGFP-Ub-R-CRT cDNAs, the coding sequences of Ub-CRT and R-CRT were amplified by PCR from pHUE-CRT-FLAG and pHUE-R-CRT-FLAG expression vectors, respectively (11). The forward primer (same for both amplification reactions) contained an EcoRI restriction site (underlined), a Kozak sequence (boldfaced), and the Ub sequence (italicized): 5Ј-CCGGAATTCCCCGCCGCCACC-ATGCAGATCTTTGTGAAGACCCTCACTGG-3Ј. The reverse primer (the same for both amplification reactions) contained a KpnI restriction site (underlined) and the CRT sequence (italicized): 5Ј-CGGGGTACCCCCAGCTCGTCCTTGGCCTG GCC-3Ј. The PCR products were digested with EcoRI and KpnI endonucleases and cloned into pEGFP-N1 expression vector. All constructs were confirmed by DNA sequencing.
Mutant Construction and Site-directed Mutagenesis-For generation of the pEGFP-C146A-CRT mutant, we digested the pEYFP-C146A-CRT plasmid (13) with EcoRI and KpnI endonucleases and cloned the fragment containing the coding sequence of human full-length C146A-CRT into pEGFP-N1 expression vector.
For cloning of the pEGFP-Ub-E1V-CRT mutant, we used the site-specific mutagenesis method with overlap extension (27). Construction of the E1V mutation (numbering starts from the mature CRT) required two mutagenic primers, two flanking oligonucleotides, and three PCR. One pair of primers was used to amplify the DNA containing the mutation site together with downstream sequences. The reverse primer contained the mutation (underlined and boldfaced) introduced into the wildtype (WT) template cDNA of Ub-CRT: 5Ј-AAGTAGACGGC-AGGCACTCCACCGCGGAGGCGCAACACC-3Ј. The forward primer (F2) was the same as for the Ub-CRT WT sequence described under "Plasmids" (above) and contained an EcoRI restriction site, a Kozak sequence, and the Ub sequence. The second pair of primers was used to amplify the DNA that contained the mutation site together with upstream sequences. The forward primer of this pair contained the mutation (underlined and boldfaced) introduced into the template DNA and was complementary to reverse primer: 5Ј-GCCTCCGCGGT-GGAGTGCCTGCCGTCTACTTCAAGGAGC-3Ј. The reverse primer (R2) was the same as for the Ub-CRT WT sequence described under "Plasmids" (above) and contained a KpnI restriction site and the CRT sequence. To amplify the overlapping DNA fragments, both sets of primers were used in two separate amplification reactions. The mutation of interest was located in the overlap region and, therefore, in both sets of the resulting amplified fragments. The full-length Ub-E1V-CRT DNA was amplified by a third PCR, mixing the overlapping fragments and using primers F2 and R2 that bind to the 3Ј and 5Ј extremes of the initial fragments, respectively. The PCR products were digested with EcoRI and KpnI endonucleases and cloned into pEGFP-N1 expression vector. All constructs were confirmed by DNA sequencing.
Establishment of Stable Cell Lines-CHO cells were transfected with pEGFP-Ub-CRT, pEGFP-Ub-R-CRT, or pEGFP-Ub-E1V-CRT. Stably transfected cell lines were selected in the presence of 0.5 mg/ml G418 (Phytotechnology Laboratories; Overland Park, KS). To obtain stable cell lines that uniformly expressed CRT-EGFP, R-CRT-EGFP, or E1V-CRT-EGFP, we isolated populations with intermediate levels of EGFP expression by fluorescence-activated cell sorting (FACS Aria IIu; BD Biosciences) after treatment with MG132 (Sigma).
Immunofluorescence-Cells were grown on glass coverslips to 60% confluence, and immunofluorescence was stimulated as described previously (7,13). Nuclei were stained with DAPI.
Protein Half-life Assay-Stable CHO cell lines expressing CRT-EGFP, R-CRT-EGFP, or E1V-CRT-EGFP were treated for 9 h with MG132 to cause accumulation of EGFP fusion proteins. Alternatively, ATE1 Ϫ/Ϫ cells were transfected transiently with pEGFP-Ub-CRT or pEGFP-Ub-R-CRT, cultured for 6 h, and treated with MG132 for the indicated times in the Figs. 8 and 9 legends. In both cases cells were washed 4 times, cultured for 1 h to allow diffusion of the inhibitor, and washed again. CHO cells were incubated for various durations with CHX or MG132. Proteins were separated by SDS-PAGE, and levels of EGFP fusion proteins and R-CRT were monitored by Western blotting.
In Vivo Ubiquitination Assay-To increase Ub levels and the sensitivity of Ub detection, CHO cells were transfected transiently with pcDNA3-HA-Ub or co-transfected with pcDNA3-HA-Ub and pEGFP-Ub-R-CRT or pEGFP-Ub-E1V-CRT. Six hours after transfection, cells were cultured with proteasome inhibitor for 16 -18 h. Cell lysates were prepared using lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Nonidet P-40, 0.1% deoxycholate) plus 5 mM iodoacetamide, 5 mM N-ethylmaleimide, 10 M MG132, and 1:100 protease inhibitor mixture (Calbiochem) to inhibit deubiquitinating enzymes and proteases and then sonicated briefly on ice. The remaining steps were performed at 4°C. Lysates were precleared by centrifugation at 13,000 ϫ g for 10 min, and the supernatants were subjected to immunoprecipitation for EGFP. The protein samples were separated by SDS-PAGE and immunoblotting with anti-HA and anti-GFP antibodies (Abs).
Immunoprecipitation-All steps were performed at 4°C. Cell lysates were incubated on an orbital shaker with 50 l of protein G-Sepharose beads (GE Healthcare) for 2 h to reduce nonspecific protein binding. Beads were pelleted for 20 s at 2300 rpm. To immunoprecipitate EGFP fusion proteins, supernatants were incubated with 2 l of anti-GFP for 1 h, added with 50 l of protein G-Sepharose beads, and incubated on a rocker for 4 h. Protein G-Sepharose⅐Ab complexes were pelleted at 2300 rpm for 20 s and washed twice in 500 l of lysis buffer and once in washing buffer (50 mM Tris-HCl, pH 7.5, 250 mM NaCl, 0.1% Nonidet P-40, 0.05% deoxycholate). Protein G-Sepharose⅐Ab complexes were resuspended in sample buffer 2ϫ. Total cell lysate and bound and unbound fractions were subjected to SDS-PAGE, and proteins were transferred to nitrocellulose membrane for immunoblotting.
Cell Viability Assay-Cell viability was determined by the colorimetric technique using resazurin (28) (Sigma). In each assay, 10,000 cells were plated per well in 96-well plates. Cells were treated at various intervals (see Fig. 6A) with 10 M MG132 or vehicle. After this treatment, 2 l of 1 mg/ml resazurin solution were added to 200 ml of medium per well. Plates were incubated at 37°C for 4 h, and the optical density was measured at 600 nm wavelength in a FL X800 Microplate Reader (BioTek Instruments, Inc.) Assessment of Cell Death-Cells/sample were collected as in López Sambrooks et al. (11). Briefly, cells were washed in PBS, pelleted, and resuspended in buffer (10 mM HEPES/NaOH, pH 7.4, 140 mM NaCl, 2.5 mM CaCl 2 ) containing propidium iodide to identify dead cells. Cells were analyzed by flow cytometry on a FACSCantoII cytometer (BD Biosciences) using FlowJo software.
Imaging and Image Analysis-Images were acquired on a laser scanning confocal microscope (FluoView 1000; Olympus; Center Valley, PA) with a 60ϫ oil immersion lens, NA 1.42. At least 200 cells were examined for each condition. Fiji software was used to measure cell fluorescence and calculate the corrected total cell fluorescence with correction for background fluorescence. Blot images acquired from the Odyssey IR-imaging system were analyzed using Odyssey Application Software v.2.1 (LI-COR).
Statistical Analysis-Significance was tested by one-way or two-way analysis of variance using the GraphPad Prism 5.01 program (GraphPad Software). One-way and two-way analysis of variance was followed, respectively, by the Newman-Keuls and Bonferroni multiple comparison test. Differences were considered significant for p Ͻ 0.05. Half-lives were calculated using nonlinear regression and one-phase decay analysis.

R-CRT Accumulation Results from Proteasomal Inhibition-
Arginylation is considered a primary destabilizing residue according to the N-end rule for targeting proteins to proteasomal degradation (19). To evaluate the role of arginylation in removal of intracellular R-CRT, we incubated NIH 3T3 cells in the presence of a potent reversible proteasome inhibitor (MG132) and measured the cellular content of endogenous cytoplasmic R-CRT by immunofluorescence at intervals using anti-R-CRT Ab. This antibody specifically recognizes endogenous R-CRT (7). After 1 h of MG132 treatment, the R-CRT level was twice as high as that of control (non-treated) cells (p Ͻ 0.001) (Fig. 1, A and B). After 2 h of treatment, the level was slightly (p Ͻ 0.05) higher than that at 1 h (Fig. 1, A and B). There was a possibility that these findings resulted from inhibition of cysteine protease by MG132 (31). We, therefore, performed similar studies with lactacystin, an irreversible specific proteasome inhibitor that does not affect cysteine proteases (32), and obtained similar results (Fig. 2).
MG132 inhibits proteasomal degradation and also leads to a heat-shock response that induces expression of ER chaperones (33). Whether MG132 promotes retrotranslocation of ER proteins is unknown. The increase of R-CRT after proteasome inhibition could, therefore, be due to either increased CRT synthesis or retrotranslocation. Our subsequent experiments were designed to distinguish between these two possibilities.
We showed previously that R-CRT content increases when cells are treated with an ER stressor (Tg) (7,12). To determine whether such an R-CRT increase is affected by proteasomal inhibition, NIH-3T3 cells were treated with Tg in the presence and absence of MG132. R-CRT level in the Tg-treated cells was increased to a similar degree regardless of MG132 treatment (Fig. 1, C and D). However, when Tg was removed after 20 min of treatment, the increased R-CRT level was maintained only in cells treated with MG132 for 40 or 60 min (Fig. 1, C and D). The R-CRT level resulting from treatment with Tg alone was not increased by the addition of MG132 (Fig. 1D: 20Ј Tg, white versus black bar), indicating that MG132 did not activate CRT retrotranslocation. The sustained increase of R-CRT in the presence of MG132 after Tg removal was, therefore, due to impaired proteasomal degradation. Because cyt-CRT accumulation results from retrotranslocation of stored ER-CRT, it is not possible to evaluate CRT degradation by using a protein synthesis inhibitor to block CRT synthesis. Taken together, these findings suggest that CRT is arginylated after retrotranslocation from the ER to the cytoplasm (7,12), and R-CRT is then removed by the proteasomal degradation system.
Cyt-CRT Is Degraded Rapidly by Proteasomes in an Arginylation-independent Manner-To determine whether arginylation of CRT is required for its proteasomal degradation, we evaluated CRT degradation under conditions that precluded N-terminal arginylation and analyzed proteasomal degradation of CRT, R-CRT, and a non-arginylatable CRT (E1V-CRT) expressed directly in the cytoplasm. Stable CHO cell lines expressing three chimeric proteins (Ub-CRT-EGFP, Ub-R-CRT-EGFP, Ub-E1V-CRT-EGFP) were used. Cleavage of the Ub moiety by cellular C-terminal hydrolases was confirmed by Western blotting (Fig. 3). Under immunofluorescence micros- , not significant. Cells were analyzed by double immunofluorescence using anti-R-CRT and SG marker anti-TIA-1 pAbs. Co-localization of proteins was monitored by confocal microscopy. SG formation was observed upon MG132 treatment, and R-CRT was found to be co-localized with SGs as shown by yellow pseudo-color in the merge images.  copy, stably transfected cells showed EGFP fluorescence levels slightly higher than the autofluorescence of nontransfected cells, most likely because of rapid degradation of the expressed proteins (Fig. 4, A-C). In cells treated with MG132, CRT-EGFP, R-CRT-EGFP, and E1V-CRT-EGFP, levels increased at different times (Fig. 4D). Immunocytochemical analysis (with an anti-R-CRT Ab) of R-CRT-EGFP-and CRT-EGFP-transfected cells treated with MG132 revealed similar accumulations of R-CRT (Fig. 4E). In contrast, R-CRT was not detected in E1V-CRT-EGFP-transfected cells, consistent with the presumed inability of N-terminal Val to incorporate Arg (Fig. 4, C-E). Although all three variants showed CRT accumulation in the presence of MG132, Western blotting with anti-GFP Ab revealed differences in their accumulation rates (Fig. 4F). An increase of R-CRT-EGFP expression was observed after 2 h of MG132 treatment, whereas cells expressing non-arginylatable E1V-CRT-EGFP or arginylatable CRT-EGFP required 4 and 6 h of MG132 treatment, respectively (Fig. 4F). In CRT-EGFP-expressing cells treated with MG132, protein accumulation was detected with anti-GFP at essentially the same time that R-CRT accumulation was detected by Western blotting with anti-R-CRT Ab (Fig. 4F). Taken together, these findings indicate that proteasome inhibition results in accumulation of R-CRT-EGFP, CRT-EGFP, and E1V-CRT-EGFP and suggested that (i) CRT arginylation is not required for its degradation, and (ii) R-CRT-EGFP and E1V-CRT-EGFP are less susceptible to proteasome degradation than is CRT-EGFP.
Effect of the N-terminal Amino Acid on CRT Half-life-The observed accumulation of CRT-EGFP, R-CRT-EGFP, and E1V-CRT-EGFP after MG132 treatment could be explained by differential rates of synthesis (34). To evaluate this possibility, we determined the half-life of each protein in the presence of the protein synthesis inhibitor CHX. Stably transfected cells as described in the preceding section were incubated with MG132 to accumulate the corresponding protein (CRT-EGFP, R-CRT-EGFP, E1V-CRT-EGFP), MG132 was removed, cells were incubated further with CHX, and EGFP clearances were determined at various times by Western blotting (Fig. 5A). Clearance was more pronounced for CRT-EGFP than for R-CRT-EGFP or E1V-CRT-EGFP (Fig. 5, A and C). No changes in clearance were observed when cells were incubated with CHX in the presence of MG132 (Fig. 5, B and D). Half-lives in the presence of CHX were ϳ1.3 h for the mixture of CRT-EGFP and R-CRT-EGFP present in CRT-EGFP cells, ϳ0.7 h for the non-arginylated CRT-EGFP, ϳ2.0 h for R-CRT-EGFP, and ϳ4.8 h for E1V-CRT-EGFP (Fig. 5E); i.e. the half-life of nonarginylated CRT-EGFP was 65% shorter than that of R-CRT-EGFP and 85% shorter than that of E1V-CRT-EGFP. The half-life of R-CRT from CRT-EGFP and R-CRT-EGFP, as determined using anti-R-CRT antibody, was ϳ2.0 h (Fig. 5F), a similar value than that obtained employing anti-GFP antibody (Fig. 5, C and E).
MG132 has been reported to induce apoptosis in some cases (35). It is, therefore, conceivable that this inhibitor affects CRT levels indirectly via some apoptosis-related mechanism.
To test this possibility, we cultured CHO cells expressing CRT-EGFP, R-CRT-EGFP, or E1V-CRT-EGFP in the presence of MG132 and analyzed cell viability by the colorimetric resazurin assay. No changes were observed in viability at least up to 24 h of MG132 treatment (Fig. 6A). Moreover, no significant increase in cell death was detected by flow cytometry of propidium iodidestained cells at the times studied (Fig. 6B).
These findings demonstrate that CRT-EGFP, R-CRT-EGFP, and E1V-CRT-EGFP undergo proteasomal degradation and that arginylation of CRT is not required for its proteasomal degradation. Because CRT-EGFP is arginylated by Ate1 and some CRT-EGFP remains arginylation-free (Fig. 7), the half-life of the mixture of non-arginylated CRT-EGFP and R-CRT- EGFP is presumed to lie between that of R-CRT-EGFP and E1V-CRT-EGFP. However, the degradation of the mixture of non-arginylated CRT-EGFP and R-CRT-EGFP is faster than that of R-CRT-EGFP or E1V-CRT-EGFP (Fig. 5E), indicating that non-arginylated CRT-EGFP should be degraded more efficiently.
Degradation of R-CRT and CRT in Cells Lacking Ate1-The mutation introduced into CRT for preparation of E1V-CRT may, in addition to preventing its arginylation, induce conformational changes in the protein. To test this possibility, we examined CRT-EGFP degradation in ATE1 Ϫ/Ϫ cells (9, 36). These cells were transfected with Ub-CRT-EGFP or Ub-R-CRT-EGFP and treated with MG132 for 18 h to allow cytoplasmic accumulation of CRT-EGFP or R-CRT-EGFP. MG132 was removed, cells were cultured with CHX for 2 h, and protein expression was analyzed by Western blotting. In agreement with the results shown in Fig. 5, CRT-EGFP and R-CRT-EGFP were decreased by treatment with CHX alone but unaffected by treatment with CHX plus MG132 (Fig. 8, A and B). The CHX-induced decrease was significantly smaller (Ͻ40%; p Ͻ 0.05) for R-CRT-EGFPversus CRT-EGFP-expressing cells (Fig. 8B). The simplest interpretation of these findings is that arginylation of CRT impairs its degradation and that different proteasomal pathways are involved in the degradation of CRT versus R-CRT. To rule out the possibility that EGFP contributes to the degradation of these variants, ATE1 Ϫ/Ϫ cells were transfected with Ub-CRT-FLAG or Ub-R-CRT-FLAG and treated with MG132 for 12 h to allow cytoplasmic accumulation of CRT-FLAG or R-CRT-FLAG. MG132 was removed, cells were cultured with CHX for 1 or 2 h, and protein level was analyzed by Western blotting. Results for proteins fused with FLAG versus EGFP at the C terminus of isoforms were similar (Fig. 9, A-D), indicating that the CRT sequence encloses the signal for degradation and that the signal is not affected by EGFP fusion at the C terminus.
R-CRT-EGFP Is Degraded by the Ub Proteasome System-The finding that arginylatable CRT-EGFP is degraded more rapidly than are R-CRT-EGFP or E1V-CRT-EGFP suggests that the isoforms are subject to different proteasomal mechanisms. We, therefore, examined the possible involvement of the Ub-proteasome system in these differences. CHO cells were cotransfected to express HA epitope-tagged Ub and either R-CRT or E1V-CRT, and ubiquitination status was examined. Six hours after co-transfection, cells were incubated with MG132 for 18 h to allow accumulation of Ub conjugates. R-CRT-EGFP and E1V-CRT-EGFP were then immunoprecipitated from lysed cells with anti-GFP Ab (Fig. 10A) and analyzed by Western blotting with anti-HA Ab. HA-Ub conjugates were detected only for R-CRT (Fig. 10B), suggesting that arginylation is necessary for an efficient ubiquitination of R-CRT and that nonarginylatable CRT is degraded via a Ub-independent proteasomal pathway.
Why Does Arginylation of CRT Not Follow the N-end Rule?-According to the N-end rule, Arg is a primary destabilizing residue that targets proteins for degradation by the Ub-protea- some system (19,20). However, we found that the half-life of CRT, which has a Glu residue at its N terminus, is shorter than that of R-CRT. What is the cause of the higher half-life of R-CRT? We demonstrated previously that R-CRT but not CRT is recruited to SGs and that R-CRT moves to the plasma membrane in response to stress, acting as a pro-apoptotic signal (11). Arginylation may, therefore, protect CRT from proteasomal degradation under stress conditions by altering its metabolic fate. The dimerization of R-CRT is of a higher magnitude than that of CRT. Disulfide-bridged dimerization is essential for scaffolding of larger SGs under heat shock (13).
The above findings suggest that dimerization of R-CRT is another way to avoid proteasomal degradation. To evaluate this possibility, we examined R-CRT accumulation in cells incubated in the absence versus presence of (i) MG132 and (ii) dimerization conditions. Because homodimers of CRT are formed under non-reducing conditions by exposure of free SH-Cys-146 (13,37,38), we generated a CRT mutant (C146A-CRT-EGFP) that cannot dimerize. CRT knock-out fibroblasts (CRT Ϫ/Ϫ ) were transfected with C146A-CRT-EGFP or WT CRT (WT-CRT-EGFP) that are directed to the ER. Transfected cells were analyzed by immunocytochemistry using anti-R-CRT Ab (Fig. 11). Cells transfected with C146A-CRT and incubated in the absence of MG132 showed R-CRT levels 25% lower than those of cells transfected with WT-CRT-EGFP (Fig. 11, A,  B, and E). R-CRT levels in MG132-treated cells were 2-fold higher than those in MG132-nontreated cells (Fig. 11E), regardless of the dimerization capability of the transfected protein. This finding is consistent with the idea that R-C146A-CRT remains in a monomeric form susceptible to proteasomal degradation. Similar results were obtained when NIH-3T3 cells were cultured in the presence of dithiothreitol to induce disulfide bridge breakage, and R-CRT levels were determined in the presence versus absence of MG132 (data not shown). These findings indicate that R-CRT dimerization prevents its proteasomal degradation and accounts for its reduced turnover rate (Fig. 5).
In summary, the dimerization capability of different CRT isoforms is correlated with their susceptibility to proteasomal degradation. In contrast to the N-end rule, which predicts a longer half-life for CRT than for R-CRT, our results indicate that CRT arginylation is responsible for the longer half-life of R-CRT, even though this isoform is the substrate of the Ubproteasome system.

Discussion
The present study was focused on the degradation mechanism whereby cells modulate cytoplasmic CRT and R-CRT levels. Our previous studies have shown that R-CRT is directed to SG under stress conditions in which Ca 2ϩ levels are decreased (7,12,39) and that R-CRT may reach the plasma membrane if the stress persists (11). It was suggested that CRT from plasma membrane is degraded in the lysosome pathway (40). However, the degradation pathway of internal CRT is unclear, and that of R-CRT is completely unknown. Afshar et al. (24) claimed that cyt-CRT is not a proteasomal substrate because it did not detect ubiquitinated cyt-CRT in cells. On the other hand, Labriola et al. (25) suggested that CRT is degraded in the cytosol by proteasome because, after exposure of cells to an inhibitor of ER Ca 2ϩ -ATPase, cyt-CRT level was reduced by CHX treatment and increased by MG132 treatment. Our present findings indicate that both CRT and R-CRT are proteasomal substrates, although the degradation rate of R-CRT is lower than that of CRT (Figs. 4,5,8,and 9). The same trend was observed for the endogenous protein ( Figs. 1 and 2), indicating that R-CRT halflife is not affected by other PTMs on CRT after passage through the ER.
R-CRT accumulation in cells is a consequence of proteasome inhibition (Figs. 1 and 2). Such accumulation occurs primarily during the first hour of MG132 exposure, suggesting that cyt-CRT is regulated at the retrotranslocation level. Our findings show clearly that the accumulation is due to blocking of proteasomal activity rather than to increased retrotranslocation or CRT synthesis (Fig. 1, C and D), consistent with previous findings that CRT retrotranslocation is independent of proteasomal activity (24). ER stress does not appear to promote R-CRT accumulation, because MG132 treatment of cells produced only a slight accumulation of reporter cytoplasmic substrates (41).  Arginylatable CRT, R-CRT isoform, and non-arginylatable mutant E1V-CRT expressed in cytoplasm had very different half-lives: longest for E1V-CRT (ϳ4.8 h), intermediate for R-CRT (ϳ2.0 h), and shortest for CRT (ϳ0.7 h) (Fig. 5). R-CRT was ubiquitinated, but E1V-CRT was not (Fig. 10). These findings, taken together, could be interpreted as following the N-end rule, in that arginylation of proteins facilitated their rec-ognition by E3 Ub ligases and was followed by proteasomal degradation (19,20). However, ubiquitination in this case did not confer a shorter half-life to R-CRT as predicted by the N-end rule. Rather, our findings are consistent with proteasomal degradation of CRT by two distinct mechanisms, one Ub-dependent and the other Ub-independent. Similar results have been obtained for other proteins (e.g. ornithine decarboxylase, c-FOS, p21 Waf1/Cip1 ) that undergo proteasomal degradation in the absence of Ub modification (21,42). However, it cannot be discarded that both isoforms of CRT could be degraded by the ubiquitin proteasome system with a difference in the efficiency of their ubiquitination. Further studies of the ubiquitination of R-CRT will be useful to elucidate the detailed mechanism of proteasomal targeting for its degradation.
The proteasomal pathway involves several coexisting types of proteasomes (26S, 20S, immunoproteasome, etc.) (43). 26S proteasomes, which account for ϳ15% of total proteasomes, degrade natively folded and functional proteins in a primarily Ub-dependent manner (43). In contrast, 20S proteasomes, which account for ϳ40% of total proteasomes, degrade damaged/misfolded proteins in a Ub-independent manner (43). Different proteasome complexes may, therefore, be responsible for the strikingly different half-lives of CRT, R-CRT, and E1V-CRT. Ubiquitinated R-CRT is most likely degraded by 26S proteasomes, whereas non-ubiquitinatable E1V-CRT is degraded by some other type. Because CRT can be arginylated by cellular Ate1, it may be degraded by more than one type of proteasomal complex, accounting for its lower half-life in cells that express CRT-EGFP. In Ate1 Ϫ/Ϫ cells expressing CRT or R-CRT, degradation of the two proteins is related to differences in half-life similar to those observed in the other experimental systems used (Figs. 8 and 9). CRT expressed in Ate1 Ϫ/Ϫ cells remains  arginylation-free, similarly to E1V-CRT expressed in stably transfected CHO cells. However, the half-life of the E1V-CRT isoform is longer than that of R-CRT (Fig. 5), suggesting that CRT turnover is altered by point mutation of the Glu residue.
In regard to the molecular determinants that target CRT to proteasomal degradation, we hypothesize that other PTMs such as phosphorylation (44) or the presence of PEST sequences in the CRT primary structure could be involved. Previous reports suggest that intrinsically disordered regions, unfolding, and oxidative damage are potential determinants of proteasomal degradation (21,45). Further studies are needed to clarify the molecular contributions of such determinants to CRT degradation.
Arginylation of CRT affects its subcellular localization (7,11). We observed that proteasomal inhibition resulted in accumulation of R-CRT in SGs (Fig. 1). Because only R-CRT is associated with SGs (7), we presume that these structures interfere with proteasomal degradation of R-CRT. Arginylation may also promote molecular rearrangements of the cyt-CRT population, e.g. by facilitating homodimer formation (13), and thereby block proteasomal degradation (Fig. 11). Important points remaining to be clarified are whether R-CRT is protected from degradation because the dimerized form cannot undergo polyubiquitination, or polyubiquitinated R-CRT is protected from degradation by physical association with stress granules. Previous studies indicate that proteins such as C/EBP are stabilized by homodimer or heterodimer formation (46). The longer half-life of R-CRT (Fig. 12) may result from any of the mechanisms mentioned above (ubiquitination, dimer formation, association with SGs, accumulation at the plasma membrane). Once CRT leaves the ER, it must be removed quickly from the cytoplasm because its continued presence would alter Ca 2ϩ homeostasis so as to reduce cell survival. CRT is involved in a variety of injury processes including cell stress, apoptosis, and phagocytosis. CRT must be stabilized in the cytoplasm in order to play its normal role in these processes. Arginylation of CRT is important for such stabilization as well as for the processes of SG formation, plasma membrane localization, and pro-apoptotic signaling. The regulation of CRT levels in various subcellular structures depends on control of not only its synthesis (47) but also its degradation.
The present findings demonstrate cross-talk among the events of CRT retrotranslocation, post-translational arginylation, ubiquitination, and homodimer formation. Future studies will more fully elucidate the cross-talk among PTMs that modulate the fate of CRTs.