|Home | About | Journals | Submit | Contact Us | Français|
Overexpression of high-mobility group A protein 1a (HMGA1a) causes aberrant exon 5 skipping of the Presenilin-2 (PS2) pre-mRNA, which is almost exclusively detected in patients with sporadic Alzheimer's disease. An electrophoretic mobility shift assay confirmed aberrant U1 small nuclear ribonucleoprotein particle (snRNP)-HMGA1a complex formation (via the U1-70K component), with RNA containing a specific HMGA1a-binding site and an adjacent 5′ splice site. Psoralen cross-linking analysis demonstrated that the binding of HMGA1a adjacent to the 5′ splice site induces unusually extended association of U1 snRNP to the 5′ splice site. As a result, spliceosome assembly across either the intron or the exon is arrested at an early ATP-independent stage. We conclude that the HMGA1a-induced aberrant exon skipping is caused by impaired dissociation of U1 snRNP from the 5′ splice site, leading to a defect in exon definition. The proposed molecular mechanism has profound implications for other known posttranscriptional modulation strategies in various organisms, all of which are triggered by aberrant U1 snRNP binding.
Alternative splicing is an essential process applied to generate tremendous diversity in protein products from a limited number of genes (reviewed in references 1 and 39). Misregulation of this process is also a major source of splicing defects in disease-associated genes (reviewed in reference 23). Mutations in essential cis-acting elements and disorders of trans-acting factors can cause such splicing defects (reviewed in reference 22). The former is evident in many genetic diseases that are commonly caused by point mutations in splicing signals, e.g., conserved splice site sequences and splicing enhancer elements (reviewed in references 4, 14, and 22). However, the mechanisms via which trans-acting factors cause changes in the splicing pattern of disease-associated genes, often with severe clinical consequences, are poorly understood.
The presenilin-2 (PS2) gene is one of the known Alzheimer's disease (AD)-associated genes (reviewed in reference 33). The specific aberrant splicing, skipping exon 5, generates the truncated deleterious protein PS2V, which accumulates as visible PS2V bodies at high frequency in the hippocampus of sporadic AD patients (29, 30). We established an experimental system in which this particular exon 5 skipping event can be induced by hypoxia in cultured human neuroblastoma SK-N-SH cells (29, 30). Using this cell line system, we purified and identified high-mobility group A protein 1a (HMGA1a) as a sequence-specific RNA-binding factor that is responsible for specific exon 5 skipping in the PS2 pre-mRNA (18). We found that overexpressed HMGA1a bound to the specific target sequence, GCU(G)CUACAAG, adjacent to the 5′ splice site of the PS2 pre-mRNA (18, 19). This was a surprising discovery, since HMGA1a/b (formerly known as HMG-I/Y; http://www.informatics.jax.org/mgihome/nomen/hmg_family.shtml) proteins had been classified previously as nonhistone DNA-binding proteins (reviewed in reference 26). The fact that markedly increased levels of HMGA1 were observed in brain tissues of sporadic AD patients (18), combined with experimental evidence showing that the PS2V protein impairs the pathway of the unfolded-protein response and stimulates the production of amyloid β (29, 30), demonstrates that HMGA1a-induced PS2V production is one of the risk factors for neuronal cell death in sporadic AD (reviewed in reference 11).
Interestingly, HMGA1a forms a complex with the U1 small nuclear ribonucleoprotein particle (snRNP) through its integral protein component, U1-70K (18, 19). Since U1 snRNP is an essential splicing factor that recognizes the 5′ splice site, and the U1-70K protein plays a critical role in the protein interactions with other splicing factors (reviewed in reference 20), we hypothesized that HMGA1a-induced exon skipping is triggered by the defect at the adjacent 5′ splice site. Here we elucidate the location and kinetics of U1 snRNP binding in the presence of HMGA1a, providing a crucial piece of the puzzle to propose the definitive mechanism of HMGA1-induced aberrant exon skipping that reveals a novel type of splicing modulation associated with human disease.
The template plasmid for BS1/5′SS RNA (see Fig. Fig.1)1) was constructed using an oligonucleotide-annealed DNA fragment of the following oligonucleotides: 5′-agcttCTGGGCAAGTCTAGACGTAGTACCGCTGCTACAAGGTTGGTATCg-3′ and 5′-aattcGATACCAACCTTGTAGCAGCGGTACTACGTCTAGACTTGCCCAGa-3′ (the HMGA1a-binding sequence is underlined, and the parts of flanking HindIII and EcoRI sites are indicated in lowercase). The annealed DNA fragment was subcloned into the HindIII-EcoRI site of the pSP64 vector (Promega) to obtain pSP64-BS1/5′SS. The plasmids for BS1/5′SSmut and BS1mut/5′SS RNAs (RNAs containing mutations at the 5′ splice site and at the HMGA1a-binding sequence, respectively) were constructed in a similar way (pSP64-BS1/5′SSmut and pSP64-BS1mut/5′SS, respectively). The sequences of the transcribed RNAs are shown at the top of Fig. Fig.11.
The original plasmid pcDNA3(+)-BS1-E6 (19) was subcloned into the pSP64 vector to obtain the pSP64-BS1-E6 plasmid. The plasmids for the mmW, mWm, and Wmm (see Fig. Fig.4)4) RNAs were constructed by PCR, using pSP64-BS1-E6 digested with HindIII as a template. The forward primers used were as follows: 5′-ATACaagcttCT(GCCGAGATAAG)2(GCTGCTACAAG)GTTGGTATCGAATTCGTAATCATGGTCATAGCTG-3′ for mmW, 5′-ATACaagcttCT (GCCGAGATAAG)(GCTGCTACA AG)(GCCGAGATAAG)GTTGGTATCGAATTCGTAATCATGGTCATAGCTG-3′ for mWm, and 5′-ATACaagcttCT(GCTGCTACAAG)(GCCGAGATAAG)2GTTGGTATCGAATTCGTAATCATGGTCATAGCTG-3′ for Wmm (the HindIII site is indicated in lowercase, the HMGA1a-binding sequence is underlined, and the mutated HMGA1a-binding sequence is italicized). The reverse primer used was 5′-GGGGaagcttCCGGCTCGTATGTTGTGTGG-3′ (the HindIII site is indicated in lowercase). The PCR products were digested using HindIII and were self-ligated at the HindIII site to obtain the pSP64-mmW, pSP64-mWm, and pSP64-Wmm plasmids. The antisense U1 snRNA-encoding plasmid (pSPU1-) (2) used for Northern blot detection was a kind gift from the laboratory of J. A. Steitz.
The template plasmid for the BS1mut-E6 pre-mRNA was constructed by PCR using pSP64-BS1-E6 digested with HindIII as a template. The forward primer used was 5′-ATACaagcttGGTACCGAGCTCGGATCCCC(GCCGAGATCAG)GTGAGGCCCTGAAT-3′, and the reverse primer used was 5′-GTACCaagcttAAGTTTAAACGCTAGCCAGCTTGGGTCTC-3′ (the HindIII site is indicated in lowercase, and the mutated HMGA1a-binding sequence is italicized).
The template plasmid for the iE5i RNA was constructed by PCR of human genomic DNA (Promega) using the forward primer 5′-GGTGGGaagcttCGAGGAGCAGTCAGGG-3′ (the HindIII site is indicated in lowercase) and the reverse primer 5′-GAGAgaattcTGGCTGGAGGGCAGGGC-3′ (the EcoRI site is indicated in lowercase). The resulting PCR product, which contained the PS2 exon 5 (142 bp) flanked by intronic upstream (50-bp) and downstream (28-bp) sequences, was subcloned into the HindIII-EcoRI site of pSP64 vector to obtain the pSP64-iE5i plasmid.
The pSP64-BS1/5′SS (-BS1/5′SSmut and -BS1mut/5′SS), pSP64-mmW (-mWm and -Wmm), pSPU1-, pSP64-BS1-E6 (-BS1mut-E6), and pSP64-iE5i plasmids were linearized using EcoRI, EcoRI, HindIII, XbaI, and EcoRI, respectively, and were used as the template DNAs for in vitro transcription using SP6 polymerase (New England Biolabs) as described previously (21).
Recombinant HMGA1a was expressed in Escherichia coli and purified as described previously (19). Purified U1 snRNP, which was prepared as described previously (32), was a kind gift from the laboratory of R. Lührmann.
Each reaction mixture (in a total of 15 μl) was incubated at room temperature for 10 to 15 min (for BS1/5′SS and BS1/5′SSmut RNAs) or at 37°C for 20 min (for BS1mut/5′SS RNA) in a binding buffer modified from a previously described method (8), containing 3 mM MgCl2, 50 mM KCl, 20 mM Tris-HCl (pH 7.5), 0.1 mM EDTA, 0.1% (vol/vol) Triton X-100, 10% (vol/vol) glycerol, 3 μg of bovine serum albumin, and 1 μg of tRNA (MnCl2 was replaced with MgCl2, as it was not necessary to eliminate nonspecific polyadenylation activity). RNA-protein complexes were analyzed by 5% polyacrylamide gel electrophoresis (PAGE) (acrylamide:bisacrylamide ratio = 80:1 [wt/wt]) at 4°C using 0.5× Tris-borate-EDTA (TBE) buffer (8). Dried gels were visualized using a bioimaging analyzer (Fujix BAS1000; Fujifilm).
The procedure was essentially performed using a mixture of in vitro splicing, according to the reported method (15, 36). Each reaction mixture (in a final volume of 12.5 μl) containing ~20 fmol of 32P-labeled RNA or pre-mRNA and 200 ng of a psoralen derivative (4′-aminomethyl-trioxsalen hydroxy-chloride; Calbiochem) was incubated at 37°C for the indicated time (0 to 90 min) in the presence of 3 mM ATP, 20 mM creatine phosphate, 20 mM HEPES-NaOH (pH 7.3), 3.5 mM MgCl2, 2% (wt/vol) low-molecular-weight polyvinyl alcohol (Sigma), and 3.5 μl of HeLa cell nuclear extract (CilBiotech). The mixture was irradiated with UV light (365 nm) for 15 min on ice at a distance of 3 cm from the light source (CSL-6A; Cosmo Bio). RNA was extracted with phenol, precipitated with ethanol, and fractionated by denaturing 7% or 10% PAGE, as described previously (21). Dried gels were visualized using a bioimaging analyzer (Fujix BAS1000; Fujifilm).
To identify cross-linked products of RNA or pre-mRNA that interacted with specific U snRNAs, we performed either antisense oligonucleotide-directed RNase H digestion (13) or detection by Northern blot analysis (27, 36). Oligonucleotides (15 nucleotides [nt]) complementary to the 5′ ends of the U1 or U2 snRNAs were used as described previously (13). To digest endogenous U snRNA prior to psoralen UV cross-linking, 3.5 μl of HeLa cell nuclear extract was incubated at 37°C for 20 min in a 12.5-μl reaction mixture containing 50 pmol of each antisense oligonucleotide, 2 U RNase H (Takara Bio), and 3 mM MgCl2. Northern blot detection was performed using an in vitro splicing reaction of nonradiolabeled BS1-E6 pre-mRNA that was analyzed by denaturing 10% PAGE (21). The gel was transferred to a Hybond-NX membrane (GE Healthcare) with 0.5× TBE using a semidry transfer apparatus. The blot was cross-linked by UV light at 120 mJ/cm2 (Stratalinker 2400; Stratagene) and prehybridized in a solution of 50% (vol/vol) formamide, 5× Denhardt's solution, 5× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate), 0.1% sodium dodecyl sulfate (SDS), and 200 μg/ml of salmon sperm DNA. Hybridization was performed in the same solution using a 32P-labeled anti-U1 snRNA probe (2) at 38°C overnight, and the blot was washed with 2× SSC plus 0.2% SDS twice at 50°C and visualized using a bioimaging analyzer (Fujix BAS1000; Fujifilm).
In vitro splicing was performed as described previously (21), with minor modifications. 32P-labeled pre-mRNA (~20 to 40 fmol) was incubated for the indicated time in a 25-μl reaction mixture containing 7.5 μl of HeLa cell nuclear extract (CilBiotech), 3 mM ATP, 20 mM creatine phosphate, 20 mM HEPES-NaOH (pH 7.3), and 3.5 mM MgCl2. RNA was extracted with phenol, precipitated with ethanol, and fractionated by denaturing 7% or 10% PAGE, as described previously (21).
To deplete HMGA1a protein, 50 μl of HeLa cell nuclear extract was precleared for 30 min with 50 μl of a 1:1 diluted slurry of protein A-Sepharose (GE Healthcare) in 1 ml of EBC buffer (50 mM Tris-HCl at pH 8.0, 170 mM NaCl, 0.5% Nonidet P-40, 50 mM NaF) (6), and the supernatant was incubated overnight at 4°C with 18 μl of SP2/0 antibody (negative control) or HMGA1a antibody (FL95; Santa Cruz). The mixture was incubated with 50 μl of a 1:1 diluted slurry of protein A-Sepharose for 30 min and centrifuged. A total of 7.5 μl of this supernatant was used for in vitro splicing assays.
Spliceosome assembly analyses aimed at observing E and EDE complex formations were performed as described previously (12). 32P-labeled pre-mRNA (~20 fmol) was incubated in a 10-μl reaction mixture containing 2.5 μl of HeLa cell nuclear extract, 60 mM KCl, and 4 U of RNase inhibitor (New England Biolabs). At the indicated time points for the splicing reaction, 2 μl of termination buffer (1× TBE buffer containing 20% glycerol and 2.5 mM EDTA) (40) was added to each reaction mixture, which was immediately subjected to 1.2% low-melting-point agarose (Invitrogen) horizontal gel electrophoresis at 4°C. All gels were dried and visualized using a bioimaging analyzer (Fujix BAS1000; Fujifilm).
We previously showed that HMGA1a binds U1 snRNP through U1-70K in vitro (18) and that the aberrant HMGA1a-U1 snRNP complex is formed on the HMGA1a-binding site of heterologous pre-mRNA under splicing conditions (19). To exclude the possibility of an additional factor(s) being necessary to form the HMGA1a-U1 snRNP complex, we used highly purified U1 snRNP and recombinant HMGA1a to perform an electrophoretic mobility shift assay (EMSA) with the short target RNA BS1/5′SS, which contains a single copy of the HMGA1a binding sequence and a 5′ splice site sequence (Fig. (Fig.1,1, top). We also prepared the variant RNAs BS1/5′SSmut and BS1mut/5′SS, which contain mutations at the 5′ splice site and at the HMGA1a-binding sequence, respectively (Fig. (Fig.1,1, top). HMGA1a and U1 snRNP individually bound to the target BS1/5′SS RNA in a dose-dependent manner and formed unique shifted complexes (Fig. (Fig.1,1, lanes 1 to 5). The addition of both factors together led to the appearance of a super-shifted ternary complex (Fig. (Fig.1,1, lanes 6 and 7). The U1 snRNP and HMGA1a binding was not observed with BS1/5′SSmut and BS1mut/5′SS, respectively, and the super-shifted complex did not appear with either of these RNA fragments (Fig. (Fig.1,1, lanes 8 and 9). These results confirmed that the HMGA1a-U1 snRNP complex, in the absence of other factors, forms without steric hindrance on the minimal target RNA containing the essential elements, i.e., an HMGA1a-binding sequence and the adjacent 5′ splice site.
We previously showed that overexpression of HMGA1a allows the formation of an aberrant HMGA1a-U1 snRNP complex (via binding to its 70K component) on the HMGA1a target site adjacent to the 5′ splice site, which leads to inhibition of the splicing of downstream intron (18, 19). Accordingly, we postulated that the inactivation of 5′ splice site is due to HMGA1a-induced misplacement of U1 snRNP to a possible 5′ splice site-like sequence or a pseudo 5′ splice site, located upstream of the HMGA1a-binding site. We found such 5′ splice site-like sequences in the original PS2 pre-mRNA (18) and also in the heterologous pre-mRNAs used previously (i.e., E1-BS1-E2 and E1-BS2-E2) in which HMGA1a-inducible exon skipping was reconstituted (19). On the other hand, our EMSA data indicated that the authentic 5′ splice site is required for the formation of the HMGA1a-U1 snRNP complex (Fig. (Fig.11).
Given the above-described arguments, we sought to identify the actual U1 snRNA annealing site in the presence of excess HMGA1a using psoralen-mediated UV light cross-linking with the BS1/5′SS RNA (derived from E1-BS1-E2), which contained three 5′ splice site-like sequences (Fig. (Fig.1,1, top). We observed a single cross-linked product in a HeLa cell nuclear extract (Fig. (Fig.2A,2A, lane 3), which was abolished either by targeted degradation of U1 snRNA (lane 5) or by mutation of the authentic 5′ splice site (with BS1/5′SSmut RNA) (lane 10), indicating specific cross-linking between U1 snRNA and the 5′ splice site. Remarkably, addition of recombinant HMGA1a did not diminish the U1 snRNA cross-linking signal, leading instead to a visible increase (Fig. (Fig.2A,2A, lanes 6 and 7). We could not observe any other cross-linking signals with the BS1/5′SS and BS1/5′SSmut RNAs in the presence of HMGA1a due to expected U1 snRNP binding to the upstream 5′ splice site-like sequences (Fig. (Fig.2A,2A, lanes 6, 7, 13, and 14). These results revealed that U1 snRNP binds to the authentic 5′ splice site but not to the pseudo 5′ splice sites, even in the presence of HMGA1a.
To investigate how HMGA1a causes a malfunction at the authentic 5′ splice site, where we observed the canonical U1 snRNP binding, we examined the state of psoralen cross-linking using a time-course experiment. We found that HMGA1a induced a significant change in the kinetics of U1 snRNA-5′ splice site cross-linking with the BS1/5′SS RNA substrate (Fig. (Fig.2B).2B). The cross-linking was gradually diminished and almost disappeared at 60 min without the addition of HMGA1a (Fig. (Fig.2B,2B, lanes 1 to 3). In contrast, a substantial amount of the cross-linking was retained up to 60 min after the addition of HMGA1a (Fig. (Fig.2B,2B, lanes 4 to 6). To examine whether the HMGA1a binding to the target site is a prerequisite for extended U1 snRNA-5′ splice site cross-linking, we used the BS1mut/5′SS RNA containing point mutations in the HMGA1a-binding sequence (Fig. (Fig.1,1, top). With the addition of HMGA1a, the kinetics of cross-linking was similar to those observed for the BS1/5′SS RNA without HMGA1a, and the cross-linking products were barely detectable at 60 min (Fig. (Fig.2B,2B, lanes 7 to 9; cf. to lanes 1 to 3).
This finding confirms that binding of HMGA1a to the target sequence is required for the formation of a stable complex with U1 snRNP. This observation is consistent with the following two pieces of splicing data with heterologous pre-mRNA, including a minimal HMGA1a-binding site. (i) Exon inclusion was restored by the addition of competitor oligoribonucleotides carrying the HMGA1a-binding sequence in vitro and in vivo (19). (ii) Addition of recombinant HMGA1a induces exon skipping in vitro (19), whereas immunodepletion of endogenous HMGA1a prevents exon skipping in vitro (Fig. (Fig.33).
The HMGA1a-binding sequence in the PS2 pre-mRNA is located adjacent to the 5′ splice site (Fig. (Fig.1,1, top) (18). Accordingly, we examined whether the distance between the HMGA1a-binding site and the 5′ splice site is critical to elicit the HMGA1 effect on extended U1 snRNA-5′ splice site cross-linking. We prepared mmW/5′SS, mWm/5′SS, and Wmm/5′SS RNAs, which contain three tandem sequences of either the wild-type HMGA1a binding sequence (W) or its mutated sequence (m) to systematically alter the distance from the downstream 5′ splice site while keeping the length and the surrounding sequences constant (Fig. (Fig.4,4, top). Psoralen cross-linking assays revealed that only the mmW/5′SS RNA led to extended U1 snRNA-5′ splice site cross-linking with the addition of HMGA1a (Fig. (Fig.4,4, lanes 3 to 8). Remarkably, the timely dissociation of the U1 snRNP from the 5′ splice site was restored in the mWm/5′SS and Wmm/5′SS RNAs, even with the addition of HMGA1a (Fig. (Fig.4,4, lanes 9 to 14). This result demonstrates that the positioning of the original HMGA1a-binding site, which is adjacent to the downstream 5′ splice site, is crucial for the functional contact between HMGA1a and the U1-70K component to promote extended U1 snRNP binding to the 5′ splice site.
These cross-linking experiments were performed using short RNA fragments that were not subjected to splicing. It was therefore important to examine whether the addition of HMGA1a also prevents U1 snRNP dissociation from the 5′ splice site of a splicing-competent substrate during the splicing reaction. We used the BS1-E6 pre-mRNA, which led to the HMGA1a-induced inhibition of the constitutive splicing at the first step (Fig. (Fig.5A,5A, lanes 1 to 6) (19). HMGA1a is not a general splicing inhibitor, as it had no significant effect on the splicing of the BS1mut-E6 pre-mRNA, which contains a mutated HMGA1a-binding sequence (Fig. (Fig.5A,5A, lanes 7 to 12), or on the β-globin pre-mRNA, which lacks an HMGA1a-binding sequence (19).
Psoralen cross-linking was performed in this splicing at each time point indicated. The inhibition of splicing was monitored using denaturing PAGE, whereas the generated U1 snRNA-5′ splice site cross-linked products were identified using a sensitive Northern blot analysis (Fig. (Fig.5B).5B). We found that this cross-linking signal decreased gradually and almost disappeared at 90 min without the addition of HMGA1a (Fig. (Fig.5B,5B, lanes 1 to 3), whereas the signal was retained up to 90 min after the addition of HMGA1a (Fig. (Fig.5B,5B, lanes 4 to 6). These results led us to conclude that the aberrant HMGA1a-U1 snRNP complex prevents proper U1 snRNP dissociation from the 5′ splice site during the active splicing reaction, leading to HMGA1a-binding site-dependent inhibition of splicing. Notably, it was reported that stable U1 snRNA binding to the 5′ splice site is inhibitory during the process of spliceosome assembly by hindering U1 snRNP dissociation (17).
We assumed that HMGA1a binding to U1 snRNP via the U1-70K component interfered with early spliceosome assembly, as protein interactions through the U1-70K protein play an important role at the early stage of the splicing reaction (reviewed in reference 20); thus, we first examined the effects of HMGA1a on early spliceosome assembly using the same BS1-E6 pre-mRNA by excluding ATP from the splicing reaction (Fig. (Fig.6A).6A). Native agarose gel electrophoresis showed a separation of the E complex from the nonspecific H complex (Fig. (Fig.6A,6A, lanes 1 to 4), and the addition of HMGA1a markedly impaired the formation of the E complex (lanes 5 to 8). These results indicate that the HMGA1a-induced U1 snRNP retention on the 5′ splice site abrogated the ATP-independent formation of the E complex, blocking functional communication between 5′ and 3′ splice sites via SF1, the 35-kDa subunit of U2AF (U2AF35), and SR proteins, which is essential for the commitment of the pre-mRNA to the subsequent splicing process (reviewed in reference 20). As expected, formation of the subsequent ATP-dependent spliceosomal complexes A, B, and C using the same BS1-E6 pre-mRNA was severely impaired by the addition of HMGA1a (data not shown).
Because overexpression of HMGA1a causes exon 5 skipping in the original PS2 pre-mRNA in vivo (18), it is critical to investigate the HMGA1a-induced changes in the exon definition complex that could not be examined with the single-intron substrate BS1-E6. Using substrate carrying an exon and flanking introns, the early ATP-independent complex across the exon, termed the exon-defined E (EDE) complex, was recently identified and characterized (31); therefore, we analyzed the EDE complex using the iE5i RNA, which carries full exon 5 of PS2 with the flanking polypyrimidine tract, the 3′ splice site, and the 5′ splice sites (Fig. (Fig.6B,6B, top). Native agarose gel analysis successfully detected a discrete complex separated from the nonspecific H complex that corresponded to the EDE complex (Fig. (Fig.6B,6B, lanes 1 to 4). We observed the complete disruption of this specific EDE complex formation by the addition of HMGA1a (Fig. (Fig.6B,6B, lanes 5 to 8). These results suggest that the HMGA1a-U1 snRNP complex arrests the exon definition complex across exon 5 in the PS2 pre-mRNA at an early ATP-independent stage, which is a crucial cause of HMGA1-induced exon 5 skipping.
The main aim of this study was to elucidate the mechanism that underlies aberrant skipping of the specific exon (exon 5 of the PS2 pre-mRNA) induced by overexpressed HMGA1a in patients with sporadic Alzheimer's disease (18). Based on the experimental evidence obtained from our previous studies and the present analyses, here we propose an exact mechanism for HMGA1a-inducible exon skipping, which is initiated by the prevention of normal U1 snRNP dissociation from the authentic 5′ splice site (Fig. (Fig.7).7). Given the documented stepwise process of splicing, proper association of U1 snRNP with, and its timely dissociation from, the 5′ splice site would be prerequisites for the subsequent splicing stages (reviewed in reference 20). Nevertheless, to the best of our knowledge, the present study is the first to attribute the essential cause of a human disease to the impairment of U1 snRNP dissociation induced by a trans-acting factor.
In the original PS2 pre-RNA, the formation of a stable HMGA1a-U1 snRNP complex via the U1-70K component was postulated to block the interaction between U1-70K and U2AF35, which is mediated by their binding to SR proteins (38, 41, 42). This interaction promotes communication between adjacent 5′ and 3′ splice site pairs across either intron or exon, termed “intron bridging” and “exon definition,” respectively, which play a critical role in the recognition of bona fide introns and exons (Fig. (Fig.7A)7A) (reviewed in references 20 and 28). Here we provided evidence that hyperstable HMGA1a-U1 snRNP binding via the U1-70K protein interferes with the formation of the ATP-independent early spliceosome, which is responsible for either intron bridging (in a substrate carrying a single intron followed by the PS2 exon 6) or exon definition (in a substrate carrying the PS2 exon 5 with flanking intronic sequences) (Fig. (Fig.66 and and7B).7B). The observed phenomena are fully supported by the fact that overexpression of free U1-70K, interfering competitively with the formation of the HMGA1a-U1 snRNP complex, completely prevents exon 5 skipping of the PS2 pre-mRNA in vivo (18). Intriguingly, the proposed mechanism is coordinate with recent structural data on U1 snRNP; i.e., the N-terminal polypeptide of U1-70K wraps almost halfway around U1 snRNP to make contact with the U1-C protein (25), which is essential for 5′ splice site recognition and for the formation of the E complex (5, 9, 17, 37).
By extrapolating from the HMGA1a-induced mechanism proposed here, other disease genes that comprise an HMGA1a-binding sequence adjacent to the 5′ splice site may be potential targets for the splicing defects in response to the overexpressed HMGA1a, which is induced by external stresses, such as hypoxia, tumor factors, and cell growth/differentiation factors (reviewed in reference 26). Recently, we have identified HMGA1a-inducible aberrant splicing in the estrogen receptor α pre-mRNA in breast cancer cells (our unpublished data).
The HMGA1 and HMGA2 genes, which encode the HMGA1a, HMGA1b, and HMGA2 proteins, are well-known proto-oncogenes that promote tumor progression and metastasis when overexpressed in cells (reviewed in reference 26). The novel function of HMGA1a as an RNA-binding protein provides a potential paradigm for studying the inducible aberrant splicing events found in sporadic diseases and cancers that arise in the absence of mutations in the corresponding genes.
It is of considerable interest that analogous U1 snRNP-mediated systems are evolutionally conserved among different species as a strategy to modulate posttranscriptional events during gene expression (Table (Table1,1, see Roman numerals). Three out of five cases (I, II, and IV) are quite analogous in the sense that the binding to the 5′ splice site is altered via U1 snRNP-protein interaction through U1-70K, even though there is no significant homology either in these partner proteins or in their target RNA sequences. Remarkably, all cases cause dysfunction of the authentic 5′ splice site, either by shifting U1 snRNA binding to the pseudo 5′ splice site (II, IV, and V) or by hyperstabilizing U1 snRNP binding to the authentic 5′ splice site (I and III), which leads to defects either in splicing (I, II, III, and V) or in polyadenylation (IV). As a result, deleterious (I) and functional (II and V) protein isoforms are produced; otherwise, gene silencing is induced by degradation of the target mRNAs (III and IV)—eventually contributing to cause disease (I), to prevent disease (V), to gain a new function (II), and to promote feedback regulation (III and IV). These significant physiological and medical phenotypes are quite diverse; nonetheless, it is amazing that the shared mechanism was initiated by the aberrant binding of the essential splicing factor U1 snRNP.
We are grateful to G. Weber and R. Lührmann for providing the purified U1 snRNP; to J. A. Steitz for providing the antisense U1 snRNP-encoding plasmid; to D. Rio, J. Valcárcel, H. Lou, A. Macmillan, D. Wassarman, and M. L. Hastings for valuable suggestions on psoralen cross-linking and the spliceosome assay; to R. Reeves, X. Roca, and T. Manabe for critical reading of the manuscript; to S. I. Gunderson, K. W. Lynch, J. F. Cáceres, and A. R. Krainer for helpful comments prior to publication; and to M. Tohyama, T. Katayama, T. Yanase, R. Takayanagi, H. Nawata, and P. Sassone-Corsi for support and encouragement.
This work was supported by the “Collaboration with Local Communities” project as a matching fund subsidy for private universities to A.M. and a Grant-in-Aid for Scientific Research (C) to K.O. (grant 21591679) from the Ministry of Education, Culture, Sports, Science and Technology of Japan.
Published ahead of print on 1 March 2010.