The Major Astrocytic Phosphoprotein PEA-15 Is Encoded by Two mRNAs Conserved on Their Full Length in Mouse and Human*

Specific phosphoproteins are targets of numerous ex- tracellular signals received by astrocytes. One such tar-get, which we previously described, is PEA-15, a protein kinase C substrate associated with microtubules. Two cDNAs differing in the length of their 3 (cid:42) -untranslated region (3 (cid:42) UTR) were cloned from a mouse astrocytic library. Accordingly, Northern blots revealed two tran- scripts (1.7 and 2.5 kilobase pairs) abundant brain regions but also found in peripheral tissues. PEA-15-de- duced protein sequence (130 amino acids) shared no similarity with known proteins but is 96% identical to its human counterpart. In addition, several regions of the 3 (cid:42) UTR share more than 90% identity between mouse and human. Different potential regulatory sequences are found in the 3 (cid:42) UTR, which also completely includes the proto-oncogene MAT1. The high level of conservation of both the coding and the untranslated regions and the differential tissular distribution of the two transcripts of this major brain phosphoprotein suggest that not only the protein but also the 3 (cid:42) UTR of PEA-15 mRNA play a role in astrocytic functions. an Astrocytic cDNA Striatal from 16-day-old confluent monolayer devoid of contaminant cells Following total (10), and the primed an oligo(dT) 18 an Xho I site at its 5 (cid:57) end and synthesized using avian myeloblastosis virus reverse transcriptase (Boehringer Mannheim). Second strand synthesis was performed according to Gu¨bler and Hoffman’s protocol (11, The and quality of the two synthesis steps were by radioac- tivity incorporation, S 1 protection assay, and alkaline gel electrophoresis. Ligation of an Eco RI adaptor at the 5 (cid:57) end of the size-selected double strand cDNA and digestion of the Xho I site at the 3 (cid:57) end directional insertion at the corresponding sites of (cid:108) -ZAPII bacteri- ophage vector (Stratagene). The 1.5 kb, 3 respectively) were into Gigapack II (Stratagene) and grown by infection of the XL1-Blue (cid:57) (Stratagene).

Specific phosphoproteins are targets of numerous extracellular signals received by astrocytes. One such target, which we previously described, is PEA-15, a protein kinase C substrate associated with microtubules. Two cDNAs differing in the length of their 3-untranslated region (3UTR) were cloned from a mouse astrocytic library. Accordingly, Northern blots revealed two transcripts (1.7 and 2.5 kilobase pairs) abundant brain regions but also found in peripheral tissues. PEA-15-deduced protein sequence (130 amino acids) shared no similarity with known proteins but is 96% identical to its human counterpart. In addition, several regions of the 3UTR share more than 90% identity between mouse and human. Different potential regulatory sequences are found in the 3UTR, which also completely includes the proto-oncogene MAT1. The high level of conservation of both the coding and the untranslated regions and the differential tissular distribution of the two transcripts of this major brain phosphoprotein suggest that not only the protein but also the 3UTR of PEA-15 mRNA play a role in astrocytic functions.
Astrocytes have to integrate many different extracellular signals with regard to their numerous functions (1,2). For example, in the developing embryo, radial glia direct migrating neurons to their appropriate location and participate in determining their phenotype. Astrocytes also appear to be involved in the survival of mature neurons by releasing neurotrophic factors (3) and by providing neurons with nutrients (4). They also modulate neuronal transmission by removing ions and inactivating neurotransmitters. The presence on astrocytes of receptors for neurotransmitters, growth factors, hormones, and cytokines (5) allows extracellular signals to be transduced into intracellular cascades, such as the activation of specific protein kinases and phosphatases (6). Specific phosphoproteins are the targets of extracellular signals, and analysis of their function is likely to shed light on key regulatory steps involved in neuron-glial interactions.
Recently, we have described a novel protein kinase C substrate that is highly enriched in astrocytes, PEA-15 (7). This acidic 15-kDa phosphoprotein exists in vivo as three isoforms, namely N, Pa and Pb, which correspond to unphosphorylated, mono, and diphosphate forms, respectively. PEA-15 appears to be a protein required for mature astrocytic functions, because both protein levels and phosphorylation increase during ontogenesis, maximal expression being reached in the adult brain (8). Moreover, PEA-15 phosphorylation is modulated by neurotransmitters or hormones, such as noradrenaline, vasoactive intestinal peptide, and endothelin. In addition to protein kinase C, calcium-calmodulin-dependent protein kinase II also phosphorylates PEA-15 (9). 1 Thus, signals triggering increases in intracellular calcium concentrations result in an increased phosphorylation of PEA-15. The phosphorylation sites for both kinases were identified following protein purification, proteolytic cleavage, and microsequencing, and they were found to be different: LTRIPSAKK for protein kinase C and DIIRQP-SEEEIIK for calcium-calmodulin-dependent protein kinase II. Specific antibodies were raised against the two corresponding peptides that recognized PEA-15 in species ranging from fish to mammals (8), suggesting that an important regulatory feature of the protein is conserved. Finally, immunocytochemistry performed on intact astrocytes revealed that PEA-15 colocalizes with microtubules (8), 2 whereas its level of phosphorylation changes either upon depolymerization or stabilization of tubulins, suggesting that PEA-15 could be involved in the morphological plasticity of astrocytes.
In the present study, we describe the isolation of PEA-15 cDNAs from an astrocytic cDNA library, their characterization, and their tissue distribution. Cloning and comparison with the human PEA-15 cDNAs reveals the conservation of both the coding and the noncoding regions of the transcripts, suggesting the presence of important regulatory sequences that are necessary both for the structure and function of these mRNAs and for their regulation.

PEA-15 Isolation and
Microsequencing-Semipreparative two-dimensional polyacrylamide gels were performed as described previously (7). Protein spots corresponding to PEA-15 were cut out from the gel and subjected to an overnight digestion with Endolysin C. Generated internal peptides were resolved by high pressure liquid chromatography prior to microsequencing. The sequence of four peptides was determined (Jacques D'Alayer, Institut Pasteur, Paris), including the phosphorylation sites for protein kinase C and calcium-calmodulindependent protein kinase II. The longest peptide sequence obtained was SEEITTGSAWFSFLESHNK.
Construction of an Astrocytic cDNA Library-Striatal astrocytes from 16-day-old Swiss mouse embryos were grown for 3 weeks to obtain a confluent monolayer devoid of contaminant cells (8). Following total RNA extraction (10), polyadenylated RNA was purified, and the first strand of DNA was primed with an oligo(dT) 18 containing an XhoI site at its 5Ј end and synthesized using avian myeloblastosis virus reverse transcriptase (Boehringer Mannheim). Second strand synthesis was performed according to Gü bler and Hoffman's protocol (11,12). The yields and quality of the two synthesis steps were checked by radioactivity incorporation, S 1 protection assay, and alkaline gel electrophoresis. Ligation of an EcoRI adaptor at the 5Ј end of the size-selected double strand cDNA and digestion of the XhoI site at the 3Ј end allowed directional insertion at the corresponding sites of the -ZAPII bacteriophage vector (Stratagene). The phage vectors incorporating the three largest fractions (mean size 3, 2, and 1.5 kb, 3 respectively) were packed into Gigapack II Gold particles (Stratagene) and grown by infection of the XL1-Blue MRFЈ strain (Stratagene).
Library Screening with Specific PEA-15 DNA Probe-The 19-amino acid peptide cited above was used to design two degenerated oligonucleotides: AGYGARGARATHAC (amino acids SEEIT) and YTTRTTRT-GRCTYTC (amino acids ESHNK), sense and antisense, respectively. Total RNA isolated from mouse primary cultures of astrocytes was used as template in a reverse transcriptase-polymerase chain reaction (PCR) containing the two degenerated oligonucleotides.
The 57-base pair (bp) PCR product was cloned into the PCRII vector using the TA cloning kit (Invitrogen) and sequenced with Sequenase (Amersham Corp.). Its central region, from nucleotide 16 to nucleotide 42 (ACAGGCAGTGCCTGGTTTAGCTTCCTG), was chemically synthesized (Genset) and end-labeled with [␣-32 P]dCTP (DuPont NEN) in the presence of terminal deoxytransferase (Boehringer Mannheim). This labeled probe was used for the screening at high stringency of the mouse astrocytic cDNA library. After three rounds of purification, the rescue of Bluescript plasmid from -ZAPII bacteriophage vector was carried out with Exassist helper phage and Sol-R bacterial strain according to the supplier's protocol (Stratagene).
In Vitro Translation and Western Blots-Recombinant plasmid containing the 2.4-kb PEA-15 cDNA was linearized with either XhoI or BamHI to allow the synthesis of the corresponding sense or antisense RNA with T3 or T7 RNA polymerase, respectively. One g of digested matrix DNA was added per transcription reaction, directly coupled to the translation step by using the TnT coupled reticulocyte lysate system from Promega. The reaction proceeded for 90 min at 30°C in the optimal conditions given in the supplier's protocol. Radioactive detection of the translation products was achieved by adding 40 Ci of [ 35 S]methionine to the transcription-translation reaction medium and by autoradiography of the dried polyacrylamide electrophoresis gels. In other experiments, the translation products were analyzed by Western blotting. In that case, we used the tRNA nscent system (Promega) according to the supplier's indications. The translation products were electrophoresed in 15% SDS-polyacrylamide gels, transferred onto polyvinylidene difluoride membranes, incubated with the specific PEA-15 antibody, and revealed by a second anti-rabbit antibody coupled to a chemiluminescent reaction with ECL (Amersham Corp.).
Northern Blot Analysis-Total RNA from different tissues and cultured astrocytes (as described in Ref. 8) were extracted. RNA samples were electrophoresed through 1% agarose (Appligene) gels following standard procedures (13) and transferred to Hybond-N membranes (Amersham Corp.). A specific probe for the PEA-15 coding region was synthesized using PCR; the sense primer included the ATG initiation codon, whereas the antisense primer included the TGA termination codon. The expected 390-bp PCR product containing the complete mouse PEA-15 coding region was labeled using the RadPrime DNA labeling kit (Life Technologies, Inc.) and [␣-32 P]dCTP. Two additional probes spanning different regions of the PEA-15 cDNA were obtained using PCR. The first, in the central region contained nucleotides 713-1754; the second, at the 3Ј end included nucleotides 1755-2391. Specific binding was analyzed and quantified in a Packard Instantimager.
In Situ Hybridization-Brains were quickly removed from animals sacrificed by decapitation and were kept frozen until sectioning, and the in situ hybridization was performed as described previously (14). [ 35 S]Thio-UTP-labeled antisense and sense probes were made with T 7 and T 3 RNA polymerase (Stratagene), respectively, from a linearized plasmid containing the 2.4-kb PEA-15 cDNA. Sections were hybridized overnight at 50°C with the antisense and sense probes in a buffer containing 40% formamide. After RNase treatment, the sections were dehydrated, delipidated, and air dried. Standard autoradiography was carried out using NTB-3 emulsion (Kodak). Following development, slides were counterstained with hematoxylin and eosin.

RESULTS
Cloning of PEA-15 cDNAs from a Mouse Astrocytic Library-To increase the probability of cloning a full-length PEA-15 cDNA, a mouse astrocytic cDNA library was constructed (see "Materials and Methods"). A nucleotide probe was synthesized by PCR using degenerated primers whose sequences were deduced from the 19-amino acid peptide obtained after PEA-15 microsequencing. Using astrocytic cDNA as a template, the expected 57-bp PCR product was obtained, cloned, and sequenced, giving a unique central sequence of 27 bp. The 27-bp oligonucleotide containing the full match mouse PEA-15 cDNA sequence was used to screen the astrocytic cDNA library at high stringency (see "Materials and Methods"). 600,000 independent clones were screened, yielding approximately 800 positive clones. 11 out of these 800 were finally purified after three sequential rounds of screening. The insert size, assessed by PCR for all positive clones, was 2.4 kb for nine clones and 1.6 kb for two clones.
The longest open reading frame (ORF) found in the 2.4-kb cDNA is 390 bp long with a 109-bp 5Ј leader sequence and a 1891-bp 3ЈUTR ending just after a polyadenylation consensus sequence AATAAA (Fig. 1). The nine clones were found to be identical, differing only by the length of their 5Ј end. The 1.6-kb cDNA contains the same 390-bp ORF as the 2.4-kb cDNA. Its 3ЈUTR exhibited the same sequence as the 2.4-kb cDNA, except that it was truncated at 1170 bp, and presented an alternative polyadenylation signal ATTAAA, suggesting that it resulted from alternative termination and polyadenylation.
The methionine codon that initiated the 390-bp ORF is located within the nucleotide sequence GGCGTCATGG, which fulfills Kozak's criteria for a eukaryotic initiation codon (15). This probable initiation codon is preceded by an in-frame termination codon, indicating that some regulation of translation could occur at this level (16). In addition, in vitro translation, using rabbit reticulocyte lysates, resulted in the synthesis of a unique 15-kDa protein, recognized by an antibody raised against PEA-15, which is specific (Fig. 2).
The translated protein contains all of the four peptide sequences obtained by PEA-15 microsequencing including the phosphorylation sites for protein kinase C (Ser 104 ) (7) and calcium-calmodulin-dependent protein kinase II (Ser 116 ) (9). Thus, PEA-15 is a 130-amino acid protein with a predicted molecular mass of 15,054 daltons and a calculated isoelectric point of 5.12, in good agreement with previous results obtained from two-dimensional SDS-polyacrylamide gel electrophoresis migration of the endogenous protein (7). Comparison of the deduced PEA-15 sequence with protein and nucleic acid data bases did not reveal any related sequence. Analysis of PEA-15 secondary structure (Hydrophobic Cluster Analysis; Ref. 17) suggests a hydrophobic core characteristic of globular cytosolic proteins that could account for the thermostability of the protein. Two sites of interaction with microtubules can be suggested, between amino acids 98 -107 and amino acids 122-129, because they contain a motif with three lysines and one to three Underlined in the amino acid sequence are the consensus phosphorylation sites for protein kinase C and calcium-calmodulin-dependent protein kinase II, with an asterisk for the two serine phosphorylated by these two kinases. Also underlined are the two other peptides obtained after microsequencing. These sequences data are available from EMBL under accession numbers X86809 (HSPEA15) and X86694 (MMPEA15) for the human and mouse sequences, respectively. B, schematic alignments between mouse and proline described as essential for microtubule-binding of tau and MAP2 (18 -20). In addition, a 20-amino acid-long stretch just upstream of the protein kinase C phosphorylation site is homologous to a conserved nonmotor domain found in microtubule-based molecular motors and dynein-and kinesin-related proteins, previously suggested to mediate protein-protein interactions and the formation of macromolecular complexes (21).
Tissue and Cellular Distribution of the PEA-15 Gene Expression-As previously indicated, cloning and sequencing of PEA-15 cDNA demonstrated the presence of two transcripts differing in the length of their 3ЈUTR. Accordingly, Northern blot analysis of total RNA from different tissues, using a labeled probe corresponding to the specific PEA-15 coding sequence, showed two transcripts of 2.5 and 1.7 kb (Fig. 3). These two mRNAs are abundant and have a widespread distribution in the central nervous system, particularly in the spinal cord, hypothalamus, and striatum (Fig. 3A). Indeed, the two transcripts are also found more ubiquitously but at lower levels in several peripheral organs (Fig. 3B). The presence of detectable levels of PEA-15 mRNAs in peripheral tissues was surprising because Western blotting previously suggested that PEA-15 expression might be restricted to the central nervous system with the exception of eye and lung (Table I and Ref. 8). Interestingly, the relative amount of the 2.5-over the 1.7-kb transcript was very different according to the tissue or brain region considered, suggesting a regulation of the transcription termination and polyadenylation that could also influence translation efficiency (Table I).
Protein purification and microsequencing of the two phosphorylation sites within PEA-15 allowed the generation of highly specific polyclonal affinity purified antibodies (8). To determine which cells expressed PEA-15 in vivo, we compared PEA-15 immunoreactivity with that of well established specific astrocytic marker glial fibrillary acidic protein. This was performed by double-labeling of coronal sections of the adult rat nervous system. Antibodies against PEA-15 mostly labeled astrocytes in all structures from the olfactory bulb to the spinal cord (Fig. 4, A and B). In addition, some neurons were also found to be PEA-15 positive (Fig. 4, C and D). By using the 2.4-kb cDNA, an RNA probe was synthesized for in situ hybridization on brain slices. A specific labeling was clearly observed, and the distribution of the PEA-15 mRNA corresponded to the protein localization as shown for example in the parietal and pyriform cortices (Fig. 5).
Comparison between Mouse and Human cDNAs-Nucleic acid data base searches found several human partial sequences (ESTs) highly similar to regions of the mouse PEA-15 cDNAs. The genexpress program (Genethon, Evry, France) provided us with the clone c-ozg10, which we entirely sequenced, finding that it is the human counterpart of the mouse 1.6-kb cDNA. Based on c-ozg10 sequence, it was possible to align multiple partially overlapping human ESTs. This allowed us to obtain a 2385-bp sequence that is the full-length human counterpart of the 2.4-kb PEA-15 mouse cDNA and then further confirmed after reverse transcriptase-PCR performed on human postmortem brain tissue and sequencing. Northern blots using human brain extracts confirmed the expression in vivo of these human PEA-15 cDNA sequences. Two cDNAs (2.4 and 1.6 kb) were found to code PEA-15 both in human and mouse. The percentages represent the homology between mouse and human cDNAs both in the coding and the noncoding regions. MAT1 is a reported transforming cDNA from a mouse mammary tumor cDNA library.  3. Tissue distribution of PEA-15 transcripts. Northern blot of total RNA (10 g) from various rat tissues were hybridized with a 32 P-labeled probe corresponding to PEA-15 ORF. Equivalent amounts were loaded as assessed by hybridization with a glyceraldehyde-3phosphate dehydrogenase-labeled probe. Specific binding was analyzed and quantified in a Packard Instantimager (see Table I). two transcripts (data not shown). Interestingly, analysis of the sources of the libraries used to obtain 70 of these ESTs confirms the prominent expression of PEA-15 in the brain (44%), whereas significant expression also exists in many peripheral organs including placenta and liver (7% each), eye, lung, heart, endothelial cells, pancreas, testis and uterus (4% each), adrenal gland, prostate gland, kidney, and spleen (only one EST each).
Comparison of the mouse and human sequences revealed a striking degree of conservation. Like the mouse gene, two alternative polyadenylation signals were found in human sequences at bp 1536 for ATTAAA and bp 2261 for AATAAA, respectively. The PEA-15 coding sequence is 96% identical, coding for a highly conserved protein with 125 identical amino acids and only five conservative changes.
Human as well as mouse PEA-15 cDNAs have a long 3ЈUTR. Comparison of these 3ЈUTRs indicated three regions with identities greater than 90%, often over a stretch of more than 100 bp (e.g., in Fig. 1B, 390 -570, 1410 -1691, and 2180 -2300 nucleotides). Several rare and potentially regulatory motifs are found in both species in these conserved regions. For example eight GGGNGGRR repeats in the 3ЈUTR of PEA-15 mRNA are also found in the glial-specific virus JCV (22).
Further analysis of the 2.4-kb PEA-15 cDNA 3ЈUTR revealed the sequence of the proto-oncogene MAT1. MAT1 is a nucleic acid sequence of 1.7 kb recently isolated from a chemically induced mouse mammary tumor on the basis of its transforming activity in NIH 3T3 cells (23). The entire MAT1 sequence is included within the 3ЈUTR of the 2.4-kb PEA-15 cDNA (nucleotides 713-2391). The occurrence of a chimeric clone is very unlikely because no restriction enzyme sites exist around the start sit of this sequence. Furthermore, each of the 11 independently isolated mouse PEA-15 cDNA clones described here contained the MAT1 sequence, and more than 10 human ESTs encompass nucleotide 713 of PEA-15 cDNA and have additional upstream sequence; finally, we were able to directly  (cg, a), frontal (fr, a  and b) and piriform (pir, c) cortex. Note that the intensity of labeling is differential. d and e, bright field photoautoradiographs at high magnification showing reduced silver grains associated with individual cells (n, cell nucleus stained with hematoxylin). The cells are highly (arrow) or lightly (arrowhead) labeled. The sections were hybridized with antisense (d) and sense (e, control) PEA-15 35 S-UTP-labeled riboprobes. The reduced silver grains in e correspond to background. amplify the 2.4-kb PEA-15 cDNA from astrocytes by reverse transcriptase-PCR (data not shown), demonstrating that this sequence indeed exists in vivo.
To investigate the possible expression of a truncated PEA-15 mRNA that includes only the MAT1 region, additional Northern blot analyses were performed with several probes corresponding to three different regions of the 2.4-kb PEA-15 cDNA (Fig. 6). Probe A, designed as the PEA-15 coding region (bp 0 -390) revealed two messages (1.7 and 2.5 kb), as did probe B, that corresponded to the 3ЈUTR of the 1.6-kb PEA-15 cDNA (bp 613-1670, including the 5Ј end of MAT1) (Fig. 6, A and B). By contrast, probe C, which includes the 3Ј end of the 2.4-kb PEA-15 cDNA from bp 1670 to 2391 (which also includes the 3Ј end of MAT1), only hybridized with the 2.5-kb mRNA (Fig. 6C). The same results were obtained with different normal peripheral tissues, including mammary glands and several tumors (data not shown). Furthermore, as shown on Fig. 2, it was never possible to observe a 6-kDa protein translated from the PEA-15 RNA, suggesting it is only able to express PEA-15 protein. Taken altogether, these findings demonstrate that the MAT1 proto-oncogene cDNA is a partial sequence of 1678 nucleotides of the 3ЈUTR of PEA-15 not coding for another protein.

DISCUSSION
During the last two decades, numerous functions have been assigned to astrocytes, among which is their role as communicating cells (2). Intracellular phosphoproteins can be considered as targets for extracellular signals received by the cell (6). Their phosphorylation modifies their function and consequently some of the cell properties. However, very few specific astrocyte-enriched phosphoproteins are known, essentially the two main components of intermediate filaments: vimentin in immature cells and glial fibrillary acidic protein in mature astrocytes (24,25). We have previously characterized PEA-15 as one of the major phosphoproteins in cultured astrocytes (7). Taking advantage of this enrichment, an astrocytic cDNA library was constructed to increase the relative abundance of PEA-15 transcripts. The amount of PEA-15 positive clones in the library, near 1:1000, is in agreement with the high enrichment of this protein in astrocytes. We demonstrate here that indeed PEA-15 is encoded by the longest ORF found in the isolated cDNAs because: (i) it contains all of the four peptide sequences previously established from protein microsequencing, (ii) it is translated in vitro as a 15-kDa protein, and (iii) it is expressed in vivo as demonstrated by Northern blotting and in situ hybridization.
The high expression of PEA-15 in astrocytes led us to focus on the function the protein could play in these cells in particular. However, a low but significant expression of the protein was also observed in neurons and oligodendrocytes grown in primary cultures (8). Data reported now support a more ubiquitous expression of the protein in vivo because immunohistochemistry performed on brain sections clearly revealed, beside astrocytes, subpopulations of PEA-15 positive neurons, and in situ hybridization confirmed these results at the mRNA level ( Fig. 4 and 5). In addition, Northern blots indicate that expression of PEA-15 is predominant in the central nervous system; however, a significant but low level of PEA-15 transcripts is detected in several peripheral organs. This suggests additional functions for PEA-15 transcripts required in multiple cells and tissues, in addition to their specific role in astrocytes.
Phylogenetic conservation of the epitopes containing the two phosphorylation sites of PEA-15 was already suggested by Western blotting because specific antibodies allowed the detection of the protein in the brains of mammals, birds, and fish (8). Accordingly, the high homology (96%) between the human and mouse protein sequences established in the present study confirms this striking conservation and suggests a strong structural requirement for the function of the protein.
In addition to the remarkable conservation of the PEA-15 protein sequence, three highly conserved regions are found within the 3ЈUTR of its cDNAs, each greater than 100 nucleotides in length. In mouse as well as in human, the two transcripts are presumably generated by the alternative use of the polyadenylation signals, which are found a dozen nucleotides upstream of the poly(A) tail of each cloned transcript. The 2.5-kb mRNA being always the most abundant form found in the central nervous system, diversity in the 3ЈUTR of PEA-15 mRNA may result in differential stability or translation efficiency, as proposed for other eukaryotic mRNAs (for a review see Ref. 26).
3ЈUTRs are also known to contain regulatory sequences that signal mRNA localization, translational regulation, and direct degradation (27)(28)(29). Conserved sequences found in mouse and human PEA-15 cDNA 3ЈUTR are good candidates for such roles. Indeed, several infrequent regulatory motifs were found in these regions, including JCV repeats. The human JC polyomavirus (JCV) is the etiologic agent of the neurodegenerative disease progressive multiple leukoencephalopathy and replicates only in astrocytes. Several studies have established that the restricted host range of JCV to glial cells is determined at the level of viral transcription that is mediated by glial-enriched DNA-binding regulatory proteins (30,31). Two 98-bp enhancer/promoter sequences have been characterized and recently demonstrated to bind two identified proteins: Pura and YB-1 (32). Thus, the PEA-15 gene might be one of the physiological targets of such trans-activators.
In addition, the 3ЈUTR of the 2.4-kb PEA-15 cDNA contains the proto-oncogene MAT1. The MAT1 sequence was isolated from a mouse mammary tumor induced in vitro with N-methyl-N-nitrosourea and lithium and was reported to induce the oncogenic transformation of NIH-3T3 cells (23). A role for 3ЈUTRs in transformation have been previously described. For example, ␣-tropomyosin 3ЈUTR expression suppresses tumorigenicity (33,34), whereas tropomyosin isoforms are frequently missing from spontaneously arising tumors (35), and in vitro transformation with oncogenes or viruses induces suppression of tropomyosin gene expression (36). Finally, expression of a cDNA encoding full-length tropomyosin suppresses transformation (37). PEA-15's co-localization with microtubules and abundance in astrocytes could indicate that this protein plays a role in the regulation of morphological plasticity in astrocytes. Consequently, deregulation of its 3ЈUTR (which contains the proto-oncogene MAT1) could contribute to tumorigenicity. In favor of this hypothesis, the expression of the protein was found FIG. 6. Northern blot analysis of the two PEA-15 mRNAs expressed in normal tissues. Northern blot of total RNA (10 g) from various rat tissues hybridized with 32 P-labeled probe corresponding to PEA-15 ORF (A), to the central region of its cDNAs (bp 713-1754) (B), or to the 3Ј end of the 2.4-kb cDNA (C). Equivalent amounts were loaded as assessed by hybridization with a glyceraldehyde-3-phosphate dehydrogenase-labeled probe.
to be lower in proliferating cells such as C6 glioma cell line (8).
Growing evidence shows that astrocytes are able to initiate dynamic responses when stimulated in vivo by a wide variety of extracellular signals. An important cellular response consists in increases in intracellular calcium that influence many astrocytic functions including cytoskeletal rearrangement and intercellular communication through calcium waves. As a major cytosolic phosphoprotein in astrocytes regulated by multiple calcium-dependent phosphorylation pathways, PEA-15 is ideally positioned to play a major role in signal integration. The present study also suggests a high degree of regulation at the level of translation, localization, and/or transcription of its mRNA and provides new tools to investigate PEA-15's presumed role in and astrocytic growth and differentiation functions both at the cellular and tissue level.