|Home | About | Journals | Submit | Contact Us | Français|
Induction of the transcription factor CHOP (CCAAT-binding homologous protein; GADD 153) is a critical cellular response for the transcriptional control of endoplasmic reticulum (ER) stress-induced apoptosis. Upon nuclear translocation, CHOP upregulates the transcription of proapoptotic factors and downregulates antiapoptotic genes. Transcriptional activation by CHOP involves heterodimerization with other members of the basic leucine zipper transcription factor (bZIP) family. We show that the bZIP protein C/EBPβ isoform LIP is required for nuclear translocation of CHOP during ER stress. In early ER stress, LIP undergoes proteasomal degradation in the cytoplasmic compartment. During later ER stress, LIP binds CHOP in both cytoplasmic and nuclear compartments and contributes to its nuclear import. By using CHOP-deficient cells and transfections of LIP-expressing vectors in C/EBPβ−/− mouse embryonic fibroblasts (MEFs), we show that the LIP-CHOP interaction has a stabilizing role for LIP. At the same time, CHOP uses LIP as a vehicle for nuclear import. LIP-expressing C/EBPβ−/− MEFs showed enhanced ER stress-induced apoptosis compared to C/EBPβ-null cells, a finding in agreement with the decreased levels of Bcl-2, a known transcriptional control target of CHOP. In view of the positive effect of CHOP-LIP interaction in mediating their proapoptotic functions, we propose this functional cooperativity as molecular symbiosis between proteins.
The endoplasmic reticulum (ER) stress is caused by a variety of biological (viral infection, aging, or genetic mutations) and environmental factors (glucose deprivation or chemicals) (38) and has been associated with inflammation and the pathology of several diseases such as diabetes mellitus, Alzheimer's, and Parkinson's (reviewed in reference 52).
The overburdening of ER with unfolded proteins triggers a series of tightly controlled and timely defined events (reviewed in reference 38), which constitute the unfolded protein response (UPR). This cellular response is characterized by a decrease of mRNA translation, upregulation of ER-resident chaperones, and activation of ER-associated degradation (ERAD).
The PERK (PKR-like ER kinase)-mediated phosphorylation of the alpha subunit of eukaryotic initiation factor 2 (eIF2α), results in attenuation of global protein synthesis and the preferential translation of some mRNAs, most notably the one for the master regulator of the stress response, activating transcription factor 4 (ATF4) (13). The second arm of the UPR is the activation of the ER membrane-resident transcription factor ATF6, following Golgi apparatus-mediated transport and cleavage by SP1 and SP2 (site 1 and 2 proteases). The splicing of the XBP1 (X-box binding protein 1) mRNA by IRE-1 (inositol requirement 1) kinase-endonuclease is another UPR mechanism, the spliced variant coding for the active transcription factor (53). ATF6 and XBP1 modulate the transcription of genes encoding protein chaperones. After the activation of ERAD, misfolded ER proteins are retrotranslocated and degraded in the cytoplasmic compartment by the ubiquitin-proteasome system (UPS) (42). These mechanisms are initially used to relieve the protein burden of the ER so that the cell can resume its prestress functionality. However, in conditions of severe and/or persistent stress, the apoptotic program is launched.
ATF4 is induced via the PERK-eIF2α pathway and upregulates the transcription of the CCAAT-binding homologous protein (CHOP) (26). The prodeath transcription factor CHOP transcriptionally upregulates the proapoptotic factor BH-3 only (BIM) (34) and downregulates the antiapoptotic B-cell leukemia/lymphoma 2 protein (Bcl-2) (29).
The association between the ER membrane resident kinase IRE1, the Bax/Bak complex and TRAF2 (TNF receptor-associated factor 2) (19) triggers the activation of the ASK1 (apoptosis signal-regulating kinase 1), which transduces the signal to JNK (c-Jun N-terminal kinase) and p38-MAPK. JNK-mediated phosphorylation has activating effects on BIM, but inactivates Bcl-2. Activated BIM binds the Bax/Bak complex, and the consequence is the ER Ca2+ depletion. The increased influx of calcium into the mitochondria triggers the release of cytochrome c, formation of the apoptosome, and the subsequent full caspase activation (38).
ER stress conditions abruptly upregulate a plethora of transcription factors, such as ATF3 (23), ATF4 (13), ATF6 (17), CHOP (46), and XBP1 (53). Their function is initially to regulate survival genes, such as protein chaperones who alleviate the ER stress burden (38), and later on to induce proapoptotic genes or to downregulate the antiapoptotic ones (33). The interplay between the induced transcription factors plays a critical role in the outcome of the ER stress response and the fate of the cell. The proteins belonging to the basic leucine zipper transcription factor (bZIP) family are known to homo- and heterodimerize with other members of the family and the various heterodimer combinations to greatly enhance the repertoire of regulated gene targets (35). Dimerization usually occurs at the leucine zipper domain (4). Such interactions have been described between ATF4 and CHOP (41), between ATF4 and C/EBPβ (CCAAT-binding enhancer protein beta) (44), or between CHOP and C/EBPβ or C/EBPα (8, 37).
The CHOP transcription factor belongs to the bZIP family, and it was determined that it cannot effectively form homodimers (43); therefore, in order to function, it must heterodimerize with other members of the family, most notably with C/EBPβ (37). CHOP-containing heterodimers are either inactive (CHOP acting as a dominant-negative partner ) or have a decreased requirement for consensus DNA sequence binding (43). CHOP is expressed at very low levels in nonstress conditions but is robustly upregulated in ER stress. In addition, CHOP−/− mouse embryonic fibroblasts (MEFs) are more resistant to ER stress-induced apoptosis (tunicamycin treatment) than their wild-type counterparts (56). Interestingly, the same was observed for C/EBPβ−/− MEFs (56). In addition to BIM and Bcl-2, other gene targets of CHOP include growth arrest and DNA damage-inducible protein (GADD34), ER oxidoreductin 1 (ERO1α) (27), and Tribbles-related protein 3 (TRB3) (31). Key protein effectors of the UPR (ATF4, XBP-1, and ATF6) regulate CHOP, and CHOP appears to be a crucial proapoptotic factor found at the convergence point in the regulatory network used for the launching of apoptosis caused by persistent ER stress.
The nuclear translocation of CHOP is a necessary step for the induction of apoptosis. Human islet amyloid polypeptide (IAPP) was suggested to be one of CHOP nuclear import facilitators because the infection of INS-1 cells with IAPP-expressing adenovirus was correlated with increased CHOP expression, nuclear translocation, and DNA fragmentation (21).
The C/EBPβ transcription factor belongs (same as CHOP) to the bZIP family, and it was initially described to be a transcriptional regulator of the interleukin-6 (IL-6) gene (1). Although it was mostly associated (together with C/EBPα) with adipogenesis (50), it is now recognized that the roles of C/EBPβ span well beyond the differentiation of preadipocytes. Recently, it was shown to play important roles in bone formation by regulating the MafB (V-maf musculoaponeurotic fibrosarcoma oncogene homolog B) transcription factor (39). It was also determined to be a downstream mediator of ERK1/2 activation, its phosphorylation being a critical step in ovulation and luteinization-related events (7).
C/EBPβ is a proven proteasome target, and its degradation is believed to take place via a ubiquitin-independent mechanism (16). In human intestinal epithelial cells (22) and lung fibroblasts treated with the proteasome inhibitor MG132 (24) or in myotubes treated with β-lactone, the expression of all three C/EBPβ isoforms is increased (48). However, in one report, the degradation of C/EBPβ was linked to calpain activation (49).
Three functionally relevant C/EBPβ isoforms were described (most likely expressed by use of alternative start codons): liver-enriched transcriptional activating protein 1 (LAP-1), LAP-2, and liver inhibitory protein (LIP). The mouse LIP isoform is a 152-amino-acid N-terminus truncated version of LAP-1 and, because it lacks the activation domains, is widely considered an attenuator of transcription (although in one recent report it was shown to fulfill activator functions ).
In a recent publication from our lab (25), we analyzed the kinetics of the LIP isoform in early ER stress and determined that, after a sharp decrease in expression in early ER stress (attributed to proteasomal degradation and translational control), LIP protein levels recovered in mid to late ER stress and the LIP/LAP-2 ratio increased significantly. We also noticed that the protein level kinetics for the LIP and CHOP proteins during ER stress were similar. Furthermore, the nuclear levels of CHOP were severely decreased in C/EBPβ-deficient MEFs compared to their wild-type counterpart.
We sought to determine whether or not the CHOP-LIP protein-protein interaction has any physiological relevance in terms of cell death or survival during the course of ER stress. We hypothesized that the putative nuclear localization sequence (NLS) within CHOP is not functional and that the binding between the two proteins may favor the nuclear import of CHOP-LIP heterodimers, thus enabling the proapoptotic functions of CHOP. Therefore, we examined the CHOP-LIP interaction in both the cytoplasmic and nuclear compartments and analyzed the localization of the CHOP protein in stable transfected cell lines expressing the wild type and LIP mutants. In order to assign a physiological relevance to this protein-protein interaction, we compared the specific regulation of CHOP-modulated proapoptotic gene targets and the ER stress-induced cell death in MEFs lacking the C/EBPβ protein or expressing the LIP isoform only.
Thapsigargin (Tg), leptomycin B (LMB), Z-Leu-Leu-Leu-al (MG132), and cycloheximide (CHX), all of the highest purity available, were purchased from Sigma. Restriction enzymes were bought from New England Biolabs.
Cells were cultured in high glucose Dulbecco modified Eagle medium (DMEM) in the presence of penicillin (100 IU/ml), streptomycin (100 μg/ml), and 2 mM l-glutamine under a humidified atmosphere and 5% CO2 at 37°C. C6 rat glioma cells cultured in DMEM were supplemented with 5% heat-inactivated fetal bovine serum (FBS) and 5% calf serum. MEFs were cultured in DMEM supplemented with 10% FBS. The stable cell lines generated from the plasmid-transfected MEFs were selected and maintained in the medium described above, supplemented with 150 μg of hygromycin B (Invitrogen)/ml. Fugene HD (Roche Applied Sciences) was used to transfect cultured cells according to the manufacturer's instructions. Induction of ER stress was performed as described previously (25). Briefly, cells were treated with 400 nM Tg for the indicated times.
In order to obtain endogenous regulation and expression levels, we constructed plasmids expressing C/EBPβ isoforms, LIP, or LIP mutants under the control of the natural C/EBPβ promoter. The plasmids were transfected into the C/EBPβ−/− MEFs, and mass cultures grown in hygromycin-supplemented medium were generated. The pTK-Hyg plasmid (Clontech) encoding for both hygromycin and ampicillin resistance was linearized by PCR amplification with the primer pair 5′-CAAACTGTCGACCAAAAAATGC-3′ (with a SalI restriction site) and 5′-CCAAAGGTTCCCCACCAACAGC-3′. The pBS+ plasmid encoding the genomic rat coding sequence for C/EBPβ, together with 2.97 kb of its promoter and 0.48 kb of the 3′ untranslated region (6), was digested with SalI and HindIII and ligated into the pTK-Hyg. vector, previously cut with the same enzyme combination.
Site-directed mutagenesis was performed by using a QuikChange mutagenesis kit (Stratagene) and the cycling program: 95°C for 3 min, 95°C for 45 s, 55°C for 45 s, and 68°C for 10 min (with steps 2 to 4 repeated 18 times). The cycling mix contained 30 pmol of each primer and 5% dimethyl sulfoxide. DH5α competent bacteria were used for transformation.
The LIP-only plasmid was obtained by digestion with MlyI and MscI restriction enzymes, removal of the 120-bp insert containing the ATG1 and ATG2 initiation codons of the C/EBPβ mRNA, and religation of the remaining plasmid.
The C/EBPβ-NLS mutant plasmids (NLSi) were generated by site-directed mutagenesis of the NLS using the primer pair 5′-CAACAACATCGCGGTGGGCGCTAGCCGCGACAAGGCC-3′ and 5′-GGCCTTGTCGCGGCTAGCGCCCACCGCGATGTTGTTG-3′ and by using C/EBPβ- or the LIP-expressing plasmids as templates.
The CHOP expression plasmid was described elsewhere (20). The CHOP construct with the putative NLS sequence deleted (LIP-NLSΔ) was generated by site-directed mutagenesis using the primer pair 5′-AAGATCAAGGAAGACAGAGTGGTCAGTG-3′ and 5′-CACTGACCACTCTGTCTTCCTTGATCTT-3′.
The CHOP-NLSC/EBPβ plasmid was obtained following a two-step process. First, we mutagenized the CHOP gene with the following primer pairs: (i) 5′-GGAAGAGCAAGGAAGAACCGGTAAACGGAAACAGAGTGG-3′ and 5′-CCACTCTGTTTCCGTTTACCGGTTCTTCCTTGCTCTTCC-3′ with an AgeI restriction site and (ii) 5′-CAAGGAAGAACCGGTAAACTTAAGCAGAGTGGTCAGTGCCCAGC-3′ and 5′-GCTGGGCACTGACCACTCTGCTTAAGTTTACCGGTTCTTCCTTG-3′ with a restriction site for AflII. A double-stranded oligonucleotide coding for the amino acid sequence of the C/EBPβ NLS and with cohesive ends for AgeI and AflII was obtained by annealing the single-stranded oligonucleotides 5′-CCGGTGTGCGCAAGTCGCGAGACAAGGCCC-3′ and 5′-TTAAGGGCCTTGTCTCGCGACTTGCGCACA-3′. The final step in obtaining the CHOP-NLSC/EBPβ plasmid was the ligation of the double-stranded oligonucleotide into the previously AgeI- and AflII-digested mutant CHOP plasmid.
First-strand cDNA samples were synthesized from RNA samples by using the Superscript III First-Strand Synthesis SuperMix for quantitative real-time RT-PCR (qRT-PCR; Invitrogen) as described previously (25). Real-time PCR was performed using StepONE Plus (Applied Biosystems) and the Power SYBR green PCR master mix, according to the manufacturer's instructions. The PCR primers used were as follows: 5′-CTGGAAGCCTGGTATGAGGAT-3′ and 5′-CAGGGTCAAGAGTAGTGAAGGT-3′ for the detection of CHOP transcripts and 5′-ATGCCTTTGTGGAACTATATGGC-3′ and 5′-GGTATGCACCCAGAGTGATGC-3′ for the detection of the mouse Bcl-2 transcript 1. The 18S RNA primer pair was as follows: 5′-TTGACGGAAGGGCACCACCAG-3′ and 5′-GCACCACCACCCACGGAATCG-3′; the primer pair 5′-GCAAGAAGAAGATCGCCAAG-3′ and 5′-CGCTCCTCAAACTTGACCTT-3′ was used for the detection of the RPL27 mRNA. Both RPL27 and 18S rRNA served as internal controls.
Cells were seeded on gelatin-coated coverslips, treated, and washed with phosphate-buffered saline (PBS). Fixing was performed by immersing the coverslips in paraformaldehyde (4% in 1× PBS [pH 7.5]) at room temperature for 30 min. After the permeabilization step (5 min of incubation at room temperature in PBS-0.2% Triton X-100), the coverslips were blocked with 10% bovine serum albumin in PBS-0.1% Triton X-100 for 1 h at room temperature. The binding of the primary antibody (GADD 153 [Santa Cruz catalog no. sc-7351], 1:50 dilution in blocking solution) took place overnight at 4°C and, after a second washing step with PBS (three times for 5 min each time), the coverslips were incubated with the secondary antibody (goat anti-mouse Oregon green 488 [Molecular Probes catalog no. O-6380], dilution 1:300) for 2 h at room temperature. The coverslips were washed again three times for 5 min each time with PBS and mounted with SlowFade Gold antifade reagent containing DAPI (4′,6′-diamidino-2-phenylindole; Invitrogen).
Both the growth medium and the subsequently trypsinized cells were collected by centrifugation (4,000 rpm, 10 min at 4°C). Cell pellets were gently suspended in 1 ml of PBS and immediately fixed by injecting the resulting cell suspension in 10 ml of 70% ethanol with 4 h at 4°C incubation. The fixed cells were pelleted, suspended in 1 ml of 2 N HCl-0.5% Triton X-100, and incubated at room temperature for 30 min. After being washed with 1 ml of 0.1 M Na2B4O7·H2O (pH 8.5), the cells were resuspended in 1 ml of 10 μg of propidium iodide/ml in PBS, followed by incubation for 4 h at 4°C.
A Becton Dickinson flow cytometer set at 488 nm was used, and histogram plots were created using the DiVa software. The percentages of cells belonging to the different phases of cell cycle were visualized by using a scatter plot of red fluorescence and calculated by gating the sub-G1, G1, S, and G2 cell populations.
Total cell lysates were obtained by suspending the cell pellets in radioimmunoprecipitation assay buffer (50 mM Tris-HCl [pH 8], 150 mM NaCl, 1% NP-40, 0.1% sodium dodecyl sulfate [SDS], 2 mM EDTA) supplemented with protease inhibitor mix (Roche), sonication of the suspensions two times for 10 s each time on ice, and clearance by centrifugation at 13,000 rpm for 10 min at 4°C.
The fractionation into nuclear extracts and cytoplasmic fractions was described elsewhere (25). Typically, 40 μg of total cell lysate, 20 μg of nuclear extract, and 40 μg of cytoplasmic fractions were separated by SDS-PAGE. Proteins were detected on Western blots with antibodies specific for C/EBPβ, CHOP (GADD 153), and Bcl-2 (Santa Cruz Biotechnology); histone H3, caspase 3, and PARP [poly(ADP-ribose) polymerase; Cell Signaling]; tubulin (Sigma); and horseradish peroxidase-conjugated secondary antibodies (Calbiochem) and then visualized by enhanced chemiluminescence (Perkin-Elmer). All Western blot experiments were repeated at least three times unless otherwise indicated.
Immunoprecipitations were typically performed by incubating 3 μg of antibody overnight with 600 μg of lysate under rotation conditions. Immunocomplexes were pulled down with protein A-Dynabeads (Invitrogen), washed, boiled in sample buffer, and resolved on SDS gels.
Fluorescence and phase-contrast pictures were taken by using of a Leica DMI 4000 B microscope.
We have previously shown (25) that LIP protein levels decrease significantly in C6 cells treated with Tg in early ER stress. In spite of its recognized transcription factor function, LIP exhibits also cytoplasmic localization, and we sought to determine whether LIP exits the nucleus in order to be degraded in the cytoplasm during early ER stress. We pretreated C6 cells with the nuclear export inhibitor leptomycin B (LMB), followed by the induction of ER stress by Tg treatment. In C6 cells treated with Tg alone, we observed the expected decrease of the LIP isoform in early ER stress. Interestingly, the nuclear LIP levels were rescued by LMB pretreatment (Fig. (Fig.1A),1A), indicating that LIP exits the nucleus to be degraded in the cytoplasm during ER stress. This mechanism was specific for LIP because LAP levels did not change significantly during the treatments (Fig. (Fig.1A).1A). As expected, treatment of unstressed cells with LMB did not show nuclear accumulation of the protein, supporting the observation that ER stress induces nuclear exit and degradation of LIP in the cytoplasm (Fig. (Fig.1A1A).
In order to further assess this, we created a mutant in which the NLS previously described for C/EBPβ proteins (2) was rendered nonfunctional (NLSi) by substituting with alanine two critical amino acids (R238 and K239) (Fig. (Fig.1B).1B). We expressed wild-type and mutant proteins in C/EBPβ−/− MEFs, fractionated the cell lysates into nuclear and cytoplasmic extracts, and showed that both the NLSi-LIP and the NLSi-LAP proteins localized more in the cytoplasm (Fig. (Fig.1B),1B), with LIP being almost exclusively cytoplasmic. We also noticed that the mutant LIP levels in total cell lysates were significantly decreased compared to those from the wild-type protein (Fig. (Fig.1B).1B). In agreement with this observation, C/EBPβ−/− cells transfected with a LIP expression vector showed higher LIP levels than cells transfected with the LIP-NLSi isoform (data not shown). Taken together, the presented data and our previous report (25) propose that LIP is a target for proteasomal degradation in the cytoplasm of stressed and un-stressed cells.
We hypothesized that a cytoplasmically localized and thus more degradation-prone LIP would have a significantly shorter half-life than the wild-type protein, which is capable of entering the nucleus, and thus is protected from proteasomal degradation. In order to measure the half-lives of the wild-type and mutant proteins, we inhibited the protein synthesis in C/EBPβ−/− cells expressing either the wild-type or the NLSi proteins by treating them with 10 μg of CHX (an inhibitor of translation elongation)/ml. We were able to confirm the significantly shorter half-life of the NLS-inactivated LIP mutant also by using hippuristanol as an inhibitor of translation initiation (data not shown). In agreement with our hypothesis, the half-life of the LIP-NLSi isoform was significantly lower than that of the wild-type protein (19 min compared to 2.5 h, Fig. Fig.1C).1C). When we inhibited the proteasome activity in C/EBPβ-NLSi expressing cells with 10 μM MG132 in CHX (10 μg/ml) conditions, we observed that the protein level of LIP-NLSi mutant is rescued, further confirming the proteasome-mediated degradation of cytoplasmic LIP (Fig. (Fig.1D).1D). These treatment conditions did not lead to increase of the phosphorylation levels of eIF2α or to the induction of CHOP (data not shown). We concluded that the nuclear localization protects LIP from proteasomal degradation.
During the work for a related project, we noticed that the kinetics of accumulation for CHOP and LIP proteins were similar during ER stress. Also, in MEFs deficient for CHOP (CHOP−/−), the protein levels of LIP were significantly decreased in ER stress, compared to the CHOP-expressing MEFs (CHOP+/+), both in total cell extracts and in nuclear fractions (Fig. (Fig.2A).2A). Furthermore, accumulation of LIP protein levels in late ER stress (>6 h of Tg treatment) was delayed (Fig. (Fig.2A).2A). These findings indicate a possible implication of CHOP in the kinetics of LIP expression.
We decided to investigate the previously described CHOP-LIP interaction in-depth by performing coimmunoprecipitation experiments using antibodies for CHOP and C/EBPβ and LIP-expressing, ER-stressed cells. When we performed coimmunoprecipitation experiments using whole-cell lysates from cells expressing the LIP isoform only, we found that the interaction of CHOP and LIP (CHOP:LIP binding ratio being about 1:1.8) is apparent as early as 3 h of stress (Fig. (Fig.2B).2B). By using nuclear and cytoplasmic fractions, we showed that the interaction takes place in both cellular compartments (Fig. (Fig.2B).2B). To further prove the cytoplasmic association, we used coimmunoprecipitation experiments using cells expressing the LIP mutant bearing the inactivated NLS (LIP-NLSi). Because the coimmunoprecipitation experiments show that this particular LIP mutant, despite its nuclear localization deficiency, is still able to bind the CHOP protein (Fig. (Fig.2C),2C), we concluded that the CHOP-LIP protein interaction takes place also in the cytoplasm.
We showed that MEFs lacking CHOP exhibit significantly lower LIP protein levels in ER stress compared to their wild-type counterparts, suggesting stabilizing roles of CHOP toward LIP. Furthermore, CHOP physically binds LIP in both the cytoplasm and the nucleus during the Tg-induced ER stress.
Due to the importance of CHOP in the launch of ER stress-caused apoptosis, we wondered whether the LIP-CHOP interaction has any relevance for the localization and the cellular function of the latter. We checked the nuclear localization of CHOP in cells deficient for C/EBPβ (KO), cells expressing the LIP isoform only (LIP), and cells expressing all three C/EBPβ isoforms (C/EBPβ). We observed that, during early ER stress, LIP-only and C/EBPβ cells showed similar levels of nuclear CHOP, while in the nuclei of KO cells the levels of CHOP were severely decreased (Fig. (Fig.3A).3A). In addition, the levels of CHOP analyzed in total cell lysates were significantly lower in C/EBPβ−/− MEFs compared to LIP- or C/EBPβ-expressing cells (data not shown).
The higher CHOP levels in LIP-only and C/EBPβ-expressing cells were not due to increased CHOP mRNA levels (Fig. (Fig.3B),3B), compared to KO cells. Moreover, the LIP-dependent cytoplasmically localized CHOP had similar levels in all three cell lines (Fig. (Fig.3A).3A). In addition, the lower CHOP levels in KO cells, may be explained by the proteasomal degradation of CHOP during ER stress. This was supported by increased CHOP levels in KO cells treated with Tg and inhibitors of the proteasome (data not shown).
We also confirmed the LIP-dependent pattern of nuclear localization of CHOP in ER stress by performing immunocytochemistry experiments, staining the KO, LIP, and C/EBPβ cells with CHOP-specific antibodies and the nucleus-specific dye DAPI (Fig. (Fig.3C).3C). These results, together with the apparent cytoplasmic interaction between the two proteins, strongly suggest that one of the cellular roles of the LIP isoform during ER stress might be of favoring the nuclear import of CHOP.
To our knowledge, no NLS was thus far described for the CHOP protein. We carefully inspected the amino acid sequence of CHOP and observed that the sequence TRKRK, located between the amino acids 101 and 105 in the mouse sequence has the best resemblance to a classical monopartite NLS (Fig. (Fig.4A).4A). We deleted the putative NLS of CHOP and transiently cotransfected CHOP-deficient MEFs with either CHOP wild-type, CHOP-NLSC/EBPβ or CHOP-NLSΔ-constitutively expressing plasmids (under the control of the cytomegalovirus [CMV] promoter), together with pEGFP-N1 plasmid (ratio of 5:1). We did not notice any difference between the nuclear localization of the NLS-deleted CHOP and the one observed for the wild-type protein (Fig. (Fig.4B).4B). However, the CHOP mutant bearing the LIP NLS (NLSC/EBPβ) (Fig. (Fig.4B)4B) shows robust nuclear accumulation compared to the wild-type variant of CHOP, suggesting that the putative NLS of CHOP is not functional or at least that it is not very efficient. Because the predicted NLS for CHOP seems not to be genuine and because the CHOP protein is unable to homodimerize, these results support the idea of CHOP's need for an interaction partner for nuclear import.
In order to test this hypothesis, we checked the nuclear localization of CHOP during ER stress, by cellular fractionation (Fig. (Fig.4C)4C) and immunocytochemistry (Fig. (Fig.4D),4D), in the cell line expressing an exclusively cytoplasmically localized LIP isoform (LIP-NLSi). Interestingly, we found that the protein level of nuclear CHOP in LIP-NLSi cells was similar to the one from KO cells (not shown) and significantly decreased compared to the one in the LIP wild-type cells. These data suggest that, when LIP is trapped in the cytoplasm, the nuclear import of CHOP is severely impaired (Fig. 4C and D). However, cytoplasmic LIP And CHOP levels were not higher in LIP-NLSi cells compared to wild type-LIP expressing cells, suggesting that cytoplasmic trapping of the two proteins may promote their proteasomal degradation. This is also in agreement with lower CHOP levels during ER stress in total extracts of LIP-NLSi cells (data not shown).
Having confirmed that LIP binds CHOP and that this interaction enables the nuclear import of CHOP in ER stress, it was important to determine whether or not any physiological relevance may be attributed to this process. Thus, according to our hypothesis, the cells expressing the LIP isoform should be more prone to undergo apoptosis compared to KO cells, due to the increased nuclear import of CHOP and hence the earlier activation of its proapoptotic targets. We treated the three cell lines (KO, LIP, and C/EBPβ) with Tg and monitored the appearance of floating, round, highly refringent, apoptotic cells. As anticipated, after 18 h of Tg treatment, the cells expressing the LIP isoform alone and, to a lesser extent, the cells expressing all three C/EBPβ isoforms exhibited detached cells in large number, compared to the dismal number of apoptotic cells in the KO cell line (phase-contrast pictures, Fig. Fig.5A).5A). When we measured the percentage of sub-G1 (apoptotic) cells by propidium iodide staining (and by using FACS analysis), we found that Tg treatment induces apoptosis in ca. 20.5% of the counted LIP cells (a 3.9-fold increase versus untreated cells) and ca. 22% in the cells expressing all three isoforms at the 18-h time point, compared to only 7.3% (a 1.6-fold increase versus untreated cells) in KO cells (Fig. (Fig.5B).5B). Thus, it appears that the expression of the LIP isoform alone in a C/EBPβ−/− background is sufficient to efficiently induce apoptosis during ER stress in MEFs.
We also tested the Tg-induced cell death in CHOP−/− and CHOP+/+ cells and observed that after 18 h of treatment, the CHOP+/+ cells showed 3.26 times more apoptotic cells compared to the untreated group. The increase in the CHOP−/− treated cells was 1.37 times higher than the untreated (data not shown). These data are in agreement with the proapoptotic function of CHOP during ER stress (56).
At the molecular level we found that the proapoptotic (cleaved) caspase 3 is increased both in protein level and in activity (cleavage of PARP) significantly earlier in the LIP-expressing cells compared to the KO cells (12 versus 18 h) during Tg-induced ER stress (Fig. (Fig.5C),5C), strongly suggesting that the ER stress apoptotic program is launched more rapidly in cells expressing the LIP isoform.
Because CHOP is a well-characterized proapoptotic protein and because during ER stress it was shown to repress the transcription of the antiapoptotic factor Bcl-2, we compared the protein and the mRNA expression of Bcl-2 in KO versus LIP cells. We found that in LIP cells the protein expression of Bcl-2 is severely decreased during ER stress (Fig. (Fig.5D),5D), strongly suggesting that the expression of LIP is required for the CHOP-mediated repression of this gene. The mRNA levels of Bcl-2, as quantified by qRT-PCR, showed a decrease at the 9- and 12-h time points of Tg treatment, strongly implicating transcriptional regulation events in modulating Bcl-2 levels. This conclusion is also supported by the decreased Bcl-2 levels in MEFs expressing all three C/EBPβ isoforms (data not shown). Taken together, the experiments described in Fig. Fig.55 established LIP as an important contributing factor in launching of the ER stress apoptotic program.
The present work reveals new functional roles played by the C/EBPβ isoform LIP during ER stress. The transcription factor LIP lacks the transactivation domains (present in both LAP-1 and LAP-2 isoforms), and its cellular function was initially believed to be that of a gene repressor (or attenuator), acting as a dominant-negative dimerization partner (5). Expression of LIP was shown to have proproliferative roles in several breast cancer cell lines (54), (55). However, the cellular functions of LIP (especially other than those related to transcriptional regulation) and implications of its metabolism for the fate of the cell during stress conditions have thus far been insufficiently studied.
In earlier studies, we showed that in the early hours of Tg-induced ER stress in C6 cells, the protein levels of LIP decrease significantly, but the chemical inhibition of the proteasome was able to rescue its expression (25).
Nuclear localization and nuclear export sequences were previously described for C/EBPβ, and it was experimentally proven that the LAP isoforms are able to shuttle to the cytoplasm in some conditions, such as in tumor necrosis factor alpha-treated primary hepatocytes (2) or 3T3-L1 preadipocytes treated with ceramide (40). Here, we sought to determine the cellular compartment in which LIP is being degraded during the ER stress, since this might prove to be relevant for its function. Depending on the cell line, between 17 and 50% of the proteasomes are located in the nucleus (36), and nuclear proteasomes differ from the cytoplasmic ones in both the affinity and the specificity of their substrates (45). For example, the transcription factor MyoD was shown to be proteasomally degraded exclusively in the nucleus (9). On the other hand, several proteins, most notably the proapoptotic transcription factor p53, need to be nuclear exported in order to be degraded (11).
LMB is a potent inhibitor of the CRM-1 (chromosome region maintenance 1) mediated nuclear export of proteins containing leucine-rich export sequences (10) and by using it in conjunction with Tg-induced ER stress, we showed that, when nucleus localized, the LIP isoform is protected from proteasomal degradation. Conversely, the C/EBPβ mutant proteins and especially the LIP isoform with the inactivated NLS (NLSi) exhibit significantly lower protein levels in total extracts. The half-life of LIP-NLSi was significantly shorter compared to the wild-type proteins, suggesting that LIP is a target of cytoplasmic proteasomes also in nonstress conditions and that its degradation is exacerbated upon Tg treatment. Thus, in the present study we were able to uncover the mechanistic base of the proteasome-mediated degradation of LIP during ER stress, that being the nuclear-cytoplasmic shuttling, followed by the subsequent degradation in the cytoplasmic compartment.
It was previously published that CHOP avidly binds all three C/EBPβ isoforms (47) and that the trapping of LIP by CHOP seems to have an activating role for LAP-mediated induction of the IL-6 gene (15). We determined that in CHOP-deficient MEFs, the recovery of LIP protein levels during ER stress is significantly impaired, pointing toward a functional connection between the two and a stabilizing role of CHOP toward LIP. CHOP-LIP interaction takes place also in the cytoplasm, suggesting functions other than transcriptional regulation for the CHOP-LIP heterodimers. Moreover, the protein level of nuclear CHOP in LIP-expressing cells is significantly higher compared to the one observed in the C/EBPβ−/− cells, pointing toward a possible role for LIP in CHOP's cellular localization.
The nuclear import of CHOP was addressed in a few publications but, to our best knowledge, no NLS was deemed responsible for it. The N terminus of the CHOP protein (amino acids 1 to 70) was determined to be irrelevant for this process (32). The substitution of the nuclear localization-resembling sequence within CHOP with the NLS characterized for the C/EBPβ proteins led to significantly increased CHOP nuclear levels, compared to the wild-type protein or to the mutant with the deleted putative NLS, a clear indication of a more efficient nuclear import.
Thus, a tempting hypothesis, strengthened by the knowledge that CHOP is not capable of forming homodimers (43), was that, in order to be imported into the nucleus and fulfill its proapoptotic functions, CHOP needs to interact with another protein, possibly from the bZIP family and that the interacting partner should provide the heterodimers with a functional NLS. Dimerization requirement for nuclear import of transcription factors was previously described. The STAT1 (signal transducer and activator of transcription 1) transcription factor is trapped in the cytoplasm in a monomeric state and needs to be tyrosine phosphorylated as a prerequisite for its dimerization and the subsequent nuclear import (28). In a recent publication (30) it was shown that the transcription factors NeuroD1 (neurogenic differentiation 1) and E47 heterodimerize and act synergistically in order to be imported into the nucleus.
We show here that when LIP is forcibly cytoplasmically localized (in the LIP-NLSi cell line), the nuclear levels of CHOP are dramatically reduced, strongly suggesting that the LIP isoform is likely to be one of the contributing factors in the nuclear import of CHOP during ER stress.
The contribution of the C/EBPβ proteins to apoptotic cell death is still debated (3, 12, 51). In ER stress conditions, C/EBPβ seems to fulfill proapoptotic roles (56); however, the contribution of the singular C/EBPβ isoforms to the activation of the ER stress apoptotic program was thus far not addressed experimentally.
We found that LIP-only cells exhibit at the molecular level earlier appearance and activity of proapoptotic protein species (cleaved caspase 3) and decreased expression of the antiapoptotic Bcl-2 (a gene for whom CHOP has a repressor activity ) compared to KO cells during the course of ER stress.
Taken together, the above-mentioned results strongly suggest a physiological relevance for LIP in the launch of ER stress-induced apoptosis, the mechanism being, by our results and interpretations, the LIP-mediated nuclear import of CHOP.
Our work suggests that CHOP does not homodimerize (43) because the resulting NLS-deficient CHOP homodimers would be trapped in the cytoplasm and they would be targeted by the proteasomes (15). CHOP would thus be forced to heterodimerize, and its binding partner would have to provide a functional sequence for nuclear import. We described here LIP as being one of these partners. However, because some CHOP protein is detected in the nuclei of the C/EBPβ−/− cells, it is reasonable to believe that LIP is not the only one factor responsible for the nuclear import of CHOP and that, in some stress conditions, other bZIP transcription factors might substitute LIP. Adaptation phenomena of the cultured C/EBPβ−/− cells are also not to be excluded.
An important finding of the present study is that endogenous expression of the LIP isoform only is sufficient for the efficient induction of ER stress-induced apoptosis. By showing that LIP enables the nuclear import of CHOP, we were able to provide a mechanistic explanation for the apparent proapoptotic roles of LIP, strongly supported by the robust downregulation of a specifically CHOP-repressed antiapoptotic gene target, Bcl-2. Although it was previously suggested that, in order to repress Bcl-2, CHOP needs a bZIP family heterodimerization partner (29), at this point we do not know whether this partner could be LIP. A close inspection of the mouse Bcl-2 promoter region does not reveal any consensuslike putative binding sites for either CHOP or C/EBPβ, but the regulation of Bcl-2 by C/EBPβ was described in human cells (18). Even though we clearly see the association of CHOP with LIP in the nuclear fractions also, we do not know whether the nuclear LIP-CHOP heterodimers are transcriptionally active entities (one report  claims they are not) and whether the role of the LIP-CHOP interaction spans beyond the nuclear import process of the heterodimers. LIP might just be a cytoplasmic-nuclear shuttling carrier for CHOP, and a global analysis of cellular mRNA transcripts in LIP-only cells should provide answers to the regulation of LIP target genes. According to our model (Fig. (Fig.6),6), the C/EBPβ LIP isoform is a continuous target of the cytoplasmic proteasomes and is therefore mainly localized in the safe environment of the nucleus. Its degradation, probably coupled with increased nuclear export, is increased in the first hours of stress. However, once CHOP is upregulated and begins to accumulate in the cytoplasm, it binds LIP and CHOP-LIP heterodimers are being translocated to the nucleus. CHOP is then able to modulate its target genes, among others, by repressing the expression of Bcl-2 and other antiapoptotic genes (20, 25, 41).
Taken together, our data suggest that, during ER stress, LIP seems to be using CHOP for protection from degradation and CHOP seems to make use of LIP for function. Thus, because both protein species appear to mutually benefit from their interaction, the physiological outcome being apoptosis, we describe the LIP-CHOP functional cooperativity using the term protein symbiosis.
C.-B.C. dedicates this study to Roderich Brandsch (University of Freiburg, Freiburg, Germany).
C.-B.C. thanks Cristinel Sandu (Rockefeller University) for critical reading of the manuscript. We thank Ronald Wek (Indiana University) for providing the CHOP−/− MEFs and Peter F. Johnson (National Cancer Institute—Frederick) for the C/EBPβ−/− MEFs. Chuanping Wang is acknowledged for excellent technical assistance, and Elena Bevilacqua is acknowledged for fruitful discussions.
This study was supported by grants R01DK060596 and R01DK053307 to M.H. from the National Institutes of Health.
Published ahead of print on 17 May 2010.