Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The inducible nitric-oxide synthase (iNOS) is up-regulated by immunologic and inflammatory stimuli, such as lipopolysaccharides (LPS) or cytokines, resulting in increased production of nitric oxide (NO) (for review, see Rao, 2000). Induction of iNOS is part of the antimicrobial and tumoricidal actions of macrophages (Nathan and Hibbs, 1991; Cui et al., 1994) but may also have detrimental effects, such as tissue damage observed in, for example, arthritis, type I diabetes, and septic shock (Corbett and McDaniel, 1992; McCartney-Francis et al., 1993; Morikawa et al., 1999). The diversity of NO effects implies a tight regulation of its production.

Research to date has shown that regulation of the iNOS gene is complex, implicating transcriptional, post-transcriptional/translational, and post-translational mechanisms (Geller and Billiar, 1998), and several consensus sequences and corresponding trans-acting factors contributing to the transcriptional control of both the human and murine iNOS genes have been described previously (Lowenstein et al., 1993; Xie et al., 1993; de Vera et al., 1996; Chu et al., 1998).

Compared with its transcriptional control, the post-transcriptional regulation of the iNOS gene is poorly understood yet probably an important part of the overall regulation. For example, in mouse peritoneal macrophages, transforming growth factor 1 suppresses iNOS expression by decreasing iNOS mRNA stability and translational efficiency and by increasing the degradation of iNOS protein (Vodovotz et al., 1993). On the other hand, iNOS mRNA stability has been shown to increase during induction of iNOS expression (Weisz et al., 1994). In none of the cases, however, are the mechanisms behind the events known. The 3'untranslated region (UTR) of both human and murine iNOS contains conserved AU-rich sequences, known to mediate mRNA instability in many labile cytokine and proto-oncogene transcripts (Evans et al., 1994; Ross, 1995). Whether these AU-rich sequences play a role in the iNOS mRNA stability remains unknown (Geller and Billiar, 1998). In one report, Rodriguez-Pascual et al. (2000) show the binding of the Drosophila melanogaster ELAV (embryonic lethal abnormal vision)-like protein HuR to AU-rich motifs in the 3'UTR of human iNOS mRNA and speculate that it could mediate the post-transcriptional events during the cytokine-induced expression of the human iNOS.

In the present study, we analyze the interaction of cytoplasmic liver proteins with the murine iNOS mRNA. We demonstrate that inflammation affects the interaction of a multiprotein complex formation with the 3'UTR of the iNOS mRNA. The proteins in the complex seem to be heterogeneous nuclear ribonucleoprotein I (hnRNP I) (also named polypyrimidine tract binding protein, PTB) and hnRNP L, and they bind at two distinct regions of the 3'UTR, both easily accessible for trans-acting factors. The possible novel functions of hnRNP I/PTB and hnRNP L as regulators of the iNOS expression are discussed.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Reagents. D-Galactosamine was from ICN Pharmaceuticals (Costa Mesa, CA), RNase A from Roche Diagnostics Scandinavia (Bromma, Sweden), RNase T1 from Invitrogen AB (Lidingö, Sweden), Amplitaq Gold and dNTP mix from PerkinElmer Life Sciences (Boston, MA), and T7 RNA polymerase from Promega (Madison, WI). Isotopes ([-³²P]UTP and [-³²P]dCTP) were from Amersham Biosciences (Piscataway, NJ). Protease inhibitors (phenylmethylsulfonyl fluoride, leupeptine and DTT), yeast tRNA, lipopolysaccharide (Escherichia coli, serotype 0127:B8), and homoribopolymers [poly(A), poly(U), poly(C) and poly(G)] were from Sigma-Aldrich Sweden AB (Stockholm, Sweden).

Antibodies and Oligodeoxyribonucleotides. Monoclonal antibodies 4D11 (anti-hnRNP L) and 4F4 (anti-hnRNP C) were generously provided by Dr. Gideon Dreyfuss (Howard Hughes Medical Institute, University of Pennsylvania, Philadelphia, PA). Anti-hnRNP I/PTB was purchased from Zymed Laboratories (South San Francisco, CA). All of the oligodeoxyribonucleotides used in the study were from Sigma Genosys (Pampisford, Cambridgeshire, UK).

Animals. Male DBA/2J mice, 6 to 10 weeks old, were provided by Möllegaard (Ejby, Denmark). The mice were kept at the animal facility at the Biomedical Centre in Uppsala and fed chow and water ad libitum. They were allowed to acclimatize for 1 week before treatment. The mice (~25 g) were treated intraperitoneally with 100 ng of LPS and 10 mg of D-galactosamine dissolved in saline. Control mice received saline only. After 6 h of treatment, the animals were sacrificed and livers were removed. The studies were approved by the Ethical Committee in Uppsala (Sweden) with approval number C3/1 and performed accordingly.

RNA Isolation and Northern Blot Analysis. Total RNA was isolated from the liver samples using an RNeasy Midi kit (QIAGEN GmbH, Hilden, Germany) according to the manufacturer's protocol. Total RNA (20 µg) was subjected to electrophoresis in 1.2% agarose/formaldehyde gel, transferred to a Hybond-N nylon membrane (Amersham Biosciences AB, Uppsala, Sweden) and UV cross-linked before hybridization. The 4-kilobase NotI fragment, excised from a plasmid containing the murine iNOS cDNA (kindly provided by Dr. Charles J. Lowenstein, John Hopkins University, Baltimore, MD), was radiolabeled with [-³²P]dCTP using the Megaprime labeling kit (Amersham Biosciences AB). Unincorporated nucleotides were removed by using a G-50 column (Amersham Biosciences AB). Both prehybridization and hybridization were performed at 65°C in Church buffer (Church and Gilbert, 1984) modified to contain 0.25 M NaHPO₄, pH 7.2, 7% SDS, and 1 mM EDTA. Hybridization was performed overnight with 1.7 × 10⁷ cpm of radiolabeled probe. To assess equal loading of the samples, the mRNA level of the house-keeping gene glyceraldehyde-3-phosphate-dehydrogenase (GAPDH) was measured using the radiolabeled GAPDH cDNA (BD Clontech, Palo Alto, CA).

Preparation of Liver Cytoplasmic Extracts. For preparation of the crude cytoplasmic protein extracts, excised mouse liver was homogenized in 15 mM HEPES buffer, pH 7.6, containing 3 mM MgCl₂, 40 mM KCl, 1 mM DTT, 0.5 mM phenylmethylsulfonyl fluoride, 10 ng/µl leupeptine, and 0.4% igepal. The homogenate was centrifuged at 12,000g for 10 min at 4°C. The supernatant, corresponding to the cytoplasmic extract, was aliquoted and stored at 80°C until further use. Protein concentration was measured by the method of Lowry et al. (1951).

In Vitro Transcription. Using different sets of primers, cDNAs encoding various subfragments of murine iNOS 3'UTR and coding sequence (see Fig. 2, Table 1) were obtained by PCR using the iNOS-containing plasmid as a template. All PCR reactions were performed with Taq DNA polymerase. After 10 min at 95°C followed 30 cycles: 94°C, 1 min; 52°C, 1 min; 72°C, 1 min. The nucleotide positions of the primers refer to the cDNA map as published by Lowenstein et al. (1992). Primer 2 hybridizes to the pBluescript plasmid, 29 nt downstream of the NotI cloning site. All fragments produced with primer 2 were further subjected to cleavage with NotI to eliminate the plasmid sequence. Sense oligonucleotides contained 23 nt corresponding to the T7 RNA polymerase promoter. PCR-amplified products were transcribed with T7 RNA polymerase in the presence of [-³²P]UTP (800 Ci/mmol) using standard protocols (Promega). The 218- and 81-nt fragments were cut with the restriction enzymes EarI and SspI before use as templates for in vitro transcription, to generate the resulting 112- and 41-nt probes, respectively. Unincorporated nucleotides were removed by using a G-50 column (Amersham Biosciences AB). Unlabeled RNA competitors were transcribed under similar conditions, with the radiolabeled nucleotide substituted with 0.5 mM UTP.

View this table: [in this window] [in a new window]   TABLE 1 Primers used for PCR The different sets of primers used to amplify various fragments from iNOS coding sequence and 3'UTR are shown, as seen in Figure 2. T7 represents the T7 RNA polymerase promoter sequence 5'-TAATACGACTCACTATAGGGAGA-3'. The nucleotide assignments correspond to the iNOS sequence in GenBank reference number M92649 (Lowenstein et al, 1992).

UV Cross-Linking. Binding reactions were carried out for 12 min at room temperature using 200,000 cpm of ³²P-labeled RNA transcript and 25 µg of protein extracts, in a total volume of 20 µl containing 10 mM HEPES, pH 7.6, with 3 mM MgCl₂, 40 mM KCl, 5% glycerol, 1 mM DTT, and 1 µg of yeast tRNA. The reaction mixture was then placed on ice and exposed to UV light for 20 min in a Spectrolinker XL-1000 UV cross-linker (Spectronics, Westbury, NY). Subsequently, unbound RNA was digested with 2 µg of RNase A at 37°C for 20 min. The samples were denatured under nonreducing conditions and separated by SDS/PAGE (12%). Finally, the gel was dried and autoradiographed overnight. For competition experiments, the protein extract was preincubated for 5 or 10 min with the indicated unlabeled competitors.

Immunoprecipitation. The UV cross-linked complexes were immunoprecipitated essentially as described by Hamilton et al. (1993). A typical UV cross-linking was performed with 40 µg of liver cytoplasmic extract and 200,000 cpm of radiolabeled 112- or 81-nt probes, respectively. After RNase A digestion, the RNA-protein complexes were incubated with a 1:500 dilution of anti-hnRNP L (4D11), anti-hnRNP I/PTB, or anti-hnRNP C (4F4) (used as a control antibody) monoclonal antibodies for 2h at 4°C. The samples were then incubated with 30 µl of protein A-Sepharose beads (Amersham Biosciences AB) for 1 h at 4°C. The beads were recovered by brief centrifugation and washed three times in PBS. The proteins were denatured and analyzed by SDS/PAGE (12%). After drying, the gels were exposed to X-ray films.

RNA Mobility Shift Assay and Supershift Assay. Binding conditions were the same as described above, using 5 or 25 µg of liver cytoplasmic extract, as indicated. After binding, the reaction mixtures were treated with RNase T1 (20 units) for 20 min at room temperature. The RNA-protein complexes were resolved on nondenaturing 7% polyacrylamide gels (acrylamide/bisacrylamide ratio, 37.5:1) with 50 mM Tris/glycine, pH 8.8, as running buffer. The gels were dried and visualized by autoradiography. For competition experiments, unlabeled competitors were added 10 min before the addition of the radiolabeled RNA. In supershift assays, 5 µg of protein extract was incubated with the RNA probe as described above. After RNase T1 treatment, 1 µl of anti-hnRNP L (4D11) anti-hnRNP I/PTB or anti-hnRNP C (4F4) monoclonal antibodies, respectively, were added to the mixtures and incubated for another 10 min, before loading on a native gel, as described by Min et al. (1995). In several cases, lanes of one or several gels from the same UV cross-linking or RNA mobility shift experiment with identical exposure times, have been combined into one figure.

Computer Analysis. The Mfold program was used to perform computer secondary structure predictions of the iNOS 3'UTR sequence (Mathews et al., 1999; Zuker et al., 1999).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Demonstration of High-Affinity Protein Binding to the 3'UTR of the Murine iNOS mRNA, and Dependence of the RNA-Protein Complex Formation on Septic Shock. Mice treated with LPS and D-galactosamine develop septic shock (Barton and Jackson, 1993), a condition that induces the iNOS gene (Morikawa et al., 1999). This was used to mimic an infection and to achieve up-regulation of iNOS mRNA. As expected, an increase of iNOS mRNA was seen after 6 h of treatment (Fig. 1).

View larger version (29K): [in this window] [in a new window]   Fig. 1.   iNOS mRNA induction. Liver RNA (20 µg) prepared from mice (~25g), untreated () or "septic shock"-treated (LPS/Gal) (+) i.p. with LPS (100 ng) and D-galactosamine (10 mg) for 6 h, were used for Northern blot analysis. Control animals received saline vehicle only. Hybridization was performed with 1.7 × 106 cpm of radiolabeled iNOS 4-kilobase cDNA fragment. GAPDH mRNA levels are shown as controls for RNA loading.

View larger version (13K): [in this window] [in a new window]   Fig. 2.   iNOS mRNA fragments used for RNA-protein interaction analysis. The schematic of the iNOS 3'UTR, part of the coding sequence (CS) and the individual transcripts used in the RNA binding studies are shown. The oligonucleotides used for amplification in the PCR reactions are indicated on the right; see Table 1 for details of the oligonucleotides used. The number of nucleotides for each probe is indicated. P2 is an oligonucleotide located in the plasmid sequence where iNOS cDNA is cloned. All transcripts generated with P2 were therefore cut with NotI after PCR amplification, to remove the 29-nt plasmid sequence. The 218- and 81-nt templates were cut with the restriction enzymes EarI and SspI, before in vitro transcription, to generate the 112- and 41-nt probes, respectively.

View larger version (59K): [in this window] [in a new window]   Fig. 3.   Protein complex formation with the iNOS 3'UTR is sensitive to septic shock treatment. RNA gel mobility shift assay with 32P-radiolabeled iNOS 3'UTR RNA full-length probe (493 nt; see Fig. 2) incubated with 25 µg of cytoplasmic protein extract from mice untreated (-) or treated (+) with LPS (100 ng) and D-galactosamine (10 mg) for 6 h. Control animals received saline vehicle only. The third lane contains no extract. The positions of the free RNA probe and the RNA-protein complexes I () and II () are indicated.

View larger version (46K): [in this window] [in a new window]   Fig. 4.   Determination of the apparent molecular masses of the proteins binding to the iNOS 3'UTR mRNA. iNOS 3'UTR full-length RNA probe was incubated with 25 µg of cytoplasmic protein extract from untreated () or "septic shock"-treated (LPS/Gal) (+) mice, UV cross-linked, and resolved on SDS/PAGE (12%) as described under Materials and Methods. Molecular mass (MW) standards in kilodaltons are indicated on the right. The 70- and 60-kDa complexes are marked with arrows.

View larger version (50K): [in this window] [in a new window]   Fig. 5.   Specificity of the RNA-protein complex formation. A, RNA gel mobility shift assay was carried out with the 32P-radiolabeled iNOS 3'UTR RNA full-length probe (493 nt) incubated without extract in the first lane (free RNA probe) or with 25 µg of protein extract from untreated animals in the absence () or presence of 1 or 5 µg of unlabeled yeast tRNA; 0.5 or 1 µg of unlabeled iNOS 3'UTR RNA probe, or with 0.5 or 1 µg of unlabeled iNOS coding sequence (CS) RNA probe (660 nt) as shown. The unlabeled competitor was added 10 min before the radioactive probe. Complex I is shown. B, UV cross-linking assay with the 32P-radiolabeled iNOS 3'UTR RNA full-length probe (493 nt) and 25 µg of cytoplasmic protein extract from untreated mice, incubated in the absence () or presence of 1 or 5 µg of unlabeled yeast tRNA; 135 or 300 molar excess of unlabeled iNOS 3'UTR RNA probe, or with 135 or 300 molar excess of unlabeled iNOS coding sequence (CS) RNA probe, as indicated. The 60- and 70-kDa complexes are shown. The unlabeled competitor was incubated for 10 min with the protein extract before addition of the radiolabeled probe. Molecular mass (MW) standards in kilodaltons are indicated on the right.

Mapping of the Binding Regions for Proteins Involved in the 60- and 70-kDa Complexes. The strategy for mapping the protein binding sites at the iNOS 3'UTR is shown in Fig. 2 and Table 1. A total of 11 RNA probes were prepared and used in both gel mobility shift and UV cross-linking analyses. RNA probes of 158 and 348 nt, respectively, were first created from two sections of the iNOS 3'UTR sequence, as shown in Fig. 2. We detected the gel-shifted complex, resembling the native complex, only with the longer probe, as shown in Fig. 6A. Using smaller probes, the same complex was seen with the 218-nt sequence, whereas the 154-nt probe gave a higher complex also sensitive to septic shock (Fig. 6A). With the 112-nt probe, we still obtained a complex similar to that seen with the 3'UTR probe, whereas the 81-nt probe showed a profile partly similar to the 154-nt probe (Fig. 6a). It is noteworthy that with both the 348-, 218-, and 112-nt probes, the sensitivity of complex formation to septic shock, is conserved.

View larger version (67K): [in this window] [in a new window]   Fig. 6.   Mapping of the protein binding sites. A, gel mobility shift assay with the 3'UTR 493-, 158-, 348-, 218-, 154-, 112-, and 81-nt RNA probes (see Fig. 2, Table 1) incubated with 5 µg of cytoplasmic protein extract from untreated () or "septic shock" (LPS/Gal) treated (+) mice. The first lane for each RNA probe contains no protein extract. The position of the free RNA probes and the RNA-protein complexes I () and II () are shown on the left. Closed arrows for the 81-nt complexes are indicated on the right. B, UV cross-linking with 32P-radiolabeled 348-, 154-, 218-, 73-, 81-, 112-, 106-, 41-, and 43-nt RNA probes (see Fig. 2, Table 1) incubated with 25 µg of protein extract from untreated () or "septic shock" (LPS/Gal) treated (+) mice as described under Materials and Methods. Molecular mass (MW) standards in kilodaltons are indicated on the right, and the 70- and 60-kDa complexes are marked with arrows. Only proteins from untreated animals were used.

View larger version (32K): [in this window] [in a new window]   Fig. 7.   Inhibition of the 60-kDa RNA-protein complex formation with antisense oligodeoxyribonucleotides. A, the iNOS 112-nt RNA sequence with the location of the 6 antisense (AS) oligodeoxyribonucleotides AS1 to AS6 is indicated. B, UV cross-linking analysis using 25 µg of cytoplasmic protein extract from untreated mice. The 32P-radiolabeled 112-nt RNA probe was preincubated for 5 min in the absence () or presence of each of the six AS oligodeoxyribonucleotides (170 nM), as indicated, before addition of the protein extract. RNA-protein complexes were resolved on SDS/PAGE (12%). The 60-kDa complex is shown. Molecular mass (MW) standards in kilodaltons are indicated on the right.

View larger version (26K): [in this window] [in a new window]   Fig. 8.   Inhibition of the 70-kDa RNA-protein complex formation with antisense oligodeoxyribonucleotides. A, the iNOS 81-nt RNA sequence with the location of the five antisense (AS) oligodeoxyribonucleotides AS'1 to AS'5 is indicated. B, UV cross-linking analysis using 25 µg of cytoplasmic protein extract from untreated mice. The 32P-radiolabeled 81-nt RNA probe was preincubated for 5 min in the absence () or presence of each of the five AS oligodeoxyribonucleotides (170 nM), as indicated, before addition of the protein extract. RNA-protein complexes were resolved on SDS/PAGE (12%). The 70-kDa complex is shown. Molecular mass (MW) standards in kilodaltons are indicated on the right.

View larger version (72K): [in this window] [in a new window]   Fig. 9.   Homoribopolymer affinity of the 60- and 70-kDa complexes. A, UV cross-linking competition assays were carried out with the 32P-radiolabeled 112-nt fragment in the presence of 25 µg of cytoplasmic extract from untreated mice without () or with 10, 50, or 100 ng of unlabeled homoribopolymers poly(A), poly(U), poly(C), or poly(G), as indicated. Molecular mass (MW) standards in kilodaltons are indicated on the right. The 60-kDa complex is shown. B, UV cross-linking competition assays were carried out with the 32P-radiolabeled 81-nt RNA probe in the presence of 25 µg of cytoplasmic extract from untreated mice without () or with 10, 50, or 100 ng of unlabeled homoribopolymers poly(A), poly(U), poly(C), or poly(G), as shown. Molecular mass (MW) standards in kilodaltons are indicated on the right. The 70-kDa complex is shown.

View larger version (20K): [in this window] [in a new window]   Fig. 10.   Mfold computer predicted secondary structure of the 3'UTR of iNOS mRNA. The secondary structure of iNOS 3'UTR mRNA was predicted using the Mfold program (Mathews et al., 1999; Zuker et al., 1999). The structure in its most stable conformation is shown (minimum free energy dG = 109.7 kcal/mol). The putative binding sites for the proteins involved in the 60- and 70-kDa RNA-protein complexes, as defined by the covering of AS2-4 and AS'3, respectively, are indicated. Both are predicted to be situated in exposed loops, easily accessible for interactions with trans-acting factors.

Identification of the 60- and 70-kDa Proteins Binding to the 3'UTR of iNOS. Based on the characteristics of the smaller RNA-protein complex (an apparent molecular mass of 60 kDa, a high affinity for polyuridines, and a polypyrimidine-rich sequence), we postulated that this protein may be one form of the previously described hnRNP I/PTBs, 57- to 62-kDa multifunctional gene regulators (Hellen et al., 1993; Singh et al., 1995). To test this hypothesis, we performed immunoprecipitation of the UV cross-linked 112-nt RNA-protein complex. As shown in Fig. 11A, we were able to immunoprecipitate the 60-kDa complex with the anti-hnRNP I/PTB monoclonal antibody but not with the anti-hnRNP C or hnRNP L antibodies, indicating that indeed this protein could be one of the variants of the hnRNP I/PTB.

View larger version (29K): [in this window] [in a new window]   Fig. 11.   Immunoprecipitation of the proteins in the 60- and 70-kDa RNA-protein complexes. A, UV cross-linking was performed with 200,000 cpm of 32P-radiolabeled 112-nt RNA probe and 40 µg of liver cytoplasmic protein extract from untreated mice, followed by RNase A digestion. Immunoprecipitation was performed as described under Materials and Methods, without antibody () or with 1:500 dilution of the monoclonal antibodies anti-hnRNP C (C), anti-hnRNP I/PTB (I), or anti-hnRNP L (L), as indicated. A reference sample (UV cross-linking followed directly by SDS/PAGE) (Co) was included in the gel. The UV cross-linked lane was exposed on film for 2 days, whereas the immunoprecipitated samples were exposed for 1 week. Molecular mass (MW) standards in kilodaltons are indicated on the left. The 60-kDa complexes are marked with arrows. B, UV cross-linking was performed with 200,000 cpm of 32P-radiolabeled 81-nt RNA probe and 40 µg of liver cytoplasmic protein extract from untreated mice, followed by RNase A digestion. Immunoprecipitation was performed as described under Materials and Methods, without antibody () or with 1:500 dilution of the monoclonal antibodies anti-hnRNP C (C), anti-hnRNP I/PTB (I), or anti-hnRNP L (L), as indicated. A reference sample (UV cross-linking followed directly by SDS/PAGE) (Co) was included in the gel. The UV cross-linked lane was exposed on film for 2 days, whereas the immunoprecipitated samples were exposed for 1 week and analyzed using autoradiographic imaging analysis. Molecular mass (MW) standards in kilodaltons are indicated on the left. The 60- and 70-kDa complexes are marked with arrows.

View larger version (77K): [in this window] [in a new window]   Fig. 12.   Antibody supershift of the gel shift complex. The 32P-radiolabeled 3'UTR 493-nt probe incubated in the absence () or in the presence (+) of 5 µg of cytoplasmic protein extract as described for the gel mobility shift assay, was incubated without () or with 1 µl of the monoclonal antibodies anti-hnRNP C (C), anti-hnRNP I/PTB (I), or anti-hnRNP L (L), as indicated. Samples (10 µl) were loaded on the gel. The positions of the complexes I () and II () are indicated. The supershifted complex is marked on the right. The position of the free RNA probe is shown.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Several lines of evidence suggest that post-transcriptional events are important in the regulation of the iNOS expression, but only little is known of their molecular mechanisms (Vodovotz et al., 1993; Weisz et al., 1994; Rodriguez-Pascual et al., 2000). In this study, we identified two liver cytoplasmic proteins of 60 and 70 kDa, interacting with the 3'UTR of the murine iNOS mRNA, and our evidence strongly indicates that these proteins are hnRNP I/PTB and hnRNP L, respectively. Their binding to the 3'UTR of the mRNA is affected by septic shock, which suggests a role in the post-transcriptional events of iNOS induction by inflammation.

The RNA binding characteristics of the 60-kDa protein indicates that it may be one variant of the hnRNP I/PTB (Hellen et al., 1993; Singh et al, 1995). Its binding site at the iNOS mRNA is a polypyrimidine-rich sequence, showing a high nucleotide identity with known PTB consensus sequences (Singh et al., 1995): the 112-nt sequence at the iNOS 3'UTR, containing the high-affinity binding site, includes a 15-nt stretch in the 5' end with 87% identity to the known PTB consensus sequence. In addition, a region in the middle of the 112-nt sequence show 77% identity with a 26-nt sequence in GAP-43 3'UTR mRNA, which has been reported to bind a PTB-like protein (Irwin et al., 1997). Another sequence, similar to the known hnRNP I/PTB binding sites, is found 42 nt upstream of the possible hnRNP L binding site in the iNOS 3'UTR. This may explain why the 60-kDa protein was found to interact at two distinct regions of the 3'UTR (Figs. 8 and 11B).

Similarly, the size of the other protein (70 kDa) binding to iNOS 3'UTR and the 70% nt identity of its binding site at the iNOS mRNA with a known binding site of hnRNP L in vascular endothelial growth factor 3'UTR mRNA, led us to hypothesize that this protein is the hnRNP L (Shih and Claffey, 1999).

Immunoprecipitation of the UV cross-linked complexes and supershift of the native complex confirmed that the two proteins binding to the 3'UTR of the iNOS mRNA are the hnRNP I/PTB and hnRNP L or highly similar proteins. The absence of strict consensus binding sequences in the different mRNA target molecules described in the literature suggests that the proteins have some flexibility in their RNA sequence recognition. Such flexibility has been shown for the hnRNP A1 (Burd and Dreyfuss, 1994). Alternatively, the two proteins described here could be close structural analogs of the hnRNP I/PTB and L species described previously, with modified RNA binding characteristics. This is possible, particularly in the case of hnRNP I/PTB, for which several variants have been described previously (Gil et al., 1991; Markovtsov et al., 2000).

Previous studies have shown that hnRNP I/PTB interacts with hnRNP L (Hahm et al., 1998a). Our studies are in agreement with the previous observations, although they do not reveal the exact nature of the interaction. When the major complex appearing in the gel-shift assay is UV cross-linked and resolved by SDS/PAGE, two bands of approximately 60 and 70 kDa are detected, showing that both proteins are present in the native complex (data not shown). Furthermore, the poly(U) homoribopolymer efficiently prevented the formation of the entire gel shifted complex (data not shown). Yet, of the two proteins described here, only the 60-kDa protein binding to the 3'UTR was poly(U) sensitive, thus suggesting that this protein is critical for the multiprotein complex formation. Also, the lack of native complex formation with the 81-nt probe, which contains the binding site for the hnRNP L, and where the hnRNP I/PTB seems to interact only weakly, suggests that the high-affinity binding of hnRNP I/PTB with the RNA is prerequisite for the binding of hnRNP L in the native complex. It is therefore interesting to note that only the anti-hnRNP L antibody caused a supershift of the native complex. This could indicate that whereas the hnRNP I/PTB regulates the interaction of the entire complex with the RNA, the hnRNP L is more exposed at the surface of the multiprotein complex.

So far, only one trans-acting factor has been described interacting with the 3'UTR of iNOS mRNA: Rodriguez-Pascual et al. (2000) have shown that the HuR protein interacts with AUUUA rich sequences of the iNOS mRNA in human DLD-1 cells and hypothesize that the protein may take part in mRNA stabilization. We did not find any evidence for the binding of HuR to mouse iNOS mRNA, and neither of the two proteins described here had affinity for AUUUA motifs (data not shown). An explanation for these essentially different results could be that trans-acting factors interacting with iNOS mRNA can be cell- and species-specific.

The hnRNP I/PTB seems to be involved in several processes, such as alternative splicing (Singh et al., 1995), viral replication (Chung and Kaplan, 1999), alteration of RNA conformation (Huang and Lai, 1999) and cap-independent translation through internal ribosomal entry sites (Hellen et al., 1993). The protein has been shown to shuttle between the cytoplasm and the nucleus in a transcription-linked manner (Michael et al., 1995), suggesting distinct roles in the different subcellular compartments. Irwin et al. (1997) propose that the interaction of hnRNP I/PTB with the 3'UTR of the bovine brain GAP-43 mRNA could stabilize the transcript and Kim et al. (2000b) show that the hnRNP I/PTB inhibits translation of Bip (an immunoglobulin heavy-chain binding protein) dependent on internal ribosomal entry sites.

The hnRNP L is highly homologous to hnRNP I (Piñol-Roma et al., 1989) and is also found to shuttle between the nucleus and the cytoplasm in a transcription-dependent manner (Michael et al., 1995; Hahm et al., 1998b). Several functions have been assigned to hnRNP L: e.g., modulation of the human vascular endothelial growth factor labile mRNA stability under hypoxic conditions (Shih and Claffey, 1999) and the translation of hepatitis C virus mRNA (Hahm et al., 1998b).

Both hnRNP I/PTB and hnRNP L contain four loosely conserved RNA recognition motifs, and it has been shown that they may form homodimers and/or heterodimers with other hnRNPs [e.g., the hnRNP E2, I, K and L (Kim et al., 2000a)]. It has therefore been suggested that some of their effects are exerted in concert (Hahm et al., 1998a). Potential roles of the heterodimer on RNA processing, mRNA nucleocytoplasmic transport and/or translation of some mRNAs have been proposed. It is also possible that the hnRNP I/PTB-hnRNP L complex has an RNA-binding specificity different from the individual proteins and thus exert a different function compared with the individual proteins (Hahm et al., 1998a).

In conclusion, we describe for the first time the interaction of hnRNP I/PTB and hnRNP L with the murine iNOS mRNA and the modulation of this interaction by inflammation, a strong inducer of the iNOS expression. The modulated interaction with the iNOS mRNA suggests a novel function for hnRNP I/PTB and hnRNP L. We do not know how septic shock affects the affinity of the proteins for the iNOS mRNA, but, based on their previously described functions, we can speculate on their possible roles in the iNOS induction. The hnRNP I/PTB could inhibit translation during resting conditions alone, as shown in the case of the Bip mRNA (Kim et al., 2000b), or concertedly with hnRNP L (Kim et al., 2000a). Alternatively, the two proteins may participate in the rapid iNOS mRNA degradation under noninduced conditions. Indeed, we detect a strong RNA-protein complex in resting cells where the iNOS transcripts are hardly detectable, even though significant transcription of the iNOS gene is going on (de Vera et al., 1996; Linn et al. 1997; Rodriguez-Pascual, 2000). The discrepancy between the high transcription level and low amounts of iNOS mRNA in resting cells can be explained by an efficient degradation of the iNOS transcript (Rodriguez-Pascual et al., 2000). In inflammation, this could be reversed, resulting in stimulation of gene expression (at least in part) via an increase in mRNA stability. How the inflammation-modulated binding of hnRNP I/PTB and hnRNP L to the iNOS mRNA relates to the mRNA stability and other post-transcriptional events is a focal point of our future research.