The eIF2α (eukaryotic initiation factor-2α) kinase PERK (doublestranded RNA-activated protein kinase-like ER kinase) is essential for the normal function of highly secretory cells in the pancreas and skeletal system, as well as the UPR (unfolded protein response) in mammalian cells. To delineate the regulatory machinery underlying PERK-dependent stress-responses, gene profiling was employed to assess global changes in gene expression in PERK-deficient MEFs (mouse embryonic fibroblasts). Several IE (immediate-early) genes, including c-myc, c-jun, egr-1 (early growth response factor-1), and fra-1 (fos-related antigen-1), displayed PERK-dependent expression in MEFs upon disruption of calcium homoeostasis by inhibiting the ER (endoplasmic reticulum) transmembrane SERCA (sarcoplasmic/ER Ca2+-ATPase) calcium pump. Induction of c-myc and egr-1 by other reagents that elicit the UPR, however, showed variable dependence upon PERK. Induction of c-myc expression by thapsigargin was shown to be linked to key signalling enzymes including PLC (phospholipase C), PI3K (phosphatidylinositol 3-kinase) and p38 MAPK (mitogen-activated protein kinase). Analysis of the phosphorylated status of major components in MAPK signalling pathways indicated that thapsigargin and DTT (dithiothreitol) but not tunicamycin could trigger the PERK-dependent activation of JNK (c-Jun N-terminal kinase) and p38 MAPK. However, activation of JNK and p38 MAPK by non-ER stress stimuli including UV irradiation, anisomycin, and TNF-α (tumour necrosis factor-α) was found to be independent of PERK. PERK plays a particularly important role in mediating the global cellular response to ER stress that is elicited by the depletion of calcium from the ER. We suggest that this specificity of PERK function in the UPR is an extension of the normal physiological function of PERK to act as a calcium sensor in the ER.
The ER (endoplasmic reticulum) depends upon a specialized environment of high Ca2+ concentration to fold, modify and assemble newly synthesized proteins destined for secretion or insertion into one of several cellular membranes. Perturbation of the ER environment results in the inability of the ER to properly fold proteins and stimulates the UPR (unfolded protein response). Various pathophysiological conditions and pharmacological reagents that alter the Ca2+ levels, oxidizing environment, or glycosylation machinery in the ER can trigger the UPR [1–3]. Reagents that perturb Ca2+ homoeostasis in the ER are particularly potent elicitors of the UPR. The dominant folding protein, BiP/GRP78 (binding protein/glucose-regulated 78 kDa protein) plays an important role in the storage of ER Ca2+ with a stoichiometry of 1–2 moles of Ca2+/mole of BiP/GRP78 . BiP/GRP78 also acts as the principal regulator of three ER-resident transmembrane proteins, PERK (double-stranded RNA-activated protein kinase-like ER kinase) IRE-1 (inositol requiring-1) and ATF6 (activating transcription factor 6) which are primary sensors and transducers of the UPR [5–9]. A decrease in ER Ca2+ results in disassociation of BiP/GRP78 from PERK and IRE1 resulting in their activation [10,11], whereas, dissociation of BiP/GRP78 from ATF6 leads to its translocation to the Golgi, a prerequisite for its activation [12,13]. The ER is the major storage organelle in the cell for Ca2+ and mobilization of ER Ca2+ has a potent impact on a variety of calcium-sensitive signal transduction pathways in the cytoplasm and nucleus [14–16]. Thus major perturbation of ER calcium homoeostasis can have manifold effects upon stress-related signalling pathways in addition to eliciting the UPR.
In mammalian cells, the UPR is characterized by transcriptional activation of chaperone and protein-folding genes, regulation of growth arrest and apoptosis genes, and repression of global protein synthesis . Among the four ER-resident transmembrane proteins, IRE1α, IRE1β, and ATF6 play a dominant role in mediating transcriptional regulation, whereas PERK is purported to be responsible for repressing global protein synthesis via phosphorylation of the α subunit of eIF2α (eukaryotic initiation factor 2α) [18,19]. Phosphorylation of eIF2α, on the other hand, can also indirectly control gene transcription by positively regulating the translation of transcription factors as has been shown for mammalian ATF4 and yeast GCN (general control of non-derepressible) 4 [20,21]. The translational derepression of ATF4 upon increased phosphorylation of eIF2α by PERK in turn leads to the up-regulation of CHOP/GADD153 (CCAAT/enhancer-binding protein-homologous protein/growth arrest and DNA damage-inducible protein)  and ATF3  expression. Thus by activating ER-resident protein kinases, cells respond to ER stress via intracellular signalling pathways to regulate gene expression. However, the relevant signalling pathways remain to be elucidated.
Signal transduction from the ER to the cell nucleus could be mediated by similar signal cascades associated with plasma membrane-initiated cell signalling. ER stress, for example, is coupled to the activation of stress-activated protein kinases [23,24]. These kinases, including JNK (c-Jun N-terminal kinase) and p38 MAPK (mitogen-activated protein kinase) are known to be activated through a cascade of kinase activities preferentially triggered by physical stresses and inflammatory cytokines, which distinguish them from the ERK (extracellular signal-regulated kinase) pathway [25,26]. JNK and p38 MAPK are activated by a diverse array of ER stress-inducing agents such as thapsigargin, tunicamycin, and DTT (dithiothreitol), which cause depletion of ER Ca2+, inhibition of N-linked glycosylation of proteins, or impairment of disulphide bond formation, respectively [23,27–30]. In addition, several IE (immediate-early) genes, including c-myc and egr-1, have been shown to be transcriptionally activated as a consequence of ER-stress [31–35]. In as much as the ER-resident transmembrane proteins play a major role in ER adaptation, their activities may also be required for the broader cellular adaptations to ER stress mediated by activation of JNK and p38 MAPK pathways and the IE genes which all impact cell proliferation and viability.
Recently we and others discovered that PERK, an ER transmembrane kinase, is required for the normal growth and viability of the endocrine and exocrine pancreas [36,37] leading us to propose that PERK may regulate signal transduction pathways underlying growth control. To investigate this question further we have established MEFs (mouse embryonic fibroblasts) from Perk−/− mice  and have identified genes that exhibit an aberrant response to ER stress. We show in the present study that the normal response of c-myc and egr-1 expression to the loss of Ca2+ homoeostasis in the ER requires PERK.
Moreover, PERK is required for the activation of JNK and p38 MAPK induced by the loss of ER Ca2+ homoeostasis, but PERK is not strictly required for the activation of JNK and p38 MAPK when ER stress is elicited by others reagents that induce the UPR. Thus PERK plays a particularly important role in mediating the global cellular response to ER stress that is elicited by the depletion of Ca2+ from the ER. We suggest that this specificity of PERK function in the UPR is an extension of the normal physiological function of PERK to act as a Ca2+ sensor in the ER.
MEFs were isolated from Perk+/+ and Perk−/− embryos at days 14–15 of gestation . The cells were cultured in Dulbecco's modified Eagle's medium (Invitrogen) containing 10% (v/v) FBS (foetal bovine serum) at 37 °C in a humidified 5% CO2 atmosphere.
cDNA microarray analysis
The total RNA used for microarray analysis was isolated using Atlas Pure Total RNA Labelling System (Clontech Laboratories, Inc.) according to the manufacturer's instructions. The isolated RNA was further treated with DNase I (Ambion, Inc.) for DNA decontamination. The Perk+/+ and Perk−/− MEFs at passage 2 were grown to 90% confluence and treated with or without 1 μM thapsigargin for 1 h before the RNA was isolated. Total RNAs for standard RT-PCR (reverse transcription-PCR) analysis were prepared by TRI reagent (Sigma).
The Clontech Atlas Mouse Nylon Arrays 1.2 and 1.2 II containing 2350 genes in total were employed in this study. The 32P-labelled probe was prepared from 5–8 μg of total RNA and hybridization was performed according to Clontech protocols. The hybridized array was exposed to a Kodak phosphor-image screen for 2–4 days and scanned and visualized using a Storm 860 PhosphorImager (Molecular Dynamics) at 100 μm resolution. The intensity of hybridization signals was quantified using Clontech AtlasImage 2.01 software. The fold change of each gene (cDNA) between two samples, for example mutant versus wild-type or treated versus untreated, was calculated after normalization to the sum of the signal values over background of all genes on the array. Two separate sets of treatment and RNA isolation were used for the assay. In total, four compared values of an individual gene were obtained by cross-comparison between these two sets of samples. The differentially expressed genes in the MEFs were selected only when they consistently showed at least a 2 fold change from multiple calculations.
The mRNA levels of genes of interest were analysed by one-step RT-PCR using Promega Access RT-PCR System. In a standard 20 μl reaction volume, 0.1 μg of total RNA was added for the analysis and 3–5 μl of final reaction mixture was electrophoresed on a 1.5% agarose gel. The DNA fragment was visualized by ethidium bromide staining. The primer sequences (Clontech 7832-1) were designed by Clontech to be used for the RT-PCR analysis of genes on the Atlas Array. The aliquots were removed from each reaction after 13–40 PCR cycles (when visible products first appeared and before they reached a saturation plateau). The cDNA intensities were quantified by ImageQuant 5.1 software (Molecular Dynamics) and compared between samples. The expression of egr-1 and c-myc was also evaluated by real-time quantitative PCR (TaqMan, Applied Biosystems). The total RNA template for reverse transcription was purified from treated cells using the RNeasy mini kit (Qiagen). cDNA was reverse-transcribed using Moloney-murine leukaemia virus polymerase and random hexamer priming (both from Promega). Real-time quantitative PCR reactions contained 0.5–1.0 μg of input cDNA and 2 μM of the appropriate primer pairs. SYBR Green detection (Eurogentec) was used to quantify the amplification of egr-1 and c-myc. The amplification of tubulin in each cDNA sample was used to normalize the data for the test genes.
Kinase inhibitors analysis
The majority of kinase inhibitors were obtained from Calbiochem. MEFs were pretreated with inhibitors for 30 min before thapsigargin was added to the culture. The concentration used for various kinase inhibitors was: PD98059 (ERK kinase inhibitor), 20 μM; U73122 [PLC (phospholipase C) inhibitors], 10 μM; KT5720 (protein kinase A inhibitor), 500 nM; LY294002 [PI3K (phosphatidylinositol 3-kinase) inhibitor], 20 μM; SB203580 (p38 MAPK inhibitor), 20 μM. The less active PLC inhibitor U73343 was obtained from Sigma and was also used at 20 μM.
Stress treatment and immunoblot analysis
MEFs were grown on 35 mm plates to 90% confluence and switched to fresh media containing 0.5% FBS for 18–24 h. Cells were then treated with a variety of stress-inducing agents, including thapsigargin (1 μM), DTT (10 mM), tunicamycin (10 μg/ml), UV light (200 J/m2), anisomycin (10 μg/ml), and TNF-α (tumour necrosis factor-α) (10 ng/ml). At the indicated time after treatment, cells were washed with PBS and lysed by the addition of 80 μl of cold RIPA buffer [1% NP 40 (Nonidet P40), 0.5% sodium deoxycholate, 0.1% SDS, 1×PBS, pH 8.0] containing 1× protease and phosphatase inhibitor cocktails (Sigma). Cells were scraped from the plate and incubated on ice for 10 min. Total cell lysates were obtained after centrifugation at 14000 g for 10 min at 4 °C and the protein concentration was determined by Bio-Rad protein assay reagent. Proteins (15–20 μg) were separated by SDS 4–15% gradient PAGE and transferred to PVDF membrane. Immunoblotting was performed as previously described . Primary antibodies used are mouse anti-eIF2α (kindly provided by Scot Kimball, Pennsylvania State University, PA, U.S.A.), 1:2000; rabbit anti-phospho-eIF2α (Biosource International), 1:1000; rabbit anti-JNK, 1:1000; rabbit anti-(p38 MAPK), 1:1000; rabbit anti-(phospho-p38 MAPK), 1:1000; rabbit anti-ERK, 1:1000; rabbit anti-phospho-ERK, 1:1000 (all from Cell Signaling Technology); mouse anti-phospho-JNK, 1:200 (sc-6254, Santa Cruz Biotechnology). Secondary antibodies used are horseradish peroxidase conjugated anti-mouse or anti-rabbit antibodies (1:5000, Jackson Immuno Research). The immuno-reactive signal was detected by ECL® Plus reagents (Amersham–Pharmacia Biotech) and visualized by a Storm 860 PhosphorImager. Quantification of the signals was conducted using ImageQuant 5.1 software.
PERK is required for the thapsigargin-induced expression of multiple IE genes
To determine the global regulatory effects of PERK on gene expression, we have employed cDNA microarray analysis to examine differences in the expression of 2350 mouse mRNAs in Perk−/− versus Perk+/+ MEFs. We identified 31 genes with a difference of greater than 2 fold expression between Perk+/+ and Perk−/− MEFs under normal growth conditions, and 27 genes were differentially expressed upon treatment with thapsigargin to elicit the UPR. Several mitogenesis-related IE genes, including those encoding c-myc, c-jun, egr-1, fra-1, TIS21 (NGF inducible protein) and tristetraproline, displayed PERK-dependent expression in MEFs upon thapsigargin treatment (Table 1). Except for c-jun whose expression was down-regulated in Perk−/− MEFs, the expression of all other IE genes was differentially up-regulated in Perk+/+ MEFs. RT-PCR analysis of the mRNA levels of selected IE genes such as egr-1 and c-myc confirmed that their expression was induced by thapsigargin in a PERK-dependent manner (Figure 1A). The expression of CHOP, previously shown to be positively regulated by PERK in response to thapsigargin [20,39], was significantly reduced in Perk−/− MEFs by both microarray and RT-PCR analysis (results not shown).
|GenBank® accession number||Gene/protein name||Ratio MEF+/+/MEF−/−|
|GenBank® accession number||Gene/protein name||Ratio MEF+/+/MEF−/−|
Differential expression of c-myc and egr-1 in Perk mutant MEFs treated with thapsigargin
Induction of IE gene expression is usually transient and thus disruption of PERK function may only change their temporal expression rather than ablating their expression. To address this issue, we analysed the mRNA levels of c-myc and egr-1 at various time points after thapsigargin treatment. Standard RT-PCR analysis revealed that whereas c-myc mRNA levels in Perk+/+ MEFs continued to rise during 2 h of thapsigargin treatment, their levels in Perk−/− MEFs showed no increase but declined after 1 h of treatment (Figure 1B). Thus PERK is required for thapsigargin-induced increase of c-myc expression. During the first 30 min after the addition of thapsigargin, egr-1 expression increased in both Perk+/+ and Perk−/− cells. In contrast, wild-type MEF egr-1 mRNA levels continued to increase from 30 min to 2 h, though egr-1 mRNA levels in Perk−/− MEFs dramatically declined (Figure 1B). Therefore PERK is required to sustain the induction and/or stability of the egr-1 mRNA.
We also examined the response of the IE genes to CPA (cyclopiazonic acid), another SERCA (sarcoplasmic/ER Ca2+-ATPase) inhibitor, as compared with the response to thapsigargin using a more sensitive real-time PCR method for quantifying the RNA levels. In the presence of 10 μM CPA, c-myc mRNA levels increase at least 2 fold within 1 h and decline slightly, at 2 h in Perk+/+ cells while levels in Perk−/− remain for the most part, unchanged over the course of the treatment (Figure 1C). We also observe an approximately 2 fold PERK-dependent increase in egr-1 mRNA levels in response to CPA at 1 h; however, there is a significant reduction after 2 h of treatment. It has been shown that both CPA and thapsigargin are specific inhibitors for the SERCA pump in the ER membrane and initiate the release of calcium from the same intracellular stores. However, they do differ in their affinity for SERCA and elicit unique patterns of intracellular calcium transients, which could result in differences in the kinetics of responses to them [40,41]. The temporal differences we observe in IE mRNA up-regulation in response to CPA as compared with thapsigargin may reflect these subtle functional differences; however, overall it is clear that both SERCA inhibitors will up-regulate IE gene expression at the mRNA level in a PERK-dependent manner.
Various ER-stress stimuli result in differential PERK-dependent expression of c-myc and egr-1
c-myc can also be induced upon tunicamycin treatment in MEFs , suggesting that c-myc induction may be a common event under ER stress. To determine if the induction of c-myc is PERK-dependent in cells stressed with other reagents that elicit the UPR, we analysed the mRNA level of c-myc in wild-type and Perk−/− MEFs treated with tunicamycin and DTT. We observed that c-myc mRNA levels were increased up to 2 fold 1 h after tunicamycin treatment in both Perk+/+ and Perk−/− MEFs (Figure 2A), indicating that PERK is dispensable for the induction of c-myc expression by tunicamycin. The c-myc mRNA level continued to increase beyond the first hour of tunicamycin treatment in wildtype cells, whereas the Perk−/− cells exhibited a reduction (Figure 2A). In cells treated with DTT, the level of c-myc mRNA was not increased until 2 h after treatment in wild-type cells (Figure 2B). By contrast, no significant change in c-myc mRNA levels was observed in Perk−/− MEFs upon DTT treatment (Figure 2B), indicating that the induction of c-myc expression by DTT is PERK-dependent. Together, these results indicate that the PERK-dependent induction of c-myc expression is specific to certain but not all ER stress stimuli.
In parallel, we also analysed the expression of egr-1 under the ER stresses elicited by tunicamycin and DTT. Tunicamycin could induce similar transient elevation of mRNAs in both Perk+/+ and Perk−/− MEFs (Figure 2A), indicating that PERK is not necessary for the induction of egr-1 expression by tunicamycin. Unlike tunicamycin, DTT did not cause a significant change in egr-1 mRNA levels in Perk+/+ MEFs (Figure 2B). In Perk−/− MEFs, however, egr-1 mRNA levels were greatly decreased at 1 h after DTT treatment (Figure 2B). These results indicate that although PERK is not essential for ER stress-induced expression of egr-1, it is required to maintain the steady state mRNA levels of egr-1.
The expression of c-myc and egr-1 induced by tunicamycin and DTT
Thapsigargin-induced c-myc expression is mediated through multiple signal transduction pathways
The induction of IE genes is correlated with a variety of signal transduction pathways . To identify the specific signal transduction pathways underlying the PERK-dependent activation of c-myc by thapsigargin in MEFs, we employed an array of specific enzyme inhibitors to block major signalling pathways. We found that thapsigargin-induced expression of c-myc was significantly repressed by PLC inhibitor U73122, and partially blocked by the PI3K inhibitor LY294002 and the p38 MAPK inhibitor SB203580 (Figure 3). By contrast, the ERK kinase inhibitor PD98059 and the protein kinase A inhibitor KT5720 were unable to abrogate the thapsigargin-induced increase of c-myc expression. U73343, which is often used as a negative control for the PLC inhibitor U73122 had no effect on thapsigargin-induced expression of c-myc (results not shown). These results indicate that signal transduction pathways associated with PLC, PI3K and p38 MAPK activities are necessary for the thapsigargin-induced expression of c-myc in MEFs.
Effects of various enzyme inhibitors on the expression of c-myc induced by thapsigargin
PERK is required for thapsigargin-induced activation of JNK and p38 MAPK but not ERK
ER stress induces the activation of p38 MAPK [24,28–30] and our finding that an inhibitor of p38 MAPK markedly diminished the induction of c-myc by thapsigargin, prompted us to investigate whether activation of p38 MAPK is PERK-dependent. The level of phosphorylated p38 MAPK was increased 2–3 fold in Perk+/+ but not in Perk−/− MEFs after thapsigargin treatment (Figure 4), indicating that activation of the p38 MAPK signal pathway by thapsigargin is PERK-dependent.
Effects of thapsigargin on the level of phosphorylated p38 MAPK, JNK and ERK
In view of the cross-talk between different MAPK signalling pathways, we examined activation of ERK and JNK upon thapsigargin treatment. We did not observe an induction of ERK phosphorylation upon thapsigargin treatment in either Perk+/+ or Perk−/− MEFs. In contrast, phosphorylation of JNK was induced 2 fold in normal cells but not in Perk−/− MEF cells (Figure 4).
PERK mediates the activation of JNK and p38 MAPK through distinct signalling pathways from those used by other non ER-stress stimuli
JNK and p38 MAPK are activated by a number of cellular stresses in addition to those that perturb the function of the ER. To determine whether the activation of JNK and p38 MAPK by other cellular stresses also requires PERK function, we treated Perk+/+ and Perk−/− MEFs with UV light, anisomycin and TNF-α and measured the levels of phosphorylated JNK and p38 MAPK. We found that the activation of JNK and p38 MAPK by UV, anisomycin and TNF-α was very similar between Perk+/+ and Perk−/− MEFs (Figure 5). All three of these stresses led to a predicted strong activation of both p38 and JNK. These results indicate that the components involved in JNK and p38 MAPK activation are not generally defective in Perk−/− MEFs, but rather PERK is required for their activation in the context of specific ER stresses. Phosphorylation of eIF2α was induced in cells treated with UV but not in cells treated with anisomycin or TNF-α (Figure 5A). The GCN2 eIF2α kinase has been shown to be responsible for phosphorylation of eIF2α upon UV induction [43,44].
Effects of UV light, anisomycin and TNF-α on the phosphorylation of p38 MAPK and JNK in MEFs
Only specific ER stresses can trigger PERK-dependent activation of JNK and p38 MAPK
In addition to thapsigargin, several other ER stress-inducing agents such as DTT and tunicamycin can also activate JNK [23,27]. To determine whether the PERK-dependent activation of stress-activated MAPKs is a universal consequence of ER stresses, we measured the levels of phosphorylated JNK and p38 MAPK in both Perk+/+ and Perk−/− MEFs treated with DTT or tunicamycin. DTT treatment significantly stimulated a 2 fold PERK-dependent phosphorylation of eIF2α within the first 30 min (Figure 6A); however, only modest stimulation of JNK and p38 MAPK phosphorylation was observed in the Perk+/+ MEFs (1.4 fold for JNK and 1.6 fold for p38 MAPK). The relatively low level of JNK and p38 MAPK phosphorylation induced by DTT in wild-type cells was not observed until after 2 h despite a large induction in eIF2α phosphorylation after only 30 min. No induction of JNK and p38 MAPK phosphorylation was observed in Perk−/− MEFs (Figure 6A). Compared with other pharmacological reagents that can induce the UPR, tunicamycin triggered a relatively low level of eIF2α phosphorylation in Perk+/+ MEFs (Figure 6B). Moreover, tunicamycin induced a modest level of JNK and p38 MAPK phosphorylation in both Perk+/+ and Perk−/− cells.
Effects of DTT and tunicamycin on the phosphorylation of p38 MAPK and JNK
In summary, among the three reagents that induce the UPR, only disruption of ER calcium homoeostasis by treatment of thapsigargin led to a strong induction of JNK and p38 MAPK phosphorylation that was PERK-dependent and highly correlated with eIF2α phosphorylation. By contrast, DTT elicited a high level of PERK-dependent eIF2α phosphorylation but this was associated with a delayed and weak induction of JNK and p38 MAPK phosphorylation, whereas tunicamycin stimulated a higher level of JNK and p38 MAPK phosphorylation but was independent of PERK and was not correlated with eIF2α phosphorylation.
The UPR stimulates the expression of ER resident chaperones to adapt to increased demands for protein folding in the ER. Because of the central role the ER plays in the function and viability of the cell, ER perturbations also impact upon a diverse array of other cellular functions. Underlying the general cellular response to ER stress are the stress-activated kinase pathways and the induction of IE genes, which can also be activated by non-ER mediated stresses. We found that the ER-resident transmembrane kinase PERK is required for either the induction or maintenance of the induced state of c-myc and egr-1 gene expression under specific ER stresses. Of the ER stresses tested, each differentially impacts the expression of c-myc and egr-1. When ER calcium homoeostasis is disrupted via inhibition of the ER SERCA calcium pump by thapsigargin, c-myc is dramatically induced only in Perk+/+ MEFs. Thapsigargin treatment has an early stimulatory effect on egr-1 expression in both wild-type and Perk−/− mutant MEFs. At later time points, however, egr-1 mRNA levels continue to increase in wild-type cells but dramatically decline in Perk−/− cells resulting in a ten fold difference at 2 h. Similar complexity in response to two other UPR inducing reagents, DTT and tunicamycin was observed. PERK is required for the activation of c-myc upon DTT treatment but not upon tunicamycin treatment. However, after two h of tunicamycin treatment under our experimental condition, the expression of c-myc declined substantially in Perk−/− MEFs but continued to increase in Perk+/+ MEFs. A similar phenomenon was observed for the expression of egr-1 in response to thapsigargin and DTT treatment; egr-1 expression radically declined in Perk−/− MEFs in contrast with wild-type cells in which egr-1 continued to increase or remain at high levels. Thus for most ER stresses, PERK is required either for the initial activation of IE gene expression or for maintaining the activated state.
The IE genes such as egr-1 and c-myc play crucial roles in regulating both cell proliferation and apoptosis [45–47]. CHOP/GADD153, a mediator of apoptosis, is also induced as part of the UPR and its transcription is in part regulated by PERK via translational control of ATF-4, a CHOP transcriptional activator . PERK is required for the developmental regulation of IGF-1  and ER-stress induced translocation of the nuclear factor-κB transcription factor to the nucleus . Thus PERK regulates an array of factors that control growth and adaptation to stress. PERK activity is also modulated under non-stressed conditions, as a function of stimulus-coupled secretion that elicits calcium mobilization from the ER (B. C. McGrath and D. R. Cavener, unpublished work). Under normal physiological conditions that stimulate secretion and ER calcium mobilization, PERK may regulate factors which in turn modulate secretory capacity in concert with changes in developmental and physiological demand for secretory proteins. This function may be particularly important in the major secretory cells of the pancreas and skeletal system which exhibit gross dysfunctions in mice and humans deficient in PERK [37,48,49].
A plethora of extracellular and intracellular stimuli can activate the IE genes suggesting that PERK may be required for only specific pathways which mediate responses related to ER stress. We found that inhibition of PLC significantly represses c-myc induction by thapsigargin. PLC catalyses the cleavage of phosphatidylinositol 4,5-bisphosphate to generate the messengers DAG (diacylglycerol) and IP3 (inositol 1,4,5-trisphosphate). IP3, in turn, is required to activate the ER IP3 receptor which releases Ca2+ from the ER. We suggest that the induction of c-myc expression by thapsigargin requires the concerted action of inhibiting the SERCA, which normally pumps cytoplasmic Ca2+ into the ER, and activation of the IP3 receptor, which releases Ca2+ from the ER. Alternatively, c-myc induction may instead depend upon any one of several signalling and regulatory functions impacted by the generation of IP3 and DAG messengers.
In addition, we showed that thapsigargin-induced PERK-dependent expression of c-myc can be partially repressed by inhibitors of PI3K and p38 MAPK (Figure 3). PERK may be required to transduce signalling initiated by ER stress to multiple regulatory pathways which co-operate to regulate c-myc expression. Alternatively, PLC, PI3K and p38 MAPK may be functionally linked together in controlling c-myc expression. In support of the latter hypothesis, PLC-β has been shown to interact with both p38 MAPK and its upstream kinase, MAPK kinase 3 . In addition, phosphorylation of p38 MAPK lies downstream of PI3K activation , suggesting a functional relationship exists between these proteins.
The stress-activated protein kinases JNK and p38 MAPK were previously shown to be induced by ER stress [23,24] and given our finding that c-myc induction by thapsigargin is partially dependent upon p38 MAPK activity, we investigated the potential role of PERK in regulating p38 MAPK and JNK. We found that PERK is indeed required for the activation of JNK and p38 MAPK upon treatment with thapsigargin, an inhibitor of the SERCA calcium pump. However, the activation of JNK and p38 MAPK by other ER stressors is independent or only partially dependent upon PERK. Moreover, the activation of JNK and p38 MAPK by extracellular and cytoplasmic stressors is independent of PERK. Thus PERK appears to play a restrictive role in regulating the activation of stress-activated kinases, JNK and p38 MAPK, upon disruption of ER Ca2+ homoeostasis. For other ER stress-inducing reagents (e.g. DTT and tunicamycin) the activation of PERK and phosphorylation of eIF2α is not required for the activation of JNK and p38 MAPK, with the possible exception of DTT which was shown to induce a modest and delayed phosphorylation of JNK and p38 MAPK only in Perk+/+ cells. Thus in general, activation of PERK and phosphorylation of eIF2α is insufficient to activate JNK and p38 MAPK. We speculate that the release of ER calcium stores upon thapsigargin treatment stimulates other calcium sensitive pathways, which act in concert with PERK to activate p38 and JNK. Other ER stressors, which directly disrupt protein folding or abrogate protein glycosylation, may lead to the activation of the stress-activated protein kinases by different pathways than those involved in perturbation of Ca2+ homoeostasis.
The JNK and p38 MAPK pathways are particularly sensitive to disruption of intracellular Ca2+ signalling [51–56], and we propose that PERK plays an essential role to integrate ER-stress response and calcium-sensitive signalling pathways. In addition to its primary role in protein folding and quality control, the ER is also the major storage organelle for Ca2+. Mobilization of ER Ca2+ plays a fundamental role in regulating Ca2+ signalling cascades and exocytotic release of secretory granules . Moreover, disruption of intracellular Ca2+ homoeostasis or perturbation of Ca2+ signalling underlies a vast array of human diseases [14,15,57–59]. PERK may serve as an important cellular signalling mediator during the fluctuation of ER lumenal calcium, which integrates the lumenal function of protein folding and assembly with the dynamic physiological regulation of intracellular calcium.
It is shown that four ER transmembrane proteins – PERK, IRE1α, IRE1β and ATF6 – act in concert to mediate the UPR. The ER resident BiP/GRP78 negatively regulates each of these four proteins and consequently dissociation of BiP/GRP78 activates their function [10–13], suggesting that these four regulatory proteins could be activated simultaneously to activate the cascade of events comprising the UPR. One or more of these ER transmembrane proteins may functionally interact to mediate specific aspects of the UPR. For example, IRE1 has been shown to be required for the activation of JNK in MEFs upon treatment with thapsigargin and tunicamycin . We show in the present study that thapsigargin, but not tunicamycin-induced activation of JNK is dependent on a functional PERK. Together, these findings suggest that PERK and IRE1 are both essential for the JNK activation upon disruption of ER Ca2+ homoeostasis.
In conclusion, the PERK eIF2α kinase is required to mediate the cellular responses to the disruption of ER Ca2+ homoeostasis, and to a more limited extent, other ER stresses. We propose that these stress-related functions of PERK are an extension of the major role of PERK in regulating normal physiological processes mediated by Ca2+ mobilization and signalling. The direct downstream signalling components of PERK that mediate its ER-stress related functions remain to be discovered.
This study was supported by grant R01DK062049 (to D.R.C.) from the National Institutes of Health.
activating transcription factor 4
binding protein/glucose-regulated 78 kDa protein
CCAAT/enhancer-binding protein-homologous protein
early growth response factor-1
eukaryotic initiation factor-2α
extracellular signal-regulated kinase
foetal bovine serum
growth arrest and DNA damage-inducible protein
general control of non-derepressible
c-Jun N-terminal kinase
mitogen-activated protein kinase
mouse embryonic fibroblast
pancreatic eIF-2α kinase
double-stranded RNA-activated protein kinase-like ER kinase
unfolded protein response