Translational Pausing Ensures Membrane Targeting and Cytoplasmic Splicing of XBP1u mRNA

See allHide authors and affiliations

Science  04 Feb 2011:
Vol. 331, Issue 6017, pp. 586-589
DOI: 10.1126/science.1197142


Upon endoplasmic reticulum (ER) stress, an endoribonuclease, inositol-requiring enzyme-1α, splices the precursor unspliced form of X-box–binding protein 1 messenger RNA (XBP1u mRNA) on the ER membrane to yield an active transcription factor (XBP1s), leading to the alleviation of the stress. The nascent peptide encoded by XBP1u mRNA drags the mRNA–ribosome–nascent chain (R-RNC) complex to the membrane for efficient cytoplasmic splicing. We found that translation of the XBP1u mRNA was briefly paused to stabilize the R-RNC complex. Mutational analysis of XBP1u revealed an evolutionarily conserved peptide module at the carboxyl terminus that was responsible for the translational pausing and was required for the efficient targeting and splicing of the XBP1u mRNA. Thus, translational pausing may be used for unexpectedly diverse cellular processes in mammalian cells.

Most proteins are believed to function after release from the mRNA–ribosome–nascent chain (R-RNC) complex, but there are some exceptions. Before its release from the R-RNC complex, the nascent protein X-box–binding protein 1 (XBP1u) recruits the XBP1u mRNA to the membrane (1). This event increases the efficiency of cytoplasmic splicing of the XBP1u mRNA by a transmembrane protein IRE1α upon endoplasmic reticulum (ER) stress (Fig. 1A). The IRE1α-mediated splicing reaction removes a 26-base nucleotide chain (intron) from the unspliced mRNA (XBP1u) (Fig. 1B) (2, 3). Because the splicing reaction causes a translational frameshift, the spliced mRNA (XBP1s) is translated into the XBP1s protein carrying the C-terminal half that is different from that of the XBP1u protein (Fig. 1B). Thus, the XBP1s protein, but not the XBP1u protein, functions as an active transcription factor to alleviate ER stress. Conversely, a hydrophobic region 2 (HR2), which is a membrane-targeting signal, is present only on the XBP1u protein, allowing the XBP1u protein and its mRNA to be recruited to the membrane (Fig. 1, A and B) (1).

Fig. 1

Translational pausing of the XBP1u mRNA. (A) Our model for nascent chain–mediated membrane-targeting of the XBP1u mRNA. N, N terminus. (B) Scheme of XBP1 protein variants. White box outlined in black, the first-segment; red box and white box outlined in blue, the second segments of XBP1u and XBP1s, respectively; hatched region, HR2; dotted region, CTR; yellow box, an artificially fused sequence (124–amino acid residues); black box, the sequence encoded by an intron. (C) XBP1 variant proteins, produced for the indicated periods in vitro, were separated by NuPAGE gels and analyzed by immunoblotting to detect the indicated epitopes. (D) In vitro translation mixtures (15-min reaction) were further incubated with puromycin (1 mg/ml) (Puro) or cycloheximide (1 mg/ml) (CHX) at 30°C for 5 min, then analyzed as described in (C). u, F-XBP1u-H; FS, F-XBP1u-FS-H; s, F-XBP1s-H. Black arrowheads, full-length products; white arrowheads, paused intermediates; asterisks, nonspecific bands.

Presumably, HR2 needs to become exposed from the ribosome tunnel to recruit the XBP1u R-RNC complex to the membrane (Fig. 1A). However, the C-terminal region (CTR) of the XBP1u protein (Fig. 1B) following HR2 is only 53 amino acids long. This means that the XBP1u protein may be fully synthesized and released from the R-RNC complex immediately after exposure of HR2 from the ribosome tunnel, which has a length equivalent to 40 residues (4). Therefore, translational elongation of the XBP1u protein may need to be arrested or paused after exposure of HR2 to allow sufficient time for this protein to recruit the R-RNC complex.

In the case of the bacterial SecM protein, elongation arrest has been demonstrated by detection of the corresponding peptidyl-tRNA species (5). We detected that of the XBP1 proteins by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) under the neutral pH condition (NuPAGE) to preserve peptidyl-tRNA ester bonds. The XBP1s and XBP1u cDNAs were each engineered to carry an N-terminal FLAG (DYKDDDDK) and a C-terminal hemagglutinin (HA) epitope (F-XBP1s-H and F-XBP1u-H) and were translated in vitro with the use of a rabbit reticulocyte lysate (Fig. 1C). In vitro translation of F-XBP1s-H yielded a band that was specifically recognized by an antibody to the N-terminal FLAG tag (anti-FLAG). Because this band was also detected by antibody to the C-terminal HA tag (anti-HA), it represents the full-length F-XBP1s-H, indicating no apparent translational arrest with F-XBP1s-H. However, in the case of F-XBP1u-H, a diffused low-mobility band (often seen as a doublet band) appeared only on the anti-FLAG–stained blot before the appearance of the full-length products. Because the intensity of the diffused low-mobility band decreased in a time-dependent manner, the molecules in this band probably constitute intermediates before yielding the full-length F-XBP1u-H. These intermediates probably represent nascent F-XBP1u-H polypeptides with covalently attached tRNA, because puromycin and pancreatic ribonuclease (RNase A), but not cycloheximide, broke them to yield free F-XBP1u-H peptide fragments (white arrowheads in Fig. 1D and fig. S1), which migrated almost as equally as or slightly faster than the full-length F-XBP1u-H. The band shift from 48 to ~35 kD accounted for the loss of the tRNA moiety. Furthermore, our results from XBP1u-FS, which is an artificial fusion of the XBP1u protein with an unrelated 124–amino acid polypeptide (Fig. 1B), is consistent with our idea that the XBP1u ORF intrinsically carries a signal that pauses translation at a specific region of the XBP1u protein (Fig. 1, C and D, and supporting online material text). These results strongly suggest that the translation of the XBP1u mRNA is paused just before translational termination.

We then addressed the primary structural requirement for the translational pausing. First, we investigated the importance of two highly hydrophobic regions on the XBP1u: HR1 and HR2 (fig. S2A). However, deletion or hydrophobic-to-charged–amino acid substitutions of HR2, as well as deletion of HR1, caused no reduction of the translational pausing (fig. S2, A to C) (6).

Next, we focused on the C-terminal moiety of the XBP1u protein, because it is completely different from that of the XBP1s protein (Fig. 1B). We engineered an in-frame fusion of Renilla reniformis luciferase (Rluc) to the C-terminal moiety of the XBP1u protein to generate Rluc-XC88-H (Fig. 2A). In vitro translation of Rluc-XC88-H yielded a considerable amount of intermediates caused by pausing (Fig. 2B). Serial truncation analysis demonstrated that the C-terminal 26–amino acid region, but not HR2, contributes to the translational pausing (Fig. 2, A and B). This region is evolutionarily conserved (Fig. 2C), and alanine-scanning mutagenesis revealed that 14 amino acid residues, which are highly conserved during evolution, contribute to the translational pausing (Fig. 2C and fig. S3). A more detailed time-course experiment of in vitro translation indicated that Leu246→Ala246 (L246A) or W256A (7) mutation caused almost complete loss of the tRNA-attached intermediate (Fig. 2D). Moreover, the full-length products accumulated faster when either of these mutations was introduced. Conversely, the S255A mutation extended the pausing duration.

Fig. 2

CTR is responsible for the translational pausing. (A) Rluc-XBP1u fusion proteins. The numbers of N-terminal residues of the fused XBP1u are indicated at the top. Black box, intron; hatched boxes, HR2; green circles, HA tags; a. a., amino acid. (B) The products from the indicated constructs, synthesized in vitro for 15 min in the presence of 35[S]methionine and 35[S]cysteine, were separated as in Fig. 1C and visualized using a Fuji BAS2500 phosphor imager. (C) Comparison of the sequences of CTRs in vertebrates. Orange-shaded regions, conserved amino acid residues; black triangle, amino acid residues in which alanine-scanning mutations abolish translational pausing; S255, white triangle. The S255-to-A255 mutation augmented the translational pausing. (D) Indicated variants of the F-XBP1u-H protein were analyzed as described in (B), except that samples were treated with RNase A before electrophoresis (lower panels). (E) XBP1u was engineered to carry the N-terminal three-tandem HA epitope and the nonsplicing mutation (HA3-XBP1u [nonsplicing]) (6). Twenty-four hours after transfection of HEK293T cells with expression plasmids encoding indicated variants, protein lysates were prepared, subjected to RNase A treatment, and analyzed by immunoblotting (6). See the legend to Fig. 1 for a description of the black and white arrowheads.

To examine whether the pausing intermediates exist in vivo, we transiently expressed an XBP1u construct in human embryonic kidney (HEK) 293T (HEK293T) cells. The XBP1u protein partly existed as the pausing intermediate, which was digested by RNase A to yield a peptide fragment that migrated almost as equally as the full-length XBP1u protein (Fig. 2E). In addition, as observed with the in vitro translation experiments, the L246A and W256A mutations abrogated the translational pausing, whereas the S255A mutation strengthened it. Thus, the translational pausing of XBP1u also occurred in mammalian cells.

To characterize the translational-pausing R-RNC complex, we next employed a mass-tagging assay using polyethylene glycol–maleimide (Mal-PEG) (6). This procedure involved adding Mal-PEG to the in vitro–translated samples of F-XBP1u-H, which includes three cysteine residues (C204, C215, and C247) (fig. S4A). Mal-PEG reacts to form a covalent bond with sulfhydryl groups. Further, Mal-PEG is unable to access a cysteine residue buried inside of the ribosome tunnel (8), suggesting that, in the tRNA-attached intermediate, C204, but not C247, was exposed outside of the ribosome tunnel, and C215 may be close to the exit site of the tunnel (fig. S4B). These observations are consistent with our model that HR2 is exposed outside of the ribosome tunnel to drag the translational-pausing R-RNC complex to the membrane. Conversely, it is likely that the amino acids required for the translational pausing (between Y241 and N261) (Fig. 2C) are buried inside the ribosome tunnel.

Translational pausing appeared to be required for the efficient membrane-recruitment of the XBP1u protein itself, not just for that of XBP1u mRNA. We performed in vitro translation of the XBP1u mRNA in the presence of microsomal membrane derived from canine pancreas (CMM) (Fig. 3B). The membrane fraction was then recovered by centrifugation and separated by conventional Laemmli SDS-PAGE, which yielded full-length protein bands and peptide-moiety bands of the paused intermediates (Fig. 3B). Addition of CMM into the in vitro translation mixture shifted the XBP1u peptide products from soluble fractions to pellet fractions, indicating membrane recruitment of these peptides. However, the pausing-defective mutants L246A and W256A accumulated considerable amounts of the XBP1u peptides in the soluble fractions, even in the presence of CMM, suggesting that the translational pausing is required for the efficient membrane targeting of the XBP1u protein itself.

Fig. 3

Translational pausing contributes to the membrane binding of the nascent XBP1u peptide. (A) After in vitro translation with the indicated templates, the products were treated with or without RNase A, as described in Fig. 2D. “No-term” mutants carry deletions of HA and the termination codon. (B) Membrane-binding assay of the XBP1u protein. The indicated versions of F-XBP1u-H were produced as in (A) in the presence or absence of CMM, and separated by ultracentrifugation to analyze pellet (P) and soluble (S) fractions by Laemmli SDS-PAGE and autoradiography. (C) In vitro translation with the indicated templates and RNase A treatment were performed as described in Fig. 2D. dHR2 indicates a complete deletion of HR2 from the F-XBP1u-no-term variants. (D) Membrane-binding assay of the indicated variants. See the legend to Fig. 1 for a description of the black and white arrowheads.

To further examine the role of translational pausing in the membrane binding of the XBP1u protein, we deleted the termination codon from the XBP1u cDNA and performed the in vitro translation assay. As expected (9), this deletion caused translational arrest of even the pausing-defective mutants L246A and W256A (Fig. 3A). These forced elongation-arrested versions of L246A and W256A were almost completely recruited to the membrane (Fig. 3B). Thus, elongation pausing or arrest, per se, is important for membrane binding of the XBP1u protein. As for the wild-type (WT) XBP1u protein (1), membrane binding of forced elongation-arrested mutants required HR2, because a deletion of HR2 completely abolished membrane binding of the XBP1 peptides (Fig. 3, C and D).

We next investigated the importance of translational pausing in vivo. The WT and pausing-mutant XBP1u cDNAs were transiently expressed in HEK293T cells. The cytosolic and membrane fractions were then separated by plasma-membrane permeabilization, with the use of digitonin, to determine the subcellular localization of the product mRNAs. Membrane-localization (ML) values (1)—which express how an mRNA is enriched in the membrane fraction compared with control cytosolic mRNA of glyceraldehyde-3-phosphate dehydrogenase—were determined (Fig. 4, A and B). The ML value of the hyper-pausing mutant S255A was almost the same as that of the wild type, whereas those of the pausing-defective mutants L246A and W256A were strongly diminished. Thus, the translational pausing of the XBP1u mRNA contributes to its membrane localization.

Fig. 4

Translational pausing is required for efficient membrane targeting and cytoplasmic splicing of the XBP1u mRNA. (A) Twenty-four hours after transfection of HEK293T cells with expression plasmids encoding the indicated variants, cellular localization of mRNA products was examined (6). SS- denotes the calreticulin’s N-terminal signal peptide. GAPDH, glyceraldehyde-3-phosphate dehydrogenase. (B) The ML value for each mRNA shown in (A) presented as mean ± SD of triplicate measurements (6). (C) Transfected cells as in (A) were treated with or without 0.5 μg/ml of thapsigargin (Tg) for 4 hours. Total mRNAs were then analyzed by reverse transcription polymerase chain reaction to monitor cytoplasmic splicing of the XBP1u variant mRNAs. (Right panel) Splicing efficiency expressed as mean ± SD (error bars) of triplicate experiments. u, unspliced; s, spliced. (D) HEK293T cells expressing WT or W256A version of XBP1u mRNA were treated with 0.2 μg/ml of Tg for indicated periods. (Right panel) Splicing efficiency.

Membrane targeting of the XBP1u mRNA is important for its efficient splicing by IRE1α (1). We examined the role of the pausing for the splicing of the XBP1u mRNA by using the pausing mutants. The L246A and W256A mutations decreased splicing efficiency of XBP1u mRNA (Fig. 4, C and D), indicating that the pausing is indeed important for the efficient splicing of the mRNA. Nevertheless, the pausing-defective XBP1u mRNAs exhibited considerable splicing efficiency when they were artificially distributed to the membrane by in-frame insertion of a nucleotide sequence encoding an N-terminal signal peptide of an ER-luminal protein calreticulin (Fig. 4, A and B, and fig. S5). Thus, the membrane localization of the XBP1u mRNA contributes to its efficient splicing by IRE1α.

A CTR-dependent translational pausing temporarily froze the XBP1u R-RNC complex in which HR2 protrudes from the ribosome tunnel. Because HR2 has membrane-binding ability, the resulting XBP1u R-RNC was recruited to the membrane, allowing the XBP1u mRNA to be efficiently spliced by IRE1α. This is an interesting case of translational pausing mechanism of an mRNA that is involved in the transport of the mRNA in eukaryotes.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S6

Table S1


References and Notes

  1. Materials and methods and other supporting material are available on Science Online.
  2. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.
  3. We thank J. Iida-Hashimoto and N. Fujimoto for technical assistance and K. Ito, S. Chiba, and H. Yoshida for discussions. This work was supported by Grants-in-Aid for Scientific Research on Priority Areas (K.K., Y.K.) and Scientific Research B (K.K.) of KAKENHI from the Ministry of Education, Culture, Sports, Science and Technology of Japan. K.Y. is a special postdoctoral fellow supported by Global Center of Excellence program from the Japan Society for the Promotion of Science.
View Abstract

Navigate This Article