- Email: [email protected]
Characterization of two promoters that regulate alternative transcripts in the microtubule-associated protein (MAP) 1A gene1
Biochimica et Biophysica Acta 1518 (2001) 260^266
www.bba-direct.com
Promoter paper
Characterization of two promoters that regulate alternative transcripts in the microtubule-associated protein (MAP) 1A gene1 Atsuo Nakayama *, Takayuki Odajima, Hideki Murakami, Naoyoshi Mori, Masahide Takahashi Department of Pathology, Nagoya Univerty School of Medicine, 65 Tsurumai-cho, Showa-ku, Nagoya 466-8550, Aichi, Japan Received 4 October 2000; received in revised form 4 January 2001 ; accepted 10 January 2001
Abstract We cloned and characterized the mouse gene for microtubule-associated protein (MAP) 1A, an important protein for neuronal morphology and mitotic spindle formation. We also investigated the 5P untranslated region of the gene to characterize the promoter units. Two alternative transcripts different in the 5P region were identified by 5P RACE. Both transcripts were principally observed in the brain. Genomic cloning revealed that exons 1, 2, and 4 generate the 5P part of a long transcript, whereas exons 3 and 4 generate a short transcript. Putative 5P and intronic promoters flanking exons 1 and 3, respectively, are GC-rich and lack a canonical TATA box. DNase I footprinting from mouse cells revealed that several potential cis-elements were occupied by nuclear proteins. A reporter assay system in conjunction with a number of deletion and mutation constructs was used to test the two putative promoters. Both putative promoters showed transactivity and their function was dependent upon Sp1 sites. In addition, an NF-1 site, an HNF3B site, and an AP-1/ATF site were necessary for basal promoter activity of the intronic promoter. Our data provide insight into the regulatory mechanisms that govern the expression of the MAP1A gene. ß 2001 Elsevier Science B.V. All rights reserved. Keywords : Microtubule-associated protein 1A gene; Promoter; Gene structure ; Alternative transcript
Microtubule-associated proteins (MAPs) are co-puri¢ed with re-polymerized microtubules from brain homogenates [1] and play an important role in determining neuronal morphology and the balance between plasticity and stability of neuronal processes [2,3]. Fibrous MAPs are also involved in the stabilization of the mitotic spindle [4,5]. MAP 1A and 1B are major ¢lamentous high molecular MAPs, which are especially abundant in brain [6^9]. The structures of these two MAPs are closely related. However, MAP 1A and 1B are encoded by distinct genes and map to mouse chromosome 2 (Mtap1) and 13 (Mtap5), respectively [10]. The temporal expression patterns of these two MAPs, which are transcriptionally regulated, are complementary during central nervous system development [10]. Studies on the MAP1B gene and its transcripts have revealed the structure and the transcriptional regulatory units of this gene. In addition to a major transcript driven by two TATA boxes, both of which reside 5P of
* Corresponding author. Fax: +81-52-744-2091; E-mail : [email protected] 1 GenBank accession numbers AF182208^182213.
exon 1 [11], truncated forms of the transcripts, which start from exons 3U and 3A, have been identi¢ed in human, rat, and mouse [12,13]. Thus, the MAP1B gene may have multiple promoter units regulating the unique expression pattern of the gene. In contrast, the structure of the MAP1A gene has not been well characterized and its functional promoter has not been determined [14]. Here, we report the isolation of the mouse MAP1A gene and identi¢cation of two promoters that regulate the expression of the alternative transcripts. We obtained a 5P coding fragment of mouse MAP1A cDNA by RT-PCR from mouse brain total RNA with primers whose sequences are conserved between rat and human. The ampli¢ed 664 bp fragment was 95.2% identical with the corresponding rat cDNA (GenBank accession number M38196) and 89.5% with the human cDNA (GenBank accession number U38291). The amino acid sequence was completely identical with the amino-terminal sequence of the rat MAP1A peptide spanning amino acids 4^224 [8]. Based on the sequence information of this fragment, we performed 5P RACE using the 5P RACE system version 2 (Gibco-BRL, Rockville, MD, USA). Reverse transcribed product prepared from mouse brain RNA,
0167-4781 / 01 / $ ^ see front matter ß 2001 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 4 7 8 1 ( 0 1 ) 0 0 1 7 3 - 7
BBAEXP 91521 10-4-01
A. Nakayama et al. / Biochimica et Biophysica Acta 1518 (2001) 260^266
by priming with primer R1384 (5P-GCC CGA TGT TTG ATT TTG CTG-3P), was subjected to oligo-dCTP tailing and the following PCR ampli¢cation. The primers used in the PCR were AUAP supplied by the manufacturer and L1261 (5P-ATC TTG TTC TGG GGA GCA TTT-3P) in the ¢rst round. UAP supplied by the manufacturer and L1106 (5P-CTT TAC TAT TGC CTG CCC ACT T-3P) were used in the second round. After two rounds of ampli¢cation, two major RACE products, 1.1 kb and 1.4 kb in length, were obtained. Sequencing revealed that the 5P 80 nucleotides of the 1.1 kb product and 5P 373 nucleotides of the 1.4 kb product were di¡erent (Fig. 1A, upper panel), while about 1 kb of the sequences were identical (GenBank accession numbers AF182212 and AF182213). To con¢rm that the two transcripts, a short transcript corresponding to the 1.1 kb RACE product and a long transcript corresponding to the 1.4 kb product, are ex-
261
pressed in mouse tissues, we performed RT-PCR analysis. Primer pairs SU2 (5P-GAC TCT TCC CTG AAG CTG CC-3P) and CL273/566 (5P-AAG GAC CTG GGC GAA GTT TT-3P) were designed to amplify the short transcript, and LU179 (5P-GCT GGG ACC TGC GGA AAT AC-3P) and CL273/566 were used in ampli¢cation of the long transcript. Both transcripts were highly expressed in brain (Fig. 1A), and they were also expressed in mouse embryos (12.5 days post conception). The short transcript was also expressed at low levels in heart, muscle, and several other organs. However, the long transcript was not detectable in these organs. Although both transcripts were principally expressed in brain, the short transcript had a broader tissue-speci¢c expression pattern. We performed primer extension to obtain additional information on the 5P end nucleotide(s) of the transcripts using the Primer Extension System (Promega, Madison,
6
Fig. 1. Distribution of the 5P alternative transcripts and determination of their 5P ends. (A) A schematic representation of the two mouse MAP1A transcripts is shown. The common regions of the two transcripts are represented as open bars. Regions speci¢c for the short transcript are shaded and regions speci¢c for the long transcript are hatched. In-frame ATGs are indicated as open triangles (P), and inframe stop codons are indicated as solid triangles (S). Primers used for RT-PCR are represented as arrows below. 10 Wl of the PCR products from mouse brain (B), heart (H), lung (Lu), liver (Li), spleen (S), kidney (K), intestine (I), uterus (U), ovary (O), muscle (M), 12.5dpc embryo (E), Neuro2A (N2), and NIH3T3 (NIH) were analyzed on a 1.5% agarose gel. The upper panel shows products from the long transcript (L), and the middle panel shows products from the short transcript (S). At the bottom, actin bands are shown as a control for equal loading (A). (B) Identi¢cation of the 5P end of the short transcript. Primers PE1 (5PCGC AAA GAG GCT CCG CCG GGT TCT GA-3P) and PE2 (5PCCG GAG TTG TCT CCA TGG CAA CGC TG-3P) were designed to anneal with nucleotides 126^151 and 66^91 of the long 5P RACE product, respectively. Primer PE3 (5P-TGG TCT CGA GGG TCT CAC TCT GGT GG-3P) corresponds to nucleotides 97^122 of the short RACE product. A schematic representation of the primers along with the two transcripts is illustrated at the top. A single 122 nucleotide band of extended product was identi¢ed on the sequencing gel, when primer PE3 was annealed with 10 Wg of mouse brain poly(A) RNA (lane 3). No extended products could be seen with primers PE1 and PE2 (lanes 1 and 2). Sequencing ladders of the corresponding genomic regions primed with PE1 and PE2 were run on both sides as molecular size markers. (C) RNase protection assay determining the 5P end of the long transcript. A genomic fragment from 3465 to +82 relative to the putative transcription initiation nucleotide of exon 1, determined by 5P RACE, was subcloned into pGEM-7Zf vector (Promega). This subclone was used as a template to make a 32 P-labeled antisense RNA probe. Schematic representations of the antisense RNA probe, the in vitro transcribed RNA from the long RACE product (ivt LT), and its truncated form (ivt TLT) are shown. RNA fragments protected by 10 Wg (lane 5) and 50 Wg (lanes 6 and 7) of mouse brain total RNA exhibited the same banding pattern as RNA fragments from in vitro transcribed RNA from the long RACE product (lane 4). For comparison, in vitro transcribed RNA from the 5P truncated long RACE product was also assayed (lane 3). Discrete protected bands could not be identi¢ed with 50 Wg [8] of yeast tRNA (lane 1), the in vitro transcribed short transcript (lane 2), 10 Wg of mouse liver total RNA (lane 8), or 50 Wg of mouse liver total RNA (lane 9).
BBAEXP 91521 10-4-01
262
A. Nakayama et al. / Biochimica et Biophysica Acta 1518 (2001) 260^266
Fig. 2. Structure of the 5P portion of the MAP1A gene. (A) Exons 1^6 (open bars) and BamHI (B), EcoRI (E), and XhoI (X) sites are illustrated. The 3P end of exon 6 has not been determined. Inserts of isolated phage clones are represented as lines at the top. Splicing patterns resulting in the 5P alternative transcripts are shown below. Potential coding regions are shaded. (B) Exons 1^6 of the MAP1A gene are aligned with exons 1^5 of the mouse MAP1B gene [13]. Each exon is represented as a box. The numbers inside the boxes indicate the exon sizes in nucleotides. The splicing patterns resulting in the short transcript for MAP1A and a truncated form for MAP1B are indicated by solid lines connecting the exons. The splicing patterns resulting in the long transcript for MAP1A and MAP1B are indicated by broken lines.
WI, USA). We identi¢ed a single 122 nucleotide band of extended product for the short transcript (Fig. 1B, lane 3). Although both primer extension and 5P RACE depend upon reverse transcription without premature cessation, our results suggested that the 5P end nucleotide identi¢ed by 5P RACE was the major transcription initiation nucleotide of the short transcript. We could not detect extension products for the long transcript despite using two di¡erent primers (Fig. 1B, lanes 1 and 2). Our inability to detect extension products may be attributed to a high stacking 5P sequence in the long transcript. We performed RNase protection assays using the RPA II (Ambion, Austin, TX, USA) to determine the transcription initiation nucleotide(s) of the long transcript. A genomic fragment covering the 5P part of exon 1 and the 5P £anking region was subcloned and used as a template to make an antisense RNA probe (Fig. 1C, upper panel; see Figs. 2 and 3 for the genomic structure and sequence). Mouse brain RNA protected the antisense RNA probe while mouse liver RNA did not (Fig. 1C, lanes 5^9). In vitro transcribed RNA corresponding to the long RACE product resulted in a single protected band (Fig. 1C, lane 4), but in vitro transcribed RNA from the short RACE product and yeast tRNA did not protect the antisense probe (Fig. 1C, lanes 1 and 2). The fragments protected by brain RNA and in vitro transcribed RNA from the long RACE product were the same size. For comparison, in vitro transcribed RNA from the 5P truncated long RACE product was also examined, and it protected a shorter fragment in
the expected size (Fig. 1C, lane 3). This result con¢rmed that the 5P end of the long RACE product was the major transcription initiation site of the long transcript. To isolate MAP1A genomic clones, we screened 2U106 clones of a mouse genomic lambda phage library [15] with the 664 bp PCR fragment described above as a probe. Two overlapping clones, N4 and M1, out of 14 positive clones isolated were analyzed. Mapping of restriction enzyme sites, Southern blotting, and sequencing yielded the 5P genomic structure of the gene (Fig. 2A, GenBank accession numbers AF182208^182211). The two transcripts di¡erent in the 5P region appear to be splicing variants. Exons 1, 2, and 4 correspond to the 5P region of the long transcript, while exon 3 spliced with exon 4 correspond to the 5P region of the short transcript. The intron^exon boundaries (Fig. 2A) satis¢ed the AG-GT rule with the exception of the 5P boundary of exon 3. Comparison of the 5P organization of the MAP1A and MAP1B genes revealed that their exon^intron structures are quite similar
Fig. 3. Nucleotide sequences of the MAP1A promoter regions and potential cis-acting elements. (A) Nucleotide sequence of exon 1 and £anking regions. The 5P end of exon 1 is indicated by an arrow (s) and numbered as +1. The base sequence of exon 1 is in uppercase letters and the sequence of the 5P upstream region and part of intron 1 are in lowercase letters. Potential Sp1 binding sites (Sp1(A), (B), (C)) and a Y/CCAAT box (Y/CCAAT) are boxed. Open triangles (P) indicate the 5P ends of the deletion constructs used in the luciferase assay. Their names are shown above. (B) Nucleotide sequence of exon 3 and parts of introns 2 and 3. The 5P end of exon 3 is indicated as in (A). The sequence of exon 3 is in uppercase letters and the sequences of introns 2 and 3 are in lowercase letters. Potential cis-elements and 5P ends of the deletion constructs are indicated as in (A).
BBAEXP 91521 10-4-01
A. Nakayama et al. / Biochimica et Biophysica Acta 1518 (2001) 260^266
(Fig. 2B) [13]. The sizes of exons 2, 4, and 5 of MAP1A were identical with those of exons 2, 3, and 4 of MAP1B. The nucleotide sequences were 53.9^59.0% identical when aligned (GenBank accession number X51386). Exons 1 of MAP1A and 1B share 54.6% sequence identity, however, the precise size of MAP1B exon 1 is undetermined. This ¢nding supports the concept that the two genes are derived from a common ancestral gene [8]. Sequencing of the 5P £anking regions of exons 1 and 3 revealed that both of the promoter regions are GC-rich and lack a canonical TATA box. The sequences were further analyzed with MatInspector V2.2 using the TRANSFAC 4.0 database at http://transfac.gbf.de [16] to identify potential cis-elements. Three Sp1 sites, at 3107, 3102, and 380, (Sp1(A), Sp1(B), and Sp1(C)) and a Y/CCAAT box, at 361 relative to the transcription initiation nucleotide of exon 1, were identi¢ed (v0.95 core similarity and v0.90 matrix similarity ; Fig. 3A). Two Sp1 sites at 364 and 353 (Sp1(a) and Sp1(b)), a CCAAT box at 3152, an NF-1 site at 3123, an HNF3B site at 3111, and an AP-1/ ATF site at 389 relative to the putative transcription initiation site of exon 3 were identi¢ed (Fig. 3B). We performed DNase I footprinting to determine which of these potential cis-elements are bound by nuclear proteins. Neuro2A, mouse neuroblastoma cells, and NIH3T3, mouse ¢broblast cells, have been reported to express MAP1A protein [4] and were con¢rmed to express both the 5P alternative transcripts. However, the long transcript was expressed at a very low level in NIH3T3 (Fig. 1A). We prepared nuclear proteins from these two cell lines as previously described [17]. Three Sp1 sites and the Y/CCAAT box in the 5P promoter were protected from DNase I digestion by NIH3T3 nuclear proteins and the Sp1(C) site and the Y/CCAAT box were protected by Neuro2A nuclear proteins (Fig. 4A). The overlapping Sp1(A) and Sp1(B) sites were not protected even with 20 Wg of Neuro2A nuclear extract, when a DNA fragment end-labeled in the coding strand was used as the probe (not shown). However, the Sp1(B) site was protected with 10 Wg of Neuro2A nuclear extract, when a DNA fragment end-labeled in the non-coding strand was employed (Fig. 4A, a bottom picture). These results suggest that the three Sp1 sites and the Y/CCAAT box in the 5P promoter region are functional cis-elements. When the intronic promoter region was analyzed by DNase I footprinting, a long segment including the CCAAT box, the NF-1, HNF3B, and AP1/ATF sites was protected by nuclear proteins from NIH3T3 and Neuro2A (Fig. 4B). The Sp1(b) site was protected by NIH3T3 extracts, whereas DNase I hypersensitive bands were only observed at the Sp1(a) site. Thus, most of the potential elements in the intronic promoter seemed to be involved in assembling nuclear trans-factors. We generated luciferase reporter constructs to examine whether the two promoter regions have functional transactivities, and to determine which potential elements are
263
Fig. 4. Nuclear extracts from NIH3T3 and Neuro2A contain binding proteins to the cis-elements of MAP1 promoters. (A) DNase I footprinting analysis of the 5P promoter region. A DNA fragment spanning from 3141 to +13 relative to the transcription initiation nucleotide of exon 1 was 32 P-labeled at the 3P end of the coding strand and was incubated with either 10 Wg (10) or 3 Wg (3) of nuclear proteins from NIH3T3 or Neuro2A. The incubation was carried out in 100 Wl of binding bu¡er (20 mM Tris, 1 mM EDTA, 10% glycerol, 0.1% NP-40, 2 mM MgCl2 , 1 mM dithiothreitol, 2% polyvinyl alcohol, and 1 Wg of poly(dI-dC)) at room temperature for 20 min. DNase I digestion was performed on ice for 1 min. 10 Wg of BSA (B) was used as a control. The Maxam^Gilbert sequencing G ladder of the probe is used as a molecular size marker (G). The positions of potential cis-elements are indicated on the left (S(A): Sp1(A) site, S(B): Sp1(B) site, S(C): Sp1(C) site, and Y/C: Y/CCAAT box). A limited gel image using the DNA fragment labeled at the 3P end of the non-coding strand is shown at the bottom of the right panel. Regions protected from DNase I digestion and DNase I hypersensitive sites are indicated by hatched bars and open triangles, respectively. (B) DNase I footprinting analysis of the MAP1A intronic promoter region. The results are represented as in (A). The DNA fragment spanning 3158 to +56 relative to the transcription initiation nucleotide of exon 3 was labeled at the 3P end of the non-coding strand. Thus the G ladder represents C positions in the coding strand. Protected regions and positions of potential cis-elements are indicated as in (A) (S(a): Sp1(a) site, S(b): Sp1(b) site, A: AP-1/ATF site, H: HNF3B site, N: NF-1 site, and C: CCAAT box).
critical for the transactivities. The genomic fragments inserted into the pGL3 Basic vector (Promega) are shown in Fig. 5. Constructs MN24, PN1.8, and SpN20 were cloned by utilizing restriction sites of the MAP1A gene indicated in Fig. 5. Constructs p1(3109), p3(3158), and p3(3136) were made by inserting PCR fragments, with appropriate restriction sites engineered at both ends, into the pGL3
BBAEXP 91521 10-4-01
264
A. Nakayama et al. / Biochimica et Biophysica Acta 1518 (2001) 260^266
Fig. 5. Functional analysis of the 5P and intronic promoters in mouse cell lines. (A) The luciferase reporter assay with 5P deletion and mutant constructs suggested that the 5P promoter activity depended upon multiple Sp1 binding sites. The names of the constructs and schematic drawings of genomic fragments inserted in the reporter vector are shown on the left. Exon 1 (open bar), restriction sites (MluI, PstI, and NcoI) employed to generate the constructs, and potential cis-elements are illustrated in a partial MAP1A genomic map at the top. Numbers are relative to the transcription initiation nucleotide of exon 1. Genomic fragments represented by arrows positioned in front of the luciferase reporter gene in the pGL3 Basic vector. Restriction fragments (MN24 and PN1.8) and 5P deletion fragments (p1(3109), p1(393), p1(358), and p1(343)) were tested for their transactivities. Reporter constructs with mutations in the Sp1(A) site (p1(3109)mSp1A), in the Sp1(B) site (p1(3109)mSp1B), in the Sp1(C) site (p1(3109)mSp1C), and in the Y/CCAAT box (p1(3109)mNFY) were also analyzed. Open bars represent the average relative luciferase activities given as activity `X' fold above the promoterless pGL3 Basic vector in NIH3T3 and Neuro2A. *P 6 0.0001. (B) The intronic promoter activity depends on other elements as well as Sp1 binding sites. A partial MAP1A genomic map surrounding exon 3 is shown at the top on the left. Exon 3 (open bar), restriction sites, and potential cis-elements are indicated as in (A). Numbers are relative to the transcription initiation site of exon 3. Names of constructs, schematic drawings of DNA fragments inserted in the reporter vector, and relative luciferase activities are presented as in (A).
Basic vector. Additional 5P deletion clones (p1(393), p1(358), and p1(343), and p3(3109), p3(394), p3(368), p3(351), and p3(340)) were made from p1(3109) and p3(3158), respectively, using the Erase-a-Base System (Promega). Transfections and luciferase assays were performed as previously described [18]. A 2.4 kb genomic fragment, containing the 5P £anking region of exon 1, transactivated the reporter gene in NIH3T3 and, less e¡ectively, in Neuro2A (Fig. 5A, MN24). Several shorter genomic fragments di¡ering in their 5P ends showed similar or increased activity in both cell lines (Fig. 5A, PN1.8, p1(3109), and data not shown).
A 5P deletion construct missing the overlapping Sp1(A) and Sp1(B) sites resulted in a 70% reduction of transactivity relative to p1(3109), and a deletion construct missing all three Sp1 sites resulted in a 94% reduction of transactivity, in NIH3T3 (Fig. 5A, p1(393) and p1(358)). Similarly, 42% and 85% reductions of transactivity were observed with these deletion constructs in Neuro2A. These results suggest that the basal promoter activity resides in a 51 nucleotide region from 3109 to 358, which contains the three Sp1 sites. Although the p1(358) construct contained most of the Y/CCAAT box between 361 and 346, this element was insu¤cient for basal transactivity. How-
BBAEXP 91521 10-4-01
A. Nakayama et al. / Biochimica et Biophysica Acta 1518 (2001) 260^266
ever, a deletion construct eliminating this entire element resulted in a small but statistically signi¢cant reduction in transactivity (Fig. 5A, p1(343)). We introduced a series of mutations into the p1(3109) construct by changing the core sequences from CCCCGCCC to CCCtGCag in the Sp1 sites and from ACCAATG to ACtAgTG in the Y/CCAAT box with the QuikChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA). The disruption of the Sp1(A) site and the disruption of the Sp1(C) site reduced promoter activity in NIH3T3 (Fig. 5A, p1(3109)mSp1A and p1(3109)mSp1C), which was consistent with the data from the deletion experiments. However, the disruption of the Sp1(B) site did not a¡ect the transactivity (Fig. 5A, p1(3109)mSp1B). In Neuro2A, a signi¢cant reduction of the promoter activity was observed only by disruption of the Sp1(C) site. The experiments with the constructs bearing mutations in the Sp1(C) site con¢rmed that the Sp1(C) site is important for the basal activity of the 5P promoter. In contrast, the disruption of a single core site in either of the overlapping Sp1(A) or Sp1(B) sites did not signi¢cantly a¡ect the basal promoter activity. These data suggest that the overlapping Sp1(A) and Sp1(B) sites may be functionally redundant. Using similar techniques, we determined that the basal promoter activity resided in a 96 nucleotide region from 3136 to 340 relative to the putative initiation site of exon 3 (Fig. 5B). 5P deletion constructs, bearing stepwise deletions of the NF-1, HNF3B, AP-1/ATF, and Sp1(a) sites, resulted in a progressive reduction of transactivity in the two cell lines (Fig. 5B, p3(3136), p3(3109), p3(394), p3(368), and p3(351)). Neither p3(351) nor p3(340) transactivated the reporter gene, indicating that Sp1(b) alone has no promoter activity. However, the disruption of the Sp1(b) site in the basal intronic promoter resulted in a signi¢cant reduction of transactivity, especially in NIH3T3 (Fig. 5B, p3(3136) and p3(3136)mSp1b). Thus the Sp1(b) site is necessary but not su¤cient for the basal promoter activity. The disruptions of the Sp1(a) site and other elements by changing the sequences from AAGGCCAAG to AAGaattcG in the NF-1 site, from AGAAACAGTT to AGctAgcGTT in the HNF3B site, and from CCTCGTCACG to CCTgGTacCG in the AP1 site, also a¡ected the transactivity compared with the parental wild type constructs, p3(3136), in NIH3T3 (Fig. 5B, p3(3136)mNF1, p3(3136)mHNF, p3(3136)mAP1, and p3(3136)mSp1a). However, the disruptions of the Sp1(a) and NF-1 sites did not signi¢cantly a¡ect transactivity in Neuro2A (Fig. 5B, p3(3136)mNF1 and p3(3136)mSp1a). These results indicate that the HNF3B and AP-1/ATF sites contribute to the basal promoter activity of the intronic promoter in both NIH3T3 and Neuro2A. However, NF-1 and Sp1(a) are important for the basal promoter activity only in NIH3T3. Thus, the basal promoter activity of the intronic promoter depends on the additional elements as well as the Sp1 sites. In conclusion, our results suggest that the MAP1A gene
265
contains two TATA-less promoters that regulate the expression of two alternative transcripts, a novel long transcript and a short transcript homologous to rat MAP1A. Both luciferase reporter assays and DNase I footprinting suggested that both of the basal promoter activities depended upon Sp1 binding sites. RT-PCR demonstrated that the 5P alternative transcripts are expressed abundantly in brain, which is consistent with observations that MAP1A protein is highly expressed in neurons at the histological level [19]. However, expression of the short transcript was not restricted in a cell- or tissue-speci¢c manner (Fig. 1A and unpublished observation). Transactivity of the intronic promoter in both Neuro2A and NIH3T3 cells seemed to correlate well with the expression pattern of the short transcript in these cells. In contrast, the novel long transcript was preferentially expressed in Neuro2A cells, while the 5P promoter driving its expression demonstrated transactivity in both Neuro2A and NIH3T3. It is possible that the 5P promoter has regulatory elements that confer neuronal cell-speci¢c expression of the long transcript outside the region examined in this study. Further investigation of the two promoters and extensive study of the expression patterns of the two transcripts are necessary to fully understand the regulatory mechanisms that govern MAP1A expression. We are grateful to Dr. Taniguchi for kindly providing a mouse genomic library. This work was supported in part by a grant for COE research from the Ministry of Education, Science, Culture, and Sports of Japan. References [1] A. Matus, Microtubule-associated proteins : their potential role in determining neuronal morphology, Annu. Rev. Neurosci. 11 (1988) 29^44. [2] A. Matus, Sti¡ microtubules and neuronal morphology, Trends Neurosci. 17 (1994) 19^22. [3] J. Nunez, I. Fischer, Microtubule-associated proteins (MAPs) in the peripheral nervous system during development and regeneration, J. Mol. Neurosci. 8 (1997) 207^222. [4] G.S. Bloom, F.C. Luca, R.B. Vallee, Widespread cellular distribution of MAP-1A (microtubule-associated protein 1A) in the mitotic spindle and on interphase microtubules, J. Cell Biol. 98 (1984) 331^340. [5] P.W. Baas, Microtubules and neuronal polarity: lessons from mitosis, Neuron 22 (1999) 23^31. [6] Y. Shiomura, N. Hirokawa, The molecular structure of microtubuleassociated protein 1A (MAP1A) in vivo and in vitro, An immunoelectron microscopy and quick-freeze, deep-etch study, J. Neurosci. 7 (1987) 1461^1469. [7] R. Sato-Yoshitake, Y. Shiomura, H. Miyasaka, N. Hirokawa, Microtubule-associated protein 1B molecular structure, localization, and phosphorylation-dependent expression in developing neurons, Neuron 3 (1989) 229^238. [8] A. Langkopf, J.A. Hammarback, R. Mu«ller, R.B. Valle, C.C. Garner, Microtubule-associated proteins 1A and LC2: two proteins encoded in one messenger RNA, J. Biol. Chem. 267 (1992) 16561^ 16566. [9] R. Fukuyama, S.I. Rapoport, Brain-speci¢c expression of human microtubule-associated protein 1A (MAP1A) gene and its assignment to human chromosome 15, J. Neurosci. Res. 40 (1995) 820^825.
BBAEXP 91521 10-4-01
266
A. Nakayama et al. / Biochimica et Biophysica Acta 1518 (2001) 260^266
[10] C.C. Garner, A. Garner, G. Huber, C. Kozak, A. Matus, Molecular cloning of microtubule-associated protein 1 (MAP1A) and microtubule-associated protein 5 (MAP1B): identi¢cation of distinct genes and their di¡erential expression in developing brain, J. Neurochem. 55 (1990) 146^154. [11] D. Liu, I. Fischer, Two alternative promoters direct neuron-speci¢c expression of the rat microtubule-associated protein 1B gene, J. Neurosci. 16 (1996) 5026^5036. [12] L.L. Lien, C.A. Feener, N. Fischbach, L.M. Kunkel, Cloning of human microtubule-associated protein 1B and the identi¢cation of a related gene on chromosome 15, Genomics 22 (1994) 273^280. [13] W. Kutschera, W. Zauner, G. Wiche, F. Propst, The mouse and rat MAP1B genes: genomic organization and alternative transcription, Genomics 49 (1998) 430^436. [14] J.K. Fink, S.M. Jones, C. Esposito, J. Wilkowski, Human microtubule-associated protein 1a (MAP1A) gene genomic organization. cDNA sequence, and developmental- and tissue-speci¢c expression, Genomics 35 (1996) 577^585.
[15] M. Taniguchi, S. Yuasa, H. Fujisawa, I. Naruse, S. Saga, M. Mishina, T. Yagi, Disruption of semaphorin III/D gene causes severe abnormality in peripheral nerve projection, Neuron 19 (1997) 519^ 530. [16] K. Quandt, K. Frech, H. Karas, E. Wingender, T. Werner, MatInd and MatInspector ^ New fast and versatile tools for detection of consensus matches in nucleotide sequence data, Nucleic Acids Res. 23 (1995) 4878^4884. [17] J.D. Dignam, R.M. Lebovitz, R.G. Roeder, Accurate transcription initiation by RNA polymerase II in a soluble extract from isolated mammalian nuclei, Nucleic Acids Res. 11 (1983) 1475^1489. [18] Y. Iwata, A. Nakayama, H. Murakami, K. Iida, T. Iwashita, N. Asai, M. Takahashi, Characterization of the promoter region of the RFP gene, Biochem. Biophys. Res. Commun. 261 (1999) 381^384. [19] T.A. Schoenfeld, L. McKerracher,.R. Ober, B. Vallee, MAP1A and MAP1B are structurally related microtubule associated proteins with distinct developmental patterns in the CNS, J. Neurosci. 9 (1989) 1712^1730.
BBAEXP 91521 10-4-01
www.bba-direct.com
Promoter paper
Characterization of two promoters that regulate alternative transcripts in the microtubule-associated protein (MAP) 1A gene1 Atsuo Nakayama *, Takayuki Odajima, Hideki Murakami, Naoyoshi Mori, Masahide Takahashi Department of Pathology, Nagoya Univerty School of Medicine, 65 Tsurumai-cho, Showa-ku, Nagoya 466-8550, Aichi, Japan Received 4 October 2000; received in revised form 4 January 2001 ; accepted 10 January 2001
Abstract We cloned and characterized the mouse gene for microtubule-associated protein (MAP) 1A, an important protein for neuronal morphology and mitotic spindle formation. We also investigated the 5P untranslated region of the gene to characterize the promoter units. Two alternative transcripts different in the 5P region were identified by 5P RACE. Both transcripts were principally observed in the brain. Genomic cloning revealed that exons 1, 2, and 4 generate the 5P part of a long transcript, whereas exons 3 and 4 generate a short transcript. Putative 5P and intronic promoters flanking exons 1 and 3, respectively, are GC-rich and lack a canonical TATA box. DNase I footprinting from mouse cells revealed that several potential cis-elements were occupied by nuclear proteins. A reporter assay system in conjunction with a number of deletion and mutation constructs was used to test the two putative promoters. Both putative promoters showed transactivity and their function was dependent upon Sp1 sites. In addition, an NF-1 site, an HNF3B site, and an AP-1/ATF site were necessary for basal promoter activity of the intronic promoter. Our data provide insight into the regulatory mechanisms that govern the expression of the MAP1A gene. ß 2001 Elsevier Science B.V. All rights reserved. Keywords : Microtubule-associated protein 1A gene; Promoter; Gene structure ; Alternative transcript
Microtubule-associated proteins (MAPs) are co-puri¢ed with re-polymerized microtubules from brain homogenates [1] and play an important role in determining neuronal morphology and the balance between plasticity and stability of neuronal processes [2,3]. Fibrous MAPs are also involved in the stabilization of the mitotic spindle [4,5]. MAP 1A and 1B are major ¢lamentous high molecular MAPs, which are especially abundant in brain [6^9]. The structures of these two MAPs are closely related. However, MAP 1A and 1B are encoded by distinct genes and map to mouse chromosome 2 (Mtap1) and 13 (Mtap5), respectively [10]. The temporal expression patterns of these two MAPs, which are transcriptionally regulated, are complementary during central nervous system development [10]. Studies on the MAP1B gene and its transcripts have revealed the structure and the transcriptional regulatory units of this gene. In addition to a major transcript driven by two TATA boxes, both of which reside 5P of
* Corresponding author. Fax: +81-52-744-2091; E-mail : [email protected] 1 GenBank accession numbers AF182208^182213.
exon 1 [11], truncated forms of the transcripts, which start from exons 3U and 3A, have been identi¢ed in human, rat, and mouse [12,13]. Thus, the MAP1B gene may have multiple promoter units regulating the unique expression pattern of the gene. In contrast, the structure of the MAP1A gene has not been well characterized and its functional promoter has not been determined [14]. Here, we report the isolation of the mouse MAP1A gene and identi¢cation of two promoters that regulate the expression of the alternative transcripts. We obtained a 5P coding fragment of mouse MAP1A cDNA by RT-PCR from mouse brain total RNA with primers whose sequences are conserved between rat and human. The ampli¢ed 664 bp fragment was 95.2% identical with the corresponding rat cDNA (GenBank accession number M38196) and 89.5% with the human cDNA (GenBank accession number U38291). The amino acid sequence was completely identical with the amino-terminal sequence of the rat MAP1A peptide spanning amino acids 4^224 [8]. Based on the sequence information of this fragment, we performed 5P RACE using the 5P RACE system version 2 (Gibco-BRL, Rockville, MD, USA). Reverse transcribed product prepared from mouse brain RNA,
0167-4781 / 01 / $ ^ see front matter ß 2001 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 4 7 8 1 ( 0 1 ) 0 0 1 7 3 - 7
BBAEXP 91521 10-4-01
A. Nakayama et al. / Biochimica et Biophysica Acta 1518 (2001) 260^266
by priming with primer R1384 (5P-GCC CGA TGT TTG ATT TTG CTG-3P), was subjected to oligo-dCTP tailing and the following PCR ampli¢cation. The primers used in the PCR were AUAP supplied by the manufacturer and L1261 (5P-ATC TTG TTC TGG GGA GCA TTT-3P) in the ¢rst round. UAP supplied by the manufacturer and L1106 (5P-CTT TAC TAT TGC CTG CCC ACT T-3P) were used in the second round. After two rounds of ampli¢cation, two major RACE products, 1.1 kb and 1.4 kb in length, were obtained. Sequencing revealed that the 5P 80 nucleotides of the 1.1 kb product and 5P 373 nucleotides of the 1.4 kb product were di¡erent (Fig. 1A, upper panel), while about 1 kb of the sequences were identical (GenBank accession numbers AF182212 and AF182213). To con¢rm that the two transcripts, a short transcript corresponding to the 1.1 kb RACE product and a long transcript corresponding to the 1.4 kb product, are ex-
261
pressed in mouse tissues, we performed RT-PCR analysis. Primer pairs SU2 (5P-GAC TCT TCC CTG AAG CTG CC-3P) and CL273/566 (5P-AAG GAC CTG GGC GAA GTT TT-3P) were designed to amplify the short transcript, and LU179 (5P-GCT GGG ACC TGC GGA AAT AC-3P) and CL273/566 were used in ampli¢cation of the long transcript. Both transcripts were highly expressed in brain (Fig. 1A), and they were also expressed in mouse embryos (12.5 days post conception). The short transcript was also expressed at low levels in heart, muscle, and several other organs. However, the long transcript was not detectable in these organs. Although both transcripts were principally expressed in brain, the short transcript had a broader tissue-speci¢c expression pattern. We performed primer extension to obtain additional information on the 5P end nucleotide(s) of the transcripts using the Primer Extension System (Promega, Madison,
6
Fig. 1. Distribution of the 5P alternative transcripts and determination of their 5P ends. (A) A schematic representation of the two mouse MAP1A transcripts is shown. The common regions of the two transcripts are represented as open bars. Regions speci¢c for the short transcript are shaded and regions speci¢c for the long transcript are hatched. In-frame ATGs are indicated as open triangles (P), and inframe stop codons are indicated as solid triangles (S). Primers used for RT-PCR are represented as arrows below. 10 Wl of the PCR products from mouse brain (B), heart (H), lung (Lu), liver (Li), spleen (S), kidney (K), intestine (I), uterus (U), ovary (O), muscle (M), 12.5dpc embryo (E), Neuro2A (N2), and NIH3T3 (NIH) were analyzed on a 1.5% agarose gel. The upper panel shows products from the long transcript (L), and the middle panel shows products from the short transcript (S). At the bottom, actin bands are shown as a control for equal loading (A). (B) Identi¢cation of the 5P end of the short transcript. Primers PE1 (5PCGC AAA GAG GCT CCG CCG GGT TCT GA-3P) and PE2 (5PCCG GAG TTG TCT CCA TGG CAA CGC TG-3P) were designed to anneal with nucleotides 126^151 and 66^91 of the long 5P RACE product, respectively. Primer PE3 (5P-TGG TCT CGA GGG TCT CAC TCT GGT GG-3P) corresponds to nucleotides 97^122 of the short RACE product. A schematic representation of the primers along with the two transcripts is illustrated at the top. A single 122 nucleotide band of extended product was identi¢ed on the sequencing gel, when primer PE3 was annealed with 10 Wg of mouse brain poly(A) RNA (lane 3). No extended products could be seen with primers PE1 and PE2 (lanes 1 and 2). Sequencing ladders of the corresponding genomic regions primed with PE1 and PE2 were run on both sides as molecular size markers. (C) RNase protection assay determining the 5P end of the long transcript. A genomic fragment from 3465 to +82 relative to the putative transcription initiation nucleotide of exon 1, determined by 5P RACE, was subcloned into pGEM-7Zf vector (Promega). This subclone was used as a template to make a 32 P-labeled antisense RNA probe. Schematic representations of the antisense RNA probe, the in vitro transcribed RNA from the long RACE product (ivt LT), and its truncated form (ivt TLT) are shown. RNA fragments protected by 10 Wg (lane 5) and 50 Wg (lanes 6 and 7) of mouse brain total RNA exhibited the same banding pattern as RNA fragments from in vitro transcribed RNA from the long RACE product (lane 4). For comparison, in vitro transcribed RNA from the 5P truncated long RACE product was also assayed (lane 3). Discrete protected bands could not be identi¢ed with 50 Wg [8] of yeast tRNA (lane 1), the in vitro transcribed short transcript (lane 2), 10 Wg of mouse liver total RNA (lane 8), or 50 Wg of mouse liver total RNA (lane 9).
BBAEXP 91521 10-4-01
262
A. Nakayama et al. / Biochimica et Biophysica Acta 1518 (2001) 260^266
Fig. 2. Structure of the 5P portion of the MAP1A gene. (A) Exons 1^6 (open bars) and BamHI (B), EcoRI (E), and XhoI (X) sites are illustrated. The 3P end of exon 6 has not been determined. Inserts of isolated phage clones are represented as lines at the top. Splicing patterns resulting in the 5P alternative transcripts are shown below. Potential coding regions are shaded. (B) Exons 1^6 of the MAP1A gene are aligned with exons 1^5 of the mouse MAP1B gene [13]. Each exon is represented as a box. The numbers inside the boxes indicate the exon sizes in nucleotides. The splicing patterns resulting in the short transcript for MAP1A and a truncated form for MAP1B are indicated by solid lines connecting the exons. The splicing patterns resulting in the long transcript for MAP1A and MAP1B are indicated by broken lines.
WI, USA). We identi¢ed a single 122 nucleotide band of extended product for the short transcript (Fig. 1B, lane 3). Although both primer extension and 5P RACE depend upon reverse transcription without premature cessation, our results suggested that the 5P end nucleotide identi¢ed by 5P RACE was the major transcription initiation nucleotide of the short transcript. We could not detect extension products for the long transcript despite using two di¡erent primers (Fig. 1B, lanes 1 and 2). Our inability to detect extension products may be attributed to a high stacking 5P sequence in the long transcript. We performed RNase protection assays using the RPA II (Ambion, Austin, TX, USA) to determine the transcription initiation nucleotide(s) of the long transcript. A genomic fragment covering the 5P part of exon 1 and the 5P £anking region was subcloned and used as a template to make an antisense RNA probe (Fig. 1C, upper panel; see Figs. 2 and 3 for the genomic structure and sequence). Mouse brain RNA protected the antisense RNA probe while mouse liver RNA did not (Fig. 1C, lanes 5^9). In vitro transcribed RNA corresponding to the long RACE product resulted in a single protected band (Fig. 1C, lane 4), but in vitro transcribed RNA from the short RACE product and yeast tRNA did not protect the antisense probe (Fig. 1C, lanes 1 and 2). The fragments protected by brain RNA and in vitro transcribed RNA from the long RACE product were the same size. For comparison, in vitro transcribed RNA from the 5P truncated long RACE product was also examined, and it protected a shorter fragment in
the expected size (Fig. 1C, lane 3). This result con¢rmed that the 5P end of the long RACE product was the major transcription initiation site of the long transcript. To isolate MAP1A genomic clones, we screened 2U106 clones of a mouse genomic lambda phage library [15] with the 664 bp PCR fragment described above as a probe. Two overlapping clones, N4 and M1, out of 14 positive clones isolated were analyzed. Mapping of restriction enzyme sites, Southern blotting, and sequencing yielded the 5P genomic structure of the gene (Fig. 2A, GenBank accession numbers AF182208^182211). The two transcripts di¡erent in the 5P region appear to be splicing variants. Exons 1, 2, and 4 correspond to the 5P region of the long transcript, while exon 3 spliced with exon 4 correspond to the 5P region of the short transcript. The intron^exon boundaries (Fig. 2A) satis¢ed the AG-GT rule with the exception of the 5P boundary of exon 3. Comparison of the 5P organization of the MAP1A and MAP1B genes revealed that their exon^intron structures are quite similar
Fig. 3. Nucleotide sequences of the MAP1A promoter regions and potential cis-acting elements. (A) Nucleotide sequence of exon 1 and £anking regions. The 5P end of exon 1 is indicated by an arrow (s) and numbered as +1. The base sequence of exon 1 is in uppercase letters and the sequence of the 5P upstream region and part of intron 1 are in lowercase letters. Potential Sp1 binding sites (Sp1(A), (B), (C)) and a Y/CCAAT box (Y/CCAAT) are boxed. Open triangles (P) indicate the 5P ends of the deletion constructs used in the luciferase assay. Their names are shown above. (B) Nucleotide sequence of exon 3 and parts of introns 2 and 3. The 5P end of exon 3 is indicated as in (A). The sequence of exon 3 is in uppercase letters and the sequences of introns 2 and 3 are in lowercase letters. Potential cis-elements and 5P ends of the deletion constructs are indicated as in (A).
BBAEXP 91521 10-4-01
A. Nakayama et al. / Biochimica et Biophysica Acta 1518 (2001) 260^266
(Fig. 2B) [13]. The sizes of exons 2, 4, and 5 of MAP1A were identical with those of exons 2, 3, and 4 of MAP1B. The nucleotide sequences were 53.9^59.0% identical when aligned (GenBank accession number X51386). Exons 1 of MAP1A and 1B share 54.6% sequence identity, however, the precise size of MAP1B exon 1 is undetermined. This ¢nding supports the concept that the two genes are derived from a common ancestral gene [8]. Sequencing of the 5P £anking regions of exons 1 and 3 revealed that both of the promoter regions are GC-rich and lack a canonical TATA box. The sequences were further analyzed with MatInspector V2.2 using the TRANSFAC 4.0 database at http://transfac.gbf.de [16] to identify potential cis-elements. Three Sp1 sites, at 3107, 3102, and 380, (Sp1(A), Sp1(B), and Sp1(C)) and a Y/CCAAT box, at 361 relative to the transcription initiation nucleotide of exon 1, were identi¢ed (v0.95 core similarity and v0.90 matrix similarity ; Fig. 3A). Two Sp1 sites at 364 and 353 (Sp1(a) and Sp1(b)), a CCAAT box at 3152, an NF-1 site at 3123, an HNF3B site at 3111, and an AP-1/ ATF site at 389 relative to the putative transcription initiation site of exon 3 were identi¢ed (Fig. 3B). We performed DNase I footprinting to determine which of these potential cis-elements are bound by nuclear proteins. Neuro2A, mouse neuroblastoma cells, and NIH3T3, mouse ¢broblast cells, have been reported to express MAP1A protein [4] and were con¢rmed to express both the 5P alternative transcripts. However, the long transcript was expressed at a very low level in NIH3T3 (Fig. 1A). We prepared nuclear proteins from these two cell lines as previously described [17]. Three Sp1 sites and the Y/CCAAT box in the 5P promoter were protected from DNase I digestion by NIH3T3 nuclear proteins and the Sp1(C) site and the Y/CCAAT box were protected by Neuro2A nuclear proteins (Fig. 4A). The overlapping Sp1(A) and Sp1(B) sites were not protected even with 20 Wg of Neuro2A nuclear extract, when a DNA fragment end-labeled in the coding strand was used as the probe (not shown). However, the Sp1(B) site was protected with 10 Wg of Neuro2A nuclear extract, when a DNA fragment end-labeled in the non-coding strand was employed (Fig. 4A, a bottom picture). These results suggest that the three Sp1 sites and the Y/CCAAT box in the 5P promoter region are functional cis-elements. When the intronic promoter region was analyzed by DNase I footprinting, a long segment including the CCAAT box, the NF-1, HNF3B, and AP1/ATF sites was protected by nuclear proteins from NIH3T3 and Neuro2A (Fig. 4B). The Sp1(b) site was protected by NIH3T3 extracts, whereas DNase I hypersensitive bands were only observed at the Sp1(a) site. Thus, most of the potential elements in the intronic promoter seemed to be involved in assembling nuclear trans-factors. We generated luciferase reporter constructs to examine whether the two promoter regions have functional transactivities, and to determine which potential elements are
263
Fig. 4. Nuclear extracts from NIH3T3 and Neuro2A contain binding proteins to the cis-elements of MAP1 promoters. (A) DNase I footprinting analysis of the 5P promoter region. A DNA fragment spanning from 3141 to +13 relative to the transcription initiation nucleotide of exon 1 was 32 P-labeled at the 3P end of the coding strand and was incubated with either 10 Wg (10) or 3 Wg (3) of nuclear proteins from NIH3T3 or Neuro2A. The incubation was carried out in 100 Wl of binding bu¡er (20 mM Tris, 1 mM EDTA, 10% glycerol, 0.1% NP-40, 2 mM MgCl2 , 1 mM dithiothreitol, 2% polyvinyl alcohol, and 1 Wg of poly(dI-dC)) at room temperature for 20 min. DNase I digestion was performed on ice for 1 min. 10 Wg of BSA (B) was used as a control. The Maxam^Gilbert sequencing G ladder of the probe is used as a molecular size marker (G). The positions of potential cis-elements are indicated on the left (S(A): Sp1(A) site, S(B): Sp1(B) site, S(C): Sp1(C) site, and Y/C: Y/CCAAT box). A limited gel image using the DNA fragment labeled at the 3P end of the non-coding strand is shown at the bottom of the right panel. Regions protected from DNase I digestion and DNase I hypersensitive sites are indicated by hatched bars and open triangles, respectively. (B) DNase I footprinting analysis of the MAP1A intronic promoter region. The results are represented as in (A). The DNA fragment spanning 3158 to +56 relative to the transcription initiation nucleotide of exon 3 was labeled at the 3P end of the non-coding strand. Thus the G ladder represents C positions in the coding strand. Protected regions and positions of potential cis-elements are indicated as in (A) (S(a): Sp1(a) site, S(b): Sp1(b) site, A: AP-1/ATF site, H: HNF3B site, N: NF-1 site, and C: CCAAT box).
critical for the transactivities. The genomic fragments inserted into the pGL3 Basic vector (Promega) are shown in Fig. 5. Constructs MN24, PN1.8, and SpN20 were cloned by utilizing restriction sites of the MAP1A gene indicated in Fig. 5. Constructs p1(3109), p3(3158), and p3(3136) were made by inserting PCR fragments, with appropriate restriction sites engineered at both ends, into the pGL3
BBAEXP 91521 10-4-01
264
A. Nakayama et al. / Biochimica et Biophysica Acta 1518 (2001) 260^266
Fig. 5. Functional analysis of the 5P and intronic promoters in mouse cell lines. (A) The luciferase reporter assay with 5P deletion and mutant constructs suggested that the 5P promoter activity depended upon multiple Sp1 binding sites. The names of the constructs and schematic drawings of genomic fragments inserted in the reporter vector are shown on the left. Exon 1 (open bar), restriction sites (MluI, PstI, and NcoI) employed to generate the constructs, and potential cis-elements are illustrated in a partial MAP1A genomic map at the top. Numbers are relative to the transcription initiation nucleotide of exon 1. Genomic fragments represented by arrows positioned in front of the luciferase reporter gene in the pGL3 Basic vector. Restriction fragments (MN24 and PN1.8) and 5P deletion fragments (p1(3109), p1(393), p1(358), and p1(343)) were tested for their transactivities. Reporter constructs with mutations in the Sp1(A) site (p1(3109)mSp1A), in the Sp1(B) site (p1(3109)mSp1B), in the Sp1(C) site (p1(3109)mSp1C), and in the Y/CCAAT box (p1(3109)mNFY) were also analyzed. Open bars represent the average relative luciferase activities given as activity `X' fold above the promoterless pGL3 Basic vector in NIH3T3 and Neuro2A. *P 6 0.0001. (B) The intronic promoter activity depends on other elements as well as Sp1 binding sites. A partial MAP1A genomic map surrounding exon 3 is shown at the top on the left. Exon 3 (open bar), restriction sites, and potential cis-elements are indicated as in (A). Numbers are relative to the transcription initiation site of exon 3. Names of constructs, schematic drawings of DNA fragments inserted in the reporter vector, and relative luciferase activities are presented as in (A).
Basic vector. Additional 5P deletion clones (p1(393), p1(358), and p1(343), and p3(3109), p3(394), p3(368), p3(351), and p3(340)) were made from p1(3109) and p3(3158), respectively, using the Erase-a-Base System (Promega). Transfections and luciferase assays were performed as previously described [18]. A 2.4 kb genomic fragment, containing the 5P £anking region of exon 1, transactivated the reporter gene in NIH3T3 and, less e¡ectively, in Neuro2A (Fig. 5A, MN24). Several shorter genomic fragments di¡ering in their 5P ends showed similar or increased activity in both cell lines (Fig. 5A, PN1.8, p1(3109), and data not shown).
A 5P deletion construct missing the overlapping Sp1(A) and Sp1(B) sites resulted in a 70% reduction of transactivity relative to p1(3109), and a deletion construct missing all three Sp1 sites resulted in a 94% reduction of transactivity, in NIH3T3 (Fig. 5A, p1(393) and p1(358)). Similarly, 42% and 85% reductions of transactivity were observed with these deletion constructs in Neuro2A. These results suggest that the basal promoter activity resides in a 51 nucleotide region from 3109 to 358, which contains the three Sp1 sites. Although the p1(358) construct contained most of the Y/CCAAT box between 361 and 346, this element was insu¤cient for basal transactivity. How-
BBAEXP 91521 10-4-01
A. Nakayama et al. / Biochimica et Biophysica Acta 1518 (2001) 260^266
ever, a deletion construct eliminating this entire element resulted in a small but statistically signi¢cant reduction in transactivity (Fig. 5A, p1(343)). We introduced a series of mutations into the p1(3109) construct by changing the core sequences from CCCCGCCC to CCCtGCag in the Sp1 sites and from ACCAATG to ACtAgTG in the Y/CCAAT box with the QuikChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA). The disruption of the Sp1(A) site and the disruption of the Sp1(C) site reduced promoter activity in NIH3T3 (Fig. 5A, p1(3109)mSp1A and p1(3109)mSp1C), which was consistent with the data from the deletion experiments. However, the disruption of the Sp1(B) site did not a¡ect the transactivity (Fig. 5A, p1(3109)mSp1B). In Neuro2A, a signi¢cant reduction of the promoter activity was observed only by disruption of the Sp1(C) site. The experiments with the constructs bearing mutations in the Sp1(C) site con¢rmed that the Sp1(C) site is important for the basal activity of the 5P promoter. In contrast, the disruption of a single core site in either of the overlapping Sp1(A) or Sp1(B) sites did not signi¢cantly a¡ect the basal promoter activity. These data suggest that the overlapping Sp1(A) and Sp1(B) sites may be functionally redundant. Using similar techniques, we determined that the basal promoter activity resided in a 96 nucleotide region from 3136 to 340 relative to the putative initiation site of exon 3 (Fig. 5B). 5P deletion constructs, bearing stepwise deletions of the NF-1, HNF3B, AP-1/ATF, and Sp1(a) sites, resulted in a progressive reduction of transactivity in the two cell lines (Fig. 5B, p3(3136), p3(3109), p3(394), p3(368), and p3(351)). Neither p3(351) nor p3(340) transactivated the reporter gene, indicating that Sp1(b) alone has no promoter activity. However, the disruption of the Sp1(b) site in the basal intronic promoter resulted in a signi¢cant reduction of transactivity, especially in NIH3T3 (Fig. 5B, p3(3136) and p3(3136)mSp1b). Thus the Sp1(b) site is necessary but not su¤cient for the basal promoter activity. The disruptions of the Sp1(a) site and other elements by changing the sequences from AAGGCCAAG to AAGaattcG in the NF-1 site, from AGAAACAGTT to AGctAgcGTT in the HNF3B site, and from CCTCGTCACG to CCTgGTacCG in the AP1 site, also a¡ected the transactivity compared with the parental wild type constructs, p3(3136), in NIH3T3 (Fig. 5B, p3(3136)mNF1, p3(3136)mHNF, p3(3136)mAP1, and p3(3136)mSp1a). However, the disruptions of the Sp1(a) and NF-1 sites did not signi¢cantly a¡ect transactivity in Neuro2A (Fig. 5B, p3(3136)mNF1 and p3(3136)mSp1a). These results indicate that the HNF3B and AP-1/ATF sites contribute to the basal promoter activity of the intronic promoter in both NIH3T3 and Neuro2A. However, NF-1 and Sp1(a) are important for the basal promoter activity only in NIH3T3. Thus, the basal promoter activity of the intronic promoter depends on the additional elements as well as the Sp1 sites. In conclusion, our results suggest that the MAP1A gene
265
contains two TATA-less promoters that regulate the expression of two alternative transcripts, a novel long transcript and a short transcript homologous to rat MAP1A. Both luciferase reporter assays and DNase I footprinting suggested that both of the basal promoter activities depended upon Sp1 binding sites. RT-PCR demonstrated that the 5P alternative transcripts are expressed abundantly in brain, which is consistent with observations that MAP1A protein is highly expressed in neurons at the histological level [19]. However, expression of the short transcript was not restricted in a cell- or tissue-speci¢c manner (Fig. 1A and unpublished observation). Transactivity of the intronic promoter in both Neuro2A and NIH3T3 cells seemed to correlate well with the expression pattern of the short transcript in these cells. In contrast, the novel long transcript was preferentially expressed in Neuro2A cells, while the 5P promoter driving its expression demonstrated transactivity in both Neuro2A and NIH3T3. It is possible that the 5P promoter has regulatory elements that confer neuronal cell-speci¢c expression of the long transcript outside the region examined in this study. Further investigation of the two promoters and extensive study of the expression patterns of the two transcripts are necessary to fully understand the regulatory mechanisms that govern MAP1A expression. We are grateful to Dr. Taniguchi for kindly providing a mouse genomic library. This work was supported in part by a grant for COE research from the Ministry of Education, Science, Culture, and Sports of Japan. References [1] A. Matus, Microtubule-associated proteins : their potential role in determining neuronal morphology, Annu. Rev. Neurosci. 11 (1988) 29^44. [2] A. Matus, Sti¡ microtubules and neuronal morphology, Trends Neurosci. 17 (1994) 19^22. [3] J. Nunez, I. Fischer, Microtubule-associated proteins (MAPs) in the peripheral nervous system during development and regeneration, J. Mol. Neurosci. 8 (1997) 207^222. [4] G.S. Bloom, F.C. Luca, R.B. Vallee, Widespread cellular distribution of MAP-1A (microtubule-associated protein 1A) in the mitotic spindle and on interphase microtubules, J. Cell Biol. 98 (1984) 331^340. [5] P.W. Baas, Microtubules and neuronal polarity: lessons from mitosis, Neuron 22 (1999) 23^31. [6] Y. Shiomura, N. Hirokawa, The molecular structure of microtubuleassociated protein 1A (MAP1A) in vivo and in vitro, An immunoelectron microscopy and quick-freeze, deep-etch study, J. Neurosci. 7 (1987) 1461^1469. [7] R. Sato-Yoshitake, Y. Shiomura, H. Miyasaka, N. Hirokawa, Microtubule-associated protein 1B molecular structure, localization, and phosphorylation-dependent expression in developing neurons, Neuron 3 (1989) 229^238. [8] A. Langkopf, J.A. Hammarback, R. Mu«ller, R.B. Valle, C.C. Garner, Microtubule-associated proteins 1A and LC2: two proteins encoded in one messenger RNA, J. Biol. Chem. 267 (1992) 16561^ 16566. [9] R. Fukuyama, S.I. Rapoport, Brain-speci¢c expression of human microtubule-associated protein 1A (MAP1A) gene and its assignment to human chromosome 15, J. Neurosci. Res. 40 (1995) 820^825.
BBAEXP 91521 10-4-01
266
A. Nakayama et al. / Biochimica et Biophysica Acta 1518 (2001) 260^266
[10] C.C. Garner, A. Garner, G. Huber, C. Kozak, A. Matus, Molecular cloning of microtubule-associated protein 1 (MAP1A) and microtubule-associated protein 5 (MAP1B): identi¢cation of distinct genes and their di¡erential expression in developing brain, J. Neurochem. 55 (1990) 146^154. [11] D. Liu, I. Fischer, Two alternative promoters direct neuron-speci¢c expression of the rat microtubule-associated protein 1B gene, J. Neurosci. 16 (1996) 5026^5036. [12] L.L. Lien, C.A. Feener, N. Fischbach, L.M. Kunkel, Cloning of human microtubule-associated protein 1B and the identi¢cation of a related gene on chromosome 15, Genomics 22 (1994) 273^280. [13] W. Kutschera, W. Zauner, G. Wiche, F. Propst, The mouse and rat MAP1B genes: genomic organization and alternative transcription, Genomics 49 (1998) 430^436. [14] J.K. Fink, S.M. Jones, C. Esposito, J. Wilkowski, Human microtubule-associated protein 1a (MAP1A) gene genomic organization. cDNA sequence, and developmental- and tissue-speci¢c expression, Genomics 35 (1996) 577^585.
[15] M. Taniguchi, S. Yuasa, H. Fujisawa, I. Naruse, S. Saga, M. Mishina, T. Yagi, Disruption of semaphorin III/D gene causes severe abnormality in peripheral nerve projection, Neuron 19 (1997) 519^ 530. [16] K. Quandt, K. Frech, H. Karas, E. Wingender, T. Werner, MatInd and MatInspector ^ New fast and versatile tools for detection of consensus matches in nucleotide sequence data, Nucleic Acids Res. 23 (1995) 4878^4884. [17] J.D. Dignam, R.M. Lebovitz, R.G. Roeder, Accurate transcription initiation by RNA polymerase II in a soluble extract from isolated mammalian nuclei, Nucleic Acids Res. 11 (1983) 1475^1489. [18] Y. Iwata, A. Nakayama, H. Murakami, K. Iida, T. Iwashita, N. Asai, M. Takahashi, Characterization of the promoter region of the RFP gene, Biochem. Biophys. Res. Commun. 261 (1999) 381^384. [19] T.A. Schoenfeld, L. McKerracher,.R. Ober, B. Vallee, MAP1A and MAP1B are structurally related microtubule associated proteins with distinct developmental patterns in the CNS, J. Neurosci. 9 (1989) 1712^1730.
BBAEXP 91521 10-4-01