HMGA1a: sequence-specific RNA-binding factor causing sporadic Alzheimer’s disease-linked exon skipping of presenilin-2 pre-mRNA T Original XXX Aberrant Manabe Articles splicing etAuthors al. of © PS2 pre-mRNA by HMGA1a Blackwell Malden, Genes GTC © 1356-9597 Journal 2007tocompilation The USA Cells Publishing Inc 2007 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.
Takayuki Manabe1,a,b, Kenji Ohe2,a,c, Taiichi Katayama1,d, Shinsuke Matsuzaki1, Takeshi Yanagita1, Hiroaki Okuda1, Yoshio Bando3, Kazunori Imaizumi4, Raymond Reeves5, Masaya Tohyama1 and Akila Mayeda2,c,* 1
Department of Anatomy and Neuroscience, Graduate School of Medicine, Osaka University, Suita, Osaka 565-0871, Japan Department of Biochemistry and Molecular Biology, University of Miami Miller School of Medicine, Miami, FL 33136-1019, USA 3 Department of Anatomy, Asahikawa Medical College, Asahikawa, Hokkaido 078-8510, Japan 4 Division of Molecular and Cellular Biology, Department of Anatomy, Faculty of Medicine, University of Miyazaki, Kiyotake, Miyazaki 889-1692, Japan 5 Department of Biochemistry and Biophysics, School of Molecular Biosciences,Washington State University, Pullman,WA 99164-4660, USA 2
Aberrant exon 5 skipping of presenilin-2 (PS2) pre-mRNA produces a deleterious protein isoform PS2V, which is almost exclusively observed in the brains of sporadic Alzheimer’s disease patients. PS2V over-expression in vivo enhances susceptibility to various endoplasmic reticulum (ER) stresses and increases production of amyloid-β peptides.We previously purified and identified high mobility group A protein 1a (HMGA1a) as a trans-acting factor responsible for aberrant exon 5 skipping. Using heterologous pre-mRNAs, here we demonstrate that a specific HMGA1a-binding sequence in exon 5 adjacent to the 5′ splice site is necessary for HMGA1a to inactivate the 5′ splice site. An aberrant HMGA1a–U1 snRNP complex was detected on the HMGA1a-binding site adjacent to the 5′ splice site during the early splicing reaction. A competitor 2′-O-methyl RNA (2′-O-Me RNA) consisting of the HMGA1a-binding sequence markedly repressed exon 5 skipping of PS2 pre-mRNA in vitro and in vivo. Finally, HMGA1a-induced cell death under ER stress was prevented by transfection of the competitor 2′-O-Me RNA. These results provide insights into the molecular basis for PS2V-associated neurodegenerative diseases that are initiated by specific RNA binding of HMGA1a.
Introduction Alzheimer’s disease (AD) is a neurodegenerative disorder with several pathological characteristics, specifically severe Communicated by: Yoshikazu Nakamura *Correspondence: E-mail:
[email protected] a These authors contributed equally to this work and should be considered jointly as first author. b Present address: Center for Developmental Biology, RIKEN Kobe Institute, Kobe, Hyogo 650-0047, Japan. c Present address: Institute for Comprehensive Medical Science, Fujita Health University, Toyoake, Aichi 470-1192, Japan. d Present address: Department of Anatomy and Neuroscience, Hamamatsu University School of Medicine, Hamamatsu, Shizuoka 431-3192, Japan.
neuronal loss, glial proliferation, extracellular deposition of senile plaques composed of amyloid-β, and deposition of intracellular neurofibrillary tangles (reviewed in Selkoe 2001; Blennow et al. 2006). Most of the AD cases are sporadic, with more than 15 million people affected worldwide. However, the cause of sporadic AD, involved with aging and environmental risk factors, is poorly understood and its therapeutic approach is not yet well developed. The presenilin-1 (PS1) together with the highly homologous presenilin-2 (PS2) genes are known to be associated with the pathology of AD (reviewed in Katayama et al. 2004). These genes generate transmembrane proteins located in the endoplasmic reticulum (ER), which are involved in the production of amyloid-β through its γ-secretase activity and also associated with the unfolded
DOI: 10.1111/j.1365-2443.2007.01123.x © 2007 The Authors Journal compilation © 2007 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.
Genes to Cells (2007) 12, 1179–1191
1179
T Manabe et al.
protein response that is a defense mechanism against ER stress.We found that an aberrantly spliced PS2V mRNA, generated by an exon 5 skipping event of the PS2 gene transcript, is significantly expressed in the brains of sporadic AD patients compared with those of normal controls (Sato et al. 1999). This exon 5 skipping of PS2 pre-mRNA causes a frameshift and generates a premature termination codon in the downstream exon 6 (Sato et al. 1999). Expected nonsense-mediated mRNA decay of this PS2V mRNA is probably incomplete and thus results in the production of the deleterious truncated PS2V protein. The PS2V protein accumulates in intracellular inclusion bodies termed PS2V bodies, and they are specifically observed in pyramidal cells of the cerebral cortex and the hippocampus of sporadic AD patients during the early stages (Sato et al. 2001; Manabe et al. 2002). PS2V protein is also detected in the frontal lobe of some bipolar disorder and schizophrenia patients (Smith et al. 2004). In vitro experiments indicate that PS2V protein impairs the signaling pathway of the unfolded protein response which sensitizes cells to various ER stresses, and significantly stimulates the production of both amyloid β40 and β42 (Sato et al. 2001). PS2V also changes the conformation of tau protein, which is a major component of neurofibrillary tangles (Nishikawa et al. 2004). All these results suggest that aberrant splicing of the PS2 pre-mRNA is involved in neurodegenerative disorders including sporadic AD. Using human neuroblastoma SK-N-SH cells, we established an in vivo system in which this exon 5 skipping event in PS2 pre-mRNA can be induced by hypoxia (Sato et al. 1999, 2001). We could purify and identify high mobility group A protein 1a (HMGA1a; formerly HMG-I: ) that directly binds to specific sequences, i.e. GCU(G)CUACAAG, adjacent to the 5′ splice site of the PS2 pre-mRNA (Manabe et al. 2003).We showed that hypoxia induces HMGA1a overexpression in neuronal cells but not in non-neuronal cells (Manabe et al. 2003). Together with the fact that HMGA1a also binds U1 snRNP via its 70K protein, we postulated that this site-specific HMGA1a–U1 snRNP complex interferes with functional U1 snRNP binding to the adjacent 5′ splice site, and thus results in exon 5 skipping of the PS2 pre-mRNA (Manabe et al. 2003). Consistent with this hypothesis, the deletion of this HMGA1a-binding sequence was found to prevent exon 5 skipping in vivo (Higashide et al. 2004). This was a surprising finding because HMGA1 proteins, including both the HMGA1a (formerly HMG-I) and HMGA1b (formerly HMG-Y) alternatively spliced isoforms, have been well documented as non-histone 1180
Genes to Cells (2007) 12, 1179–1191
DNA-binding proteins (reviewed in Reeves 2001). Importantly, we found elevated expression of HMGA1 proteins in the hippocampus of sporadic AD patients (Manabe et al. 2003). Taken together, we proposed that sustained hypoxia in the brain could be an initial trigger to induce HMGA1-mediated production of the deleterious PS2V protein, which would be one of the risk factors leading to eventual neuronal cell death observed in brains of sporadic AD patients (Manabe et al. 2003). Interestingly, aluminum-induced oxidative stress, which is a recognized risk factor for neurodegenerative disorders including AD, also induces HMGA1a expression and subsequent PS2V production (Matsuzaki et al. 2004). Taking advantage of in vitro splicing with heterologous model pre-mRNAs, we provide evidence here that aberrant HMGA1a–U1 snRNP complex formation on the HMGA1a target sequence is responsible for exon 5 skipping of the PS2 pre-mRNA. Given the sequence-specific RNA-binding property of HMGA1a, we demonstrate that oligoribonucleotides that contain the HMGA1a target sequence competitively repress aberrant exon 5 skipping in vitro and in vivo, and eventually prevent neuronal cell death under ER stress.
Results Recombinant HMGA1a protein induces exon skipping of heterologous pre-mRNA containing the HMGA1a-binding sequence in vitro
We previously purified and identified HMGA1a protein based on its sequence-specific binding to exon 5 of the PS2 gene transcript (Manabe et al. 2003). We found that over-expression of HMGA1 proteins in vivo, induced either by hypoxia or by transient transfection of HMGA1, causes aberrant exon 5 skipping of PS2 mini-gene transcript (Manabe et al. 2003). To determine whether sequence-specific HMGA1a binding is essential for exon 5 skipping, we performed in vitro splicing with a heterologous model pre-mRNA. We used purified recombinant HMGA1a (r-HMGA1a) protein that was expressed in Escherichia coli (Reeves & Nissen 1999). The mobility of E. coli r-HMGA1a on SDS-PAGE was somewhat faster than that of native human HMGA1a (Fig. 1A), which is consistent with known post-translational modifications in the native protein (reviewed in Reeves 2001). We have reported that deletions of both 5′ and 3′ flanking introns do not affect exon 5 splicing patterns of PS2 pre-mRNA under both normal and hypoxic stress conditions in vivo (Higashide et al. 2004). We found that exon 5 of PS2 pre-mRNA is sufficient to cause its
© 2007 The Authors Journal compilation © 2007 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.
Aberrant splicing of PS2 pre-mRNA by HMGA1a
skipping by r-HMGA1a protein using a δ-crystallin mini-gene containing the entire PS2 exon 5 in vitro (data not shown). Here, we tested whether the HMGA1abinding sequence is sufficient to induce exon skipping. We previously identified an HMGA1a-binding site (two conserved tandem-repeat sequences) at the 3′ end of exon 5, which is adjacent to the 5′ splice site (Fig. 1B; Manabe et al. 2003). We therefore inserted this tandem repeat alone at the 3′ end of the middle exon of the model plasmid construct DUP51, which consists of three exons and two identical introns of the human β-globin gene (Dominski & Kole 1991). This heterologous transcript (G1-BS2-G2 pre-mRNA) exclusively generates an exon-inclusion spliced product in vitro using a HeLa cell nuclear extract (Fig. 1C, 0 pmol). However, addition of increasing amounts of r-HMGA1a to the reaction gradually induced the exon-skipping product whereas the exon-inclusion product decreased (12.5–50 pmol). We did not observe such HMGA1ainduced exon skipping with the original DUP51 premRNA, which has no HMGA1a-binding site but has a middle exon of similar length (data not shown; see Fig. 3). Our results with the heterologous pre-mRNA demonstrate that the HMGA1a-binding sequence is sufficient for HMGA1a to cause exon skipping. HMGA1a specifically binds to the HMGA1abinding site of heterologous pre-mRNA in the splicing reaction
Figure 1 Recombinant HMGA1a (r-HMGA1a) protein induces exon skipping of heterologous pre-mRNAs including the HMGA1a-binding sequence in the middle exon. (A) Purified r-HMGA1a expressed in Escherichia coli and the native HMGA1a purified from hypoxia-induced neuroblastoma (SK-N-SH) cells. These purified proteins were analyzed by 12% SDS-PAGE followed by Coomassie Blue staining. (B) The HMGA1a-binding site adjacent to the 5′ splice site identified in exon 5 of PS2 pre-RNA (Manabe et al. 2003). The consensus sequences of the tandem repeat, GCU(G)CUACAAG, are underlined (10 nt and 11 nt). (C) In vitro splicing of heterologous G1-BS2-G2 premRNA. The structure of the pre-mRNA is shown schematically at the top (black shading indicates tandem repeats of the HMGA1a-binding sequence). Indicated amounts (0–50 pmol) of r-HMGA1a were added in the splicing reactions (in 25 μL). RNA products were extracted and analyzed by denaturing PAGE and autoradiography.
To confirm whether specific HMGA1a binding is essential to cause observed exon skipping in the heterologous pre-mRNA, we mapped the HMGA1a-binding region in the pre-mRNA during in vitro splicing. Since UV cross-linking assays showed that only a single repeat of HMGA1a-binding sequence is required for HMGA1a binding (Manabe et al. 2003), here we used the heterologous G1-BS1-G2 pre-mRNA containing a single repeat of the HMGA1a-binding sequence. To map the HMGA1a-binding site, site-directed RNase H cleavage was performed in the splicing reaction followed by UV cross-linking and immunoprecipitation (Fig. 2). We designed antisense oligonucleotide that hybridizes to two sites in the G1-BS1-G2 pre-mRNA to generate three cleaved RNA fragments (A–C) by endogenous RNase H activity present in the HeLa cell nuclear extract (upper panel). The splicing reaction was carried out in the absence of ATP, so that U1 snRNP can bind more stably to the substrate. A control antibody (IgG) could not immunoprecipitate any of the RNA fragments (Fig. 2, lower panel). Anti-HMGA1a antibody immunoprecipitated only the
© 2007 The Authors Journal compilation © 2007 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.
Genes to Cells (2007) 12, 1179–1191
1181
T Manabe et al.
Figure 2 HMGA1a protein specifically binds to the HMGA1abinding site of heterologous pre-mRNA in the splicing reaction. Structure of heterologous G1-BS1-G2 pre-mRNA, two positions of antisense oligonucleotide hybridized and the generated RNA fragments (A–C; ~283, ~142 and ~201 nt) by RNase H are shown schematically at the top (the exact RNase H-cleavage sites were not determined). After in vitro splicing without ATP, RNA products were subjected to RNase H digestion with the antisense oligonucleotide followed by UV cross-linking and immunoprecipitation with control antibody (IgG), anti-HMGA1 and antiU1-70K antibodies. Immunoprecipitated RNA fragments were extracted and analyzed by denaturing PAGE and autoradiography. A portion of total RNase H digest, which includes undigested (666 nt) and partially digested (~356 and ~452 nt) products, was also loaded (Input).
‘B’ fragment that contains the HMGA1a-binding site with the adjacent 5′ splice site. This result suggests specific HMGA1a binding to the target site in an ATPindependent splicing reaction. Anti-U1-70K antibody immunoprecipitated not only the ‘A’ fragment, which contains the 5′ splice site of the upstream intron, but also the ‘B’ fragment. We previously demonstrated that HMGA1a binds to U1 snRNP via U1-70K protein, and this binding is not affected by micrococcal nuclease treatment (Manabe et al. 2003). Moreover, over-expression of free U1-70K restored functional U1 snRNP binding to the 5′ splice site (Manabe et al. 2003). Together, detection of U1 snRNP in the ‘B’ fragment could be due to protein–protein interaction between RNA-bound 1182
Genes to Cells (2007) 12, 1179–1191
Figure 3 HMGA1a protein represses constitutive splicing of pre-mRNA including the HMGA1a-binding sequence. (A,B) In vitro splicing of the constitutively spliced pre-mRNAs, PS2 BS1-E6 and β-globin E1-E2. Structures of the pre-mRNAs are shown schematically at the top (black shading indicates a single repeat of the HMGA1a-binding sequence). Indicated amounts (0–50 pmol) of purified HMGA1a protein were added in the splicing reactions (in 25 μL). RNA products were extracted and analyzed by denaturing PAGE and autoradiography. Positions of the pre-mRNAs, lariat intermediates and spliced mRNAs are schematically indicated.
HMGA1a and U1 snRNP. Our results indicate sitespecific formation of an HMGA1a–U1 snRNP complex on the HMGA1a-binding site during the early ATPindependent step of the splicing reaction. HMGA1a-mediated exon skipping is due to dysfunction at the 5′′ splice site
Next we examined whether the observed site-specific formation of an HMGA1a–U1 snRNP complex
© 2007 The Authors Journal compilation © 2007 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.
Aberrant splicing of PS2 pre-mRNA by HMGA1a
Figure 4 The 2′-O-Me RNA containing the minimal HMGA1a-binding sequence competes with the target HMGA1a-binding site during splicing and restores exon inclusion. The structure of heterologous G1BS1-G2 pre-mRNA is shown schematically at the top (black shading indicates a single repeat of the HMGA1a-binding sequence). The sequences of 2′-O-Me RNAs (11 nt) with HMGA1a-binding sequence and its mutant for control are shown (asterisks indicate the mutated ribonucleotides). Using different batches of HeLa cell nuclear extracts, two sets of in vitro splicing (in 25 μL) were performed with indicated amounts (0–100 pmol) of each 2′-O-Me RNA. RNA products were extracted and analyzed by denaturing PAGE and autoradiography.
interferes with functional U1 snRNP binding to the adjacent 5′ splice site. If this is the case, constitutive splicing should be repressed due to the inactive 5′ splice site. To test this prediction, we constructed a minimal PS2 mini-gene substrate consisting of a single repeat of the HMGA1a-binding sequence followed by shortened intron 5 and exon 6 sequences (Fig. 3A, upper panel). This transcript (BS1-E6) was spliced in vitro, and the effect of HMGA1a protein addition was examined. We observed a marked repression of constitutive splicing between exons 5 and 6 in an HMGA1a concentrationdependent manner (Fig. 3A, lower panel). Significant decrease of the lariat intermediate, rather than its accumulation, is consistent with splicing inhibition at the first step leading to 5′ splice site cleavage. We did not observe an HMGA1a-dependent splicing repression by HMGA1a in other constitutive splicing substrates such as β-globin and δ-crystallin pre-mRNAs that do not possess HMGA1a-binding sequences (Fig. 3B; data not shown). More direct examination with BS1E6 pre-mRNA including mutated HMGA1a-binding sequence is in progress. It was previously reported that the 5′ splice site recognition of an internal exon is an important determinant for exon inclusion (reviewed in Berget 1995). Taken together, we conclude that the exon 5 skipping event of the original PS2 pre-mRNA observed in vivo (Manabe et al. 2003) and exon skipping of the heterologous pre-mRNAs observed in vitro (Fig. 1C) are due to dysfunction at the 5′ splice site of the skipped exon.
Competitor oligoribonucleotides with an HMGA1abinding sequence restore exon inclusion of heterologous pre-mRNA in vitro
We found that the specific HMGA1a binds to a single repeat of the HMGA1a-binding sequence during splicing reaction, and it inactivates the adjacent 5′ splice site (Figs 2 and 3). We thus predict that the oligoribonucleotides containing only a single repeat of the HMGA1abinding sequence (Fig. 4, upper panel) can compete with the G1-BS1-G2 pre-mRNA that contains a single repeat sequence of the HMGA1a-binding site. We used modified 2′-O-methyl oligoribonucleotides (2′-O-Me RNA), which are very stable during in vitro splicing reactions performed with crude nuclear extracts (Mayeda et al. 1990). Using UV cross-linking assays, we have confirmed that 2′-O-Me RNA containing the HMGA1abinding sequence inhibits specific HMGA1a binding to the target region of PS2 pre-mRNA in vitro (data not shown; Manabe et al. 2003). Because of the shorter internal exon of the heterologous G1-BS1-G2 pre-mRNA (than that of the G1-BS2-G2 in Fig. 1C), it generates more exon-skipping products than exon-inclusion products in vitro with HeLa cell nuclear extracts that contain endogenous HMGA1a protein (Fig. 4, 0 pmol 2′-O-Me RNA). Therefore, this G1-BS1-G2 pre-mRNA (rather than G1-BS2-G2) is suitable to see possible depletion effect of endogenous HMGA1a. The addition of 11-nt 2′-O-Me RNA containing only a minimal HMGA1a-binding sequence effectively competed with the target site of G1-BS1-G2
© 2007 The Authors Journal compilation © 2007 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.
Genes to Cells (2007) 12, 1179–1191
1183
T Manabe et al.
pre-mRNA and gradually restored the exon-inclusion product in a concentration-dependent manner (Fig. 4, left panel, 25–100 pmol). In contrast, this competitive effect was not observed with the mutant 2′-O-Me RNA (left panel, 25–100 pmol). We observed a more drastic effect of the 2′-O-Me RNA with another batch of nuclear extract, suggesting different amounts of endogenous HMGA1a and/or other related factor(s) (right panel, 50–100 pmol). These results indicate that the 2′-O-Me RNA acts as a decoy to sequester endogenous HMGA1a, thereby leading to the exon inclusion. Transfected 2′′-O-Me RNA is effectively incorporated into the nucleus of neuroblastoma SK-N-SH cells
Our in vitro splicing assays demonstrate that the HMGA1abinding sequence is essential to cause HMGA1a-mediated exon skipping (Figs 1C and 3A), and thus a competitive repression of exon skipping by decoy 2′-O-Me RNA was observed in vitro (Fig. 4). Next, it is important to confirm whether decoy 2′-O-Me RNA can also competitively repress aberrant exon 5 skipping (PS2V generation) in wild-type PS2 pre-mRNA observed originally in vivo (Manabe et al. 2003). Prior to testing the effect of transfected decoy 2′-OMe RNA on splicing in vivo, we checked the efficiency of 2′-O-Me RNA incorporation into SK-N-SH cells and its subsequent stability in the cell nucleus (Fig. 5). We prepared 32P-labeled decoy 2′-O-Me RNA to monitor transfection efficiency into cells and its further delivery into the nucleus by measuring radioactivity of total cell lysates, cytosolic fractions and nuclear fractions, respectively. 2′-O-Me RNA was introduced effectively into the SK-N-SH cells and successfully delivered into the nucleus in a dose-dependent manner (Fig. 5A). Since radioactivity from degraded 2′-O-Me RNA also could contribute towards the readings in these experiments, we checked the integrity of incorporated 2′-O-Me RNA in the nuclear fraction by PAGE and autoradiography. For our purpose, the delivered decoy 2′-O-Me RNA must be stable in the nucleus until a majority of PS2 pre-mRNA is spliced in vivo, i.e. for at least 24 h after transfection (Cáceres et al. 1994; Sakashita et al. 2004). We detected substantial amounts of intact decoy 2′-O-Me RNA that has the same length as the starting 2′-O-Me RNA even 48 h after transfection (Fig. 5B).We conclude that decoy 2′-O-Me RNA incorporated into the nucleus remains sufficiently stable to perform possible repression of PS2V splicing in vivo. 1184
Genes to Cells (2007) 12, 1179–1191
Figure 5 The 2′-O-Me RNA is efficiently transfected into the SK-N-SH cells and remains stable in the nucleus. (A) Indicated final concentration (in culture medium) of 32P-labeled decoy 2′O-Me RNA (the sequence in Fig. 6A) was transiently transfected into SK-N-SH cells.The radioactive counts of cell lysates, nuclear fractions and cytosolic fractions were measured by a liquid scintillation counter. The measured counts (c.p.m./plate) and the percentages of the input radioactivity are shown in the upper and lower panels, respectively. (B) 32P-labeled decoy 2′-O-Me RNA (final 1.5 μg/mL) was transiently transfected into SK-NSH cells and the nuclear fraction was prepared 48 h after transfection. Input decoy 2′-O-Me RNA (as a marker) and the nuclear fractions were analyzed by denaturing PAGE and autoradiography.
© 2007 The Authors Journal compilation © 2007 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.
Aberrant splicing of PS2 pre-mRNA by HMGA1a
Figure 6 Transfected decoy 2′-O-Me RNA specifically represses hypoxia-induced PS2V and subsequent neuronal cell death under ER stress in vivo. (A) The sequences of decoy and control 2′-O-Me RNAs (derived from no. 1 probe; Manabe et al. 2003).The HMGA1atarget site of PS2 pre-RNA is aligned on the top. (B) Effect of transfected decoy and control 2′-O-Me RNAs on PS2 pre-mRNA splicing in vivo. SK-N-SH cells were transfected with indicated final concentrations of 2′-O-Me RNAs and exposed to normoxia or hypoxia. Total RNAs from these cells were assayed by RT-PCR and the products were analyzed by PAGE with ethidium bromide staining. Data of six independent experiments are expressed as the mean ± S.E. Asterisks indicate a significant difference from control (no addition of 2′-O-Me RNA): *P < 0.05; **P < 0.001. (C) Results of cell viability assays are shown with histogram (expressed as a percent of living cells in control; column 1). SK-N-SH cells were transfected with HMGA1a-expression plasmid together with either decoy or control 2′-O-Me RNA (final concentration in μg/mL) followed by treatment for 10 h with tunicamycin (TM; final concentration in 1 μg/mL). The MTS assay was performed after transfection for 24 h. Indicated values of four independent experiments are expressed as the mean ± S.E. Asterisks indicate a significant difference from control (no addition of 2′-O-Me RNA): *P < 0.05; **P < 0.01.
Decoy 2′′-O-Me RNA represses hypoxia-induced PS2V in vivo but does not affect the DNA-binding activity of HMGA1a
We examined whether transient transfection of decoy 2′-O-Me RNA (Fig. 6A) can repress the expression of hypoxia-induced endogenous PS2V in SK-N-SH cells. We found that the generation of PS2V mRNA was dramatically reduced by the transfection of decoy 2′-OMe RNA, and the reduction was dependent on the dose of the 2′-O-Me RNA (Fig. 6B, lower panel). In contrast, the control 2′-O-Me RNA (Fig. 6A), which has no HMGA1a-binding sequence, has no effect on the generation of PS2V (Fig. 6B, upper panel). We previously confirmed that the PS2V protein product was indeed induced under hypoxia and the level of PS2V mRNA is proportionally reflected in the level of PS2V protein (Manabe et al. 2003). These results demonstrate that the transfected decoy 2′-O-Me RNA specifically
inhibits hypoxia-induced exon 5 skipping and thus represses production of PS2V in vivo. Our current in vitro and in vivo results clearly indicate a novel function for HMGA1a protein as an RNA-binding protein. Since HMGA1 proteins are well-known DNAbinding proteins that regulate gene transcription by forming specific complexes with other transcription factors on promoter regions (reviewed in Reeves 2001), it was important for us to test whether there are any deleterious effects of excess decoy 2′-O-Me RNA on the DNA-binding activity of HMGA1a, which might be physiologically important. Numerous studies have established that HMGA1 proteins preferentially bind to the minor groove of ATrich DNA sequence (reviewed in Reeves 2001), but few have systematically analyzed DNA substrates for the high-affinity binding consensus sequence. We therefore determined the optimal DNA-binding sequences for HMGA1a by using an in vitro DNA selection assay,
© 2007 The Authors Journal compilation © 2007 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.
Genes to Cells (2007) 12, 1179–1191
1185
T Manabe et al.
and obtained an apparent consensus octamer sequence (double stranded), (G/A)G(T/A)(A/T)(A/T)TTT (T. Manabe and M. Tohyama, unpublished data). We then examined the effects of our decoy 2′-O-Me RNA on this sequence-specific DNA binding of HMGA1a. Employing electrophoretic mobility shift analyses with nuclear extracts of HMGA1a-transfected SK-N-SH cells, we observed strong binding of HMGA1a to this consensus DNA sequence even in the presence of 1000-fold molar excess of decoy 2′-O-Me RNA (data not shown). These results confirmed that the inhibitory effects of decoy 2′-O-Me RNA are specific to HMGA1a binding to the target RNA (Fig. 1B), and do not affect its binding to the target DNA (T. Manabe and M. Tohyama, unpublished data). Decoy 2′′-O-Me RNA prevents SK-N-SH cell death under ER stress
Previous in vitro evidence showed that PS2V-expressing cells become sensitive to several ER stresses (Sato et al. 1999, 2001). Since hypoxia-induced PS2V generation is caused by HMGA1a over-expression (Manabe et al. 2003), we expect that the decoy 2′-O-Me RNA sequestering HMGA1a away from the target HMGA1abinding site would prevent eventual cell death under ER stress caused by tunicamycin. To test this hypothesis, here we used a cell viability assay system with HMGA1a-transfected SK-N-SH cells. We have checked that PS2V is significantly expressed in HMGA1a-transfected SK-N-SH cells (data not shown; Manabe et al. 2003). Transfection of HMGA1a followed by tunicamycin treatment resulted in markedly increased levels of cell death, whereas no significant cell death was detected either with HMGA1a transfection alone or when cells were treated with tunicamycin without HMGA1a transfection (Fig. 6C, columns 1–4; Manabe et al. 2003). Remarkably, transfection of the same decoy 2′-O-Me RNA used for in vivo splicing assays (Fig. 6A) significantly reduced the level of induced cell death in a dose-dependent manner, whereas the control 2′-O-Me RNA did not (Fig. 6C, columns 4–10). We conclude that the decoy 2′-O-Me RNA, competing with the target HMGA1a-binding site and repressing PS2V generation, is also effective in preventing PS2V-induced neuronal cell death under ER stress.
Discussion A mistake in essential pre-mRNA splicing process often results in serious diseases. Indeed it is estimated that ~15% of all point mutations in human genetic diseases 1186
Genes to Cells (2007) 12, 1179–1191
causes a pre-mRNA splicing defect (reviewed in Krawczak et al. 1992). However, we do not know much about trans-acting factors that cause changes in the splicing pattern with pathological consequences. We purified HMGA1a protein based on its sequencespecific RNA-binding activity and identified it as an inducible trans-acting factor causing aberrant exon 5 skipping of PS2 pre-mRNA, which produces a deleterious PS2V protein observed in the brains of sporadic AD patients (Manabe et al. 2003). Our in vitro splicing data here provide evidence to support the proposed working model: the site-specific HMGA1a–U1 snRNP complex formation interferes with functional U1 snRNP binding to the adjacent 5′ splice site (Fig. 7A). The results of in vitro and in vivo competition assays with decoy oligoribonucleotides are consistent with our proposed pathogenic pathway of sporadic AD in the absence of any mutation in the PS2 gene (Fig. 7A and B). The molecular mechanism of HMGA1a-mediated aberrant exon skipping
To elucidate the molecular mechanism underlying specific aberrant splicing observed in PS2 pre-mRNA, we reconstituted the exon-skipping event in vitro with heterologous pre-mRNAs. In vitro splicing results demonstrate that only a minimal HMGA1a-binding sequence is essential to cause HMGA1a-dependent exon skipping. However, our results do not rule out the possibility of an HMGA1a-independent mechanism that might be involved also in the exon 5 skipping of native PS2 premRNA. It was recently reported that an upstream stem– loop structure in exon 5 plays a role in exon 5 skipping under hypoxic stress conditions, although either the trans-acting factor(s) involved in, or the molecular mechanism responsible for, exon 5 skipping are not known yet (Higashide et al. 2004). It is reasonable to speculate that including such a stem–loop as an additional cis-element might enhance modest efficiency of exon skipping in our heterologous substrates that contain only a minimal HMGA1a-binding sequence. In this study, we focused on the HMGA1a-dependent mechanism of the aberrant splicing.We demonstrate that HMGA1a-induced exon skipping is due to inhibition of the 5′ splice site event of the corresponding exon. Our key observation is that an aberrant HMGA1a–U1 snRNP complex is formed specifically on the HMGA1atarget site of the heterologous substrate.We found previously that HMGA1a and U1 snRNP bind through the U1-70K protein (but not through RNA), and that exon 5 skipping of PS2 pre-mRNA was repressed by overexpression of U1-70K protein (Manabe et al. 2003).
© 2007 The Authors Journal compilation © 2007 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.
Aberrant splicing of PS2 pre-mRNA by HMGA1a
These results together strongly support our proposed 5′ splice site repression model (Fig. 7A). It is evident that a large excess of HMGA1a protein, which is induced under pathological conditions, is an initial condition that causes possible aberrant splicing. We previously showed that aberrant PS2 exon 5 skipping occurs in neuronal cells only if HMGA1a is overexpressed either by hypoxia or by transfection of HMGA1a (Manabe et al. 2003). A low level of HMGA1a in neuronal cells, under physiological conditions, does not cause the HMGA1a-binding to the target sequence and subsequent aberrant exon 5 skipping. PS2V proteins were not detected in age-matched control brains; yet there is a basal level of HMGA1a expression (Manabe et al. 2003). Deleterious HMGA1a–U1 snRNP complex on the HMGA1a-binding site may be formed only if the concentration of HMGA1a exceeds a certain threshold. Since HeLa cells (derived from cervical carcinomas) express high levels of endogenous HMGA1a protein (Bandiera et al. 1998; Reeves et al. 2001), we could detect HMGA1a–U1 snRNP complex on the HMGA1a-binding site in vitro with a HeLa cell nuclear extract. We estimated the amount of HMGA1a in HeLa cells by immunoblotting as 1.0–1.5 × 108 molecules/cell, which is over 2000-fold more abundant than in normal (non-transformed) epithelial cells (Y. Li, J. E. Adair and R. Reeves, unpublished data; Reeves et al. 2001). The estimated amount of U1 snRNP (snRNA) in HeLa cells is ~1.0 × 106 molecules/cell (Weinberg & Penman 1968), which is 100- to 150-fold less than that of HMGA1a. Therefore, a simple concentrationdependent competition model, or the HMGA1a positionindependent U1 snRNP sequestering model, cannot explain why a large excess of HMGA1a protein is necessary to inactivate the 5′ splice site by aberrant HMGA1a–U1 snRNP complex on PS2 pre-mRNA. An important clue for answering this question is the fact that the HMGA1a– U1 snRNP complex is not formed nearby the 5′ splice site without an adjacent HMGA1a-binding sequence even with abundant HMGA1a protein (see fragment A Figure 7 (A) A molecular model for the mechanism of hypoxiainduced PS2V generation and its repression by decoy 2′-O-Me RNA containing the HMGA1a-binding sequence.The following lines of evidence support the model: (i) hypoxia induces overexpression of HMGA1a protein that subsequently leads to the generation of PS2V (Manabe et al. 2003). (ii) HMGA1a binds to the target site adjacent to the 5′ splice site and it recruits U1 snRNP through U1-70K protein (Fig. 2; Manabe et al. 2003). (iii) The aberrant U1 snRNP–HMGA1a complex causes inactivation of the 5′ spice site (Fig. 3) and subsequent exon 5 skipping in vitro and in vivo (Fig. 1; Manabe et al. 2003). (B) A proposed HMGA1a-mediated process in the development of sporadic AD. In the brain, sustained hypoxia (e.g. microinfarction that closely
correlates with a history of AD patients) and oxidative stress (e.g. chronic exposure to aluminum that is one of the known risk factors of AD) could be potential triggers to induce HMGA1amediated production of deleterious PS2V protein (Manabe et al. 2003; Matsuzaki et al. 2004). In vitro experiments suggested that over-expressed PS2V protein impairs the signaling pathway of the unfolded protein response and extracellular PS2V significantly enhances the production of amyloid-β (Sato et al. 1999, 2001). Here we showed that decoy 2′-O-Me RNA, which specifically represses aberrant exon 5 skipping of PS2 pre-mRNA, could prevent eventual neuronal cell death under ER stress in vivo.
© 2007 The Authors Journal compilation © 2007 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.
Genes to Cells (2007) 12, 1179–1191
1187
T Manabe et al.
in Fig. 2). It is likely that HMGA1a binding to the HMGA1-binding adjacent to the 5′ splice site is a prerequisite to make a stable and inhibitory HMGA1– U1 snRNP complex via the U1-70K protein. Even though a large excess of HMGA1a can bind to free U1 snRNP as we observed (Manabe et al. 2003), this binding may be unstable, and unbound U1 snRNP can still properly accesses the 5′ splice site (without an adjacent HMGA1a-binding site). This argument is well supported by the fact that HMGA1a protein has not been detected so far by mass spectrometry analyses of splicing complexes assembled with HeLa cell nuclear extracts in vitro (reviewed in Jurica & Moore 2003).The complex formation between HMGA1a and free U1 snRNP might not be stable enough to exist persistently in assembled spliceosomes in HeLa nuclear extracts. This idea well explains the fact that HMGA1a is unable to cause exon skipping indiscriminately, while it is highly dependent on the presence of a specific HMGA1abinding site adjacent to the 5′ splice site. Intriguingly, our postulated model is analogous to the following two distinct repression mechanisms observed in pre-mRNA processing, although there is no significant homology either in these proteins or these target RNA sequences. (i) In Drosophila, PSI protein (AB motif ) specifically binds U1 snRNP via its 70K protein (Prorich sequence), recruits U1 snRNP to the upstream pseudo 5′ splice site, inactivates the 5′ splice site and causes retention of the downstream intron of P-element pre-mRNA (Labourier et al. 2001; Ignjatovic et al. 2005). (ii) In bovine papillomavirus, poly (A) polymerase specifically binds U1 snRNP via its 70K protein, recruits U1 snRNP to the 5′ splice site sequence located downstream of the polyadenylation signal and inhibits polyadenylation of the late gene expression (Gunderson et al. 1998). In the PS2 pre-mRNA, we cannot rule out the possibility that there is U1 snRNA interaction with possible 5′ splice site-like sequence (or pseudo 5′ splice site) located upstream of the HMGA1a-binding site. However, possible complementary sequences to U1 snRNA nearby the HMGA1a-binding site in the original PS2 pre-mRNA are not highly conserved as that of Drosophila P-element pre-mRNA. We found even less complementary sequences in our heterologous pre-mRNAs: E1-BS1(2)-E2 and BS1-E6. It is likely that a prerequisite HMGA1a–U1 snRNP complex on the HMGA1abinding site might facilitate U1 snRNA annealing to the possible weak 5′ splice site-like sequence, if any. Our study provides a reasonable molecular mechanism to explain the aberrant exon 5 skipping event of the PS2 gene transcript, in which mutations are not the primary cause of a neurodegenerative disorder such as sporadic 1188
Genes to Cells (2007) 12, 1179–1191
AD. A responsible RNA-binding factor was unexpectedly identified as HMGA1a, which is a previously well-characterized DNA-binding protein (Manabe et al. 2003). It is of interest to determine the distinct HMGA1a domains responsible for target RNA and DNA sequences.The fact that hypoxia-induced HMGA1a protein accumulates in typical nuclear speckles with the endogenous authentic splicing factor SC35 clearly indicates a role of HMGA1a as a potential splicing regulator (Manabe et al. 2003). Based on our model (Fig. 7A), we expect that other disease-associated genes that possess an HMGA1a-binding sequence near a 5′ splice site are subject to splicing defects in response to the over-expression of HMGA1a—which is induced by various extra- and intra-cellular signals; e.g., external stress, tumor factors and cell growth/differentiation factors (reviewed in Reeves 2001). Given the fact that HMGA1a is known as an oncogene (reviewed in Reeves 2001; Faustino & Cooper 2003), its potential role in changes of alternative splicing patterns associated with neoplasia and metastasis is not surprising. Pathological significance and medical implication of PS2V repression by decoy oligoribonucleotides
The nature of sequence-specific RNA binding of HMGA1a immediately pointed to a possible competitive inhibition approach using decoy oligoribonucleotides containing the target sequence. Here we demonstrated that competitor 2′-O-Me RNA containing the HMGA1a-binding sequence significantly inhibits exon skipping in vitro, represses PS2V production in vivo and eventually prevents neuronal cell death under ER stress in vivo.These results indicate that specific HMGA1a binding to the target site is an initial step leading to eventual PS2V-associated neuronal cell death—supporting our proposed model (Fig. 7B). To examine the pathological significance of our model, an appropriate animal model system is critical. We have already found that PS2V transgenic mice die at early embryonic stage, indicating that the PS2V protein is highly deleterious (N. Sato and M.Tohyama, unpublished observation). This observation is consistent with the fact that the aberrantly spliced isoform PS2V eventually causes neuronal cell death even though its expression is at low levels (reviewed in Katayama et al. 2004). We are currently preparing conditional transgenic mice in which PS2V and HMGA1a expression can be induced at the adult stage. Results obtained with such transgenic mice will directly examine the validity of our proposed molecular model and, consequently, determine its potential medical significance.
© 2007 The Authors Journal compilation © 2007 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.
Aberrant splicing of PS2 pre-mRNA by HMGA1a
The fact that the decoy 2′-O-Me RNA is capable of preventing neuronal cell death triggered by the expression of deleterious PS2V protein implies the potential therapeutic application of such decoy oligoribonucleotides. Although our study is preliminary to indicate the practicality of the clinical approach, we can point out important advantages based on its unique mechanism of inhibition. In contrast to gene therapies with short interfering RNA (siRNA), ribozyme and antisense oligoribonucleotides that are prone to fully inhibiting the function of target protein (reviewed in Khan & Lal 2003; Wood et al. 2003; Hannon & Rossi 2004; Jaeger & Banks 2004), our strategy with decoy oligoribonucleotides targets only the specific HMGA1a-binding sequence, and does not prevent either the expression of HMGA1 protein or the HMGA1a-binding activity to DNA, which are potentially important for the regulation of transcription (reviewed in Reeves 2001). Therefore, the side effects using the decoy oligoribonucleotides should be minimal, which is an important consideration in the therapeutic application. Recently, we have succeeded in detecting PS2V protein in cerebrospinal fluids obtained from sporadic AD patients, but not in those from controls, by enzyme-linked immunosorbent assay using specific antibodies against PS2V protein (T. Katayama and M.Tohyama, unpublished observation). The screening of sporadic AD patients at various stages revealed that PS2V protein appears even at the early stages (T. Katayama and M.Tohyama, unpublished observation).We expect that PS2V protein will be a reliable diagnostic marker of sporadic AD detectable at early stages of the disease.
Experimental procedures Preparation of recombinant and native HMGA1a proteins The full-length cDNA of HMGA1a from pcDNA3(+)-HMGA1a plasmid (Manabe et al. 2003) was amplified by PCR with flanking NdeI and BamHI restriction sites, and the amplified DNA fragment was subcloned into the corresponding restriction sites of pSBETc vector (Schenk et al. 1995) to generate pSBETc-HMGA1a plasmid. The pSBETc-HMGA1a was transformed into E. coli BL21(DE3)/plysS (EMD Biosciences, La Jolla, CA), grown in 2-L culture, and the expression was induced with 0.7 mm IPTG. Purification of r-HMGA1a was performed essentially as described previously (Reeves & Nissen 1999). Briefly, the bacterial pellet was washed and extracted by 50 mL of 5% perchloric acid and 0.1% Triton X-100. The extracted HMGA1a was precipitated by six volumes of the alcohol mixture (ethanol : methanol : acetone : water = 50 : 25 : 25 : 1).The pellet was washed with the same mixture and resuspended in buffer A [20 mm HEPES–NaOH (pH 8.0),
0.2 mm EDTA, 6 m urea, 5% (v/v) glycerol and 1 mm DTT] with 0.5 mm PMSF and dialyzed in the same buffer. The dialysate was loaded on a Mono Q column (1 mL; GE Healthcare BioSciences, Piscataway, NJ), washed with buffer A and eluted with a 40-mL linear gradient from 0 to 0.6 M NaCl in buffer A. HMGA1a eluted in a fraction with 0.12–0.14 m NaCl. The final purified fraction was dialyzed against buffer B [20 mm HEPES– NaOH (pH 8.0), 0.1 m KCl, 0.2 mm EDTA, 5% (v/v) glycerol, 1 mm DTT and 0.5 mm PMSF]. Preparation and purification of native HMGA1a protein from hypoxia-induced SK-N-SH cells were described previously (Manabe et al. 2003).
In vitro splicing assays To construct β-globin-base heterologous plasmids (for E1-BS2E2 and E1-BS1-E2 pre-mRNAs), overlap-extension PCRs were performed on the duplicated human β-globin plasmid DUP51 (Dominski & Kole 1991) using primers including XbaI-linkers and double/single-repeat sequences of the HMGA1a-binding site. The amplified fragment was then cleaved with XbaI and self-ligated. The verified replaced sequences of the middle exons of E1-BS2-E2 and E1-BS1-E2 are 5′-GCTGCTGGGCAAGTCTAGACGTGCTCTACAAGTACCGCTGCTACAAG-3′ and 5′-GCTGCTGGGCAAGTCTAGACGTAGTACCGCTGCTACAAG-3′, respectively (HMGA1a-binding sequences are underlined; see Fig. 1B). To construct partial PS2 plasmid with a single intron (for BS1E6 pre-mRNA), BamHI fragment (spanning 69 bp of intron 5 and 68 bp of exon 6) was excised from pcDNA3(–)-No. 0 plasmid (Manabe et al. 2003) and then subcloned into BamHI site of pBluescript II SK vector (Stratagene, La Jolla, CA). The excised EcoRI fragment was then subcloned into EcoRI site of pcDNA3(+)-No. 11 (including HMGA1a-binding site; Manabe et al. 2003), which leaves 18 bp of the pBluescript vector sequence in the intron 5. These plasmids were linearized by appropriate restriction enzymes (BamHI for E1-BS1(2)-E2 and XbaI for BS1-E6 plasmids) and used as template DNAs for in vitro transcription. GpppG-capped 32P-labeled pre-mRNA substrates were prepared by in vitro transcription with SP6 RNA polymerase as described (Mayeda & Krainer 1999b). For BS1-E6 pre-mRNA, the fulllength transcript was purified by preparative electrophoresis on a 5.5% polyacrylamide/7 m urea gel for the in vitro splicing assays as described (Mayeda & Krainer 1999b). HeLa cells were purchased in the medium (National Cell Culture Center, Minneapolis, MN) and incubated for 2–4 h at 37 °C prior to the preparation of nuclear extracts (Mayeda & Krainer 1999a). In vitro splicing reactions in 25 μL were performed as described (Mayeda & Krainer 1999b) with the following modifications. To optimize exon-skipping activity in our model heterologous pre-mRNA, E1-BS2-E2 pre-mRNA (20 fmol) was mixed with r-HMGA1 protein (indicated amounts), adjusted to 30 mm KCl concentration and pre-incubated at room temperature for 5 min. Then HeLa cell nuclear extracts (8.5 μL) together with other splicing reagents were added and incubated at 30 °C for 3–4 h
© 2007 The Authors Journal compilation © 2007 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.
Genes to Cells (2007) 12, 1179–1191
1189
T Manabe et al. (E1-BS2-E2 pre-mRNA) or for 2 h (other pre-mRNAs). For the competitive splicing assays, the indicated amount of 2′-O-Me RNA (see Fig. 4 for the sequences; purchased from IBA, Göttingen, Germany) were added with E1-BS1-E2 pre-mRNA (20 fmol) and incubated under standard splicing conditions at 30 °C for 2 h. RNA products were analyzed by electrophoresis on a 5.5% polyacrylamide/7 m urea gel and autoradiography as described (Mayeda & Krainer 1999b).
HMGA1a mapping by RNase H digestion and immunoprecipitation In vitro splicing reactions were set up as described (Mayeda & Krainer 1999b) except for the exclusion of ATP to stabilize U1 snRNP binding. After in vitro splicing for 5 min at 30 °C, the reaction mixtures were irradiated with 254 nm UV light at 0.36 J using a UV cross-linker (Model 1800, Stratagene, La Jolla, CA). The reaction mixtures were put on ice and antisense DNA oligonucleotide 5′-ACGTCTAGACTTGCCCAGCAGCCTAAG-3′ (Operon Biotechnologies, Huntsville, AL) was added, and incubated for another 5 min at 30 °C. Endogenous RNase H activity in HeLa nuclear extract cleaved E1-BS1-E2 pre-mRNA (which contains two target sites) into three fragments (see Fig. 2). Anti-HMGA1 (anti-HMG-I/Y) rabbit antibody, anti-U1 snRNP70K goat antibody, and normal rabbit IgG (Santa Cruz Biotechnology, Santa Cruz, CA) were microdialyzed against buffer B (see above) prior to using them in the immunoprecipitation.The reaction mixtures diluted with 250 μL of IP buffer (Hanamura et al. 1998) were rocked with these antibodies for 2 h and for a further 1 h with the addition of protein A-Sepharose (GE Healthcare Bio-Sciences). Pellets were washed 3 times with IP buffer. The RNA in the final pellets was extracted with phenol and chloroform, ethanol precipitated with 1 μg glycogen and analyzed by electrophoresis on a 10% polyacrylamide/7 m urea gel as described (Kim et al. 2001).
Transient transfection of SK-N-SH cells and preparation of the cell fractions/extracts SK-N-SH cell culture under hypoxic stimulation and transient transfection were performed essentially as described (Sato et al. 1999). Transient transfections of HMGA1a and 2′-O-Me RNA were carried out using Lipofectamine and Oligofectamine (Invitrogen, Carlsbad, CA), respectively. Nuclear and cytosolic fractions from cultured SK-N-SH cells were prepared as described (Manabe et al. 2000). Nuclear extracts were prepared from the above nuclear fractions as described (Yoneda et al. 1999).
In vivo splicing assays In vivo splicing assays with SK-N-SH cells in normoxia or hypoxia condition are performed as described (Manabe et al. 2003). Decoy and control 2′-O-Me RNAs (see Fig. 6A for the sequences) were purchased (Japan Bio Services, Saitama, Japan). Preparation of total RNA from SK-N-SH cells and the following RT-PCR analyses were performed as described (Sato et al. 1999; Manabe et al. 2003).
1190
Genes to Cells (2007) 12, 1179–1191
Analysis of cell viability SK-N-SH cells were transfected with 0.5 μg/mL of pcDNA3(+)HMGA1a plasmid (Manabe et al. 2003) and 0.5–1.5 μg/mL of 2′O-Me RNA (decoy and control; see Fig. 6A for the sequences), followed by treatment with 1 μg/mL tunicamycin for 10 h. After transfection for 24 h, the cell viability was analyzed by MTS assay according to the manufacturer’s protocol (CellTiter 96 Aqueous One Solution Cell Proliferation assay kit; Promega, Madison, WI).
Acknowledgements We greatly thank A. Arakawa for technical assistance; R.K. Fujimura, C. Jain, T. Venkataraman, R.H. Warren and J.F. Cáceres for critical reading of the manuscript; R. Lührman, D.C. Rio, H. Suzuki, T. Tani and M. Ohno for helpful suggestions; K.E. Rudd, M.L. King and A.R. Krainer for encouragement; M.P. Deutscher for generous support. T.M., T.K. and M.T. were supported by the Japan Science and Technology Agency ( JST) as a research project in the Innovation Plaza Osaka. T.K. was supported by Grantin-Aid for Scientific Research from the Ministry of Health, Labor and Welfare of Japan. K.O. and A.M. were supported by the Florida Biomedical Research Program Grant (BM031) from FDH and the Developmental Research Grants from the Sylvester Comprehensive Cancer Center. A.M. was a research member of the Sylvester Comprehensive Cancer Center.
References Bandiera, A., Bonifacio, D., Manfioletti, G., Mantovani, F., Rustighi, A., Zanconati, F., Fusco, A., Di Bonito, L. & Giancotti, V. (1998) Expression of HMGI(Y) proteins in squamous intraepithelial and invasive lesions of the uterine cervix. Cancer Res. 58, 426–431. Berget, S.M. (1995) Exon recognition in vertebrate splicing. J. Biol. Chem. 270, 2411–2414. Blennow, K., de Leon, M.J. & Zetterberg, H. (2006) Alzheimer’s disease. Lancet 368, 387–403. Cáceres, J.F., Stamm, S., Helfman, D.M. & Krainer, A.R. (1994) Regulation of alternative splicing in vivo by overexpression of antagonistic splicing factors. Science 265, 1706–1709. Dominski, Z. & Kole, R. (1991) Selection of splice sites in pre-mRNAs with short internal exons. Mol. Cell. Biol. 11, 6075–6083. Faustino, N.A. & Cooper, T.A. (2003) Pre-mRNA splicing and human disease. Genes Dev. 17, 419–437. Gunderson, S.I., Polycarpou-Schwarz, M. & Mattaj, I.W. (1998) U1 snRNP inhibits pre-mRNA polyadenylation through a direct interaction between U1 70K and poly(A) polymerase. Mol. Cell 1, 255–264. Hanamura, A., Cáceres, J.F., Mayeda, A., Franza, B.R. Jr. & Krainer, A.R. (1998) Regulated tissue-specific expression of antagonistic pre-mRNA splicing factors. RNA 4, 430–444. Hannon, G.J. & Rossi, J.J. (2004) Unlocking the potential of the human genome with RNA interference. Nature 431, 371–378. Higashide, S., Morikawa, K., Okumura, M., Kondo, S., Ogata,
© 2007 The Authors Journal compilation © 2007 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.
Aberrant splicing of PS2 pre-mRNA by HMGA1a M., Murakami, T., Yamashita, A., Kanemoto, S., Manabe, T. & Imaizumi, K. (2004) Identification of regulatory cis-acting elements for alternative splicing of presenilin 2 exon 5 under hypoxic stress conditions. J. Neurochem. 91, 1191–1198. Ignjatovic, T., Yang, J.C., Butler, J., Neuhaus, D. & Nagai, K. (2005) Structural basis of the interaction between P-element somatic inhibitor and U1-70k essential for the alternative splicing of P-element transposase. J. Mol. Biol. 351, 52 –65. Jaeger, L.B. & Banks, W.A. (2004) Antisense therapeutics and the treatment of CNS disease. Front. Biosci. 9, 1720 –1727. Jurica, M.S. & Moore, M.J. (2003) Pre-mRNA splicing: awash in a sea of proteins. Mol. Cell 12, 5 –14. Katayama, T., Imaizumi, K., Manabe, T., Hitomi, J., Kudo, T. & Tohyama, M. (2004) Induction of neuronal death by ER stress in Alzheimer’s disease. J. Chem. Neuroanat. 28, 67 –78. Khan, A.U. & Lal, S.K. (2003) Ribozymes: a modern tool in medicine. J. Biomed. Sci. 10, 457 – 467. Kim, V.N., Kataoka, N. & Dreyfuss, G. (2001) Role of the nonsensemediated decay factor hUpf 3 in the splicing-dependent exon– exon junction complex. Science 293, 1832 –1836. Krawczak, M., Reiss, J. & Cooper, D.N. (1992) The mutational spectrum of single base-pair substitutions in mRNA splice junctions of human genes: causes and consequences. Hum. Genet. 90, 41–54. Labourier, E., Adams, M.D. & Rio, D.C. (2001) Modulation of P-element pre-mRNA splicing by a direct interaction between PSI and U1 snRNP 70K protein. Mol. Cell 8, 363 – 373. Manabe, T., Kitayama, T., Ogita, K. & Yoneda, Y. (2000) Differential expression and phosphorylation of particular fos family members by kainate in nuclear and cytosolic fractions of murine hippocampus. Neuroscience 100, 453 – 463. Manabe, T., Katayama, T., Sato, N., Gomi, F., Hitomi, J., Yanagita, T., Kudo, T., Honda, A., Mori, Y., Matsuzaki, S., Imaizumi, K., Mayeda, A. & Tohyama, M. (2003) Induced HMGA1a expression causes aberrant splicing of Presenilin-2 pre-mRNA in sporadic Alzheimer’s disease. Cell Death Differ. 10, 698 –708. Manabe, T., Katayama, T., Sato, N., Kudo, T., Matsuzaki, S., Imaizumi, K. & Tohyama, M. (2002) The cytosolic inclusion bodies that consist of splice variants that lack exon 5 of the presenilin-2 gene differ obviously from Hirano bodies observed in the brain from sporadic cases of Alzheimer’s disease patients. Neurosci. Lett. 328, 198 – 200. Matsuzaki, S., Manabe, T., Katayama, T., Nishikawa, A., Yanagita, T., Okuda, H., Yasuda, Y., Miyata, S., Meshitsuka, S. & Tohyama, M. (2004) Metals accelerate production of the aberrant splicing isoform of the presenilin-2. J. Neurochem. 88, 1345–1351. Mayeda, A., Hayase, Y., Inoue, H., Ohtsuka, E. & Ohshima, Y. (1990) Surveying cis-acting sequences of pre-mRNA by adding antisense 2′-O-methyl oligoribonucleotides to a splicing reaction. J. Biochem. 108, 399 – 405. Mayeda, A. & Krainer, A.R. (1999a) Preparation of HeLa cell nuclear and cytosolic S100 extracts for in vitro splicing. Methods Mol. Biol. 118, 309–314.
Mayeda, A. & Krainer, A.R. (1999b) Mammalian in vitro splicing assays. Methods Mol. Biol. 118, 315–321. Nishikawa, A., Manabe, T., Katayama, T., Kudo, T., Matsuzaki, S., Yanagita, T., Okuda, H., Bando, Y. & Tohyama, M. (2004) Novel function of PS2V: change in conformation of tau proteins. Biochem. Biophys. Res. Commun. 318, 435–438. Reeves, R. (2001) Molecular biology of HMGA proteins: hubs of nuclear function. Gene 277, 63–81. Reeves, R., Edberg, D.D. & Li, Y. (2001) Architectural transcription factor HMGI( Y) promotes tumor progression and mesenchymal transition of human epithelial cells. Mol. Cell. Biol. 21, 575–594. Reeves, R. & Nissen, M.S. (1999) Purification and assays for high mobility group HMG-I(Y) protein function. Methods Enzymol. 304, 155–188. Sakashita, E., Tatsumi, S., Werner, D., Endo, H. & Mayeda, A. (2004) Human RNPS1 and its associated factors: a versatile alternative pre-mRNA splicing regulator in vivo. Mol. Cell. Biol. 24, 1174–1187. Sato, N., Hori, O., Yamaguchi, A., Lambert, J.C., ChartierHarlin, M.C., Robinson, P.A., Delacourte, A., Schmidt, A.M., Furuyama, T., Imaizumi, K., Tohyama, M. & Takagi, T. (1999) A novel presenilin-2 splice variant in human Alzheimer’s disease brain tissue. J. Neurochem. 72, 2498–2505. Sato, N., Imaizumi, K., Manabe, T., et al. (2001) Increased production of β-amyloid and vulnerability to endoplasmic reticulum stress by an aberrant spliced form of presenilin 2. J. Biol. Chem. 276, 2108–2114. Schenk, P.M., Baumann, S., Mattes, R. & Steinbiss, H.-H. (1995) Improved high-level expression system for eukaryotic genes in Escherichia coli using T7 RNA polymerase and rare ArgtRNAs. Biotechniques 19, 196–200. Selkoe, D.J. (2001) Alzheimer’s disease results from the cerebral accumulation and cytotoxicity of amyloid β-protein. J.Alzheimers Dis. 3, 75–80. Smith, M.J., Sharples, R.A., Evin, G., McLean, C.A., Dean, B., Pavey, G., Fantino, E., Cotton, R.G., Imaizumi, K., Masters, C.L., Cappai, R. & Culvenor, J.G. (2004) Expression of truncated presenilin 2 splice variant in Alzheimer’s disease, bipolar disorder, and schizophrenia brain cortex. Brain Res. Mol. Brain Res. 127, 128–135. Weinberg, R.A. & Penman, S. (1968) Small molecular weight monodisperse nuclear RNA. J. Mol. Biol. 38, 289–304. Wood, M.J., Trulzsch, B., Abdelgany, A. & Beeson, D. (2003) Ribozymes and siRNA for the treatment of diseases of the nervous system. Curr. Opin. Mol.Ther. 5, 383–388. Yoneda, Y., Ogita, K., Azuma, Y., Ikeda, M., Tagami, H. & Manabe, T. (1999) N-methyl-d-aspartate signaling to nuclear activator protein-1 through mechanisms different from those for kainate signaling in murine brain. Neuroscience 90, 519–533. Received: 19 April 2007 Accepted: 12 July 2007
© 2007 The Authors Journal compilation © 2007 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.
Genes to Cells (2007) 12, 1179–1191
1191
