The Stress Response in Chinese Hamster Ovary Cells REGULATION OF ERp72 AND PROTEIN DISULFIDE ISOMERASE EXPRESSION AND

Expression of the glucose-regulated proteins (GRPs), GRP’78 and GRP94, is induced by a variety of stress conditions including treatment of cells with tunica- mycin or the calcium ionophore A23 187. The stimulus for induction of these resident endoplasmic reticulum (ER) proteins appears to be accumulation of misfolded or underglycosylated protein within the ER. We have studied the induction of mRNAs -encoding two other resident ER proteins, ERp72 and protein disulfide iso- merase (PDI), during the stress response in Chinese hamster ovary cells. ERp72 shares amino acid sequence homology with PDI within the presumed cata- lytic active sites. ERp72 and, to a lesser degree, PDI were induced by treatment of Chinese hamster ovary cells with tunicamycin or A23187. These results identify ERp72 as a member of the GRP family. Stable high level overproduction of ERp72 or PDI from recombinant expression vectors did not alter the stitutive of other GRPs. High level overexpression resulted in secretion of the over- produced protein specifically but not other resident ER proteins. This suggests that the ER retention mechanism is mediated by more specific interactions than just KDEL sequence recognition.

This suggests that the ER retention mechanism is mediated by more specific interactions than just KDEL sequence recognition.
Native disulfide bond formation occurs in the endoplasmic reticulum (ER)' during and shortly after translocation of secretory proteins into that compartment. A resident protein of the ER, protein disulfide isomerase (PDI), is involved in the formation of disulfide bonds (for review, see Freedman (1984)). In vitro experiments using reduced and denatured proteins have shown that PDI increases the rate of formation of native disulfide bonds and catalyzes the rearrangement of non-native bonds (for review, see Freedman et al., 1989). Recently, it has been shown that the presence of PDI within dog pancreas microsomes during in vitro protein synthesis is necessary for disulfide bond exchange reactions to occur (Bulleid and Freedman, 1988).
Other resident ER proteins may also be involved in the processing and transport of secretory proteins within that compartment. Proteins which are major components of the ER of murine plasmacytoma cells have been identified (Lewis * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 USC. Section 1734 solely to indicate this fact.
$ To whom correspondence should be addressed.
The amino acid sequence KDEL present at the carboxyl terminus of many resident ER proteins is required for efficient retention of these proteins within that compartment (Munro and Pelham, 1987). GRP78, GRP94, and PDI have KDEL at their carboxyl termini, while ERp72 has the related sequence KEEL (Mazzarella and Green, 1987;Mazzarella et al., 1990;Munro and Pelham, 1986). The presence of this sequence may serve as a generalized signal for retention in the ER. However, the retention mechanism is not absolute for all KDEL-containing proteins (Zagouras and Rose, 1989).
Since GRPs are postulated to interact with proteins transiting the ER, especially under conditions of misfolding or aggregation, it is possible that other resident ER proteins may also be members of the GRP family. It has been suggested but not proven that GRP58 may be the same protein as PDI (Whelan and Hightower, 1985). ERp72, a luminal ER protein of murine plasmacytoma cells, has been shown to share amino acid sequence homology with PDI at regions believed to constitute the active sites of PDI (Mazzarella et al., 1990). The induction of proteins with disulfide isomerase activity, along with the induction of GRP78 and GRP94, may be part of the cellular response to the presence of misfolded protein in the ER.
We have examined the induction of ERp72 and PDI mRNAs following exposure of Chinese hamster ovary (CHO) cells to tunicamycin or the calcium ionophore A23187, treatments which induce GRP expression (Lee, 1987). ERp72 mRNA and protein were induced following treatment with A23187 or inhibition of N-linked glycosylation with tunicamycin. PDI mRNA was also induced by both treatments but at a much reduced level compared with ERp72. These results suggest that regulation of expression of ERp72 and perhaps PDI is similar to that of GRP78 and GRP94 and indicate that ERP72 is a member of the GRP family.
To study the potential role of increased expression of ERp72 or PDI, these proteins were constitutively expressed at high levels in CHO cells. Overexpression of either protein did not alter the expression or stress-mediated induction of other GRPs. Interestingly, overexpression of PDI or ERp72 resulted in detectable secretion only of the overexpressed protein rather than of all KDEL-containing proteins. This result suggests that retention of resident ER proteins may involve protein-specific signals in addition to the KDEL retention mechanism. Further studies will utilize these overexpressing cells to help define the roles of ERp72 and PDI in secretion and the stress response independently of increased expression of the other GRPs. medium cell extracts were prepared (Dorner et al., 1987)   . All restriction enzymes were obtained from New England Biolabs. Cell Lines-The DHFR-deficient cell line CHO DUKX-Bll designated CHO which requires nucleosides in the medium for growth has been described elsewhere (Urlaub and Chasin, 1980). The DHFRdeficient CHO DUKX cell line exnressing the elucocorticoid recentor designated CHO-GRA was derived as pre;ious& described (Israel'and Kaufman, 1989). Cell Line Deriuation-CHO-GRA cells were transfected with pMT2-PDI or pMT2-72 by the calcium phosphate precipitation procedure and selected for DHFR expression in increasing concentrations of methotrexate (Kaufman, 1990 (Chirgwin et al., 1979). RNA (6 pg) was electrophoresed on 1% agarose gels containing 7.4% formaldehyde and 20 mM sodium phosphate, pH 6.5, transferred to nitrocellulose, and hybridized to nick-translated DNA fragments at high stringency (5 X SSC, 65 "C) by standard techniques (Maniatis et al., 1982). DNA restriction fragments used as probes were: hamster GRP78, 1.6-kbp PstI-EcoRI fragment from p3C5 (Ting et al., 1987); hamster GRP94,  (Bole et al., 1986) or anti-ERp72 polyclonal rabbit antiserum (Lewis et al., 1985) as described elsewhere (Dorner and Kaufman, 1990 Assay for PDI AC&&y-The insulin reduction assay was used to measure PDI activity (Tomizawa, 1962 (Chang et al., 1987;Olden et al., 1979). To ascertain whether PDI or ERp72 mRNAs were also induced by this stress condition, CHO cells were treated with tunicamycin for 8 or 24 h. Total cellular RNA was subjected to Northern analysis, and the levels of mRNAs for GRP78, GRP94, PDI, and ERp72 were determined. The level of GRP78 mRNA increased 33-fold after 8 h of treatment with tunicamycin and 28-fold after 24 h of treatment as measured by hybridization to a GRP78-specific probe (Fig. L4). GRP94 mRNA levels, as detected with a GRP94-specific probe, increased 16-and 29-fold following 8-and 24-h tunicamycin treatment, respectively (Fig. 1B).
ERp72 mRNA increased lo-fold after 8-h treatment and 13-fold after 24-h treatment as measured by hybridization to ERp72 D.
-mm PDI an ERp'lZ-specific probe (Fig. 1C). Hybridization with a PDIspecific probe showed an increase in PDI mRNA of only 3fold after 8 h and 4-fold after 24-h treatment with tunicamycin (Fig. 1D). Actin mRNA levels, measured by hybridization to an actin-specific probe, were used to quantitate the equivalence of the gel load and normalize quantitation of the level of induction. Actin mRNA levels were equivalent except for a slight decline at the 24-h tunicamycin time point (data not shown). These results showed that ERp72 mRNA and, to a lesser extent, PDI mRNA were induced by tunicamycin treatment as were GRP78 and GRP94 mRNAs ( Table I).
The calcium ionophore A23187 affects the permeability of cellular membranes to calcium. Previous studies have shown that the GRP78 and GRP94 genes are activated at the transcriptional level by treatment with A23187, resulting in increased levels of mRNA for these proteins (Resendez et al., 1985). We examined whether PDI and ERp72 mRNA levels could also be elevated by A23187 treatment. CHO cells were treated for 6 or 12 h with A23187. Northern analysis of total RNA was performed. Hybridization with a GRP78-specific probe revealed a 20-and 30-fold increase in the level of GRP78 mRNA after 6 and 12 h of treatment, respectively (Fig. 24). The GRP94 mRNA level was increased 5-fold after 6 h and 11-fold after 12 h of A23187 treatment as detected by hybridization with a GRP94-specific probe (Fig. 2B).
ERp72 mRNA levels showed 3.6-and 5-fold increases after 6-and 12-h treatment, respectively (Fig. 2C). After 6-h treatment, PDI mRNA levels were not increased and a 2-fold induction of PDI mRNA occurred after 12-h treatment (Fig.  20). Thus, ERp72 mRNA was induced similarly to GRP78 and GRP94 mRNAs after tunicamycin or A23187 treatment, while PDI mRNA showed only moderate induction after tunicamycin treatment but not A23187 treatment (Table I).
Protein Induction after Tunicamycin or A23187 Treatment-To determine whether the induction of ERp72 mRNA correlated with increased ERp72 protein synthesis, ERp72 protein synthesis was analyzed by immunoprecipitation of extracts of [35S]methionine pulse-labeled CHO cells treated with tunicamycin for 24 h, treated with A23187 for 12 h, or left untreated. For comparison and as a positive control for induction, GRP78 protein levels were also examined. Immunoprecipitation with an anti-BiP (GRP78) antibody showed loo-fold increased GRP78 synthesis after A23187 treatment (Fig. 3, lane 2) or tunicamycin treatment (Fig. 3, lane 3) compared with constitutive levels (Fig. 3, lane 1)). Immunoprecipitation with an anti-ERp72 antibody showed 6.5-fold increased levels of ERp72 synthesis after A23187 treatment (Fig. 3, lane 5) and 13-fold increased levels after tunicamycin treatment (Fig. 3, lane 6)) compared with constitutive levels (Fig. 3, lane 4). Thus, under these conditions ERp72 protein synthesis was increased along with GRP78 protein synthesis consistent with the mRNA induction.
Overexpression of Murine ERP72 and PDI in CHO Celk-Cells overexpressing ERp72 or PDI independently of the other GRPs were generated by transfection of CHO-GRA cells with expression vectors pMT2-PDI or pMT2-72. RNA transcribed from the adenovirus major late promoter on these vectors contains the adenovirus tripartite leader at the 5' end, ERp72 or PDI coding sequences, followed by a DHFR coding sequence within the 3' end of the mRNA. This dicistronic mRNA expresses sufficient DHFR to allow selection of transfected DHFR-deficient CHO-GRA cells by growth in the absence of nucleosides . Cells harboring amplification of the transfected DNA were obtained by growth in increasing concentrations of methotrexate (Kaufman, 1990).
Transfected CHO-GRA cells selected in 0.02 FM methotrexate were examined by immunoblot analysis for stable expression of ERp72 and PDI. Equal amounts of cell extract were probed with rabbit antisera directed against murine ERp72 or PDI. The cell line designated 72Al transfected with pMT2-72 displays elevated levels of ERp72 protein (Fig. 4A with the endogenous level of CHO-GRA cells (Fig. 4A, lane 3). The cell line designated 59A6 transfected with pMT2-PDI displays elevated levels of PDI protein (Fig. 4A, lane 2) compared with CHO-GRA cells (Fig. 4A,  lane 3). Elevated levels of ERp72 or PDI could also be detected by Coomassie Blue stain consistent with the immunoblot analysis (data not shown).
We next examined the levels of all four resident ER proteins in the overproducing cell lines. Equal amounts of cell extracts and corresponding volumes of 24-h conditioned medium were immunoblotted using a combination of antibodies which recognizes ERp72, PDI, GRP78, and GRP94. Examination of cell extracts revealed that 72Al cells overexpressing ERp72 (Fig. 4B, lane 2) or 59A6 cells overexpressing PDI (Fig. 4B, lane 1) did not display elevated levels of GRP78 or GRP94 compared with CHO-GRA cells (Fig. 4B, lane 3). Thus, overexpression of ERp72 or PDI did not alter intracellular expression levels of other GRPs. We also examined whether the retention mechanism for KDEL-containing proteins could be perturbed by overexpression of ERp72 or PDI resulting in secretion of these proteins. In the conditioned medium of 72Al cells only ERp72 was detected and not GRP78, GRP94, or PDI (Fig. 4B, lane 5). Similarly, only PDI was detected in the conditioned medium of 59A6 cells (Fig. 4B, lane 4). NO KDEL-containing proteins were detected in the conditioned medium of CHO-GRA cells (Fig. 4B, lane 6). While secretion of all GRPs was not observed, overexpression resulted in detectable secretion specifically of the overexpressed protein.
PDI Activity in 59A6 and 72Al Cells-Disulfide isomerase activity can be measured by the disruption of the disulfide bonds between the two chains of insulin by the release of the fi chain as trichloroacetic acid-soluble material (Tomizawa, 1962). Homogenates of 72A1,59A6, and CHO cells were mixed with '251-labeled insulin and the release of p chain monitored (Table II). 72Al cells overexpressing ERp72 did not show increased reduction activity. In contrast, a 5-fold increase in reduction activity per mg of protein was observed in PDI overexpressing 59A6 cells compared with CHO-GRA cells.
Stress Response of Overexpressing Cells-We next examined whether constitutive overexpression of a single GRP would alter the induction of the other GRPs following stress elicited by tunicamycin treatment. 72A1, 59A6, and CHO-GRA cells were treated for 6 h with tunicamycin and total cellular RNA analyzed by Northern blotting. Hybridization to a GRP78specific probe showed a 14-fold induction of GRP78 mRNA in CHO-GRA cells (Fig. 5A, lanes 1 and 2), a 12-fold induction in 59A6 cells ( (Fig. 5A, lanes 3 and 4), and an II-fold induction in 72Al cells (Fig. 5A, lanes 5 and 6). Hybridization to a GRP94-specific probe showed that the level of induction of GRP94 mRNA was also similar in 59A6 cells (4.6-fold) (Fig .   TABLE II PDI assay of CHO cells Cell homogenates were prepared from parental CHO-GRA, 72A1, and 59A6 cells. Assay for PDI activity was the insulin reduction assay in which the PDI-mediated disruption of the disulfide bonds of insulin is monitored by the trichloroacetic acid solubility of free fl chain (Tomiazwa, 1962). '?-Insulin was incubated with cell homogenates (10 pg) at 37 "C. After incubation the reaction was precipitated with trichloroacetic acid and the trichloroacetic acid-soluble material counted in a y counter. Specific activity was determined by dividing nanomoles of insulin degraded by the amount of total cell homogenate (10 pg) used in the assay. 5B, lanes 3 and 4) and 72Al cells (3.7-fold) (Fig. 5B, lanes 5 and 6) compared with CHO-GRA cells (4-fold) (Fig. 5B, lanes  1 and 2). The level of actin RNA was used to quantitate the induction level (Fig. 5C). Thus, stable overexpression of ERp72 or PDI had no effect on the level of induction of either GRP78 or GRP94. Constitutive mRNA levels for GRP78 and GRP94 were similar between all cell lines except for a less than e-fold increase in GRP78 mRNA observed in 72Al cells.

DISCUSSION
Expression of GRPs is increased by conditions which block normal processing and secretion of proteins (Lee, 1987). These conditions result in the accumulation of misfolded or aggregated protein in the ER. We therefore examined whether other resident ER proteins, PDI and ERp72, were coordinately induced with GRP78 and GRP94. PDT facilitates disulfide bond interchange and thus is involved in protein folding in the ER. PDI contains two homologous regions related to amino acid sequences of thioredoxin (Edman et al., 1985). These regions of PDI are believed to constitute the active sites involved in disulfide bond interchange (Freedman et al., 1989). ERp72 contains three copies of this conserved sequence, and the spacing between two copies is nearly identical to that found in PDI (Mazzarella et al., 1990). While a disulfide isomerase activity has not yet been demonstrated for ERp72, this homology is suggestive that it may also facilitate disulfide bond formation to promote protein folding or unfolding.
Since PDI and ERp72 may play a role in protein folding in the ER, we studied whether they are induced by stress conditions which result in the accumulation of misfolded protein in the ER. Treatment of CHO cells with tunicamycin or the calcium ionophore A23187 produced a significant increase in ERp72 mRNA levels. The -fold induction of ERp72 mRNA was comparable with that observed for GRP94 mRNA. We observed that, while PDI mRNA levels were moderately increased after tunicamycin treatment, little induction was observed after A23187 treatment. Previously it has been reported that PDI protein levels are not increased after A23187 treatment (Mater and Koch, 1986). Increases in protein levels are usually disproportionately less than the induction level of mRNA for the GRPs* (Watowich and Morimoto, 1988), and so the 2-fold mRNA induction after A23187 treatment which we observed may not produce a detectable increase in PDI protein. We did observe increased ERp72 protein synthesis levels after A23187 and tunicamycin treatment. Thus, in the case of ERp72, induction of mRNA correlated with increased levels of protein.
Our data indicate that ERp72 mRNA levels may be regulated in a similar manner to GRP78 and GRP94 after tunicamycin or A23187 treatment. These observations indicate that ERp72 is a member of the GRP family. Since induction by calcium ionophore treatment is a distinctive feature of GRPs (Lee, 1987), we cannot definitively include PDI in the GRP family. PDI levels are elevated in normal cells engaged in high levels of synthesis of secreted proteins (Freedman, 1989),' but we observed little stress induction. This suggests that PDI functions during synthesis of normal proteins to form correct disulfide bonds but that increased levels of PDI are not part of the cellular response to the accumulation of misfolded protein. In contrast, ERp72 is induced similarly to other GRPs by stress conditions. Since ERp72 displays PDI homology, we speculate that ERp72 may interact with misfolded or denatured protein in the ER as part of a refolding 'A. J. Dorner, L. C. Wasley, P. Raney, and R. J. Kaufman, unpublished observations. or unfolding process after stress.
To further identify the functions and activities of ERp72 and PDI, we have generated CHO cell lines which overexpress the murine proteins individually. As measured by the ability to disrupt the disulfide bonds of the insulin heterodimer and release the p chain (Tomizawa, 1962), overexpression of ERp72 in 72Al cells did not result in increased insulin reduction activity in cell homogenates. We expect that a disultide isomerase activity of ERp72 may be observed under different assay conditions and exhibit specificities different than PDI. Alternate assay conditions are presently being evaluated to measure ERp72 activity. Homogenates of 59A6 cells displayed increased levels of insulin reduction activity which correlated with the increased level of PDI protein. This shows that the expressed PDI has an appropriate activity following overexpression in CHO cells.
With these cell lines we have initiated studies on the effect of overexpression of ERp72 or PDI in the absence of increased levels of the other GRPs. As detected by immunoblot analysis, murine ERp72 was expressed at a very elevated level in 72Al cells compared with the endogenous CHO protein. This high level of ERp72 did not result in increased expression of GRP78, GRP94, or PDI. In addition, high levels of ERp72 do not appear to be toxic and can be maintained by the cell in the absence of stress conditions. Immunoblot analysis of 59A6 cells showed high constitutive levels of PDI could also be maintained by the cell without inducing a stress response. Thus, stable overexpression of a single resident ER protein does not signal the cell to increase expression of other resident proteins. In particular, overexpression of ERp72, a GRP, did not induce expression of other GRPs. This suggests that the level of individual stress proteins in the ER is not part of the induction signal. Furthermore, the presence of high levels of ERp72 or PDI did not trigger an induction of GRPs due to misfolding or disruption of normal ER processes.
It is of interest that only ERp72 could be detected in the conditioned medium of 72Al cells and only PDI detected in the conditioned medium of 59A6 cells. It is possible that high level expression of a protein normally retained in the ER results in saturation of the retention mechanism and a low level of secretion of all KDEL-containing proteins. If the retention mechanism were a generalized one involving all KDEL-containing proteins (Monro and Pelham, 1987), then it would follow that saturation by overexpression of ERP72 or PDI would also result in secretion of a proportion of GRP78 and GRP94. Instead we have observed only secretion of the overexpressed protein. This suggests that the retention mechanism for ER-localized proteins may be more selective than just recognition of the KDEL sequence. This is consistent with a recent study in which addition of the KDEL sequence to two secretory proteins resulted in a retardation but not inhibition of secretion (Zagouras and Rose, 1989). Additional control over retention and secretion may also involve proteinprotein interactions or other signals specific for each protein.
Stable overexpression of a GRP has not previously been reported. We examined whether elevated levels of ERp72 or PDI would have an effect on the strength of the cellular stress response by measuring RNA levels for GRP94 and GRP78 following tunicamycin treatment. If an elevated level of PDI or ERp72 could reduce the amount of misfolded protein, then the induction of GRPs might be reduced. The induction levels of GRP94 and GRP78 were similar for the overexpressing cells compared with the parental CHO cells. This suggests that increased levels of either PDI or ERp72 alone could not mitigate the stress of blocking N-linked glycosylation by tunicamycin treatment. In addition, overexpression did not Stress Response in Chin kese Hamster Ovary Cells perturb the normal signaling mechanism which induces GRP expression. These cell lines will be further utilized as hosts for the expression of mutant and normal secreted proteins to determine the functions of ERp72 and PDI in the cell's secretory pathway under normal and stress conditions.