Stress granule formation mediates the inhibition of colonic Hsp70 translation by interferon-γ and tumor necrosis factor-α

Abstract

Mucosal inflammation, through cytokines such as interferon-γ (IFN-γ) and tumor necrosis factor-α (TNF-α), has many effects on the intestinal epithelium, including selective translational inhibition of the cytoprotective protein heat shock protein 70 (Hsp70). To further elucidate the mechanisms underlying this effect, we examined the role of stress granules in mediating the actions of these proinflammatory cytokines. Using conditionally immortalized young adult mouse colonic epithelial cells, we demonstrate that IFN-γ and TNF-α, which upregulate eukaryotic initiation factor-α (eIF-2α) phosphorylation and reduce Hsp70 translation, significantly enhance stress granule formation in heat-shocked intestinal epithelial cells. The IFN-γ and TNF-α effects in upregulation of stress granule formation and downregulation of Hsp70 were eIF-2α dependent, and the effect could be negated by blocking eIF-2α phosphorylation with use of an RNA-dependent protein kinase inhibitor. Correspondingly, IFN-γ and TNF-α increased binding of cytoplasmic proteins to the 3′-untranslated region of Hsp70 mRNA, suggesting specific recruitment of Hsp70 to stress granules as the mechanism of IFN-γ and TNF-α inhibition of Hsp70 translation. We thus report a novel linkage between inflammatory cytokine production, stress granule formation, and Hsp70 translation inhibition, providing additional insights into the response of intestinal epithelial cells to inflammatory stress.

intestinal heat shock proteins are highly conserved, multifunctional proteins essential for maintenance of gut health, and their altered expression can render the mucosa more susceptible to sustained and severe injury and possibly malignant transformation (4, 8, 9, 16, 21, 30). These proteins, particularly Hsp70 and Hsp25, are highly regulated by a complex set of extracellular cues, including gut-derived microbial signals such as LPS and short-chain fatty acids (SCFAs) (2, 14, 15, 23, 26). In colitis, however, inducible heat shock proteins, particularly Hsp70, are significantly and selectively downregulated, a process that renders the mucosa more susceptible to injury and stress. We have previously shown that this effect is in part mediated by the increased levels of proinflammatory cytokines such as interferon-γ (IFN-γ) and tumor necrosis factor-α (TNF-α) that stimulate a translational downregulation of Hsp70. These cytokines stimulate the phosphorylation of the RNA-dependent protein kinase (PKR), which in turn phosphorylates eukaryotic initiation factor-α (eIF-2α), resulting in reduced Hsp70 protein translation in inflammatory bowel diseases and experimental colitis (8, 9).

Stress granules are cytoplasmic structures that serve as dynamic triage sites for proteins and mRNA during stresses. They are associated with cell survival during environmental insults. Phosphorylation of eIF-2α leads to stress granule formation. Under these conditions, nonfunctional ribosome components, including untranslated mRNAs, 40S ribosomal subunits, various translation initiation factors, and RNA-binding proteins, accumulate in the cytoplasm to form dense aggregates (1, 12, 13, 18).

Messenger RNAs are recruited into the stress granules by selective interactions between RNA binding proteins and the 3′-untranslated regions (3′-UTRs) of specific mRNA; thus the formation of stress granules is recognized as a mechanism of mRNA-specific translational regulation (1, 13, 18). Interestingly, few studies have specifically looked at the structure and function of stress granules in intestinal epithelial cells. Given our previously published finding of enhanced phosphorylation of eIF-2α in colitis, we hypothesized that stress granules may be involved in IFN-γ- and TNF-α-induced inhibition of Hsp70 translation (8, 9).

In this study, we demonstrate that IFN-γ and TNF-α, which upregulate eIF-2α phosphorylation and reduce Hsp70 translation, significantly enhance stress granule formation in heat-shocked intestinal epithelial cells. The IFN-γ and TNF-α effects in upregulation of stress granule formation and downregulation of Hsp70 were eIF-2α dependent and the effect could be negated by blocking eIF-2α phosphorylation using PKR inhibitor (PKR-I). Correspondingly, IFN-γ and TNF-α increased binding of cytoplasmic proteins to the 3′-UTR of Hsp70 mRNA, thus suggesting specific recruitment of Hsp70 to stress granules as the mechanism of IFN-γ and TNF-α inhibition of Hsp70 translation.

MATERIAL AND METHODS

Cell culture.

Conditionally immortalized young adult mouse colonic epithelial cells (YAMC) were used for all studies (a generous gift from Dr. Robert Whitehead, Vanderbilt University, Nashville, TN) (32). The YAMC cell line is one of the very few noncancerous colonic epithelial cell lines that have been established. Hsp70 expression is minimal in YAMC cells but can be consistently induced by heat stress, SCFAs, and LPS. Therefore, the YAMC cell line is a suitable in vitro model for studying the regulatory mechanisms of colonic epithelial Hsp70 expression in colitis (8, 9). YAMCs were grown at 33°C as previously described. Cells were fed IFN-γ-free medium and switched to the nonpermissive condition of 37°C for 16–24 h. During this time, the SV40 large T antigen is no longer produced and any remaining protein misfolds because of a temperature-sensitive mutation at amino acid 58 (tsA). During this period, cells were treated with the cytokines murine IFN-γ (200 U/ml) and TNF-α (100 ng/ml) for 8 h prior to heat shock and harvest. For in vitro assessment of Hsp70 production, cells were heated to 42°C for 23 min and returned to 37°C for 120 min, inducing maximal Hsp induction (21). To inhibit phosphorylation of eIF-2α, the PKR-I (Calbiochem, San Diego, CA) was added 30 min before heat shock induction in some experiments (8, 10).

In all experiments, cells were harvested by rinsing twice and scraping into ice-cold PBS. Cells were then pelleted (14,000 g × 20 s) and lysed for RNA and protein extraction as described in the following section.

Western blot analysis.

For analysis of total cell proteins, YAMCs were homogenized in lysis buffer [10 mM Tris pH 7.4, 5 mM MgCl₂, complete protease inhibitor cocktail (Roche Molecular Biochemicals, Indianapolis, IN), 50 U/ml DNAse (Amersham, Piscataway, NJ), and 50 U/ml RNase (Ambion, Austin, TX)]; 10 μl was removed for protein analysis using the bicinchoninic acid method. To the remainder, 3× Laemmli stop solution was added followed by heating to 65°C for 10 min.

Cytoplasmic and nuclear protein lysates were harvested by using NE-PER nuclear and cytoplasmic extraction reagents (Pierce, Rockford, IL) according to the manufacturer's instruction. Each fraction was resuspended in lysis buffer as described above.

Protein (20 μg) was separated by SDS-PAGE and transferred to polyvinylidene difluoride membranes in 25 mM Tris, pH 8.8; 192 mM glycine; 15% vol/vol methanol. Membranes were blocked with 5% wt/vol nonfat dry milk in Tween-Tris-buffered saline (TTBS). Primary antibodies for Hsp70 (SPA810; Stressgen), Hsc70 (SPA815; Stressgen), β-actin (AAN01; Cytoskeleton, Denver, CO), eIF-2α (9722; Cell Signaling, Beverly, MA), phosphorylated eIF-2α Ser51 (9721; Cell Signaling), and TIA-1 (Santa Cruz Biotechnology), were added and incubated overnight at 4°C. Membranes were washed with TTBS, incubated with horseradish peroxidase-conjugated species-appropriate secondary antibodies (Jackson Immunoresearch, West Grove, PA) for 1 h at room temperature, and developed by use of an enhanced chemiluminescence system (Supersignal; Pierce).

Quantification of images was done by scanning densitometry with NIH Image J 1.54 software (National Institutes of Health, Bethesda, MD).

Real-time PCR for Hsp70 mRNA.

Total RNA was extracted from pelleted YAMCs by Trizol (Invitrogen, Grand Island, NY) according to the manufacturer's instructions. Complementary DNA was synthesized by using SuperScript II (Invitrogen) and random hexanucleotide primer. The mouse Hsp70 (NM_010478) sequence was downloaded from GenBank. The forward and reverse primers used were as follows: mouse Hsp70, 5′-TATGCCTTCAACATGAAGAGCGCC-3′ and 5′-CTTGTCCAGCACCTTCTTCTTGTC-3′; mouse GAPDH, 5′-GGCAAATTCAACGGCACAGT-3′ and 5′-AGATGGTGATGGGCTTCCC-3′. Real-time PCR was performed by using an iCycler (Bio-Rad) with iQSYBR Green PCR supermix (Bio-Rad). The two-step quantification cycling protocol was used. The C_t value was defined as the cycle number at which the fluorescence crosses a fixed threshold above the baseline. As a relative quantitation, fold changes were measured by the ΔΔC_t method. For each sample, the C_t value of Hsp70 mRNA was measured and compared with the GAPDH endogenous control as ΔC_t (ΔC_t = C_{t Hsp70} − C_{t GAPDH}). The fold change of Hsp70 mRNA in the unknown sample (ΔC_{t Unknown}) relative to that in the control sample (ΔC_{t Control}) was determined by 2^−ΔΔCT, where ΔΔC_t = ΔC_{t Unknown} − ΔC_{t Control} (28).

Luciferase reporter assays.

YAMC cells were transiently transfected with modified pGL3-promoter constructs (E1761, Promega) and pRL-TK plasmids (Renilla luciferase driven by thymidine kinase promoter, E2241, Promega) by using TransIT LT-1 (Mirus; Madison, WI) transfection reagent according to the manufacturer's recommendation. Four hours after transfection, cells were subjected to IFN-γ (200 U/ml) and TNF-α (100 ng/ml) for 8 h prior to heat shock. Cells were harvested by shaking in 500 μl of active lysis buffer (Promega). Firefly and Renilla luciferase activity in the lysate was determined with a dual-luciferase reporter assay system, according to the manufacturer's instructions (Promega). Triplicate samples were assayed for firefly luciferase activity normalized to Renilla (20).

RNA electromobility shift assay.

YAMCs were scraped and pelleted immediately after heat induction. Cytoplasmic protein lysates were harvested by using NE-PER nuclear and cytoplasmic extraction reagents (Pierce) according to the manufacturer's instructions. ³²P-labeled mouse Hsp70.1 (NM_010478.2) 3′-UTR RNAs, as shown in the Fig. 3A diagram, were transcribed from DNA templates using mMessage mMachine (Ambion) according to the manufacturer's instructions. Ten micrograms of cytoplasmic lysates were incubated with 10,000 cpm RNA probe and 150 μg of heparin (Sigma) for 15 min at room temperature. Then 10 U of RNase T1 (Invitrogen) were added and the incubation continued for 15 min. Samples were electrophoresed through a 4% (wt/vol) native PAGE using Tris-borate-EDTA buffer. Gels were dried at 65°C and protein-RNA complexes were visualized by autoradiography (20).

Fig. 3.

RNA-dependent protein kinase (PKR) inhibitor (PKR-I) inhibited stress granule assembly and reversed the effects of IFN-γ and TNF-α on Hsp70 translation. A: immunofluorescence micrographs of stress granule. YAMCs were treated with IFN-γ (200 U/ml) and TNF-α (100 ng/ml) for 8 h and with PKR-I (30 nM) for 30 min and then subjected to heat shock, followed by staining for TIA-1 (green), eIF-4E (red), and DAPI (blue). Colocalization of TIA-1 (green) and eIF-4E (red) in the cytoplasm indicates the stress granules (yellow). YAMC cells without heat stress were also analyzed. Image shown is representative of 4 individual experiments. B: quantification of percentage cells that exhibit stress granules. Results are means ± SE, n = 4, *P < 0.05. C: Western blots of Hsp70, Hsc70, phospho-eIF-2α, and eIF-2α YAMC cells were treated with IFN-γ and TNF-α for 8 h prior to heat shock or heat shock alone. Increasing doses of PKR-I (0, 10, and 30 nM) were added 30 min before heat shock. Results are means ± SE, n = 4.

Immunofluorescence staining.

YAMC cells plated on glass coverslips were fixed in 4% paraformaldehyde for 10 min and then 100% vol/vol methanol at −20°C for 10 min. Fixed cells were permeabilized by incubation in 0.1% Triton X-100 in PBS for 15 min and incubated in 10% bovine serum albumin in PBS for 1 h at room temperature. Cells were incubated with primary antibodies, anti-TIA-1 (sc-1751, Santa Cruz Biotechnology, Santa Cruz, CA) and anti-eIF-4E (Cell Signaling), overnight at 4°C. For visualization of TIA-1 and eIF-4E, coverslips were incubated with both Cy2 and Cy5 conjugated AffiniPure donkey IgG (Jackson Immunoresearch Laboratories) in PBS for 1 h at room temperature. Coverslips were incubated in 4′-6-diamidino-2-phenylindole (DAPI) for 5 min at room temperature for nuclear staining and then mounted on glass slides using a solution of n-propyl gallate in 50% glycerol (Sigma, St. Louis, MO) and examined using an Olympus DSU Spinning Disk Confocal microscope system. All images were acquired by using the Olympus system and Slidebook 5.0 software with standard image processing techniques (6, 12). To calculate the percentage of cells with stress granule formation, 100 cells were counted in more than five microscopic fields on every slide.

In vitro Hsp70 overexpression.

CMV-Hsp70 expression construct was generated by inserting a full-length mouse Hsp70.1 coding sequence (1929 bp) into a pCMV-HA mammalian expression vector (Clontech, Mountain View, CA). To overexpress Hsp70, YAMC cells were transiently transfected with CMV-Hsp70 constructs using TransIT LT-1 transfection reagent according to the manufacturer's recommendation. Sixteen hours after transfection, cells were subjected to IFN-γ (200 U/ml) and TNF-α (100 ng/ml) for 8 h prior to Western blot analysis and immunofluorescence staining after heat shock.

Immunoprecipitation.

Cytoplasmic protein lysates were harvested from 1 × 10⁶ YAMC cells by using NE-PER nuclear and cytoplasmic extraction reagents. Lysates were incubated overnight at 4°C with 2 μg anti-TIA-1 antibody (same as used in Western blot) coupled to 50 μl protein G-Sepharose (Sigma). Immunoprecipitates were pelleted by centrifugation, washed four times with NT2 buffer [50 mM Tris·HCl (pH 7.4), 150 mM NaCl, 1 mM MgCl₂, and 0.05% Nonidet P-40]. RNA was extracted from immunoprecipitated lysates by Trizol and reverse transcribed into cDNA for real-time PCR as described above (11).

Statistical analysis.

Results are presented as means ± SE for the indicated number of experiments. The results of multiple experiments were analyzed by Student's t-test or ANOVA with Bonferroni correction for multiple comparisons.

RESULTS

Effect of cytokines on Hsp70 induction in YAMC cells.

IFN-γ and TNF-α are proinflammatory cytokines specifically increased in colitis. YAMC cells were used to analyze the cellular effects of these proinflammatory cytokines on Hsp70 expression. After IFN-γ (200 U/ml) and TNF-α (100 ng/ml) pretreatment for 8 h, Hsp70 was induced by heat shock. As expected, as shown in Fig. 1, A and B, Hsp70 was significantly increased after heat shock compared with basal levels. However, IFN-γ and TNF-α pretreatment dampened Hsp70 expression compared with cells treated with heat shock alone.

Fig. 1.

IFN-γ and TNF-α downregulate Hsp70 protein expression, but not mRNA, in young adult mouse colonic epithelial (YAMC) cells. YAMC cells were treated with IFN-γ (200 U/ml) and TNF-α (100 ng/ml) for 8 h and then stimulated with heat shock (HS). Cells were harvested at 60 min after heat shock. Cells without heat shock were harvested as basal. A: Western blots of Hsp70, Hsc70, β-actin, phospho-eIF-2α (p-eIF2α), and eukaryotic initiation factor-α (eIF-2α). Image shown is representative of 4 individual experiments all with similar results. B: densitometry results of Western blot of Hsp70 normalized to Hsc70. C: Hsp70 mRNA abundance was analyzed by real-time PCR. Results are means ± SE, n = 4. *P < 0.05 compared with heat shock alone by Student's t-test.

To determine the effect of IFN-γ and TNF-α on transcription, total cellular Hsp70 mRNA levels were measured by real-time PCR (Fig. 1C). Interestingly, high levels of Hsp70 mRNA were induced 60 min after heat shock in samples either with or without IFN-γ and TNF-α pretreatment. Therefore, the inhibition of Hsp70 by IFN-γ and TNF-α treatment did not appear to be at the transcriptional level.

IFN-γ and TNF-α increase stress granule assembly in heat-shocked YAMC cells.

Consistent with our previously published data, the decrease in Hsp70 protein expression was accompanied by a concomitant increase in the level of phosphorylated eIF-2α without altering the levels of total eIF-2α (Fig. 1A). As it is known that phosphorylated eIF-2α is associated with stress granule formation, and that stress granule formation is associated with translational regulation (1, 12), we investigated assembly of stress granules in YAMC cells. YAMC cells were fixed 60 min after heat shock and immunostained for TIA-1 (green), the RNA binding protein necessary for stress granule formation, and eIF-4E (red), a member of a group of small ribosomal subunits known to be components of stress granules (6, 12). Colocalization of TIA-1 and eIF-4E (yellow) indicates stress granule formation (Fig. 2A arrow). YAMC cells without heat shock did not demonstrate stress granule formation (Fig. 2A). Thermal stress alone did produce stress granules in a small portion (less than 30%) of the cells; however, stress granule formation was significantly increased in the cells pretreated with IFN-γ and TNF-α prior to heat shock (Fig. 2B) to over 80% of cells with stress granules in cytoplasm (P < 0.05). These TIA-1 and eIF-4E associated stress granules in YAMC cells have comparable microscopic appearance to those reported in other cell lines (6, 12).

Fig. 2.

IFN-γ and TNF-α pretreatment enhanced stress granule formation in heat shock-induced YAMC cells. YAMC cells were treated with IFN-γ (200 U/ml) and TNF-α (100 ng/ml) for 8 h and then subjected to heat shock. Cells without heat shock treatment were analyzed as basal. A: immunofluorescence micrographs of stress granule assembly. Cells were fixed 60 min after heat shock, stained for TIA-1 (green), eIF-4E (red), and 4′-6-diamidino-2-phenylindole (DAPI; blue) as described inmaterial and methods. Colocalization of TIA-1 and eIF-4E in cytoplasm indicates stress granules (yellow, arrow). B: quantification of percentage of cells that exhibit stress granules (SGs). Results are means ± SE, n = 4. *P < 0.05 compared with heat shock alone by analysis of variance using a Bonferroni correction. C: Western blots of TIA-1, eIF-4E, and β-actin in whole cell lysates from cells with heat shock and IFN-γ and TNF-α (+) pretreatment. D: Western blot of TIA-1 and eIF-4E in nuclear and cytoplasmic lysates from cells with heat shock and IFN-γ and TNF-α (+) pretreatment. Images shown are representative of 4 individual experiments. E: densitometry results of Western blots of TIA-1 and eIF-4E.

The cytokines did not affect the total levels of TIA-1 or eIF-4E in heat-stressed cells (Fig. 2C) or the distribution of eIF-4E; however, the distribution of TIA-1 between the cytoplasm and the nucleus was altered (Fig. 2D). With IFN-γ and TNF-α pretreatment, a significant portion of TIA-1 was translocated from the nucleus to the cytoplasm in heat-shocked YAMC cells (Fig. 2E).

Stress granule assembly in response to IFN-γ, TNF-α, and heat shock precedes inhibition of Hsp70 expression.

Our data demonstrate that the dual stress of inflammatory cytokines (IFN-γ, TNF-α) and heat shock is associated with stress granule formation and decreased Hsp70 translation. However, two possibilities exist. Cytokine-mediated intestinal inflammation may lead to stress granule formation and Hsp70 reduction, or cytokine-mediated Hsp70 reduction may render the enterocytes more susceptible to cellular stress, with the subsequent formation of stress granules. If our hypothesis that stress granule formation assembly precedes Hsp70 inhibition is valid, then inhibition of stress granule assembly in the face of inflammatory cytokine and heat shock stress would be expected to preserve Hsp70. Conversely, if Hsp70 inhibition precedes stress granule assembly, then preservation of Hsp70 should decrease stress granule development. We evaluated both of these possibilities.

First, to inhibit stress granule formation, YAMC cells were pretreated with a small molecule inhibitor of PKR, PKR-I (as demonstrated in the schematic Fig. 8) (10). As shown in Fig. 3C, PKR-I treatment decreased IFN-γ and TNF-α induced eIF-2α phosphorylation in a dose-dependent fashion in heat-shocked cells. This degree of inhibition is comparable to our previously published experiments using small interfering RNA to knockdown PKR and block eIF-2α phosphorylation (8). PKR-I treatment (30 nM) after cytokine and heat shock treatment concomitantly decreased the percentage of cells with stress granule presence (P < 0.05, Fig. 3, A and B), and rescued Hsp70 translation previously inhibited by IFN-γ and TNF-α. PKR-I had a dose-dependent effect on rescuing cytokine-inhibited Hsp70 translation, with a maximal rescue to over 70% at a concentration of 30 nM (Fig. 3C). PKR-I had no effect on cells with heat shock only.

Fig. 8.

Model of inflammation-associated translational downregulation of Hsp70 and stress granule formation. As shown at left, under basal conditions or after recovery in response to heat stress alone, eIF-2α induces formation of the translational initiation complex on Hsp70 mRNA. This translation complex includes other translational elements including eIF-4E, 40S, and 60S ribosome. Thus substantial Hsp70 proteins are continuously produced from the formed polyribosome. In contrast, as shown on the right, if eIF-2α is phosphorylated by PKR, formation of the initiation complex is inhibited by phospho-eIF-2α, which is a competitive inhibitor to the nonphosphorylated eIF-2α in Hsp70 translation. RNA-binding proteins, like TIA-1, then bind to the 3′-UTR of untranslated Hsp70 mRNA to further inhibit the Hsp70 translation. TIA-1 autoaggregation leads to the accumulation of these nonfunctioning complexes at discrete cytoplasmic foci known as stress granules. Our data demonstrate that, with heat stress alone, a small portion of cells develop stress granules. The substantial amount of Hsp70 mRNA induced by heat shock overrides the mRNA recruitment and translational inhibition in stress granules. However, the stress of the proinflammatory cytokines, IFN-γ and TNF-α, in addition to heat stress leads to stress granule formation in almost every cell, thus significantly sequestering Hsp70 mRNA in an untranslated state. IFN-γ and TNF-α leads to phosphorylation of eIF-2a by enhanced phosphorylation of PKR, an action that can be blocked by PKR-I. IFN-γ and TNF-α also promote the relocation of RNA-binding proteins, like TIA-1, from the nucleus to the cytoplasm, which further enhances the stress granule assembly.

Next, to examine the effect of high Hsp70 protein levels on stress granule formation, YAMC cells were transfected with a CMV-Hsp70 expression construct to increase Hsp70 levels prior to treatment with cytokines and heat stress. As shown in Fig. 4A, substantial expression of Hsp70, comparable to levels induced by heat shock, was detected in cells transfected with the CMV-Hsp70 plasmid. The LT-1 transfection reagent alone was used as control and did not alter Hsp70 expression. Additional treatment of IFN-γ and TNF-α also did not alter the Hsp70 level in transfected cells.

Fig. 4.

High Hsp70 levels did not block IFN-γ and TNF-α upregulated stress granule formation in heat shock-stressed YAMC cells. A: YAMC cells were transfected with CMV-Hsp70 expression construct 16 h before IFN-γ and TNF-α treatment. Cells were harvested for Western blots of Hsp70 and Hsc70. Cells without treatment, with heat shock alone and treated with LT-1 transfection reagent alone were analyzed as controls. B: immunofluorescence micrographs of stress granule. CMV-Hsp70-transfected YAMCs were treated with IFN-γ (200 U/ml) and TNF-α (100 ng/ml) for 8 h and then subjected to heat shock, followed by staining for TIA-1 (green), eIF-4E (red), and DAPI (blue). Colocalization of TIA-1 (green) and eIF-4E (red) in the cytoplasm indicates the stress granules (yellow). YAMC cells treated with LT-1 transfection reagent along were analyzed as control. Image shown is representative of 4 individual experiments. C: quantification of percentage cells that exhibit stress granules. Results are means ± SE, n = 4, *P < 0.05.

Stress granule formation was analyzed by immunofluorescence staining (Fig. 4B). In YAMC cells treated with the LT-1 transfection reagent alone as control, thermal stress alone produced stress granules in 40% of the cells, but stress granule formation was significantly increased by pretreatment with IFN-γ and TNF-α to 90% of cells (P < 0.05, Fig. 4C). In CMV-Hsp70-transfected cells, which maintained high Hsp70 levels, heat shock alone induced stress granule formation in about half of the cells. Again, cytokine pretreatment still increased stress granule formation to almost 100% of the cells (Fig. 4, B and C).

Increased binding of regulatory proteins to Hsp70 3′-UTR after IFN-γ and TNF-α treatment.

Stress granule formation begins with the binding of RNA proteins to the AU-rich regions of 3′-UTR repeats of specific mRNA. To document specific involvement of Hsp70 in stress granules, cytoplasmic lysates from YAMC cells after IFN-γ and TNF-α pretreatment and heat shock were incubated with radiolabeled in vitro transcribed RNAs corresponding to either the 3′ end 300 nucleotides of Hsp70.1 3′-UTR or to the complete 3′-UTR (650 nucleotides), followed by incubation with RNase to degrade unbound RNA. As shown in Fig. 5A, the full-length Hsp70.1 3′-UTR contained a 20-nucleotide AU-rich region, a potential RNA binding protein binding site, while the partial 3′-UTR probe did not. In contrast to the probe only, RNA-protein complexes with reduced mobility in a nondenaturing gel were observed in Fig. 5B. More binding was noted with use of the complete 3′-UTR probe than with the partial 3′-UTR probe, suggesting the requirement of AU-rich regions for binding. Correspondingly, more partial probes without protein binding were digested into fractions at the bottom of the gel when using the partial 3′-UTR probe. When the complete 3′-UTR probe was used, the binding of proteins to RNA increased with heat shock treatment. This binding was further enhanced by pretreatment with IFN-γ and TNF-α. Thus RNA binding protein binding characteristic of stress granule formation was specifically noted with Hsp70 mRNA in intestinal epithelial cells.

Fig. 5.

IFN-γ and TNF-α upregulated binding of cytoplasmic proteins to the Hsp70 mRNA AU-rich 3′-untranslated region (UTR) after heat shock. A: schematic of RNA probes corresponding to the complete Hsp70.1 3′-UTR and the partial 3′ end 300 nucleotides of the 3′-UTR. B: YAMC cells were treated with IFN-γ (200 U/ml) and TNF-α (100 ng/ml) for 8 h and then subjected to heat shock. Cytoplasmic lysates were prepared 1 h after heat shock. Binding of cytoplasmic proteins to Hsp70.1 mRNA 3′-UTR was analyzed by RNA-EMSA using radiolabeled Hsp70.1 3′-UTR probes. Image shown is representative of 4 individual experiments.

Hsp70 3′-UTR decreases protein translation in heat-shocked cells with cytokine pretreatment.

To determine whether increased binding to the AU-rich region of the Hsp70 3′-UTR corresponded to the previously noted decrease in Hsp70 protein translation, YAMC cells were transfected with a Hsp70.1 3′-UTR construct fused downstream of the firefly luciferase gene (pGL3–3′-UTR), as shown in the Fig. 6A schematic. A pRL-TK vector expressing Renilla luciferase was cotransfected as a control for transfection efficiency. Luminescence served as a measure of mRNA translation (20). In contrast to heat shock alone, pretreatment with IFN-γ and TNF-α significantly decreased measured luminescence (Fig. 6B) but did not change luciferase mRNA levels (Fig. 6C), indicating decreased translation of the chimeric luciferase mRNAs. This finding demonstrates that the AU-rich region of the Hsp70 3′-UTR important for stress granule formation is also important for the effect of IFN-γ and TNF-α on Hsp70 translational inhibition.

Fig. 6.

Hsp70 mRNA 3′-UTR facilitated translational inhibition by IFN-γ and TNF-α after heat shock. A: schematic of luciferase reporter assay constructs with and without the Hsp70 mRNA 3′-UTR. Luciferase vectors, pGL3/pGL3-Hsp70 3′-UTR and pRL-TK, were used in transient transfections. Sixteen hours later, cells were treated with IFN-γ (200 U/ml) and TNF-α (100 ng/ml) for 8 h and then subjected to heat shock. B: the relative firefly luciferase activity was normalized to Renilla luciferase. Luminescence was measured as an indication of differences in mRNA translation. C: firefly luciferase mRNA was measured by real-time PCR and normalized to Renilla luciferase mRNA. Results are means ± SE, n = 4. *P < 0.05 pGL3-Hsp70 3′-UTR compared with pGL3 by paired t-test.

IFN-γ and TNF-α increased cytoplasmic TIA-1-associated Hsp70 mRNA in heat-shocked cells.

Our data suggest that IFN-γ and TNF-α induce stress granule formation as well as binding of RNA binding proteins to specific AU-rich regions of Hsp70 mRNA; thus we next sought to demonstrate specific incorporation of Hsp70 into stress granules. TIA-1 and TIA-1 associated ribonucleoprotein (RNP) complexes characteristic of stress granules were immunoprecipitated by using an anti-TIA-1 antibody (Santa Cruz Biotechnology) from cytoplasmic lysates of YAMC. RNP-associated Hsp70 mRNA was analyzed by real-time PCR. Total cytoplasmic Hsp70 mRNA and total cytoplasmic mRNA were also measured (Fig. 7). After heat shock, Hsp70 mRNA was increased in TIA-1-associated RNP complexes. Compared with heat shock alone, IFN-γ and TNF-α pretreatment significantly further increased Hsp70 mRNA associated with TIA-1 in the cytoplasm (Fig. 7A). In contrast, although total cytoplasmic Hsp70 mRNA levels were greatly induced after heat shock, they were not further altered by the addition of IFN-γ and TNF-α. Total cytoplasmic mRNA levels were similar in all the samples. These results demonstrate that IFN-γ and TNF-α increases the recruitment of Hsp70 mRNA to the TIA-1-associated RNP complex component of stress granules.

Fig. 7.

IFN-γ and TNF-α increased cytoplasmic TIA-1-associated Hsp70 mRNA in YAMC cells after heat shock. YAMC cells were treated with IFN-γ (200 U/ml) and TNF-α (100 ng/ml) for 8 h and then stimulated with heat shock (42°C for 23 min). Cytoplasmic lysates were harvested. A: TIA-1 associated ribonucleoprotein (RNP) complexes were immunoprecipitated from cytoplasmic lysates. Hsp70 mRNA associated with the RNP complex was measured by real-time PCR and normalized to the heat shock-alone sample as fold change. B: total cytoplasmic Hsp70 mRNA. C: GAPDH mRNA in the cytoplasmic lysate as control. Results are means ± SE, n = 4. *P < 0.05 compared with heat shock alone by analysis of paired t-test.

DISCUSSION

Regulation of protein expression is a pivotal mechanism for cellular responses to different conditions and can occur at several levels, including mRNA transcription, assembly of the ribosomes that initiate translation, and posttranslation (7, 19, 25). Under various stress conditions, including heat, altered pH, or osmolarity, translational regulation is particularly important to allow stress proteins to be rapidly induced and preferentially expressed to ensure survival under severe situations (7, 19). Inflammation is a specific stress condition particularly relevant to intestinal physiology in which a variety of cytokines and chemokines are present that alter protein expression (5, 22).

Hsp70 plays a critical cytoprotective role in intestinal epithelial cells. Studies have demonstrated the greatest Hsp70 expression in the areas of the gastrointestinal tract exposed to the most extreme conditions: the stomach with extremely low pH and the colon with constant high exposure to intestinal bacteria (2, 14). In contrast, small intestinal cells that do not normally express Hsp70 begin exposure if surgical manipulation increases bacterial exposure (2). In addition to stabilizing the actin skeleton and protecting tight junctions, Hsp70 is also known to halt cell division, slow cell growth, and prevent apoptosis, which are additional cell protective processes (3, 4, 24, 27, 30, 31). It seems incongruous, then, that our previous paper demonstrated that in cell culture, animal models, and human tissue, proinflammatory cytokine (IFN-γ and TNF-α) exposure decreased Hsp70 protein expression after heat shock. This occurred without an alteration in transcription. Decreased de novo protein synthesis of Hsp70 by radiolabeled amino acid incorporation and absence of effect on protein degradation by pulse-chase experiments were consistent with translational regulation of Hsp70 (8).

Our present studies are notable for demonstrating increased Hsp70 recruitment to stress granules as the mechanism of translational downregulation. Experiments were conducted in one of the very few noncancerous colonic epithelial cell models, YAMC cells. This cell line retains key properties of colonic epithelia, including epithelial specific keratin expression and brush border-related enzyme expression (32). Hsp70 can be consistently induced after heat shock in YAMC cells, which makes it a suitable in vitro model for further analysis of detailed mechanisms of our previous in vivo finding in both mouse and human colonic mucosa of inflammatory cytokine mediated Hsp70 translational inhibition (8, 9).

Stress granules are cytoplasmic structures into which proteins and mRNA enter and leave during stress. They are associated with cell survival during environmental insults (1, 12). As demonstrated in Fig. 8, phosphorylation of eIF-2α is activated by distinct types of environmental stress and leads to stress granule formation (29). Phosphorylated eIF-2α competitively inhibits the nonphosphorylated eIF-2α and decreases levels of the eIF-2-GTP-tRNA^met complex required for loading the initiator methionine onto the 48S preinitiation complex required to begin translation. Absence of eIF2-GTP-tRNA^met allows the RNA binding proteins TIA-1 and TIAR to bind the mRNAs in the 40S ribosomal complex, leading to polysome disassembly. Autoaggregation of the COOH-termini of the TIA proteins and nonfunctioning ribosomal components leads to formation of stress granules (1, 12, 25).

Our present study reveals that the IFN-γ and TNF-α-induced decrease in Hsp70 protein expression is accompanied by an increase in the level of phosphorylated eIF-2α (Fig. 1A) (8). We have confirmed that stress granule production precedes Hsp70 inhibition by inhibiting PKR with the chemical compound PKR-I (Fig. 3C), which decreased stress granule formation (Fig. 3A) and was able to rescue Hsp70 translation inhibition by cytokines. As confirmation, we used a CMV-Hsp70 expression construct to force increased levels of Hsp70. With increased Hsp70 levels, IFN-γ and TNF-α still increased stress granule formation (Fig. 4), proving that Hsp70 effects are downstream of the effect of IFN-γ and TNF-α on stress granule formation.

We also demonstrated that cytokine treatment induced another early event in stress granule formation: specific increased RNA binding to AU-rich regions of the 3′-UTR repeat of Hsp70 mRNA (Fig. 5). The importance of this binding to Hsp70 translational inhibition was demonstrated in luciferase assays. YAMC cells were transfected with constructs with and without the Hsp70 mRNA 3′-UTR. Luminescence served as a measure of mRNA translation. Inflammatory cytokine and heat shock stress inhibited luminescence when the 3′-UTR was part of the construct (Fig. 6). This is consistent with our previously published in vivo findings (9). In wild-type mice, with the 3′-UTR present on the Hsp70 gene, colonic Hsp70 protein expression was significantly reduced in active dextran sulfate sodium (DSS) colitis. However, in transgenic mice with colonic Hsp70 expression from a villin promoter regulated “UTR-less” transgene, Hsp70 protein expression was not reduced in active DSS colitis. These in vivo results thus confirm that the 3′-UTR is necessary for translational downregulation of Hsp70 in intestinal inflammation. Lastly, immunoprecipitation of TIA-1 associated ribonucleoprotein complexes characteristic of stress granules demonstrated increased stress granule formation with specific incorporation of Hsp70 mRNA after IFN-γ and TNF-α treatment (Fig. 7). Thus multiple lines of evidence point to IFN-γ and TNF-α exposure, leading to stress granule formation and subsequent recruitment of Hsp70 to the stress granules in intestinal epithelial cells, resulting in sequestration of Hsp70 mRNA from translation.

During stress, it is critical that proteins to protect viability receive the highest translational priority to optimize translational efficiency. Our results differ from those seen in plants and HeLa cells under the stress condition of heat shock (6, 12). These studies demonstrated specific exclusion of Hsp70 from the stress granules so that in these cells Hsp70 was preferentially translated. Our study is the first to specifically examine the role of stress granules in intestinal epithelial cells, specifically under the stress of exposure to inflammatory cytokines. Under nonlethal stress conditions such as heat shock or exposure to intestinal flora, significantly high levels of Hsp70 mRNA override the inhibitory regulation of stress granules, so that Hsp70 mRNAs outside stress granules are available and sufficient for Hsp70 translation. With the additional stress of inflammation, increased numbers of stress granules, which form in almost every cell, recruit more Hsp70 mRNAs and decrease Hsp70 production.

Studies have shown that stress granule mRNA is translationally inactive but can be translated after the cells have recovered from stress, thus preserving the most critical functions for ongoing survival (12, 18). It is possible that Hsp70 is a more critical housekeeping protein for intestinal epithelial cells than other cell types and is recruited to stress granules for preservation once the stress has passed. In our study, some Hsp70 is translated even with cytokine pretreatment, but just at diminished levels compared with heat shock alone. The combination of heat stress and inflammatory cytokine exposure may represent a more severe injury. Alternatively, it is possible that the effects of Hsp70 are not optimal for damaged cells. Decreased Hsp70 levels may prevent inhibition of apoptosis, allowing necessary removal of damaged cells. Our model represents acute injury, and we postulate that under these conditions, rapid removal of damaged cells is warranted and stress granules may be important for self-preservation.

In our above described DSS-induced acute colitis model, using a villin promoter regulated “UTR-less” Hsp70 transgenic mouse, cytokine-mediated intestinal inflammation was seen, but progressive downregulation of colonic Hsp70 protein expression was not detected (9). These transgenic Hsp70 mice without colonic Hsp 70 downregulation had lower endoscopic and inflammation scores after DSS treatment than did the wild-type mice, which did demonstrate translational downregulation of Hsp70 with increasing severity of mucosal inflammation, suggesting that Hsp70 protects enterocytes from the inflammatory damages. Under chronic conditions, decreased translation of Hsp70 could be detrimental to overall healing and promote colitis. Our cell culture model does not recapitulate the complex interactions of immune and epithelial cells resulting in in vivo colitis models, and it is likely that there is a difference in the role of Hsp70 in acute and chronic inflammation.

Our study reveals a novel link between inflammatory cytokines and stress granule formation in intestinal epithelial cells, leading to a specific decrease in Hsp70 translation. This study enhances our understanding of regulation of Hsp processing. Further understanding of the cues of prioritization of translation will provide additional insights into the pathogenesis of inflammatory bowel diseases and means of appropriately optimizing cytoprotective effects.

GRANTS

This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants DK-47722 (E. B. Chang), DK-38510 (E. B. Chang), the Digestive Disease Research Core Center DK-42086 of the University of Chicago, and Research Training Awards (S. Hu) from the Crohn's and Colitis Foundation of America.

DISCLOSURES

None of the authors have any conflict of interest to disclose.