Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
2023 Journal of Cell Science 113, 2023-2034 (2000) Printed in Great Britain © The Company of Biologists Limited 2000 JCS0962 Identification of a novel C-terminal variant of βII spectrin: two isoforms of βII spectrin have distinct intracellular locations and activities Nandini V. L. Hayes1, Catherine Scott1, Egidius Heerkens2, Vasken Ohanian2, Alison M. Maggs3, Jennifer C. Pinder3, Ekaterini Kordeli4 and Anthony J. Baines1,* 1Department of Biosciences, University of Kent, Canterbury, Kent, CT2 7NJ, England 2Department of Medicine, University of Manchester, Oxford Road, Manchester, M13 9WL, England 3MRC Muscle and Cell Motility Unit, The Randall Institute, King’s College London, 26-29 Drury Lane, London 4Institut J. Monod, CNRS, Universite Paris 7, Tour 43, 2 place Jussieu, 75251 Paris cedex 05, France WC2B 5RL, England *Author for correspondence (e-mail: [email protected]) Accepted 20 March; published on WWW 10 May 2000 SUMMARY It is established that variations in the structure and activities of βI spectrin are mediated by differential mRNA splicing. The two βI spectrin splice forms so far identified have either long or short C-terminal regions. Are analogous mechanisms likely to mediate regulation of βII spectrins? Thus far, only a long form of βII spectrin is reported in the literature. Five human expressed sequence tags indicated the existence of a short splice variant of βII spectrin. The occurrence and DNA sequence of the short C-terminal variant was confirmed by analysis of human and rat cDNA. The novel variant lacks a pleckstrin homology domain, and has 28 C-terminal residues not present in the previously recognized longer form. Transcripts of the short C-terminal variant (7.5 and 7.0 kb) were most abundant in tissues originating from muscle and nervous system. Antibodies raised to a unique sequence of short C-terminal variant recognized 240 kDa polypeptides in cardiac and skeletal muscle and in nervous tissue; in cerebellum and forebrain, additional 270 kDa polypeptides were detected. In rat heart INTRODUCTION Spectrin is a cytoskeletal protein essential for the determination of cell shape, the resilience of membranes to mechanical stress, the positioning of particular transmembrane proteins within the plane of a membrane, and the organization of organelles and molecular traffic (reviewed in e.g. Bennett and Gilligan, 1993; De Matteis and Morrow, 1998; Holleran and Holzbaur, 1998). These functions require interactions with a wide variety of protein and lipid ligands, among which are actin and diverse integral and peripheral membrane proteins. It is becoming increasingly clear that functions of spectrin in different cell types or cellular compartments are mediated by isotypes of spectrin that are the products of several spectrin genes, or that vary structurally though differential mRNA splicing. Spectrin is tetramer of two α and two β chains (for a review of spectrin structure, see Viel and Branton, 1996). The α and and skeletal muscle, both long and short C-terminal forms of βII spectrin localized in the region of the Z line. The central region of the sarcomere, coincident with the M line, was selectively labeled with antibodies to the short Cterminal form. In cerebellum, the short form was not detectable in parallel fibers, structures in which the long form was readily detected. In cultured cerebellar granule neurons, the long form was dominant in neurites, with the short form being most abundant in cell bodies. In vitro, the short form was found to lack the binding activity for the axonal protein fodaxin, which characterizes the C-terminal region of the long form. Subcellular fractionation of brain revealed that the short form was scarcely detectable in postsynaptic density preparations, in which the long form was readily detected. We conclude that variation in the structure of the C-terminal regions of βII spectrin isoforms correlates with their differential intracellular targeting. Key words: Spectrin, Muscle, Heart, Neuron β chains lie loosely twined side by side to form anti-parallel dimers that self-associate to give the tetramer. Currently two mammalian genes encoding α spectrin polypeptides (α1 and αII) are known, as are four genes for β spectrins (sptb1 encoding βI spectrin, sptb2 encoding βII, sptbn2 encoding βIII and elf) (Hu et al., 1992; McMahon et al., 1987; Mishra et al., 1998; Ohara et al., 1998; Sahr et al., 1990; Stankewich et al., 1998; Winkelmann et al., 1988). The archetypal spectrin, the form found in red blood cells, is composed of αI and βI polypeptides. The β chains are of great interest since they contain most of the binding activities of the spectrins. The N-terminal region contains tandem calponin homology domains that bind F-actin (Carugo et al., 1997). Most of the length of the chain is made up of several approx. 106-residue repeats that form triple helices, the most C-terminal of which is a partial repeat that is a site for association with the α chain (see Viel and Branton, 2024 N. V. L. Hayes and others 1996). The C-terminal region after the last (partial) helical repeat is a source of structural variation between β spectrins. The archetypal red cell βI spectrin has a short C-terminal region of 52 amino acid residues. When βI gene products were molecularly cloned from muscle, a different C-terminal region was detected (Winkelmann et al., 1990a). This form was longer (about 230 residues). To distinguish the long and short forms, Winkelmann and Forget (1993) proposed that the forms should be given subtype (Σ) designations in the order of their discovery: thus the short form is βIΣ1; the long form is βIΣ2. The long C-terminal region contains within it a pleckstrin homology (PH) domain (Musacchio et al., 1993). Cloning of βII and βIII spectrins revealed similar C termini that included PH domains (Hu et al., 1992; Ohara et al., 1998; Stankewich et al., 1998). PH domains are present in a variety of proteins involved in signal transduction, as well as cytoskeletal proteins (Rebecchi and Scarlata, 1998). The interaction of the PH domain with inositol phospholipids seems to regulate some of spectrin’s membrane interactions (Godi et al., 1998). About 100 amino acid residues link the last (partial) helical repeat to the PH domain. This is the most non-conserved region of sequence between the β spectrins, and is a ligand binding site. These residues in the long C-terminal form of βII spectrin (residues 2087-2198) have a specific interaction with the neuronal protein fodaxin, a protein which does not interact with either the long or the short forms of βI spectrin (Hayes and Baines, 1994; Hayes et al., 1997). The differential mRNA splicing that gives rise to the long and short forms of βI spectrin occurs about half way through the region (Winkelmann et al., 1990a), so this region varies not just between products of different spectrin genes, but also between different of products of the same gene. By contrast to the activities of long C-terminal forms, the short C-terminal region of the erythrocyte βI spectrin has no documented ligand binding activities; however multiple residues unique to this short β chain are substrates for casein kinase (Harris and Lux, 1980; Pedroni et al., 1993). Phosphorylation of erythrocyte βI spectrin is associated with control of erythrocyte mechanical function (Manno et al., 1995). The long and short forms of βI spectrin therefore differ in terms of ligand binding and regulatory properties. In this paper, we consider whether βII spectrins, like βI, have a short, functionally distinct, variant? Taking advantage of the expressed sequence tag (EST) database, we have identified a sequence that we show is contiguous with the βII gene. This sequence encodes a variant that is structurally analogous to the short C-terminal region of human erythrocyte βI. While this work was in progress the sequences of two products of the fourth β spectrin gene, ELFs 1 and 3, were described (Mishra et al., 1998, 1999). The C-terminal regions of these proteins are extremely similar in both nucleotide and amino acid sequence to the ‘short’ variant of βII spectrin that we have identified. The common C-terminal portions of products of βII and elf genes seem to represent a novel structural domain characteristic of products of these genes. This structural domain is rich in potential phosphorylation sites, but is functionally distinct from the ‘long’ form of βII spectrin in that it lacks the PH domain, and also the site for interaction with the protein fodaxin (Hayes et al., 1997). Antibodies selective for long and short forms of βII spectrin reveal that while there is extensive overlap in the intracellular location of the two proteins in muscle cells and neurons, they are not always coincident. In particular skeletal muscle and heart M line regions contain the short form; the distal portions of cerebellar granule cell neurites are enriched in the long form. Only the long form is retained in preparations of post-synaptic densities. The respective C termini appear to correlate with differential compartmentalization of isoforms. We conclude that a recurring theme in the biology of most (if not all) mammalian β spectrins is likely to be functional variation mediated by variant C termini. MATERIALS AND METHODS Procedures and reagents Standard molecular biology methods were used (Sambrook et al., 1989) except where stated. Human skeletal muscle cDNA was obtained from Cambridge Biosciences, UK. Human genomic DNA was kindly given by Dr S. Eber (Georg-August-Universitat, Gottingen, Germany). Unless otherwise indicated all chemicals and reagents were obtained from Sigma (Poole, Dorset, UK). The antifodaxin monoclonal antibody, DR1, was used as described previously (Hayes and Baines, 1994). FITC-conjugated AffiniPure donkey antiguinea pig IgG (Jackson Immunoresearch Laboratories, Inc) and TRITC-conjugated goat anti-rabbit, TRITC-conjugated anti-mouse IgG and FITC-conjugated anti-rabbit IgG (Vector Laboratories, Peterborough, UK) were used as secondary antibodies for immunofluorescence. Peroxidase-conjugated secondary antibodies for immunoblotting were from Dako (Cambridge, UK). Database searches and sequence alignments Database searches were made using the NCBI BLAST server available over the World Wide Web at http://www.ncbi.nlm.nih.gov/BLAST/. Potential splice sites in genomic sequences were identified using Splice Site Prediction by Neural Network http://www.fruitfly.org/seq_tools/splice.html. GCG Wisconsin Package version 8 (GeneticsComputerGroup, 1994) was used for all other sequence analyses. cDNA encoding the C-terminal region of the variant βII spectrin cDNA encoding the C-terminal region of the short variant of βII spectrin was obtained using polymerase chain reaction (PCR). In each PCR, one primer represented the known βΙΙ spectrin sequence, the other was designed to be specific for the putative variant βII, based on the EST sequences. Primer 1 represented the 5′ end of the known C-terminal region. Primer 2 was designed from a consensus of EST sequences (see Results) for the short variant, and is complementary to sequence in the 3′ untranslated region. Primer 3 represented sequence encoding the 14th triple helical repeat. (a) PCR using primers 1 and 2 The sequence of the sense primer was as follows: CGTAGGATCCGTGCGCAGACAGCAAG (primer 1). A BamHI site (underlined) was included at the 5′ end to facilitate cloning. The sequence of the antisense primer was GGAATTCCAGAGGATTTGGAAAGGG (primer 2), and included an EcoRI site (underlined). These primers were used to amplify either human skeletal muscle cDNA (2-10 ng per reaction, giving a 0.3 kb product) or human genomic DNA (0.1 µg per reaction, giving a 1 kb product). Amplification was performed using Taq polymerase (Promega) for cDNA or Advantage cDNA polymerase (Clontech) for genomic DNA. The PCR conditions were as follows: 1 cycle of 95°C for 5 minutes, 55°C for 1 minute and 72°C for 1 minute; 30 cycles of 95°C for 30 seconds, 55°C for 1 minute and 72°C for 1 minute; and 1 cycle of 95°C for 30 seconds, 55°C for 1 minute and 72°C for 5 minutes. The products were ligated into the TA cloning vector pGEM-T Easy βII spectrins 2025 (Promega) and transformed into E. coli JM 109. Representative plasmids were isolated and sequenced on both strands in triplicate by the Advanced Biotechnology Centre, Charing Cross and Westminster Medical School, London. (b) PCR using primers 3 and 2 The sequence of the sense primer (primer 3) was as follows: TTAGGGATCCGCAGTCCAAAGTGGATAAAC, and included a BamHI site (underlined). The antisense primer was the same as that described above (primer 2). Advantage cDNA polymerase (Clontech) was used according to the manufacturer’s recommendations and the following three-step PCR protocol was carried out: 1 step of 94°C for 5 minutes; 5 cycles of 94°C for 30 seconds, 72°C for 6 minutes; and 25 cycles of 94°C for 30 seconds, 68°C for 6 minutes. The products were ligated into the TA cloning vector pGEM-T Easy (Promega) and transformed into E. coli JM 109. Representative plasmids were sequenced on both strands by MWG-Biotech UK Ltd, Milton Keynes, UK. Preparation of short βII spectrin C-terminal fusion protein To prepare a glutathione-S-transferase (GST) fusion protein of the short βII spectrin C-terminal region, the 0.3 kb cDNA fragment was ligated into the prokaryotic expression vector pGEX-2T (Pharmacia) and transformed into E. coli JM 109. A 36 kDa GST-fusion protein was induced and purified by glutathione affinity chromatography by methods described for other GST-fusion proteins (Hayes et al., 1997). The recombinant polypeptide was judged pure by the criterion of a single band on 12% SDS-PAGE. Recombinant fragments of long βI and βII spectrins A GST-fusion protein of the long C-terminal region of βI spectrin was prepared as previously described (Hayes et al., 1997). This protein contains residues 2009-2329, and is designated GST-βIΣ2(2009-2329). The fusion protein was cleaved by thrombin to liberate the 35kDa βIΣ2(2009-2329) C-terminal polypeptide. GST was removed from the cleavage mixture by glutathione affinity chromatography. A GST fusion protein of residues 2087-2198 of the long form of βII spectrin was prepared as previously described (Hayes et al., 1997). This is designated GST-βIIΣ1(2087-2198). This contains the region of βII spectrin that binds the axonal protein fodaxin. A modified pGEMEX-1 plasmid containing cDNA encoding amino acid residues 2060-2333 of the human long form of βΙΙ spectrin was also used. Details of this plasmid and purification of the 30 kDa recombinant protein, βΙΙ(2060-2333), from E. coli BL21 (DE3) pLysS are given in Davis and Bennett (1994) and Hayes et al. (1997). Northern blots Total RNA was extracted from various tissues by acid guanidinium thiocyanate-phenol-chloroform extraction (Chomczynski and Sacchi, 1987). Equivalent amounts of RNA (10 µg) were fractionated by denaturing agarose gel electrophoresis, transferred to Hybond-N (Amersham), fixed by ultraviolet illumination and processed for northern blot analysis. cDNA was labelled by random priming using hexaprimers and [α-32P]dCTP for 24 hours at 37°C. Labelled DNA was heat denatured prior to using as a probe. Membranes were prehybridised for 2 hours followed by overnight hybridisation with the appropriate cDNA probe. Both steps were carried out at 65°C. Subsequently, membranes were washed and the signal was visualized using an Instant Imager (Canberra-Packard). Production of antibodies Antibody to the short form of βII spectrin was raised in guinea pigs to a peptide antigen. The peptide NSRRTASDQPWSGL, which represents a unique sequence at the very C-terminal of this spectrin isotype, was synthesized. The peptide was conjugated to keyhole limpet hemocyanin (KLH, Pierce, Chester, UK) via the crosslinker 1ethyl-3-[3-dimethylaminopropyl] carboiimide hydrochloride (EDC). Peptide (2 mg dissolved in 0.1 M MES, pH 4.5) was incubated at room temperature for 2 hours with 2 mg KLH and 0.5 mg EDC with constant shaking. Non-conjugated peptide was separated from KLHconjugate using a 9.1 ml PD-10 Sephadex G25 desalting column (Pharmacia) and filtered using a 0.2 µm Millipore filter. 200 µg of peptide conjugate was mixed with Freund’s complete adjuvant for the initial immunization and with Freund’s incomplete adjuvant for subsequent booster injections at 3-week intervals. The serum from the fourth boost was affinity purified on short βII spectrin C-terminal fusion protein coupled to Activated CH Sepharose 4B (Pharmacia). The affinity-purified antibody was diluted 1:200 for immunoblots and 1:50-1:100 for immunofluorescence. Antibodies to the long form of βII spectrin were raised in rabbits to recombinant βΙΙ(2060-2333) using an immunization protocol essentially the same as described above. Antiserum was affinity purified on βΙΙ(2060-2333) conjugated to Sepharose. It was used at 1:500 on immunoblots and for immunofluorescence at 1:50-1:100. Electrophoresis and blotting methods Various rat tissues were dissected and rapidly powdered in liquid nitrogen in a mortar and pestle. Boiling 5× sample buffer (Laemmli, 1970) was added to the powdered tissue, centrifuged and the supernatant used for SDS-PAGE (Laemmli, 1970) on 6% or 7% polyacrylamide gels. Purified spectrin C-terminal proteins were analysed by 12% SDS-PAGE. Immunoblot analysis was carried out as described by Nicol et al. (1997). Immunohistochemistry of muscle and cerebellar sections 0.5-1 µm semi-thin frozen sections of adult rat extensor digitorum longus (EDL) or diaphragm were prepared and analysed as described by Kordeli et al. (1998). Cerebellum or heart were processed essentially as described previously (Rayner and Baines, 1989). Cerebellar granule cells Granule cells were cultured from 7-day-old rats by standard methods (as given in Nath et al., 1996), and were used in immunofluorescence after 14 days in culture. The cells were fixed in methanol at −20°C for 20 minutes, and processed for immunofluorescence as above. Preparation and use of affinity columns Purified C-terminal GST-βIIΣ1(2087-2198) spectrin, C-terminal GSTβIIΣ2(2087-2167) spectrin (both obtained as described above), GST and bovine serum albumin affinity columns were prepared using Activated CH-Sepharose 4B (Pharmacia) according to the manufacturer’s recommendations. These columns were kept at 4°C and used within 2 months of being made. Conditions for affinity chromatography were the same as previously described (Hayes and Baines, 1994). Briefly, an enriched preparation of rat brain fodaxin was dialysed into buffer A (10 mM sodium phosphate, pH 7.5, 2 mM EGTA, 0.05% Tween 20, 0.5 mM DTT and 0.02% sodium azide), filtered using a 0.45 µm pore filter and applied to the affinity column. The column was washed with 30 column volumes of buffer A. Bound proteins were eluted in buffer A plus 1.2 M KBr. The results were analysed by SDS-PAGE followed by immunoblotting. Preparation of post-synaptic densities Post-synaptic densities were prepared from pig brain by the method of Carlin et al. (1980). RESULTS Detection of an alternatively spliced form of βII spectrin The two splice variants of βI spectrin differ at a region towards the C terminus after the last (partial) triple helical repeat, so 2026 N. V. L. Hayes and others we reasoned that an analogous variation might occur in βII. To purposes of this report we will continue to refer to variants of test this hypothesis, the database of expressed sequence tags βII spectrin with long and short C-terminal regions. (dbEST) was examined. The novel C-terminal region is strikingly similar to ELF1 dbEST was searched using tBLASTn (Altschul et al., 1990) and ELF3, two products of elf, a different β spectrin gene with residues 2087-2194 of human βII (βG) spectrin (GenBank (Mishra et al., 1998, 1999). The nucleotide sequences were M96803). These residues encode a region that links between the helical (A) Segments 15-19 of long (βIIΣ1) and short (βIIΣ2) C-terminal variants. repeating part of βII spectrin and the 1680 YAGLKDLAEERRGKLDERHRLFQLNREVDDLEQWIAEREVVAGSHELGQDYEHVTMLQER PH domain. Five human and one rat βIIΣ1 βIIΣ2 1 YAGLKDLAEERRGKLDERHRLFQLNREVDDLEQWIAEREVVAGSHELGQDYEHVTMLQER ESTs (GenBank entries zl37f06.s1, zl37f06.r1, yi59b05.s1, vy96c01.r1, βIIΣ1 1740 FREFARDTGNIGQERVDTVNHLADELINSGHSDAATIAEWKDGLNEAWADLLELIDTRTQ 61 FREFARDTGNIGQERVDTVNHLADELINSGHSDAATIAEWKDGLNEAWADLLELIDTRTQ vr25f01.r1, yp50d05.s1) indicated the βIIΣ2 existence of an alternatively spliced C βIIΣ1 1800 ILAASYELHKFYHDAKEIFGRIQDKHKKLPEELGRDQNTVETLQRMHTTFEHDIQALGTQ terminus. The ESTs represented clones βIIΣ2 121 ILAASYELHKFYHDAKEIFGRIQDKHKKLPEELGRDQNTVETLQRMHTTFEHDIQALGTQ from a variety of tissues. Primers for PCR were designed to βIIΣ1 1860 VRQLQEDAARLQAAYAGDKADDIQKRENEVLEAWKSLLDACESRRVRLVDTGDKFRFFSM 181 VRQLQEDAARLQAAYAGDKADDIQKRENEVLEAWKSLLDACESRRVRLVDTGDKFRFFSM amplify part of the predicted coding βIIΣ2 region of this spectrin. One primer was βIIΣ1 1920 VRDLMLWMEDVIRQIEAQEKPRDVSSVELLMNNHQGIKAEIDARNDSFTTCIELGKSLLA based on the known βII spectrin βIIΣ2 241 VRDLMLWMEDVIRQIEAQEKPRDVSSVELLMNNHQGIKAEIDARNDSFTTCIELGKSLLA sequence (the 5′ end being codon 2087); 1980 RKHYASEEIKEKLLQLTEKRKEMIDKWEDRWEWLRLILEVHQFSRDASVAEAWLLGQEPY the other was designed to hybridize with βIIΣ1 301 RKHYASEEIKEKLLQLTEKRKEMIDKWEDRWEWLRLILEVHQFSRDASVAEAWLLGQEPY the EST-predicted sequence in the 3′ βIIΣ2 Spectrin repeats ←↓→ C-terminal region untranslated region. Amplification of βIIΣ1 2040 LSSREIGQSVDEVEKLIKRHEAFEKSAATWDERFSALERLTTLELLEVRRQQEEEERKRR human skeletal muscle first strand βIIΣ2 361 LSSREIGQSVDEVEKLIKRHEAFEKSAATWDERFSALERLTTLELLEVRRQQEEEERKRR cDNA gave a single product of the 2100 PPSPEPSTKVSEEAESQQQWDTSKGEQVSQNGLPAEQGSPRMAETVDTSEMVNGATEQRT predicted size (310 bp) (GenBank βIIΣ1 421 PPSPEPSTKVSEEAESQQQWDTSKGEQVSQNGLPAEQGSPRVSYRSQTYQNYKNFNSRRT AJ005694). Amplification of rat skeletal βIIΣ2 muscle, kidney and brain cDNA gave βIIΣ1 2160 SSKESSPIPSPTSDRKAKTALPAQSAATLPARTQETPSAQMEGFLNRKHEWEAHNKKASS identical PCR products, except for a βIIΣ2 481 ASDQPWSGL*.................................................. single nucleotide difference in the 2220 RSWHNVYCVINNQEMGFYKDAKTAASGIPYHSEVPVSLKEAVCEVALDYKKKKHVFKLRL 3′ untranslated region (GenBank βIIΣ1 ............................................................ AJ242018). The amino acid sequences βIIΣ2 of the novel βII C-terminal are identical βIIΣ1 2280 NDGNEYLFQAKDDEEMNTWIQAISSAISSDKHEVSASTQSTPASSRAQTLPTSVVTITSE in rat and human and confirmed the EST βIIΣ2 ............................................................ prediction. 2340 SSPGKREKDKEKDKEKRFSLFGKKK* Amplification of cDNA, from the βIIΣ1 βIIΣ2 .......................... 14th spectrin repeat (codon 1680) to the 3′ end of the new sequence, confirmed that the cDNA derived from βΙΙ (βG) spectrin; the nucleotide sequence was (B) Last helical repeat and C-terminal regions of βIΣ1, βIIΣ2, and ELF3. identical to the human sequence 1 IKEKLLQLTEKRKEMIDKWEDRWEWLRLILEVHQFSRDASVAEAWLLGQEPYLSSREIGQ between codons 1680 and 2140, and βIIΣ2 ELF3 1 IKEKLLQLTEKRKEMIDKWEDRWEWLRLILEVHQFSRDASVAEAWLLGQEPYLSSREIGQ included the ankyrin binding site in the βIΣ1 1 IREKLQQVMSRRKEMNEKWEARWERLRMLLEVCQFSRDASVAEAWLIAQEPYLASGDFGH 15th repeat. After this the two sequences 61 SVDEVEKLIKRHEAFEKSAATWDERFSALERLTTLELLEVRRQQEEEERKRRPPSPEPST show little identity, and the predicted βIIΣ2 61 SVDEVEKLIKRHEAFEKSAATWDERFSALERLTTLELLEVRRQQEEEERKRRPPSPDPNT coding region of the new sequence stops ELF3 βIΣ1 61 TVDSVEKLIKRHEAFEKSTASWAERFAALEKPTTLELKE..RQIAE....RPAEETGPQE after 28 further codons (Fig. 1A). Because the novel C-terminal βIIΣ2 121 KVSEEAESQQQWDTSKGEQVSQNGLPAEQGSPRVSYRSQTYQNYKNFNSRRTASDQPWSG 121 KVSEEAES.QQWDTSKGDQVSQNGLPAEQGSPRVSYRSQTYQNYKNFNSRRTASDHSWSG sequence represents the second subtype ELF3 115 EEGETAGEAPVSHHAATERTSPVSLWSRLSSSWESLQPEPSHPY................ of βII spectrin, it would be βIIΣ2 βIΣ1 spectrin in the nomenclature of βIIΣ2 181 L Winkelmann and Forget (1993). Since ELF3 180 M . the long form of βII was the first to be βIΣ1 discovered, it is βIIΣ1. Because of the relative order of discovery, this is Fig. 1. Sequence of the novel βII spectrin short C-terminal isoform, in comparison to (A) the long C-terminal isoform of βII (βG) and (B) to the short C-terminal isoforms of βI spectrin and opposite to the nomenclature of βI ELF3. In this figure, human βII and βI sequences are marked as follows: short C-terminal form spectrin where Σ1 is the short form. In of βII βIIΣ2; long C-terminal form of βII βIIΣ1; short C-terminal region of βI βIΣ1. Regions of the nomenclature of Hu et al. (1992), the the sequence are indicated in the figure. Note that the novel C-terminal variant is identical to short variant would be βG246. We feel the long C-terminal isoform of βII until residue 2140, after which they diverge. The novel Cthat the nomenclature for the non- terminal sequence is very similar (although not identical) to the C-terminal sequence of mouse specialist is confusing, and for the ELF3, but dissimilar to the short C-terminal sequence of human βI spectrin. βII spectrins 2027 Fig. 2. Genomic organization of exons encoding the short C-terminal βII spectrin. To examine the genomic organization of the two βII spectrin isoforms, PCR was used to amplify human genomic DNA encoding the respective C termini. PCR products were analysed by electrophoresis on 0.9% agarose gels, which were stained with ethidium bromide. PCR analysis of (A) DNA encoding the 3′ coding regions of the long C-terminal isoform of βII (lane 1) or the short C-terminal isoform of βII (lane 2) from human genomic DNA: 3 kb and 1 kb bands were obtained respectively. The 3 kb band was isolated, and used as a template for PCR with primers for the short C-terminal. A 1 kb band resulted (lane 3). These data indicate that the 3′ coding region of the short C-terminal form is contained within the region of the sbtb2 gene that contains the exons encoding the 3′ coding sequence of the long C-terminal form. (B) Generation of long and short C-termini by differential exon usage. Schematic diagram representing the genomic DNA sequence that encodes the two βII spectrin isotypes. An intron interrupts the coding region between codons 2119 and 2120. An exon starting at 2120 reads through to the stop codon of the short C-terminal isoform (asterisk). A splice donor site within this exon exists at codon 2140 (marked spl). Usage of this site removes the region containing this stop codon and joins to an exon encoding the remainder of the C-terminal of the long isoform. 90% identical. Fig. 1B shows comparison of the amino acid sequences of the novel C-terminal region with mouse ELF3 and human erythrocyte β spectrin (βIΣ1) C-terminal regions. The novel βII C-terminal and the ELF3 C-terminal regions seem to represent a common sequence domain for these two spectrins, clearly distinct from the short βI spectrin C-terminal region. We also noted an EST from adult rat (GenBank AI030132) that is 98% identical to the mouse elf3 3′ coding sequence, and which almost certainly derives from the rat elf gene. This EST is only 91% identical to the novel rat βII short sequence. This is consistent with rats having both the sbtb2 and elf genes. The novel βII C-terminal region is encoded by the same gene as the long form of βII spectrin The different C-terminal sequences of βI spectrin derive from alternative mRNA splicing within an exon (Winkelmann et al., 1990a). To determine if this was also the case with βII spectrin, human genomic DNA was amplified by PCR using primer pairs in which one primer encoded the common region of both spectrins, and the other was specific for the 3′ coding regions of either the long or short C-terminal regions (see Materials and Methods). Amplification of genomic DNA encoding the long form gave a product of approx. 3 kb (Fig. 2A, lane 1); correspondingly, the primers representing the short form gave a product of approx. 1 kb (Fig. 2A, lane 2). The 3 kb product was purified and reamplified with primers for the short form. This gave a product of approx. 1 kb (Fig. 2A, lane 3), indistinguishable by electrophoresis and restriction mapping from the product of direct amplification of the original genomic DNA. The region encoding the short form is evidently contained within the region of the sptb2 gene encoding the long form. The 1 kb product was sequenced (GenBank AJ238723; shown schematically in Fig. 2B). The coding region of βII Cterminal region is interrupted between codons 2119/2120 by a 678 bp intron. The following exon encodes the C-terminal region of the new form. Within this exon, a splice donor site is predicted (TCCACGGGTTAGTTA – marked spl in Fig. 2B). This is at the point in the sequence where the new and original forms vary. We infer that, as with βI spectrin (Winkelmann et al., 1990a), a splicing event can occur within the exon that encodes the C-terminal region of the short form. The splicing event removes the stop codon present in the new form, and allows the translation of a longer form of βII spectrin. Expression of the mRNA for the short form in rat tissues To investigate the expression of mRNA encoding the short Cterminal sequence in the tissues of rat, northern blot analysis was used. Total RNA from a variety of tissues was hybridized under stringent conditions with a 142 bp human cDNA probe, which comprised a sequence unique to the short form and did not overlap with the long form (codons 2144-2167 of the short form). Fig. 3 shows that two transcripts of approx. 7.5 kb (a strong signal) and 7.0 kb (a much weaker signal) were found in multiple tissues, being especially prominent in cardiac and brain tissue, but detectable in mesenteric arteries, aorta and lung tissue. Note that the last three of these have a significant smooth muscle component. For comparison, Hu et al. (1992) detected widespread expression of an approx. 9.4 kb transcript of the long form of βII (βG) spectrin. They found expression in brain was relatively high while that in heart was much lower: we have also confirmed this data using RT-PCR (not shown). Northern blotting also distinguishes expression of mRNA encoding short C-terminal βII from that encoding ELF3. In particular, elf3 mRNA (9.0 kb) is longer than the short βII, and abundant in lung, kidney and liver (Mishra et al., 1999), in which our northern analysis detected little or no short Cterminal βII mRNA. The more sensitive RT-PCR method was also used to detect expression of the short C-terminal variant. Low levels of expression were detected in kidney, liver and intestinal epithelia, possibly reflecting the smooth muscle component of blood vessels in these tissues. Sequencing of RT-PCR products confirmed that these were the short variant of βII spectrin and not an elf gene product. Antibodies to the two forms of βII spectrin The sequences of the two forms of βII spectrins diverge at amino acid residue 2139. The following 28 amino acids in the short C-terminal βII spectrin variant are unique. To generate an antibody against the short C-terminal form, a synthetic 2028 N. V. L. Hayes and others Fig. 3. Northern blot analysis of short C-terminal βII spectrin transcripts. Total RNA was prepared from the indicated rat tissues, fractionated by denaturing agarose gel electrophoresis (10 µg RNA per lane) and transferred to Hybond-N. The filter was probed with 32P-labelled cDNA encoding only the unique region of the short isoform of βII spectrin. Bands of approx. 7.5 and 7.0 kb (indicated) were detected in several tissues, the strongest expression being in heart (all four quadrants) and brain. peptide was prepared representing the C-terminal 14 residues from this sequence (see Fig. 1 and Materials and Methods). A guinea pig antiserum was prepared as indicated in Materials and Methods, and affinity purified on a GST-fusion protein of the C-terminal 80 residues. Specificity of this antibody for the short form was confirmed by immunoblotting with recombinant fragments of the long and short forms of βII, as well as the long form of βI. GST-short βII C-terminal spectrin fusion protein was specifically recognized by the anti-peptide antibody (Fig. 4, middle panel, lane 1) but recombinant long form βII fragment (GST-βIIΣ1(2087-2198)) was not (Fig. 4, middle panel, lane 2); neither was any immunoreactivity observed to recombinant βI long form fragment (βIΣ2(2009-2329)) (Fig. 4, middle panel, lane 3). Rabbit antiserum against the long C-terminal form of βII spectrin was produced using recombinant polypeptide representing the C-terminal region of the long form of βII (βIIΣ1(2087-2198)). Antibody to the long form detects the recombinant long form C-terminal fragment (recombinant βIIΣ1(2087-2198)) (Fig. 4, right hand panel, lane 2) but not the short C-terminal fragment (lane 1), even though the fusion proteins share a sequence of 42 amino acids at their respective N termini (see Fig. 1). The antibody to the long form also showed no reactivity with the C-terminal fragment of the long form of βI spectrin (recombinant βIΣ2(2009-2329)) (lane 3). The antibodies therefore recognize unique sequences in these spectrin isotypes and can therefore distinguish between the two alternatively spliced forms of βII spectrin. No reaction was observed either by western blotting or immunofluorescence with erythrocyte spectrin (βIΣ1 – data not shown). Bearing in mind the extent of sequence identity between the Fig. 4. Immunoblot analysis of antibodies to βII spectrin isotypes. The figure shows replicate samples of recombinant β spectrin C-terminal fragments after electrophoresis on 12% SDS-polyacrylamide gels. Gels were either stained with Coomassie Blue (left), or transferred to nitrocellulose paper and probed with antibodies to the short C-terminal form (middle) or to the long C-terminal form of βII (right). The protein samples were (1) GST-βIIΣ2(2087-2167) representing the C terminus of the short isoform of βII (2) GST-βIIΣ1(2087-2198) representing the C terminus of the long isoform of βII and (3) βIΣ2(2009-2329) representing the C terminus of the long isoform of βI. The two anti-βII spectrin antibodies appear specific for their respective isotypes. short C-terminal βII spectrin variant and ELF1/3, it seemed likely that antibodies to the short C-terminal βII spectrin variant might be cross-reactive between these antigens. A synthetic peptide was prepared using the equivalent sequence of ELF1/3. This displaced approx. 60% of the reactivity from GST-βIIΣ2 fusion protein in ELISA (not shown). This indicates that antibodies to the short form will show some cross-reactivity with ELF1/3. However, when we attempted to detect the 27 kDa ELF1 protein that has been reported in brain, kidney and liver using immunoblots (not shown) no reactivity was found, indicating that for practical purposes this antibody does not efficiently recognize the ELF proteins. Using immunoblots calibrated with the GST-short βII Cterminal construct, the abundance of the short form of βII spectrin was measured at approximately 2.5 pmol/mg protein in a brain membrane fraction. This compares with estimates of total spectrin content in brain membrane fraction of 30 pmol spectrin tetramer/mg protein (Davis and Bennett, 1984). Comparison between the distribution of the two Cterminal variants in rat muscle and nervous tissues The distributions of the short and long C-terminal variants were investigated in rat tissues (Fig. 5) by immunoblot. Immunodetection using antibodies to the long form revealed a single approx. 280 kDa band in heart, hind limb skeletal muscle, forebrain, cerebellum and spinal cord (Fig. 5, middle panel, lanes 1-5). Antibodies against the short C-terminal variant detected a strong band at approximately 240 kDa in heart (Fig. 5, lane 1, right hand panel). In hind limb skeletal muscle (lane 2) a single 240 kDa immunoreactive product was observed, but in forebrain (lane 3) only a single 270 kDa band was observed. In the cerebellum (lane 4) two bands of 240 kDa and 270 kDa were seen; in spinal cord (lane 5) only the 240 kDa band was observed. All immunoreactivity disappeared if 100-fold excess of the peptide antigen was present in the βII spectrins 2029 Fig. 5. Immunoblot analysis of βII spectrin isotypes in rat muscle and neuronal tissues. Replicate 6% SDSpolyacrylamide gels of tissue homogenates (lane 1, heart; lane 2, skeletal muscle; lane 3, forebrain; lane 4, cerebellum; lane 5, spinal cord) were eitherstained with Coomassie Blue (left) or transferred to nitrocellulose paper and probed with antibodies to the long C-terminal form (middle) or to the short C-terminal form of βII (right). Antibodies to the long C-terminal region of βII detect a 270 kDa antigen in each of the tissues. Antibodies to the short C-terminal region of βII detect an approx. 240 kDa band in heart, skeletal muscle cerebellum and spinal cord; an additional band at approx. 270 kDa is detected in forebrain and cerebellum. primary antibody (data not shown). These data indicate that both variants of βII spectrin can be present in the same tissue. Selective incorporation of the short C-terminal βII spectrin variant into muscle M line regions, but both forms are Z line components The immunoblotting data suggested that both variants are expressed in heart, skeletal muscle and nervous tissue. Indirect immunofluorescence was used to compare the distribution of the two βΙΙ spectrins in rat skeletal muscle (Fig. 6) and in ventricular heart muscle (Fig. 7). In longitudinal sections of skeletal muscle (extensor digitorum longus; Fig. 6A-F), both βII spectrin variants were at the Z-line level, as indicated by counterstaining with antibodies to α-actinin. In addition punctate staining with antibodies to the long form was also observed. As with heart, antibodies to the short form showed clear labelling throughout the depth of the M-line level; no M-line labelling was detected for the long form (Fig. 6C,F, insets). Specificity of the antibodies on these sections was confirmed by preabsorption of the antibodies with relevant antigen, which abolished specific staining (not shown). In cross-sections of diaphragm (not shown) neither βII spectrin was especially concentrated at or near the plasma membrane, and in particular, no labelling was detected at neuromuscular junctions (NMJs), again in clear distinction to βI spectrin. Both βII spectrin antigens were detected at the Z line in cardiac tissue (Fig. 7). In confocal microscopy, their distribution overlapped with anti-α-actinin (a wellcharacterised Z disc component). However, in addition to Zline staining, a longitudinal myoplasmic distribution was observed for both βII polypeptides, although this was more obvious with antibodies to the long form. The most striking difference was selective staining of the central region of the sarcomere with antibodies to the short form: this staining is coincident with the M line. It was unlikely in heart that either form was associated with the T-tubule system, as a similar regular striated staining pattern was observed in cardiac atrial tissue, which does not contain many T-tubules (data not shown). No sarcolemmal or costameric labelling was detected with either antibody (not shown), clearly distinguishing βII from βI spectrin in muscle (Wood and Slater, 1998; Zhou et al., 1998). In some sections immunoreactivity with blood vessels was detected. In these cases, antibodies to both the βII isotypes reacted with the smooth muscle cells lining the vessels, but not with erythrocytes within the blood vessels. Retention of the short C-terminal variant near the cell soma of cerebellar granule neurons In frozen sections of cerebellar molecular and granule cell layers (Fig. 8A), the long C-terminal variant was detected in the granule cell layer, the Purkinje cells and the parallel fibres of the molecular layer. By contrast, little if any immunoreactivity for the short βII C-terminal variant was found in the parallel fibres, although cerebellar granule cells and Purkinje cell bodies were recognized by the antibody (Fig. 8B). Purkinje cell dendrites were not stained. This indicates that the short βII C-terminal variant is probably restricted to regions close to cell bodies. In the white matter tracts, axons were abundantly labelled with antibodies to the long form, but staining with antibodies to the short form was comparatively weak (not shown). In cultures of primary rat cerebellar granule neurons (Fig. 8C-E) both isoforms were present in individual cells, but the short C-terminal βII variant had a more restricted distribution than the long form. The long C-terminal form was present in many neurites, but the short βII variant was predominantly in the cell soma, with a distribution in neurites primarily limited to their proximal segments. Many fine processes were noted in the culture that only reacted with antibodies to the long form, and not at all with the short βII variant. The distribution of the short βII variant in cell bodies appeared very diffuse, and was not particularly associated with plasma membranes. In both sections of cerebellum and primary granule neuronal cultures, the short C-terminal variant was restricted more closely to the cell body than the long C-terminal variant. Fodaxin does not interact with the C-terminal region of the short form of βII spectrin Fodaxin is an axonal protein that interacts with the long Cterminal form of βII spectrin (Hayes et al., 1997), and is also present in cerebellar parallel fibres and white matter axons (Rayner and Baines, 1989). Fodaxin therefore codistributes with the long, but not the short C-terminal βII. The region of the long form of βII spectrin that interacts with fodaxin comprises residues 2087-2198, i.e. the region of variation 2030 N. V. L. Hayes and others between the long and short C-terminal forms. Given this, it was important to test whether or not the short form retained the fodaxin binding activity. To assay binding, affinity chromatography was used with affinity columns made from recombinant C-terminal fragments of each isoform. An enriched preparation of fodaxin from rat cerebral cortex (Hayes et al., 1997) was prepared for affinity chromatography and applied to an affinity column made by coupling a GST fusion protein of the short C-terminal region to Sepharose (Fig. 9). Fodaxin was detected using the monospecific monoclonal antibody DR1. Fig. 9A shows that, as expected, fodaxin interacted with the affinity column made from the long C-terminal form. Fodaxin did not appear in the breakthrough fractions from this column, and was only recovered by eluting the column with a dissociating buffer containing 1.2 M KBr. By contrast, when fodaxin was applied to the affinity column of the short C-terminal form (Fig. 9B), it was found in the breakthrough fractions only and was not detectable in the wash or fractions that had been eluted with 1.2 M KBr. These results indicate that the Cterminal region of the short form does not interact with fodaxin under these conditions and demonstrate a difference in activity between the C-terminal regions of the two βII spectrin variants. The short βII variant is not retained in postsynaptic density preparations One further comparison of the properties of the two βII spectrin variants was made by examining their relative retention in preparations of postsynaptic densities (PSDs). PSDs are enriched in spectrin (Carlin et al., 1983), including βII spectrin, which appears to form part of an apparatus bridging certain receptors to cytoplasmic actin (Wechsler and Teichberg, 1998). PSD preparations from pig brain were compared for their content of the long and short forms of βII spectrin (Fig. 10). Coomassie Blue stained gels of equal protein loadings of brain homogenate and a PSD preparation are shown in Fig. 10, lanes 1 and 2. Corresponding immunoblots of PSDs were probed with antibodies to the two forms (long form, lanes 3 and 4; short form, lanes 5 and 6), with the known PSD protein PSD-95 as a standard PSD marker (lanes 7 and 8). Note the enrichment of PSD-95 in the PSD preparation. The 280 kDa band of the long C-terminal βII variant was readily detected in the PSD preparation, but the short C-terminal variant was clearly depleted. The long C-terminal region of βII spectrin correlates with targeting to and/or retention in PSDs, and these data suggest that there are PSDspecific interactive ligands for this variant. DISCUSSION In this paper we describe a novel variant of βII spectrin. This variant is derived from the same gene (sptb2) as the previously published βII (βG) spectrin (Hu et al., 1992; Ma et al., 1993), but differs through alternative mRNA splicing near the 3′ end of the coding region. The predicted protein sequence shown in Fig. 1 is shorter than the first βII sequence to be published, and contains no PH domain. The site of variation occurs about half way between the last (partial) triple helical repeat and the PH domain. The new sequence is only 28 amino acid residues long, and is generally hydrophilic in character. It is comparatively rich in amino acid residues carrying hydroxyl groups (ser, thr and tyr: 10 of these in total). Strikingly, the C-terminal region from the end of the helical repeats to the short C terminus is nearly identical in amino acid sequence to the corresponding regions in two products of the elf gene, the polypeptides ELF1 and ELF3 (Mishra et al., 1998, 1999) (Fig. 1). Indeed the nucleotide sequences are about 90% identical. There is no question, though, that the cDNA that we have characterized comes from the gene that encodes βII spectrin (the sptb2 gene). First, the sequence of the region of the coding sequence upstream of the variant region is absolutely identical to the published βII sequence, but only about 90% Fig. 6. Immunofluorescence analysis of βII spectrin isoforms in rat skeletal muscle. 0.5 µm semi-thin frozen longitudinal sections of EDL skeletal muscle were double-labelled with antibodies to (A) the short C-terminal variant of βII spectrin or (D) the long C-terminal variant of βII spectrin, and compared with antibodies to (B,E) α-actinin. The insets in each image show higher magnification views of the regions outlined in the main image. There is extensive coincidence of the two stains for both isoforms of βII spectrin and α-actinin in the Z-line region (yellow in merged image, C,F) but there is selective staining for the short Cterminal form at the level of the M line (arrows) and in longitudinal elements (C). Νo staining for the long C-terminal variant is detectable at the level of the M-line. Scale bar, 10 µm. βII spectrins 2031 Fig. 7. Immunofluorescence analysis of βII spectrin isotypes in rat cardiac muscle. Longitudinal frozen sections of formaldehyde-fixed cardiac muscle (4 µm thickness) were labelled with antibodies to (A) the short C-terminal variant of βII spectrin or (B) the long C-terminal variant of βII spectrin. As with skeletal muscle (Fig. 6), the major striations labelled with both antibodies are Z lines. In addition, labelling at the M-line level was noted selectively with antibodies to the short form, indicated by arrowheads in A. Scale bar, 10 µm. identical to elf. Second, the coding region for the novel C terminus is within a 3 kb genomic DNA region that contains sequence encoding both C termini (Fig. 2). The exon that encodes the end of the novel reading frame contains within it a predicted splice site at the point of sequence variation between the long and short forms. This is similar to the exon organization in the gene encoding βI (Winkelmann et al., 1990b). Third, a rat EST that appears to encode an elf product can be distinguished from the rat cDNA sequence for our variant βII (noted in Results). In comparison with βII variation, no elf gene product has yet been published that has a PH domain. A major concern in our investigations of the short βII Cterminal mRNA and polypeptide is the degree of similarity to elf gene products. It was conceivable that our investigations of these might have been complicated by cross-hybridization of cDNA probes, or cross-reaction of antibodies. As indicated in Results, we view our probes as being specific for sptb2 products. Our RT-PCR is clearly specific for cDNA encoding βII since we have sequenced representative PCR products. In northern blots (Fig. 3), the size of the mRNA encoding the βII spectrin short C-terminal variant was smaller than the reported size (Mishra et al., 1999) of ELF3 mRNA, and the tissue distribution clearly differed. In particular, where heart contains little ELF3 mRNA, this is one of the most abundant sources of the short βII C-terminal variant. Antibodies to the short βII variant show limited cross-reactivity with an ELF peptide in ELISA, but do not detect ELF1 in blots of tissues, nor do they crossreact detectably with βI or the long C-terminal form of βII (Fig. 4). All these data indicate that we have only used probes specific for the relevant spectrin. The reaction of two bands in brain tissue with the antibodies to the short C-terminal variant was noted in Fig. 5: one was approx. 240 kDa, the other approx. 270 kDa. The 240 kDa form was also found in other tissues, and its size correlates well with the predicted size of the short-terminal variant βII polypeptide Fig. 8. βII spectrin isotypes in cerebellum and cultured cerebellar granule cells. The distributions of (A) long and (B) short C-terminal βII isoforms in rat cerebellum granule and molecular layers were analysed by immunofluorescence. Note that both isoforms are present in cell bodies in the granule cell layer (gc) and in Purkinje cell bodies (pc). However, the shorter isoform is not detectable in the parallel fibres (pf) of the molecular layer, unlike the long form. Primary cultures of rat cerebellar granule cells were analysed by immunofluorescence using (C) antibodies to the long C-terminal region of βII or (D) antibodies to the short C-terminal region of βII followed by FITC-conjugated or TRITC-conjugated secondary antibodies, respectively. A merged image of C and D is shown in (E). Note that the long C-terminal isoform is much more extensively distributed in the neuronal processes than the short C-terminal form (C compared to D) and that the short form is more prominent in the perikarya and neurite proximal region than in the neurite distal region. (246 kDa). The molecular identity of the approx. 270 kDa band is not clear, but it is not likely to be ELF3, for the reasons noted above. One possibility is that it represents another splice variant of βII spectrin that has not yet been characterized. Immunofluorescence (Figs 6-8) indicates that the two βII spectrin isoforms have a distribution that overlaps in any one cell, but is not identical. In skeletal and cardiac muscle, both isoforms are present at the level of the Z line, but only the Mline level contains the short C-terminal variant (Figs 6, 7). Cterminal variation therefore correlates with intracellular targeting. 2032 N. V. L. Hayes and others Fodaxin eluted (arbitrary units) Fig. 9. The long, but not the short, C(A) βIIΣ 1 segment 19L (B) βIIΣ 2 segment 19 terminal isoform of βII spectrin interacts with fodaxin. The interaction of fodaxin and 100 100 the two βII spectrins was assayed by affinity chromatography. Affinity columns made 80 80 from the indicated spectrin constructs were used to test the interaction with sheep brain 60 60 fodaxin. 0.5 ml of fodaxin was loaded onto (A) a column of a recombinant fragment of 40 40 the long C-terminal region of βII spectrin coupled to Sepharose or (B) a column of a 20 20 recombinant fragment of the short Cterminal region of βII spectrin coupled to 0 0 Sepharose. The start of the x axis (volume) 0 10 20 30 40 0 10 20 30 40 on these graphs marks the point of loading. The sample was allowed to run into the Volume (ml) Volume (ml) column, which was washed with binding buffer until the point indicated by the arrow. At this point the buffer was changed to a dissociating buffer containing 1.2 M KBr. Fractions eluting from the column were monitored by immunoblotting and densitometry. Fodaxin is retained only on the long C-terminal region affinity column, not on the short C-terminal region affinity column, or on other control columns (not shown), indicating selective binding of fodaxin to one spectrin isoform. The distribution of the βII spectrin isoforms in muscle clearly differs from the βI isoform. The βI spectrin is concentrated at the sarcolemma (e.g. Porter et al., 1997; Zhou et al., 1998) and enriched in neuromuscular junctions (Wood and Slater, 1998). The novel βII spectrin we describe here is enriched in internal structures, not sarcolemma or neuromuscular junctions (NMJs). We specifically examined NMJs in diaphragm using α-bungarotoxin and were unable to detect colocalizing βII short C-terminal variant. No immunofluorescence from erythrocytes in blood vessels in our sections was detected with this antibody, indicating no crossreactivity with erythrocyte βI spectrin. Zhou et al. (1998) reported little reaction of anti-β non-erythroid spectrin with internal structures of muscle; however, a number of other authors (e.g. Isayama et al., 1993) have reported spectrins associated with internal muscle structures. Since the exact isotype specificity of the antibodies used by Zhou et al. (1998) was not reported in their paper, it is difficult to make comparisons between their results and those reported here. Nervous tissue is an abundant source of both isotypes of βII spectrin, and tissue with a muscle component (especially skeletal muscle and heart) contains readily detected short C-terminal βII variant (Figs 3, 5). Although the short variant is detectable by RT-PCR in tissues characterized by the presence of, for example, polarized epithelia (kidney or intestinal epithelium) it is a minor product in these tissues. The short βII is expressed most abundantly in electrically excitable tissue. The short C-terminal βII variant exists at about 2.5 pmol/mg protein in brain membrane fraction (see Results). Total brain spectrin tetramer is estimated at 30 pmol/mg, i.e. total β spectrin is about 60 pmol/mg. The short βII is therefore about 4% of total β spectrin. Bearing in mind the restricted nature of its distribution, and that there are four β spectrin genes, the fact that it is a comparatively small proportion of total β spectrin is not surprising. We might estimate that in the cerebellar granule/Purkinje cell layer, the short variant is perhaps three to four times more concentrated than in total homogenate, i.e. representing 12-15% of total β spectrin, a significant proportion in the context of a cell layer documented to have products of βI and βIII gene products, as well as another variant of βII. Fig. 10. Enrichment of the long C-terminal form of βII spectrin in post-synaptic density preparations, but depletion of the short form. The presence of the two βII spectrin isoforms in preparations of postsynaptic densities was investigated by immunoblot. The figure shows replicate gels of total pig brain homogenate (1, 3, 5, 7) and postsynaptic densities (2, 4, 6, 8). Total protein loadings in each lane were the same. The gels were stained with Coomassie Blue (lanes 1 and 2), or transferred to nitrocellulose and probed with antibodies to the long form of βII spectrin (lanes 3, 4), antibodies to the short form of βII spectrin (lanes 5, 6) or to the known post-synaptic protein PSD-95 (lanes 7, 8). Note the enrichment of PSD-95 in the preparation. The long isoform of βII spectrin is moderately enriched in the PSD preparation, relative to starting homogenate, but the short isoform is strongly depleted. The positions of marker proteins (kDa) are shown. In cerebellum, the distributions of both the long and short βII isoforms differ from each other and the long βI spectrin. Malchio-Dialbedi et al. (1993) reported that the long Cterminal isoform of βI, βIΣ2 spectrin, was enriched in granule cells and Purkinje cell dendrites; it was abundant in PSDs. In contrast and in agreement with the overall distribution of βII βII spectrins 2033 reported by Clark et al. (1994) the long βII was abundant in axons (cerebellar parallel fibres, cultured granule cell neurites). The short βII had a more restricted distribution: the cell bodies of proximal neurite segments of granule cells, and Purkinje cell bodies were positive for the short βII; PSDs were depleted of short βII. In this context, it is of interest to note that PSDs contain N-methyl-D-aspartate (NMDA) receptors. NMDA receptors have been reported to bind βII spectrin (Wechsler and Teichberg, 1998). Since the long, but not the short, βII isoform is present in PSDs, it may be that NMDA receptors form one of the classes of direct membrane binding site for the Cterminal region of βII spectrin reported by Lombardo et al. (1994) and Davis et al. (1994). βIII spectrin has been reported to be associated with presynaptic structures (Sakaguchi et al., 1998). Different β spectrin complements seem to characterize the various compartments of cerebellar neurons. The intra-neuronal compartmentation of the two βII forms correlates also with their comparative binding activities. Fodaxin is an axonal protein of the membrane skeleton in many neurons (Rayner and Baines, 1989). In particular it is present in the parallel fibers of the cerebellar molecular layer that are axons of granule cells (Hayes and Baines, 1994; Rayner and Baines, 1989). Fig. 9 indicates that only the long form can bind fodaxin; this correlates with the enrichment of this form in parallel fibres. Correspondingly, the short βII is restricted primarily to cell body structures, and does not bind fodaxin. βIΣ2 also does not bind fodaxin (Hayes et al., 1997), again corresponding with the lack of colocalisation. All these data indicate an exquisite level of compartmentation of spectrin isoforms in the nervous system, which correlates with the selective nature of their interactions. All these observations point to the variant C-terminal regions correlating with differential subcellular targeting. The simplest deduction is that the C termini act as targeting modules for their respective polypeptides. This has a precedent in the observation that a β spectrin associated with Golgi is targeted to it by interactions of the PH domain (Godi et al., 1998). It will be important to test the hypothesis that the short βII spectrin C-terminal functions as, for example, an M-line targeting module in striated muscle. Likewise, it will be interesting to test if the βII long C-terminal is a PSD-targeting module. This work was supported by a project grant to A.J.B. from the Biotechnology and Biological Sciences Research Council, and equipment grants from the Wellcome Trust. C.S. is the recipient of a BBSRC Special Studentship. Work in V.O.’s laboratory is supported by the British Heart Foundation. A.J.B. and E.K. acknowledge a British Council Alliance grant. This work benefited from the BBSRC’s Seqnet facility. We thank Ms Shweta Singh for supplying cerebellar granule cell cultures. REFERENCES Altschul, S. F., Gish, W., Miller, W., Myers, E. W. and Lipman, D. J. (1990). Basic Local Alignment Search Tool. J. Mol. Biol. 215, 403-410. Bennett, V. and Gilligan, D. M. (1993). The spectrin-based membrane skeleton and micron-scale organization of the plasma-membrane. Annu. Rev. Cell Biol. 9, 27-66. Carlin, R. K., Bartelt, D. C. and Siekevitz, P. (1983). Identification of fodrin as a major calmodulin-binding protein in postsynaptic density preparations. J. Cell Biol. 96, 443-448. Carlin, R. K., Grab, D. J., Cohen, R. S. and Siekevitz, P. (1980). Isolation and characterization of postsynaptic densities from various brain regions: enrichment of different types of postsynaptic densities. J. Cell Biol. 86, 831845. Carugo, K. D., Banuelos, S. and Saraste, M. (1997). Crystal structure of a calponin homology domain. Nat. Struct. Biol. 4, 175-179. Chomczynski, P. and Sacchi, N. (1987). Single-step method of RNA isolation by acid guanidinium thiocyanate- phenol-chloroform extraction. Anal. Biochem. 162, 156-159. Clark, M. B., Ma, Y. P., Bloom, M. L., Barker, J. E., Zagon, I. S., Zimmer, W. E. and Goodman, S. R. (1994). Brain alpha-erythroid spectrin – identification, compartmentalization, and beta-spectrin associations. Brain. Res. 663, 223-236. Davis, J. and Bennett, V. (1984). Brain ankyrin. A membrane associated protein with binding sites for spectrin, tubulin and the cytoplasmic domain of the erythrocyte anion channel. J. Biol. Chem. 259, 13550-13559. Davis, L. H. and Bennett, V. (1994). Identification of two regions of βG spectrin that bind to distinct sites in brain membranes. J. Biol. Chem. 269, 4409-4416. De Matteis, M. A. and Morrow, J. S. (1998). The role of ankyrin and spectrin in membrane transport and domain formation. Curr. Opin. Cell Biol. 10, 542-549. GeneticsComputerGroup (1994). Program Manual for the Wisconsin Package, Version 8. 575 Science Drive, Madison, Wisconsin, USA 53711. Godi, A., Santone, I., Pertile, P., Devarajan, P., Stabach, P. R., Morrow, J. S., DiTullio, G., Polishchuk, R., Petrucci, T. C., Luini, A. and DeMatteis, M. A. (1998). ADP ribosylation factor regulates spectrin binding to the Golgi complex. Proc. Natl. Acad. Sci. USA 95, 8607-8612. Harris, H. W., Jr. and Lux, S. E. (1980). Structural characterization of the phosphorylation sites of human erythrocyte spectrin. J. Biol. Chem. 255, 11512-11520. Hayes, N. V. L. and Baines, A. J. (1994). Axonal membrane-skeletal protein A60: association with a brain spectrin binding activity, and entry into cerebellar axons at a stage after the initiation of axonal growth. J. Neurochem. 62, 300-306. Hayes, N. V. L., Phillips, G. W., Carden, M. J. and Baines, A. J. (1997). Definition of a sequence unique in beta II spectrin required for its axonspecific interaction with fodaxin (A60). J. Neurochem. 68, 1686-1695. Holleran, E. A. and Holzbaur, E. L. F. (1998). Speculating about spectrin: new insights into the Golgi-associated cytoskeleton. Trends Cell Biol. 8, 2629. Hu, R. J., Watanabe, M. and Bennett, V. (1992). Characterization of human brain cDNA-encoding the general isoform of beta-spectrin. J. Biol. Chem. 267, 18715-18722. Isayama, T., Goodman, S. R. and Zagon, I. S. (1993). Localization of spectrin isoforms in the adult mouse heart. Cell Tissue Res. 274, 127-133. Kordeli, E., Ludosky, M. A., Deprette, C., Frappier, T. and Cartaud, J. (1998). AnkyrinG is associated with the postsynaptic membrane and the sarcoplasmic reticulum in the skeletal muscle fiber. J. Cell Sci. 111, 21972207. Laemmli, U. K. (1970). Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680-685. Lombardo, C. R., Weed, S. A., Kennedy, S. P., Forget, B. G. and Morrow, J. S. (1994). Beta II-spectrin (fodrin) and beta I epsilon 2-spectrin (muscle) contain NH2- and COOH-terminal membrane association domains (MAD1 and MAD2). J. Biol. Chem. 269, 29212-29219. Ma, Y. P., Zimmer, W. E., Riederer, B. M. and Goodman, S. R. (1993). The complete amino acid sequence for brain β-spectrin (β-fodrin) – relationship to globin sequences. Mol. Brain Res. 18, 87-99. Malchio-Dialbedi, F., Ceccarini, M., Winkelmann, J. C., Morrow, J. S. and Petrucci, T. C. (1993). The 270-kDa splice variant of erythrocyte β-Spectrin (βIΣ2) segregates in-vivo and in-vitro to specific domains of cerebellar neurons. J. Cell Sci. 106, 67-78. Manno, S., Takakuwa, Y., Nagao, K. and Mohandas, N. (1995). Modulation Of Erythrocyte-Membrane Mechanical Function By Beta-Spectrin Phosphorylation and Dephosphorylation. J. Biol. Chem. 270, 5659-5665. McMahon, A. P., Giebelhaus, D. H., Champion, J. E., Bailes, J. A., Lacey, S., Carritt, B., Henchman, S. K. and Moon, R. T. (1987). cDNA cloning, sequencing and chromosome mapping of a non-erythroid spectrin, human alpha-fodrin [published erratum appears in Differentiation 1987; 34, 241]. Differentiation 34, 68-78. Mishra, L., Cai, T., Levine, A., Weng, D., Mezey, E., Mishra, B. and Gearhart, J. (1998). Identification of elf1, a beta-spectrin, in early mouse liver development. Int. J. Dev. Biol. 42, 221-224. 2034 N. V. L. Hayes and others Mishra, L., Cai, T., Yu, P., Monga, S. P. and Mishra, B. (1999). Elf3 encodes a novel 200-kD beta-spectrin: role in liver development. Oncogene 18, 353364. Musacchio, A., Gibson, T., Rice, P., Thompson, J. and Saraste, M. (1993). The PH domain: a common piece in the structural patchwork of signalling proteins. Trends Biochem. Sci. 18, 343-348. Nath, R., Raser, K. J., Stafford, D., Hajimohammadreza, I., Posner, A., Allen, H., Talanian, R. V., Yuen, P. W., Gilbertsen, R. B. and Wang, K. K. W. (1996). Nonerythroid alpha-spectrin breakdown by calpain and interleukin 1beta-converting-enzyme-like protease(s) in apoptotic cells – contributory roles of both protease families in neuronal apoptosis. Biochem. J. 319, 683-690. Nicol, S., Rahman, D. and Baines, A. J. (1997). Ca2+-dependent interaction with calmodulin is conserved in the synapsin family: Identification of a highaffinity site. Biochemistry 36, 11487-11495. Ohara, O., Ohara, R., Yamakawa, H., Nakajima, D. and Nakayama, M. (1998). Characterization of a new beta-spectrin gene which is predominantly expressed in brain. Mol. Brain Res. 57, 181-192. Pedroni, S., Lecomte, M. C., Gautero, H. and Dhermy, D. (1993). Heterogeneous phosphorylation of erythrocyte spectrin beta chain in intact cells. Biochem. J. 294, 841-846. Porter, G. A., Scher, M. G., Resneck, W. G., Porter, N. C., Fowler, V. M. and Bloch, R. J. (1997). Two populations of beta-spectrin in rat skeletal muscle. Cell Motil. Cytoskel. 37, 7-19. Rayner, D. and Baines, A. J. (1989). A novel component of the axonal cortical cytoskeleton, A60, defined by a monoclonal antibody. J. Cell Sci. 94, 489500. Rebecchi, M. J. and Scarlata, S. (1998). Pleckstrin homology domains: a common fold with diverse functions. Annu. Rev. Biophys. Biomol. Struct. 27, 503-528. Sahr, K. E., Laurila, P., Kotula, L., Scarpa, A. L., Coupal, E., Leto, T. L., Linnenbach, A. J., Winkelmann, J. C., Speicher, D. W., Marchesi, V. T. and Forget, B. G. (1990). The complete cDNA and polypeptide sequences of human erythroid alpha spectrin. J. Biol. Chem. 265, 4434-4443. Sakaguchi, G., Orita, S., Naito, A., Maeda, M., Igarashi, H., Sasaki, T. and Takai, Y. (1998). A novel brain-specific isoform of beta spectrin: isolation and its interaction with Munc13. Biochem. Biophys. Res. Commun. 248, 846-851. Sambrook, J., Fritsch, E. F. and Maniatis, T. (1989). Molecular Cloning. A Laboratory Manual. Cold Spring Harbor Laboratory Press, NY. Second edition. Stankewich, M. C., Tse, W. T., Peters, L. L., Ch’ng, Y., John, K. M., Stabach, P. R., Devarajan, P., Morrow, J. S. and Lux, S. E. (1998). A widely expressed betaIII spectrin associated with Golgi and cytoplasmic vesicles. Proc. Natl. Acad. Sci. USA 95, 14158-14163. Viel, A. and Branton, D. (1996). Spectrin: on the path from structure to function. Curr. Opin. Cell Biol. 8, 49-55. Wechsler, A. and Teichberg, V. I. (1998). Brain spectrin binding to the NMDA receptor is regulated by phosphorylation, calcium and calmodulin. EMBO J. 17, 3931-3939. Winkelmann, J. C., Costa, F. F., Linzie, B. L. and Forget, B. G. (1990a). Beta spectrin in human skeletal muscle. Tissue-specific differential processing of 3′ beta spectrin pre-mRNA generates a beta spectrin isoform with a unique carboxyl terminus. J. Biol. Chem. 265, 20449-20454. Winkelmann, J. C., Costa, F. F., Linzie, B. L. and Forget, B. G. (1990b). β-spectrin in human skeletal muscle – tissue-specific differential processing of 3′ β-spectrin pre-messenger RNA generates a β-spectrin isoform with a unique carboxyl terminus. J. Biol. Chem. 265, 20449-20454. Winkelmann, J. C. and Forget, B. G. (1993). Erythroid and nonerythroid spectrins. Blood 81, 3173-3185. Winkelmann, J. C., Leto, T. L., Watkins, P. C., Eddy, R., Shows, T. B., Linnenbach, A. J., Sahr, K. E., Kathuria, N., Marchesi, V. T. and Forget, B. G. (1988). Molecular cloning of the cDNA for human erythrocyte betaspectrin. Blood 72, 328-334. Wood, S. J. and Slater, C. R. (1998). beta-Spectrin is colocalized with both voltage-gated sodium channels and ankyrinG at the adult rat neuromuscular junction. J. Cell Biol. 140, 675-684. Zhou, D., Ursitti, J. A. and Bloch, R. J. (1998). Developmental expression of spectrins in rat skeletal muscle. Mol. Biol. Cell 9, 47-61.