Regulation of Tyrosine Hydroxylase mRNA Stability by Protein-binding, Pyrimidine-rich Sequence in the 3 * -Untranslated Region*

The stability of mRNA for tyrosine hydroxylase (TH), the rate-limiting enzyme in catecholamine synthesis, is regulated by oxygen tension in the pheochromocytoma-derived PC12 cell line. We previously identified a pyri-midine-rich 27-base-long protein-binding sequence in the 3 * -untranslated region of TH mRNA that is associated with hypoxia-inducible formation of a ribonucleoprotein complex (hypoxia-inducible protein-binding site (HIPBS)). In this study, we show that HIPBS is an mRNA stabilizing element necessary for both constitutive and hypoxia-regulated stability of TH mRNA. The mutations within this sequence that abolish protein binding markedly decrease constitutive TH mRNA stability and ablate its hypoxic regulation. A short fragment of TH mRNA that contains the wild-type HIPBS confers the increased mRNA stability to the reporter chloramphenicol acetyltransferase mRNA. However, it is not sufficient to confer hypoxic regulation. The HIPBS element binds two isoforms of a 40-kDa poly(C)-binding protein (PCBP). Hypoxia induces expression of the isoform 1, PCBP 1 , but not the isoform 2, PCBP 2 , in PC12 cells. Tyrosine hydroxylase (TH), 1 the rate-limiting enzyme in the biosynthesis of catecholamines, is expressed in specific populations of neurons in the central and peripheral nervous systems, in the neuroendocrine cells of the adrenal medulla m m M were eluted overnight at 4 a of the outside cup (50 m M Tris m M EDTA). The and outside compartments were separated by dialysis mem- brane with a molecular cut-off of 12,000–14,000 (Spectra/por*2; Spectrum Medical Industries). The eluted proteins were concentrated using Microcon-10 microconcentrators (Amicon), resuspended in the sample buffer, and subjected to SDS-PAGE analysis. Northwestern and Western Analysis— Proteins were separated on 9% SDS-PAGE and transferred onto nitrocellulose with a semidry The blotted proteins were renatured in buffer containing 10 m M Tris, 50 m M NaCl, m M EDTA, 3 Denhardt’s 1 m M DTT, and 1 mg/ml heparin for 1 h at room The blots were then hybridized with 1 3 10 6 cpm of 5 9 -labeled HIPBS oligoribo- nucleotide in the same binding buffer without DTT or heparin, but with E. coli tRNA (20 m g/ml), overnight at room temperature. The blots were then washed briefly in the same buffer and exposed to For Western blot analysis, blots were stripped in PBS with SDS for 1 h at 37 °C. The blots were blocked in Tris-buffered saline 1 Tween 20 with nonfat milk and were with the in Tris-buffered saline 1 Tween The was

Tyrosine hydroxylase (TH), 1 the rate-limiting enzyme in the biosynthesis of catecholamines, is expressed in specific populations of neurons in the central and peripheral nervous systems, in the neuroendocrine cells of the adrenal medulla and carotid body, and in cultured cell lines such as the pheochromocytomaderived PC12 cell line. Regulation of TH gene expression at the level of gene transcription is well documented. Recently, there has been growing evidence that TH mRNA is also regulated at the level of mRNA stability. TH mRNA is a stable message with a half-life that varies from 9 to 16 h in various subclones of PC12 cells (1)(2)(3). It is enhanced during differentiation of neuroblastoma cells (1) and during stimulation of the protein kinase C pathway in PC12 cells (2). In contrast, the stability of TH mRNA does not change in PC12 cells during stimulation of TH mRNA expression by dexamethasone or forskolin (3). A recent study demonstrated substantial differences in basal TH mRNA turnover rates between different neuronal populations from as short a time as 6 -7 h, in the dopaminergic neurons of the arcuate nucleus, to as long as 11-23 h, in the dopaminergic midhypothalamic neurons (4). In addition, TH mRNA is destabilized in the dopaminergic cells of the arcuate nucleus in a manner that corresponds to the rhythmic output displayed by these neurons (4).
Our laboratory demonstrated that hypoxia augments the stability of TH mRNA in PC12 cells (5). We identified a 27base-long pyrimidine-rich sequence within the TH mRNA 3Јuntranslated region (UTR) (1552-1578 bases of TH mRNA) that binds protein factors in a hypoxia-inducible manner (hypoxia-inducible protein-binding sequence (HIPBS)) in PC12 cells (6,7), catecholaminergic cells of the superior cervical ganglia, and the dopaminergic cells of the carotid body (8). Mutational analysis revealed that the optimal protein-binding site is represented by the motif (U/C)(C/U)CCCU within the pyrimidine-rich sequence, where the underlined cytidines are the core binding site (7). This motif is conserved in TH mRNAs from different species. These results imply that formation of the ribonucleoprotein complex associated with HIPBS may be involved in physiological regulation of TH mRNA stability in catecholaminergic cells. Thus, identification of the role of this element in the regulation of TH mRNA stability is critically important to further investigations in this area.
In the present study, we investigated the role of HIPBS in regulating the stability of TH mRNA during normoxia and hypoxia. We show that HIPBS is a stabilizing element necessary for maintaining constitutive and hypoxia-regulated TH mRNA half-life. We also show that the HIPBS-binding protein is represented by two isoforms of a 40-kDa poly(C)-binding protein (PCBP, also known as ␣CP or hnRNPE (9 -12)), and we demonstrate that expression of the PCBP 1 isoform is induced by hypoxia in PC12 cells.

EXPERIMENTAL PROCEDURES
Materials-All chemicals were purchased from either Fisher or Sigma and enzymes from Promega or Life Technologies, Inc.
Plasmid Constructs-The 5Ј-marked TH cDNA constructs were obtained by inserting full-length TH cDNA (see Fig. 1A), which had either a wild-type or mutated protein-binding site, into the BamHI site of pcDNA3 expression vector (Invitrogen). Expression of TH mRNA from these vectors results in exogenous TH transcripts containing 25 bases of * This work was supported in part by National Institutes of Health Grants HL51078 and HL58687 and American Heart Association Grantin-aid 9750110N. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The the TH 5Ј-UTR region and an additional 26-base-long sequence (BamHI-HindIII) upstream from the plasmid DNA (see Fig. 1A).
The chimeric CAT-TH constructs were generated by inserting the KpnI-ApaI fragment of TH 3Ј-UTR that contains either a wild-type or mutated protein-binding site into the respective restriction sites behind the chloramphenicol acetyltransferase (CAT) gene within the pcDNA3CAT vector (see Fig. 3A). The derived chimeric mRNAs are referred to, respectively, as CAT-wtTH (wild-type) or CAT-mutTH (mutant).
Stable Transfections-Transfections were performed using Lipo-fectAMINE (Life Technologies Inc.), according to the manufacturer's protocol. Briefly, 10 g of plasmid DNA and 10 l of LipofectAMINE were combined in a polystyrene tube containing 800 l of Dulbecco's modified Eagle's medium/F-12 without serum or antibiotic. This mixture was added to 60% confluent PC12 cells growing on a 100-mm 2 dish. After 5-7 h, fetal bovine serum and antibiotics were added to achieve a concentration identical to that used for maintaining the cells. The cells were allowed to recover for 72 h and were passaged 1:4 into a selective medium containing 400 g/ml G418 (Life Technologies Inc.). After 8 -12 weeks, the colonies were pooled together and used for analysis. Each pool of stably transfected cells contained 20 -100 individual colonies.
Ribonuclease Protection Assay (RPA)-RNA from transfected cells was extracted with TRI reagent (MRC, Inc., Cincinnati, OH) and was digested with RNase-free DNase (Ambion), phenol-chloroform-extracted, and precipitated. The marked TH antisense probe was transcribed from the SP73 transcription vector (Promega) containing the HindIII-PstI fragment of the 5Ј-region of marked TH cDNA (see Fig. 1A) using SP6 polymerase. The CAT (see Fig. 3A) and 18 S antisense riboprobes were transcribed from TRI-CAT or T7-18 S templates (Ambion) using T7 polymerase. TH and CAT probes were labeled to the high specific activity using [␣-32 P]UTP (specific activity Ͼ 800 Ci/mmol; NEN Life Science Products) at 42°C for 1 h, heated to 95°C, and digested with RNase-free DNase. A single RNA band of appropriate size was obtained by gel purification. The 18 S probe was labeled to lower specific activity (2.5-3 ϫ 10 5 cpm/g) by including 1 mM unlabeled UTP in the transcription reaction (Maxiscript, Ambion). Ten micrograms of total RNA were hybridized with 0.2 ϫ 10 6 cpm of CAT probe or 0.5 ϫ 10 6 cpm of TH probe and 0.1 ϫ 10 6 cpm of 18 S probe in the hybridization buffer (80% formamide, 40 mM PIPES, pH 6.4, 0.4 M NaCl, 1 mM EDTA) at 42°C in a dry oven overnight. Samples were then digested with RNase A/T1 mixture at 37°C for 30 min, RNases were inactivated with proteinase K, and RNA was phenol-chloroform-extracted and precipitated. Heat-denatured samples were run on a denaturing polyacrylamide gel (8% acrylamide with 7 M urea) and exposed to x-ray film for 18 -48 h. Radioactive bands were identified and quantified using a PhosphorImager system (Molecular Dynamics). Constitutive expression of the exogenous TH mRNA is presented as percent of the endog-enous TH mRNA expression measured in the same experiment. Expression of the endogenous TH and CAT mRNAs was directly normalized to the radioactivity measured in the 18 S ribosomal band. The measurements of mRNA half-life (t1 ⁄2 ) were performed using actinomycin D (5 g/ml) as described previously (5).
Electroelution of Proteins from Ribonucleoprotein Complexes-The binding reactions were performed using 15 g of unlabeled HIPBS oligoribonucleotide and 1.5-2 mg of protein extract from PC12 cells in a 300-l reaction volume. The control reactions were identical except for the RNA. Each 300-l reaction was loaded into a 35-mm 2 -wide well of a preparative nondenaturing gel. One well contained the binding reaction with the labeled RNA for further identification of the RNA⅐protein complex. The gel was exposed to x-ray film, and the fragments of gels containing unlabeled complexes co-migrating at the same level as the labeled complexes were excised, macerated, and loaded into the elution The inside and outside compartments were separated by dialysis membrane with a molecular weight cut-off of 12,000 -14,000 (Spectra/por*2; Spectrum Medical Industries). The eluted proteins were concentrated using Microcon-10 microconcentrators (Amicon), resuspended in the sample buffer, and subjected to SDS-PAGE analysis.
Northwestern and Western Analysis-Proteins were separated on 9% SDS-PAGE and transferred onto nitrocellulose with a semidry blotter (Bio-Rad). The blotted proteins were renatured in buffer containing 10 mM Tris, pH, 7.5, 50 mM NaCl, 1 mM EDTA, 1ϫ Denhardt's solution, 1 mM DTT, and 1 mg/ml heparin for 1 h at room temperature. The blots were then hybridized with 1 ϫ 10 6 cpm of 5Ј-labeled HIPBS oligoribonucleotide in the same binding buffer without DTT or heparin, but with E. coli tRNA (20 g/ml), overnight at room temperature. The blots were then washed briefly in the same buffer and exposed to x-ray film. For Western blot analysis, blots were stripped in PBS with 1% SDS for 1 h at 37°C. The blots were blocked in Tris-buffered saline ϩ Tween 20 with 5% nonfat milk and were probed with the primary antibody in Tris-buffered saline ϩ Tween 20 with 5% dry milk overnight. The signal was visualized by exposing the blots to chemiluminescence reagents (Amersham). Chicken anti-PCBP antibody that recognizes PCBP 1 and PCBP 2 isoforms was a gift from S. A. Liebhaber (11). A specific rabbit anti-PCBP 1 antibody against the peptide sequence of PCBP 1 (amino acids 201-220, GenBank TM accession number U24223 (9)) was raised and affinity-purified by Alpha Diagnostic. A specific anti-PCBP 2 anti-body (GenBank TM accession number X78136) was a gift from A. V. Gamarnik and R. Andino (12).
Statistical Analysis of the Data-The mRNA half-lives were calculated for each experiment from linear regression lines fitted to all time points for each individual experiment. The average t1 ⁄2 is given as mean Ϯ S.D. (n Յ 3) or S.E. (n Ͼ 3) for all experiments. Unpaired t tests or analysis of variance were used to calculate statistical significance between the groups. In a graphical representation, the logarithmic values of the RNA concentration for each time point are shown, and a straight line is fitted for all the points in the experimental group by linear regression using Origin 3.5 linear regression algorithms.

Stability of TH mRNA Is Decreased by Mutation within the
Pyrimidine-rich Site-To determine the role of HIPBS in the regulation of TH mRNA stability, we stably expressed, in PC12 cells, the full-length exogenous TH mRNA with either wildtype or mutated HIPBS. The exogenous mRNAs were marked by a short sequence located in the 5Ј-end of the message (Hin-dIII-BamHI, Fig. 1A) to differentiate exogenous from endogenous TH mRNAs and to measure both in the same sample using RPA (Fig. 1B).
First we analyzed the constitutive expression of exogenous as compared with endogenous TH mRNA. The steady-state level of exogenous wild-type TH mRNA was at 64.8 Ϯ 6.9% (n ϭ 5) of the endogenous TH mRNA (Fig. 1, B, lanes 4 and 5, and C). In contrast, the steady-state level of the mutated TH mRNA was significantly decreased and was only 37.3 Ϯ 1% (n ϭ 4, p Ͻ 0.01) of the endogenous TH mRNA (Fig. 1, B, lanes 6 and 7, and C).
Next we analyzed degradation rates of the wild-type and mutated exogenous TH mRNAs after inhibition of transcription with actinomycin D (Fig. 2). During normoxia the t1 ⁄2 of the wild-type TH mRNA was 9.2 Ϯ 0.4 h (n ϭ 3; Fig. 2A, solid squares), and it increased almost 2-fold to 17.2 Ϯ 0.5 h (n ϭ 2, p Ͻ 0.01; Fig. 2A, open squares) during hypoxia. This is similar to the half-life measured for endogenous TH mRNA (7.5 Ϯ 0.4 h (n ϭ 5) in normoxia and 15 Ϯ 0.7 h (n ϭ 5) in hypoxia; p Ͻ 0.01). The half-life of the mutated TH mRNA was significantly lower at 4 Ϯ 0.8 h (n ϭ 3, p Ͻ 0.01; Fig. 2B, solid squares), and it not only failed to increase during hypoxia but showed a small decrease to 2.9 Ϯ 0.3 h (n ϭ 5, p ϭ 0.035, Fig. 2B). These data indicate that HIPBS is an mRNA stabilizing element.
HIPBS Is an mRNA Stabilizing Element in the Context of a Heterologous mRNA-To further characterize the role of HIPBS in regulation of mRNA stability, we stably expressed in PC12 cells chimeric mRNA that have fragment of TH 3Ј-UTR containing HIPBS cloned downstream from the CAT reporter gene (Fig. 3A). The chimeric mRNA had either a wild-type (CAT-wtTH mRNA) or mutated, inactive (CAT-mutTH mRNA) HIPBS site.
Steady-state levels of the CAT-wtTH mRNA were substantially increased compared with CAT mRNA lacking the TH insert (Fig. 3B, lanes 4 -6 compared with lane 3). In contrast, steady-state levels of the CAT-mutTH mRNA (lane 7) were significantly lower than either CAT-wtTH mRNA (lanes 4 -6) or CAT mRNA alone (lane 3). Importantly, the change in the constitutive levels of chimeric mRNAs was also reflected in the comparable differences in the levels of the derived protein measured using CAT assays (not shown).
Measurements of the degradation rates of CAT, CAT-wtTH, and CAT-mutTH mRNAs after transcriptional inhibition with actinomycin D (Fig. 4) show that insertion of the wtTH sequence into the 3Ј-UTR of the CAT gene increased the half-life of the chimeric mRNA from 5 Ϯ 0.2 (n ϭ 3) to 11.6 Ϯ 1 h (n ϭ 5, p Ͻ 0.01). On the other hand, insertion of the same sequence with a mutation that abolished protein binding (7) resulted in a substantial destabilization of the chimeric mRNA and a decrease in the mRNA t1 ⁄2 to 2.3 Ϯ 0.05 h (n ϭ 2). This is significantly lower (p Ͻ 0.01) than the half-life of the CAT-wtTH mRNA.
To determine whether expression of chimeric CAT-wtTH mRNA is regulated during hypoxia, the mRNA levels were measured during normoxia and hypoxia in the absence or presence of actinomycin D. Unexpectedly, hypoxia failed to regulate expression of the CAT-wtTH mRNA in PC12 cells (data not shown). Further, even insertion of the full-length 3Ј-UTR and 5Ј-UTR of TH mRNA downstream and upstream, respectively, from the CAT gene did not confer hypoxic regulation to this chimeric mRNA (data not shown). Thus, the HIPBS is an mRNA stabilizing element that is necessary but not sufficient for the O 2 -dependent regulation of TH mRNA turnover. Moreover, additional regulatory elements involved in the hypoxic regulation are most likely located within the coding region of TH mRNA.
Characterization of the RNA⅐Protein Complex-The TH mRNA⅐protein complexes were analyzed by UV light crosslinking of TH mRNA⅐protein-binding reactions and separation by SDS-PAGE under reducing conditions (100 mM DTT). Two large complexes were identified at approximately 80 and 50 kDa (Fig. 5A, lane 1). Addition of poly(U) competitor blocked, to a large extent, the 80-kDa complex, revealing a major complex at 50 -55 kDa (Fig. 5A, lane 2), but it did not affect formation of the complex identified in the RNA gel shift assays as shown previously (6). Formation of the 50 -55-kDa complex in the presence of poly(U) was abolished by addition of the unlabeled 162-base-long TH transcript, HIPBS oligoribonucleotide, or poly(C) RNA as competitors (Fig. 5A, lanes 3-5). A 50 -55-kDa complex was formed in the absence of poly(U) when labeled HIPBS oligoribonucleotide was used as a probe (Fig. 5A, lane  6). The complex was not formed when the TH mRNA mutated within the protein-binding site was used in the binding reaction (Fig. 5B). This mutant did not form a complex detectable in the gel retardation assays (7). The 50 -55-kDa complex was induced by hypoxia in PC12 cells 2.6 Ϯ 0.3-fold (n ϭ 3, p Ͻ 0.02) (Fig. 5C). Thus, the 50 -55-kDa ribonucleoprotein complex is the main complex associated with HIPBS and is regulated by hypoxia.
Identification of the Protein Factor in the Complex with the Pyrimidine-rich Sequence in TH mRNA-Because poly(C) RNA is such an effective competitor of the complex formation (6, 7), poly(C)-agarose affinity was used to purify the binding factor from the S-100 protein extracts from PC12 cells (Fig. 6A). Proteins bound to poly(C)-agarose were fractionated by elution with increasing concentrations of KCl (0.5-2 M). The presence of the TH mRNA-binding activity was determined by Northwestern assay with 32 P-labeled HIPBS (Fig. 6A, left). The 40-kDa HIPBS-binding protein was identified in the eluates (Fig.  6A, lanes 2-5). This protein was further identified as the 40-kDa PCBP (also known as ␣CP or hnRNPE (10 -12)) with a specific polyclonal antibody (Fig. 6A, lanes 7-10).
Protein fractions eluted with the salt gradient were pooled and analyzed for their ability to form complexes with TH mRNA. The combined eluted fractions formed a 50 -55-kDa complex with TH transcript that migrated with the same mobility (Fig. 6B, lane 4) as the complex formed by the total S-100 fraction (Fig. 6B, lane 2) analyzed by UV cross-linking and SDS-PAGE. Extracts treated with poly(C)-agarose were depleted of the TH mRNA-binding activity (Fig. 6B, lane 3).
To confirm that the purified poly(C)-binding protein is present in the HIPBS-associated protein complex, proteins were electroeluted from the unlabeled complexes identified by the gel shift assay, separated on SDS-PAGE, and screened by Northwestern analysis with HIPBS probe, followed by Western analysis with a specific antibody against the two isoforms of PCBP (Fig. 7A). As controls, the proteins were eluted from corresponding gel slices in which the binding reaction did not include RNA. HIPBS bound to a protein factor electroeluted from the complexes formed in the presence of RNA (Fig. 7A, Only the 50 -55-kDa complex was formed when HIPBS oligoribonucleotide was used as a probe (lane 6). B, the 50 -55-kDa complex (arrow) is formed by wild-type (WT) TH transcript (lane 1) but not by TH mRNA mutated in the protein-binding region (MUT, lane 2). C, the formation of the 50 -55-kDa complex is induced in cells exposed to 5% O 2 (hypoxia, lane 2) as compared with cells exposed to 21% O 2 (normoxia, lane 1). lane 2) but not to the proteins electroeluted from the gel slices in the absence of RNA (Fig. 7A, lane 1). Subsequent reprobing of the same blot with an antibody specific for the isoforms PCBP 1 and PCBP 2 revealed that both isoforms are actually present in the complex (Fig. 7A, lanes 4 and 6, respectively). PCBP 2 migrates slightly higher than PCBP 1 detected by Western blot, which may represent some post-translational modifications and a somewhat higher molecular weight of PCBP 2 (a nine-amino acid difference).
To determine whether hypoxia induces expression of the PCBP isoforms in PC12 cells, cytoplasmic protein extracts were analyzed by Western blot using antibodies specific for each isoform. A 2.4 Ϯ 0.3-fold induction in the expression of PCBP 1 was measured in four independent sets of PC12 protein extracts (Fig. 7B). In contrast, no induction (or inhibition) of PCBP 2 was measured (data not shown). To confirm the increased amount of PCBP 1 protein in the complex, proteins were electroeluted from the HIPBS-associated complexes formed by normoxic or hypoxic extracts (Fig. 7C). Indeed, proteins elec-troeluted from the complexes formed with the hypoxic extracts showed enrichment for the PCBP 1 (Fig. 7C, lane 2). These data indicate that the PCBP 1 isoform is critical for hypoxic induction of the 50 -55-kDa complex associated with TH mRNA. DISCUSSION The major finding from this study is that HIPBS is an mRNA stabilizing element required for constitutive and hypoxia-regulated control of TH mRNA. A four-point mutation that abolished the protein-binding site within the full-length TH mRNA resulted in a 2-fold destabilization of the mutated mRNA and a corresponding 2-fold decrease in mRNA steady-state levels. In addition, mutation of HIPBS abolished the O 2 -dependent regulation of TH mRNA stability. We further demonstrated that a short fragment of the TH 3Ј-UTR containing the HIPBS was sufficient to confer augmented mRNA stability on the heterologous mRNA, which in turn resulted in augmented steadystate levels of the chimeric mRNA and derived protein. Importantly, the mutated HIPBS conferred destabilization to chimeric CAT-TH mRNA. The decrease in the half-life because of the mutation in the protein-binding site was very similar for both the chimeric CAT-TH mRNA or full-length exogenous TH mRNA.
Interestingly, whereas HIPBS is a necessary element for the O 2 -dependent stabilization of TH mRNA in the context of the full-length TH mRNA and whereas it binds a protein factor in a hypoxia-inducible manner, HIPBS alone is not sufficient to confer hypoxic regulation to the CAT mRNA. Neither are the full-length 3Ј-and 5Ј-UTRs of TH mRNA. This result is similar to previously published data regarding hypoxic regulation of VEGF mRNA stability (13). In that study, the full-length 3Ј-UTR of VEGF mRNA did not confer hypoxic regulation on the heterologous mRNA (13). Thus, it is possible that additional necessary regulatory elements are located within the coding region of hypoxia-regulated mRNAs. Work is in progress to identify these additional elements in TH mRNA.
Sequences similar to that in TH mRNA were reported to regulate the stability of other mRNAs. First, the stability of the erythroid-specific ␣ 2 -globin mRNA is regulated by three short cytidine-rich repeats similar to the TH protein-binding motif (14). Second, a pyrimidine/cytidine-rich sequence within the collagen ␣1(I) mRNA is involved in the regulation of the stability of this mRNA and forms a similar ribonucleoprotein complex with protein factors (15). Mutations in these elements destabilize globin and collagen ␣1(I) mRNAs. In our study, however, we show not only that the mutations destabilize TH mRNA but also that the short fragment of mRNA containing the protein-binding site is capable of conferring increased mRNA stability to a heterologous mRNA.
The increased expression of chimeric CAT-TH mRNA was accompanied by a corresponding increase in expression of functional protein (data not shown), an indication that translation of the chimeric mRNA was most likely not affected. This observation is important because pyrimidine-rich sequences binding similar protein factors were reported to regulate translation (12,16,17).
Analysis of the RNA⅐protein complexes after UV light crosslinking and SDS-PAGE analysis revealed the presence of one major broad 50 -55-kDa complex. This finding extends and complements the results of our previous study (6), in which we detected two TH mRNA⅐protein complexes: the major 74-and a minor 53-kDa complex. We currently view the 50 -55-kDa complex as the major one. The importance of this complex became clear only after competition with poly(U), which blocked protein binding at 80 kDa, and after analysis of complexes formed specifically with HIPBS oligoribonucleotide in comparison with the complexes formed with the longer TH transcript. Thus, the  2, 4, and 6) or a control gel slice that contains only proteins but no RNA (lanes 1, 3, and 5) by Northwestern (NW) assay using labeled HIPBS RNA (lanes 1 and 2) and by Western (W) using specific anti-PCBP 1 (lanes 3 and 4) or anti-PCBP 2 (lanes 5 and 6) antibodies, respectively. B, Western blot analysis of two independent sets of normoxic and hypoxic cytoplasmic extracts demonstrating the strongest (3-fold, lanes 1 and 2) and the weakest (1.8-fold, lanes 3 and 4) induction of the PCBP 1 by hypoxia. Lane 5, assessment of the antibody specificity using antigenic peptide in the Western reaction. C, Western blot analysis of protein factors electroeluted from the HIPBS-associated ribonucleoprotein complex formed with proteins from hypoxic (5% O 2 , lane 2) as compared with normoxic (21% O 2 , lane 1) PC12 cells.
80-kDa complex most likely represents nonspecific binding of protein factors to sequences other than HIPBS fragments within the 162-base-long TH transcript. The 50 -55-kDa complex should correspond, after subtracting the molecular mass corresponding to the 28 bases of RNA (8 kDa), to the protein factor with an approximate molecular mass of 42-47 kDa.
In view of the strong affinity of the binding proteins for poly(C) RNA, poly(C) RNA-agarose was used to purify the protein. A similar procedure was used previously to purify the protein factor binding to the cytidine-rich elements in ␣ 2 -globin mRNA (9,10). As expected, we purified the 40-kDa poly(C)binding protein that corresponds to PCBP (12), also known as ␣CP (9, 10) or hnRNPE (17,18), as identified by the specific antibodies. The eluate from the poly(C)-agarose restored fully formation of the complex with HIPBS RNA or the TH transcript in the binding reactions. This finding differs from previously reported data showing that the poly(C)-eluted proteins did not restore formation of the complex associated with ␣ 2globin stability element and was thus considered necessary but not sufficient for the complex formation (10). Further, formation of the globin-associated complex seems to require additional protein factors such as AUF1 (hnRNPD) protein (19). So far we did not find any evidence for the presence of AUF protein in the HIPBS-associated complex. 2 The protein-binding motifs within the TH and ␣ 2 -globin mRNAs show some potentially important differences. Whereas the essential PCBP-binding motifs are very similar in the two mRNAs (11), in the case of TH mRNA the binding motif is located in the middle of a long pyrimidine-rich sequence, and in the case of ␣ 2 -globin mRNA the short binding motifs are separated by purine-rich sequences. This difference may affect the constitution of the ribonucleoprotein complex associated with the protein-binding motifs and result in RNA-specific regulation of mRNA stability.
Interestingly, we measured an approximately 2-fold induction in expression of PCBP 1 but not PCBP 2 in cytoplasmic extracts obtained from hypoxic cells. That corresponds to the similar enrichment of PCBP 1 in the electroeluates from the ribonucleoprotein complexes formed with the hypoxic protein extracts and to the quantitative increase in the complex formation. This difference results from an absolute enrichment for PCBP 1 , as similar increases were measured using total cellular lysates, and immunocytochemistry did not show any intracellular translocation in the PCBP 1 content. 2 This difference in the hypoxic inducibility of the two isoforms supports the observation that although PCBP 1 and PCBP 2 have high degrees of sequence homology, they show some specific functional differences, such as efficiency of interacting with AUF1 protein (19) or efficiency to regulate poliovirus translation (12).
The molecular mechanism by which binding of PCBP to the HIPBS element regulates mRNA stability, especially in the context of increased TH mRNA stability during hypoxia, is presently unknown. A possible mechanism would be protection of a nuclease cleavage site within the region associated with HIPBS in the TH mRNA. Protein binding to the wild-type sequence may protect this site from nuclease activity. When protein binding is prevented by mutation, the site becomes accessible to cleavage, resulting in a rapid degradation of mRNA. The fact that a fragment of TH mRNA containing mutated HIPBS confers decreased mRNA stability to the chimeric mRNA may favor such a possibility. On the other hand, formation of the complex associated with the globin mRNA stability element indicates its potential role in stabilization via the poly(A) tails (20,21). To evaluate this possibility, a determination of the length of poly(A) tails of TH mRNA during normoxia and hypoxia is under way.