Index Introduction Experimental Procedures Fish Cell culture and transfection cDNA cloning Homologous recombination of fosmid vector clones and microinjection of fosmid constructs PCR and plasmid construction Northern blot hybridization Immunoblotting Immunofluorescence Metabolic labeling Results Identification of UPR mediators in medaka IRE1 pathway PERK pathway ATF6 pathway Regulator of the ER chaperone BiP Visualization of UPR activation in medaka fish Discussion Acknowledgments References

Introduction

Newly synthesized secretory and transmembrane proteins are folded and assembled in the endoplasmic reticulum (ER). Although the ER incorporates an efficient protein quality control system, unfolded proteins may still accumulate in the ER under a variety of conditions collectively termed ER stress. Eukaryotic cells cope with ER stress by activating the unfolded protein response (UPR), which consists of translational and transcriptional programs (Mori, 2000; Ron and Walter, 2007). The ER stress signal is sensed and transduced across the ER membrane toward the nucleus by a transmembrane protein(s) in the ER. The number of ubiquitously expressed sensors/transducers has increased with evolution, one (Ire1p) in Saccharomyces cerevisiae, three (ire-1, pek-1 and atf-6) in Caenorhabditis elegans and Drosophila melanogaster and five (IRE1α, IRE1β, PERK, ATF6α and ATF6β) in mammals (Mori, 2009). Although IRE1β in mice is expressed in the gut only (Bertolotti et al., 2001), we characterize it here as a ubiquitous sensor/transducer on the basis that medaka IRE1β appears to be expressed ubiquitously (see the RESULTS section). Among the various programs activated in response to ER stress, the most conserved program is transcriptional induction of ER-localized molecular chaperones and folding enzymes (hereafter collectively termed ER chaperones), as well as components of the ER-associated degradation (ERAD) machinery, both of which help maintain homeostasis of the ER. Interestingly, transcriptional induction of ER chaperones and ERAD components is mediated by Ire1p orthologues in yeast (Cox et al., 1993; Mori et al., 1993; Travers et al., 2000), worm (Shen et al., 2005) and fly (Hollien and Weissman, 2006), but by ATF6α in mammals (Adachi et al., 2008; Wu et al., 2007; Yamamoto et al., 2007). Thus, the major regulator of ER quality control proteins has switched during evolution from Ire1p orthologues to ATF6α (Mori, 2009). We have hypothesized that this switch occurred with the development of vertebra, on the basis that chordates such as Ciona intestinalis and Branchiostoma lanceolatum express three ubiquitous sensors/transducers as do worm and fly, whereas teleosts such as zebrafish (Danio rerio) and medaka (Oryzias latipes) express five ubiquitous sensors/transducers, as do mammals.

Here, to better understand the evolutional development of the UPR, we investigated whether mammalian UPR signaling pathways are conserved in medaka fish.

Experimental Procedures

Fish

Medaka southern strain cab was used as wild-type. Fishes were maintained in a re-circulating system with a 14:10 h light: dark cycle at 27.5°C. Fluorescence of transgenic fish was observed and quantified with a Leica M205 FA fluorescent stereomicroscope.

Cell culture and transfection

Medaka OLCAB-e3 cells (Hirayama et al., 2006) were cultured in L15 medium supplemented with 20% fetal calf serum, 10 mM HEPES, and antibiotics (100 U/ml penicillin and 100 μg/ml streptomycin) at 33°C. OLCAB-e3 cells, at 8×10⁶ cells in a 100-mm dish with 16 μg of plasmid for immunoblotting and northern blot hybridization and at 2×10⁵ cells in a 24-well plate with 4 μg of plasmid for immunofluorescence, were transfected by electroporation using a Microporator (Digital Bio).

cDNA cloning

cDNAs containing the entire open reading frames (ORFs) of medaka IRE1α, IRE1β, PERK, ATF6α, ATF6β, XBP1, ATF4-1, ATF4-2, CHOP and BiP were obtained by RT-PCR using total RNA prepared from OLCAB-e3 cells and the following primers, designed based on information in the Ensemble genome browser: 5'-AGCGTTGGGACAACATTTCG-3' and 5'-GTTAGCTTGCACAGCTGTCC-3' for IRE1α, 5'-TGGATTTAAAGCGGTACCGC-3' and 5'-TTTACAAAAGCAATAGCAGC-3' for IRE1β, 5'-TCCTTAAGACACCGACACG-3' and 5'-AGGGGTCACCGAGGGTCACC-3' for PERK, 5'-ATCGCTACAAAGATGTCGACAG-3' and 5'-GACGTCTGGTTTTTGCTTCG-3' for ATF6α, 5'-GCCTCGTTGTGGCTGCTTCG-3' and 5'-GTCACCATGGCAACGCCACC-3' for ATF6β, 5'-TCATGGTGGTGGTCGCAACGG-3' and 5'-TTAGAACTTTGTGGGCAGAGG-3' for XBP1, 5'-AGGACCGAACCTCTCGACTT-3' and 5'-CAGTCTCCACCAACAGCTCA-3' for ATF4-1, 5'-CCCAACATCGTCACACACTC-3' and 5'-CGTAATGCGGACAAAACGTA-3' for ATF4-2, 5'-AGATGACTGCCGAGTGGCTAC-3' and 5'-CATTTTGCTCCACACACACC-3' for CHOP and 5'-CGCATTAAGCGAAGAAGCTC-3' and 5'-TGGTGGCAAGAGGATAACAA-3' for BiP. Amplified fragments were extensively sequenced. DDBJ accession numbers are as follows: IRE1α (AB667982), IRE1β (AB667983), PERK (AB667981), ATF6α (AB667979), ATF6β (AB667980), XBP1 (AB667985), ATF4-1 (AB667986), ATF4-2 (AB667987), CHOP (AB667988), and BiP (AB667984).

Homologous recombination of fosmid vector clones and microinjection of fosmid constructs

To generate fluorescent reporter constructs, various fosmid vector clones containing UPR mediator genes were obtained from National BioResource Project (NBRP) Medaka. Fosmid IDs are as follows: GOLWFno486_i19 for IRE1α, GOLWFno383_i16 for IRE1β, GOLWFno475_h05 for PERK, GOLWFno86_n07 for ATF6α, GOLWFno92_a12 for ATF6β, and GOLWFno434_k20 for BiP.

These fosmid vector clones were recombined twice using SW102, an Escherichia coli strain that was used for the BAC homologous recombination system (Lee et al., 2001). To perform recombination, SW102 cells containing each fosmid vector clone were incubated at 42°C for 20 min, made competent by the Inoue method (Sambrook et al., 1989), and transformed by the targeting DNA fragment consisting of the 5' arm-INSERT-3' arm. First recombination was to insert the I-SceI meganuclease recognition site into the fosmid vector backbone. Thus, the region encompassing the I-SceI site and ampicillin-resistant gene (kindly provided by Dr. S. Higashijima) was amplified by PCR using the primers 5'-GAAGATCACTTCGCAGAATAAATAAATCCTGGTGTCCCTGTTGATACCGGACTCAGATCTCGAG-3' and 5'-CAATTCTCATGTTTGACAGCTTATCATCGAATTTCTGCCATTCATCCGCTGGTAGCGGTGGTTT-3' as 5' arm and 3' arm, respectively, corresponding to the fosmid vector backbone sequence. Second recombination was to insert the enhanced green fluorescent protein (EGFP) cassette into each first exon. In this way, the region encompassing the EGFP gene, polyadenylation site, and kanamycin-resistant gene was amplified using the following primers as the 5' arm and 3' arm, respectively, corresponding to each first exon sequence. 5'-TATGTAGTCAGCGGCCGTACGGATAAACCAACGGTTAGGGACCAAAGAATATGGTGAGCAAGGGCGAGGA-3' and 5'-GTCATACGTTCAATAATACGAGTGTAAGTCTTGCTAATTTATACACCTACTCGACCAGTTGGTGATTTTG-3' for P_IRE1α-EGFP, 5'-GTTCGGCGGCTTTAATTCTCCGGACCTTGAAACGTAGTGTTTCGGGGGTCATGGTGAGCAAGGGCGAGGA-3' and 5'-GTGGTGTAAACAAAAGCAGAAACAGTTGCACACTTTGACGAAGTTTACCTACCAGTTGGTGATTTTGAAC-3' for P_IRE1β-EGFP, 5'-CGGCCAGTGGCATGTGAATATCATCCTTCACTAATGTGCCAGTTTTACCAATGGTGAGCAAGGGCGAGGA-3' and 5'-GCCAACAATGTCCTCCTAGCTAAACAAGTGTCGCGCTATTTCGGTTACCTTCGACCAGTTGGTGATTTTG-3' for P_PERK-EGFP, 5'-CGGGTTTGATTCGCTTGAGCCTGATAAGTCACTAATGTATCGCTACAAAGATGGTGAGCAAGGGCGAGGA-3' and 5'-TTGGCCCCTTGTCTGAATCTTTTTCTAGAAAGAACGAGCGCCCCCCTTACTCGACCAGTTGGTGATTTTG-3' for P_ATF6α-EGFP, 5'-CGCTGACTGTCTGTCTGCAGACTTTCCAAACACGCACCGCTTCTGTGGCGATGGTGAGCAAGGGCGAGGA-3' and 5'-CACACCCAAAGCCAAGCAGCCCCCCCATCGCTACTTTCACGTCATCTTACTCGACCAGTTGGTGATTTTG-3' for P_ATF6β-EGFP, 5'-TCCTTCCTTTGCAGGTTTTTTTTCTCGTTCTGTGAGCAGAGGCGCGAAAGATGGTGAGCAAGGGCGAGGA-3' and 5'-CATATTAAATGCTTTCGACAAACTACAACCAATCAGAACGTGCTGCTTACACCAGTTGGTGATTTTGAAC-3' for P_BiP-EGFP.

Each fosmid construct was diluted to 10 ng/μl with 0.5× Yamamoto’s buffer, 0.5×I-SceI buffer and 0.05% phenol red, and injected into eggs at their one-cell stage together with 0.25 U/μl of I-SceI (New England Biolabs) using a FemtoJet (Eppendorf).

PCR and plasmid construction

To determine the presence of mRNA for IRE1α, IRE1β and PERK in OLCAB-e3 cells, PCR was performed using OLCAB-e3 cDNA or medaka genomic DNA prepared from adult caudal fin as template using the following primers: 5'-CAGTTGTTCAAGGTGACTGG-3' and 5'-GTTAGCTTGCACAGCTGTCC-3' for IRE1α, 5'-TGGATTTAAAGCGGTACCGC-3' and 5'-GTAGAGCAGAGGGCTGTTGG-3' for IRE1β, 5'-GCCTCGATCAGGAACAGAGA-3' and 5'-GTCGTTCCAGGGGTCCTCGTTTGTC-3' for PERK. To examine ER stress-induced splicing of XBP1 mRNA, RT-PCR was performed using 5'-TGCTGACAGAAAATGAGGAAC-3' and 5'-CCAACAGCAGATCAGACTC-3' as primers.

Constructions of plasmids was performed according to standard procedures (Sambrook et al., 1989). pCMV-myc (CLONTECH) was used to express ATF6α(1–356), ATF6β(1–364), pXBP1(U) and pXBP1(S), each tagged with c-myc epitope at the N-terminus. pcDNA 3.1(+) (Invitrogen) was used to express ATF6α(1–361) untagged.

Northern blot hybridization

Total RNA was extracted from OLCAB-e3 cells by the acid guanidinium/phenol/chloroform method using Isogen (Nippon Gene). Northern blot hybridization was performed according to standard procedures (Sambrook et al., 1989). Digoxigenin-labeled cDNA probes were prepared using PCR according to the manufacturer’s instructions (Roche) and hybridized with RNA electrophoresed and blotted on a membrane. Subsequent reaction with anti-digoxigenin antibody (Roche) and treatment with the chemiluminescent detection reagent CDP-star (GE Healthcare Biosciences) were performed according to the manufacturers’ specifications. Chemiluminescence was visualized using an LAS-3000mini LuminoImage analyzer (Fuji Film).

Immunoblotting

OLCAB-e3 cells were harvested with a rubber policeman and collected by centrifugation. Cell pellets were lysed in SDS sample buffer (50 mM Tris/HCl, pH 6.8, containing 100 mM dithiothreitol, 2% SDS and 10% glycerol) containing protease inhibitor cocktail (Nacalai Tesque) and 10 μM MG132, and immediately boiled for detection of CHOP, ATF4-2, ATF6α, c-myc epitope, and KDEL in BiP. Alternatively, collected cell pellets were suspended in sucrose buffer (10 mM HEPES, pH 7.9, containing 50 mM NaCl, 0.5 M sucrose, 0.1 mM EDTA, 0.5% Triton X-100, and 100 mM dithiothreitol) containing protease inhibitor cocktail and 10 μM MG132, and separated into nuclear and postnuclear fractions by centrifugation at 1,000 rpm for 5 min. The nuclear fraction was lysed in SDS sample buffer and immediately boiled for detection of XBP1. Immunoblotting analysis was performed by standard procedures (Sambrook et al., 1989) using Western Blotting Luminol Reagent (Santa Cruz). Chemiluminescence was detected using an LAS-3000mini LuminoImage analyzer (Fuji Film). Rabbit anti-human CHOP and anti-c-myc epitope polyclonal antibodies were purchased from Santa Cruz Biotechnology and Medical and Biological Laboratories (MBL), respectively. Mouse anti-KDEL monoclonal antibody was obtained from Stressgen. Anti-medaka ATF6α, XBP1 and ATF4-2 antisera were raised against the N-terminal regions of ATF6α (51–283), XBP1 (1–74) and ATF4-2 (166–361), respectively.

Immunofluorescence

OLCAB-e3 cells grown on coverslips were transiently transfected with c-myc-epitope-tagged expression plasmids. Cells were fixed with 3.7% paraformaldehyde for 10 min, and permeabilized with 0.2% Triton X-100 for 10 min. They were then incubated with mouse anti-c-myc epitope monoclonal antibody (9E10, Santa Cruz) for 1 h, and then with FITC-conjugated anti-mouse IgG antibody (Cappel) for 1 h at 37°C. Coverslips were mounted with VECTASHIELD Mounting Medium (Vector Laboratories) containing 50 μg/ml DAPI. Images were acquired using a Leica TCS SP2 confocal microscope system.

Metabolic labeling

OLCAB-e3 cells were treated with various concentrations of thapsigargin or dithiothreitol for 1 h, and pulse labeled for 10 min prior to harvest with 4.1 Mbq/dish EXPRE³⁵S³⁵S protein labeling mixture (PerkinElmer). Radiolabeled cells were lysed in SDS sample buffer and boiled. Samples were subjected to SDS-PAGE and radioactivity was analyzed using a FLA3000G FluoroImage analyzer (Fuji Film).

Results

Identification of UPR mediators in medaka

Using information provided by the Ensemble genome browser (http://www.ensembl.org/Oryzias_latipes/Info/Index), we have cloned and sequenced cDNA of medaka UPR mediators, namely IRE1α, IRE1β, PERK, ATF6α, ATF6β, XBP1, CHOP, and BiP, which correspond to those encoded by mammalian genomes in a one-to-one manner, as described in Experimental Procedures (Fig. 1). Indeed, the amino acid sequence (233–274) of human IRE1α is closer to the corresponding region of medaka IRE1α than to that of medaka IRE1β, whereas the amino acid sequence (344–376) of human IRE1β is closer to the corresponding region of medaka IRE1β than to that of medaka IRE1α (Fig. 1A). Similarly, the amino acid sequence (448–483) of human ATF6α is closer to the corresponding region of medaka ATF6α than to that of medaka ATF6β, whereas the amino acid sequence (496–529) of human ATF6β is closer to the corresponding region of medaka ATF6β than to that of medaka ATF6α (Fig. 1C). The exception is that the medaka genome contains two ATF4 genes, designated here ATF4-1 and ATF4-2, instead of the one ATF4 gene in mammals (Fig. 1E). We also found that the medaka genome contains two ATF5 genes instead of the one ATF5 gene in mammals (data not shown). Sequence comparison indicates that the two medaka ATF4 and two ATF5 genes are orthologues of the mammalian ATF4 and ATF5 genes, respectively.

View Details

Fig. 1. Schematic comparison of medaka and human UPR mediators. Domain structures are shown schematically for IRE1α and IRE1β (A), PERK (B), ATF6α and ATF6β (C), XBP1 (D), ATF4 (F), CHOP (F) and BiP (G). Abbreviations are as follows. SS, signal sequence; TMD, transmembrane domain; Kinase, protein kinase; RNase, ribonuclease; bZIP, basic leucine zipper; AD, activation domain; HSP70, heat shock protein of 70 kDa; and KDEL, ER retention signal. XBP1 is expressed as either pXBP1(U), the unspliced form, or pXBP1(S), the spliced form. The N-terminal region of human CHOP with no domain assignment is 44% identical to the N-terminus of medaka CHOP (F). Amino acid sequences in particular regions of medaka and human IRE1α and IRE1β as well as ATF6α and ATF6β are aligned (right in A and C). The numbers to the right of the respective gene name represent entire percent identity between human and medaka orthologues, and the numbers beneath various domains represent the percent identity of the particular domain between human and medaka orthologues.

To determine whether medaka UPR sensors/transducers are ubiquitously expressed, we made transgenic medaka in which the expression of EGFP was driven by a promoter of each UPR sensor/transducer. To this end, a fosmid vector clone containing an entire or a part of each UPR sensor/transducer gene in its ~30 kb insert was obtained from National BioResource Project (NBRP) Medaka (http://www.shigen.nig.ac.jp/medaka/). In the case of IRE1α, the fosmid vector clone GOLWFno486_i19 contains a 33.1 kb insert in which three genes exist, namely the 3' part of TEX2, the entire IRE1α and the 5' part of GAA (Fig. 2A): thus, this fosmid vector clone contains both the entire upstream and downstream regions of the IRE1α gene. The first exon of the IRE1α gene was replaced by a cassette consisting of the EGFP gene, polyadenylation site derived from bovine growth hormone, and kanamycin-resistant gene. The resulting recombinant fosmid vector clone was microinjected into a one-cell stage embryo to make transgenic medaka using I-SceI meganuclease, which mediates the integration of only one or a few copies of the transgene in tandem (Thermes et al., 2002). Because the EGFP gene was under the control of the IRE1α promoter, as implied by the description P_IRE1α-EGFP in Fig. 2A, the level and pattern of EGFP expression in transgenic fish should reflect those of endogenous IRE1α. Two transgenic fish lines carrying the P_IRE1α-EGFP gene were maintained without feeding to avoid autofluorescence from food after hatching, and analyzed for EGFP expression at 1 day post-hatching. As shown in Fig. 2A, EGFP driven by the IRE1α promoter was expressed throughout the entire body (the other line also showed similar pattern), suggesting that IRE1α is expressed ubiquitously in medaka, as in mammals.

View Details

Fig. 2. Ubiquitous expression of five UPR sensors/transducers in medaka. A fosmid vector clone containing the IRE1α (A), IRE1β (B), PERK (C), ATF6α (D) or ATF6β (E) gene was obtained, and the first exon of the respective gene was replaced with a cassette consisting of the EGFP gene, polyadenylation site (polyA), and kanamycin-resistant gene (KmR). Transgenic fishes carrying the PIRE1α-EGFP (A), PIRE1β-EGFP (B), PPERK-EGFP (C), PATF6α-EGFP (D) or PATF6β-EGFP (E) fusion gene were analyzed for EGFP expression at 1 day post-hatching (dph).

Similarly, EGFP driven by a promoter of IRE1β, PERK, ATF6α or ATF6β, as implied by the descriptions P_IRE1β-EGFP (Fig. 2B), P_PERK-EGFP (Fig. 2C), P_ATF6α-EGFP (Fig. 2D) or P_ATF6β-EGFP (Fig. 2E), was expressed throughout the entire body; although the results with one transgenic line each are shown, similar results were obtained with three other lines for IRE1β and two other lines for ATF6α and ATF6β. Because fluorescence levels in these transgenic fishes were much lower than that observed with P_IRE1α-EGFP transgenic fish, abdominal regions were omitted as they contained yolk, which exhibited high endogenous fluorescence. These results are consistent with those in mammals, except with regard to IRE1β; mammalian IRE1β is known to be expressed only in the gut (Bertolotti et al., 2001). Because the fosmid vector clone GOLWFno383_i16 used to create transgenic medaka contains both the entire upstream and downstream regions of the IRE1β gene (Fig. 2B), it is very likely that the fosmid vector clone contains all information necessary to determine the expression pattern of IRE1β in medaka. To confirm this, we examined whether IRE1β mRNA was expressed in medaka cell culture line OLCAB-e3, which was established from medaka embryo. As shown in Fig. 3A, RT-PCR analysis showed that cDNA fragments corresponding to not only IRE1α mRNA but also IRE1β mRNA were indeed amplified from RNA which had been prepared from OLCAB-e3 cells. These fragments were not considered to represent contamination of genomic DNA because the use of genomic DNA prepared from the adult caudal fin as template resulted in the amplification of a DNA fragment of different size for IRE1α or no amplification of a DNA fragment of the same size for IRE1β. These results suggested that IRE1β mRNA appears to be expressed ubiquitously in medaka, in contrast to the case in mammals.

View Details

Fig. 3. Conservation of the IRE1 pathway. (A) Total RNA was extracted from OLCAB-e3 cells and subjected to RT-PCR analysis using a pair of primers specific to IREα or IRE1β, which amplified exons 20–22 or 1–6, respectively, as indicated on the right. Medaka genomic DNA prepared from adult caudal fin was also used as template to show that the fragment obtained was not derived from amplification of genomic DNA. (B) ORF1 and ORF2 present in medaka XBP1 mRNA are shown schematically. Numbers indicate nucleotide positions, with the transcriptional start site set at 1. The red box indicates the intron removed by ER stress-induced unconventional splicing. The arrows indicate the positions of primers used to detect splicing of XBP1 mRNA in response to ER stress. (C) Stem-loop structures around the splice sites in medaka and mouse XBP1 mRNA are shown. The arrows indicate the cleavage sites in the seven-nucleotide loops, and six nucleotides indispensable for the cleavage reaction are boxed. (D) OLCAB-e3 cells were treated with 2 μg/ml tunicamycin (Tm) or 300 nM thapsigargin (Tg) for the indicated periods. Total RNA was extracted and subjected to RT-PCR analysis using the primers shown in (B). pCMV-myc-medaka pXBP1(S) and pCMV-myc-medaka pXBP1(U) were used as template for control. The asterisk indicates a non-specific band. (E) OLCAB-e3 cells were untreated or treated with 2 μg/ml tunicamycin (Tm) or 300 nM thapsigargin (Tg) for 8 h. MG132 (10 μM) was added to the culture for 2 h prior to harvest. Nuclear extracts were prepared and subjected to immunoblotting using anti-medaka XBP1 antibody. (F) OLCAB-e3 cells transiently transfected with pCMV-myc-medaka pXBP1(S) or pCMV-myc-medaka pXBP1(U) were fixed and stained with anti-c-myc antibody (left) or DAPI (middle). Human and medaka nuclear exclusion signal (NES) present in pXBP1(U) are compared. Identical amino acids are indicated by asterisks.

We next examined whether three (IRE1, PERK and ATF6) signaling pathways in the UPR are activated in response to ER stress in medaka, as in mammals.

IRE1 pathway

In metazoan, ER stress-induced activation of IRE1 initiates unconventional (spliceosome-independent) splicing of XBP1 mRNA to produce active transcription factor XBP1 (Calfon et al., 2002; Shen et al., 2001; Yoshida et al., 2001). Medaka XBP1 mRNA contains two large ORFs which partially overlap each other, as in mouse and human XBP1 mRNA (Fig. 3B). The overlapping regions of both medaka and mouse contain two characteristic stem-loop structures, and the three nucleotides in the seven-nucleotide loop which are known to be essential for ER stress-induced cleavage by IRE1 are completely conserved (Fig. 3C). RT-PCR analysis of mRNA prepared from unstressed OLCAB-e3 cells with the primers surrounding the two stem loop structures produced a band of 249 bp, as expected (Fig. 3D, lanes 3 and 6). When OLCAB-e3 cells were treated with tunicamycin, an inhibitor of protein N-glycosylation, or thapsigargin, an inhibitor of Ca²⁺ pump in the ER, a smaller fragment was produced (Fig. 3D, lanes 4, 5, 7, 8). Sequencing of the smaller fragment revealed that the two stem loops were cleaved at the expected positions, resulting in the release of a 26-nucleotide-intron (Fig. 3C). Immunoblotting analysis using an anti-medaka XBP1 antibody we raised revealed induction of the spliced form of XBP1 protein, designated pXBP1(S), after treatment with tunicamycin or thapsigargin (Fig. 3E, lanes 2 and 4). These results indicate that the medaka IRE1-XBP1 pathway is activated in response to ER stress, as in mammals.

Unspliced XBP1 mRNA is translated to produce the unspliced form of XBP1, designated pXBP1(U). pXBP1(U) and pXBP1(S) share the basic-leucine zipper domain but their C-terminal regions differ as a result of ER stress-induced splicing, which causes a frame shift at the splice site. pXBP1(S) enters the nucleus to activate transcription of its target genes (Fig. 3F, panel a). In contrast, pXBP1(U) shuttles between the nucleus and cytoplasm (Fig. 3F, panel d), due to the presence of a nuclear exclusion signal localized in the pXBP1(U)-specific C-terminal region (Fig. 3F, bottom). pXBP1(U) targets pXBP1(S) and pATF6(N), the active form of ATF6 (see below), to proteasome-mediated degradation. Medaka cells are therefore able to shut transcription down quickly when ER stress is resolved, as in mammals (Yoshida et al., 2006).

PERK pathway

In mammals, PERK activated in response to ER stress phosphorylates the α subunit of eukaryotic translational initiation factor 2 (eIF2α), resulting in general translational attenuation. This attenuation paradoxically induces translation of the transcription factor ATF4, which induces transcription of its target genes, including the gene encoding the pro-apoptotic transcription factor CHOP (Ron, 2002).

RT-PCR analysis revealed that PERK mRNA was expressed in OLCAB-e3 cells (Fig. 4A). Because the phosphorylated form of medaka eIF2α was not detected by commercially available anti-human or mouse phosphorylated eIF2α antibody, we used pulse labeling to measure whether protein synthesis is attenuated in response to ER stress. As shown in Fig. 4B, protein synthesis rate was gradually decreased after treatment of OLCAB-e3 cells with thapsigargin or tunicamycin (data not shown) in a dose-dependent manner. Stronger attenuation was observed when OLCAB-e3 cells were treated with dithiothreitol, which strongly malfolds proteins in the ER by reducing disulfide bridges.

View Details

Fig. 4. Conservation of the PERK pathway. (A) RT-PCR analysis was carried out as in Fig. 3A using a pair of primers specific to PERK, which amplified exons 3 and 4, as indicated on the right. (B) OLCAB-e3 cells were treated with thapsigargin (Tg) or dithiothreitol (DTT) at the indicated concentrations for 1 h and pulse labeled with 35S-methionine and cysteine for 10 min prior to harvest. Cell lysates were prepared, subjected to SDS-PAGE, and autoradiographed (left panel). The radioactivity of all bands was determined and normalized with the value in untreated cells. The means from three independent experiments with standard deviations (error bars) are plotted against the drug concentration (right panel). (C) Structures of mouse ATF4 mRNA and medaka ATF4-1 and ATF4-2 mRNAs are schematically shown. Arrows denote small ORFs present upstream of the ATF4 ORF indicated by the white box. Numbers indicate nucleotide positions, with the translation start site set at 1. (D) OLCAB-e3 cells were untreated or treated with 2 μg/ml tunicamycin (Tm) for 8 h. MG132 (10 μM) was added to the culture for 2 h prior to harvest. Cell lysates were prepared and subjected to immunoblotting using anti-medaka ATF4-2 or anti-human CHOP antibody.

The paradoxical translational induction of ATF4 is mediated by small ORFs localized upstream of the mammalian ATF4 gene (uORF1 and uORF2, Fig. 4C). Essentially identical structures are found on both the medaka ATF4-1 and ATF4-2 genes (Fig. 4C). Accordingly, tunicamycin-induced translational attenuation resulted in the induction of ATF4-2 (detected with an anti-medaka ATF4-2 antibody we raised, which does not cross reacts with ATF4-1) and its target CHOP (detected by anti-human CHOP antibody), as shown in Fig. 4D. These results demonstrate that the medaka PERK-ATF4 pathway is activated in response to ER stress, as in mammals.

ATF6 pathway

In mammals, both ATF6α and ATF6β are constitutively synthesized as type II transmembrane proteins in the ER, designated pATF6α(P) or pATF6β(P). Upon ER stress, pATF6α(P) and pATF6β(P) relocate from the ER to the Golgi apparatus to be cleaved by Site-1 and Site-2 proteases. The N-terminal fragments liberated from the membrane, designated pATF6α(N) and pATF6β(N), enter the nucleus and activate transcription of their target genes (Mori, 2009).

Medaka endogenous ATF6α was detected as a protein of ~90 kDa in unstressed OLCAB-e3 cells by immunoblotting using an anti-medaka ATF6α antibody we raised (Fig. 5A, lanes 1 and 4). Treatment with tunicamycin or thapsigargin resulted in detection of a protein of ~55 kDa (Fig. 5A, lanes 3, 5, and 6). This migrated to the same position as ATF6α(1–361) (lane 7), which represented the N-terminal fragment of ATF6α produced by cleavage at the Site-2 site. It should be noted that the unglycosylated form of precursor ATF6α, designated pATF6α(P*), was detected in tunicamycin- but not thapsigargin-treated OLCAB-e3 cells (Fig. 5A, lane 3). c-myc-tagged ATF6α(1–356) and c-myc-tagged ATF6β(1–364), representing pATF6α(N) and pATF6β(N), respectively, were localized in the nucleus when overexpressed by transfection in OLCAB-e3 cells (Fig. 5B). These results indicate that medaka ATF6 is constitutively synthesized as a glycosylated transmembrane protein in the ER and activated by ER stress-induced proteolysis, which allows the cleaved fragment to enter the nucleus, as in mammals.

View Details

Fig. 5. Conservation of the ATF6 pathway. (A) OLCAB-e3 cells were treated with 2 μg/ml tunicamycin (Tm) or 300 nM thapsigargin (Tg) for the indicated periods. OLCAB-e3 cells were also transiently transfected with pcDNA-medaka ATF6α(1–361). MG132 (10 μM) was added to the culture for 2 h prior to harvest. Cell lysates were prepared and subjected to immunoblotting using anti-medaka ATF6α antibody. Migration positions of pATF6α(P), unglycosylated pATF6α(P*), and pATF6α(N) as well as molecular weight markers are shown. Both pATF6α(P) and pATF6α(P*) were detected as doublets and the bands migrating faster likely represent partial degradation products of pATF6α(P) and pATF6α(P*), because they were not detected in medaka embryonic fibroblasts prepared from ATF6α-knockout medaka (data not shown). Schematic structures of pATF6α(P), pATF6β(P), pATF6α(N), pATF6β(N) and pATF6α(1–361) are shown on the right. Amino acid sequences of medaka ATF6α and ATF6β around the Site-1 and Site-2 cleavage sites (shown in red) are aligned at bottom. Transmembrane regions are boxed. (B) OLCAB-e3 cells transiently transfected with pCMV-myc-medaka pATF6α(1–356) or pCMV-myc-medaka pATF6β(1–364) were fixed and stained with anti-c-myc antibody (left panel) or DAPI (middle panel).

Regulator of the ER chaperone BiP

We found that the major ER chaperone BiP was induced in response to treatment of OLCAB-e3 cells with tunicamycin at the level of both mRNA and protein, as in mammals (Fig. 6A and 6B). We then explored which is involved in the transcriptional induction of BiP, ATF6α, ATF6β, or XBP1, by overexpressing their active forms in OLCAB-e3 cells. Overexpression of the c-myc-tagged version of ATF6α(1–356) or ATF6β(1–364), equivalent to pATF6α(N) or pATF6β(N), respectively, resulted in detection of the respective nuclear protein by immunoblotting, the levels of which were increased by treatment of OLCAB-e3 cells with the proteasome inhibitor MG132 for 4 h prior to harvest (Fig. 6C, lanes 2, 3, 7 and 8). Overexpression of the c-myc-tagged version of pXBP1(S) or pXBP1(U) resulted in detection of the respective protein by immunoblotting, the levels of which were markedly increased by treatment with MG132 (Fig. 6C, lanes 4, 5, 9, and 10). Importantly, overexpression of ATF6α(1–356) markedly increased BiP at both the mRNA and protein levels (Fig. 6C, lanes 2 and 7), whereas that of ATF6β(1–364) increased BiP at the mRNA and protein levels only slightly (Fig. 6C, lanes 3 and 8), despite the fact that the expression level of ATF6β(1–364) was much higher than that of ATF6α(1–356) (Fig. 6C, compare lane 2 with lane 3 and lane 7 with lane 8). In contrast, overexpression of pXBP1(S) or pXBP1(U) showed little effect on BiP at either the mRNA or protein level (Fig. 6C, lanes 4, 5, 9 and 10). These results strongly suggest that the transcriptional induction of BiP in medaka is mainly mediated by ATF6α, as in mammals.

View Details

Fig. 6. Effect of overexpression of active forms of ATF6α, ATF6β and XBP1 on levels of BiP mRNA and BiP protein. (A) OLCAB-e3 cells were untreated or treated with 2 μg/ml tunicamycin (Tm) for 8 h. Total RNA was prepared and analyzed by northern blot hybridization using a DIG-labeled probe specific to BiP or GAPDH. (B) OLCAB-e3 cells were treated with 2 μg/ml tunicamycin (Tm) for the indicated periods. Cell lysates were prepared and analyzed by immunoblotting using anti-KDEL antibody. (C) OLCAB-e3 cells were transiently transfected with pCMV-myc (mock), pCMV-myc-medaka pATF6α(1–356), pCMV-myc-medaka pATF6β(1–364), pCMV-myc-medaka pXBP1(S), or pCMV-myc-medaka pXBP1(U). MG132 (10 μM) was added to aliquots of the culture for 4 h prior to harvest (lanes 6–10). OLCAB-e3 cells were also untreated or treated with 2 μg/ml tunicamycin (Tm) for 8 h (lanes 11 and 12). Seventy-two hours after transfection, total RNA was prepared and analyzed by northern blot hybridization using a DIG-labeled probe specific to BiP, and cell lysates were prepared and analyzed by immunoblotting using anti-c-myc or anti-KDEL antibody.

Visualization of UPR activation in medaka fish

We finally examined whether the activation status of UPR can be visualized by making transgenic medaka expressing EGFP. For this purpose, the first exon of the BiP gene was replaced with the EGFP cassette used in Fig. 2 in a fosmid vector (Fig. 7A), which was injected into a one-cell stage embryo. Examination of transgenic medaka carrying the P_BiP-EGFP gene showed that EGFP was ubiquitously expressed, with particularly high expression in the brain, liver and gut (Fig. 7A), suggesting that ER stress was generated physiologically in such tissues during development and that ATF6α-mediated induction of BiP occurred accordingly. Importantly, tunicamycin treatment increased fluorescence intensity 2.0-, 2.8- and 4.2-fold in muscle, intestine and liver, respectively (Fig. 7B). Tunicamycin might not have reached brain regions. These results indicate that this EGFP reporter is very useful in monitoring UPR activity in living fish.

View Details

Fig. 7. Visualization of BiP level in living fish. (A) A fosmid vector clone containing the entire BiP gene was obtained, and the first exon of the BiP gene was replaced with the EGFP cassette used in Fig. 2. Transgenic fish carrying the PBiP-EGFP gene was analyzed for EGFP expression at 1 dph. (B) A transgenic fish carrying the PBiP-EGFP gene was analyzed for EGFP expression at 1 dph, and then treated with 2 μg/ml tunicamycin (Tm). Twenty-four hours later, the same fish was analyzed for EGFP expression. Fluorescence intensity in muscle, intestine or liver was measured before (n=5) or after (n=8) tunicamycin treatment of transgenic (EGFP+) fishes. The means with standard deviations (error bars) are shown. *p<0.001. Fluorescence intensity in nontrasgenic fishes (EGFP-) was negligible compared with that in transgenic fishes carrying the PBiP-EGFP gene.

Discussion

Here, we demonstrate that the UPR signaling pathways are very well conserved between medaka and mammals, with all three pathways, namely the IRE1-XBP1, PERK-ATF4 and ATF6 pathways, activated in medaka, as in mammals. In particular, overexpression experiments suggested that the ATF6α pathway is a major regulator of the ER chaperone BiP in medaka, as in mammals. This differs from findings in non-vertebrate model organisms, such as yeast, worm and fly, in which the IRE1 pathway plays a major role in the transcriptional induction of BiP. We therefore propose the term vertebrate UPR to discriminate these systems.

The transparent nature of medaka fish up to the juvenile stage allows us to monitor the activation status of the UPR in living fish by fluorescent microscopy without dissection of the abdomen. Using this approach, we will be able to gain insight into more naive activation of the UPR, which may be either tissue- or pathway-specific. Indeed, we also obtained data suggesting that the ATF6 pathway is activated during physiological development of the brain, liver and gut. Microarray analysis combined with subsequent morpholino knockdown experiments will illustrate what kinds of protein cause physiological ER stress during development of such tissues. We are now constructing transgenic medaka which allows monitoring of the activation status of the IRE1 pathway using ER stress-induced splicing of XBP1 mRNA, as utilized in mammals (Iwawaki et al., 2004) as well as transgenic medaka which allows monitoring of the activation status of the PERK pathway using the induction of ATF4 or CHOP by ER stress.

The genome project for medaka fish has now been completed (Kasahara et al., 2007) and a reverse genetic approach has been established (Taniguchi et al., 2006), allowing us to identify knockout medaka deficient in various UPR mediators. Morpholino-mediated gene knockdown and overexpression mediated by mRNA microinjection have also been established. Medaka is hardy and prolific, and rearing costs are reasonable. These characteristics combined with our present results, identify medaka as a highly useful vertebrate model for comprehensive analysis of the biology and physiology of the UPR.

Acknowledgments

We thank Ms. Kaoru Miyagawa and Ms. Yayoi Yamamoto for their technical and secretarial assistance. We are grateful to Dr. S. Higashijima at the National Institutes of Natural Sciences for providing materials used in recombination of the fosmid vector clone, and Dr. Kiyoshi Naruse at the National Institute for Basic Biology providing materials of NBRP medaka. This work was financially supported in part by grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan (20247026 to K. M.). Tokiro Ishikawa is a recipient of Research Fellowships from the Japan Society for the Promotion of Science for Young Scientists.