Article
Cloning of mammalian Ire1 reveals diversity in the ER stress responses
Xiao-Zhong Wang1,
Heather P. Harding1,
Yuhong Zhang1,
Ethel M. Jolicoeur1,
Masahiko Kuroda1 and David Ron1
- Skirball Institute of Biomolecular Medicine, the Departments of Medicine, Cell Biology and the Kaplan Cancer Center, NYU Medical Center, New York, NY 10016, USA
Correspondence to:
David Ron, E-mail: ron@saturn.med.nyu.edu
Received 19 June 1998; Accepted 5 August 1998; Revised 31 July 1998
Abstract
Cells modify their gene expression pattern in response to stress signals emanating from the endoplasmic reticulum (ER). The well-characterized aspect of this response consists of the activation of genes that encode protein chaperones and other ER resident proteins, and is conserved between mammals and yeast. In mammalian cells, however, ER stress also activates other pathways, including the expression of the transcription factor CHOP/GADD153 and its downstream target genes. ER stress is also linked to the development of programmed cell death, a phenomenon in which CHOP plays an important role. Here we report on the cloning of a murine homolog of yeast IRE1, an essential upstream component of the ER stress-response in yeast. The mammalian Ire1 is located in the ER membrane and its over-expression in mammalian cells activates both the endogenous ER chaperone GRP78/BiP and CHOP-encoding genes. Over-expression of a dominant-negative form of Ire1 blocks the induction of GRP78/BiP and CHOP in response to the ER stress induced by tunicamycin treatment. Over-expression of murine Ire1 also leads to the development of programmed cell death in transfected cells. These results indicate that a single upstream component, Ire1, plays a role in multiple facets of the ER stress-response in mammalian cells.
Keywords:
- chaperones,
- endoplasmic reticulum,
- membrane protein,
- protein folding,
- protein kinase
Introduction
A significant number of noxious circumstances lead to accumulation of unfolded intermediates in the endoplasmic reticulum (ER) and trigger a stress response (Lee, 1987; Gething and Sambrook, 1992). This response, referred to as the unfolded protein response (UPR), was initially characterized in terms of the transcriptional induction of a set of mRNAs, all of which have in common the fact that they encode ER resident proteins that are thought to participate in the folding of newly synthesized proteins. These include the genes encoding the ER chaperones GRP78/BiP and GRP94, protein disulfide isomerase (PDI) and the PDI-like ERp72 (Kozutsumi et al., 1988; Dorner et al., 1990; Little et al., 1994). Defined in such terms, the UPR can be easily rationalized in terms of the need to rectify a perturbation in protein folding in the ER by increasing the synthesis of components that promote the process of proper folding in that compartment.
In the yeast Saccharomyces cerevisiae, the UPR is effected by a signaling cascade in which the upstream-most component known to date is an ER-associated type I transmembrane protein kinase, Ire1p/Ern1p (Cox et al., 1993; Mori et al., 1993), a protein that also bears considerable homology to the regulated endoribonuclease RNaseL (Bork and Sander, 1993; Sidrauski and Walter, 1997). Ire1p senses the perturbed environment in the ER, presumably through its lumenal domain. This leads to activation of downstream signaling by a process that is thought to depend on oligomerization and trans-phosphorylation of the kinase domain (Shamu and Walter, 1996; Welihinda and Kaufman, 1996). This leads to activation of a sequence-specific endoribonucleolytic activity, resident in the C-terminal non-lumenal side of Ire1p (Sidrauski and Walter, 1997). The latter cleaves a pre-formed precursor mRNA that encodes the transcription factor Hac1p at two sites, excising a small intron (Cox and Walter, 1996; Nikawa et al., 1996; Kawahara et al., 1998). The exons of Hac1 are joined together by a non-canonical splicing reaction that is brought about by tRNA ligase (Sidrauski et al., 1996). This leads to enhanced translation of the spliced Hac1 mRNA (Chapman and Walter, 1997; Kawahara et al., 1997). The product of this translation, Hac1p, then binds to a UPR response elements in the promoters of the yeast homolog of GRP78 (a gene known as KAR2) and two yeast PDI-like genes (Cox and Walter, 1996; Mori et al., 1998). Deficiencies in components of this pathway lead to increased sensitivity of the yeast to agents that induce the UPR, supporting the hypothesis that the UPR is geared to attenuating ER stress. In mammalian cells also, there is evidence favoring such a role. Over-expression of GRP78/BiP attenuates the cell-death-promoting effects of calcium ionophores that induce the UPR (Morris et al., 1997), and interfering with the induction of GRP78/BiP, by means of antisense constructs, renders the cells more sensitive to conditions that trigger a UPR (Little and Lee, 1995).
In mammalian cells, ER stress may also activate genes and pathways that are not directly linked to the expression of resident ER proteins. For example, treatment of cells with the glycosylation inhibitor tunicamycin, a potent inducer of the UPR, leads to the induction of growth arrest and programmed cell death in many cell types (Larsson et al., 1993; Pérez-Sala and Mollinedo, 1995; Dricu et al., 1996). A similar response is observed in cells that harbor temperature-sensitive mutations in components of the protein glycosylation pathway (Nakashima et al., 1993; Silberstein et al., 1995): switch to the non-permissive temperature leads to the induction of the UPR and triggers growth arrest and programmed cell death. These events could be due to distant downstream consequences of the perturbation in ER function. However, recent evidence suggests that this outcome may be due to an active process directly triggered by the UPR: the gene encoding the transcription factor CHOP (also known as GADD153) is not normally expressed in cells but is potently induced by stress. Initially thought to be responsive to DNA-damage and growth arrest (Fornace et al., 1989), subsequent studies have uncovered a strong correlation between the development of ER stress and induction of CHOP. Known inducers of the UPR, such as tunicamycin, thapsigargin, the calcium ionophore A23187 and reducing agents such as dithiothreitol, all induce CHOP (Price and Calderwood, 1992; Halleck et al., 1997). CHOP is also induced by depriving cells of nutrients such as glucose and oxygen, a situation that is also associated with the induction of a UPR (Price and Calderwood, 1992; Carlson et al., 1993).
More direct evidence favoring a link between a primary ER signal and CHOP induction was provided by the observation that GRP78/BiP over-expression attenuates CHOP induction in response to stress (Wang et al., 1996). At the physiological level, over-expression of CHOP attenuates the process of adipocytic differentiation in 3T3-L1 cells (Batchvarova et al., 1995) and a gain of function mutation in CHOP is invariably associated with human myxoid liposarcoma (Crozat et al., 1993; Rabbitts et al., 1993). Under some circumstances, CHOP over-expression can lead to cell-cycle arrest (Barone et al., 1994), and can promote programmed cell death (Friedman, 1996). Knock-out of the CHOP gene in mouse cells and tissues leads to decreased programmed cell death in response to agents that induce the UPR (Zinszner et al., 1998). Collectively, these observations suggest a role for CHOP in modulating cellular phenotypes in response to ER stress. Therefore, it is of considerable interest to study the link between CHOP induction and other, more classical facets of the UPR. Here we describe an analysis of the role of one newly identified component of that signaling cascade, mammalian Ire1. Our results suggest that in mammalian cells, the UPR is not limited to the induction of ER resident proteins and it may exert pleiotrophic effects on cellular phenotypes through the induction of components such as CHOP.
Search of the sequence databases with the yeast Ire1p sequence revealed a predicted Caenorhabditis elegans homolog (Q09499). Based on their similarity to that protein, we identified two different ESTs that encode potential mammalian homologs of Ire1p, one murine and one human. The murine EST was used to isolate a full-length coding region clone from a murine colon cDNA library. The encoded protein of 911 amino acids has an N-terminal hydrophobic stretch of 18 residues followed by a 403 amino acid sequence that exhibits 33.5% identity to the lumenal domain of C.elegans Ire1p and 19.9% identity to the yeast protein. A non-conserved 19 residue hydrophobic stretch (aa 429–447), predicted to serve as a transmembrane domain, precedes the C-terminal region. The latter is 40% identical to the kinase/endonuclease effector domain of yeast and C.elegans Ire1p (Figure 1A and B), and 20% identical to human RNaseL (not shown). The primary amino acid sequence of the clone is consistent with it being a mammalian homolog of yeast Ire1p. In both man and mouse, the gene encoding Ire1 is immediately adjacent to that of polo-like kinase (plk), and thus maps to the mouse locus at 59 cM on chromosome seven. The human gene was independently mapped on a radiation hybrid panel to the syntenic region on chromosome 16 (Figure 1C).
The domain organization of mIre1 is consistent with that of a type I transmembrane protein. An expression plasmid that encodes for the full-length protein, tagged at its C-terminus with a Myc epitope, was transfected into COS1 cells. Immunostaining of permeabilized cells with the tag-specific monoclonal antibody, 9E10, revealed a fine reticular cytoplasmic pattern that overlapped closely with the ER marker, ribophorin 2 (Figure 1D). This finding is consistent with the encoded protein being an ER resident protein, as proposed for yeast Ire1p. Our assay does not detect the endogenous protein, and therefore we can not formally exclude the possibility that retention in the ER is an artifact due to over-expression of a protein that is normally targeted to another exocytic compartment. However, we note similar expression patterns in different cells types (see Figure 5B) and across varying levels of expression including very low levels, supporting but not proving the physiological relevance of this localization.
Yeast Ire1p, when over-expressed in yeast cells, triggers the activation of the UPR and increases the expression of Kar2, the yeast homolog of GRP78/BiP (Cox et al., 1993; Mori et al., 1993). We therefore tested the ability of mIre1 over-expression to trigger similar increase in mammalian GRP78/BiP expression. 293T cells transfected with an expression plasmid for Ire1 exhibit a substantial increase in GRP78/BiP mRNA (Figure 1E). This increase is still significantly less than that observed in response to tunicamycin, a potent inducer of the UPR. This assay, together with the subcellular localization and the primary amino-acid sequence homology strongly supports a functional homology between the mouse protein and yeast Ire1p.
To determine whether over-expression of mIre1 can induce CHOP mRNA, the aforementioned mIre1 expression plasmid was transfected into COS1 cells and the cells were stained with a CHOP-specific antiserum. mIre1-expressing cells expressed high levels of CHOP protein, whereas adjacent non-expressing cells stained negative for CHOP (Figure 2A). The ability of mIre1-overexpression to activate endogenous CHOP was confirmed in two other assays. A Northern blot of RNA from 293T cells over-expressing mIre1 revealed strong induction of endogenous CHOP mRNA (Figure 2B, lane 3). Over-expression of immunoglobulin 
heavy chain, that is retained in the ER in the absence of its corresponding light chain and can even induce the UPR under some circumstances (Lenny and Green, 1991), does not induce CHOP expression in our system (Figures 2A, panels h, i and j, and 2B, lane 4). The immunoglobulin heavy chain and mIre1 closely co-localize when expressed in the same cell, supporting the validity of this control (Figure 2A, panels f and g). A membrane-spanning derivative of Ire1, truncated at the C-terminus and lacking the effector domain, when expressed at similar levels, is likewise incapable of activating CHOP (Figure 3A; data not shown). These results indicate that the induction of CHOP by Ire1 is not simply a consequence of the over-expression of an ER lumenal protein. Next, mIre1 was over-expressed in a Chinese hamster ovary (CHO) cell-line that contains an integrated CHOP–green fluorescent protein (GFP) reporter transgene; activity of GFP increased significantly (Figure 2C). This suggests that the effect of mIre1 on CHOP expression is mediated at a transcriptional level.
In yeast, over-expression of a truncated form of Ire1p that lacks a functional effector domain has been shown to interfere with the propagation of the signal leading to the induction of KAR2 (Mori et al., 1993; Shamu and Walter, 1996). To determine if the lumenal domain of mIre1, when severed from the C-terminal putative effector domain, could exhibit similar dominant-negative effects on the mammalian UPR, an expression plasmid encoding a derivative of mIre1 that was truncated and tagged immediately C-terminal of the trans-membrane domain (Ire1-
C) was transfected into COS1 cells. The truncated protein is expressed at high level and is retained in the ER. Treatment of transfected cells with tunicamycin followed by double-staining for CHOP and Ire1-C, consistently revealed lower levels of CHOP in cells that express Ire1-C (Figure 3A, compare cells marked by arrows with adjacent unmarked cells). Similar results were observed in 293T cells transfected with Ire1-C and treated with tunicamycin; levels of CHOP and GRP78/BiP induction in these cells are 
50% lower than in cells transfected with empty plasmid (Figure 3B). These experiments indicate that the lumenal domain of mIre1, when over-expressed in cells, can interfere with the activity of the endogenous inducer(s) of the UPR.
The results presented so far do not permit one to distinguish between a situation in which mIre1 over-expression activates the UPR by triggering a lumenal event that is then transduced through the endogenous signaling cascade, and a situation in which the effector domain of the over-expressed mIre1 protein participates in propagating the signal. To address this issue, we over-expressed the isolated C-terminal domain of mIre1 in COS1 cells and measured its ability to induce endogenous CHOP expression. Two derivatives of the protein were prepared: Gal4–Irec, in which the C-terminus was fused to the DNA-binding domain of the yeast DNA-binding protein GAL4 (a domain that is known to impart nuclear localization; Silver et al., 1984), and GST–Irec in which the same domain of mIre1 was fused to Schistosoma japonica glutathione S-transferase (a cytoplasmic protein). Both fusion proteins were expressed at high levels in the transfected COS1 and 293T cells. Surprisingly, both were retained in the cytoplasm and both were able to potently activate the endogenous CHOP and, to a lesser degree, the endogenous GRP78/BiP genes (Figure 4A and B). The level of induction by the two proteins differed between COS1 and 293T cells, most probably reflecting differences in their expression level in these two cell types. While it is impossible to exclude that a small portion of GST–Irec or Gal4–Irec make their way into the nucleus, this result suggests that mIre1 can execute its role as a transducer of ER stress through a cytoplasmic signaling node. This conclusion holds true regardless of the mechanism by which these truncated proteins activate CHOP expression.
A mutant form of GST–Irec containing the K536A mutation was fully active in this assay (data not shown). However, a truncation that removes the C-terminal endonuclease homology domain of mIre1 significantly attenuated but did not entirely abolish the induction of CHOP (Figure 4C). In yeast, a similar deletion of the C-terminus of Ire1p resulted in a protein that was unable to rescue Kar2 induction in a ire strain (Shamu and Walter, 1996). All other GST fusion proteins tested, including GST alone, fusions between GST and the cytosolic stress-activated kinases SEK and SAPK
as well as a fusion between the stress-activated and apoptosis-inducing transcription factor ATF3 were expressed at similar or higher levels than GST–Irec and did not induce endogenous CHOP (Figure 4C). These experiments artificially uncouple CHOP induction from a primary ER event and render it highly unlikely that the gene is induced through the activity of an ER stress transducer distinct from Ire1. This conclusion is valid regardless of whether the over-expressed cytosolic derivatives described above effect Ire1 action directly or if they do so by activating the endogenous protein.
In the course of conducting the over-expression experiments described above we noted the presence of many floating and crenated cells in the plates transfected with the mIre1 expression plasmid, but not in the control plates. This phenomenon was more pronounced in the CHO cells than in the COS1 or 293T cells. We also noted on the FACScan that the mIre1-expressing cells were smaller than average (compare the forward scatter with the level of GFP-expressing cells in Figure 2C), suggesting that they may be undergoing programmed cell death. In addition, we were unable to recover clones of cells stably over-expressing mIre1, but were able to recover clones over-expressing the truncated form of the protein (data not shown). Noting that conditions known to trigger the UPR are also associated with the induction of programmed cell death (see Introduction), we examined the potential role of mIre1 in that process. To this end, a CHO cell-line expressing a LacZ reporter gene under the control of the CHOP regulatory sequences was used. Incorporation of the mIre1 expression plasmid leads to induction of LacZ and blue staining of the fixed cells with X-gal. When we compared the morphology of the blue cells in a plate transfected 48 h earlier with mIre1 with that of the same cells expressing LacZ from a transfected LacZ-expression plasmid (CMV–LacZ), we noted that the blue cells in the Ire1-transfected plate were smaller and rounded-up, protruding from the plane of the cellular monolayer. The CMV–LacZ transfected cells, on the other hand, were flat and otherwise indistinguishable from the neighboring non-transfected cells (Figure 5A). To examine the nuclear morphology associated with these changes we fixed the transfected CHO cells, and stained them simultaneously with the Myc-tag antibody to reveal the mIre1-expressing cells and the karyophilic dye H33258 to stain the chromatin. mIre1-expressing cells were reproducibly smaller and exhibited condensed chromatin, a feature of programmed cell death. Cells transfected with the control -chain expression plasmid had normal morphology and chromatin pattern (Figure 5B).
To validate these results, the Ire1 cDNA was co-expressed with a membrane-targeted form of green fluorescent protein, farnesylated enhanced fluorescence protein (GFPfarn). CHO cells transfected with this plasmid were stained with propidium iodide, and analyzed for DNA content and expression of GFPfarn by dual channel FACScan. As controls, the CHO cells were left untransfected or transfected with the empty GFPfarn vector. The cells co-expressing GFPfarn and Ire1 contained a significant sub-population with reduced DNA content ('hypo-diploid' cells), consistent with cells undergoing programmed cell death. A small increase in the proportion of such dying cells is also observed in the population transfected with the empty GFPfarn vector and is presumably due to the toxicity of the Lipofectamine transfection procedure (Figure 5C). These results support a role for mIre1 in the induction of programmed cell death in response to conditions that trigger the UPR.
The mIRE1 cDNA described here is a good candidate for encoding a mammalian ortholog of yeast Ire1p: beyond the similarity in primary amino-acid sequence, the similar predicted domain structure and membrane topology, the experimental results indicate that mIre1 is an ER-resident protein that is capable of activating GRP78/BiP expression in the absence of other signals when over-expressed in mammalian cells. Its ability to also activate transcription of the CHOP gene provides evidence for the existence of a common signaling node that coordinates at least two different read-outs for ER stress: one that culminates in the induction of ER chaperones and another that induces non-ER resident proteins such as the transcription factor CHOP. The CHOP knock-out mice have normal inducibility of GRP78/BiP (Zinszner et al., 1998), indicating that the two responses represent a bifurcation of one signaling pathway downstream of IRE1, as opposed to one pathway that operates in a linear fashion. The fact that CHOP is itself required for the activation of a set of genes that are induced by the UPR (Wang et al., 1998), implies that in mammalian cells Ire1 controls a rather diverse and arborized stress response. Based on the yeast model, this dual regulation could come about if both the GRPs and CHOP are co-regulated by the mammalian Hac1 equivalent(s). However, the data presented here by no means exclude a more complex relationship between Ire1 and CHOP. For example, Ire1 may induce a secondary modification in ER structure or function which then leads to CHOP activation through a parallel signaling pathway initiating from the ER. The hypothesized role of yeast Ire1p in regulating endomembrane biogenesis (Cox et al., 1997) implies that signaling downstream of Ire1 may modulate aspects of ER function that go beyond the mere production of ER chaperones. These might include modifications that impact on CHOP induction. The identification of the hypothesized mammalian Hac1-equivalent(s) will hopefully resolve the details of the epistatic relationship between IRE1 and CHOP.
One facet of the complexity of the UPR, demonstrated here, is the association between activated Ire1 signaling and the induction of cell death. This is a feature that is likely to be unique to metazoans and can be rationalized in terms of the fact that ER stress is often associated with conditions that are severely damaging to cells. Therefore, linking the ER stress response to programmed cell death through Ire1 may be beneficial under certain circumstances. The observation that CHOP plays a role in promoting programmed cell death, suggests that it may serve as link between Ire1 signaling and the induction of that response (Zinszner et al., 1998).
The mechanism of induction of CHOP and GRP78/BiP by mIre1 over-expression is not clear. In yeast it has been proposed that over-expression of Ire1p leads to a constitutive dimer/oligomer formation even in the absence of ER stress. This, in turn, promotes trans-phosphorylation of Ire1p and activates the endonucleolytic activity of the protein which then signals downstream. Supporting this model is the observation that a lysine to alanine mutation in the conserved ATP binding lobe of the kinase domain of Ire1p abolishes the ability of the over-expressed yeast protein to constitutively activate the UPR (Cox et al., 1993; Mori et al., 1993; Shamu and Walter, 1996). The same mutation also abolishes the ability of the protein to rescue a null allele of yeast IRE1 and the expressed protein may even have dominant-negative activity on the pathway (Mori et al., 1993), again pointing to a pivotal role for the kinase activity in yeast Ire1p function. Surprisingly, an mIre1 derivative, GST–Irec (Figure 4B and C) that bears a homologous mutation, K536A, is at least as active as the wild-type protein in terms of inducing CHOP and GRP78/BiP when over-expressed in mammalian cells (data not shown). Assuming that these truncated derivatives of mIre1 function through the activity of their effector domain, these findings suggest that over-expression in mammalian cells somehow abrogates the requirement for trans-phosphorylation.
The mammalian system described here also reveals an additional feature of Ire1 signaling, namely the possibility of uncoupling the C-terminal effector domain of the protein from its regulatory lumenal domain. The transient transfection paradigm is particularly well suited to this kind of experiment because it is not encumbered by the negative selection that would probably impact the stable expression of a constitutively active Ire1 fragment. Surprisingly, over-expression of the effector domain of mIre1 in the cytoplasm elicits high-level activation of CHOP. At the very least, the results presented here suggest that the signaling pathway can also be accessed from the cytoplasmic side. This conclusion holds true regardless of whether the GST–Irec fusion protein activates the UPR by mimicking activated Ire1 and splicing a mammalian Hac1-like substrate (as suggested by the diminished activity of the truncated protein that lacks the endonuclease portion of mIre1; Figure 4C, lane 3), or by competing for an inhibitory factor or de-localizing the endogenous mIre1 (as suggested by the residual activity of the endonuclease deleted GST–Irec nuc fusion protein; Figure 4C, lane 3). The recent identification of a phosphatase that can bind to and dephosphorylate yeast Ire1p and attenuate the UPR (Welihinda et al., 1998) provides one plausible mechanism by which such dominant negative effects of GST–Irec might take place. Either way, a component(s) of the signaling pathway is likely to be cytosolic. Even a model whereby the substrate mRNA resides in the cytoplasm and is cleaved there by Ire1 is not so heretical, given that in yeast, the ligase that effects the last step of the splicing reaction, tRNA-ligase, is localized to the nuclear pore (Clark and Abelson, 1987) and can presumably be accessed from both sides of that structure.
Our studies also hint at the possibility that the non-lumenal domain of mIre1 may positively interact with a cytoplasmic factor: it has previously been demonstrated that the Gal4 DNA-binding domain serves as a dominant nuclear localization domain that leads to the translocation of fusion proteins to the nucleus (Silver et al., 1984). The Gal4–Irec fusion was made in an effort to promote stronger signaling by localizing the effector domain of mIre1 to that compartment, and was based on the prevailing model that stipulates a nuclear site of action for the Ire1 endonuclease. Unexpectedly, that fusion protein is selectively retained in the cytoplasm and activates the downstream target genes no better than GST–Irec. It will be interesting to learn if the C-terminal domain of mIre1 is capable of leading to the cytoplasmic retention of other nuclear import substrates. Finally, it is important to point out that all the conclusions presented here have been drawn from genetic gain-of-function over-expression experiments. A more definitive understanding of the role of mIre1 in the mammalian UPR will come from studying deficiency mutations in that gene.
During the review of this manuscript, a paper describing the cloning of a human homolog of yIRE1 was published (Tirasophon et al., 1998). Analysis of the EST database and cloning of a partial cDNA of the mouse orthologue of that human gene informs us that mice and humans have at least two IRE1 genes. The murine locus corresponding to the gene described in the aforementioned manuscript is now referred to as Ern1 and the locus containing the gene described here is Ern2.
cDNA library construction, cDNA cloning and chromosomal mapping of Ire1
An oligo-dT primed murine colon cDNA library was constructed according to the manufacturer's instructions in a 
-ZapII vector (Stratagene). The unamplified library was screened by hybridizing to the insert of the mouse EST clone vo50h02, obtained from Genome Systems. Thirty-seven positive clones were identified among 4 million insert-containing recombinants screened (implying an mRNA abundance of 1:100 000). The cDNA was sequenced from several overlapping clones to generate a full-length coding region sequence that has been deposited in the DDBJ/EMBL/GenBank (accession No. AF071777). To map the chromosomal localization of the human Ire1 gene, a pair of primers based on the sequence of the 3' end of the human EST clones z189c06 and ng38c07 were designed (sense: 5'-CACAAG ACTGGC AGAGGC TGGGCA-3'; antisense: 5'-CTGTCC TTTCCT TGGCTT TATGCA CA-3'). When used under the following PCR conditions; 94°C (5') 
[94°C (1), 55°C (1), 72°C (1)] X30 72°C (10), these amplify the human but not the hamster allele. These were used to screen the GeneBridge 4 radiation hybrid panel (Research Genetics) and the following vector was obtained: 00000 00101 00000 01001 11000 00110 00100 01010 11100 10001 20000 01110 01101 00200 01100 00000 00111 11001 000, corresponding to the address presented in Figure 1C. The 3' ends of the aforementioned human ESTs also contain a perfect match to the last 55 base pairs from the extreme 3' end of the human PLK cDNA, suggesting a tail-to-tail orientation of IRE1 and PLK on human chromosome 16. Three different genomic PACs containing the murine Ire1 gene also specifically hybridized to the plk cDNA, supporting a similar arrangement in the mouse genome (data not shown). plk is mapped in the mouse to chromosome 7 at 59 cM (Clay et al., 1997), a region syntenic to the address obtained here for the human IRE1.
Expression vector construction
Mouse Ire1 was tagged with a Myc epitope (9E10) at its C-terminus by overlapping PCR (Squinto et al., 1990) using the following primers: 'mIre(c).9E10', 5'-GAGATC AGCTTC TGCTCG CCTCCG CTCTTT GTGGCA TCTG-3'; and '9E10. 3'U', 5'-GGGGCT CGAGTC ACAGAT CCTCCT CAGAGA TCAGCT TCTGCT C-3'. The full-length coding region of mIre1 with the C-terminal 9E10 tag was cloned as EcoRI–XhoI fragment into pCDNA3 (Invitrogen). The same strategy was to used to generate a truncated form of mIre1, 'Ire1-C'. The overlapping primer used in PCR; 'mIre(1574).9E10', 5'-GAGATC AGCTTC TGCTCC CCACGG CCCAGC ACA-3', truncates the protein at Ala518, just 3' of the transmembrane domain. The EcoRI–XhoI fragment encoding the mIre1 ER lumenal domain and transmembrane domain was then cloned into EcoRI–SalI site of pBabe-puro or pBabe-EGFP. To generate the in-frame fusion of mIre 1 cytoplasmic domain with Gal4 DNA-binding domain or GST, XbaI (in-frame with the SpeI site of pEBG; Sánchez et al., 1994) and EcoRI sites (in-frame with EcoRI site of pSG424; Sadowski and Ptashne, 1989) were introduced by PCR into the mIre1 cDNA at a position corresponding to aminoacid Ser453 using the primer 'mIre.1380.XbaI–EcoRI', 5'-GAGGTC TAGAGA ATTCAG CCCTTC AGCACC TGCT-3' and the primer '9E10. 3'U', as described above. The corresponding fragment was then cloned into the indicated recipient plasmids. The HindIII–XbaI fragment containing the entire protein coding region of Gal4–Ire1c was then transferred into pCDNA3 to achieve higher level of expression. The GST–Ire1c deleted of the nuclease domain, GST–Ire1c nuc, was constructed by deleting the 3' end of mIre1 at BamHI2196. This truncates the mIre1 protein at amino acid Asp725. The GST–SEK and GST–SAPK expression plasmids (Sánchez et al., 1994) were a gift of Len Zon (Harvard) and the GST–ATF3 plasmid has been previously described (Wang et al., 1998). The plasmid used for co-expressing Ire1 and GFPfarn was constructed in a pCDNA3 backbone. Briefly, because wild-type Aequorea victoria GFP is extracted from cells by the permeabilization step required for propidium iodide (PI) staining (Kalejta et al., 1997), we tagged the enhanced GFP cDNA (Clontech) with the 20 C-terminal residues of Ha-Ras which encode the farnesylation sequence and target the GFPfarn to the plasma membrane, leading to its retention during PI staining. The GFPfarn coding region was used to replace the neor gene in pCDNA3, creating pCDNA3.GFPfarn. The Ire1 coding region was introduced into this plasmid as an EcoRI–XhoI fragment, downstream of the CMV promoter.
RNA isolation and Northern blot analysis
Total mouse colon RNA for cDNA synthesis was prepared by GTC lysis and CsCl ultra centrifugation. Poly(A)+ RNA was isolated by two passes over an oligo-dT cellulose column (Molecular Research Center, Inc.). For Northern blot analysis, total RNA (20 g/lane) was fractionated on a 1.2% formaldehyde agarose gel and transferred onto HyBond-N nylon membrane (Amersham). The 32P murine CHOP, BiP and 
-tubulin cDNA fragments were hybridized to the blot as previously described (Wang et al., 1998).
Cell culture, transfection and treatment
COS-1, 293T and CHO K1 cells were originally obtained from ATCC. They were cultured in Dulbecco's modified Eagle's medium (DMEM) or DMEM:F12 (CHO cells) in the presence of 10% fetal calf serum (Atlanta Biological). The construction of the CHOP–GFP and CHOP–LacZ CHO reporter cells used in Figures 2C and 5A will be described in detail elsewhere (H.P.Harding and D.Ron, in preparation). The murine CHOP genomic fragment used was 8.5 kb in length and its 3' end corresponds to the PmlI site in exon 3, nine nucleotides 5' to the initiation methionine of CHOP. Cells were treated with tunicamycin (1 g/ml) for 3 h. 5' CHOP–LacZ reporter CHO K1 cells were treated with 2 M thapsigargin for 2 h. Tunicamycin and thapsigargin were purchased from Sigma (St Louis, MO). 293T cells were transfected by the calcium phosphate precipitation method using 7 g plasmid DNA for each 60 mm plate. COS-1 cells were transfected by DEAE-dextran method using 3 g of plasmid DNA per 60 mm plate. CHO K1 cells carrying the GFP– or LacZ–CHOP-reporter genes were transfected with 1 g plasmid DNA per 35 mm plate by the Lipofectamine plus method (Gibco-BRL). Cells were harvested at 48 h after transfection for either immunocytochemistry, Western blot, Northern blot or FACScan analysis as previously described (Wang et al., 1998). Transfection efficiency was monitored by FACScan analysis of the GFPfarn expression, either as a marker in bi-cistronic expressing vector or by co-transfection of CMV–EGFP marker. Propidium iodide analysis of DNA content was performed by FACScan, as previously described (Zinszner et al., 1998).
Antibodies, Western blot and immunocytochemistry
The various constructs of Ire1 were tagged with a Myc epitope at their C-terminus, replacing the stop codon by the tag. The expression of the recombinant protein was detected by the murine anti-human Myc monoclonal antibody 9E10 (Evan et al., 1985). To detect endogenous CHOP in COS-1 cells, rabbit anti-CHOP polyclonal antiserum (Ron and Habener, 1992) was used in all experiments except Figure 2A (bottom panel) and Figure 4A, where the murine anti-CHOP monoclonal antibody 9C8 (Batchvarova et al., 1995) was used to avoid either the cross-reactivity of the secondary antibodies (Figure 2A, bottom) or reactivity of the rabbit anti-CHOP with GST (Figure 4A and B). Rabbit anti-ribophorin 2 polyclonal antibody (a gift of Gert Kreibich, NYU, USA) was used at a dilution of 1:200 to visualize the ER staining pattern. To reveal the various GST fusion proteins in Figure 4C, rabbit GST-reactive antisera was used for Western blot analysis. To detect protein expression at the single-cell level, cells were fixed with 3.7% formaldehyde phosphate-buffered saline (PBS) for 5 min, and permeabilized with 0.25% Triton X-100 for 5 min and methanol for 5 min. The cells were then stained for the presence of protein with the indicated antibodies.
We thank Michael Green, Len Zon and Gert Kreibich for reagents, and Ed Skolnik, David Levy, Jan Sap, Steve Hubbard and members of their labs and ours for suggestions, advice and critique. This project was supported by research grants ES08681 and DK47119, and training grant CA73181 to H.P.H. D.R. is a Stephen Birnbaum Scholar of the Leukemia Society of America.
Barone MV, Crozat AY, Tabaee A, Philipson L and Ron D (1994) CHOP (GADD153) and its oncogenic variant, TLS-CHOP, differ in their ability to induce G1/S arrest. Genes Dev, 8, 453–464. | Article | PubMed | ISI | ChemPort |
Batchvarova N, Wang X-Z and Ron D (1995) Inhibition of adipogenesis by the stress-induced protein CHOP (GADD153). EMBO J, 14, 4654–4661. | PubMed | ISI | ChemPort |
Bork P and Sander C (1993) A hybrid protein kinase-RNase in an interferon-induced pathway? FEBS Lett, 334, 149–152. | Article | PubMed | ISI | ChemPort |
Carlson SG, Fawcett TW, Bartlett JD, Bernier M and Holbrook NJ (1993) Regulation of the C/EBP-related gene, gadd153, by glucose deprivation. Mol Cell Biol, 13, 4736–4744. | PubMed | ISI | ChemPort |
Chapman RE and Walter P (1997) Translational attenuation mediated by an mRNA intron. Curr Biol, 7, 850–859. | Article | PubMed | ISI | ChemPort |
Clark MW and Abelson J (1987) The subnuclear localization of tRNA ligase in yeast. J Cell Biol, 105, 1515–1526. | Article | PubMed | ISI | ChemPort |
Clay FJ, Ernst MR, Trueman JW, Flegg R and Dunn AR (1997) The mouse Plk gene: structural characterization, chromosomal localization and identification of a processed Plk pseudogene. Gene, 198, 329–339. | Article | PubMed | ISI | ChemPort |
Cox JS, Chapman RE and Walter P (1997) The unfolded protein response coordinates the production of endoplasmic reticulum protein and endoplasmic reticulum membrane. Mol Biol Cell, 8, 1805–1814. | PubMed | ISI | ChemPort |
Cox JS, Shamu CE and Walter P (1993) Transcriptional induction of genes encoding endoplasmic reticulum resident proteins requires a transmembrane protein kinase. Cell, 73, 1197–1206. | Article | PubMed | ISI | ChemPort |
Cox JS and Walter P (1996) A novel mechanism for regulating activity of a transcription factor that controls the unfolded protein response. Cell, 87, 391–404. | Article | PubMed | ISI | ChemPort |
Crozat AY, Åman P, Mandahl N and Ron D (1993) Fusion of CHOP to a novel RNA-binding protein in human myxoid liposarcoma with t (12;16) (q13;p11). Nature, 363, 640–644. | Article | PubMed | ISI | ChemPort |
Dorner AJ, Wasley LC, Raney P, Haugejorden S, Green M and Kaufman RJ (1990) The stress response in Chinese hamster ovary cells. Regulation of ERp72 and protein disulfide isomerase expression and secretion. J Biol Chem, 265, 22029–22034. | PubMed | ISI | ChemPort |
Dricu A, Carlberg M, Wang M and Larsson O (1996) Inhibition of N-linked glycosylation using Tunicamycin causes cell death in malignant cells: role of down regulation of the Insulin-like growth factor 1 receptor in induction of apoptosis. Cancer Res, 57, 543–548. | ISI |
Evan G, Lewis G, Ramsay G and Bishop J (1985) Isolation of monoclonal antibodies specific for human c-myc proto-oncogene product. Mol Cell Biol, 5, 3610–3616. | PubMed | ISI | ChemPort |
Fornace AJ, Neibert DW, Hollander MC, Luethy JD, Papathanasiou M, Fragoli J and Holbrook NJ (1989) Mammalian genes coordinately regulated by growth arrest signals and DNA-damaging agents. Mol Cell Biol, 9, 4196–4203. | PubMed | ISI | ChemPort |
Friedman AD (1996) GADD153/CHOP, a DNA damage-inducible protein, reduced CAAT/enhancer binding protein activities and increased apoptosis in 32D c13 myeloid cells. Cancer Res, 56, 3250–3256. | PubMed | ISI | ChemPort |
Gething MJ and Sambrook J (1992) Protein folding in the cell. Nature, 355, 33–45. | Article | PubMed | ISI | ChemPort |
Halleck MM, Holbrook NJ, Skinner J, Liu H and Stevens JL (1997) The molecular response to reductive stress in LLC-PK1 renal epithelial cells: coordinate transcriptional regulation of gadd153 and grp78 genes by thiols. Cell Stress Chaper, 2, 31–40. | Article | ISI | ChemPort |
Kalejta RF, Shenk T and Beavis AJ (1997) Use of a membrane-localized green fluorescent protein allows simultaneous identification of transfected cells and cell cycle analysis by flow cytometry. Cytometry, 29, 286–291. | Article | PubMed | ISI | ChemPort |
Kawahara T, Yanagi H, Yura T and Mori K (1997) Endoplasmic reticulum stress-induced mRNA splicing permits synthesis of transcription factor Hac1p/Ern4p that activates the unfolded protein response. Mol Biol Cell, 8, 1845–1862. | PubMed | ISI | ChemPort |
Kawahara T, Yanagi H, Yura T and Mori K (1998) Unconventional splicing of HAC1/ERN4 mRNA required for the unfolded protein response. Sequence-specific and non-sequential cleavage of the splice sites. J Biol Chem, 273, 1802–1807. | Article | PubMed | ISI | ChemPort |
Kozutsumi Y, Segal M, Normington K, Gething MJ and Sambrook J (1988) The presence of malfolded proteins in the endoplasmic reticulum signals the induction of glucose-regulated proteins. Nature, 332, 462–464. | Article | PubMed | ISI | ChemPort |
Larsson O, Carlberg M and Zetterberg A (1993) Selective killing induced by an inhibitor of N-linked glycosylation. J Cell Sci, 106, 299–307. | PubMed | ISI | ChemPort |
Lee AS (1987) Coordinate regulation of a set of genes by glucose and calcium ionophores in mammalian cells. Trends Biochem Sci, 12, 20–23. | Article | ISI | ChemPort |
Lenny N and Green M (1991) Regulation of endoplasmic reticulum stress proteins in COS cells transfected with immunoglobulin heavy chain cDNA. J Biol Chem, 266, 20532–20537. | PubMed | ISI | ChemPort |
Little E and Lee AS (1995) Generation of a mammalian cell line deficient in glucose-regulated protein stress induction through targeted ribozyme driven by a stress-inducible promoter. J Biol Chem, 270, 9526–9534. | Article | PubMed | ISI | ChemPort |
Little E, Ramakrishnan M, Roy B, Gazit G and Lee AS (1994) The glucose-regulated proteins (GRP78 and GRP94): functions, gene regulation and applications. Crit Rev Eukaryot Gene Expr, 4, 1–18. | PubMed | ChemPort |
Mori K, Ma W, Gething MJ and Sambrook J (1993) A transmembrane protein with a cdc2+/CDC28-related kinase activity is required for signaling from the ER to the nucleus. Cell, 74, 743–756. | Article | PubMed | ISI | ChemPort |
Mori K, Ogawa N, Kawahara T, Yanagi H and Yura T (1998) Palindrome with spacer of one nucleotide is characteristic of the cis-acting unfolded protein response element in Saccharomyces cerevisiae. J Biol Chem, 273, 9912–9920. | Article | PubMed | ISI | ChemPort |
Morris JA, Dorner AJ, Edwards CA, Hendershot LM and Kaufman RJ (1997) Immunoglobulin binding protein (BiP) function is required to protect cells from endoplasmic reticulum stress but is not required for the secretion of selective proteins. J Biol Chem, 272, 4327–4334. | Article | PubMed | ISI | ChemPort |
Nakashima T, Sekiguchi T, Kuraoka A, Fukushima K, Shibata Y, Komiyama S and Nishimoto T (1993) Molecular cloning of a human cDNA encoding a novel protein, DAD1, whose defect causes apoptotic cell death in hamster BHK21 cells. Mol Cell Biol, 13, 6367–6374. | PubMed | ISI | ChemPort |
Nikawa J, Akiyoshi M, Hirata S and Fukuda T (1996) Saccharomyces cerevisiae IRE2/HAC1 is involved in IRE1-mediated KAR2 expression. Nucleic Acids Res, 24, 4222–4226. | Article | PubMed | ISI | ChemPort |
Pérez-Sala D and Mollinedo F (1995) Inhibition of N-linked glycosylation induces early apoptosis in human promyelocytic HL-60 cells. J Cell Physiol, 163, 523–531. | Article | PubMed | ISI
Price B and Calderwood S (1992) Gadd45 and Gadd153 messenger RNA levels are increased during hypoxia and after exposure of cells to agents which elevate the levels of glucose-regulated proteins. Cancer Res, 52, 3814–3817. | PubMed | ISI | ChemPort |
Rabbitts TH, Forster A, Larson R and Nathan P (1993) Fusion of the dominant negative transcription regulator CHOP with a novel gene FUS by translocation t (12;16) in malignant liposarcoma. Nature Genet, 4, 175–180. | Article
Ron D and Habener JF (1992) CHOP, a novel developmentally regulated nuclear protein that dimerizes with transcription factors C/EBP and LAP and functions as a dominant negative inhibitor of gene transcription. Genes Dev, 6, 439–453. | Article | PubMed | ISI | ChemPort |
Sadowski I and Ptashne M (1989) A vector for expressing GAL4 (1–147) fusions in mammalian cells. Nucleic Acids Res, 17, 7539. | Article | PubMed | ISI | ChemPort |
Sánchez I, Hughes R, Mayer B, Yee K, Woodget J, Avruch J, Kyriakis J and Zon L (1994) Role of SAPK/ERK kinase-1 in the stress-activated pathway regulating transcription factor c-Jun. Nature, 372, 794–797. | PubMed | ISI | ChemPort |
Shamu CE and Walter P (1996) Oligomerization and phosphorylation of the Ire1p kinase during intracellular signaling from the endoplasmic reticulum to the nucleus. EMBO J, 15, 3028–3039. | PubMed | ISI | ChemPort |
Sidrauski C and Walter P (1997) The transmembrane kinase Ire1p is a site-specific endonuclease that initiates mRNA splicing in the unfolded protein response. Cell, 90, 1031–1039. | Article | PubMed | ISI | ChemPort |
Sidrauski C, Cox JS and Walter P (1996) tRNA ligase is required for regulated mRNA splicing in the unfolded protein response. Cell, 87, 405–413. | Article | PubMed | ISI | ChemPort |
Silberstein S, Collins P, Kelleher D and Gilmore R (1995) The essential OST2 gene encodes the 16-kD subunit of the yeast oligosaccharyltransferase, a highly conserved protein expressed in diverse eukaryotic organisms. J Cell Biol, 131, 371–383. | Article | PubMed | ISI | ChemPort |
Silver PA, Keegan LP and Ptashine M (1984) Amino terminus of the yeast GAL4 gene product is sufficient for nuclear localization. Proc Natl Acad Sci USA, 81, 5951–5955. | Article | PubMed | ChemPort |
Squinto SP, Aldrich TH, Lindsay RM, Morrissey DM, Panayotatos N, Bianco SM, Furth ME and Yancopoulos GD (1990) Identification of functional receptors for ciliary neurotrophic factor on neuronal cell lines and primary neurons. Neuron, 5, 757–766. | Article | PubMed | ISI | ChemPort |
Tirasophon W, Welihinda AA and Kaufman RJ (1998) A stress response pathway from the endoplasmic reticulum to the nucleus requires a novel bifunctional protein kinase/endoribonuclease (Ire1p) in mammalian cells. Genes Dev, 12, 1812–1824. | Article | PubMed | ISI | ChemPort |
Wang X-Z, Lawson B, Brewer J, Zinszner H, Sanjay A, Mi L, Boorstein R, Kreibich G, Hendershot L and Ron D (1996) Signals from the stressed endoplasmic reticulum induce C/EBP homologous protein (CHOP/GADD153). Mol Cell Biol, 16, 4273–4280. | PubMed | ISI | ChemPort |
Wang X-Z, Kuroda M, Sok J, Batchvarova N, Kimmel R, Chung P, Zinszner H and Ron D (1998) Identification of novel stress-induced genes downstream of chop. EMBO J, 17, 3619–3630. | Article | PubMed | ISI | ChemPort |
Welihinda AA and Kaufman RJ (1996) The unfolded protein response pathway in Saccharomyces cerevisiae. Oligomerization and trans-phosphorylation of Ire1p (Ern1p) are required for kinase activation. J Biol Chem, 271, 18181–18187. | Article | PubMed | ISI | ChemPort |
Welihinda AA, Tirasophon W, Green SR and Kaufman RJ (1998) Protein serine/threonine phosphatase Ptc2p negatively regulates the unfolded-protein response by dephosphorylating Ire1p kinase. Mol Cell Biol, 18, 1967–1977. | PubMed | ISI | ChemPort |
Zinszner H, Kuroda M, Wang X, Batchvarova N, Lightfoot RT, Remotti H, Stevens JL and Ron D (1998) CHOP is implicated in programmed cell death in response to impaired function of the endoplasmic reticulum. Genes Dev, 12, 982–995. | Article | PubMed | ISI | ChemPort |
