HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 277, Issue 33, 29983-29991, August 16, 2002

This Article Abstract Full Text (PDF) All Versions of this Article: 277/33/29983    most recent M112121200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Nielsen, A. L. Articles by Jørgensen, A. L. Search for Related Content PubMed PubMed Citation Articles by Nielsen, A. L. Articles by Jørgensen, A. L. Social Bookmarking                      What's this?

A New Splice Variant of Glial Fibrillary Acidic Protein, GFAP, Interacts with the Presenilin Proteins^*

Anders Lade Nielsen^§¶, Ida E. Holm, Marianne Johansen, Bjarne Bonven^§, Poul Jørgensen^§, and Arne Lund Jørgensen

From the  Department of Human Genetics and the ^§ Department of Molecular Biology, University of Aarhus and the  Department of Neurology, Aarhus University Hospital, DK-8000 Aarhus C, Denmark

Received for publication, December 19, 2001, and in revised form, May 29, 2002

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

We describe a new human isoform, GFAP, of the intermediary filament protein GFAP (glial fibrillary acidic protein). GFAP mRNA is the result of alternative splicing and a new polyadenylation signal, and thus GFAP has a new C-terminal protein sequence. This provides GFAP with the capacity for specific binding of presenilin proteins in yeast and in vitro. Our observations suggest a direct link between the presenilins and the cytoskeleton where GFAP is incorporated. Mutations in GFAP and presenilins are associated with Alexander disease and Alzheimer's disease, respectively. Accordingly, GFAP should be taken into consideration when studying neurodegenerative diseases.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Glial fibrillary acidic protein (GFAP)¹ belongs to class III of the intermediate filament (IF) proteins that have a characteristic structure composed of a highly conserved central -helical rod domain flanked by nonhelical head and tail domains (1, 2). GFAP is most abundant in mature astrocytes, and the expression is induced by astrocytic activation and accordingly during aging. Two other class III IF proteins, nestin and vimentin, dominate in immature astrocytes, whereas vimentin and GFAP are the main IF proteins in mature astrocytes. The human GFAP is a 432-amino acid-long polypeptide of 55 kDa encoded by the GFAP gene on chromosome 17q21 (3, 4). The gene has nine exons and extends over 10 kb with nine exons (5). GFAP shows a high degree of homology among species (3, 5). Regulatory elements directing astrocyte-specific transcription have been identified in both the human and mouse GFAP genes (6-8).

The astrocytic mRNA of 2.9 kb represents the dominating GFAP isoform (GFAP) in the central nervous system (3, 4). Three additional minor isoforms, termed GFAP, -, and -, have been described in the rodent. Isoform GFAP transcription starts 169 nucleotides upstream of GFAP and was described in Schwann cells of the peripheral nervous system (9, 10); GFAP mRNA is about 2.4 kb and lacks exon 1 but includes the last 126 nucleotides of intron 1. GFAP is expressed outside the brain (11); the mRNA of GFAP is 4.2 kb and includes all 9 GFAP exons and, in addition, the last 1255 bp of intron 7 spliced in frame to exon 7 (12). The derived hypothetical GFAP protein has a new tail domain where the normal C-terminal 42 amino acids encoded by exons 8 and 9 have been replaced by the 33 amino acids encoded by intron 7 sequences used as exon (12).

Missense mutations in the tail domain and the rod domain of GFAP have been implicated in the neurodegenerative process of Alexander disease where astrocytes accumulate GFAP-containing cytoplasmic aggregates (13). Here we describe a novel human GFAP isoform designated GFAP. GFAP has a novel C-terminal tail domain, the integrity of which is required for binding to the Alzheimer's disease-associated transmembrane proteins presenilin 1 and 2 in yeast and in vitro (14, 15). Thus, GFAP represents a functionally distinct isoform in terms of presenilin interactions, and this finding suggests a direct link between presenilins and the cytoskeleton.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Plasmids-- Details on individual plasmid constructs, which were all verified by sequencing, are available upon request. Human cDNA for GFAP was cloned by PCR from a brain cDNA library. Point mutations were generated as described for QuickChange site-directed mutagenesis (Stratagene). For yeast two-hybrid assays, DBD and AAD fusion proteins were expressed from the yeast multicopy plasmids pBTM116m (16) and pGAD10 (CLONTECH), respectively. His₆-tagged constructs were obtained by subcloning the indicated cDNAs into pRSETB (Invitrogen). For GST pull-down experiments and far Westerns, cDNA was subcloned into pGEX2TK (Amersham Biosciences). For mammalian expression pSG5 (16), pcDNA3 (Invitrogen) or green fluorescent fusion vector pEGFP (CLONTECH) was used.

cDNA Library Screening and Yeast Transactivation Assays-- A yeast GAL4 activation domain tagged cDNA library derived from human fetal brain (CLONTECH) was introduced by LiAc transformation into the Saccharomyces cerevisiae L40 strain expressing the fusion protein LexA-PS-1-(1-85) from the pBTM116m vector. Approximately 6 × 10⁶ yeast transformants were selected on TrpLeu plates containing 2 mM 3-aminotriazole. After 5 days His clones that have lacZ expression on X-gal indicator plates were isolated. Library plasmids were rescued in Escherichia coli strain JM110 (leuB) and introduced back into L40 expressing either unfused LexA or different LexA fusion proteins. Positives for interaction specifically with LexA-PS-1-(1-85) were DNA sequence analyzed. Yeast L40 transformants were grown exponentially for about five generations in selective medium. Yeast extracts were prepared and assayed for -galactosidase activity essentially as described by Rose et al. (17).

Northern Blotting and RT-PCR-- For Northern blotting experiments a mouse 298-bp fragment specific for exon 7a was amplified by PCR, purified, and randomly labeled with radioactivity. The probe was hybridized onto a mouse tissue MTN blot (CLONTECH) overnight at 65 °C in 5× SSPE (0.9 M NaCl, 0.05 M sodium phosphate, 4 mM EDTA), 0.02 mg/ml carrier DNA, 5× Denhardt's solution, and 0.5% (w/v) SDS. The filter was washed twice for 10 min at room temperature in 2× SSPE, 0.1% (w/v) SDS, twice for 15 min at 65 °C with 1× SSPE, 0.1% (w/v) SDS, and for high stringency twice for 10 min at 65 °C with 0.1× SSPE, 0.1% (w/v) SDS. The blot was revealed by autoradiography. For RT-PCR mouse brain RNA was prepared according to the protocol for the TRI Reagent (Sigma). RT-PCR was done according to the protocol for the Titanium One-step RT-PCR kit (CLONTECH). Briefly, 1 µg of mouse brain RNA was reverse-transcribed at 50 °C for 1 h and subsequently PCR-amplified (94 °C, 30 s; 62 °C, 30 s; and 68 °C, 1 min) for the indicated number of cycles. The following primers were used for the PCR: mGFAP exon 7 forward (PE7f), CATCACCATTCCTGTACAGACTTTC; mGFAP exon 8 reverse (PE8r), CCACGATGTTCCTCTTGAGGTG; and mGFAP exon 7a reverse (PE7ar), CCATTTACAATCTGGTGAGCCTG. The RT-PCR products were analyzed by 2% agarose gel electrophoresis.

In Vitro Binding Assays-- GST and GST-PS-1-(1-85) fusion proteins were expressed in E. coli XL1-blue, purified on glutathione-Sepharose beads, and GST pull-down assays performed essential as described (18, 19). His₆ epitope-tagged GFAP fusion proteins were expressed in E. coli BL21(DE3) and purified on Ni²⁺-chelating columns (Amersham Biosciences). Labeling of GST fusion proteins and far Western experiments were done exactly as described (20).

Immunological Methods-- GFAP antibody was raised in rabbit against a fusion protein GST-GFAP-(390-431) consisting of GST fused to amino acids 390-431 of GFAP. The polyclonal antibody was designated pAb-GFAP. In Western blotting the antibody was used in a 1:800 dilution. Green fluorescent protein (GFP) antibody (Roche Molecular Biochemicals) was used in a 1:1000 dilution. His tag antibody (Santa Cruz Biotechnology) was used in a 1:2000 dilution. In epi-immunofluorescence experiments pAb-GFAP was used in a 1:300 dilution; FLAG antibody 1B11 (18) was used in concentration 1:750, and fluorescein isothiocyanate or TRITC-labeled goat anti-rabbit or rabbit anti-mouse secondary antibodies (Molecular Probes) were diluted 1:200. For epi-immunofluorescence analysis cells were grown in slide flasks (Nunc), and to induce cytoskeletal rearrangement medium was changed to serum-free media 24 h after transfection. After a further 24 h of incubation, cells were washed twice in PBS, and fixed in 2% para-formaldehyde, 0.1% glutaraldehyde, 0.1% Triton X-100 for 30 min on ice. The fixated cells were washed in PBS, and antibody incubations were done in PBS, 5% fetal calf serum for 1 h at 4 °C.

To generate the Triton-insoluble cell fraction enriched in cytoskeletal proteins, cells were washed twice in PBS and scraped off the culture dishes. After centrifugation at 2000 × g for 5 min, the cell pellet was homogenized in 170 mM NaCl, 600 mM KCl, 1% (w/v) Triton X-100, 6 mM sodium phosphate (pH 7.4), 1 mM EDTA, and protease inhibitor mixture (homogenate fraction, HO). After centrifugation at 8000 × g for 10 min the supernatant was restored (TSE fraction), and the pellet was washed twice in PBS and resuspended in SDS-PAGE buffer (TIE fraction). Triton-insoluble extracts from brain material were similarly prepared except for the inclusion of an additional Triton extraction step.

Accession Number-- The GenBank^TM accession number for the human GFAP exon 7a sequence is AJ306447.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

GFAP Interacts with Presenilin 1 in Yeast and in Vitro-- We used the yeast two-hybrid system to identify cDNAs encoding proteins with a capacity for interacting with PS-1. A fusion between the DBD of the LexA protein and the N-terminal 85-amino acid cytoplasmic region of PS-1, DBD-PS-1-(1-85), was used as a bait to screen a library of human fetal brain cDNAs fused to the yeast GAL4 activation domain (AAD). Around 6 × 10⁶ yeast transformants were screened, and those that grew on selective histidine-deficient yeast plates and showed positive X-gal indication were isolated. The isolated cDNAs were used for yeast retransformation, and cDNAs were sequenced from transformants that were positive for interaction with PS-1, but negative for interaction with an unfused DBD or DBD fusions to unrelated proteins. Five cDNAs isolated were shown by BLAST sequence homology searches to have high homology to the DNA sequence of the gene coding for the human GFAP. Four of the five cDNA isolates were identical and encoded an entire GFAP protein except for the three N-terminal residues. They were named GFAP46. The sequence of the fifth cDNA isolate, named GFAP21, was shorter and included in the GFAP46 sequence.

In a liquid -galactosidase assay we tested the capacity of AAD-GFAP21 and AAD-GFAP46 to interact with DBD-PS-1-(1-85) or DBD-PS-2-(1-93) and thereby activated an integrated LexA cis-element regulated yeast lacZ promoter. DBD-PS-2-(1-93) and DBD-PS-1-(1-85) interacted equally well with AAD-GFAP21 and AAD-GFAP46 (Fig. 1a). No measurable increase in -galactosidase activity was observed by co-transforming AAD-GFAP21 and AAD-GFAP46 with DBD fused to the PS-1 cytoplasmic loop (PS-1-(257-376)), PS-1 C-terminal cytoplasmic domain (PS-1-(405-446)), or unrelated bait constructs (Fig. 1a). By DBD and AAD domain swapping a similar interaction was observed between DBD-GFAP46 and AAD-PS-1-(1-85) (data not shown).

View larger version (23K): [in this window] [in a new window]   Fig. 1.   a, GFAP interacts with presenilins. Co-expression of AAD-GFAP cDNAs and DBD-PS-1-(1-85) or DBD-PS-2-(1-93) in yeast strain L40 activates -galactosidase expression. L40 was transformed with high copy number plasmids containing DBD fusions to various parts of PS-1, PS-2, HP1, lamin, or no insert. PS-1-(1-85) corresponds to the first 85 amino acids of the cytoplasmic N-terminal region, PS-1-(257-376) to the cytoplasmic loop, and PS-1-(405-446) to the C-terminal region of the PS-1 molecule. Strains were retransformed with AAD alone or AAD fusions to GFAP cDNA identified in the yeast two-hybrid screen. Double transformants were assayed for -galactosidase activity expressed from an integrated lacZ gene transcriptionally regulated by LexA-binding sites. -Galactosidase activities are the averages from three independent transformants assayed the same day. b, PS-1 binds to GFAP46 in vitro. Purified N-terminal His-tagged GFAP46, His-GFAP46, was incubated in a batch assay with GST (2nd lane) or the fusion protein GST-PS-1-(1-85) (3rd lane) bound to glutathione S-Sepharose beads. Bound GFAP was monitored by Western blotting using an His tag antibody. The 1st lane shows 1/10 the amount of input His-GFAP46 (indicated by an arrow).

To examine a direct interaction between GFAP46 and PS-1, a fusion protein of GST and PS-1 amino acids 1-85, GST-PS-1-(1-85), was expressed in E. coli and purified. GFAP46 was expressed in E. coli as a fusion protein with a His₆ tag (His-GFAP46). The His₆ tag was used for purification of the protein by nickel column chromatography. The purified protein components were used in a GST pull-down assay. GST and GST-PS-1-(1-85) were bound to an glutathione S-Sepharose matrix and thereafter incubated with His-GFAP46. After extensive washings, the retained protein was analyzed by SDS-PAGE followed by Western blotting and visualized by an antibody against the His₆ tag. The input lane (Fig. 1b) corresponds to 1/10 of the material used in each GST pull-down assay. GST-PS-1-(1-85) retained a substantial amount of His-GFAP46 ensuring a direct interaction in vitro.

GFAP46 and GFAP21 Represent a New Splice Product, GFAP, of the GFAP Gene-- The sequences of GFAP21 and GFAP46 cDNA inserts revealed identity to the 5'-region of the normal human GFAP transcript, GFAP, but differed completely in the 3'-region (Fig. 2) (3, 4). This indicated that the GFAP21 and GFAP46 cDNAs represent a new mRNA splice product of the human GFAP gene which we designated GFAP. BLAST homology searches identified homology between the 3'-region of the GFAP21 and GFAP46 cDNAs and sequences included in a newly identified rat 4.2-kb GFAP splice variant designated GFAP (12). Sequence analysis showed that the divergence between GFAP and GFAP was the result of usage of a part of GFAP gene intron 7 as an exon in GFAP (Fig. 2b) (5). The new consensus splice acceptor site was identified in the genomic sequence (Fig. 2c). The new exon utilized in GFAP was designated exon 7a. A perfect match with the polyadenylation signal, AATAAA, was present in the extreme 3'-end of exon 7a, denoted pA (Fig. 2). The sequence of GFAP46 cDNA showed polyadenylation 26 bases downstream of the polyadenylation signal in exon 7a which therefore was functional in human cells (Fig. 2b). Moreover, the sequence of the mouse GFAP gene showed that splice acceptor sites corresponding to the alternative exon 7a splicing and the polyadenylation signal were evolutionary conserved (Fig. 2c). By usage of pA, exon 8 and exon 9 are skipped and the size of the human and mouse GFAP mRNAs will be 1.8 and 2.5 kb, respectively. Note that in mouse the GFAP and GFAP mRNAs will have roughly same molecular weight. GFAP mRNA was detected solely in the brain in Northern blot experiments by using human or mouse brain mRNA probed with exon 7a sequences (Fig. 2d and data not shown). The relative expression levels of the two GFAP isoforms were determined by RT-PCR. Mouse brain RNA was in a coupled RT-PCR amplified with either a GFAP-specific, a GFAP-specific, or a mixed primer set. The RT-PCR was run for a variable number of cycles to ensure reaction points in which the PCR amplification was exponentially increasing. The relative expression level of GFAP mRNA was estimated to be about 20-fold less than that of GFAP mRNA (Fig. 2e).

View larger version (41K): [in this window] [in a new window]   Fig. 2.   Sequence characteristics and splice pattern of the new human GFAP variant, GFAP. a, graphical representation (not scaled) of the 3'-GFAP gene structure. Exons are indicated by rectangles, and the length of the exons and introns is shown in the lower part of the figure. The GFAP transcription product includes exon 7a, whereas GFAP, -, and - transcripts utilize exons 8 and 9. As a result of the new polyadenylation signal in exon 7a, the transcript is truncated by 3.35 kb compared with a readthrough to the polyadenylation signal, pA, in exon 9. b, DNA sequence of the region, designated exon 7a, in the GFAP46 and GFAP21 cDNAs that is not present in GFAP. An arrow indicates the 3'-end of the GFAP21 cDNA. The consensus polyadenylation signal, pA, in exon 7a is underlined, and a part of the poly(A) stretch of GFAP46 is included in the sequence. The in-frame translated peptide of GFAP that replaces the normal GFAP C terminus encoded by exons 8 and 9 is shown above the DNA sequence. c, DNA homology of the GFAP exon 7a splice acceptor site (upper panel) and polyadenylation signal (lower panel) between human and mouse. The abbreviations used are as follows: y, pyrimidine; r, purine; u, uracil; and n for either pyrimidine or purine. Exon sequences are indicated by capital letters, and conserved residues are indicated by a vertical line. The splice consensus sequence is shown above the sequence, and the branch point is indicated by a dot. d, GFAP is identified in the brain by Northern blotting experiments. By PCR a 298-bp probe specific for the mouse exon 7a was amplified, radioactively labeled, and used as a probe on a multiple tissue Northern blot with mRNA from different mouse tissues. e, GFAP represents a minor GFAP mRNA species. The relative mRNA amounts of GFAP and GFAP were determined by RT-PCR using mouse brain RNA. Primers specific for the two GFAP isoforms (PE7f and PE8r for GFAP and PE7f and PE7ar for GFAP) were either individually used or mixed in the indicated number of PCR cycles. The PCR products were analyzed by 2% agarose gel electrophoresis. Sizes of DNA marker in lane M are indicated to the left. f, sequence homology between human and mouse protein sequences encoded by exon 7a of GFAP (upper part) and the exons 8 and 9 of GFAP (lower panel). Identical amino acids are indicated by a vertical line and conservative amino acid changes indicated by dots.

The alternative splicing of GFAP mRNA was translated into a unique tail domain of GFAP. The 43-amino acid C-terminal tail region of GFAP encoded by exon 8 and exon 9 in GFAP was replaced by the 42 amino acids encoded by exon 7a (Fig. 2f). This GFAP-specific tail domain showed only little homology to GFAP or other IF protein sequences (Fig. 2f).

Mapping of the Sequences in PS-1 Required for GFAP Interaction-- To detect the region in PS-1 responsible for the GFAP interaction, we introduced a series of N- and C-terminal deletions and missense mutations in DBD-PS-1-(1-85) and assayed for interaction with AAD-GFAP-(3-431) in the two-hybrid system (Fig. 3a). As exemplified by the fusion protein DBD-PS-1-(66-85), deletion of the first 65 of the N-terminal residues does not influence the AAD-GFAP interaction, and DBD-PS-1(1-85del-(66-72)), in which a highly acidic stretch of amino acids (66-72) has been deleted, also retained the capacity for interaction. By contrast, removal of only three amino acids from the C-terminal end, as exemplified by DBD-PS-1-(1-82), completely abolished interaction. The same effect was seen if these residues were changed by single or double missense mutations. We noted that interaction was abolished in the non-conservative amino acid substitutions V82K and V82E but was retained in DBD-PS-1(V82L) which carried a conservative amino acid substitution associated with familial Alzheimer's disease. Also DBD-PS-1(A79V) that is associated with Alzheimer's disease had no effect on GFAP binding (Fig. 3a).

View larger version (50K): [in this window] [in a new window]   Fig. 3.   PS-1 interaction requires residues encoded from the alternative spliced sequences in GFAP. a, GFAP interaction requires a short motif in PS-1. Deletions and point mutations expressed as DBD-PS-1 fusion proteins were tested for the capacity to interact with AAD-GFAP-(3-431). -Galactosidase activities are the average values from three independent transformants assayed the same day. The location of introduced point mutations are marked by asterisks. b, in vitro mapping of GFAP sequences required for PS-1 interaction. Various deletion mutants of GFAP or full-length GFAP were expressed as fusion proteins to the His tag in E. coli BL21(DE3) and purified in the presence of 8 M urea by nickel chromatography. Equal amounts of proteins were separated by SDS-PAGE and afterward blotted unto nitrocellulose filters. Filters were used in far Western analysis with a denaturation-renaturation cycle with guanidinium HCl followed by blocking with bovine serum albumin and GST protein. The filter was incubated with 100,000 cpm/ml 32P-labeled GST-PS-1-(1-85), extensively washed, and analyzed by autoradiography. c, Western blotting of the proteins from c. Nitrocellulose filters with 1/50 the amount of loaded protein in b were processed for Western blotting and revealed by a His tag antibody. Control lane indicated the loading of equivalent amounts of bovine serum albumin. d, yeast two-hybrid mapping of GFAP sequences required for PS-1 interaction. Deletions and point mutations in AAD-GFAP fusion proteins were assayed for the interaction with DBD-PS-1-(1-85). The complete coding region of GFAP was tested in a similar manner for PS-1 interaction. -Galactosidase activities are the average values from three independent transformants assayed the same day.

The Alternative Exon 7a in GFAP Is a Determinant for PS-1 Interaction in Vitro and in Yeast-- To map the amino acids in GFAP which take part in the PS-1 interaction, we utilized the far Western assay to monitor protein-protein interactions. Various deletions of GFAP were expressed in E. coli as fusion proteins to a His tag, purified, and blotted onto a nitrocellulose filter. The filter was probed with ³²P-labeled GST-PS-1-(1-85), and after an extensive wash, the retained radioactivity was monitored by autoradiography. As expected, GST-PS-1-(1-85) interacts with His-GFAP-(3-431) and His-GFAP-(204-431) which correspond to the sequences of GFAP46 and GFAP21, respectively (Fig. 3b, left panel). Labeled GST did not interact with these proteins (data not shown). His-GFAP-(309-431) and His-GFAP-(349-431) were negative for GST-PS-1-(1-85) interaction showing that sequences in the C-terminal end of the coiled-coil region, which is conserved among the different GFAP splice variants, are required for PS-1 interaction (Fig. 3b, left panel). Because His-GFAP-(204-390) was also negative for GST-PS-1-(1-85) interaction, the GFAP-specific C-terminal tail sequences were also required for PS-1 interaction (Fig. 3b). Purified full-length GFAP was found not to interact with GST-PS-1-(1-85) as expected because it did not include the PS-1 interaction region (Fig. 3b, right panel). By Western blotting using an antibody against the histidine tag, it was verified that an equal amount of proteins was used for the far Western analysis (Fig. 3c).

We also used the yeast two-hybrid system to define the GFAP residues required for PS-1 interaction at more physiological conditions. Consistent with the far Western results AAD-GFAP-(204-431) interacted with DBD-PS-1-(1-85), whereas deletion of GFAP sequences in the C-terminal end of the coiled-coil region in both AAD-GFAP-(309-431), AAD-GFAP-(349-431), and AAD-GFAP-(390-431) completely abolished the interaction (Fig. 3d). The region between amino acids 204 and 309 includes the linker 1-2, coiled-coil 2A, and the beginning of coiled-coil 2B. The deletion of exon 7a encoded amino acids in AAD-GFAP-(204-390) also abolished PS-1 interaction (Fig. 3d). In conclusion, the PS-1 interaction domain in GFAP is large or bipartite and requires both the GFAP-specific tail sequence as well as sequences overlapping with the coiled-coil 2 and linker 1-2 shared by GFAP and GFAP. Full-length GFAP was also tested for interaction with PS-1 and in agreement with the above mapping data and the far Western results (Fig. 3b) no interaction was observed between DBD-PS-1-(1-85) and AAD-GFAP.

GFAP Is an Expressed Protein in Vivo and Incorporates into Filaments-- 293 cells were transfected with expression plasmids encoding GFP alone or fused to the N-terminal of GFAP (pGFP-GFAP) or GFAP (pGFP-GFAP). Cellular extracts were analyzed by Western blotting using either an antibody against GFP or a pAb-GFAP that was raised in rabbit against the 42-amino acid C-terminal tail region specific for the GFAP isoform (Fig. 4a). Although the GFP antibody detected both GFP fusion proteins, the GFAP antibody recognizes only GFAP (lane 3) thus ensuring the antibody specificity. The Western blot in Fig. 4b shows GFAP expression in 293 cells transfected with the expression vector pSG5 without insert (lane 1), with the GFAP insert (lane 2), and a total cellular extract from the astrocyte-derived cell line SVG(P12) (lane 3). The antibody pAb-GFAP detected in lane 3 an endogenous band of the expected GFAP size (55 kDa) co-migrating with transfected untagged GFAP (lane 2). When the antibody was pre-adsorbed with the GFAP antigen the endogenous band of the SVG(P12) extract disappeared, whereas pre-adsorption with GST did not affect reactivity of the antibody (Fig. 4c). By fractionation of SVG(P12) cells to obtain the Triton X-100-insoluble extract (TIE) enriched in cytoskeletal proteins, GFAP localization was determined to be in the TIE fraction in agreement with GFAP being an intermediate filament protein (Fig. 4d). Also in a porcine brain extract a protein with the expected size of GFAP was observed in the TIE fraction enriched in cytoskeletal proteins together with a 35-kDa immunoreactivity of unknown origin (Fig. 4e) Thus the alternative GFAP splicing seems to be reflected in the expression of the corresponding GFAP protein in vivo.

View larger version (29K): [in this window] [in a new window]   Fig. 4.   GFAP is expressed as a protein that is integrated into filaments. a, Western blot of 293 cells transiently transfected with an expression vector encoding GFP (lane 1), N-terminal GFP tagged GFAP (lane 2), or N-terminal GFP tagged GFAP (lane 3). An antibody specific for GFP was used in the left panel. The antibody pAb-GFAP was used in the right panel. b, Western blot of cell extract from 293 cells transiently transfected with an expression vector without insert (lane 1), an GFAP expression vector encoding untagged protein (lane 2), and cell extract from the astrocyte-derived cell line SVG(P12) (lane 3). The antibody pAb-GFAP detects a prominent 55-kDa band in lanes 2 and 3. c, pre-adsorbing of pAb-GFAP abolishes reactivity to the SVG(P12) cell extract. The pAb-GFAP without pre-adsorption (lane 1), pAb-GFAP pre-adsorbed with GST-GFAP, and pAb-GFAP pre-adsorbed with GST alone (lane 3). A recognized band of 55 kDa indicated by an arrow disappeared by the affinity adsorption (lane 2). d, SVG(P12) cells were homogenized, and the Triton X-100-insoluble fraction enriched in cytoskeletal proteins prepared. Total homogenate (HO), Triton X-100-soluble material from the extraction (TSE), and the Triton X-100-insoluble fraction (TIE) in amounts corresponding to an equal number of input cells were subjected to 10% SDS-PAGE. The presence of GFAP was determined by Western blotting with the pAb-GFAP antibody. e, a porcine brain slice was homogenized, and the Triton X-100-insoluble fraction enriched in cytoskeletal proteins was prepared as in d except for the inclusion of an additional Triton X-100 extraction (TIE2). The presence of GFAP was determined by Western blotting with the pAb-GFAP antibody. U.P. indicates an observed band of unknown origin. f, native GFAP assembly into filaments. N2A cells were transfected with the pcDNA3 vector encoding entire untagged GFAP. Cells were processed for immunostaining with pAb-GFAP followed by a fluorescein isothiocyanate-labeled secondary antibody. Labeled cells were visualized by epi-immunofluorescence microscopy. No staining was observed without the primary antibody, and only vague staining was observed for untransfected cells at the antibody concentrations used.

To examine the subcellular localization of GFAP, we transfected the mammalian N2A neuroblastoma cell line with an expression vector encoding full-length GFAP. The GFAP protein was detected by pAb-GFAP and a fluorescence-labeled anti-rabbit antibody. In the transfected N2A cells GFAP was incorporated into a structural network further supporting that GFAP is an intermediate filamentous protein (Fig. 4f).

GFAP and GFAP tails strongly diverge as seen in Fig. 2. The GFAP tail has been shown previously to be involved in correct filament incorporation (27). To compare GFAP and GFAP incorporation into filaments, we fused the cDNAs to N- and C-terminal GFP tags. N2A cells were transfected with the GFP constructs, and the subcellular localization was examined by fluorescence microscopy. Unfused GFP was dispersed throughout the entire cell (Fig. 5f). GFAP fusion proteins carrying either N- or C-terminal GFP tags were localized to a filamentous network (Fig. 5, a and b). Also N- or C-terminal GFP-tagged GFAP was located in filamentous networks in the transfected N2A cells (Fig. 5, c and d). The fluorescence pattern of the neurofilament protein NFL, GFP-tagged at its N terminus, is shown in Fig. 5e as a cytoskeletal reference.

View larger version (98K): [in this window] [in a new window]   Fig. 5.   GFAP is assembled into filaments. N2A cells were transfected with GFP-tagged GFAP proteins, neurofilament protein NFL, or unfused GFP. The transfected cells were visualized by epi-fluorescence microscopy. a, GFAP tagged in the N-terminal by GFP; b, GFAP tagged in the C-terminal by GFP; c, GFAP tagged in the N-terminal by GFP; d, GFAP tagged in the C-terminal by GFP; e, neurofilament protein NFL tagged in the N-terminal by GFP; and f, unfused GFP.

To examine subcellular co-localization of GFAP and GFAP, a double labeling approach was utilized where N-terminal GFP-tagged GFAP and untagged GFAP were transfected into N2A cells. GFAP was specifically labeled by the pAb-GFAP antibody and stained with a red fluorescent-labeled secondary antibody (Fig. 6). By merging the GFP and red fluorescent-stained cells, a perfect overlap in localization was observed indicating that the two GFAP isoforms were polymerized into the same filamentous structures.

View larger version (37K): [in this window] [in a new window]   Fig. 6.   GFAP and GFAP co-localize and GFAP partly co-localizes with PS-1. N2A cells were transfected with the indicated expression vectors, and transfected cells were visualized by epi-fluorescence microscopy. a and b, cells were transfected by GFAP tagged in the N terminus by GFP and stained with pAb-GFAP and a red fluorescent secondary antibody. c and d, cells were transfected by untagged GFAP and stained with pAb-GFAP and a red fluorescent secondary antibody. e-g, GFAP and GFAP co-localize. N2A cells were co-transfected with GFAP GFP tagged in the N terminus and untagged GFAP. Cells were stained with pAb-GFAP and a red fluorescent secondary antibody. h and i, cells were transfected by GFAP tagged in the N terminus by GFP and stained with a FLAG antibody and a red fluorescent secondary antibody. j and k, cells were transfected by FLAG-PS-1 and stained with FLAG antibody and a red fluorescent secondary antibody. l-n, GFAP and PS-1 partly co-localize. N2A cells were co-transfected with GFAP GFP-tagged in the N terminus and FLAG epitope-tagged PS-1. Cells were stained with FLAG antibody and the red fluorescent secondary antibody. a, c, e, h, j, and l, images obtained in the green wavelength; b, d, f, i, k, and m, images obtained in the red wavelength; and g and n, merged images from e and f and l and m, respectively.

Furthermore, we examined the co-localization between GFAP and PS-1. For this N-terminal GFP-tagged GFAP and N-terminal FLAG epitope-tagged PS-1 were transfected into N2A cells. The PS-1 localization was determined by an anti-FLAG antibody and a red fluorescent-labeled secondary antibody. As already extensively documented, PS-1 localized to the perinuclear region and cytoplasmic granules. Consequently, we observed that a subpopulation of the GFAP pool co-localized with PS-1 (Fig. 6).

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Glial fibrillary acidic protein, GFAP (isoform GFAP), is one of the main intermediary filament proteins of the astrocytic cytoskeleton. Here we describe a novel isoform of human GFAP, designated GFAP. The corresponding GFAP mRNA is the result of alternative splicing where exon 8 and exon 9 of the GFAP transcript are substituted by a new exon, designated exon 7a, located in intron 7 of the GFAP gene. Exon 7a is flanked by splice consensus sequences in the 5'-end and a polyadenylation signal in the 3'-end. In addition we identified the corresponding mouse GFAP transcript that is expressed predominantly in the brain (Fig. 2). GFAP mRNA was determined to be about 20-fold less expressed than GFAP mRNA (Fig. 2), a ratio in accordance with the relative abundance of EST clones of each isoform. The GFAP mRNA-derived human protein, GFAP, is 431 amino acids long, one amino acid shorter than GFAP, and has a deduced molecular mass of 55 kDa. The GFAP protein was identified in central nervous system derived cell lines and in a porcine brain extract and had Triton X-100 extraction characteristics as expected of a cytoskeletal protein (Fig. 4). The head and rod regions of GFAP are identical to the head and rod regions of GFAP. But the tail region of GFAP encoded by exon 7a is completely different from the tail region of GFAP encoded by exons 8 and 9. This difference in the tail regions suggested different functions of GFAP and GFAP, and it might be significant that the tail region of GFAP is almost 100% conserved, whereas the tail in GFAP allows for 25% divergence between the human and mouse sequences (Fig. 2f).

It has been suggested recently (21-23) that PS-1, in particular the 30-kDa N-terminal fragment of the protease processed PS-1, interacts with cytoskeletal proteins. High expression of both PS-1 and GFAP is observed in astrocytes associated with cerebral infarction and astrocytoma (24) and in reactive astrocytes surrounding the senile plaques of Alzheimer's disease (25). The GFAP transcript was identified by screening a human fetal brain cDNA library for translation products capable of binding PS-1 (Fig. 1). The tail of GFAP is indispensable for the PS-1 binding in yeast and in vitro and cannot be replaced by the exon 8- and exon 9-encoded tail of GFAP (Fig. 3). Moreover, sequences overlapping coiled-coil region 2 common to GFAP and GFAP are required for PS-1 binding.

IF proteins form polymers where the rod domain promotes formation of a coiled-coil dimer between two parallel IF proteins. The dimers associate in an antiparallel manner to form a nonpolarized tetrameric substructure where coiled-coil region 1 from one dimer is associated with coiled-coil 2 from the other dimer. The tetrameric structure appears to be the fundamental subunit from which the IF is assembled. The head domain seems to govern both end-to-end and lateral associations, whereas the tail domain may project from the surface of the filament and mediate interaction with other cellular components (1, 2). The variable tail domain may thus confer cell-specific property of the IF proteins (26). Accordingly, PS-1 and PS-2 interact with the GFAP tail and coiled-coil region 2. Such an interaction would leave coiled-coil region 1 free to interact with coiled-coil region 2 of another anti-parallel dimer during fiber formation and link a presenilin-containing membrane to the cytoskeleton. Note that a fraction of PS-1 is associated with the -cadherin-catenin complex, which serves as a cytoskeletal attachment site on the plasma membrane and has a function in cell-cell communication (28, 29). Similarly, the observed co-localization between a subpopulation of GFAP and PS-1 might be in agreement with a function of the GFAP and PS-1 interaction as a linkage between different structures within the cell.

The RDG motif present in the tail of GFAP has been implicated in correct filament formation based on C-terminal deletion studies and is evolutionarily conserved in type III IF proteins (27). The RDG motif is absent from the tail region of GFAP. However, in transfected cells GFAP can assemble into filaments (Figs. 4 and 5). This filament formation was indistinguishable from the filament formation with GFP-tagged GFAP. Accordingly, co-localization was observed between GFAP and GFAP in transfected N2A cells (Fig. 6). The observed filamentous incorporation could be due to heteromeric assembly with endogenous intermediate filament proteins. It should be noted that we have been unable to monitor direct interactions between the neurofilament proteins and GFAP at least in the yeast two-hybrid system, whereas GFAP and GFAP interact with each other (data not shown).

Amino acids 1-65 of the PS-1-(1-85) fragment can be removed without any effect on the binding capacity of GFAP, whereas amino acids 83-85 are essential for the binding capacity. Also valine at position 82 seems critical because nonconservative substitution abolished binding to GFAP. We note that the sequence motif in PS-1 required for GFAP interaction is highly conserved between presenilin proteins throughout evolution. Because of the PS-1 association with familial Alzheimer's disease, we also tested the pathogenic conservative substitutions valine to leucine at position 82 and alanine for valine at position 79 and found no effect on the GFAP binding. This was not unexpected because the pathogenic effect of these two mutations is not different from that of the other known disease-causing mutations scattered outside the PS-1-(1-85) region.

Mutation in the GFAP gene has recently been associated with Alexander disease, in which GFAP-containing inclusion bodies in astrocytes is a pathological hallmark (13). The identified mutations are located in the tail and rod domains of GFAP and the latter therefore expressed in GFAP. Our observations suggest that the isoform GFAP is relevant to consider in studies of neurodegenerative diseases.

	ACKNOWLEDGEMENT

We thank Dr. P. Fraser, Toronto, Canada, for PS-1 and PS-2 cDNAs.

	FOOTNOTES

^* This work was supported by the Danish Medical Research Council Ældreforskning II, Grant 9502112, the Danish Natural Sciences Research Council Grant 9901846, and the Novo Nordisk Foundation.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The nucleotide sequence(s) reported in this paper has been submitted to the GenBank^TM/EBI Data Bank with accession number(s) AJ306447.

^¶ To whom correspondence should be addressed: Dept. of Human Genetics, Bartholin Bldg., University of Aarhus, DK-8000 Aarhus C, Denmark. Tel.: 45 89421678; Fax: 45 86123173; E-mail: aln@mbio.aau.dk.

Published, JBC Papers in Press, June 10, 2002, DOI 10.1074/jbc.M112121200

	ABBREVIATIONS

The abbreviations used are: GFAP, glial fibrillary acidic protein; IF, intermediate filament; PS, presenilin; DBD, DNA binding domain; AAD, acidic activation domain; GST, glutathione S-transferase; His, His₆ tag; pAb, polyclonal antibody; pA, polyadenylation signal; GFP, green fluorescent protein; PBS, phosphate-buffered saline; RT, reverse transcriptase; X-gal, 5-bromo-4-chloro-3-indolyl--D-galactopyranoside; TRITC, tetramethylrhodamine isothiocyanate; TIE, Triton X-100-insoluble extract.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This article has been cited by other articles:

				J. Blechingberg, S. Lykke-Andersen, T. H. Jensen, A. L. Jorgensen, and A. L. Nielsen Regulatory mechanisms for 3'-end alternative splicing and polyadenylation of the Glial Fibrillary Acidic Protein, GFAP, transcript Nucleic Acids Res., December 3, 2007; 35(22): 7636 - 7650. [Abstract] [Full Text] [PDF]

				A. L. Nielsen and A. L. Jorgensen Self-assembly of the Cytoskeletal Glial Fibrillary Acidic Protein Is Inhibited by an Isoform-specific C Terminus J. Biol. Chem., October 1, 2004; 279(40): 41537 - 41545. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) All Versions of this Article: 277/33/29983    most recent M112121200v1