Advertisement

Ankyrin

A NEW ANKYRIN GENE WITH NEURAL-SPECIFIC ISOFORMS LOCALIZED AT THE AXONAL INITIAL SEGMENT AND NODE OF RANVIER (∗)
  • Ekaterini Kordeli
    Footnotes
    Affiliations
    Howard Hughes Medical Institute and Departments of Cell Biology and Biochemistry, Duke University Medical Center, Durham, North Carolina 27710
    Search for articles by this author
  • Stephen Lambert
    Correspondence
    To whom correspondence and reprint requests should be addressed. Tel.: 919-684-3105; Fax: 919-684-3590
    Footnotes
    Affiliations
    Howard Hughes Medical Institute and Departments of Cell Biology and Biochemistry, Duke University Medical Center, Durham, North Carolina 27710
    Search for articles by this author
  • Vann Bennett
    Affiliations
    Howard Hughes Medical Institute and Departments of Cell Biology and Biochemistry, Duke University Medical Center, Durham, North Carolina 27710
    Search for articles by this author
  • Author Footnotes
    ∗ This research is funded in part by National Institutes of Health Grant 537DK29808. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore by 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™/EMBL Data Bank with accession number(s) U13616.
    § Both authors contributed equally to this work.
    Supported by a postdoctoral fellowship from the National Multiple Sclerosis Society. Present address: Department of Supramolecular and Cellular Biology, Institute Jacques Monod, University of Paris, Paris, France. Tel.: 33-1-4427-4223.
      We have characterized a new ankyrin gene, expressed in brain and other tissues, that is subject to extensive tissue-specific alternative mRNA processing. The full-length polypeptide has a molecular mass of 480 kDa and includes a predicted globular head domain, with membrane- and spectrin-binding activities, as well as an extended “tail” domain. We term this gene ankyrinG based on its giant size and general expression. Two brain-specific isoforms of 480 kDa and 270 kDa were identified that contain a unique stretch of sequence highly enriched in serine and threonine residues immediately following the globular head domain. Antibodies against the serine-rich domain and spectrin-binding domain revealed labeling of nodes of Ranvier and axonal initial segments. Ankyrin-binding proteins also known to be localized in these specialized membrane domains include the voltage-dependent sodium channel, the sodium/potassium ATPase, sodium/calcium exchanger, and members of the neurofascin/L1 family of cell adhesion molecules. The neural-specific ankyrinG polypeptides are candidates to participate in maintenance/targeting of ion channels and cell adhesion molecules to nodes of Ranvier and axonal initial segments.

      INTRODUCTION

      Ankyrins are peripheral membrane proteins believed to interconnect integral proteins with the spectrin-based membrane skeleton (reviewed in (
      • Lambert S.
      • Bennett V.
      ) and (
      • Bennett V.
      ). Two ankyrin genes have been characterized so far: ankyrinR and ankyrinB. AnkyrinR was originally characterized as a component of the erythrocyte membrane skeleton. In the rat brain, ankyrinR is localized to the plasma membrane of a subpopulation of postmitotic neurons(
      • Lambert S.
      • Bennett V.
      ). The second gene, ankyrinB, exists as two developmentally regulated alternatively spliced variants with molecular sizes of 220 kDa and 440 kDa. The 440-kDa isoform contains a predicted extended “tail” domain in addition to the membrane and spectrin-binding domains and is targeted to the plasma membranes of unmyelinated and premyelinated axons(
      • Otto E.
      • Kunimoto M.
      • McLaughlin T.
      • Bennett V.
      ,
      • Kunimoto M.
      • Otto E.
      • Bennett V.
      ,
      • Chan W.
      • Kordeli E.
      • Bennett V.
      ).
      Ankyrin has been shown to associate with the voltagedependent sodium channel in vitro and to co-localize with this molecule at nodes of Ranvier, axonal initial segments, and the neuromuscular junction(
      • Srinivasan Y.
      • Elmer L.
      • Davis J.
      • Bennett V.
      • Angelides K.
      ,
      • Kordeli E.
      • Davis J.
      • Trapp B.
      • Bennett V.
      ,
      • Kordeli E.
      • Bennett V.
      ,
      • Flucher B.E.
      • Daniels M.P.
      ). It is generally believed that the maintenance of highly localized concentrations of the voltage-dependent sodium channel at the axonal initial segments and nodes of Ranvier is important to the initiation and propagation of the saltatory action potential. An ankyrin isoform at the node of Ranvier was initially identified by immunofluorescence using an antibody raised against erythrocyte ankyrinR which showed cross-reactivity with other members of the ankyrin gene family(
      • Kordeli E.
      • Davis J.
      • Trapp B.
      • Bennett V.
      ). As the nodal isoform of ankyrin was still present in ankyrinR-deficient mice (carrying the nb mutation) and was not recognized by antibodies specific for ankyrinB or ankyrinR, it was concluded that the ankyrin isoform present at the node of Ranvier was the product of an unidentified ankyrin gene(
      • Kordeli E.
      • Bennett V.
      ).
      In this study, we describe the cDNA sequence of a third ankyrin gene with alternatively spliced isoforms expressed in brain as well as a variety of other tissues. The two largest protein isoforms, which contain an unusual serine-rich sequence, are expressed only in nervous tissue. Specific antibodies raised against this serine-rich sequence stain axonal initial segments and nodes of Ranvier in cryosections from the rat brain. The two novel ankyrin isoforms represent the first specialized cytoplasmic protein components of these physiologically important membrane domains.

      EXPERIMENTAL PROCEDURES

       Isolation and Characterization of the AnkyrinG 480-kDa cDNA

      A human frontal cortex ZapII library (Stratagene) was screened with an antibody raised against ankyrinR as described(
      • Sambrook J.
      • Fritsch E.F.
      • Maniatis T.
      ). Positive cDNAs were identified by peroxidase staining combined with avidin-biotin enhancement (Vector Laboratories). Six positive cDNAs were subcloned using in vivo excision according to the manufacturer's protocol, and five of these cDNAs showed identity with areas of the ankyrinB sequence(
      • Otto E.
      • Kunimoto M.
      • McLaughlin T.
      • Bennett V.
      ). One 3-kb
      The abbreviations used are: kb
      kilobase(s)
      bp
      base pair(s).
      cDNA, however, consisted of a unique nucleotide sequence encoding an area of the ankyrinG membrane-binding domain. This cDNA was labeled with [P]dCTP using the primer extension method (Multiprime System, Amersham) and used to screen a human fetal brain stem gt11 library (a gift from Dr. J. Keene, Dept. of Microbiology, Duke University). Positive cDNAs were subcloned into pBluescript (Stratagene) for sequencing. Multiple rescreening of this library was required to obtain the complete ankyrinG cDNA sequence. Rat ankyrinG sequence was obtained by screening a rat brain library (Clontech) with human ankyrinG cDNAs. Positive cDNAs were confirmed by sequencing and comparison with the human ankyrinG sequence.

       Northern Blot Analysis

      Total RNA was derived from the indicated rat tissues using RNazol (Cinna/Bioctex) and poly(A) RNA prepared as described(
      • Sambrook J.
      • Fritsch E.F.
      • Maniatis T.
      ). 20 μg of poly(A) RNA was fractionated in 1% formaldehyde/agarose gels and transferred to nitrocellulose filters(
      • Sambrook J.
      • Fritsch E.F.
      • Maniatis T.
      ). Filters were hybridized with P-labeled rat ankyrinG cDNAs and washed at 65°C with 0.1 × SSC, 0.1% SDS, before autoradiography.

       Antibody Preparation

      Antibodies against the rat ankyrinG spectrin-binding domain were raised against four peptides from the rat ankyrinG sequence (corresponding to amino acid residues 935-945, 945-969, 1092-1101, and 1224-1240 of the human ankyrinG sequence, Fig. 1). These peptides were coupled to rabbit serum albumin and injected into rabbits to raise antiserum. The peptides represented sequences of the ankyrinG spectrin-binding domain not found in the other ankyrin genes. To affinity-purify specific antibodies from the antiserum, rat ankyrinG cDNA b (nucleotides 2913-5094) from the ankyrinG spectrin-binding domain (Fig. 2A) was subcloned into pGEMEX (Promega Biotech) and expressed as a fusion protein with the viral gene 10 protein. The recombinant protein was purified and immobilized on Sepharose CL-6B (Pharmacia). Antibodies to rat ankyrinG spectrin-binding domain peptides were affinity-purified against the immobilized recombinant protein.
      Figure thumbnail gr1a
      Figure 1:cDNA sequence of ankyrinG 480 kDa. The putative polyadenylation sequence is boxed. The derived amino acid sequence is also shown, using the single letter code, numbered in italics. The ANK repeats of the membrane-binding domain are shown in bold and underlined. The serine-rich domain is also shown in bold and boxed. The peptide sequences used in the production of antibodies to the spectrin-binding domain are shown in bold italics.
      Figure thumbnail gr1b
      Figure 1:cDNA sequence of ankyrinG 480 kDa. The putative polyadenylation sequence is boxed. The derived amino acid sequence is also shown, using the single letter code, numbered in italics. The ANK repeats of the membrane-binding domain are shown in bold and underlined. The serine-rich domain is also shown in bold and boxed. The peptide sequences used in the production of antibodies to the spectrin-binding domain are shown in bold italics.
      Figure thumbnail gr1c
      Figure 1:cDNA sequence of ankyrinG 480 kDa. The putative polyadenylation sequence is boxed. The derived amino acid sequence is also shown, using the single letter code, numbered in italics. The ANK repeats of the membrane-binding domain are shown in bold and underlined. The serine-rich domain is also shown in bold and boxed. The peptide sequences used in the production of antibodies to the spectrin-binding domain are shown in bold italics.
      Figure thumbnail gr2
      Figure 2:Comparison of ankyrinG 480 kDa with ankyrinB 440 kDa and predicted structure. A, dot matrix alignment of the derived amino acid sequence of ankyrinG 480 kDa against that of ankyrinB 440 kDa. Each dot represents a minimum identity of 60% over a window of 8 residues. Also shown is the relative position of three rat cDNAs (a, b, and c). B, hydrophilicity profile and predicted model for the ankyrinG 480 kDa structure.
      To raise antibodies against the serine-rich domain (Fig. 2A), rat ankyrinG cDNA c (nucleotides 5028-6042) was subcloned into pGEMEX, and the recombinant fusion protein was purified and injected into rabbits. The resulting antiserum against the rat ankyrinG serine-rich domain was affinity-purified using immobilized recombinant protein after initially depleting the serum of gene 10 antibodies using recombinant gene 10 protein coupled to Sepharose CL-6B.
      Immunoblot analysis of crude membrane samples was carried out as described previously(
      • Lambert S.
      • Bennett V.
      ), with bound antibodies detected using I-labeled protein A and autoradiography.

       Immunocytochemistry

      Adult rats were anesthetized and perfused with 150 mM NaCl, 10 mM sodium phosphate, pH 7.5, and 50 units/ml heparin, followed by 2% paraformaldehyde in the same buffer. Nervous tissue was removed and fixed for a further 2 h, before cryoprotection and freezing(
      • Lambert S.
      • Bennett V.
      ). 4-μm cryosections were cut and mounted on Vectabond-treated glass slides before incubation with primary antibodies at 5 μg/ml, overnight at 4°C. Sections were extensively washed and visualized by indirect immunofluorescence(
      • Kordeli E.
      • Davis J.
      • Trapp B.
      • Bennett V.
      ).

      RESULTS

       Isolation and Characterization of a New Ankyrin cDNA (AnkyrinG) from the Human Brain

      To identify the ankyrin isoform present at the node of Ranvier, a human frontal cortex cDNA library was screened with an antibody raised against erythrocyte ankyrinR, that had previously been used in immunofluorescence studies of the node of Ranvier(
      • Kordeli E.
      • Davis J.
      • Trapp B.
      • Bennett V.
      ). The frontal cortex contains the cell bodies of neurons that give rise to myelinated axons, while showing limited expression of ankyrinR(
      • Lambert S.
      • Bennett V.
      ). Screening the library yielded a 3-kb cDNA encoding a unique ankyrin sequence, which was then used as a probe to isolate overlapping cDNAs from a human fetal brain stem library. This cloning strategy eventually culminated in a 14,783-bp contiguous cDNA sequence, that terminated in a poly(A) sequence and had a polyadenylation signal (AATAAA) 20 bp upstream of the poly(A)+ sequence (Genbank accession number U13616) The cDNA sequence contained a single open reading frame encoding a protein of 4377 amino acids with a predicted mass of 480 kDa. The cDNA and derived amino acid sequence is shown in Fig. 1. We have termed this protein ankyrinG 480 kDa, the G reflecting the giant size of this gene and its generalized expression in multiple tissues. As with other members of the ankyrin gene family, ankyrinG contained 24 copies of a 33-amino acid motif, the ANK repeat.
      AnkyrinG is more closely related to ankyrinB than to ankyrinR with an overall homology of 71% amino acid identity when compared with ankyrinB and 57% amino acid identity when compared with ankyrinR. Fig. 2A shows a dot matrix alignment of the derived amino acid sequence of ankyrin480 kDa against that of ankyrin440 kDa. As with all members of the ankyrin gene family, the two proteins show extensive homology in their membrane- (74%amino acid identity) and spectrin-binding (67%amino acid identity) domains and only small areas of homology in their carboxyl-terminal domains.
      Unlike ankyrinR, ankyrinB and ankyrinG have a predicted “tail” domain inserted between their spectrin and carboxyl-terminal domains. Biophysical studies of recombinant polypeptides derived from ankyrinB 440 kDa suggest that much of this domain is largely unstructured and has the configuration of an extended random coil. Comparisons of this domain with other random coil polypeptides of similar sizes such as MAP2 suggest that this domain may be approximately 200 nm in length (
      • Chan W.
      • Kordeli E.
      • Bennett V.
      ). Although the “tail” domain for ankyrinG shows only 20% amino acid identity with that of ankyrinB, the sequence of this domain also shows features consistent with an extended structure and has a hydrophilicity profile (Fig. 2B) and amino acid composition similar to that of the ankyrinB 440-kDa tail domain(
      • Chan W.
      • Kordeli E.
      • Bennett V.
      ). As shown in the dot matrix alignment (Fig. 2A), the ankyrinG tail domain contains multiple small stretches of homology with the ankyrinB 440-kDa tail domain. These homologies and the strong identity between the membrane- and spectrin-binding domains of these molecules suggest the evolution of ankyrinG or ankyrinB as the result of a gene duplication event.
      The ankyrinB 440-kDa tail domain contains a large number of predicted sites for phosphorylation, particularly by the enzymes casein kinase 2 and protein kinase C(
      • Chan W.
      • Kordeli E.
      • Bennett V.
      ). Although the tail domain of ankyrinG also contains a large number of sites for casein kinase 2, it is missing many of the potential sites for protein kinase C, which are located in a group of fifteen 12-amino acid repeats present at the amino-terminal end of the ankyrinB 440-kDa tail domain. These repeats are not found in ankyrinG.
      The 480-kDa ankyrinG also contains a novel domain of approximately 40 kDa (Fig. 1, residues 1478-1908), which causes a displacement in the dot matrix alignment (Fig. 2A). This domain is serine-rich (35% serine and threonine residues) and extremely conserved between rat and human (87% identity). A data base search indicates that the serine-rich domain shares sequence homology with a number of glycosylated proteins such as agglutinin and mucin.
      Features of the ankyrinG 480-kDa sequence are summarized in the model shown in Fig. 2B. This model predicts a globular head domain consisting of the membrane- and spectrin-binding domains and a long extended unstructured tail domain. This model is based on previous biophysical studies of the membrane- and spectrin-binding domains of ankyrinR(
      • Hall T.G.
      • Bennett V.
      ) and on similar studies of the tail region of ankyrinB 440 kDa(
      • Chan W.
      • Kordeli E.
      • Bennett V.
      ), as mentioned earlier.
      As all further studies were carried out using rat tissues, three cDNAs representing 4.2 kb of the rat ankyrinG sequence were obtained from a rat brain cDNA library using human cDNAs as probes. The three rat cDNAs covered nucleotides 789-1881 (a), 2913-5094 (b), and 5028-6042 (c) of the human ankyrinG sequence. Homologies between rat and human ankyrinG over the 4.2 kb were typically of the order of 70% identity at the nucleic acid level. The relative positions of the three rat cDNAs in the ankyrinG molecule are underlined in Fig. 2A.

       Tissue-specific Expression of AnkyrinG Isoforms

      High stringency Northern blot analysis of ankyrinG expression with probes against regions coding for the membrane-binding domain, spectrin-binding domain, and serine-rich domain in four different rat tissues reveals five different transcripts (Fig. 3B). The pattern of expression suggests tissue-specific alternative mRNA processing. Two of the transcripts (15 kb and 10 kb) are brain-specific, react with probes against membrane- and spectrin-binding domains, and are the only transcripts recognized by a probe from the serine-rich domain (Fig. 3C). The 14.8-kb contiguous cDNA sequence shown in Fig. 1 is probably represented by the 15-kb transcript. Given the similarities between the ankyrinG and ankyrinB genes, it is likely that the serine-rich and tail domains of ankyrinG are removed by alternative mRNA processing similar to the removal of the tail domain in ankyrinB(
      • Otto E.
      • Kunimoto M.
      • McLaughlin T.
      • Bennett V.
      ,
      • Chan W.
      • Kordeli E.
      • Bennett V.
      ). Removal of the serine-rich domain and tail sequence from the 15-kb message would result in a 7-kb ankyrinG transcript encoding a protein of 190 kDa. AnkyrinG transcripts of 7 kb are detected in the lung (lane 2) and kidney (lane 3). A novel ankyrin cDNA from the mouse kidney has been cloned by Peters and Colleagues (
      • Peters L.L.
      • Lu F.M.
      • John K.
      • Higgins A.
      • Yialamas M.
      • Turtzo L.
      • Otsuka A.
      • Lux S.E.
      ), which encodes a protein of 190 kDa. This cDNA represents an alternatively spliced isoform of the murine ankyrinG gene (referred to as ankyrin 3).
      Figure thumbnail gr3
      Figure 3:Northern blot analysis of ankyrinG transcripts. Poly(A) RNA (20 μg) from the rat testis (lane 1), lung (lane 2), kidney (lane 3), and brain (lane 4) was fractionated in 1% formaldehyde/agarose gels and transferred to nitrocellulose. Filters were hybridized with P-labeled rat ankyrinG cDNAs from the membrane-binding domain (A), spectrin-binding domain (B), and serine-rich domain (C).
      The cDNA probe from the spectrin-binding domain also hybridizes under stringent conditions with two smaller transcripts of 5.5 kb and 4.2 kb (Fig. 3B). As transcripts of this size were not observed in kidney mRNA, hybridized with probes from ankyrinB or ankyrinR(
      • Otto E.
      • Kunimoto M.
      • McLaughlin T.
      • Bennett V.
      ), we conclude that these smaller transcripts are genuine products of the ankyrinG gene produced by alternative mRNA processing. In contrast, the probe encompassing ANK repeats 5-15 does not hybridize to the 5.5-kb transcript and hybridizes only weakly to the 4.2-kb transcript (Fig. 3A). This suggests that the smaller isoforms may be missing some or all of their membrane-binding domain. Multiple transcripts of ankyrinG (ankyrin 3) have also been noted by Peters et al. (13).
      To further elucidate the expression of ankyrinG isoforms, antibodies were raised against the ankyrinG spectrin-binding (@SpBd) and serine-rich (@SRd) domains. The use of these antibodies in immunoblot analysis of multiple rat tissues is shown in Fig. 4. In crude membrane fractions from the rat brain (Fig. 4A, lane 1), antibody against the serine-rich domain recognizes two polypeptides of 480 kDa and 270 kDa (Fig. 4A, lane 1), which presumably are encoded by the 15-kb and 10-kb transcripts. These polypeptides are not seen in the kidney (lane 2), lung (lane 3), testes (lane 4), spleen (lane 5), liver (lane 6), or heart (lane 7). Antibody against the serine-rich domain also recognizes a doublet of bands of approximately 90 kDa in size, present in all tissues studied (data not shown). These bands are not recognized by the spectrin-binding domain antibody, however, and immunofluorescence with antibody against the serine-rich domain in rat tissues other than the brain shows no discernible staining.
      Figure thumbnail gr4
      Figure 4:Immunoblot analysis of ankyrinG expression in adult rat tissues. Tissue homogenates and crude membrane samples were prepared for SDS-polyacrylamide gel electrophoresis from the rat brain (lane 1), kidney (lane 2), lung (lane 3), testes (lane 4), spleen (lane 5), liver (lane 6), and heart (lane 7). Equivalent loadings of membrane proteins were subject to immunoblot analysis using antibodies raised against the serine-rich (@SRd) (A) and spectrin-binding (@SpBd) (B) domains. Lane 1′ shows a 5× longer exposure of lane 1 blotted with @SpBd.
      In contrast to immunoblot analysis with antibody against the serine-rich domain, analysis of the same tissue samples with antibody against the spectrin-binding domain identified a number of different polypeptides (Fig. 4B). A range of major polypeptides from 190 kDa to 72 kDa are seen in multiple tissues, supporting the results from Northern blot analysis that suggest tissue-specific alternative mRNA processing. Fig. 4B, lane 1′, shows that the antibody also recognizes the 480-kDa and 270-kDa ankyrinG proteins in the brain upon longer exposure of the autoradiograph. These large proteins were not seen in any of the other tissues (lanes 2-7) when using longer autoradiograph exposures, nor were any previously undetected immunoreactive peptides observed with these exposure times.

       Localization of Neural-specific AnkyrinG Isoforms to the Axonal Initial Segment and Node of Ranvier

      Affinity-purified antibodies against the serine-rich and spectrin-binding domains were used to localize ankyrinG isoforms in the nervous system by immunofluorescence. Fig. 5, A and C, shows the localization of ankyrinG isoforms containing the serine-rich domain to the node of Ranvier in both peripheral (Fig. 5A) and central (Fig. 5C) nerves. Although this staining appears to be confined to the node of Ranvier, lower concentrations of ankyrinG may also underlie the internodal axolemma. Fig. 5, B and D, confirms localization of ankyrinG isoforms to the node of Ranvier using an antibody against the spectrin-binding domain. This antibody also stains bundles of unmyelinated axons in the rat sciatic nerve, presumably due to the presence of other ankyrinG isoforms lacking the serine-rich domain. Such staining was observed previously in the sciatic nerve with the erythrocyte ankyrinR antibody(
      • Kordeli E.
      • Davis J.
      • Trapp B.
      • Bennett V.
      ). Antibody against the serine-rich domain also stains ankyrinG isoforms at axonal initial segments, in the hippocampus (Fig. 6, A and C), the cerebellum (Fig. 6B), and the cerebral cortex (Fig. 6D). Staining is seen on the plasma membrane of these structures (Fig. 6C) and is restricted to the initial segment of the axon (Fig. 6E). AnkyrinG is also seen on small axonal-like structures of the granular cell layer (Fig. 6B) that may represent the initial segments from unmyelinated axons of granule cells. Subaxolemmal densities underlying the initial segment have been observed in unmyelinated as well as myelinated axons(
      • Peters A.
      • Palay S.L.
      • Webster H.de F.
      ).
      Figure thumbnail gr5
      Figure 5:Immunofluorescence localization of ankyrinG isoforms at the node of Ranvier. 4-μm cryosections of the rat sciatic nerve (A, B, and D) or spinal cord white matter (C) were stained with either @SRd (A and C) or @SpBd (B and D). A′ and D′ are DIC micrographs of the corresponding immunofluorescence. Arrows indicate the nodes of Ranvier. Arrowheads (B) indicate bundles of unmyelinated axons in the sciatic nerve. Bars represent 10 μm.
      Figure thumbnail gr6
      Figure 6:Immunofluorescence localization of ankyrinG isoforms at the axonal initial segment. 4-μm cryosections from the CA3 region of the hippocampus (A), cerebellum (B), and cortex (D) were stained with @SRd. Arrows indicate staining of the initial segments of neurons. Large arrowheads indicate Ranvier nodes observed in the corpus callosum (cc) and white matter (wm) of the cerebellum. Small arrowheads (B) appear to represent staining of the unmyelinated axons of granule cells (gc). C represents the insert area of A in greater detail. E shows an individual Purkinje cell (pc) from another area of the cerebellum, in greater detail. The unstained axon is indicated by open arrowheads. Bars represent 10 μm.

      DISCUSSION

      This study presents the complete cDNA sequence of a new ankyrin gene (ankyrinG), which has 480-kDa and 270-kDa neural-specific isoforms localized at axonal initial segments and nodes of Ranvier. An unusual feature of the neural-specific ankyrins is a 40-kDa serine-rich domain with sequence similarity to mucins and glycoproteins. 480-kDa ankyrinG is closely related to 440-kDa ankyrinB, an isoform of the major ankyrin gene in the brain. Both molecules possess predicted extended tail domains subject to alternative splicing(
      • Chan W.
      • Kordeli E.
      • Bennett V.
      ). Currently, we are not able to distinguish whether the two 480-kDa and 270-kDa isoforms detected by antibodies against the serine-rich domain are both present at the node and/or axonal initial segments. However, both polypeptides are expected to contain membrane-binding and spectrin-binding domains based on Northern and immunoblot analysis. Ankyrin-binding proteins also known to be localized in these specialized membrane domains include the voltage-dependent sodium channel(
      • Srinivasan Y.
      • Elmer L.
      • Davis J.
      • Bennett V.
      • Angelides K.
      ,
      • Waxman S.G.
      • Ritchie J.M.
      ), the Na/K-ATPase(
      • Waxman S.G.
      • Ritchie J.M.
      ,
      • Nelson W.J.
      • Veshnock P.J.
      ), Na/Ca exchanger(
      • Waxman S.G.
      • Ritchie J.M.
      ,
      • Li Z.
      • Burke E.P.
      • Frank J.S.
      • Bennett V.
      • Philipson K.
      ), and members of the neurofascin/L1 family of cell adhesion molecules(
      • Davis J.Q.
      • McLaughlin T.
      • Bennett V.
      ). The neural-specific ankyrinG polypeptides are candidates to participate in the maintenance/targeting of ion channels and cell adhesion molecules to nodes of Ranvier and axonal initial segments. It is of interest that initial visualization of nodes of Ranvier by electron microscopy revealed a submembrane specialization(
      • Waxman S.G.
      • Ritchie J.M.
      ,
      • Peters A.
      ). Neural-specific forms of ankyrinG may represent a component of this structure(
      • Ichimura T.
      • Ellisman M.H.
      ).
      The role of ankyrinG and ankyrinB 440-kDa isoforms in myelination and the establishment of nodes of Ranvier remains to be determined. Unmyelinated axons contain 440-kDa ankyrinB, which is down-regulated as myelination takes place(
      • Chan W.
      • Kordeli E.
      • Bennett V.
      ). Development of membrane specializations at the node of Ranvier appear early in myelination(
      • Wiley-Livingston C.A.
      • Ellisman M.H.
      ), suggesting that a number of ankyrin isoforms may be present in the axolemma during myelination. Joe and Angelides(
      • Joe E.H.
      • Angelides K.
      ), using an antibody raised against erythrocyte ankyrin, have observed that clustering of the voltage-dependent sodium channel occurs independently of ankyrin. The development of isoform-specific antibodies will permit evaluation of the role of ankyrin in axonal differentiation and clustering of ion channels at the initial segment and node of Ranvier.
      Alternatively spliced isoforms of ankyrinG are expressed in a number of tissues (Figure 2:, Figure 3:)(
      • Peters L.L.
      • Lu F.M.
      • John K.
      • Higgins A.
      • Yialamas M.
      • Turtzo L.
      • Otsuka A.
      • Lux S.E.
      ). Some of these isoforms are missing areas of their membrane-binding domain based on Northern blot analysis, although they all contain a spectrin-binding domain (Fig. 3) (
      • Peters L.L.
      • Lu F.M.
      • John K.
      • Higgins A.
      • Yialamas M.
      • Turtzo L.
      • Otsuka A.
      • Lux S.E.
      ). A similar feature has been noted in mRNA transcripts of ankyrinB(
      • Kunimoto M.
      • Otto E.
      • Bennett V.
      ). Smaller isoforms missing the membrane-binding domain might be involved in functions different from that proposed for ankyrin as a membranecytoskeleton linker and may even localize to subcellular compartments other than the plasma membrane.
      Ankyrin has been implicated in the establishment of cellular polarity, particularly in cultured kidney epithelial cells(
      • Nelson W.J.
      ). The tissue-specific expression of ankyrinG isoforms suggests that this ankyrin gene is able to use a number of different combinations of functional domains, that may have a role to play in the establishment of specialized membrane domains as well as currently unanticipated functions.

      Acknowledgements

      We gratefully acknowledge the technical assistance of S. Scott Carpenter.

      REFERENCES

        • Lambert S.
        • Bennett V.
        Eur. J. Biochem. 1993; 211: 1-6
        • Bennett V.
        J. Biol. Chem. 1992; 267: 8703-8706
        • Lambert S.
        • Bennett V.
        J. Neurosci. 1993; 13: 3725-3735
        • Otto E.
        • Kunimoto M.
        • McLaughlin T.
        • Bennett V.
        J. Cell Biol. 1991; 114: 241-253
        • Kunimoto M.
        • Otto E.
        • Bennett V.
        J. Cell Biol. 1991; 115: 1319-1331
        • Chan W.
        • Kordeli E.
        • Bennett V.
        J. Cell Biol. 1993; 123: 1463-1473
        • Srinivasan Y.
        • Elmer L.
        • Davis J.
        • Bennett V.
        • Angelides K.
        Nature. 1988; 333: 177-180
        • Kordeli E.
        • Davis J.
        • Trapp B.
        • Bennett V.
        J. Cell Biol. 1990; 110: 1341-1352
        • Kordeli E.
        • Bennett V.
        J. Cell Biol. 1991; 114: 1243-1259
        • Flucher B.E.
        • Daniels M.P.
        Neuron. 1989; 3: 163-175
        • Sambrook J.
        • Fritsch E.F.
        • Maniatis T.
        Molecular Cloning: A Laboratory Manual. 2nd Ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY1989
        • Hall T.G.
        • Bennett V.
        J. Biol. Chem. 1987; 262: 10537-10545
        • Peters L.L.
        • Lu F.M.
        • John K.
        • Higgins A.
        • Yialamas M.
        • Turtzo L.
        • Otsuka A.
        • Lux S.E.
        Mol. Biol. Cell. 1993; 4: 56a
        • Peters A.
        • Palay S.L.
        • Webster H.de F.
        The Fine Structure of the Nervous System. Vol. 92. W. B. Saunders, Philadelphia1976
        • Waxman S.G.
        • Ritchie J.M.
        Ann. Neurol. 1993; 33: 121-136
        • Nelson W.J.
        • Veshnock P.J.
        Nature. 1987; 328: 533-536
        • Li Z.
        • Burke E.P.
        • Frank J.S.
        • Bennett V.
        • Philipson K.
        J. Biol. Chem. 1993; 268: 11489-11491
        • Davis J.Q.
        • McLaughlin T.
        • Bennett V.
        J. Cell Biol. 1993; 121: 121-133
        • Peters A.
        Q. J. Exp. Physiol. 1966; 51: 229-236
        • Ichimura T.
        • Ellisman M.H.
        J. Neurocytol. 1991; 20: 667-681
        • Wiley-Livingston C.A.
        • Ellisman M.H.
        Dev. Biol. 1980; 79: 334-355
        • Joe E.H.
        • Angelides K.
        Nature. 1992; 356: 333-335
        • Nelson W.J.
        Science. 1992; 258: 948-955