Dispatch R121

Splicing: HACking into the unfolded-protein responseCaroline E. Shamu

Unfolded proteins in the endoplasmic reticulum ofSaccharomyces cerevisiae trigger a specialized RNAsplicing event that allows the subsequent translation ofthe Hac1p transcription factor. This splicing can bereconstituted with Ire1p, a transmembrane kinase thathas a site-specific RNase activity, and tRNA ligase.

Address: Department of Cell Biology, Harvard Medical School, 240Longwood Avenue, Boston, Massachusetts 02115, USA.E-mail: shamu@bcmp.med.harvard.edu

Current Biology 1998, 8:R121–R123http://biomednet.com/elecref/09609822008R0121

© Current Biology Ltd ISSN 0960-9822

The unfolded-protein response (UPR) is induced whenmisfolded or improperly assembled proteins accumulate inthe endoplasmic reticulum (ER). It results in the increasedexpression of ER-resident enzymes, such as the molecularchaperone BiP and protein disulfide isomerase (PDI), thataid the correct folding and assembly of proteins. The UPRinduces production of these protein-folding enzymes by up-regulating the transcription of the genes that encode them.

Although the UPR seems to be conserved among all eukary-otes, it has been best studied in the budding yeast Saccha-romyces cerevisiae, where the basic outline of the UPRpathway between the ER and the nucleus is now known [1].The accumulation of misfolded proteins in the ER somehowactivates Ire1p, a transmembrane serine/threonine kinasethat lies in the ER membrane with its amino terminus in theER lumen and its carboxy-terminal kinase domain either inthe nucleus or in the cytoplasm [2,3]. When activated, Ire1poligomerizes and autophosphorylates [4]. This triggers pro-duction of Hac1p, the transcription factor that binds to theunfolded protein response element (UPRE).

The 22 bp UPRE is present in the promoters of genes reg-ulated by the UPR [5–8] and is necessary and sufficient toput a linked gene under the control of the UPR. Hac1p isexpressed only after a 252 nucleotide intron is spliced fromthe HAC1 mRNA [7,9]. HAC1u (uninduced) mRNA, whichbears the intron, is present in cells under normal growthconditions, whereas HAC1i (induced) mRNA, which hasbeen spliced and which expresses the active transcriptionfactor, accumulates only in cells carrying out the UPR.Activation of Ire1p causes cleavage of HAC1u mRNA; thesplicing reaction is completed by tRNA ligase, whichcarries out the ligation step to form HAC1i mRNA [10].

The UPR pathway is the first case in which tRNA ligasehas been implicated in pre-mRNA splicing. By this and

other criteria, the HAC1 mRNA splicing reaction obviouslydiffers from conventional spliceosome-mediated pre-mRNA splicing [10,11]. Thus, a challenge in the field hasbeen to identify the endonuclease that cleaves HAC1u

mRNA. This was accomplished recently by Sidrauski andWalter [11], who have demonstrated conclusively that Ire1pis itself the site-specific endonuclease for HAC1 mRNA.

That Ire1p might have endoribonuclease activity wasinitially suggested by the strong homology of the Ire1p taildomain — the last 133 amino acids of the protein — to thepresumed nuclease domain of RNase L [10,12]. RNase L,also known as 2-5A-dependent RNase, is an interferon-inducible endonuclease thought to be involved in theantiviral and antiproliferative aspects of interferon action inmammalian cells. Like Ire1p, RNase L oligomerizes whenactivated, and its putative nuclease domain is located at theend of the protein, just after its kinase domain [13,14].

Using the HAC1 mRNA intron and bacterially expressedfragments of Ire1p, Sidrauski and Walter [11] showed thata carboxy-terminal fragment of Ire1p — including boththe kinase and the tail domains — has endonuclease activ-ity in vitro, and cleaves HAC1 mRNA at the correct splicesites. Other nucleases, including tRNA endonuclease andRNase A, do not produce specific cleavage products of theintron. Interestingly, the Ire1p kinase domain and anadenosine nucleotide (ATP, ADP or AMPPNP, but notGTP) are required for nuclease activity; the Ire1p taildomain alone does not carry out the cleavage reaction.This is consistent with the observation that both the Ire1pkinase and tail domains are required in vivo for inductionof the UPR [3,4]. The exact role of the kinase domain inthe endonuclease reaction remains to be determined.

By adding purified yeast tRNA ligase to their in vitroreaction, Sidrauski and Walter [11] were able to reconsti-tute HAC1 mRNA splicing completely. The intron wasexcised and the mRNA was correctly religated quiteefficiently. This demonstrates that splicing of HAC1mRNA is a remarkably simple reaction, requiring onlyIre1p and tRNA ligase. Moreover, because tRNA ligaseacts in this pathway immediately after Ire1p, and becausetRNA ligase is localized in the nucleus, near nuclear pores[15], it seems likely that the active form of Ire1p accumu-lates in the inner nuclear membrane and splices HAC1mRNA before it is exported to the cytoplasm.

Sidrauski and Walter [11] also report that single pointmutations at either of the HAC1u mRNA cleavage sitesaffect cleavage of only the mutated site, suggesting that

the two ends of the intron are cleaved independently. Inthis respect, the Ire1p-dependent splicing mechanismdiffers from spliceosome-mediated mRNA splicing, wherecleavage of the 5′ splice junction obligatorily precedescleavage of the 3′ splice junction. It is, however, similar totRNA splicing, where the ends of the tRNA intron arecleaved independently [16]. Ire1p is known to be anoligomer when active [4,17], and very similar stem–loopstructures have been predicted for the two splice junctionsof HAC1u mRNA [11]. Thus, Ire1p might bind to theHAC1u mRNA intron as a dimer, with one subunit cleav-ing each splice site. Furthermore, dimerization of Ire1pmight be important in holding the HAC1 exons close toeach other for accurate and efficient re-ligation.

How does removal of the HAC1 mRNA intron promoteexpression of Hac1pi, the protein encoded by HAC1i

mRNA? HAC1u mRNA is present in cells under normal

growth conditions at approximately the same concentra-tion as that of HAC1i mRNA in cells undergoing the UPR.But accumulation of Hac1pu, the protein encoded byHAC1u mRNA, is not detectable in wild-type cells [7,9].The role of the HAC1 intron in regulating expression ofHac1p was examined in a recent Current Biology paper byChapman and Walter [18].

Three possible explanations for the failure of Hac1pu toaccumulate were considered. First, in the absence of theUPR, the Hac1pu protein might be produced but might bevery susceptible to degradation. This possibility was initiallyfavored [7], but experiments in cells engineered to expressonly Hac1pi or Hac1pu have demonstrated that the two dif-ferent forms of the transcription factor are equally stable,each with a half-life of approximately two minutes [9,18]. Asecond possibility was that unspliced HAC1u mRNA isretained in the nucleus, and that removal of the intron was

R122 Current Biology, Vol 8 No 4

Figure 1

Few unfolded proteins in ER Many unfolded proteins in ER

Transcription of genes encoding ER-resident chaperones

HAC1i mRNA

tRNA ligase

Hac1pi

ER lumen

Cytoplasm

Unfolded proteins

Transport and translation

Current Biology

Inactive Ire1p

Activated Ire1p endonuclease

Transport

Translation of HAC1u

mRNA is attenuated

Nuclear pore

Nuclear membrane

Nucleus

HAC1u mRNA

A model for the unfolded protein response (UPR) pathway in yeast.Ire1p, a transmembrane kinase with endonuclease activity, is somehowactivated by accumulation of unfolded proteins in the ER. ActivatedIre1p cleaves the intron from HAC1u mRNA. The splicing reaction iscompleted by tRNA ligase and the resulting HAC1i mRNA is exportedto the cytoplasm for translation. Hac1pi is made and inducestranscription of genes encoding ER-resident chaperones. In the

absence of the UPR, HAC1u mRNA is transported to the cytoplasmand its translation is initiated; the presence of the intron, however,causes attenuation of translation and the HAC1u transcriptaccumulates in the cytoplasm coated with stalled ribosomes. Note thatthe localization of Ire1p has not been determined; but, as depictedhere, it is possible that it lies near nuclear pores. See text for details.

necessary to allow export and then translation in the cyto-plasm. Chapman and Walter [18] ruled out this possibilityby examining the localization of HAC1 mRNA directly,using in situ hybridization. They found that both HAC1u andHAC1i mRNAs are transported efficiently to the cytoplasmand that neither transcript ever accumulates in the nucleus.

A third possibility is regulation at the level of translation.Previously, Cox and Walter [7] had shown that the HAC1u

mRNA is found on polyribosomes, suggesting that itstranslation is at least initiated. Indeed, Chapman andWalter [18] found evidence that the amino terminus ofHac1p is actively translated in cells not expressing thefull-length transcription factor. They transformed cellswith a plasmid encoding Hac1p with the hemagglutinin(HA) epitope tag fused to its amino terminus, grew thecells under conditions in which the UPR was not induced,and made extracts. Immunoprecipitation with anti-HAantibodies precipitated HAC1 mRNA from the cellextract, presumably because the epitope was produced byand bound to ribosomes translating the HAC1u mRNA.

Chapman and Walter [18] conclude that translationelongation, and not initiation, is inhibited by the presence ofthe HAC1u intron. This makes sense given that the HAC1i

and HAC1u mRNAs have identical 5′ ends and thus must berecognized by identical translation initiation complexes.Furthermore, it means that the UPR can be triggered veryrapidly, by changing the activity of a single enzyme, Ire1p.The other components required for the response, forexample the proteins necessary to transcribe the HAC1 geneand to begin translation of the HAC1 mRNA, seem to beactive prior to induction of the UPR (Figure 1).

The primary control over Hac1p function thus appears tobe at the level of translation. A very small amount ofHac1pu is probably made by a few ribosomes that escaperegulation by the intron. Apart from its low levels,however, this ‘uninduced’ form of the protein promoteslittle UPRE-dependent transcription. Hac1pu is less tran-scriptionally active than Hac1pi [9,18], even though thetwo proteins differ in only the last few amino acids at theircarboxyl termini [7].

The mechanism by which the HAC1 intron regulates trans-lation is unknown, but our understanding of translationalcontrol will no doubt be expanded by studying this novelexample. It is possible that the HAC1 intron may eventuallyprovide a useful means by which to regulate the expressionof other eukaryotic genes. Chapman and Walter [18] havealready shown that the HAC1 intron — with small parts ofthe adjacent HAC1 exons — can attenuate translation of aheterologous mRNA (encoding the green fluorescentprotein, GFP) when placed in the 3′ untranslated regionafter the GFP stop codon [18]. Although regulated splicingof the associated transcript seems to require additional

sequences from the HAC1 exons, it seems only a matter oftime before the minimal sequences required for translationattenuation and for Ire1p-dependent splicing are identified.

As we learn more about the UPR, it seems to be a cellbiology course instructor’s dream pathway. The pathwayfollows a tour through many different cellular compart-ments — from the ER to the nucleus, to the cytoplasm,and back — and utilizes an unexpectedly wide variety ofregulatory mechanisms. Beware, students, it might appearin a question on your next exam!

AcknowledgementsI thank Maho Niwa, Ted Powers and Peter Sorger for comments on themanuscript, and acknowledge the support of a postdoctoral fellowship fromthe Jane Coffin Childs Memorial Fund for Medical Research.

References1. Shamu CE: Splicing together the unfolded protein response. Curr

Biol 1997, 7:R67-R70.2. Cox JS, Shamu CE, Walter P: Transcriptional induction of genes

encoding endoplasmic reticulum resident proteins requires atransmembrane protein kinase. Cell 1993, 73:1197-1206.

3. Mori K, Ma W, Gething MJ, Sambrook JF: A transmembrane proteinwith a cdc2+/CDC28-related kinase activity is required forsignaling from the ER to the nucleus. Cell 1993, 74:743-756.

4. Shamu CE, Walter P: Oligomerization and phosphorylation of theIre1p kinase during intracellular signaling from the endoplasmicreticulum to the nucleus. EMBO J 1996, 15:3028-3039.

5. Mori K, Sant A, Kohno K, Normington K, Gething MJ, Sambrook JF: A22 bp cis-acting element is necessary and sufficient for theinduction of the yeast KAR2 (BiP) gene by unfolded proteins.EMBO J 1992, 11:2583-2593.

6. Kohno K, Normington K, Sambrook J, Gething MJ, Mori K: Thepromoter region of the yeast KAR2 (BiP) gene contains aregulatory domain that responds to the presence of unfoldedproteins in the endoplasmic reticulum. Mol Cell Biol 1993, 13:877-890.

7. Cox JS, Walter P: A novel mechanism for regulating activity of atranscription factor that controls the unfolded protein response.Cell 1996, 87:391-404.

8. Mori K, Kawahara T, Yoshida H, Yanagi H, Yura T: Signalling fromendoplasmic reticulum to nucleus: transcription factor with abasic-leucine zipper motif is required for the unfolded protein-response pathway. Genes Cells 1996, 1:803-817.

9. Kawahara T, Yanagi H, Yura T, Mori K: Endoplasmic reticulumstress-induced mRNA splicing permits synthesis of transcriptionfactor Hac1p/Ern4p that activates the unfolded protein response.Mol Biol Cell 1997, 8:1845-1862.

10. Sidrauski C, Cox JS, WalterP: tRNA ligase is required for regulatedmRNA splicing in the unfolded protein response. Cell 1996,87:405-413.

11. Sidrauski C, Walter P: The transmembrane kinase Ire1p is a site-specific endonuclease that initiates mRNA splicing in theunfolded protein response. Cell 1997, 90:1031-1039.

12. Bork P, Sander C: A hybrid protein kinase–RNase in an interferon-induced pathway? FEBS Lett 1993, 334:149-152.

13. Dong B, Silverman RH: 2-5A-dependent RNase moleculesdimerize during activation by 2-5A. J Biol Chem 1995, 270:4133-4137.

14. Dong B, Silverman RH: A bipartite model of 2-5A-dependentRNase L. J Biol Chem 1997, 272:22236-22242.

15. Clark MW, Abelson J: The sub-nuclear localization of tRNA ligasein yeast. J Cell Biol 1987, 105:1515-1526.

16. Phizicky EM, Greer CL: Pre-tRNA splicing: variation on a theme orexception to the rule? Trends Biochem Sci 1993, 18:31-34.

17. Welihinda AA, Kaufman RJ: The unfolded protein responsepathway in Saccharomyces cerevisiae: oligomerization and trans-phosphorylation of Ire1p (Ern1p) are required for kinaseactivation. J Biol Chem 1996, 271:18181-18187.

18. Chapman RE, Walter P: Translational attenuation mediated by anmRNA intron. Curr Biol 1997, 7:850-859.

Dispatch R123