Register for Mechanobiology 2016!

Endoplasmic reticulum stress impairs IL-4/IL-13 signaling through C/EBPβ-mediated transcriptional suppression
Angela M. Arensdorf, D. Thomas Rutkowski


Activation of the unfolded protein response (UPR) by endoplasmic reticulum (ER) stress culminates in extensive gene regulation, with transcriptional upregulation of genes that improve the protein folding capacity of the organelle. However, a substantial number of genes are downregulated by ER stress, and the mechanisms that lead to this downregulation and its consequences on cellular function are poorly understood. We found that ER stress led to coordinated transcriptional suppression of diverse cellular processes, including those involved in cytokine signaling. Using expression of the IL-4/IL-13 receptor subunit Il4ra as a sentinel, we sought to understand the mechanism behind this suppression and its impact on inflammatory signaling. We found that reinitiation of global protein synthesis by GADD34-mediated dephosphorylation of eIF2α resulted in preferential expression of the inhibitory LIP isoform of the transcription factor C/EBPβ. This regulation was in turn required for the suppression of Il4ra and related inflammatory genes. Suppression of Il4ra was lost in Cebpb−/− cells but could be induced by LIP overexpression. As a consequence of Il4ra suppression, ER stress impaired IL-4/IL-13 signaling. Strikingly, Cebpb−/− cells lacking Il4ra downregulation were protected from this signaling impairment. This work identifies a novel role for C/EBPβ in regulating transcriptional suppression and inflammatory signaling during ER stress.


ER stress results from an accumulation of unfolded proteins in the organelle. It is induced by numerous physiological and pathophysiological stimuli including viral infections, calcium or glucose depletion, oncogenesis, metabolic flux and differentiation of certain cell types (Rutkowski and Hegde, 2010). The UPR attempts to restore ER homeostasis and improve ER functionality through transcriptional and post-transcriptional mechanisms (Ron and Walter, 2007).

Three ER-resident transmembrane receptors initiate UPR signaling: pancreatic ER eIF2α kinase (PERK), inositol requiring enzyme 1α (IRE1α) and activating transcription factor 6α (ATF6α). The pathways initiated by each of these culminate in production of transcriptional activators. Phosphorylation of PERK leads to phosphorylation of eIF2α and subsequent translation of the transcription factor ATF4; active IRE1α splices Xbp1 mRNA to allow production of the XBP1 transcription factor; and ATF6α, itself a transcription factor, is transported to the Golgi apparatus where it is liberated by regulated intramembrane proteolysis (Ron and Walter, 2007). These canonical UPR-regulated transcription factors activate transcription of genes involved in ER protein folding, processing and degradation (Rutkowski and Kaufman, 2007). Yet UPR activation also leads to substantial gene suppression, and the pathways involved in such suppression and the consequences thereof are largely unknown.

The mechanisms of UPR-mediated gene suppression are likely not attributable to direct regulation by ATF4, ATF6 or XBP1, as these are transcriptional activators. Rather, the best characterized mechanism for suppression is regulated IRE1-dependent decay (RIDD), which occurs when IRE1α directly cleaves ER-associated mRNAs, i.e. those that encode ER targeting sequence-bearing proteins (Hollien et al., 2009; Hollien and Weissman, 2006). However, only a subset of mRNAs suppressed during ER stress encode such targeting sequences; therefore other mechanisms must contribute as well. The UPR-regulated transcription factor C/EBP-homologous protein (CHOP) can act as a transcriptional repressor by sequestering other C/EBP family members (Ron and Habener, 1992). A number of metabolic genes are suppressed during ER stress; at least some of these genes require CHOP for their suppression (Chikka et al., 2013; Rutkowski et al., 2008), and CHOP blunts the upregulation of others (Lopez et al., 2007; Su and Kilberg, 2008). However, CHOP can also act as a transcriptional activator (Li et al., 2009; Marciniak et al., 2004; Ohoka et al., 2005; Ubeda et al., 1996); therefore it is not clear to what extent CHOP truly contributes to extensive gene suppression. Transcription can also be suppressed by the titration of coactivators such as CRTC2 and PGC1α by UPR-regulated transcription factors away from their normal interacting partners (Wang et al., 2009; Wu et al., 2011), though again, these mechanisms have only been definitively linked to a small number of genes. Thus, the vast majority of suppressed genes have not been tied to a specific mechanism.

Regulation of lipid metabolism upon ER stress provides stark evidence that coordinated gene downregulation has physiological consequences. UPR activation in the liver leads to transcriptional suppression of fatty acid catabolism and lipoprotein biogenesis, causing pronounced steatosis particularly in animals with a pervasive ER stress burden (Rutkowski et al., 2008). It stands to reason that, just as ER protein folding and processing are enhanced by the collective action of many genes upregulated by the UPR, so also gene suppression is aimed at the regulation of discrete cellular processes.

Our goal was to identify physiological processes on which UPR-mediated gene suppression impinges, and begin to uncover the mechanisms behind such regulation. Here we show that IL-4/IL-13 signaling is suppressed during ER stress by eIF2α-dependent regulation of the transcription factor C/EBPβ.


ER stress leads to the suppression of genes involved in inflammatory signaling cascades

The presence of gene suppression during ER stress has been well documented in a number of published microarray studies (Hollien et al., 2009; Marciniak et al., 2004; Rutkowski et al., 2008; Wu et al., 2007); however, the mechanisms involved and the cellular processes affected remain largely unexplored. Using previously published data sets and pathway enrichment analysis, we sought to identify physiological processes that were targeted by UPR-mediated gene suppression.

Gene expression changes induced by the UPR differ depending on the cell type, duration and severity of the stress; therefore to identify suppression events of potential physiological relevance, we focused on the genes regulated and the processes enriched under different ER stress conditions.

The effect of strong, acute stress was tested using wild-type mouse embryonic fibroblasts (MEFs) treated with a high dose of the N-linked glycosylation inhibitor tunicamycin (TM) for 8 hours (Marciniak et al., 2004). The scatter plot of these data highlights that ∼50 percent of the TM-regulated genes were suppressed (Fig. 1A). Among the upregulated genes were well-known UPR targets such as Bip, Chop and Edem, while downregulated genes included known RIDD targets such as Bloc1s1, Pdgfrb and Mrc2 (Fig. 1A). Unsurprisingly, pathway analysis of upregulated genes revealed an enrichment of genes involved in protein synthesis, folding and processing (Fig. 1B). In contrast, a diverse group of signaling pathways was enriched among the downregulated genes, with little direct connection to ER protein folding or processing (Fig. 1B).

Fig. 1.

Microarray analysis reveals suppression of inflammatory signaling genes during ER stress. (A) Scatter plot of genes differentially regulated (fold change >1.5, P<0.05) in wild-type MEFs treated with vehicle or 2 µg/ml TM for 8 hours. Each square represents a probe set taken from a previously published microarray (Marciniak et al., 2004). Genes of interest are indicated. (B) The list of upregulated genes (top) and downregulated genes (bottom) from the microarray summarized in A were subjected to GO process enrichment analysis, and the top seven processes that were significantly enriched (P<0.01) after correction for false discovery are shown. (C) Scatter plot generated as in A using wild-type MEFs treated with vehicle or 50 ng/ml TM for 24 hours (Wu et al., 2007). (D) GO process enrichment analysis conducted as in B using microarray data summarized in C. (E) Wild-type MEFs were treated for 24 hours with 100 ng/ml TM or vehicle, and expression of the indicated genes involved in cytokine signaling pathways was determined by qRT-PCR. Error bars here and elsewhere represent means ± s.e.m. from three independently treated samples. All experiments shown were performed 2–5 times. All qRT-PCR data were normalized against two housekeeping genes (Btf3 and Ppia).

The effect of long-term mild ER stress was tested using wild-type MEFs treated with a low dose of TM for 24 hours (Wu et al., 2007). Under these conditions, ∼20% of the TM-regulated genes were suppressed (Fig. 1C). Again, Bip, Chop and Edem were among the upregulated genes, while Mrc2 and Pdgfrb were among the downregulated genes (Bloc1s1 did not reach the cutoff for inclusion; Fig. 1C). Pathway analysis of these gene sets was consistent with Fig. 1B, with the upregulated genes containing pathways involved in protein processing in the ER, and the downregulated genes encompassing various signaling cascades (Fig. 1D). Although the enriched signaling cascades differed between the microarrays, processes involved in cytokine and inflammatory signaling were present in both. This finding was of particular note because of the emerging links between ER stress and inflammatory states such as obesity, atherosclerosis and inflammatory bowel disease (Adolph et al., 2012; Hotamisligil, 2010; Seimon and Tabas, 2009). It led us to further pursue the effects and consequences of ER stress on the expression of genes involved in cytokine signaling.

Quantitative RT-PCR (qRT-PCR) from a new experiment confirmed the ER-stress-dependent downregulation of a number of the genes present in these enriched inflammatory signaling pathways (Fig. 1E). Among these were Il4ra and Il13ra1, encoding components of the IL-4 and IL-13 receptors. IL-4 and IL-13 are proinflammatory cytokines produced by TH2 cells, and the IL4RA subunit is common to both the IL-4 and IL-13 receptor (Wills-Karp and Finkelman, 2008). The functions of IL-4 and IL-13 overlap somewhat and include B cell class switching, alternative activation of macrophages, and induction of fibrosis (Hershey, 2003; Wills-Karp and Finkelman, 2008; Wynn, 2003). Suppression of both mRNAs suggested that IL-4 and IL-13 signaling might be affected by ER stress. Because Il4ra was suppressed to a greater degree than Il13ra1, we chose to follow the mechanism and consequences of its suppression.

ER stress suppresses Il4ra expression in multiple cell types

To determine the extent to which Il4ra suppression represented a canonical feature of UPR activation, we followed its expression in response to different ER stressors, and in different cell types. Wild-type MEFs treated with either TM or the SERCA inhibitor thapsigargin (TG) showed significant and sustained suppression of Il4ra after eight or more hours of ER stress; and unlike the RIDD target Bloc1s1, this suppression showed no signs of attenuation at later time points (Fig. 2A). This suppression was mirrored in IL4RA protein expression (Fig. 2B). Il4ra downregulation was also elicited by overexpression of an ER client protein, myocilin (MYOC), a mutated form of which in particular (Y437H) elicits ER stress (Zode et al., 2011) (Fig. 2C). The effect of ER stress on Il4ra mRNA expression was also not unique to MEFs; the same effect was observed in I.29μ+ immortalized B cells (Fig. 2D), and, albeit modestly, in A549 adenocarcinomic human alveolar basal epithelial cells (Fig. 2E). These data suggested that the suppression of Il4ra is a general feature of UPR activation. We thus set about identifying the mechanism responsible for this suppression.

Fig. 2.

Suppression of Il4ra is a canonical feature of UPR activation. (A) qRT-PCR analysis of Bloc1s1 and Il4ra from wild-type MEFs treated with 500 ng/ml TM or 50 nM TG for the indicated times. (B) IL4RA expression in wild-type MEFs treated with 50 nM TG for 16 hours was detected by immunoblotting. Calnexin (CNX) was used as a loading control. Here and elsewhere, each similarly labeled lane represents an independently treated sample. (C) NIH 3T3 cells were transfected with a vector expressing tdTomato, or constructs expressing either wild-type myocilin or a Y437H mutant along with dsRed. Red fluorescent cells were sorted by FACS, and tdTomato-expressing cells were left untreated or treated with 500 ng/ml TM for 16 hours. Expression of Chop and Il4ra was determined by qRT-PCR. (D) I.29μ+ B cells were treated with 500 ng/ml TM or 50 nM TG for 16 hours, followed by qRT-PCR to quantify expression of the indicated genes. (E) A549 alveolar epithelial cells were treated with 500 ng/ml TM for 16 hours and gene expression was determined by qRT-PCR. Here and elsewhere, *P<0.05.

The UPR regulates Il4ra expression through a RIDD-independent mechanism

IL4RA is a transmembrane protein, making its mRNA a potential target of the RIDD pathway of mRNA degradation. However, in contrast to the behavior of the known RIDD target Bloc1s1, inhibition of transcription by actinomycin D (ActD) was sufficient to attenuate Il4ra expression to an extent comparable to ER stress (Fig. 3A). In addition, while Bloc1s1 suppression during ER stress was lost in Ire1α/− cells, Il4ra suppression was not (Fig. 3B). These data exclude RIDD and other IRE1α-dependent mechanisms as the mode of Il4ra suppression. Similarly, Il4ra was suppressed by ER stress in Chop/− cells, excluding CHOP-dependent processes (Fig. 3C). These data point to a heretofore unknown mechanism for Il4ra downregulation.

Fig. 3.

UPR-mediated suppression of Il4ra does not depend on the RIDD pathway or CHOP. (A) Wild-type MEFs were treated with 50 nM TG or 10 µg/ml actinomycin D (ActD) for 12 hours, and gene expression was determined by qRT-PCR. (B) qRT-PCR analysis of Ire1α+/+ and Ire1α−/− MEFs treated with 500 ng/ml TM for 16 hours. The results from each TM-treated group are expressed relative to the untreated group (UT) of the same genotype. (C) Wild-type or Chop−/− MEFs were treated for 16 hours with 50 nM TG and analyzed by qRT-PCR as in B.

The UPR-mediated transcriptional suppression of Il4ra requires C/EBPβ

If UPR-mediated suppression of Il4ra occurs through a transcriptionally repressive mechanism, we would expect its promoter region to be ER stress responsive. Indeed, a 2.5 kb genomic region encompassing 1785 bp upstream and 715 bp downstream of the Il4ra transcriptional start site (TSS) was sufficient to confer stress-dependent downregulation of a β-galactosidase (lacZ) reporter (Fig. 4A). A comparable degree of downregulation was maintained when nearly the entire region upstream of the TSS was deleted, but not when a 504 bp region downstream, within the first Il4ra intron, was deleted (Fig. 4B,C). These results suggest Il4ra is suppressed transcriptionally.

Fig. 4.

UPR-mediated transcriptional repression of Il4ra requires C/EBPβ. (A) A 2.5 kb genomic region of the Il4ra promoter was sufficient to confer ER-stress-induced transcriptional repression on a β-galactosidase (lacZ) reporter. Forty-eight hours after transfection, cells were treated with 500 ng/ml TM for 16 hours. Data were normalized against co-transfected constitutively expressed luciferase (NT), and β-galactosidase (lacZ) activity is expressed on a log2 scale relative to the untreated (UT) condition. (B) Schematic of lacZ reporters generated using Il4ra genomic regions, labeled in base pairs relative to the TSS. (C) Expression of β-galactosidase (LacZ) from the constructs shown in B was assessed as in A, with the stress-dependent expression given relative to the untreated condition for each construct. (D) Potential C/EBPβ binding sites, identified by PROMO in or near the +211 to +715 bp region of the Il4ra promoter are indicated by arrowheads. (E) Binding of C/EBPβ to +211 to +715 bp region of the Il4ra promoter was determined by chromatin immunoprecipitation using C/EBPβ antiserum or control IgG antibody. Enrichment was quantified by qRT-PCR; the underline in D denotes the PCR amplicon. (F,G) qRT-PCR analysis of the indicated genes in Cebpb+/+ and Cebpb−/− MEFs treated with TG (50 nM) for 16 hours was performed as in Fig. 3.

Bioinformatic sequence analysis of the +211 to +715 region of the Il4ra promoter did not reveal any potential sites for the canonical UPR-regulated transcription factors ATF4, XBP1 and ATF6α. However, there were several potential binding sites for C/EBPβ (Fig. 4D), which is linked to ER stress through eIF2α phosphorylation (Li et al., 2008). Binding of C/EBPβ to this genomic region of Il4ra was confirmed by chromatin immunoprecipitation (Fig. 4E). Stress-dependent suppression of Il4ra was completely lost in Cebpb/− MEFs, confirming the functional relevance of C/EBPβ (Fig. 4F). C/EBPβ also regulated the expression of at least two other inflammatory genes, Pkd1 and Lif, during ER stress, suggesting that the C/EBPβ-dependent mechanism responsible for Il4ra suppression acts more broadly (Fig. 4G). However, this mechanism cannot account for all UPR-mediated gene suppression, as the regulation of two non-inflammatory genes, Crip1 and Acat2 – involved in zinc transport and lipid metabolism, respectively – were not affected by the deletion of C/EBPβ.

eIF2α phosphorylation and dephosphorylation are necessary for C/EBPβ (LIP) expression and Il4ra regulation

Translational regulation of C/EBPβ yields three isoforms: LAP*, LAP and LIP. The LAP isoforms have been characterized as transcriptional activators and the LIP isoform as a repressor (Descombes and Schibler, 1991; Zahnow, 2009). LIP is a truncated form of the C/EBPβ protein, synthesized when the LAP and LAP* start codons are bypassed in favor of the LIP start codon (Calkhoven et al., 2000). While eIF2α phosphorylation has been shown to be essential in regulating the preponderance of these forms (Li et al., 2008), thus linking C/EBPβ to ER stress, a role for C/EBPβ in the transcriptional control of Il4ra has not been described. Thus, we tested whether modulation of eIF2α phosphorylation influenced C/EBPβ translation in the predicted way, and whether this influence corresponded with Il4ra transcriptional regulation.

Consistent with previous work (Li et al., 2008), in wild-type cells, long-term (16 hours) ER stress led to an increase in LIP expression while the expression of LAP remained constant (Fig. 5A). This increase in LIP has been shown to result from enhanced protein synthesis at later stages of the stress response, along with enhanced stability of LIP protein (Li et al., 2008). Genetic ablation of eIF2α phosphorylation prevented this increase in LIP expression (Fig. 5A), and the suppression of Il4ra (Fig. 5B). In contrast, the deletion of Atf6α had no effect on either LIP expression (Fig. 5C) or Il4ra suppression (Fig. 5D). In addition to eliminating an essential role for ATF6α in Il4ra suppression, these data suggest that the level of LIP correlates with the level of eIF2α phosphorylation and suppression of Il4ra, and that this suppression is compromised when eIF2α phosphorylation is ablated.

Fig. 5.

Translational inhibition and resumption regulate C/EBPβ (LIP) and Il4ra expression. (A) Wild-type (S/S) MEFs and MEFs homozygous for an allele encoding nonphorphorylatable eIF2α (A/A) were treated with 50 nM TG for 16 hours, and the LAP/LAP* (these two forms could not be resolved from each other) and LIP forms of C/EBPβ were detected by immunoblotting, along with phosphorylated eIF2α and CNX as a loading control. (B) eIF2α S/S and A/A cells were treated as in A and expression of Bip, Chop and Il4ra was determined by qRT-PCR. (C,D) Wild-type and Atf6α−/− MEFs were treated and analyzed as in A and B. (E,F) Wild-type and Gadd34−/− MEFs were treated and analyzed as in A and B. (G,H) Wild-type MEFs were deprived of methionine for the indicated times, and cells were analyzed by immunoblotting or qRT-PCR as in A and B.

The importance of eIF2α regulation was also seen in Gadd34/− MEFs. As expected, these cells, which lack the regulatory subunit of the PP1 phosphatase that dephosphorylates eIF2α (Novoa et al., 2001), showed dramatically enhanced eIF2α phosphorylation. However, LIP expression fell below detectable levels in these cells, and LAP was diminished as well (Fig. 5E). These findings are consistent with the previously reported short half-lives of LAP and LIP (Li et al., 2008); perpetuation of eIF2α translational inhibition would prevent resumption of their synthesis. Concordantly, Il4ra suppression was completely lost in Gadd34/− cells; indeed, Il4ra was actually upregulated by stress (Fig. 5F). Thus, both eIF2α phosphorylation and subsequent dephosphorylation are necessary for Il4ra suppression.

Though necessary for Il4ra suppression, phosphorylation and dephosphorylation per se were not sufficient. Depriving cells of the essential amino acid methionine, which leads to eIF2α phosphorylation by activation of the GCN2 pathway (Sood et al., 2000), did not lead to a significant increase in LIP expression at any time (Fig. 5G) and did not suppress Il4ra expression (Fig. 5H). Absent methionine, cells would be unable to efficiently resume protein synthesis even after eIF2α dephosphorylation. Taken together, these data point to a translational advantage of LIP that allows its expression to increase upon eIF2α dephosphorylation and resumption of protein synthesis.

In all of these experiments, Il4ra mRNA levels inversely correlated with LIP expression and showed no consistent relationship with LAP expression. We found that overexpression of LIP alone was sufficient to suppress Il4ra expression to an extent similar to control-transfected cells treated with TG (Fig. 6A,B). ER stress further increased LIP expression in LIP-transfected cells, and also resulted in a further diminishment of Il4ra expression (Fig. 6B). The overexpression of another bZIP transcription factor, ATF2, did not suppress Il4ra expression, suggesting that this effect is LIP-dependent and not due to mere transcription factor overexpression (Fig. 6C).

Fig. 6.

Expression of C/EBPβ (LIP) is sufficient to suppress Il4ra expression. (A) Wild-type MEFs were nucleofected with a vector encoding either GFP or the LIP form of C/EBPβ. Immunoblotting confirmed overexpression of LIP. Samples shown were taken from the same gel with uniform image processing. (B) qRT-PCR showing expression of Il4ra in response to treatment of transfected cells with 50 nM TG for 16 hours. (C) qRT-PCR showing expression of Il4ra in wild-type MEFs nucleofected with GFP, ATF2 or LIP expression constructs. N.S., not significant.

Transcriptional repression of Il4ra suppresses IL-4 and IL-13 signaling

Lastly, we wished to test whether transcriptional suppression of Il4ra influenced IL-4/IL-13 signaling during ER stress. Because Il4ra encodes an essential receptor subunit of both the IL-4 and IL-13 receptors, we predicted that the decrease in IL4RA protein expression would impair signaling. Stimulation of the IL-4 or IL-13 receptor leads to JAK1/2-dependent phosphorylation of the transcription factor STAT6 (Rawlings et al., 2004); thus to test this hypothesis, the level of IL-4/IL-13 signaling was quantified through the level of p-STAT6 expression. ER stress decreased IL-4 signaling in I.29μ+ B cells (Fig. 7A) and IL-13 signaling in A549 cells (Fig. 7B), as predicted given their suppression of Il4ra expression (Fig. 2D,E).

Fig. 7.

C/EBPβ is required for suppression of IL-4/IL-13 signaling during ER stress. (A) I.29μ+ immortalized B cells were treated with 500 ng/ml TM for 16 hours, and then 50 ng/ml IL-4 was added for 1 hour. Levels of phosphorylated and total STAT6 were detected by immunoblotting. (B) A549 alveolar epithelial cells were treated with 500 ng/ml TM for 16 hours, followed by 50 ng/ml IL-13 for 1 hour. STAT6 phosphorylation was assessed as in A. (C) Wild-type MEFs were treated with 5 or 50 nM TG for 16 hours, and mRNA expression was assessed by qRT-PCR. (D) Wild-type MEFs were treated with 5 nM TG or vehicle for 16 hours, and lysates were harvested and digested with Endo H or PNGase F as indicated. Digestion reactions were analyzed by immunoblotting. The ER-resident glycoprotein TRAPα served as a positive control for Endo H sensitivity. (E) Wild-type MEFs were treated with 5 or 50 nM TG for 16 hours, and then 50 ng/ml IL-13 was added for 1 hour. STAT6 phosphorylation was assessed by immunoblotting as in A. (F) Quantification of p-STAT6 from the immunoblot shown in E, normalized to total STAT6 level. (G) Wild-type and Cebpb−/− MEFs were treated with 50 nM TG or vehicle as in E.

Since IL4RA is an ER-client protein, its trafficking through the secretory pathway could be disrupted by ER stress, and thus impair IL-4/IL-13 responsiveness. To evaluate the contribution of transcriptional repression to IL-4/IL-13 signaling, we therefore had to mitigate the influence of ER stress on trafficking through the secretory pathway. To accomplish this, we needed to find stress conditions that were sufficient to activate the UPR and induce Il4ra transcriptional suppression, but that would not severely disrupt protein trafficking. We took advantage of the fact that low doses of ER-stress-inducing agents allow for UPR activation absent overt cytotoxicity (Rutkowski et al., 2006). Such treatment should suppress Il4ra expression while being less likely to grossly impair secretory pathway function. Indeed, suppression of Il4ra was first noticed in microarray data from cells treated for 24 hours with 50 ng/ml TM (Fig. 1C), to which cells are capable of long-term adaptation and which is 10–20-fold lower than the typical experimental dose of TM (Rutkowski et al., 2006). In a similar vein, 5 nM TG also activated the UPR and elicited Il4ra suppression to an extent comparable to the 50 nM dose used throughout this work (Fig. 7C).

We next used acquisition of resistance to digestion by endoglycosidase H (Endo H) to assess the efficiency of IL4RA trafficking. N-linked glycoproteins are sensitive to digestion by Endo H when they reside within the ER or cis-Golgi, but become resistant due to glycan trimming once they pass ER quality control and are trafficked past the cis-Golgi. In contrast, the endoglycosidase PNGase F cleaves all N-linked glycans regardless of cellular localization (Schwarz and Aebi, 2011). Analysis of vehicle-treated or 5 nM TG-treated MEF-protein lysates, while confirming overall diminished expression of IL4RA upon ER stress, revealed only modest impairment of IL4RA maturation (Fig. 7D). However, this low dose of TG was sufficient to suppress IL-13 signaling to an extent comparable to that of 50 nM TG, and that likely exceeded any effect due to the marginal impairment of IL4RA trafficking (Fig. 7E,F).

These data are consistent with the hypothesis that UPR-mediated suppression of IL-4/IL-13 signaling occurs through the transcriptional repression of Il4ra. If our hypothesis is correct, then this signaling impairment should not occur in the absence of C/EBPβ, since these cells fail to suppress Il4ra expression upon stress (Fig. 4F). Indeed, ER stress did not markedly impair IL-13 signaling in Cebpb/− MEFs (Fig. 7G). These data show that C/EBPβ plays an essential role in impairing IL-13 signaling during ER stress, and support our hypothesis that this impairment occurs through transcriptional suppression of Il4ra rather than through non-specific disruption to the secretory pathway.


UPR-mediated gene downregulation accounts for 20–50% of gene regulation during ER stress (Fig. 1A,C). The processes enriched in this group of suppressed genes are diverse, encompassing structural, biosynthetic and signal transductive cellular processes. The mechanisms involved, and the cellular functions impacted by this regulation have not been extensively studied. Here, we followed the impact of ER stress on IL-4/IL-13 signaling through the regulation of Il4ra, an essential component of the IL-4 and IL-13 receptors. We demonstrated that ER stress impaired IL-4/IL-13 signaling in multiple cell types, and this suppression was dependent on the transcription factor C/EBPβ. Through the regulation of Il4ra, we identified a novel pathway for UPR-mediated transcriptional repression dependent on the regulation of the C/EBPβ transcriptionally repressive isoform, LIP.

Taken together with previous work showing that eIF2α phosphorylation promotes expression of LIP during ER stress (Li et al., 2008), we propose the following model for suppression of Il4ra and other C/EBPβ-dependent genes (Fig. 8): ER stress first leads to eIF2α phosphorylation through PERK activation, inhibiting global protein synthesis (Harding and Ron, 1999; Shi et al., 1998). The effect of eIF2α phosphorylation alone is reduction in translation of both LAP and LIP isoforms of C/EBPβ (Fig. 5E). Thus, in contrast to ATF4, the translation of which is a direct consequence of inefficient ribosome assembly brought on by eIF2α phosphorylation (Harding et al., 2000), LIP translation appears not to be regulated by eIF2α phosphorylation per se. Rather, eIF2α dephosphorylation by the GADD34/PP1 complex leads to resumption of protein synthesis (Novoa et al., 2001), which promotes expression of LIP by increased synthesis, attenuated degradation, or both (Li et al., 2008). LIP then suppresses transcription of Il4ra, leading to decreased receptor expression and diminished responsiveness to exogenous IL-4 or IL-13.

Fig. 8.

The UPR-mediated suppression of Il4ra occurs through the regulation of C/EBPβ (LIP). Model for the UPR-mediated repression of IL-4/IL-13 signaling. See Discussion for details.

In general, LIP acts as a transcriptional inhibitor by dimerizing with other bZIP transcription factors (Vinson et al., 2002), including LAP (Descombes and Schibler, 1991), C/EBPα (Tsukada et al., 2011), ATF4 (Vallejo et al., 1993) and CHOP (Fawcett et al., 1996; Ron and Habener, 1992). Direct binding of C/EBPβ to the Il4ra promoter (Fig. 4E) and the lack of an effect upon removal of the UPR-regulated bZIPs ATF6α, CHOP and XBP1 (by virtue of IRE1α deletion) makes suppression through a LIP–LAP interaction or a LIP-C/EBPα interaction most likely, though other possible inhibitory heterodimers cannot yet be excluded. Indeed, even though LIP overexpression alone was sufficient to downregulate Il4ra expression, ER stress had a further suppressive effect (Fig. 6B). This effect could be simply explained by the observation that ER stress results in an increase in LIP expression even when LIP is overexpressed (Fig. 6A). However, it might instead result from stress-dependent production of ATF4, the binding and activity of which is attenuated by LIP (Li et al., 2008). Alternatively, there might be some other stress-dependent modification of C/EBPβ – for example phosphorylation (Nakajima et al., 1993; Trautwein et al., 1993). An augmenting effect of ER stress, beyond merely increasing LIP expression and perhaps independent of C/EBPβ entirely, might also account for the comparatively modest effects of ER stress on an Il4ra-responsive lacZ reporter (Fig. 4B). While C/EBPβ is clearly necessary for Il4ra suppression, whether it is truly fully sufficient remains an open question. In any case, this is the first time to our knowledge that C/EBPβ has been shown to be involved in ER stress-dependent regulation of Il4ra expression and inflammatory signaling, and it is consistent with the general role of C/EBPβ as a regulator of immune and inflammatory gene expression (Tsukada et al., 2011).

Our data also illustrate the importance of GADD34-mediated dephosphorylation of eIF2α and the resetting of translation efficiency in regulating LIP levels. In agreement with earlier work (Li et al., 2008), LIP does not accumulate until later times (4 hours and longer) after UPR activation (data not shown), and GADD34-mediated dephosphorylation of eIF2α is required for LIP production (Fig. 5E). This is in contrast to the eIF2α-regulated translation of ATF4, which occurs as early as 20 minutes after exposure to ER stress (Harding et al., 2000) and which is blocked by the activity of GADD34 (Novoa et al., 2001). Thus, it seems unlikely that LIP is preferentially synthesized by the same mechanism as ATF4, which involves inefficient ribosome assembly. One possibility is that the action of GADD34 reduces eIF2α phosphorylation to levels that are only modestly elevated relative to unstressed cells, and that this modestly elevated amount of phosphorylated eIF2α has effects on eIF2B availability and mRNA translation that are distinct from those induced by higher levels of phosphorylated eIF2α. Alternatively, eIF2α phosphorylation might be required simply to clear existing pools of C/EBPβ through translational inhibition, given the short half-life of LAP and especially LIP; then resumption of protein synthesis due to GADD34 action along with enhanced stability of LIP (Li et al., 2008) might combine to result in the LIP increases seen here.

The UPR-mediated suppression of Il4ra attenuates the ability of a cell to respond to IL-4 or IL-13. IL-4 and IL-13 are pro-inflammatory cytokines secreted from TH2 immune cells in response to parasitic infections and allergens (Denburg, 1998; Wynn, 2003). These cytokines, which have partially overlapping functions based on their shared IL4RA receptor component, stimulate JAK/STAT signaling cascades, and most cell types in the body are responsive to one or the other. The effects of IL4RA ligation vary based on cell type and range from stimulation of immune responses (TH2 cell differentiation, B cell proliferation, eosinophil recruitment and IgE class switching) to enhancement of secretory processes (mucus production, collagen synthesis and chemokine expression) to induction of cellular structural changes (smooth muscle contraction and Goblet cell hyperplasia) (Hershey, 2003; Holgate, 2011; Wills-Karp and Finkelman, 2008; Wynn, 2003). Due to these effects, IL-4/IL-13 signaling has been extensively studied in connection to inflammatory diseases such as asthma and inflammatory bowel disease (IBD) (Mitchell et al., 2010; Skinnider et al., 2002; Wynn, 2003). ER stress has been linked to diseases of inflammation including atherosclerosis, IBD and obesity (Adolph et al., 2012; Hotamisligil, 2010; Seimon and Tabas, 2009). While there are many possible mechanisms by which ER stress could influence inflammatory signaling, including through modulation of the JNK and NF-KB pathways (Hotamisligil, 2010; Kitamura, 2011), it is tempting to speculate that a dysregulated UPR might fail to suppress IL-4/IL-13 signaling, leaving cells hyperresponsive. Conversely, in physiological (as opposed to pathological) circumstances of UPR activation, suppression of IL-4/IL-13 signaling might be counterproductive. One example would be B cell differentiation, which is known to be accompanied by UPR activation (Gass et al., 2002; van Anken et al., 2003); suppression of IL-4 signaling would cripple class-switching to IgG1 (Rothman, 1993). It is noteworthy that B cell differentiation activates a non-canonical PERK-independent response (Gass et al., 2008; Masciarelli et al., 2010), which would circumvent the effects of PERK on Il4ra expression. Ultimately, it will be interesting to test whether animals with genetically altered UPR signaling, particularly in the PERK pathway, are differentially sensitive to the varied biological effects of IL-4 and IL-13.

Our current investigation into the regulation of Il4ra identified an unexpected cellular consequence of UPR-mediated gene suppression–suppression of IL-4/IL-13 signaling. Il4ra is but one of many genes suppressed during ER stress, and it is likely that such broad suppression is the combined effect of several mechanisms including the one elucidated here. Uncovering these mechanisms promises to substantially expand the understanding of the UPR as a central homeostatic response intimately intertwined in cellular physiology and pathophysiology.

Materials and Methods

Reagents and cell lines

TM and TG were purchased from EMD Millipore (Darmstadt, Germany); and ActD was purchased from Sigma Aldrich (Saint Louis, MO). IL-13 and IL-4 were from Peprotech (Rocky Hill, NJ). MEFs were harvested as described previously (Rutkowski et al., 2006) from Atf6α+/− (Wu et al., 2007), Ire1α+/− (Lee et al., 2002), Chop+/− (Zinszner et al., 1998) and eIF2α S/A (Scheuner et al., 2001) heterozygote intercrosses.

Gadd34−/− and Gadd34+/+ (Novoa et al., 2003) MEFs were generously provided by Dr David Ron (University of Cambridge); and the Cebpb−/− and Cebpb+/+ (Sterneck et al., 1997) MEFs were kindly provided by Dr Peter Johnson (NIH). The NIH 3T3 MEF line was from ATCC (Manassas, VA). The A549 cell line was donated by Dr John Engelhardt (University of Iowa), and the I.29μ+ B cell line was provided by Dr Linda Hendershot (St. Jude Children's Research Hospital).

Cell culture

MEFs were isolated and cultured as previously described (Rutkowski et al., 2006). The I.29μ+ B cell line was cultured as described in (Ma et al., 2010) at 5% CO2. The A549 cell line was maintained in DMEM/F12 1∶1 (Invitrogen) media supplemented with 10% fetal bovine serum (FBS; Hyclone), Glutamax, sodium pyruvate, and penicillin streptomycin (Invitrogen) at 37°C in a 5% CO2 incubator. ActD (10 µg/ml) was used for a total of 12 hours with media and drug replacements after 6 hours. Amino acid deprivation experiments were conducted as described previously (Li et al., 2008) with minor alterations to culture media. Met-/Cys-high glucose DMEM (Invitrogen) was supplemented with 10% dialyzed FBS, glutamine, non-essential amino acids, sodium pyruvate and penicillin streptomycin (Invitrogen).

DNA transfection

For myocilin overexpression, NIH 3T3 cells were transfected using Lipofectamine 2000 (Invitrogen) with one of three constructs: CMV-tdTomato, CMV-wild-type myocilin (MYOC) or CMV-mutant myocilin (MYOC-Y437H) according to the manufacturer's instructions. Forty-eight hours after transfection, the cells were treated with vehicle or TM (500 ng/ml) for 16 hours. Flow cytometry was used to isolate cells expressing the transfected protein tdTomato (control vector) or dsRed (myocilin constructs). RNA was isolated from the sorted cells, and qRT-PCR analysis conducted. The myocilin constructs were a gift from Dr Val Sheffield (University of Iowa).

Primary MEFs were transfected with GFP, LIP or ATF2 expression constructs using an Amaxa Nucleofector II and a Mouse/Rat Hepatocyte Nucleofector Kit (Lonza, cat no. VPL-1004) according to manufacturer's protocol. The cellular expression of GFP confirmed a transfection efficiency greater than 60%. The C/EBPβ (LIP) construct was originally designed and characterized by Descombes and Schibler (Descombes and Schibler, 1991) and was purchased from Addgene (construct no. 12561). The ATF2 construct described in (Makino et al., 2006) was a gift from Dr Shunsuke Ishii (University of Tsukuba).

RNA and protein analysis

RNA isolation and qRT-PCR were conducted as described previously (Rutkowski et al., 2006). qRT-PCR primer sequences can be found in supplemental material Table S1. Pathway analysis was performed using FunNet (Prifti et al., 2008).

Total cellular protein lysates were prepared and immunoblotting conducted as described previously (Rutkowski et al., 2006) with minor alterations. Endo H sensitivity assays were conducted as described previously (Wu et al., 2007). Lysates were separated on 9% Tris-tricine gels, and blots developed using a ChemDoc-IT imaging system (UVP). Antibodies were from various suppliers: C/EBPβ (C19) (Santa Cruz; sc-150), p-eIF2α (Invitrogen; 44728G), Calnexin (Enzo Life sciences; ADI-SPA-865), IL4RA (Santa Cruz; sc-165974), STAT6 (Cell Signaling, no. 9362), p-STAT6 (Cell Signaling, no. 9361) and TRAPα (generously provided by Dr Ramanujan Hegde, MRC laboratory, Cambridge).

Promoter analysis

The Il4ra promoter (−1785/+715 bp relative to the TSS) was amplified from genomic DNA and cloned into pcDNA5-TO-lacZ vector (Invitrogen), replacing the CMV-tet-operator promoter. Promoter truncations were made using standard cloning techniques. Transfections were carried out in NIH 3T3 MEFs using Lipofectamine 2000 (Invitrogen). Promoter activity was assessed using the Dual Light Assay Kit (Tropix) according to the manufacturer's instructions. Transcription factor binding prediction was conducted using PROMO software (Farré et al., 2003; Messeguer et al., 2002).

Chromatin immunoprecipitation

ChIP assays were preformed as described in (Hitchler et al., 2006) with minor modifications. Briefly, MEFs were fixed with 1% formaldehyde for 10 minutes at room temperature. The formaldehyde was then quenched by adding 1 M glycine for 10 minutes at room temperature. The cells were scraped and collected in 1× PBS. The cells were then pelleted and resuspended in 500 µl sonication buffer (50 mM Tris-HCl pH 8.1, 10 mM EDTA and 1% SDS and 1 mM PMSF). Samples were sonicated to produce DNA bands between 100 and 1000 bp. Samples were then diluted (1∶10) in dilution buffer (0.01% SDS, 1.1% Trition X-100, 1.2 mM EDTA, 16.7 mM Tris-HCl pH 8.1, 167 mM NaCl and 1 mM PMSF). Samples were precleared using 50 µl protein G agarose (Pierce) with BSA (500 ng/ml) for 1 hour at 4°C. Approximately 1/100th of sample was removed for input control and the remainder was split into two equal aliquots and immunoprecipitated by C/EBPβ (C19, Santa Cruz) or non-specific IgG (12–370; Millipore) overnight at 4°C. Washing, elution, and cross-linking reversal were carried out according to the aforementioned protocol. DNA was purified using a standard phenol/chloroform extraction protocol. Samples were then analyzed by real-time RT-PCR with an annealing temperature of 58°C.

Signaling experiments

Mouse embryonic fibroblasts and A549 cells were plated in 6-well dishes. Twenty-four hours after plating, culture medium was replaced with DMEM/F12 1∶1 (Invitrogen) supplemented with Serum Replacement (Peprotech, SR-100), Glutamax, sodium pyruvate and penicillin streptomycin (Invitrogen); TM, TG or vehicle were added at that time. After 16 hours of ER stress treatment, 50 ng/ml of IL-4 or IL-13 was added to the media for 1 hour and protein lysates were harvested.


We would like to thank P. Johnson for the Cebpb MEF lines, D. Ron for the Gadd34 MEF lines, J. Engelhardt for the A549 cell line and L. Hendershot for the I.29μ+ B cell line. We also thank the Sigmund, Engelhardt, Strack and Choudhury labs at the University of Iowa for sharing equipment and reagents; V. Sheffield and G. Zode for the myocilin constructs; S. Ishii for the ATF2 construct; P. Gonzalez-Alegre, C. Yeaman, R. Domann, B. Banfi and A. Choudhury for advice and critical comments; and D. Kelpsch, H. Tyra and D. McCabe for technical assistance.


  • Author contributions

    The experiments were conceived and designed by A.M.A. and D.T.R.; A.M.A. conducted the experiments. A.M.A. and D.T.R. analyzed all data and prepared the manuscript.

  • Funding

    This work was funded by the National Institute of Diabetes and Digestive and Kidney Diseases [grant number R01 DK084058 to D.T.R.]. Deposited in PMC for release after 12 months.

  • Supplementary material available online at

  • Accepted May 23, 2013.


View Abstract