A Novel Role for Protein Kinase Kin2 in Regulating HAC1 mRNA Translocation, Splicing, and Translation

Abstract

A signaling network called the unfolded protein response (UPR) resolves the protein-folding defects in the endoplasmic reticulum (ER) from yeasts to humans. In the yeast Saccharomyces cerevisiae, the UPR activation involves (i) aggregation of the ER-resident kinase/RNase Ire1 to form an Ire1 focus, (ii) targeting HAC1 pre-mRNA toward the Ire1 focus that cleaves out an inhibitory intron from the mRNA, and (iii) translation of Hac1 protein from the spliced mRNA. Targeting HAC1 mRNA to the Ire1 focus requires a cis-acting bipartite element (3′BE) located at the 3′ untranslated leader. Here, we report that the 3′BE plays an additional role in promoting translation from the spliced mRNA. We also report that a high dose of either of two paralogue kinases, Kin1 and Kin2, overcomes the defective UPR caused by a mutation in the 3′BE. These results define a novel role for Kin kinases in the UPR beyond their role in cell polarity and exocytosis. Consistently, targeting, splicing, and translation of HAC1 mRNA are substantially reduced in the kin1Δ kin2Δ strain. Furthermore, we show that Kin2 kinase domain itself is sufficient to activate the UPR, suggesting that Kin2 initiates a signaling cascade to ensure an optimum UPR.

INTRODUCTION

The endoplasmic reticulum (ER) plays a central role in many cellular processes, including protein folding, sorting, and secretion (1). Stresses that disrupt ER function cause an overaccumulation of unfolded or misfolded proteins in the ER, creating a state referred to as “ER stress” (2, 3). To adapt to ER stress, cells activate a network of signaling pathways collectively called the unfolded protein response (UPR). The UPR provides adaptive mechanisms to withstand the ER stress while (i) reducing the overall rate of protein synthesis, (ii) accelerating the protein folding capacity of the ER by increasing the expression of ER-resident chaperones, and (iii) facilitating the retrograde translocation of proteins from the ER to the cytoplasm for degradation (4, 5).

Dual kinase/RNase Ire1 is a UPR sensor conserved from yeasts to humans (6–9). Ire1 is composed of a luminal domain (Ire1^LD) and a cytoplasmic domain (Ire1^cyto) connected by an ER transmembrane (TM) region. Ire1^cyto contains a protein kinase domain and an RNase domain known as the KEN (kinase extension nuclease) domain (6). In resting cells, Ire1^LD is bound to the ER-resident chaperon Kar2 (10, 11). During ER stress conditions, unfolded proteins titrate Kar2 from the luminal domain, and Ire1 proteins aggregate to form an oligomer across the ER membrane, which is observed as a punctate fluorescent dot (Ire1 focus) (8, 12). In these oligomers, both kinase and KEN domains are activated. The active KEN domain then cleaves out an inhibitory intron from the HAC1 pre-mRNA in yeast cells (13) or from the XBP1 pre-mRNA in human cells (14). Two exons are then joined together to form a mature mRNA (15, 16) that yields a functional Hac1 or Xbp1 protein. Hac1/Xbp1 is a transcription factor that induces expression of genes that participate in the UPR signaling pathway.

Under normal growth conditions, the HAC1 pre-mRNA remains in a translationally repressed state in the cytoplasm, and under conditions of ER stress, pre-mRNA is targeted toward the Ire1 focus that cleaves out the inhibitory intron (12). Targeting is a complex process that requires a cis-acting 3′-bipartite element (3′BE) (Fig. 1) located at the 3′ untranslated region (3′UTR) of mRNA (12); however, the underlying mechanism of the 3′BE-mediated targeting of the translationally repressed mRNA remains unclear. Here we report that two 3′BE nucleotides (GG positions 1143 to 1144 [GG_1143-1144]) (i) guide targeting of HAC1 pre-mRNA toward the Ire1 foci, (ii) promote splicing of HAC1 pre-mRNA, and (iii) increase the efficiency of translation from the matured mRNA. These findings led us to propose that a 3′BE-ribonucleoprotein (RNP) complex centered on nucleotides GG_1143-1144 regulates both targeting and splicing of HAC1 pre-mRNA and subsequently promotes translation of matured mRNA.

FIG 1.

Nucleotides GG_1143-1144 play an important role in regulating HAC1 mRNA targeting and splicing. (A) Schematic representation of the HAC1 pre-mRNA. The 7-methylguanosine (m⁷G) cap, 5′- and 3′UTRs, exons, intron, and polyadenine (An) tail are shown. The 3′-bipartite element (3′BE) is shown in a box. Two conserved RNA motifs (5′-U₁₁₄₁GGCGC₁₁₄₈-3′ in red and 5′-G₁₁₈₀CGAC₁₁₈₄-3′ in green) are predicted to interact with each other (12), thus forming a helix-bulge (HB) structure. The adenine of the start codon AUG is assigned position 1. Other nucleotides are numbered accordingly and shown on top of the sketch. (B) Mutations of the nucleotides GG_1143-1144 impair yeast cell growth on the tunicamycin medium. The hac1Δ yeast strains expressing the indicated HAC1 alleles were serially diluted and spotted on SD medium alone and SD medium containing tunicamycin. (C) Mutations of the GG_1143-1144 nucleotides reduce HAC1 mRNA splicing. Total RNA was isolated from yeast strains indicated in panel B, and RT-PCR was used to analyze spliced (HAC1^s) and unspliced (HAC1^u) HAC1 transcripts. The relative levels of HAC1^s transcript were estimated by measuring the ratio of HAC1^s and total (HAC1^u plus HAC1^s) band intensities (measured by NIH ImageJ software), normalizing the ratio with the respective intensity of the 28S rRNA band, and then turning the resulting ratio value into percentage. (D) The deletion of the 3′BE reduces the ER stress response. A hac1Δ ire1Δ pIRE1-YFP yeast strain expressing the indicated HAC1-NRE variant was tested for growth on SD and tunicamycin media. (E) Analysis of HAC1 mRNA transcripts. RT-PCR was used to analyze spliced (HAC1^s) and unspliced (HAC1^u) HAC1 transcripts in cells indicated in panel D. (F) The GG_1143-1144CC mutations impair HAC1 mRNA colocalization with Ire1 protein. An ire1Δ hac1Δ strain coexpressing three plasmid-borne genes IRE1-YFP, the ND-GFP2 gene, and the indicated WT HAC1 or its derivative, HAC1-NRE-ΔBE or HAC1-NRE-GG_1143-1144CC was grown under an ER stress condition. Ire1 (red dots) and HAC1 mRNA (green dots) foci were visualized by confocal microscopy.

In an effort to identify the components of the proposed 3′BE-RNP complex, we conducted a dosage-suppressor screen using a crippled 3′BE construct (i.e., GG_1143-1144 mutated to CC [HAC1-GG_1143-1144CC]). We discovered that either of two paralogue kinases Kin1 and Kin2 (17–19) but not their kinase-dead derivatives overcame the defective UPR associated with the HAC1-GG_1143-1144CC allele. Consistently, we observed that the targeting, splicing, and translation of HAC1 mRNA were substantially reduced in the kin1Δ kin2Δ double null strain. These results collectively define a novel role for Kin kinases in molecular response to ER stress in addition to their central role in regulating cell polarity and exocytosis (19).

Next, we showed by in vivo imaging studies that the kinase Kin2 was predominantly located in the cytoplasm. This observation, together with the previous observation that both Kin1 and Kin2 are membrane-associated kinases (20), indicates that Kin kinases are endomembrane kinases. Furthermore, we showed that the Ire1-mediated ER stress response was reduced but not completely abolished in the kin1Δ kin2Δ strain and that the expression of the Kin2 kinase domain itself was sufficient to activate the ER stress response. Taken together, we conclude that the Kin kinases act in the yeast UPR not only by modulating the Ire1-Hac1 pathway but also by constituting an independent signaling cascade that augments the available avenues by which cells adapt to ER stress.

MATERIALS AND METHODS

Yeast strains and gene disruption.

The KanMX cassette in either the hac1::KanMX, ire1::KanMX, or kin1::KanMX strain was replaced by a NatMX gene by standard protocol of PCR-mediated gene disruption. In the resulting hac1::NatMX strain, either the IRE1, KIN1, or KIN2 gene was disrupted by a KanMX gene to create hac1Δ ire1Δ, hac1Δ kin1Δ, and hac1Δ kin2Δ strains, respectively. The KIN2 gene of the kin1::NatMX strain was disrupted by a KanMX cassette to create a kin1Δ kin2Δ strain. The strains used for this study are shown in Table 1.

TABLE 1.

List of strains used in this study

Strain type	Description	Source (reference)
WT	MATa his3-Δ1 leu2-Δ0 met5-Δ0 ura3-Δ0	Research Genetics
Mutant
hac1Δ	MATa his3-Δ1 leu2-Δ0 met5-Δ0 ura3-Δ0 hac11::kanMX	Research Genetics
ire1Δ	MATa his3-Δ1 leu2-Δ0 met5-Δ0 ura3-Δ0 ire1::kanMX	Research Genetics
kin1Δ	MATa his3-Δ1 leu2-Δ0 met5-Δ0 ura3-Δ0 kin1::kanMX	Research Genetics
kin2Δ	MATa his3-Δ1 leu2-Δ0 met5-Δ0 ura3-Δ0 kin2::kanMX	Research Genetics
hac1Δ kin1Δ	MATa his3-Δ1 leu2-Δ0 met5-Δ0 ura3-Δ0 hac11::NatMX kin1::kanMX	This study
hac1Δ kin2Δ	MATa his3-Δ1 leu2-Δ0 met5-Δ0 ura3-Δ0 hac11::NatMX kin2::kanMX	This study
ste2Δ	MATa leu2-3,112 ura3-52 his3-Δ1 trp1 ste2::leu2 sst1-Δ5	Raicu et al. (23)
ire1Δ hac1Δ	MATa his3-Δ1 leu2-Δ0 met5-Δ0 ura3-Δ0 ire1::kanMX hac11::NatMX	Mannan et al. (21)
kin1Δ kin2Δ	MATa his3-Δ1 leu2-Δ0 met5-Δ0 ura3-Δ0 kin1::NatMX Kin2::kanMX	This study

Plasmids.

The appropriate open reading frame (ORF) (IOC2, APC9, HRT3, ERG27, YLR099W, or ICT1) along with a 0.5-kb upstream promoter and 0.5-kb downstream terminator was PCR amplified from plasmid p59. The PCR product was then inserted into a high-copy-number 2μ plasmid. The nucleolin domain that recognizes the nucleolin recognition element (NRE) (residues 294 to 454) was fused to a monomeric form of green fluorescence protein 2 (GFP2) and expressed under a constitutive promoter and terminator (ADH1). Mutation was created by site-directed mutagenesis using standard protocols. The list of plasmids used in this study is provided in Table 2.

TABLE 2.

List of plasmids used in this study

Plasmid	Description	Source (reference)a
D4	pRS316, low-copy-number URA3 vector	Thomas E. Dever
D8	pRS426, 2μ URA3 vector	Thomas E. Dever
D63	HAC1-WT in D4-KpnI/BamHI sites	This study
D712	HAC1-GG1143-1144CC in low-copy-number LEU2 vector	This study
D713	HAC1-GG1143-1144CC in low-copy-number URA3 vector	This study
D759	HAC1-CGC1145-1147GCG in low-copy-number URA3 vector	This study
D760	HAC1-GCG1180-1182CGC in low-copy-number URA3 vector	This study
D761	HAC1-AC1183-1184CG in low-copy-number URA3 vector	This study
D64	HAC1-ΔBE (A1156-U1174) in low-copy-number URA3 vector	This study
D144	HAC1-G661C in low-copy-number URA3 vector	This study
D69	HAC1Δi in low-copy-number URA3 vector	This study
D848	HAC1Δi-G661C in low-copy-number URA3 vector	This study
D559	HAC1Δi-GG1143-1144CC in low-copy-number URA3 vector	This study
D72	IRE1 WT in low-copy-number LEU2 vector	Lee et al. (6)
D765	IOC2 in 2μ URA3 vector (BamHI-HindIII)	This study
D766	APC9 in 2μ URA3 vector (SacI-XhoI)	This study
D767	HRT3 in 2μ URA3 vector (SacI-XhoI)	This study
D768	ERG27 in 2μ URA3 vector (SpeI-KpnI)	This study
D769	YLR099W in 2μ URA3 vector	This study
D770	ICT1 in 2μ URA3 vector (SacI-XhoI)	This study
D619	KIN1 in 2μ URA3 vector	Benjamin Turk
D622	KIN1-D289A, high-copy-number URA3 vector	Patrick Brennwald
D197	KIN2 in 2μ URA3 vector	Patrick Brennwald
D733	KIN2-D248A, high-copy-number URA3 vector	This study
D51	UPRE-lacZ in high-copy-number URA3 vector	Lee et al. (6)
D763	p59 in high-copy-number URA3 vector	This study
D764	p2 in high-copy-number URA3 vector	This study
D762	p41 in high-copy-number URA3 vector	This study
D1008	HAC1-NRE in low-copy-number URA3 vector	This study
D995	ND-GFP2 in low-copy-number LEU2	This study
D102	STE2-YFP in high-copy-number TRP vector	This study
D1006	IRE1-YFP in high-copy-number LEU2 vector	Mannan et al. (21)
D625	KIN2-GFP in low-copy-number URA3 vector	Benjamin Turk
D1129	KIN2KD in pEMBLyex4	This study
D1130	HAC1-NRE-ΔBE in low-copy-number URA3 vector	This study
D1131	HAC1-NRE-GG1143-1144CC in low-copy-number URA3 vector	This study

^aAffiliations: Thomas E. Dever, National Institutes of Health, Bethesda, MD; Benjamin E. Turk, Yale University School of Medicine, New Haven, CT; Patrick Brennwald, University of North Carolina, Chapel Hill.

Genetic screen to identify the dosage suppressor of the HAC1-GG_1143-1144CC allele.

A hac1Δ strain was transformed with a low-copy-number LEU2 plasmid containing a HAC1-GG_1143-1144CC variant. The resultant hac1Δ pHAC1-GG_1143-1144CC::LEU2) strain was then transformed with a yeast genomic library in a high-copy-number URA3 plasmid and screened for tunicamycin-resistant (Tm^r) colonies. From 72 Tm^r colonies, we rescued 72 URA3 plasmids (denoted p1, p2, p3, etc., to p72). As expected, we got three populations of plasmids: (i) a set of 69 plasmids, each containing an HAC1 open reading frame (ORF); (ii) two plasmids (p2 and p7), each containing an ORF (e.g., ALG1) that bypassed the Ire1-dependent ER stress response (bypass suppressors); and (iii) one plasmid (p59) containing an ORF that rescued the ER stress response by mass action (a dosage suppressor). DNA sequence analysis revealed that the plasmid p59 contained a fragment from chromosome XII, which carried 8 ORFs (IOC2, KIN2, HRT3, CHA4, ICT1, YLR099W, ERG27, and APC9). Expression of the individual gene identifies that the kinase Kin2 alone was sufficient to restore the defective UPR associated with the HAC1-GG_1143-1144CC allele.

RNA isolation and RT-PCR.

Yeast cells were grown in an synthetic dextrose (SD) medium with appropriate nutrients at 30°C until the optical density at 600 nm (OD₆₀₀) reached ∼0.5 to 0.6. Then, dithiothreitol (DTT; 5 mM unless described otherwise) was added to the medium to induce the ER stress, and cells were grown for another 4 h. Cells were harvested, and total RNA was isolated using an RNeasy minikit (Qiagen). Purified RNA was quantified using a Nanodrop spectrophotometer (ND-1000; Thermo Scientific) and used to synthesize the first-strand cDNA with Superscript-III reverse transcriptase (Invitrogen). The synthetic cDNA was then PCR amplified using a set of primers designed to amplify a part of the HAC1 or ACT1 gene. For analysis of HAC1 transcript, we used a forward primer (5′-CGCAATCGAACTTGGCTATCCCTACC-3′) that corresponded to nucleotides +35 to 60 (where +1 denotes the adenine of the start codon AUG) and a reverse primer (5′-GGGTAGACTGTTTCCCGC-3′) that corresponded to nucleotides +604 to 621. For HAC1 mRNA splicing profile analysis, the same forward primer was used along with a reverse primer (5′-CCCACCAACAGCGATAATAACGAG-3′) that corresponded to nucleotides +1002 to 1025. For ACT1 transcript analysis, we used a forward primer (5′-CTGAAAGAGAAATTGTCCGTG-3′) that corresponded to nucleotides +919 to 939 (where +1 denotes the adenine of the start codon AUG) and a reverse primer that corresponded to nucleotides 5′-CTTGTGGTGAACGATAGATGG-3′.

Whole-cell extract preparation and Western blot analysis.

Yeast cells were grown in an SD medium with appropriate nutrients until the OD₆₀₀ reached 0.6. A 5 mM concentration of DTT was then added to the medium to induce the ER stress, and cells were harvested after 4 h. Whole-cell lysates were prepared by the trichloroacetic acid (TCA) method as described previously (21). Proteins were then fractioned by SDS-PAGE, and Western blot analysis was performed using an antibody raised against Hac1.

Confocal microscopy.

Yeast cells harboring fluorescent proteins were grown until the OD₆₀₀ reached ∼0.6. An aliquot of cells was used for in vivo imaging studies. Images were recorded using Leica TCS SP2 confocal microscope equipped with an HCX PL APO 63×/1.2-numeric-aperture water immersion objective lens. The GFP was excited at 458 nm, and emissions were recorded at 466 to 526 nm. The yellow fluorescent protein (YFP) was excited at 514 nm, and emissions were recorded at 530 to 650 nm. Colocalization studies were done using sequential scan settings to avoid cross talk between the channels. Background subtractions and processing of images were done using Leica Microsystems imaging software.

lacZ reporter assay.

The yeast strains were transformed with a plasmid containing a lacZ reporter gene under the control of a UPR element from the yeast KAR2 gene (11). Transformants were grown overnight, diluted to an OD₆₀₀ of ∼0.2 in a synthetic complete (SC) medium, and allowed to grow until the OD₆₀₀ reached ∼0.6 to 0.8. The culture was then split into two flasks: half of the culture was grown in the presence of 5 mM DTT, and the rest of the culture was grown without DTT. Cells were harvested after 4 h, protein extracts were prepared, and β-galactosidase activities were determined as described previously (21). The experiments were repeated three times. The plotted histogram represents the mean fold change with standard errors.

RESULTS

The GG_1143-1144 nucleotides within the 3′BE play a central role in guiding colocalization of HAC1 mRNA with Ire1 protein.

Under ER stress conditions, the ER-resident kinase/RNase Ire1 removes an inhibitory intron from the HAC1 pre-mRNA by cleaving two phosphodiester bonds after nucleotides G₆₆₁ and G₉₁₃ (Fig. 1A). This cleavage requires targeting of HAC1 pre-mRNA toward the ER, which is mediated by a cis-acting 3′-bipartite element (3′BE) that spans nucleotides U₁₁₂₀ to A₁₂₀₃ containing two conserved RNA motifs (5′-U₁₁₄₂GGCGCG₁₁₄₈-3′ and 5′-G₁₁₈₀CGAC₁₁₈₄-3′) (Fig. 1A) (12). To obtain new insights into targeting of HAC1 pre-mRNA, we initially defined roles of two landmark nucleotides (GG_1143-1144) by carrying out both deletion and mutational analyses of the 3′BE (Fig. 1B).

As shown in Fig. 1B, a hac1Δ strain containing an empty vector or the same vector expressing a wild-type (WT) HAC1 allele grew well on a synthetic dextrose (SD) medium (SD, lanes 1 and 2); however, growth on the SD medium containing an ER stress inducer, tunicamycin, required HAC1 gene expression (Fig. 1B, Tunicamycin, compare rows 1 and 2). This is consistent with the earlier observation that HAC1 gene function is essential for adaptation and survival of cells during ER stress (13). Substitution of a single nucleotide in the splicing site (i.e., G₆₆₁ to C) or two consecutive nucleotides together in the 3′BE (i.e., GG_1143-1144 to CC) caused severe growth defects on the tunicamycin medium (Fig. 1B, tunicamycin, rows 3 and 4), indicating that these mutations either decreased the mRNA level or diminished mRNA splicing.

To determine the level of mRNA and its splicing, yeast cells were grown in a liquid SD medium in the presence of DTT (ER stress inducer) and harvested as described in Materials and Methods. Total RNA was isolated, and reverse transcription (RT)-PCR was employed to analyze unspliced (HAC1^u) and spliced (HAC1^s) HAC1 transcripts. As expected, no HAC1 mRNA was detected in yeast cells containing the vector plasmid (Fig. 1C, RT-PCR, lane 1), and both HAC1^u and HAC1^s transcripts were detected in cells expressing a WT HAC1 allele (Fig. 1C, RT-PCR, lane 2). The relative levels of HAC1^s transcript were estimated by measuring the ratio of HAC1^s and total (HAC1^u plus HAC1^s) band intensities, normalizing the ratio with the respective intensity of the 28S rRNA band, and then turning the resulting ratio value into percentage. The HAC1^s mRNA population was about 44% in cells expressing the WT HAC1 allele, conforming to the previous observation (13) that yeast growth under an ER stress condition is directly related to HAC1 mRNA splicing. In HAC1-G₆₆₁C and HAC1-GG_1143-1144CC cells (Fig. 1C, RT-PCR, HAC1^u), the majority of HAC1 mRNA remained in unspliced forms, and only 0.7% and 14%, respectively, were in spliced forms (Fig. 1C, RT-PCR, HAC1^s, lanes 3 and 4). This decrease in the HAC1^s mRNA population suggests that the defect in growth on the medium containing tunicamycin was due to a defect in splicing of HAC1 mRNA.

To determine whether the splicing defect of HAC1-GG_1143-1144CC mRNA was related to impaired targeting of intracellular mRNA; we examined the cellular localization and the colocalization of HAC1 mRNA with Ire1 protein by live cell imaging studies. In an ire1Δ hac1Δ cell, we expressed a derivative of HAC1 mRNA (HAC1-NRE), in which a 22-nucleotide RNA module, identical to the nucleolin recognition element (NRE) (22), was inserted at the 3′UTR. In the same cell, we expressed an Ire1 variant that has been conjugated to a yellow fluorescent protein (YFP) (21). Expressions of both Ire1-YFP fusion protein and HAC1-NRE mRNA allowed the ire1Δ hac1Δ yeast cells to grow on the tunicamycin medium (Fig. 1D, WT, tunicamycin, lane 1), and in those cells, a majority (62%) of the total HAC1 mRNA population was in the spliced form (Fig. 1E, RT-PCR, HAC1^s, lane 1), suggesting that the NRE insertion did not interfere with the normal function of HAC1 gene. When the entire 3′BE was deleted from the HAC1-NRE allele (i.e., HAC1-NRE-Δ3′BE), or specifically the nucleotides GG_1143-1144 were mutated (i.e., HAC1-NRE-GG_1143-1144CC), yeast cell growth was impaired on the tunicamycin medium (Fig. 1D, tunicamycin, rows 2 and 3). Further RT-PCR analysis showed that the HAC1^s mRNA population was only 19% in the Δ3′BE cells (Fig. 1E, RT PCR, HAC1^s, lane 2) and 18% in the GG_1143-1144CC cells (Fig. 1E, RT-PCR, HAC1^s, lane 3). This decrease in HAC1^s population suggests that nucleotides GG_1143-1144 play a central role in mRNA translation, partly by influencing pre-mRNA targeting and/or splicing.

Then, we determined the possible role of nucleotides GG_1143-1144 in mRNA targeting. The indicated yeast cells (Fig. 1D) were transformed with a plasmid that expressed a chimera protein consisting of an NRE-binding nucleolin domain (ND [residues 294 to 454]) (22) and a monomeric form of the green fluorescent protein 2 (GFP2) (23). We hypothesized that the specific interaction between ND-GFP2 and NRE would result in visualization of HAC1 mRNA as discrete dots (foci) in live cells. Our hypothesis was based on a similar visualization tool for the HAC1-U1A allele (16-U1A spliceosomal RNA modules inserted at the 3′UTR), which employs a chimera protein consisting of an U1A RNA-binding protein and a green fluorescent protein (12). As shown in Fig. 1F, both HAC1 mRNA (green dots) and Ire1 protein (red dots) appeared as discrete foci in yeast cells when expressing an HAC1-NRE allele. The merged image indicated that HAC1 mRNA and Ire1 protein were largely colocalized. In contrast, colocalization of HAC1 mRNA with Ire1 protein almost disappeared in cells when expressing an HAC1-NRE-Δ3′BE or HAC1-NRE-GG_1143-1144CC allele (Fig. 1F), although each candidate mRNA and protein was observed as a discrete dot (Fig. 1F). These results suggest that nucleotides GG_1143-1144 promote targeting of HAC1 pre-mRNA toward the Ire1 protein under conditions of ER stress.

The nucleotides GG_1143-1144 promote translation of HAC1 mRNA.

The 3′UTR serves multiple functions, including modulation of mRNA stability and translational efficiency (24); thus, we examined any role of nucleotides GG_1143-1144 in HAC1 mRNA translation. An intronless HAC1 variant (HAC1ⁱ) was constructed, and then the GG_1143-1144CC mutations were incorporated into the HAC1ⁱ variant, generating an HAC1ⁱ-GG_1143-1144CC mutant. The HAC1ⁱ and HAC1ⁱ-GG_1143-1144CC mutants were individually introduced in an ire1Δ or hac1Δ strain. As shown in Fig. 2A, the ire1Δ strain containing an empty vector (row 1) or expressing a WT HAC1 (row 2), did not grow on the tunicamycin medium, but the same strain expressing an intronless HAC1ⁱ allele grew on the tunicamycin medium (row 3). These data are consistent with the earlier report (25) that the Ire1-mediated excision of the HAC1 mRNA intron is an important event in the ER stress response. Note that the ire1Δ yeast strain containing a HAC1ⁱ variant exhibited a slow-growth phenotype (Fig. 2A, yeast growth, SD medium, row 3). The slow growth was more prominent in the hac1Δ strain (Fig. 2B, row 2), whether or not the HAC1ⁱ variant contained a G₆₆₁C mutation (Fig. 2B, HAC1ⁱ-G₆₆₁C, row 3). It should be noted that in the HAC1ⁱ-G₆₆₁C allele, the native GAA codon encoding Glu-221 (E221) changes to CAA (which encodes Gln, or in the single-letter format, Q); thus the HAC1ⁱ-G₆₆₁C allele encodes a mutant Hac1-E₂₂₁Q protein. Furthermore, the hac1Δ strain indicated in Fig. 2B was transformed with a plasmid bearing a UPRE-driven lacZ reporter, and the lacZ expression was monitored (Fig. 2D) under an ER stress condition as we described earlier (6). The UPRE-driven lacZ expression was elevated even without any trigger for the ER stress (Fig. 2D, −DTT, HAC1ⁱ and HAC1ⁱ-G₆₆₁C). These results indicated that the growth defects were likely due to constitutive induction of the UPR genes and that the E221Q mutation did not alter the function of Hac1 protein. Furthermore, these results suggested that the impaired UPR as shown previously in the HAC1-G₆₆₁C cells (Fig. 1) was due exclusively to a defect in mRNA splicing.

FIG 2.

The GG_1143-1144 nucleotides promote translation of HAC1 mRNA. (A and B) The intronless HAC1 (HAC1ⁱ) variant exhibits a slow-growth phenotype. (A) An ire1Δ yeast strain containing an empty vector or the same vector bearing a wild-type HAC1 or HAC1ⁱ variant was tested for growth on SD and tunicamycin media. (B) A hac1Δ yeast strain containing an empty vector or the same vector expressing the indicated HAC1ⁱ gene or its derivatives was tested for growth on the SD medium. In the lower panels, total RNA was isolated from strains indicated in panel B, and RT-PCR was used to monitor the levels of HAC1 mRNA. (C) The GG_1143-1144CC mutations in the HAC1ⁱ allele decrease the Hac1 protein production. The whole-cell extracts were subjected to Western blot (WB) analysis using an antibody raised against the recombinant Hac1 protein. (D) The GG_1143-1144CC mutations reduce the Hac1-induced gene expression. The yeast strains indicated in panel B were transformed with a plasmid containing a lacZ reporter gene under the control of a UPR element from the yeast KAR2 gene (11). The β-galactosidase activities were monitored in those transformants in the presence (+) and absence (−) of DTT as described in Materials and Methods. The experiments were repeated three times, and the plotted histogram represents the mean fold change with standard errors.

The yeast cells expressing a HAC1ⁱ-GG_1143-1144CC mutant (Fig. 2B, SD, row 4) grew similarly to cells containing a vector plasmid (row 1) but grew much faster than cells expressing an HAC1ⁱ allele (row 2). Western blot (WB) analysis showed that the Hac1 protein production was decreased 4-fold in the HAC1ⁱ-GG_1143-1144CC cells compared with the HAC1ⁱ cells (Fig. 2C, WB, compare lanes 1 and 2). This inverse correlation between yeast growth and Hac1 protein production suggests that mutations in GG_1143-1144 nucleotides either decreased the mRNA level or hindered the efficiency of mRNA translation. RT-PCR analysis showed that the mRNA level of the HAC1ⁱ-GG_1143-1144 variant was near equal to that of the HAC1ⁱ variant (Fig. 2B, RT-PCR, compare lanes 2 and 4). The lacZ reporter expression was reduced to 50% in cells expressing an HAC1ⁱ-GG_1143-1144CC variant compared with cells expressing an HAC1ⁱ variant (Fig. 2D, compare HAC1ⁱ and HAC1ⁱ-GG_1143-1144CC, ±DTT), suggesting that the GG_1143-1144CC mutations decreased the translational efficiency of HAC1 mRNA. Collectively, our data suggest that nucleotides GG_1143-1144 promote both mRNA targeting and translation.

Overexpression of protein kinase Kin1 or Kin2 overcomes the UPR defect caused by the GG_1143-1144CC mutation in the 3′UTR.

Two cis-acting RNA motifs (5′-U₁₁₄₂GGCGCG₁₁₄₈-3′ and 5′-G₁₁₈₀CGAC₁₁₈₅-3′) within the 3′BE have been predicted to base pair in opposite direction (12), resulting in formation of a helix-bulge motif. The bulge or internal loop of an RNA secondary structure makes an obvious target for macromolecular interaction (26), leading us to hypothesize that the proposed bulge nucleotides on the HAC1-3′BE (i.e., UGG_1142-1144) (Fig. 1) serve as a core loop on which a ribonucleoprotein (RNP) is formed. The resulting 3′BE-RNP then governs targeting the pre-mRNA and subsequently promotes translation of the matured mRNA. To identify the 3′BE-RNP components, we conducted a high-copy-number suppressor screen to isolate genes that, via mass action (27), would restore the tunicamycin-resistant phenotype of yeast cells expressing an HAC1-GG_1143-1144CC allele (Fig. 1B, row 4). An hac1Δ strain was transformed with a low-copy-number LEU2 plasmid containing an HAC1-GG_1143-1144CC variant. The resultant hac1Δ pHAC1-GG_1143-44CC::LEU2 strain was then transformed with a yeast genomic library in a high-copy-number URA3 plasmid and screened for tunicamycin-resistant colonies. This screen identified that overexpression of protein kinase Kin2 restored the defective UPR associated with the HAC1-GG_1143-1144CC variant (see Materials and Methods).

The possible role of Kin2 in the UPR appeared to be a novel finding because Kin2 has been known to play a central role in cell polarity and exocytosis (19). Kin2 is a serine/threonine kinase with an N-terminal kinase domain and a C-terminal regulatory domain followed by a kinase-associated domain 1 (KA1) (17). Apart from Kin2, yeast also expresses a paralogue kinase, Kin1. Thus, our results led to several important questions: Does the kinase Kin1 possess the identical function? Does Kin2 regulate the HAC1 mRNA targeting? Does Kin2 play a role in transcription of the UPR target genes? To address the first question, we created two strains with double deletions: an hac1Δ kin1Δ strain and an hac1Δ kin2Δ strain (see Materials and Methods). Absence of either kinase (Kin1 or Kin2) in the hac1Δ strain did not affect normal growth (data not shown), indicating that the double deletion mutations do not affect the normal physiological function. Then, we introduced a 2μ URA3 plasmid into each of these double deletion strains. Also, we introduced the same plasmid carrying the KIN2 gene in the hac1Δ kin1Δ strain and the same plasmid carrying the KIN1 gene in the hac1Δ kin2Δ strain. The resulting strain was further transformed with a low-copy-number LEU2 plasmid harboring a WT HAC1, HAC1-G₆₆₁C, or HAC1-GG_1143-1144CC variant. Transformants were then tested for growth on both SD and tunicamycin media (Fig. 3A and B).

FIG 3.

Both Kin1 and Kin2 promote the Ire1/HAC1-mediated UPR. (A) Overexpression of Kin2 restores the UPR of the HAC1-GG_1143-1144CC allele. (Left) A hac1Δ kin1Δ yeast strain was transformed with a 2μ URA3 vector or the same vector carrying a WT KIN2 gene. In the resultant transformant, the indicated WT HAC1, HAC1-G₆₆₁C, or HAC1-GG_1143-1144CC was introduced. Yeast cells were then tested for growth on SD and tunicamycin media (left panels), and RT-PCR was used to analyze HAC1^u and HAC1^s mRNA populations under a condition of ER stress (right panel). (B) Overexpression of Kin1 restores the UPR of the HAC1-GG_1143-1144CC allele. (Left) An hac1Δ kin2Δ yeast strain was transformed with a 2μ URA3 vector or the same vector carrying a WT KIN1 gene. In the resultant transformant, the indicated WT HAC1, HAC1-G₆₆₁C, or HAC1-GG_1143-1144CC gene was introduced. Yeast cells were then tested for growth on SD and tunicamycin media (left panels), and RT-PCR was used to analyze the HAC1^u and HAC1^s mRNA populations under a condition of ER stress (right panel). (C) The Kin2 kinase domain function is essential for rescuing the UPR in HAC1-GG_1143-1144CC cells. A hac1Δ kin1Δ pHAC1-GG_1143-1144CC yeast strain was transformed with an empty vector and the same vector carrying the indicated WT Kin2 or Kin2-D₂₄₈A mutant. The resultant strains were tested for growth on SD and tunicamycin media (left panels), and RT-PCR was used to analyze the level of the HAC1^s mRNA population (right panel).

As shown in Fig. 3A and B, both hac1Δ kin1Δ and hac1Δ kin2Δ yeast strains harboring a wild-type HAC1 allele were able to grow on the tunicamycin medium (tunicamycin, rows 1 and 4). About 61% of the total HAC1 mRNA population in the hac1Δ kin1Δ strain and 35% in the hac1Δ kin2Δ strain were predominantly in the spliced form (Fig. 3A and B, right panels, RT-PCR, HAC1^s, lane 1). The difference in splicing percentage was most likely because the two strains responded differently under conditions of ER stress. Note that yeast cells grew more rapidly on the tunicamycin medium when Kin1 or Kin2 was overexpressed (Fig. 3A [tunicamycin, compare rows 1 and 4] and B [compare rows 1 and 4]). Consistent with the rapid growth, the HAC1 mRNA splicing was increased from 61% to 78% in the hac1Δ kin1Δ strain (Fig. 3A, right panel, RT-PCR, lanes 1 and 4) and from 35% to 48% in the hac1Δ kin2Δ strain (Fig. 3B, right panel, RT-PCR, lanes 1 and 4). Collectively, these results suggest that overexpression of either kinase promotes the HAC1 mRNA splicing, resulting in a survival advantage under ER stress.

Next, we tested whether overexpression of Kin1 or Kin2 kinase restored the splicing of the HAC1-G₆₆₁C allele (28). As expected, the splicing-deficient HAC1-G₆₆₁C allele in the hac1Δ kin1Δ or hac1Δ kin2Δ strain was unable to support cell growth on the tunicamycin medium (Fig. 3A and B, tunicamycin, row 2), even when Kin2 (Fig. 3A, row 5) or Kin1 (Fig. 3B, row 5) kinase was overexpressed. In each of these strains, less than 10% of the total HAC1 mRNA population was in the spliced form (Fig. 3A, right panel, HAC1^s, RT-PCR, lanes 2 and 5; Fig. 4B, right panel, RT-PCR, HAC1^s, lanes 2 and 5), suggesting that Kin kinases cannot rescue the splicing defect of the HAC1-G₆₆₁C allele. Overexpression of kinase Kin1 in the hac1Δ kin2Δ strain (Fig. 3B) or kinase Kin2 in the hac1Δ kin1Δ strain (Fig. 3A), each containing a HAC1-GG_1143-1144CC allele, allowed yeast cells to grow on the tunicamycin medium (Fig. 3A and B, yeast growth, tunicamycin, compare rows 3 and 6). RT-PCR analysis showed that the HAC1^s mRNA population was about 17% in the hac1Δ kin1Δ strain (Fig. 3A, right panel, RT-PCR, HAC1^s, lane 3), whereas it was 8% in the hac1Δ kin2Δ strain (Fig. 3B, right panel, RT-PCR, HAC1^s, lane 3). When Kin2 or Kin1 kinase was overexpressed, the spliced HAC1 mRNA population was increased from 17% to 45% in the hac1Δ kin1Δ strain (Fig. 3A, right panel, RT-PCR, HAC1^s, compare lanes 3 and 6) and from 8% to 42% in the hac1Δ kin2Δ strain (Fig. 3B, right panel, RT-PCR, HAC1^s, compare lanes 3 and 6). These results demonstrate that overexpression of kinase Kin1 or Kin2 overcomes both growth and splicing defects associated with GG_1143-1144CC mutations in HAC1 mRNA and confirm that Kin kinases play a role in the ER stress response.

FIG 4.

The kinase activity of Kin1 and Kin2 is required for the ER stress response. (A) The UPR is impaired in the kin1Δ kin2Δ strain. The WT yeast strain and its isogenic kin1Δ, kin2Δ, and kin1Δ kin2Δ mutants were grown, serially diluted, spotted, and tested for growth on SD and tunicamycin media. (B) Ire1, Hac1, and Kin1 Kin2 null mutant strains display differential sensitivities to tunicamycin. A yeast strain lacking ire1, hac1, or both kin1 and kin2 was grown, serially diluted (10⁰ to 10⁻³), and spotted and tested for growth on both SD and tunicamycin media. (C) The kinase-dead mutants of Kin1 and Kin2 reduce the ER stress response. A kin1 kin2Δ yeast strain expressing a WT protein (Kin1 or Kin2) or its kinase-dead mutant (Kin1-D₂₈₉A or Kin2-D₂₄₈A) was tested for growth on both SD and tunicamycin media.

Kinases Kin1 and Kin2 share a conserved kinase domain (KD) with more than 90% identity (17), indicating that the KD is likely to play a major role in ER stress response. Thus, we introduced a kinase-dead Kin2-D₂₄₈A variant into the hac1Δ kin1Δ strain harboring the HAC1-GG_1143-1144CC allele (hac1Δ kin1Δ pHAC1-GG_1143-44CC) and tested the UPR responses. The residue D₂₄₈ (Asp-248) is the catalytic aspartate of the kinase conserved HRD motif, which is known to facilitate substrate phosphorylation (29, 30). Thus, it is likely that Asp-248-to-Ala (D₂₄₈A) substitution will impair the kinase catalytic function. The Kin2-D₂₄₈A mutant (Fig. 3C, tunicamycin, row 3), like the vector plasmid (row 1), was unable to support yeast cell growth on the tunicamycin medium (Fig. 3C, tunicamycin, row 3). RT-PCR analyses showed that the HAC1^s mRNA population was about 35% in yeast cells containing a vector plasmid (Fig. 3, right panel, lane 1), increased to 61% in cells expressing a WT Kin2 (Fig. 3C, right panel, lane 2), and remained almost same as the vector control in cells expressing a Kin2-D₂₄₈A mutant (34%) (Fig. 3C, right panel, lane 3). These data suggest that the Kin2 KD plays an important role in modulating the Ire1-HAC1 pathway and/or operating a parallel signaling pathway.

The Kin signaling pathway operates in yeast during the ER stress response.

To determine the relative contribution of Kin1 or Kin2 kinases to the ER stress response, the kin1Δ or kin2Δ single deletion strain or the kin1Δ kin2Δ double deletion strain (see Materials and Methods) was grown under an ER stress condition (Fig. 4A). As shown in Fig. 4A, absence of either one (kin1Δ or kin2Δ strain) or both (kin1Δ kin2Δ strain) kinases did not have any notable growth difference on the SD medium, suggesting that their functions may be dispensable under normal physiological conditions. When those strains were grown on a medium containing 0.5 μg/ml tunicamycin (Fig. 4A), the kin1Δ strain grew like the WT strain (compare rows 1 and 2), whereas the kin2Δ strain grew slower than the WT strain (compare rows 1 and 3), indicating that one paralogue might complement the deletion of other paralogue and either one has a redundant function. The growth of the kin1Δ kin2Δ strain was severely impaired on the tunicamycin medium (Fig. 4A, tunicamycin, compare rows 1 and 4), suggesting that either kinase alone or both together contribute to the ER stress response.

Next, to investigate the possibility of an independent Kin-signaling pathway, we grew a WT strain and isogenic ire1Δ, hac1Δ, and kin1Δ kin2Δ strains on SD medium alone and SD medium containing 0.25 or 0.5 μg/ml of tunicamycin (Fig. 4B). As expected, deletion of any of these genes did not affect the normal growth on the medium without tunicamycin (Fig. 4B, left panel), indicating that none of these genes plays an essential role. Growth of both ire1Δ (Fig. 4B, row 2) and hac1Δ (Fig. 4B, row 3) strains was inhibited at 0.25 μg/ml of tunicamycin, whereas the growth of the kin1Δ kin2Δ strain was severely inhibited at 0.5 μg/ml of tunicamycin. This difference in tunicamycin sensitivities suggests that the Ire1-Hac1 signaling cascade primarily initiates the ER stress response, whereas the Kin kinases play an important additive role. To test this further, we performed a complementation test on the kin1Δ kin2Δ double deletion strain by expressing a kinase-dead Kin1-D₂₈₉A (31) or Kin2-D₂₄₈A mutant. The WT kinases (Fig. 4C, rows 2 and 4), but not the kinase-dead mutants (Fig. 4C, rows 3 and 5), supported yeast cell growth on the tunicamycin medium, confirming that the Kin kinase function was important for the ER stress response. Collectively, we conclude that the multiple signaling pathways, including active cooperation between the Ire1 and Kin signaling pathways, operate to evoke a threshold response during the ER stress.

To further investigate how Kin kinases coordinate with the Ire1-mediated ER stress responses, we monitored the levels of HAC1^s mRNA population and Hac1 protein and UPRE-driven lacZ expression in both WT and isogenic kin1Δ kin2Δ strains. The HAC1^s mRNA population was reduced to ∼50% compared with that in the WT strain (Fig. 5A, RT-PCR, HAC1^s, compare lanes 2 and 4). Consistently, production of Hac1 protein was reduced to ∼25% (Fig. 5B, WB, compare lanes 1 and 2). A robust 10-fold induction of Hac1-induced lacZ expression was observed in the WT strain, whereas a 5-fold induction was observed in the kin1Δ kin2Δ strain (Fig. 5C). This 50% reduction in the kin1Δ kin2Δ strain suggests that protein kinases Kin1 and Kin2 in an additive manner contribute to the ER stress response while integrating the Ire1-HAC1-mediated UPR.

FIG 5.

Both Kin1 and Kin2 modulate the Hac1-mediated ER stress response. (A and B) Both splicing and translation of HAC1 mRNA are reduced in the kin1Δ kin2Δ strain. (A) A wild-type (WT) strain or an isogenic kin1Δ kin2Δ mutant strain was grown in either the presence (+) or absence (−) of 20 mM DTT. Total RNA was isolated, and RT-PCR was used to analyze HAC1^s and HAC1^u mRNA transcripts. (B) Whole-cell extracts were subjected to Western blot (WB) analysis using an antibody against the Hac1 protein. (C) Expression of the Hac1-induced lacZ reporter is decreased in the kin1Δ kin2Δ strain. Expressions of lacZ were monitored in the presence (+) and absence (−) of DTT in the WT and kin1Δ in2Δ yeast strains, each bearing a UPRE-driven lacZ reporter gene.

Both clustering and colocalization of HAC1 mRNA with Ire1 protein are impaired in the kin1Δ kin2Δ strain.

To determine whether Kin1 and Kin2 kinases play a role in the focus formation of HAC1 mRNA or Ire1 proteins, we counted the number of HAC1 mRNA foci as well as Ire1 foci in a kin1Δ kin2Δ strain as we showed previously in Fig. 1F. A gross reduction in the number of HAC1 mRNA foci (Fig. 6A, compare panels for HAC1*GFP in WT and kin1Δ kin2Δ strains) was observed, but the number of Ire1-YFP foci was comparable (Fig. 6C, compare panels for Ire1-YFP in WT and kin1Δ kin2Δ strains) with the number of foci in the isogenic WT cells. Images were taken from approximately 20 different microscopic fields of view. Approximately 275 mRNA foci were visualized in 60 WT cells, whereas 131 foci were visualized in 60 kin1Δ kin2Δ yeast cells (Fig. 6B). The merged image shows that there was remarkably less colocalization of Ire1-YFP fusion protein with the HAC1-NRE mRNA (Fig. 6C, Merge panels in WT and kin1Δ kin2Δ strains). The 2-fold reduction in the HAC1 mRNA foci together with impaired colocalization of Ire1 protein and HAC1 mRNA suggests that the Kin kinases are positive regulators of the Ire1-HAC1 signaling pathway.

FIG 6.

Kinases Kin1 and Kin2 play a role in focus formation as well as in colocalization of HAC1 mRNA with Ire1 protein. (A and B) The number of HAC1 mRNA foci is reduced in the kin1Δ kin2Δ strain. The indicated WT and kin1Δ kin2Δ yeast strains coexpressing the HAC1-NRE mRNA and the ND-GFP2 fusion protein were grown under an ER stress condition. (B) The HAC1 mRNA foci were counted using a confocal microscope. We chose those cells where at least one focus was observed. Approximately 60 cells were randomly selected and analyzed from 20 different microscopic fields of view. (C) Analysis of colocalization of HAC1 mRNA and Ire1 protein. The yeast cells indicated in panel A were transformed with an Ire1-YFP fusion protein (21). The HAC1 mRNA (green dots) and Ire1 protein (red dots) were visualized by confocal microscopy. Images were merged to show their colocalization (indicated by arrows).

Kin2 is likely an endomembrane kinase.

Kin kinases being important components of the ER stress response, it is likely that they sense and respond to ER stress while residing in or anchoring on the ER membrane. Indeed, these kinases have been shown to precipitate with the membrane fraction (20), indicating that Kin kinases are likely to be associated with the plasma and/or organelle membranes. To determine the membrane association of the Kin2 protein, a GFP-tagged Kin2 variant (GFP-Kin2 fusion protein) was coexpressed with an authentic plasma-membrane-resident Ste2p (sterile-2 α protein)-YFP fusion protein in an ste2Δ yeast strain (23). Both GFP and YFP signals were monitored in yeast cells by confocal microscopy (see Materials and Methods). As shown in Fig. 7A, the YFP signal was evenly distributed on the plasma membrane, as expected, because Ste2 is a plasma membrane-bound protein. The GFP signal was abundantly present inside the cell as discrete dots (foci), indicating that the Kin2-GFP was primarily localized inside the cytoplasm. These observations, together with the previous observation that Kin kinases are precipitated with the membrane fraction (20), indicate that the Kin2 kinase is located within the inner leaflet of the plasma membrane.

FIG 7.

The kinase domain of Kin2 plays a role in the UPR. (A) Kin2 kinase is located in the cytoplasm: An ste2Δ yeast strain coexpressing Ste2-YFP and GFP-Kin2 fusion proteins was scanned using a confocal microscope. The Ste2-YFP (red circle) and GFP-Kin2 (green dots) fusion proteins were detected. (B) The deletion of both the predicted transmembrane (TM) domain and the kinase-associated domain 1 (KA1) does not impair Kin2 function. A kin1Δ kin2Δ yeast strain expressing GFP-Kin2 or its derivative GFP-Kin2-ΔKA1-ΔTM variant grew on the tunicamycin medium. (Lower panel) Confocal microscopy revealed that the GFP-Kin2-ΔKAI-ΔTM protein (green dots) displayed a punctate distribution. (C) The Kin2 kinase domain (Flag-Kin2^KD) is sufficient to overcome the defective UPR of the HAC1-GG_1143-1144CC cells. A hac1Δ kin1Δ pHAC1-GG_1143-44CC yeast strain expressing Flag-Kin2^KD or its derivative, Flag-Kin2^KD-D₂₄₈A, was tested for growth on the SD and tunicamycin media. (Lower panel) Whole-cell extracts from the indicated cells were subjected to Western blotting (WB) using an anti-Flag antibody. Nonspecific bands are shown as loading.

To identify the transmembrane (TM) domain in the Kin2 protein, we performed a comprehensive bioinformatics analysis (data not shown). The BLASTp (32) search against the PRODOM database (33) revealed that the N-terminal residues 1 to 89 in Kin2 possess a significant sequence similarity (E value, 3E−44) to a PDA0J0R3 family of membrane-bound protein kinases (17, 19, 20). Further analyses by the MEMPACK server (34) predicted that residues 29 to 47 of the Kin2 kinase constitute transmembrane regions, while the ΔG^pred program (35) indicated the same residues constitute an N-terminal ER membrane-binding region. The calculated free energy difference (ΔG value) of residues 29 to 47 was less than 10 kcal/mol, supporting the existence of a potential ER membrane-binding region. A relatively higher range of hydrophobicity of residues 29 to 47, calculated using the HELIQUEST software (36), further indicated the existence of a transmembrane region. Taken together, it appears that Kin2 contains an ER transmembrane region consisting of residues F₃₀ to Q₆₇.

Then we asked whether the predicted TM domain alone or together with the extreme C-terminal KA1 domain of Kin2 was involved in anchoring to the membrane because the KA1 of Kin2 orthologues mitogen-activated protein kinase (MAPK) and PAR1 has been reported to anchor the membrane lipid (37). We deleted residues constituting both the KA1 (residues K₁₁₁₆ to L₁₁₄₇) and TM (residues from F₃₀ to Q₆₇) domains, thus generating a GFP-Kin2-ΔTM-ΔKA1 variant. These deletion mutations did not affect the ability of Kin2 to support yeast cell growth on the tunicamycin medium (Fig. 7B, row 3), suggesting that these residues have redundant functions. Furthermore, confocal microscopy revealed that the GFP-Kin2-ΔTM-ΔKA1 variant displayed a punctate appearance similar to that of the GFP-Kin2-WT protein (Fig. 7B, lower panel), indicating that these domains might not play any role in membrane binding.

The Kin2 kinase domain itself is sufficient to activate the UPR.

The functional GFP-Kin2-ΔTM-ΔKA1 variant in fact suggested that the kinase domain of Kin2 was likely to be sufficient to induce the UPR. Thus, a kin1Δ kin2Δ strain was transformed with a Flag epitope-tagged kinase domain clone (Flag-Kin2^KD) encoding residues Arg-94 to Ala-526 from a galactose-inducible promoter. The transformant was then tested for growth on a medium containing galactose and a medium containing glucose plus tunicamycin. The high-level expression of Flag-Kin2^KD from the galactose-inducible promoter inhibited yeast cell growth, suggesting that the Flag-Kin2^KD promiscuously phosphorylates cellular protein substrates, causing a lethal phenotype. Interestingly we observed that the leaky expression of the Flag-Kin2^KD from the same galactose-inducible promoter exhibited a tunicamycin-resistant phenotype (data not shown). These results prompted us to test whether the Flag-Kin2^KD was able to restore the defective UPR associated with the HAC1-GG_1143-1144CC allele.

The Flag-Kin2^KD or its kinase-dead Flag-Kin2^KD-D₂₄₈A derivative was introduced into a hac1Δ kin1Δ strain expressing an HAC1-GG_1143-1144CC allele. The Flag-Kin2^KD mutation (Fig. 7C, tunicamycin, row 2), but not the Flag-Kin2^KD-D₂₄₈A mutation (row 3), fully restored the yeast cell growth on the tunicamycin medium. Western blot analysis showed that expression of the Flag-Kin2 ^KD-D₂₄₈A was similar to that of the WT allele (Fig. 7B, left panel, WB, compare lanes 2 and 3), suggesting that the kinase domain function is important for the stress response.

DISCUSSION

We show that the 3′BE in HAC1 mRNA promotes both mRNA targeting and translation. We also show that a high dose of either of two paralogue kinases Kin1 and Kin2 overcomes the defective UPR caused by a mutation in the 3′BE, suggesting that Kin kinases have acquired an important role in the yeast cell's response to ER stress.

The Kin kinase network in the unfolded protein response in yeast.

The Kin kinases in the budding yeast Saccharomyces cerevisiae share identical domain organization with the Kin kinase in the fission yeast Schizosaccharomyces pombe, the Par1 kinase (partitioning-defective kinase 1) in the worm Caenorhabditis elegans and the fruit fly Drosophila melanogaster, and the microtubule affinity-regulating kinase (MARK) in mammals (38). Each member of the Kin/Par1/MARK kinases contains a conserved kinase domain at the N terminus and a regulatory domain at the C terminus followed by a kinase-associated domain 1 (KA1). These kinases have been reported to regulate a myriad of cellular processes, including cellular exocytosis in the yeast S. cerevisiae, cell polarity in the worm C. elegans (39), and microtubule stability in the fly D. melanogaster (40). Here, we showed for the first time that the UPR was substantially reduced in the yeast strain that lacks both paralogue kinases Kin1 and Kin2 (Fig. 4). Moreover, we showed that either kinase but not its kinase-inactive derivative fully complements the UPR defect of the kin1Δ kin2Δ strain (Fig. 4C), suggesting that the intrinsic kinase activity is important for activation of the yeast UPR and the Kin kinase could potentially initiate a downstream signaling cascade by phosphorylating a substrate. The substrates for the Kin kinases are not known, although it has been reported that Kin kinases induce phosphorylation of the Sec9 protein in vivo (19). Sec9 is a component of t-SNARE (target-soluble N-ethylmaleimide-sensitive factor attachment protein receptor), which promotes the complex that controls the vesicle fusion during polarized exocytosis (19). Taken together, it appears that Kin kinases have pleiotropic involvement in many cellular processes, including in the ER stress response.

To differentiate whether Kin kinases simply modulate the Ire1 pathway or a distinct Kin signaling pathway operates in parallel with the canonical Ire1 signaling pathway, we analyzed yeast growth of the ire1Δ and kin1Δ kin2Δ strains under an ER stress condition (growth on the tunicamycin medium). The difference in relative sensitivities to tunicamycin (Fig. 4B) suggests the existence of a possible Kin signaling pathway. The obvious question is how does coactivation and cross talk between these two signaling pathways enhance the ER stress-signaling pathway? The existence of cross talk is obvious from our observations that the focus formation, splicing, and targeting of HAC1 mRNA were substantially reduced in the kin1Δ kin2Δ strain, and the molecular details of each these processes remain to be investigated.

Previously, it was reported that neither Kin1 nor Kin2 contains a sequence with the sufficient length and hydrophobicity to constitute a membrane-binding region (17–19). In contrast, Tibbetts et al. have shown that both Kin1 and Kin2 precipitate with the membrane fraction of the cellular organelles (20). They have further shown that both Kin1 and Kin2 proteins are oriented toward the cytoplasmic site in cells. We observed that the GFP-Kin2 fusion protein displayed a punctate pattern inside the cytoplasm (Fig. 7A and B), suggesting that Kin2 is likely an endomembrane kinase and is expected to influence the ER stress response directly by phosphorylating a substrate. Consistent with this notion, we observed that only the kinase domain of Kin2, not its kinase-dead derivative, was sufficient to fully restore the UPR defect of the kin1Δ kin2Δ strain (Fig. 4).

Kin kinases most likely recognize and phosphorylate a component of the HAC1-3′UTR-RNP. Thus, our next aim is to identify components of the HAC1-3′UTR-RNP, which will eventually enable us to sort out the Kin substrate and its role in coordinating the ER stress response and/or polarized exocytosis. Here we propose a working model for the Kin-induced signaling pathway during ER stress condition. The Kin kinases remains in an inactive state, with the kinase domain bound to the KA1 while anchoring with the ER membrane (Fig. 8). The active KD phosphorylates a substrate that either binds directly to or influences a factor to bind to the 3′UTR, and thus a functional RNP complex is formed. The resulting 3′UTR-RNP coordinates both targeting and translation of mRNA. Additionally, the Kin substrate might play a role in the Hac1-mediated transactivation of stress-responsive genes (Fig. 8).

FIG 8.

Model of activation of ER stress response by both Ire1 and Kin kinases. The ER-resident kinase/RNase Ire1 resides across the ER membrane. (Both the NH₃⁺-terminal luminal and cytoplasmic kinase/RNase domains are colored teal.) Under conditions of ER stress, Ire1 cleaves out an intron (orange line) in the HAC1 pre-mRNA. The resulting mature mRNA translates the Hac1 protein (Hac1p) that activates transcription of the UPR genes in the nucleus. In parallel, the protein kinase Kin1 or Kin2 (shown by blue with both the NH₃⁺-terminal and KA1 regions likely to anchor the endomembrane) is coactivated. There are two possible ways Kin1 and Kin2 functionally contribute to the HAC1-mediated UPR. First, Kin1/Kin2 binds directly to the 3′UTR element of HAC1 mRNA and promotes targeting to the ER stress signaling site, splicing, and translational repression. Second, Kin1/Kin2 phosphorylates a protein substrate (substrate shown by “S^kin,” and phosphorylation is indicated by “P”), a component of the 3′UTR-RNP complex that modulates both mRNA targeting and translation. Alternatively, the Kin1/Kin2 substrate constitutes a novel signaling cascade (denoted by “?” in a box) and transactivates the UPR genes in the nucleus. The sensing mechanism for Kin2 is unknown and is shown by “?”

Three major sensors (Ire1, PERK, and ATF6) in mammalian cells (41, 42) are known to signal the adaptive UPR. Each of these sensors carries a luminal domain that senses ER stress and a cytoplasmic domain that transmits ER stress to the respective transcriptional and translational machineries of the cell. Although the sensing mechanism of Kin1 or Kin2 is still unclear, our results suggest that the Kin signal pathway operates in yeast cells and is functionally coupled with the Ire1-Hac1 signal pathway to ensure an optimum level of ER stress response.

Coupled targeting and translational repression of HAC1 mRNA.

Here we showed that the 3′UTR, in particular 3′BE nucleotides GG_1143-1144, regulates HAC1 mRNA targeting to the ER stress signaling site, splicing, and translation. During mRNA translation, the 5′ cap and the 3′ poly(A) tail of eukaryotic mRNA are held together by a complex consisting of cap-binding protein eIF4G, the poly(A) binding protein (PABP), and other translation initiation factors; thereby, a pseudocircular mRNA is formed (43, 44). Such mRNA circularization is enhanced or disrupted by trans-acting factors bound to the recognition elements located in the 5′- or 3′UTR (24). Many of these RNA recognition elements, like the 3′BE in HAC1 mRNA, additionally modulate the subcellular translocation of mRNA (12) via an as yet unknown mechanism. We showed that the 3′BE directly modulated both splicing (Fig. 1) and translation of HAC1 mRNA (Fig. 2), indicating that the 3′UTR contributes to both splicing of a pre-mRNA and translation from a mature mRNA. Further studies are needed to find the mechanistic roles of the 3′BE-RNP in mRNA circularization and translational efficiency.