These authors contributed equally to this study.
Liver Biology and Pathobiology
Article first published online: 26 DEC 2007
Copyright © 2007 American Association for the Study of Liver Diseases
Volume 47, Issue 2, pages 686–697, February 2008
How to Cite
Matsui, K., Nishizawa, M., Ozaki, T., Kimura, T., Hashimoto, I., Yamada, M., Kaibori, M., Kamiyama, Y., Ito, S. and Okumura, T. (2008), Natural antisense transcript stabilizes inducible nitric oxide synthase messenger RNA in rat hepatocytes. Hepatology, 47: 686–697. doi: 10.1002/hep.22036
Sequence data: DNA Data Bank of Japan/European Bioinformatics Institute/GenBank accession numbers AB212679, AB250951, AB250952, AB260892, AB290142, and AB290143.
Potential conflict of interest: Nothing to report.
- Issue published online: 26 JAN 2008
- Article first published online: 26 DEC 2007
- Manuscript Accepted: 9 SEP 2007
- Manuscript Received: 16 APR 2007
- Ministry of Education, Culture, Sports, Science, and Technology of Japan
- Kansai Medical University
- Amino Up Chemical Co., Ltd. (Sapporo, Japan)
During inflammation, inducible nitric oxide synthase (iNOS) is induced to generate the important mediator nitric oxide (NO). Interleukin 1β (IL-1β) induces iNOS messenger RNA (mRNA), iNOS protein, and NO in rat hepatocytes. We found that the stability of iNOS mRNA changed during the induction and that the antisense (AS) strand corresponding to the 3′-untranslated region (3′UTR) of iNOS mRNA was transcribed from the iNOS gene. Expression levels of the iNOS AS transcript correlated with those of iNOS mRNA. The 1.5-kilobase region 3′-flanking to iNOS gene exon 27 was involved in IL-1β induction. Knockdown experiments suggest that sense oligonucleotides to iNOS mRNA significantly reduced iNOS mRNA levels in the hepatocytes by blocking the interaction between iNOS mRNA and the AS transcript. Overexpression of iNOS AS transcript stabilized the reporter luciferase mRNA through the fused iNOS mRNA 3′UTR. These results together with the data in a yeast RNA-hybrid assay suggested that the iNOS AS transcript interacted with iNOS mRNA and stabilized iNOS mRNA. The iNOS mRNA colocalized with the AU-rich element-binding protein HuR, a human homolog of embryonic lethal-abnormal visual protein, and heterogeneous nuclear ribonucleoprotein L (hnRNP L) in the cytoplasm of rat hepatocytes. Interaction assays further revealed that the iNOS AS transcript interacted with HuR, which interacted with hnRNP L, suggesting that iNOS mRNA, the AS transcript, and the RNA-binding proteins may mutually interact. Conclusion: The natural AS transcript of the iNOS gene interacts with iNOS mRNA and may play an important role in the stability of iNOS mRNA. This RNA-RNA interaction may be a new therapeutic target for NO-mediating inflammatory diseases. (HEPATOLOGY 2008.)
Septic shock, chronic inflammatory diseases, and liver injury up-regulate the expression of inducible nitric oxide synthase (iNOS) to generate nitric oxide (NO), which has beneficial and sometimes detrimental effects on hepatic functions.1–3 NO production correlates with the iNOS protein level. Expression of iNOS messenger RNA (mRNA) is rapidly induced in response to inflammatory stimuli. The proinflammatory cytokine interleukin-1β (IL-1β) selectively induces iNOS gene expression in rat hepatocytes, whereas bacterial endotoxin, for example, lipopolysaccharide (LPS), selectively induces iNOS gene expression in resident macrophage Kupffer cells.4, 5 IL-1β binds to type I interleukin-1 receptor (IL-1RI) on hepatocytes, and through the signal transduction pathway, IL-1RI activates transcription factors, such as nuclear factor κB (NF-κB) and CAAT/enhancer-binding protein β (C/EBPβ). These proteins bind to the iNOS gene promoter to synergistically increase transcription.6, 7 Although it was previously thought that iNOS gene expression is predominantly transcriptionally controlled, the involvement of other regulatory mechanisms has been suggested. Sodium salicylate inhibits iNOS protein synthesis and NO production, but not iNOS mRNA expression, and this indicates that sodium salicylate destabilizes iNOS mRNA at the posttranscriptional level.8 In addition, the IL-1RI gene itself is induced to augment iNOS transcription through the phosphatidylinositol-3-kinase/Akt pathway.9
The 3′-untranslated region (3′UTR) of the mRNA, especially its secondary structure, is believed to be involved in posttranscriptional regulation of mRNA stability.8, 10 In the 3′UTR of human, rat, and mouse iNOS mRNAs, there are several AU-rich elements (AREs; 5′-AUUUA-3′ or 5′-AUUUUA-3′), which appear in mRNAs encoding early-response genes, including acute phase proteins, cytokines, and some proto-oncogenes.11 ARE-binding proteins HuR, a human homolog of Drosophila embryonic lethal-abnormal visual (ELAV) protein, and AU-binding factor 1/heterogeneous nuclear ribonucleoprotein D (hnRNP D) bind to iNOS mRNA 3′UTR.10, 12 Furthermore, other hnRNPs that are involved in mRNA synthesis and maturation, such as hnRNP L and hnRNP I/polypyrimidine tract-binding protein (PTB), also bind to iNOS mRNA.10 hnRNP L preferentially binds to the regions including CA repeats13 and a 12-nucleotide (nt) sequence conserved between human vascular endothelial growth factor and mouse iNOS mRNAs.14, 15 PTB binds to a polypyrimidine tract, such as a CU-rich region of human iNOS mRNA 3′UTR.15, 16
Recently, it has been reported that the antisense (AS) strand of genes is frequently transcribed as AS transcripts, which are believed to regulate gene expression at the posttranscriptional and/or translational levels.17, 18 Robb et al.19 reported that a long, spliced AS transcript is involved in the posttranscriptional regulation of endothelial NO synthase. We expected a natural AS transcript of the iNOS gene, which might be involved in the regulation of gene expression, and tried to find it.
Materials and Methods
Cultures of Hepatocytes.
Hepatocytes were isolated from male Wistar strain rats, seeded, and incubated at 37°C overnight, as previously described.20 The next day, primary cultured hepatocytes were treated with 1 nM IL-1β (Otsuka Pharmaceutical Co., Tokushima, Japan) for the indicated time in subsequent experiments. Animal experiments were approved by the Animal Care Committee of Kansai Medical University.
Nitrite in medium was measured with NO colorimetric assay kits (Roche Diagnostics, Mannheim, Germany).
Western Blot Analysis.
Western blot analysis was performed essentially according to the previously described method.9
Total RNA was isolated by TRIzol (Invitrogen, Carlsbad, CA) and treated with TURBO DNA-free kits (Ambion, Austin, TX). Poly(A)+ and poly(A)− RNAs were fractionated by the PolyATract mRNA isolation system (Promega, Madison, WI).
Strand-Specific Reverse Transcription-Polymerase Chain Reaction (RT-PCR).
The complementary DNA (cDNA) was synthesized from total RNA with strand-specific primers, and step-down polymerase chain reaction (PCR) was performed, as previously described.21, 22 For iNOS mRNA, oligo(dT) primer for reverse transcription (RT) and CCAACCTGCAGGTCTTCGATG and GTCGATGCACAACTGGGTGAAC (5′3′ direction) were used. For iNOS AS transcript, sense primer TGCCCCTCCCCCACATTCTCT (RT) and ACCAGGAGGCGCCATCCCGCTGC and CTTGATCAAACACTCATTTTATTAAA (PCR) were used. For the internal control elongation factor 1α (EF) mRNA,23 TCTGGTTGGAATGGTGACAACATGC and CCAGGAAGAGCTTCACTCAAAGCTT were used. For cytokine-induced neutrophil chemoattractant 1 (CINC-1) mRNA,24 5′-GCCAAGCCACAGGGGCGCCCGT-3′ and 5′-ACTTGGGGACACCCTTTAGCATC-3′ were used. The mRNA levels were measured in triplicate by real-time PCR with SYBR Green I (Roche) and iCycler or Opticon2 (Bio-Rad, Hercules, CA). The copy number was normalized by EF mRNA (for iNOS mRNA) or the amount of total RNA (for iNOS AS transcript).
Rapid Amplification of Complementary DNA Ends (RACE).
RACE was performed with a cDNA PCR library kit (Takara Bio, Ohtsu, Japan).
Construction of Reporter Plasmids.
The 1.5-kilobase (kb) downstream region of rat iNOS gene exon 27 amplified by genomic PCR with ggcaagcttAAAACATACCTTCGTCCACTCCCA and cctagctaGCCAGGGGAGTTGCCATGGTGGG was reversely inserted to pGL4.10 vector (Promega) to create pRiAS-Luc1.5. The 85–base pair (bp) fragment, including a TATA box of the iNOS gene, replaced iNOS exon 27 of pRiAS-Luc1.5 to create pRiAS-Luc-1.5+TATA. Other plasmids were constructed in a similar fashion.
Construction of Plasmids for AS Overexpression.
EF promoter (kindly provided by S. Nagata, Osaka University)23 was inserted into pGL3-Basic (Promega), and then iNOS 3′UTR and SV40 polyadenylation signal (SVpA) were inserted to create reporters pEF-Luc-3′UTR and pEF-Luc-SVpA, respectively. Cytomegalovirus (CMV) promoter and iNOS AS cDNA were inserted for pCMV-AS, and the lacZ gene was inserted for pEF-LacZ-SVpA.
Transfection of Hepatocytes.
Hepatocytes were subjected to magnet-assisted transfection. Plasmid DNA was mixed with magnet-assisted transfection A (MATra-A reagent; IBA, Göttingen, Germany) and added to the wells. After a 15-min incubation on a magnetic plate, the medium was replaced. Cells were cultured overnight and treated with or without IL-1β for 4 hours. Luciferase and β-galactosidase activity was measured by PicaGene (Wako Pure Chemicals, Osaka, Japan) and Beta-Glo kits (Promega), respectively. Oligodeoxyribonucleotides blocked by phosphorothioate bonds (Operon Biotechnologies, Tokyo, Japan) were designed according to a published method.25 Sequences were as follows (an asterisk indicates a phosphorothioate bond): S1, C*A*T*TCTCTTTCCTTTGC*C*T*C; S2, G*C*C*TCATACTTCCTCAG*A*G*C (bases matched to the loop are underlined); S3, T*A*G*CTGCATTGTGTACA*G*A*T; S4, G*T*G*TATAATTCCTTGAT*G*A*A; Scr2, G*G*T*ATTGCCCACCCAAC*T*C*T; Scr3, G*G*C*TCCATATGATTAGA*T*G*T; and Scr4, G*A*T*TGTTACTTAGAGAC*T*A*T. Each oligonucleotide was mixed with MATra-A reagent. Total RNA was analyzed by RT-PCR. Transfection was performed in triplicate at least three times.
Preparation of RNA Probes.
cDNAs including a T3 promoter sequence (underlined) were PCR-amplified with TGCCCCTCCCCCACATTCTCT and aattaaccctcactaaagCTTGATCAAACACTCATTTTATTAAA (282-nt 3′UTR AS probe for northern blot analysis), aattaaccctcactaaagGCCAGAAACGTTATCATGAGGAT and CTTGATCAAACACTCATTTTATTAAA (575-nt 3′UTR sense probe for northern blotting), and aattaaccctcactaaaggTTCTCTTTCCTTTGCCTCATAC and ATCTGGGCCACTTTGCATGACT [433-nt sense probe for ribonuclease (RNase) protection assay]. These templates were transcribed in vitro with T3 RNA polymerase (Stratagene, La Jolla, CA) and digoxigenin 11 (DIG-11)–uridine triphosphate (Roche).
Northern Blot Analysis.
RNA was resolved by agarose gels, blotted onto Nytran N membrane (Whatman, Brentford, United Kingdom), and hybridized with a DIG-labeled AS probe at 68°C (iNOS mRNA) and a sense probe at 73°C (AS transcript) overnight. The membrane was washed at high stringency, incubated with anti-DIG antibody (Roche), and exposed to X-ray films with CDP-Star.
RNase Protection Assay.
A DIG-labeled 433-nt sense RNA probe was mixed with hepatocyte poly(A)− RNA or yeast transfer RNA (tRNA; 20 μg), heat-denatured, hybridized at 50°C, and digested with RNases, as previously described.22 RNA was resolved by 5% polyacrylamide/urea gel, blotted, and detected, as described above.
Construction of Plasmids in Yeast Systems.
HuR cDNA was cloned into pYESTrp3 vector (Invitrogen) to create pYESTrp-HuR. The bait vector pTFB-LexA was constructed from pTFB-1 (Dualsystems, Zurich, Switzerland) and pHybLex/Zeo-MS2 (Invitrogen). Then, hnRNP L cDNA was inserted into pTFB-LexA to create pTFB-LexA-hnRNP L and inserted into pYESTrp3 to create pYESTrp-hnRNP L. To express LexA–nuclear localization signal (NLS)–MS2 protein, pTFB-LexA-NLS-MS2 was constructed from pTFB-1, pHybLex/Zeo-MS2, and NLS oligonucleotides. For the RNA-hybrid assay, pRHP-m26E was constructed from pRH3′ (Invitrogen), pI-RED1 (Toyobo, Osaka, Japan) harboring the 3-phosphoglycerate kinase 1 promoter/terminator, and the m26 transactivation sequence.26 pYWP-MS2 was constructed from pYESTrp3, pRH3′, and pI-RED1. The iNOS 3′UTR cDNA was inserted to create a bait pYWP-MS2-3′UTR (forward) and a prey pRHP-m26E-AS (reverse).
Yeast Two-Hybrid and RNA-Hybrid Assays.
The Saccharomyces cerevisiae strain L40-ura3 (Invitrogen) was transformed by the bait pTFB-LexA-hnRNP L and a pYESTrp prey. Colonies selected in yeast complete medium deficient in leucine and tryptophan were grown in yeast complete medium further deficient in histidine. The culture was subjected to measurement of β-galactosidase activity with Beta-Glo kits. Relative β-galactosidase activity was normalized by optical density. For the RNA-hybrid assay, L40-ura3 was transformed by pTFB-LexA-NLS-MS2, a bait, and a prey.
RNA–Fluorescence In Situ Hybridization (FISH) and Immunocytochemistry.
Hepatocytes were permeabilized with 0.25% Triton X-100, fixed with 2% paraformaldehyde, and repermealibized with 2% dodecyl trimethyl ammonium chloride (Sigma-Aldrich, St. Louis, MI). Cells were processed for RNA-FISH with a DIG-labeled AS probe corresponding to iNOS 3′UTR (8 μg/mL), as previously described.27 Cells were subsequently blocked and incubated with the horseradish peroxidase–conjugated fragment antigen binding fragments from a sheep anti-DIG antibody (7500 mU/mL) at 25°C for 1.5 hours. The probes were visualized with fluorophore tyramide (1:50 dilution). Primary antibodies were rabbit anti-rat HuR (4 μg/mL; Upstate) and anti-rat hnRNP L (4 μg/mL; Santa Cruz Biotechnology, Santa Cruz, CA) immunoglobulin G's (IgGs). Tetramethyl rhodamine isothiocyanate–conjugated porcine anti-rabbit IgG antiserum (1:30 dilution) was used as the secondary antibody. Visualization was performed with an Olympus Fluoview FV300 confocal laser scanning microscope. The fluorescein isothiocyanate, tetramethyl rhodamine isothiocyanate, and Nomarski images were obtained sequentially through separate channels and merged electronically by computer software using the microscope images.
IL-1β–treated hepatocytes were fixed by 1% formaldehyde and washed with 125 mM glycine. Cells were sonicated in the presence of yeast RNA and RNase inhibitor. After centrifugation, supernatant was immunoprecipitated with Protein G-Magnetic Beads (New England Biolabs, Ipswich, MA) that were prebound with antibodies against HuR, hnRNP L, or PTB (Zymed, South San Francisco, CA). Beads were washed, digested with proteinase K at 65°C for 2 hours, and subjected to RNA isolation and RT-PCR.
IP–RT-PCR with Purified RNA.
Sonicated hepatocyte lysates were immunoprecipitated with the anti-HuR antibody that was prebound to protein G-Sepharose beads (GE Healthcare Biosciences, NJ) and then washed. Beads were mixed with hepatocyte total RNA that was separately prepared and incubated at 25°C for 2 hours. Beads were washed and subjected to RNA isolation and RT-PCR.
RNA–Electrophoretic Mobility Shift Assay (EMSA).
The Escherichia coli BL21-CodonPlus(DE3) strain (Stratagene) was transformed with pMNT-HuR-Strep or pCOLD-Strep-hnRNPL, and Strep-tagged proteins purified by Strep-Tactin chromatography (IBA) were incubated with the biotinylated iNOS AS RNA probe at 25°C for 15 minutes and resolved by agarose gel electrophoresis. If necessary, antibodies were added. The gel was blotted, and bands were detected by streptavidin–alkaline phosphatase and 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium chloride.
Statistical Analysis and Informatics.
Results in the figures are representative of at least three independent experiments yielding similar findings. Differences were analyzed by Student t test. The secondary structure was predicted with the mfold program.28
Stability of iNOS mRNA Changes During Induction.
We measured expression levels of the iNOS gene in primary cultured hepatocytes. Western blot analysis of cell extracts demonstrated that the iNOS protein increased markedly after IL-1β addition, and NO production correlated with the iNOS protein level (Fig. 1A). Real-time RT-PCR analysis revealed that the iNOS mRNA expression increased at 2 hours after IL-1β addition, peaked at 6 hours, and then decreased. To estimate the rates of synthesis and degradation of iNOS mRNA, we measured the half-life of iNOS mRNA. At 4 hours after IL-1β addition, iNOS mRNA was synthesized at the maximum rate (Fig. 1B). When the RNA synthesis inhibitor actinomycin D was added, the iNOS mRNA levels decreased, and the half-life of iNOS mRNA was 339 minutes. Prominent degradation occurred at 7 hours, and half-lives of iNOS mRNA in the absence and presence of actinomycin D were 84.9 and 60.2 minutes, respectively (Fig. 1C). In the presence of actinomycin D, the half-life before the peak was 5.6-fold longer than that after the peak, suggesting that iNOS mRNA is more stable before the induction peak.
AS Strand Corresponding to iNOS mRNA 3′UTR Is Transcribed.
We expected that the natural AS transcript of the iNOS gene would be present because the AS strand of genes is frequently transcribed.17, 18 We presumed that the iNOS AS transcript might be transcribed from exon 27, which includes the 3′UTR (Fig. 1D).
First, northern blot analysis was performed with sense and AS RNA probes to the 3′UTR (Fig. 2A). When the sense probe was hybridized at high stringency, a smear ranging from 600 to 1000 nt was detected in both total and poly(A)− RNAs of IL-1β–stimulated hepatocytes. Only this smear showed stronger intensity in response to IL-1β, and this suggests that the iNOS AS transcript gave the smear. This also suggests that these AS transcripts have no poly(A) tails. Such hybridization smears or multiple bands are often observed in analyses of sense-AS transcript pairs that are produced from the same locus on mouse chromosomes.18 In contrast, bands of iNOS mRNA were detected with the AS probe in both total and poly(A)+ RNAs of IL-1β–stimulated hepatocytes. Second, we performed an RNase protection assay to delineate the 5′-start of the AS transcription, using a sense RNA probe (Fig. 2B). The size of the band protected from RNases was about 200 nt. In contrast, the RNase protection assay to delineate the 3′-end showed several bands (data not shown). Finally, we carried out RACE to determine the nucleotide sequence of the iNOS AS transcript. Most of the 5′-RACE cDNAs started at the fifth and seventh nucleotides from the end of exon 27, whereas the 3′-ends of the 3′-RACE cDNAs varied (data not shown). These data confirmed that the AS transcript started at the end of the iNOS gene exon 27 but stopped at various sites (Fig. 2C). We concluded that an AS transcript without a poly(A) tail was transcribed from iNOS gene exon 27 in response to IL-1β.
IL-1β Induces Expression of iNOS AS Transcript.
To discriminate iNOS AS transcript from iNOS mRNA, we established and performed strand-specific RT-PCR. When the oligo(dT) primer was used for RT, only iNOS mRNA was reverse-transcribed and amplified (Fig. 1D). When a sense primer to iNOS mRNA was used for RT, we detected the iNOS AS transcript in IL-1β–stimulated hepatocytes (Fig. 2D). Expression of the AS transcript showed a peak at 6 hours after IL-1β addition. Interestingly, the expression level of iNOS AS transcript correlated with that of iNOS mRNA.
To investigate which cytokine induces the iNOS AS transcript, we added various cytokines to the hepatocytes (Fig. 2E). Only IL-1β induced the iNOS AS transcript in strand-specific RT-PCR analyses, and this indicates that IL-1β selectively induces both iNOS mRNA and AS transcript.
To clarify the regulatory elements for the AS transcription, we isolated the 3′-flanking genomic region of iNOS gene exon 27. The 300-bp region adjacent to exon 27 was GC-rich (57.3%), and one Sp1-binding site (but no TATA and CCAAT boxes) was present (Fig. 3). Within up to 35 kb downstream of the iNOS gene, only one gene is located, which encodes galectin-5, a β-galactoside–binding lectin.29 Galectin-5 is abundantly expressed in erythrocytes, but not in hepatocytes.30 Indeed, the RT-PCR analysis revealed that galectin-5 mRNA was hardly detected in rat hepatocytes regardless of IL-1β stimulation (M. Nishizawa and T. Okumura, unpublished data, 2007). Thus, it is unlikely that the galectin-5 expression is involved in the expression of iNOS AS transcript. Then, we prepared various reporter constructs, using the 2.7-kb fragment between iNOS and galectin-5 genes for the firefly luciferase assay. Constructs harboring the 1.5-kb or 2.7-kb fragment showed weak promoter activity and no IL-1β induction. It is reported that both weak and strong promoters for the AS transcripts are present in the genome-wide sense-AS transcript study.17 Thus, we replaced exon 27 with the fragment harboring a TATA box of the iNOS gene. The luciferase assay demonstrated that only the construct harboring the 1.5-kb region (1.5+TATA) showed a significant increase after IL-1β administration. We found two NF-κB–binding sites and a C/EBPβ-binding site6 within this fragment. Given all these data, the AS strand of the iNOS gene was transcribed in response to IL-1β.
iNOS AS Transcript Interacts with the 3′UTR of iNOS mRNA.
To investigate the interaction between iNOS mRNA and AS transcript, we predicted the secondary structure of iNOS mRNA and AS transcript, using the mfold program.28 Of seven predicted structures, we found four common regions designated A to D in iNOS mRNA 3′UTR, and each region included at least one stem-loop structure (Fig. 4). As expected from sequence complementarity, there were corresponding loops, Aas, Bas, Cas, and Das, in the secondary structure of the AS transcript, respectively. When a homology search in the iNOS 3′UTR was performed, we unexpectedly found cross-homologies between these loop sequences (Fig. 4B). These results suggest the possibility that iNOS mRNA and the AS transcript interact at the loops in various combinations.
To examine the function of iNOS AS transcript, knockdown experiments using iNOS sense oligonucleotides were carried out. Sense oligonucleotides S2 to S4 were designed according to the iNOS mRNA sequence and included loops that might cross-hybridize with the loops of the AS transcript (Fig. 4B). When each sense oligonucleotide (S2, S3, or S4) was introduced to hepatocytes, all of them reduced iNOS mRNA levels in strand-specific RT-PCR and real-time PCR analyses (Fig. 5A). Because the transfection efficiency of oligonucleotides was 40 to 60% by our hands (data not shown), it seemed that these sense oligonucleotides effectively reduced iNOS mRNA levels. The mixture of S2, S3, and S4 also reduced the iNOS mRNA level (data not shown). In contrast, S1 (outside of the B loop) did not reduce the level. Because AS transcript levels did not decrease, the sense oligonucleotides S2, S3, and S4 seemed to compete with iNOS mRNA. The sense oligonucleotides S1 to S5 do not include any known binding sites of HuR, hnRNP L, and PTB,12–16 or their complementary sequences. The chemokine ligand CINC-1 is also induced by IL-1β in hepatocytes, and rat CINC-1 mRNA has 11 AREs in its 3′UTR.24 However, the iNOS sense oligonucleotides did not affect the CINC-1 mRNA levels. These data together indicate that the iNOS sense oligonucleotides selectively reduce the iNOS mRNA levels.
Next, the time course of iNOS mRNA levels was measured with S4 sense and Scr4 scramble oligonucleotides (Fig. 5B). When the Scr4 scramble control was applied, iNOS mRNA levels increased, and the doubling time was 119 minutes. In contrast, when S4 was introduced, both the expression levels and synthesis rate of iNOS mRNA decreased, and the doubling time was 461 minutes; this demonstrated that S4 efficiently destabilizes iNOS mRNA. Furthermore, introduction of the S4 sense oligonucleotide to hepatocytes reduced the iNOS protein level and NO production (data not shown).
To confirm the contribution of the 3′UTR to the iNOS mRNA stability, we performed overexpression experiments of iNOS AS transcript. Both synthesis and degradation of iNOS mRNA simultaneously occur in the hepatocytes (Fig. 1). To separately monitor degradation of the mRNA, we prepared a luciferase reporter (pEF-Luc-3′UTR) driven by the EF promoter with the iNOS 3′UTR downstream of the luciferase gene (Fig. 6A). Because the EF promoter is a house-keeping, constitutive promoter,23 an mRNA synthesis rate was assumed to be constant. When the constructs constantly producing mRNAs were used, we could evaluate the stability of the luciferase-3′UTR mRNA, which may reflect the stability of the endogenous iNOS mRNA. When pEF-Luc-3′UTR was introduced to hepatocytes with the AS effector (pCMV-AS), the level of the luciferase mRNA was significantly higher than that without the AS effector (Fig. 6B). When a control reporter (pEF-Luc-SVpA) was introduced, the AS effector did not stabilize the luciferase mRNA. These data suggest that the iNOS AS transcript stabilizes the luciferase mRNA through the iNOS 3′UTR.
To test whether iNOS mRNA interacts with the AS transcript in vitro, we performed a yeast RNA-hybrid assay.26 If RNA-RNA interaction is present, the reporter lacZ gene is activated (Fig. 6C). Only the combination of iNOS mRNA bait and iNOS AS transcript prey showed increased β-galactosidase activity (Fig. 6D). Together with the data from knockdown and overexpression experiments, it is very likely that iNOS AS transcript interacts with iNOS mRNA.
iNOS mRNA Colocalizes with HuR and hnRNP L in the Cytoplasm.
It has been reported that HuR, hnRNP L, and PTB bind to different sites of iNOS mRNA12, 15, 16: HuR binds to AREs in regions C and D of human mRNA, hnRNP L binds to region D of mouse mRNA, and PTB binds to the CU-rich region between regions A and B of mouse mRNA (Fig. 4A). It was unclear whether these proteins bind to rat iNOS mRNA in IL-1β–stimulated hepatocytes. Thus, we detected iNOS mRNA by RNA-FISH and HuR and hnRNP L by immunocytochemistry (Fig. 7). When IL-1β was added, iNOS mRNA was detected in the cytoplasm, whereas HuR was detected in both the cytoplasm and nuclei (Fig. 7A,B). Merged images revealed that iNOS mRNA and HuR colocalized around the nuclei (Fig. 7C). In the absence of IL-1β (Fig. 7E,F), iNOS mRNA was not detected, and HuR staining showed in punctate signals in the nuclei. When anti-hnRNP L antibody was used, the perinuclear colocalization of iNOS mRNA and hnRNP L was also observed in the presence of IL-1β (Fig. 7K). These data suggest that HuR and hnRNP L may associate with iNOS mRNA in the cytoplasm.
The iNOS AS Transcript Also Interacts with RNA-Binding Proteins.
Next, we performed IP–RT-PCR analysis to examine whether an iNOS mRNA–AS transcript–protein complex was formed. Hepatocyte extracts were incubated with antibodies, and immunoprecipitated RNA was analyzed by strand-specific RT-PCR and real-time PCR. Both iNOS mRNA and AS transcript were detected in the immunoprecipitates (Fig. 8A), and this suggests that iNOS mRNA and AS transcript bind to HuR, hnRNP L, and PTB. To confirm the results, we performed IP–RT-PCR with purified hepatocyte RNA with anti-HuR antibody. Real-time PCR analyses of bead-bound RNA revealed that both iNOS mRNA and AS transcript were detected (Fig. 8B). These data support the idea that iNOS mRNA, the AS transcript, and RNA-binding proteins including HuR bind together.
We further analyzed the interactions among the iNOS AS transcript, HuR, and hnRNP L. It is unknown whether HuR interacts with hnRNP L, although hnRNP L interacts with PTB.15 Thus, we investigated the interaction between HuR and hnRNP L. The yeast two-hybrid assay revealed that hnRNP L bait and HuR prey gave high β-galactosidase activity (Fig. 9A). To ascertain this result, IP–western blot analysis was performed with hepatocyte extracts (Fig. 9B). When anti-hnRNP L antibody was used for IP, HuR was detected in hepatocyte extracts by western blot analysis. Taken together, these results suggest that HuR interacts directly with hnRNP L.
Then, we examined whether HuR binds to the iNOS AS transcript, which harbors two AREs (Fig. 4A). RNA-EMSA was performed with a biotinylated AS RNA probe. When HuR protein was added, a shifted band (Fig. 9C; an arrow) was observed, and nonbiotinylated AS RNA competed out. When anti-HuR antibody was added, signals of RNA-HuR bands decreased. This clearly indicates that HuR specifically binds to the iNOS AS transcript in vitro.
In contrast, all the known binding sites of hnRNP L13–15 were absent in iNOS AS transcript. To clarify whether the iNOS AS transcript indirectly interacts with hnRNP L through HuR, we performed RNA-EMSA with the iNOS AS RNA (Fig. 9D). When hnRNP L protein was added, the AS RNA band did not change the mobility (lane 2). In the presence of both hnRNP L and HuR, a shifted band appeared (lane 4; arrow). When anti–hnRNP L or anti-HuR antibody was added, the signal intensity of the band decreased (lanes 4-6). These results revealed that iNOS AS transcript bound to hnRNP L only in the presence of HuR, suggesting that iNOS AS transcript does not interact with hnRNP L without HuR. The data of IP–RT-PCR analyses (Fig. 8) also support these results. Because iNOS mRNA interacted with the AS transcript (Figs. 5 and 6) and HuR, it is likely that a complex including iNOS mRNA, the AS transcript, and the RNA-binding proteins may be formed (Fig. 9E).
Our study suggests that the natural AS transcript participates in a posttranscriptional mechanism to control mRNA stability. Both the nucleotide sequence, including AREs, and the secondary structure of iNOS mRNA 3′UTR are conserved in the rat, mouse, and human. Mouse and human iNOS AS transcripts corresponding to the 3′UTR were also detected (M. Nishizawa and T. Okumura, unpublished data, 2007). It is feasible that the iNOS AS transcript is common in various species. AS transcripts may be present also in other ARE-containing mRNAs, such as mRNAs for early-response genes encoding acute phase proteins, cytokines, and some proto-oncogenes. On the other hand, it has been reported that micro-RNA destabilizes tumor necrosis factor α (TNF-α) mRNA.31 In iNOS mRNA 3′UTR, when searching the micro-RNA sequence database miRBase, we could not find any sequences that might bind to known micro-RNA.32
The present study suggests that RNA-RNA, RNA-protein, and protein-protein interactions are involved in the formation of the iNOS mRNA–AS transcript–protein complex. Of these, the interaction between iNOS mRNA and AS transcript is important because the competition by iNOS sense oligonucleotides lowered iNOS mRNA stability, caused iNOS mRNA degradation, and reduced iNOS protein and NO production (Fig. 5). It might be possible that the sense oligonucleotides mask the binding sites of RNA-binding proteins. HuR binds ARE motifs; hnRNP L preferentially binds to the CA repeat regions and the 12-nt conserved sequence; and PTB binds to a CU-rich region of human iNOS mRNA, which is equivalent to the CU-rich region between regions A and B,12–16 as shown in Fig. 4A. The sense oligonucleotides S1 to S5 do not include any known binding sites of these RNA-binding proteins or their complementary sequences. Therefore, it is unlikely that the sense oligonucleotides directly interfere with the binding of HuR, hnRNP L, and PTB.
When iNOS AS transcript was overexpressed in the hepatocytes, it stabilized the luciferase mRNA through the iNOS 3′UTR (Fig. 6). As shown in Fig. 4B, iNOS mRNA may cross-hybridize with the AS transcript at the loops in various combinations. The intermolecular loop-loop interaction of RNA plays a major role in RNA-protein complex formation. This RNA-RNA interaction is required for the formation of bicoid mRNA-STAUFEN ribonucleoprotein particles,33 replication of human immunodeficiency virus type I,34 and assembly of splicesome and ribosome.35, 36 The present study suggests that iNOS AS transcript interacted with HuR, which then interacted with hnRNP L (Fig. 9A-D). Similarly, it would be plausible to speculate that PTB binds to the iNOS AS transcript in an indirect manner because hnRNP L binds to PTB.15 On the other hand, the iNOS mRNA interacted with not only iNOS AS transcript but also HuR (Figs. 5 and 6). Accordingly, it is possible that the iNOS AS transcript plays a role in recruiting RNA-binding proteins and then promotes protein-protein interaction in the iNOS mRNA-protein complex (Fig. 9E).
The mRNA is degraded by (1) decapping and 5′3′ exonuclease digestion and (2) deadenylation and 3′5′ exonuclease digestion.37 It is feasible that the iNOS mRNA–AS transcript–protein complex inhibits access of a deadenylation enzyme and the 3′5′ exonuclease (Fig. 9E). A functional link between the cap and the poly(A) tail37 may be involved in mRNA protection from the 5′3′ degradation. The ARE-binding protein HuR stabilizes ARE-containing mRNA, such as c-fos and IL-3 mRNAs,38 and other ARE-binding proteins are involved in the degradation of human iNOS mRNA.39 Hence, it seems likely that the iNOS mRNA–AS transcript–protein complex suppresses mRNA degradation by controlling deadenylation and decapping. Similar to the assembly of the splicesome and ribosome, the RNA-binding proteins may associate with iNOS mRNA and AS transcript in a highly ordered fashion. A further analysis of the formation of functional RNA-protein complex will clarify the detailed mechanism of iNOS mRNA stabilization.
The knockdown of iNOS AS transcript (Fig. 5) can be applied to treat inflammatory diseases including septic shock, autoimmune diseases including chronic rheumatoid arthritis, cachexia,40 and cancer. When we injected LPS/D-galactosamine into rats as a model of septic shock, both iNOS mRNA and AS transcript were detected in the liver (M. Nishizawa and T. Okumura, unpublished data, 2007). In vivo administration of iNOS sense oligonucleotides or short interfering RNA may selectively knock down iNOS mRNA, and this method may be applied to other early-response mRNAs regulated by natural AS transcripts. Therefore, the interaction between iNOS mRNA and AS transcript may be a new therapeutic target for various inflammatory diseases in which NO mediates.
- 15Inflammation modulates the interaction of heterogeneous nuclear ribonucleoprotein (hnRNP) I/polypyrimidine tract binding protein and hnRNP L with the 3′-untranslated region of the murine inducible nitric-oxide synthase mRNA. Mol Pharmacol 2002; 62: 423–431., , .