(Received for publication, July 22, 1996, and in revised form, October 30, 1996)
From the Howard Hughes Medical Institute, the
§ Department of Biological Chemistry, and the
Department of Anatomy and Cell Biology, University of Michigan
Medical Center, Ann Arbor, Michigan 48109, the ** Genetics Institute
Incorporated, Cambridge, Massachusetts 02140; and the
Department of Tumor Cell Biology, St. Jude
Children's Research Hospital, Memphis, Tennessee 38105
BiP/GRP78 is a lumenal stress protein of the endoplasmic reticulum (ER) that interacts with polypeptide folding intermediates transiting the secretory compartment. We have studied the secretion and the stress response in Chinese hamster ovary (CHO) cells that overexpress either wild-type immunoglobulin binding protein (BiP) or a BiP deletion molecule (residues 175-201) that can bind peptides and ATP but is defective in ATP hydrolysis and concomitant peptide release. Overexpressed wild-type BiP was localized to the ER and unique vesicles within the nucleus, whereas overexpressed ATPase-defective BiP was localized to the ER and cytoplasmic vesicles but was absent from the nucleus. Compared with wild-type CHO cells, overexpression of ATPase-defective BiP prevented secretion of factor VIII, a coagulation factor that extensively binds BiP in the lumen of the ER. Under these conditions factor VIII was stably associated with the ATPase-defective BiP. In contrast, the secretion of monocyte/macrophage colony stimulating factor, a protein that is not detected in association with BiP, was not affected by overexpression of ATPase-defective BiP. These results show that BiP function is not required for secretion of some proteins and suggest that some proteins do not interact with BiP upon transport through the ER.
The presence of unfolded protein in the ER induces transcription of BiP and also elicits a general inhibition of protein synthesis. Overexpression of wild-type BiP prevented the stress-mediated transcriptional induction of BiP in response to either calcium ionophore A23187 treatment or tunicamycin treatment. In contrast, overexpression of ATPase-defective BiP did not prevent the stress induction of BiP, showing that the ATPase activity is required to inhibit transcriptional induction. Overexpression of wild-type BiP, but not ATPase-defective BiP, increased survival of cells treated with A23187. The increased survival mediated by overexpressed wild-type BiP correlated with reduced translation inhibition in response to the stress condition. These results indicate that overexpressed BiP alleviated the stress in the ER to prevent BiP transcriptional induction and permit continued translation of cellular mRNAs.
The endoplasmic reticulum is the site where folding occurs for newly synthesized proteins that are destined for the cell surface. In addition, it is the principal cellular storage site for calcium. Calcium homeostasis is required for proper polypeptide folding and secretion of selective proteins (1) as well as for intracellular signaling events that occur within the cell. Within the ER1 are resident cellular proteins, such as the glucose-regulated protein of 78 kDa (2), that are also known as the immunoglobulin-binding protein (BiP) (3, 4), the glucose-regulated protein of 94 kDa (GRP94), calnexin, calreticulin, and ERp72 that associate with folding intermediates. It is proposed that these cellular ER proteins act as chaperones to prevent aggregation of polypeptide folding intermediates (5). In addition, several of these proteins are documented to be the major calcium-binding proteins in the cell (1, 6, 7). Thus, the processes that maintain calcium homeostasis and proper polypeptide folding are likely to be intimately intertwined.
Like all hsp70 family members, BiP binds ATP tightly (8) and has a weak ATPase activity (9) that is stimulated in vitro by small hydrophobic peptides (10) and that induces the release of BiP from bound polypeptides (4). Depletion of cellular ATP inhibits protein folding (11) and results in prolonged association of some proteins with BiP (12), whereas lowering ER calcium levels causes the release of T cell receptor subunits from BiP in vivo, presumably by inappropriately activating ATP hydrolysis (13). These results, together with the recent demonstration of ATP in the ER (14) and an ER ATP transport system (15), have provided evidence that the ATPase activity of BiP is important for its in vivo function. However, conditions that are optimal for in vitro ATP hydrolysis do not exist in the ER. In fact, ATP hydrolysis assayed in vitro is almost completely inhibited at pH 7.5 or with millimolar calcium concentrations (9). This has led investigators to propose that either co-stimulatory molecules exist, similar to those found in bacteria (16, 17), or that nucleotide binding, not hydrolysis, is sufficient for hsp70 activity (18).
The three-dimensional structure of an ATP-hydrolyzing fragment of hsc70, which comprises the amino-terminal 401 amino acids, was recently elucidated (19). BiP presumably forms a similar ATP-binding structure because mutation of BiP residues corresponding to those in hsc70 that are implicated in nucleotide interactions (20) severely inhibit the ATPase activity of BiP (21, 22). The BiP ATPase mutants bind immunoglobulin heavy chains but are not released in vitro by ATP (21, 22). The in vivo expression of mammalian BiP ATPase mutants (23) and yeast BiP (Kar2) mutants that map to the ATP binding domain (24) block the in vivo folding of substrate proteins, providing the first direct evidence for the involvement of BiP in protein folding and demonstrate the requirement for ATPase activity of BiP in this function.
Factor VIII is a complex plasma glycoprotein that is deficient in the X chromosome-linked bleeding disorder, hemophilia A. Although there are no known established cell lines that express factor VIII, the pathway for factor VIII secretion was elucidated from expression of the factor VIII cDNA in transfected mammalian cell lines (25). In these systems, factor VIII was inefficiently secreted as a consequence of interaction with BiP. Studies using sense and antisense BiP RNA expression vectors to elevate or reduce levels of BiP demonstrated that increased levels of BiP inhibited factor VIII secretion (26), while reduced levels of BiP improved factor VIII secretion (14, 27). The specific activity of factor VIII secreted from cells with reduced BiP levels was not detectably altered, indicating that factor VIII could fold into a biologically active form in cells with reduced BiP and suggesting that BiP interaction was not required for functional factor VIII secretion. Factor VIII release from BiP and transport out of the ER required high levels of intracellular ATP (12, 14). In contrast, the homologous coagulation protein, factor V, as well as unrelated proteins such as monocyte/macrophage colony stimulating factor (M-CSF) or von Willebrand factor, did not detectably associate with BiP, and their secretion was not inhibited by lowering cellular ATP levels (12, 28). These studies suggest that factor VIII displays a greater ATP requirement for dissociation from BiP and secretion compared to von Willebrand factor, factor V, and M-CSF.
Expression of BiP, as well as other members of the GRP gene family, is induced by the presence of unfolded proteins or unassembled protein subunits within the ER (29-31). A serine-threonine protein kinase was identified in Saccharomyces cerevisiae that is required for the transcriptional induction of the GRPs although the mechanism of kinase activation and the identity of its substrates are unknown (32, 33). Since transcription of the GRP gene family is induced upon stress conditions that result in the accumulation of unfolded protein in the ER, it is thought that members of this gene family may provide a protective mechanism to accomodate the presence of unfolded proteins. Transfection of cells with BiP antisense mRNA expression constructs or with multiple copies of the BiP glucose-regulated transcriptional element suppressed the induction of BiP without altering basal BiP levels. These cells also showed increased sensitivity to ionophore (34, 35), oxidative stress (36), hypoxia (37), and cell-mediated toxicity (38). Transcriptional induction of the GRPs in response to ER stress can also be blocked by overexpression of BiP (26). It is not known in this case if BiP protein alleviates the stress directly or interferes with the transcriptional induction of the GRP family. In this study, we show that overexpression of BiP prevents induction of GRP transcription and protects cells from ER stress, suggesting that BiP directly alleviates the ER stress. In addition, the ATPase activity of BiP was required for both of these activities. Furthermore, we studied the secretion of factor VIII and M-CSF in cells that overexpress the ATPase-defective BiP. The results show that M-CSF is not inhibited by the presence of ATPase-defective BiP and provide evidence for a BiP-ATPase independent polypeptide folding pathway for selective proteins.
The 1ADel
BiP deletes 27 amino acids from Tyr175-Glu201
from the 1A domain of the ATP binding cleft. The mutant cDNA
PstI (21) fragment was excised from pTZ and inserted into
the PstI site of the mammalian expression vector pED (39) to
derive pAD1. The wild-type BiP expression vector pEMCGRP78 was
described previously (26, 40). These vectors were electroporated into
DHFR-deficient Chinese hamster ovary (CHO) cells. Transformants were
selected for growth in the absence of nucleosides and the gene copy
number subsequently amplified by selection for growth in increasing
concentrations of methotrexate (41). Independently isolated clonal
lines were derived that express wild-type BiP (C.1 in 0.1 µM MTX and Cl.10 in 1 µM MTX) and 1ADel BiP
(AD-1-B in 0.1 µM MTX). Clonal cell lines were propagated
in complete medium lacking nucleosides and containing 10% fetal
bovine serum with penicillin and streptomycin. DHFR-deficient CHO
DUKX-B11 cells were propagated in the same medium containing thymidine,
deoxyadenosine, and adeosine. The derivation of the M-CSF (42) and
B-domain deleted factor VIII (43) expression vectors were previously
described.
Cells were plated at 200 or 2000/well into 6-well tissue culture plates in complete medium. After 24 h, cells were treated for 6 h with increasing concentrations of A23187. Cells were fed complete medium and incubated for 10-14 days. Colonies were counted after staining with crystal violet. The numbers represent the average from three independent plates and differ by less than 10%. For morphological examination, exponentially growing cells were treated with 5 µM A23187 in medium containing 10% fetal bovine serum for 4 h and were then analyzed by phase contrast microscopy and photographed using an Olympus BX60 microscope.
Exponentially growing cells were treated with A23187 in complete medium for 15 min at 37 °C. Cells were rinsed with methionine- and cysteine-free medium and then pulse-labeled in methionine- and cysteine-free Dulbecco's modified essential medium containing 50 µCi/ml [35S]methionine and [35S]cysteine (DuPont NEN) for 15 min in the presence of A23187. Cells were rinsed and harvested in Nonidet P-40 lysis buffer (44). Radioactive isotope incorporation was measured after trichloroacetic acid precipitation (44), and protein concentration was determined (45).
Transient Transfection and AnalysisControl DHFR-deficient CHO cells and AD1-B cells were transfected with pEMC-M-CSF or pMT2-LA VIII by the diethylaminoethyl dextran procedure for 4 h. At 24 h post-transfection, cells were fed with 5 ml/10-cm plate of fresh medium containing 5 mM sodium butyrate. After 18 h, some plates were treated with 10 µg/ml tunicamycin (Sigma) for 1 h. Cells were then rinsed with methionine-deficient minimal essential medium and pulse-labeled for 30 min with 100 µCi/ml [35S]-methionine (Amersham Corp.). The chase was performed by rinsing and feeding the cells with complete medium for 6 h for M-CSF and for 9 h for factor VIII. Conditioned medium was harvested, and cell extracts were prepared by lysis in Nonidet P-40 lysis buffer (30). Equal trichloroacetic acid (TCA) precipitable counts and corresponding volumes of labeled culture supernatants were immunoprecipitated with monoclonal antibodies directed against factor VIII coupled to Sepharose (F-8) (43) or against M-CSF (HM7/4.4.10) with rabbit anti-mouse and protein A-Sepharose as an immunoadsorbant. Immunoprecipitated proteins were analyzed by SDS-PAGE and autoradiography with En/HANCE (DuPont NEN). For Western blot analysis, equal amounts (20 µg) of cell extracts were electrophoresed on a reducing SDS-polyacrylamide gel and transferred to nitrocellulose. The blot was reacted with polyclonal anti-rat BiP (Stressgen Biotech Corp. Victoria, BC, Canada) and developed with goat anti-rabbit IgG conjugated to horseradish peroxidase as a secondary antibody followed by development with chloronapthol. Where indicated, CHO cells were treated with tunicamycin (1 µg/ml) for 12 h prior to preparation of cell extract.
Immunofluorescence and Electron MicroscopyCells were plated onto 35-mm tissue culture dishes containing sterile coverslips. After 48 h (before the cells reached confluency), the coverslips were removed, and the cells were fixed in 5% acetic acid, 95% ethanol. Cells were stained with polyclonal anti-BiP antibody (Affinity Bioreagents, Neshanic Station, NJ) followed by fluorescein isothiocyanate-conjugated goat anti-rabbit antibody (Southern Biotechnology Assoc., Birmingham, AL) as described (46). Tetramethylrhodamine isothiocyanate-conjugated wheat germ agglutinin (Sigma) was used to identify the Golgi compartment. Fluoresence was analyzed using dual excitation filters so that both colors (fluorescein and tetramethylrhodamine isothiocyanates) were viewed simultaneously. Fluorescence micrographs were taken with an Olympus fluorescence microscope equipped with a PM-30 automatic photomicrographic system.
Cells growing in microtiter polystyrene plates were fixed overnight in 2.5% glutaraldehyde in 0.1 M phosphate buffer and post-fixed for 2 h in 1% buffered osmium tetroxide. They were then dehydrated to absolute ethanol. In order to keep the cells as intact and undisturbed as possible, the cells were gently infiltrated with 1:3, 1:1, and then 3:1 parts of ethanol to 2-hydroxypropylmethacrylate (HPMA) each step for 4 h, resulting in final infiltration with pure HPMA. Then the cells were gently infiltrated with the same ratios of HPMA:Epon epoxy resin. The cells were finally infiltrated with pure Epon and polymerized in a 50 °C oven for 48 h. Ultrathin sections were double stained with uranyl acetate and lead citrate and examined with a Philips CM-100 transmission electron microscope.
Northern AnalysisLogarithmically growing cells were fed
medium containing 10% fetal bovine serum, 10 µg/ml tunicamycin
or 7 µM A23187 (both from Sigma) for
6 h. Total cellular RNA was isolated by the Trizol method (Life
Technologies, Inc.), and equal amounts of RNA were analyzed by
electrophoresis on 1% agarose gels containing 7.4% formaldehyde and
20 mM sodium phosphate, pH 6.5. The RNA was transferred to
nitrocellulose and hybridized with double-stranded random primed DNA
fragments in 5 × SSC (1 × SSC = 150 mM
sodium chloride, 15 mM sodium citrate) at 42 °C. The DNA
fragments used for hybridization were hamster BiP (1.6-kbp
PstI-EcoRI fragment from p3C5 (47)), chicken
-actin (1.6-kbp fragment from pA1 (48)), or hamster GRP94 (1-kbp
BamHI-EcoRV fragment from p4A3 (2)). Band
intensities were measured by PhosphorImager scanning using the
ImageQuant program (Molecular Dynamics).
Deletion of residues 175-201 (1ADel)
within hamster BiP destroyed the ATPase activity and prevented release
of bound peptides (21). To study the requirement of wild-type and
ATPase-defective BiP on protein secretion and the stress response, CHO
cell lines were derived that stably overexpress either wild-type
hamster BiP or the 1ADel BiP (Fig. 1). Analysis of two
independently isolated cell lines, C.1 and Cl.10, by Western
immunoblotting detected a significant increase in BiP protein levels
(approximately 5- and 10- fold, respectively) in cells transfected and
selected for expression of wild-type hamster BiP (Fig. 1, lanes
3 and 4) compared with the original CHO cells (Fig. 1,
lane 2). This level of BiP was similar to that observed
after treatment with tunicamycin (Fig. 1, lane 1). AD-1
cells expressed significant amounts of ATPase mutant BiP that migrated
with an approximately 3 kDa lower apparent molecular mass, as expected
from the 27-amino acid deletion (Fig. 1; lane 5). Expression
of 1ADel BiP also increased expression of the endogenous CHO cell
BiP.
The localization of the overexpressed BiP was monitored by
immunofluorescence with anti-BiP antibody using fluorescein-conjugated anti-rabbit antibody. The Golgi compartment was stained with
rhodamine-labeled wheat germ agglutinin and appeared normal in all cell
lines studied (Fig. 2A-C). Control CHO cells
demonstrated the presence of BiP within a reticular structure, the ER
(Fig. 2A). overexpressed wild-type BiP was localized to the ER although
at a much greater intensity (Fig. 2B) compared with control
CHO cells. In addition, some of the overexpressed wild-type BiP was
also detected as punctate spots within the nucleus. The 1ADel BiP
mutant was also detected in a reticular network but was not detected in
the nucleus. Additionally, the BiP mutant was localized to
membrane-enclosed vesicles in the cytoplasm (Fig. 2C),
similar to those previously noted for ATPase-defective BiP point
mutants expressed transiently in COS-1 cells (46). Electron microscopic
analysis demonstrated that the nuclear localization of wild-type BiP
correlated with the appearance of new membrane-bound vesicles, which
were likely the site of the overexpressed nuclear BiP (Fig.
2E). In contrast, the 1ADel BiP overexpression correlated
with cytoplasmic vesicles and the absence of nuclear vesicles (Fig.
2F, arrows).
Elevated BiP Inhibits Stress Induction of GRPs and Provides Survival in Response to ER Stress
The effect of wild-type and
mutant BiP overexpression on the induction of GRP transcription was
studied by treating cells with A23187, a calcium ionophore, or
tunicamycin, an inhibitor of N-linked glycosylation. Total
RNA was isolated and analyzed by Northern blot hybridization to a BiP
probe. The amount of RNA transferred to nitrocellulose was corrected
for by hybridization to an actin probe. BiP mRNA was induced 14- and 19-fold in control CHO cells upon treatment with A23187 and
tunicamycin, respectively (Fig. 3, lanes
1-3), and overexpression of wild-type BiP reduced the stress
induction to 1.3- and 1.8-fold, respectively (Fig. 3, lanes
4-6) similar to previous results (26). Overexpression of
ATPase-defective BiP increased the constitutive endogenous BiP mRNA
level by 2-fold (Fig. 3, lane 7); however, upon A23187 or
tunicamycin treatment, the mRNA level was further increased by 4.3- and 7-fold, respectively (Fig. 3, lanes 8 and 9).
These results indicate that elevated levels of a functional BiP with ATPase activity are required to inhibit the stress induction of BiP
transcription.
The effect of wild-type and mutant BiP overexpression on cell survival
upon calcium depletion from the ER after treatment with ionophore
A23187 was measured using a colony formation assay. Treatment of
control CHO cells with 5 µM A23187 reduced colony formation to approximately 30% of that observed in untreated cells (Fig. 4). In contrast, the two cell lines that
overexpress wild-type BiP were resistant to 10 µM A23187
treatment in this assay (Fig. 4; data not shown). Cells that express
ATPase-defective BiP were reproducibly slightly more sensitive to
A23187 than control CHO cells (Fig. 4) or CHO cells selected for MTX
resistance and expressing factor V, a large secreted glycoprotein (28).
These results show that overexpression of BiP confers protection to ER
stress upon calcium depletion and suggest that the ATPase activity of BiP is required for this activity.
The effect of overexpression of wild-type BiP on cell survival in
response to A23187 was dramatically evident by morphological examination after two h. Whereas cell death was observed in control CHO
cells and in ATPase mutant BiP-expressing cells, cells that overexpress
wild-type BiP did not exhibit altered morphology (Fig. 5). Under these conditions, cell death appeared to be
due to necrosis since characteristics of apoptotic cell death, DNA
fragmentation into a nucleosomal ladder, were not detected (data not
shown). Pretreatment of CHO cells with cycloheximide did not reduce
death by A23187 treatment (data not shown), indicating that protein synthesis was not required for the A23187-promoted cell death.
Overexpression of BiP Protects Against Inhibition of Protein Synthesis Mediated by A23187 Treatment
Agents that mediate ER
stress, such as ionophore or tunicamycin treatment, elicit an immediate
inhibition of protein synthesis initiation through phosphorylation of
the eukaryotic translation initiation factor 2 subunit (44, 49). We
asked whether overexpression of BiP can influence the inhibition of
protein synthesis in response to A23187 treatment. Cells were treated
with increasing concentrations of ionophore A23187 and then
pulse-labeled with [35S]methionine/cysteine in the
presence of ionophore. Treatment with 5 µM A23187
inhibited protein synthesis in control CHO cells and in ATPase-mutant
BiP-expressing CHO cells to 17 and 14% of untreated levels,
respectively (Fig. 6). In contrast, A23187 treatment of
the two wild-type BiP overexpressing cell lines inhibited protein synthesis only to 57 and 52% (Fig. 6), demonstrating that they were
partially resistant to this inhibition. Analysis of the spectrum of
proteins synthesized in the presence of A23187 did not detect translational rescue of any specific polypeptide(s) in response to
wild-type BiP overexpression, but rather, general cellular protein
synthesis was resistant to the translational inhibition (not shown).
These results show that the resistance to A23187 directly correlated
with protection of protein synthesis inhibition in response to
A23187.
BiP ATPase Activity Is Not Required for the Secretion of All Proteins
Previous studies demonstrated that overexpression of
wild-type BiP inhibited secretion of factor VIII but not of M-CSF (26). We therefore studied the effect of ATPase-defective BiP overexpression on the secretion of factor VIII and M-CSF. Control CHO or AD-1 cells
were transiently transfected with expression vectors encoding M-CSF
(42) or B domain-deleted factor VIII, which is encoded by an in-frame
deletion of 810 amino acids from the middle of the coding region
protein that does not alter the biological activity of the secreted
factor VIII (50, 51). B-domain-deleted factor VIII transiently
associates with BiP, whereas M-CSF association with BiP is not detected
(26). Cells were pulse-labeled with [35S]methionine/cysteine and chased in medium containing
excess unlabeled methionine. Samples of conditioned medium and cell
extracts were harvested for analysis by immunoprecipitation. Analysis
of B-domain deleted factor VIII detected the factor VIII primary
translation product in the extracts of CHO and AD-1 cells after the
pulse label (Fig. 7A, lanes 1 and
4). In addition, a small amount of a polypeptide,
representing BiP, could be detected in a complex with factor VIII.
Compared with the endogenous BiP, the ATPase-defective BiP was
preferentially associated with factor VIII in CHO cells expressing the
mutant BiP (Fig. 7A, lane 4). After a 9-h chase in the control CHO cells, the majority of factor VIII was detected in
the conditioned medium as species representing the single chain, a
heavy chain doublet, and a light chain doublet (Fig. 7A,
lane 3). In addition, a portion of factor VIII was retained
in the cell extract (Fig. 7A, lane 2). After the
chase period in the AD-1 cells, very little factor VIII was detected in
the conditioned medium (Fig. 7A, lane 6) but
rather was retained in the cell in a complex with BiP (Fig.
7A, lane 5). These results show that factor VIII
was not secreted from AD-1 cells and was retained in a complex with
BiP.
Analysis of M-CSF secretion from CHO cells detected the primary translation product in the cell extract after a 20-min pulse label (Fig. 7B, lane 1). This polypeptide was efficiently chased out of the cell into the conditioned medium (Fig. 7B, lanes 2 and 3). M-CSF was detected in the conditioned medium as the processed mature 40-kDa species as well as a multimeric heterogeneous glycosylated high molecular form. Analysis of M-CSF secretion in AD-1 cells also detected the primary translation product that was efficiently chased into the conditioned medium. Since the M-CSF primary translation product co-migrates with the 1ADel ATPase-deleted BiP, we studied the secretion of M-CSF in the presence of tunicamycin so that the M-CSF could be distinguished from the mutant BiP. Cells were treated with tunicamycin only transiently during the pulse radiolabeling and chase procedure so that the endogenous BiP protein levels were not significantly induced. Tunicamycin had no effect on the synthesis or secretion of M-CSF from CHO cells or AD-1 cells although the polypeptides migrated with increased mobility due to the absence of N-linked oligosaccharides (Fig. 7B, lanes 4-6). In the AD-1 cells, M-CSF was detected in the cell extract after the pulse-label and was not detected in the cells after the chase period. 1ADel deleted BiP was detected in the AD-1 cell extracts as a polypeptide that was nonspecifically precipitated with the anti-M-CSF antibody as demonstrated by immunoprecipitation of extracts from non-transfected AD-1 cells (not shown). These results show that M-CSF was efficiently secreted in cells that express the ATPase-defective BiP.
Protein translocation into, folding within, and transport out of the ER are essential steps in the biogenesis of secretory proteins, membrane proteins, and proteins destined to reside in various intracellular organelles. The factors that control these ER functions as well as regulate the synthesis and maintain the integrity of the ER are poorly understood. Numerous studies implicate BiP as providing a fundamental role in all these processes. BiP is likely the primary sensor to alterations in ER homeostasis, which result in the transcriptional induction of the genes encoding the ER-localized stress response proteins such as GRP94, ERp72, GRP170, calreticulin, as well as BiP itself (31, 52).
Yeast BiP is an essential gene (53, 54), and either inactivation or depletion of BiP results in the accumulation of untranslocated proteins in the cytosol (55). In vitro reconstitution of deficient yeast microsomes with purified BiP restores translocation (56), demonstrating that BiP is directly required for translocation. The translocating polypeptide is directly associated with BiP (57), and genetic studies revealed that yeast BiP interacts with sec63, an ER membrane protein with a DnaJ-like domain, to translocate nascent proteins across the ER membrane (58). Several attempts have been made to demonstrate a similar role for mammalian BiP in translocation (59-61), but thus far, the data have been contradictory and inconclusive. It this study, there is no evidence that expression of the BiP ATPase mutant has any effect on protein translocation. We did not detect shorter translation products or unglycosylated products for either factor VIII or M-CSF.
Overexpressed ATPase-defective mutant BiP was detected in membrane-enclosed vesicles in the cytoplasm that were similar to those previously described for BiP ATPase point mutants transiently expressed in COS monkey cells. This was previously attributed to disruption of the ER (46) and may be due to requirements for BiP function to maintain structural integrity of the ER or that the accumulation of unfolded proteins in the BiP mutant expressing cells formed aggregates resulting in the vesiculation (23). Consistent with this latter notion, we observed that overexpression of the mutant BiP induced the mRNA level for the endogenous BiP. BiP is known to be induced under conditions where unfolded protein accumulates in the ER (29, 31). In comparison with ATPase-deleted BiP overexpression, overexpression of wild-type BiP correlated with BiP-containing membrane-bound vesicles in the nucleus. This may indicate a normal physiological role for BiP in the nucleus that was uncovered by wild-type BiP overexpression, or it may indicate an aberrant process that occurs upon wild-type BiP overexpression in the absence of stress. These vesicles are remniscent of cytoplasmic granules that appeared upon forced hsp70 overexpression in Drosophila cells that apparently represented irreversibly inactivated hsp70 (62). Previous studies have identified a requirement of Kar2p for ER-membrane fusion (63). If BiP functions to promote membrane fusion events, then disruption of the BiP ATPase activity may prevent generation of BiP-containing vesicles in the nucleus.
BiP prevents aggregation of polypeptide folding intermediates, prevents secretion of unassembled, unfolded, or mutated proteins, and possibly directs abberant folded proteins to the degradative machinery. Investigations into the role of BiP in mammalian cells have supported two different, but not mutually exclusive, functions for BiP in protein folding. One hypothesis is that BiP assists protein folding by maintaining proteins in a conformation where they are folding competent (5, 64). This model is supported by the transient association of BiP with polypeptide folding intermediates destined for secretion (65-68) and an ATP dependence for proper folding, disulfide bond formation, and secretion (11, 12, 69). In contrast, other studies (26, 27, 70-72) suggest that BiP provides a retention mechanism for quality control to prevent unfolded proteins from exiting the secretory pathway and possibly enter the degradative pathway. However, despite many efforts to date, there is no direct demonstration that BiP binding and release from unfolded proteins actually catalyzes protein folding (73-75). However, recent evidence supports that BiP ATPase activity is required for immunoglobulin light chain folding and disulfide bond formation (23).
We studied secretion of two proteins that display different interactions with BiP in CHO cell lines that stably overexpress wild-type or ATPase defective BiP. B domain-deleted factor VIII transiently binds BiP but is secreted (27, 66, 76). In contrast, M-CSF is not detected in association with BiP and is very efficiently secreted (26). The secretion of B domain-deleted factor VIII was inhibited in cells expressing ATPase-defective BiP. The factor VIII translation product was detected in a stable complex with the mutant BiP. This is consistent with the conclusion that the ATPase-defective BiP was able to form a complex with factor VIII but release and secretion were blocked. However, since overexpression of wild-type BiP itself can inhibit secretion of factor VIII (26), we cannot attribute the block to factor VIII secretion to the loss of BiP ATPase activity. In addition, the molecular weight of the factor VIII indicated that the molecule was properly glycosylated and thus suggesting that the mutant BiP did not interfere with translocation or glycosylation of factor VIII. Analysis of M-CSF expression demonstrated that it was also efficiently synthesized, translocated, and glycosylated. However, in contrast to factor VIII, M-CSF was also efficiently secreted into the conditioned medium. Similarly, the observation that some immunoglobulin light chains can fold and be secreted in cells expressing BiP mutants supports the idea that the secretory pathway is not completely disrupted (23). The finding that cells expressing ATPase-defective BiP could efficiently synthesize, process, and secrete M-CSF suggests that all proteins do not require BiP function for secretion in mammalian cells. If M-CSF transiently associated with BiP and folding and release required ATP hydrolysis, then overexpression of mutant BiP should have trapped M-CSF in complexes. Such complexes were not observed, and M-CSF secretion was not inhibited. This indicates that M-CSF either does not associate with BiP or does not require ATP-dependent release. Analogous ATPase-defective mutants in the Escherichia coli BiP homologue DnaK act as dominant negative mutants when expressed with wild-type DnaK (77, 78). In this study, we show that a transfected BiP ATPase mutant is able to compete with endogenous BiP protein to bind nascent proteins in the ER. Because this binding appears to be irreversible (21, 46), we would expect that any nascent protein that the mutant BiP bound to should remain bound. Therefore, the most straightforward interpretation of the experiments on M-CSF secretion is that M-CSF never associated with BiP.
Previous studies have suggested that BiP may be required to protect cells from cellular stress. In these studies, BiP induction was inhibited by expression of antisense BiP mRNA (34, 37, 38), expression of BiP-targeted ribozymes (79), or amplification of the glucose core element upstream from the BiP promoter to titrate out transcription factors required for transcriptional activation (35, 80). However, all of these approaches also inhibited the expression of not only BiP but also of other GRP genes, so it was not possible to attribute the observed cell protection specifically to BiP induction (34, 35, 79). We have shown in the present study that overexpression of wild-type BiP can protect cells from ER stress induced by calcium depletion upon ionophore A23187 treatment and that the ATPase activity of BiP was required for this protection. Since BiP overexpression does not induce other GRP genes (26), it is most likely that BiP is directly acting to prevent or alleviate the stress. Analysis of calcium flux using fura-2 indicated no significant change in calcium flux in response to A23187 mediated by wild-type BiP overexpression compared with control CHO cells.2 These studies support the idea that calcium flux is not the sole mechanism triggering cytotoxicity upon ER stress. Taken together, the data presented here are the first to demonstrate that elevated BiP increases cell survival and that this may be mediated by directly alleviating the stress. We propose that ER stress generates unfolded proteins that may have hydrophobic patches exposed and that are susceptible to aggregation. Elevated levels of BiP could bind the exposed hydrophobic regions and prevent them from aggregation. The protection observed to ER stress mediated by BiP overexpression is analogous to the observed thermotolerance obtained by overexpression of hsp70, a homologue of BiP localized to the cytosol (81). The results show that BiP and hsp70 provide homologous protective functions.
Cells respond to malfolded proteins by induction of GRP expression (2)
and inhibition of protein synthesis (49). Upon calcium depletion from
the ER, protein folding is disrupted and protein synthesis is inhibited
by phosphorylation of the eukaryotic translation initiation factor
eIF-2 (82). The signal transduction pathways in response to ER
stress are rapidly being elucidated. Recently, it was demonstrated that
the double-stranded RNA-dependent protein kinase (PKR), a
kinase associated with the ribosomes of the rough ER, can phosphorylate
eIF-2
upon calcium depletion by ionophore, as well as by other ER
stress-inducing agents (44, 82). Studies in yeast implicate a central
role for a type 1 transmembrane Ser/Thr protein kinase
(IRE1/ERN1) that is activated by the presence of
unfolded proteins in the lumen of the ER (32, 33). IRE1 is a
non-essential gene under non-stress conditions that is required under
conditions where improperly folded proteins accumulate in the ER. The
results reported here demonstrate that overexpression of wild-type BiP,
but not ATPase defective BiP, can prevent activation of these two
kinase pathways, PKR and IRE1. Based on the putative role
for BiP in maintaining proteins in a folding competent state, we
propose that BiP prevents activation of PKR and IRE1 through
directly interacting with malfolded proteins and inhibiting their
aggregation. Future studies will be required to identify the mechanism
by which unfolded proteins activate these two stress-inducible
signaling pathways.
We thank Louise Wasley and Mariet Varban for excellent technical assistance.