3 Non-neuronal 210x10 Mr microtubule-associated protein (MAP4) contains a domain homologous to the microtubule-binding domains of neuronal MAP2 and tau

STEVEN J. CHAPIN

Department of Biology, University of California, Los Angeles, CA 90024, USA and JEANNETTE CHLOE BULINSKI

Department of Anatomy & Cell Biology, Columbia University, College of Physicians & Surgeons, 630 W. 168th St, New York, NY 10032, USA 'Address for correspondence to either author: Department of Anatomy & Cell Biology, BB-1213, Columbia University, College of Physicians & Surgeons, 630 W. 168th St, New York, NY 10032, USA

Summary A polyclonal antiserum raised against a HeLa cell amino acid repeats; this region is homologous to a microtubule-associated protein of Mr 210 000 (210 kD motif present in the microtubule (MT)-binding MAP or MAP4), an abundant non-neuronal MAP, domain of two prominent neuronal MAPs, MAP2 and was used to isolate cDNA clones encoding MAP4 tau. The pMAP4.245 sequence also encoded a series from a human fetal brain A gtll cDNA expression of unrelated repeats, located in the MAP's projection library. The largest of these clones, pMAP4.245, domain, N-terminal to the MT-binding domain. contains an insert of 4.1kb and encodes a 245 kD MAP4.245 fusion proteins bound to MTs in vitro, /J-galactosidase fusion protein. Evidence that while fusion proteins that contained only the projec- pMAP4.245 encodes MAP4 sequences includes tion domain repeats failed to bind specifically to immunoabsorption of MAP4 antibodies with the MTs. Thus, the major human non-neuronal MAP pMAP4.245 fusion protein, as well as identity of resembles two neuronal MAPs in its MT-binding protein sequences obtained from HeLa 210 kD MAP4 domain, while most of the molecule has sequences, with amino acid sequences encoded by pMAP4.245. and presumably functions, distinct from those of the The MAP4.245 cDNA hybridizes to several large neuronal MAPs. (~6-9 kb) transcripts on Northern blots of HeLa cell RNA. DNA sequencing of overlapping MAP4 cDNA clones revealed a long open reading frame contain- Key words: cDNA cloning, repeats, fusion proteins, ing a C-terminal region with three imperfect 18- heterogeneity.

Introduction and Borisy, 1975; Weingarten et al. 1975), a number of neuronal structural MAPs have been identified and Microtubule-associated proteins (MAPs) are non-tubulin extensively characterized, owing largely to their abun- proteins that interact with or are components of micro- dance and their presumed importance in MT function tubules (MTs). On the basis of our present knowledge of within the nervous system (reviewed by Vallee, 1990; and their presumed functions with MTs, MAPs can be divided Lee, 1990). For example, it is known that all of these MAPs into three categories: motor proteins, modifying enzymes are fibrous proteins and that all are extensively phos- and structural proteins. The first group includes kinesin phorylated in vivo. Three of these protein species, MAPI, and related proteins, flagellar and cytoplasmic dyneins, MAP2 and tau, each exhibit functionally independent and dynamin (reviewed by Vale and Goldstein, 1990). The domains; one domain binds to the MT wall and stimulates second group includes enzymes that post-translationally assembly, while the other projects from the MT surface. modify tubulin (reviewed by Greer and Rosenbaum, 1989). Cloning and sequencing of the neuron-specific MAPs, The third group is defined as proteins that copurify with MAP2 and tau, have revealed that both contain, near their MTs in self-assembly schemes, stimulate polymerization respective C termini, a shared MT-binding domain of and bind to MTs in vitro, colocalize to MTs in vivo, and do containing several 18-amino acid imperfect repeats (Lewis not exhibit activities appropriate for MAPs of groups one et al. 1988; Lee et al. 1988, 1989; Himmler et al. 1989). or two. For want of detailed information concerning the MAP2 immunolocalizes specifically to dendrites in neur- function of these MAPs, they have been dubbed structural onal cells (Huber and Matus, 1984; DeCamilli et al. 1984), MAPs. Since the original identification of proteins that while at least some tau forms are confined to axonal copurified with brain tubulin (Sloboda et al. 1975; Murphy locations (Papasozomenos and Binder, 1987). Neuronal Journal of Cell Science 98, 27-36 (1991) Printed in Great Britain © The Company of Biologists Limited 1991 27 structural MAPs are either absent or are present in trace escence (Bulinski and Gundersen, 1986), as well as to screen a amounts outside of the nervous system (Weatherbee et al. human fetal brain A gt-11 cDNA library that was kindly donated 1982; Bloom et al. 19$4; Olmsted, 1986). The restricted to us by Dr Rachael Neve, of University of California, Irvine distribution of these MAPs suggests that they perform a (Neve et al. 1986). Immunoreactive clones were subcloned into the phagemid vector pGEM 7ZfX+) (Promega, Madison, WI) for specialized function in the nervous system. Indeed, the restriction mapping and sequencing. unique morphology of neurons implies the need for a unique cytoskeleton. Therefore, extrapolation of the properties and functions of structural MAPs that predomi- Immunoabsorption of MAP4 antiserum with fusion nate in neuronal cells to proliferating cells of non- proteins neuronal tissues is not likely to provide an accurate Immunoabsorption of MAP antisera with A gt-11 fusion proteins was performed following a modification of the procedure of Lewis description of MAP or MT function. 4 et al. (1986); 7.6xlO bacteriophage from plaque-purified clones Isolation and characterization of MAPs from non- expressing putative MAP4 regions or from a random subset of the neuronal cells and tissues have proven to be less tractable human fetal brain library were plated onto Escherichia coli Y1090 than the study of neuronal MAPs. Purification of non- and overlayed with a nitrocellulose filter containing 10 mM neuronal MTs and identification of non-neuronal MAPs isopropyl /^D-thiogalactopyranoside for 2h at 37 °C to induce were first accomplished from HeLa cells (Bulinski and expression of fusion protein. The filters were removed, rinsed in Borisy, 1979; Weatherbee et al. 1980). The MAPs these 50 mM Tris-HCl, pH7.5, 150 mM NaCl, 0.05% Tween 20 (TBST), 3 3 and non-specific binding was inhibited by incubation in a blocking groups identified, 210xl0 Mr and 125xlO Mr species, solution (TBST containing 0.5% gelatin, 10% horse serum and copurified with tubulin in reversible assembly schemes, 5 % normal rabbit serum). Filters were then incubated for 2 h in stimulated the polymerization of tubulin in vitro (Bulinski 5 ml of a 1/1000 dilution of polyclonal MAP4 antibody. The and Borisy, 1980a) and colocalized with MTs both in vitro solution containing unbound antibody was then removed, the and in vivo as judged by both immunofiuorescence filters were washed, and bound antibody was eluted as described (Bulinski and Borisy, 19806) and immunoelectron mi- by Lewis et al. (1986). Antibody fractions obtained using putative croscopy (DeBrabander et al. 1981). Thus, these predomi- MAP4 cDNA clones and from random A clones were assayed by nant HeLa cell MAPs satisfy the criteria outlined above immunostaining Western blots and cultured cells as described by Bulinski and Gundersen (1986). for structural MAPs. The species originally named HeLa 210 kD MAP, but later christened human MAP4, is a component of most, but DNA sequencing not all, cytoplasmic and spindle MTs in cultured primate Overlapping subclones for DNA sequencing were produced using cells (Bulinski and Borisy, 1980c; Chapin and Bulinski, exonuclease III and mung bean nuclease (Henikoff, 1984) and by unpublished data). MAP4 was found to be immunologi- Bubcloning of restriction fragments. Dideoxy-sequencing reac- cally distinct from any known neuronal MAPs; in fact, tions were performed using Sequenase and instructions fromU.S . Biochemicals (Cleveland, Ohio). Difficult sequences were resolved unlike the neuronal MAPs, MAP4 was found to be an by substituting dITP for dGTP in the reactions and by sequencing abundant component of MTs in a variety of primate tissues both strands. and cell types (Bulinski and Borisy, 1980c). MAP4 homologs exist in a variety of organisms (reviewed by Production of fusion proteins and MT binding Olmsted, 1986). Thus, MAP4 most likely represents a experiments nearly ubiquitous class of non-neuronal MAPs and presumably differs both structurally and functionally Lysogens from putative MAP4 cDNA clones were produced in E. coli strain Y1O89, and extracts containing fusion protein were from its neuronal counterparts, MAPI, MAP2 and tau. prepared according to the method of Young and Davis (1983), In order to understand the structure and function of except that bacterial pellets were resuspended in MT assembly human MAP4, as well as its relationships to putative buffer containing 0.2 mM phenylmethylsulfonyl fluoride (PMSF; interspecific MAP4 homologs and to the neuronal MAPs, Chapin and Bulinski, 1990) prior to lysis. Bacterial extracts were we have isolated cDNA clones encoding most of human treated with lO/igml"1 boiled RNase A in the presence of protease inhibitors (0.2mM PMSF, lOj/gml"1 leupeptin, and MAP4 and have begun characterization of MAP4 at the 1 molecular level. Here we report that human MAP4, which lO/zgrnl" tosyl arginine methyl ester) for 20min at 37 °C, and centrifuged to remove aggregates (7min, 205 000 # at 4°C). functions in the dynamic MT cytoskeleton of proliferating Extracts (40 ;<1) were combined with assembly buffer containing cells, contains a MT-binding domain homologous to the 20/(M taxol, with or without 30 j/g taxol-stabilized MTs (prepared MT-binding domain of the neuron-specific MAPs, MAP2 from DEAE-purified bovine brain tubulin) in a final volume of and tau. 50/d, and were incubated at 37°C for 15 mm. To test the specificity of binding, an additional incubation of extract and MTs was carried out in the presence of 0.35 M NaCl. Samples were then Materials and methods centrifuged (7min, 205 000 #, at 25 °C) and equal fractions of supernatants and pellets were analyzed by SDS-PAGE and Materials immunoblotting. Coomassie Blue staining of gels verified that Except as noted, all reagents were purchased from Sigma tubulin had not been significantly depolymerized or degraded Chemical Co. (St Louis, MO), or from Fisher Scientific (Tustdn, during the incubations (not shown). CA). Immunochemicals were from Organon Teknika (West Chester, PA). Restriction enzymes were from Promega Biotech Peptide sequencing (Madison, Wl), or New England BioLabs (Beverly, MA). Radio- Proteolytic digestion of gel bands containing 210 kD MAP4 was chemicals were purchased from New England Nuclear (Boston, accomplished with Staphylococcus aureus V8 protease, according MA). to the method of Cleveland et al. (1977). Briefly, a heat-stable MAP fraction was electrophoresed, and the 210 kD band was Preparation of MAP4 protein and antisera and isolation excised and re-electrophoresed in the presence of V8 protease. ofMAP4 cDNA clones Peptides were analyzed by Coomassie Brilliant Blue staining of HeLa 210 kD MAP4 was prepared (Chapin and Bulinski, 1991) gels and by immunoblotting (Bulinski and Gundersen, 1986). and used to elicit rabbit antibodies (Bulinski and Borisy, 19806). Peptides that were immunoreactive with 210 kD MAP antibodies MAP4 antisera were used for immunoblotting and immunofluor- were subjected to gas-phase sequence analysis, following transfer

28 •S. J. Chapin and J. C. Bulinski to Immobilon membranes according to the method of Matsudaira electropheretograms of putative MAP4 fusion proteins (1987). revealed that they shared major proteolytic fragments (e.g. see Fig. IB). This is because the clones share terminal Northern blotting 5' EcoBl sites that are internal to the MAP4 coding region Total HeLa cell RNA was isolated by the guanidine thiocyanate (see Fig. 3, below). procedure (Cathala et al. 1983). RNA was electrophoresed on 0.8 % formaldehyde-agarose gels, transferred to a Genescreen mem- brane (New England Nuclear, Boston, MA), and hybridized with Verification of identity of MAP4 cDNA clones pMAP4.245 insert, which had been 32P-labelled using a random- Reactivity of fusion proteins with the MAP4 antiserum, primed DNA labelling kit obtained from Boehringer-Mannheim which reacts specifically with MAP4 and the immunologi- (Indianapolis, IN). cally related 255 kD MAP, but not with the neuronal MAPs (Fig. 1), was our initial criterion for identification of Results these clones as bona fide MAP4 cDNAs. To provide further evidence that our clones indeed encode epitopes shared Isolation of MAP4 cDNA clones with authentic human MAP4, we affinity-purified our We purified heat-stable MAPs from HeLa taxol-stabilized polyclonal MAP4 antiserum with our largest fusion MTs and prepared polyclonal antisera to 210 kD MAP4, protein (encoded by clone pMAP4.245) and with fusion and we used these antisera to screen a A gt-11 cDNA proteins produced by random clones from the human fetal library prepared from human fetal brain tissue (Neve et al. brain library. The antibody fraction that did not bind 1986). Five strongly immunoreactive clones were isolated (Unbound) and the affinity-purified (Bound/eluted) anti- from a total of 580 000 clones plated; each clone selected body fraction were each used to probe Western blots was used to prepare lysogens. Electropheretograms and containing a HeLa heat-stable MAP fraction (Fig. 2A). Western blots of the extracts prepared from two of these Antibody purified using clone pMAP4.245 immunolabeled lysogens are shown in Fig. 1, lanes a and b. A major both 210 kD and 255 kD MAP species, while the unbound component of each extract was a fusion protein immuno- fraction was depleted of reactivity with both species. reactive with both MAP4 and /J-Gal antibodies. In each Furthermore, when the affinity-purified antibody was extract, other immunoreactive proteins of lower MT were used in immunofluorescence of African green monkey also present; the latter presumably resulted from proteol- kidney (TC-7) cells, MT labeling was observed, while the ysis of the MAP4 portion of each fusion protein. Note that unbound fraction was depleted of its immunoreactivity the putative proteolytic fragments reacted with /5-Gal with MTs (Fig. 2B). In contrast, fusion proteins from antibody, and that a prominent band reactive only with random clones failed to absorb MAP4 immunoreactivity /3-Gal antibody, which exhibited approximately the Mr (Fig. 2A). These experiments demonstrate that the protein predicted for the j3-Gal portion of the fusion protein encoded by pMAP4.245 shares one or more epitopes with (100 kD), is apparent in both electropheretograms shown the 210 kD and 255 kD HeLa MAPs. Several lines of in Fig. 1 (see arrow in lanes a and b). A close comparison of evidence suggest that multiple MAP4 epitopes are B Coomassie -•MAP4 ••B-Gal -• • i i Fig. 1. Coomassie Brilliant Blue-stained gel (Coomassie) and 123abc 1 2 3a be abc anti-MAP4 (> MAP4), and anti- /3-gal O /3-gal) immunoblots of HeLa extract (lanes 1), porcine brain MAP fraction (lanes 2), 205- heat-stable HeLa MAP fraction Ganes 3), and extracts of bacterial lysogens (lanes a, b). Lanes a, pMAP4.245; b, 116- pMAP4.220; c, control extract of nonlysogenic bacteria. 97- (A) Proteins resolved on a 7.5% gel. MT standards are shown at left (xl(T3). Filled arrowhead, 66- MT of /3-Gal portion of each fusion protein. T indicates T- electrophoretic position of tubulin polypeptides. The numerical suffix in the name of each cDNA clone indicates the 45- Mr of the fusion protein it encodes. Open arrows at right show positions corresponding to 3 r 29- Mr 245xl0 and Mr 170xl0 in A and B. (B) Anti-MAP4 immunoblot of: lane a, pMAP4.245; and lane b, pMAP4.220 resolved on a 5 % gel*

MT-binding domain of MAP4 29 PMAP4.245 Random clones Bound Bound Unbound (eluted) Unbound (eluted)

255- 210- I

B pMAP4.245 Bound Fig. 2. Analysis of anti-MAP4 antibody affinity-purified using A Unbound (eluted) gt-11 fusion proteins. (A) HeLa heat-stable MAP fraction was blotted and immunostained with antibody Unbound or Bound/ eluted from nitrocellulose filters containing fusion protein of A gt- 11 clone pMAP4.245 or random clones from the human fetal brain cDNA library. Positions of 210 kD MAP4 and 266 kD MAP are shown at left. (B) Immunofluorescence staining pattern of monkey kidney (TC-7) cells. MAP4 antibody Unbound or Bound/eluted from a nitrocellulose filter containing A clone MAP4.245 fusion protein. Affinity-purification was carried out as described in Materials and methods. Bar, 10//m.

encoded by pMAP4.245. First, the MAP4 antiserum several approaches, we have established that pMAP4.245 appears to be truly polyclonal, because it reacted with a is a cDNA clone partially encoding MAP4. majority of peptides from a V8 digest of 210 kD MAP4 (not shown). Furthermore, since at least two of the immuno- Restriction mapping and sequencing of MAP4 cDNA reactive peptides that we sequenced (see below) are not clones included in our shortest fusion proteins, the proteolytic Fig. 3 shows the restriction map of clones pMAP4.245 and peptides and the short fusion proteins represent at least pMAP4.220. pMAP4.245 contains an open reading frame two non-overlapping, if not distinct, MAP4 epitopes, both of 2946 nucleotides (encoding 982 amino acids). On the of which are encoded by pMAP4.245. Regardless of how basis of the MT of the fusion protein, and the anomolous many epitopes the MAP4 antibody recognizes, pMAP4.246 SDS-PAGE behavior of heat-stable MAPs (Lee et al. 1988; appears to encode most, if not all, of the MAP4 epitopes Lewis et al. 1988), it appears that this clone encodes reactive with the antiserum, since substantial MAP4 approximately two-thirds of the MAP4 sequence. immunoreactivity was removed from the antiserum by pMAP4.220 encodes a truncated fusion protein and its 3' absorbing with pMAP4.245 fusion protein. region (Fig. 3, broken line) is distinct from pMAP4.245. To provide more direct evidence that our cDNA clones We have no independent clones confirming the 3' region encode MAP4 sequences, we obtained amino acid se- pMAP4.220, and the sequence at the position where quences from proteolytic fragments of HeLa MAP4. Two pMAP4.220 diverges from pMAP4.245 is not a sequence peptides, derived from V8 protease digestion of electro- typical of an exon/intron boundary (Padgett etal. 1986), so phoretically pure MAP4, contained N-terminal sequences it is unlikely that pMAP4.220 represents unprocessed that were found within the amino acid sequence encoded heterogeneous nuclear RNA. Instead, pMAP4.220 may be by pMAP4.245 (Fig. 4, below; boxed residues). Thus, from a cloning artifact that arose by ligation of heterologous

30 S. J. Chapin and J. C. Bulinski PDR PI P2 MT

O_ pMAP4.245

pMAP4.220

1 kb

Fig. 3. Restriction map of pMAP4.245 and pMAP4.220. The open reading frame of pMAP4.245 is denoted by a bold line. The positions of the 18-amino acid repeats of the MT-binding domain (MT) and the projection domain repeats (PDR), and the regions corresponding to peptide sequences obtained from proteolytic fragments of 210kD MAP4 (PI, P2) are indicated with brackets. The 3' end of pMAP4.220, which differs in sequence from that of PMAP4.245 (see text), is indicated by a broken line. The termination codon for the pMAP4.220 fusion protein is indicated (•). Restriction sites shown are Pstl (V), BamHl (T), SacI (•), and EcoRl (O).

E F LEEKMAYQ ETPNSOHUPE DTNFCFQPEO VVDPISTDPF 40

K M YHDDDLAO LVFPSSATAD TtlFAGQNDP LKDtYQNSPC 80

II T AVVPQGUS VEALNIPHIE SFVSPEAVAE PPOPTAVPLE 120

I A K E I E M A « E EDPPAOALE1 HHGLCT TDWA PSKETEHALA 160

K 0 MALATKTE VALAKDMESP TKLDVTIAKD HQPSHE3DHA 200

L V KOHELPTE (EVALVKDVR WPTETOVSSA tMVVLPTEIE 240

V A PAKDVTIL. KETE I A I P H D L A P J K D H G P P KEN T E 280 -II IR A tP I itl M D L A PtKDHGPPICE U t [ V P A r D L V L L S E I EVA 0 A 320 —I N D I I t t T E 1 t SAEKVAL88E T E V A L A L P P E T 360 K 0 KALPIEAE VAPVKOMAQL PETEIAPAKD VAPITVKEVG 400

(DHSPLSE TEHALGKDVT PPPEIEVVLI VCLPPEHE UO

LTEDQVPA LKTEAPLAKD 480

ME1AQTQKGI 320

S P E T I T 0 T GKKCSLPAEE 560

E T » G I A K P EEGRPVVtGT 600

TITSKAKTQP 640 680

720

R S P S T L L 760

TSTtSHKIC TTTLSGTAPA 800

S T T P R L S 840

N I K H 0 P G G GRVQIQKtKV 830

0 V K I E $ 0 K LHFKEKAQAC 920

SEAPLCPG PPACEEPAIS 960

GGGDOItEA OTLDSOIOET 982

Fig. 4. Derived amino acid sequence of clone pMAP4.245. The amino acid sequence encoded by clone pMAP4.245 is indicated in single-letter code. Numbering starts with the first amino acid encoded by this clone. A region with strong homology to MAP2 and tau is indicated by broken underlines (---) with the three 18-amino acid repeats indicated by bold underlines (^-H. Seventeen 14- amino acid repeats and a set of two 26-amino acid repeats are indicated by thin underlines ( ) and broken underlines ( ), respectively. Amino acids that correspond to two peptide sequences obtained from HeLa 210 kD MAP4 are enclosed in boxes. Four potential cell-cycle-specific phosphorylation sites (see Discussion) are indicated by brackets. sequences to MAP4-encoding sequences during construc- similar to the MT-binding motif reported within the tion of the cDNA library. Regardless of its origin, sequences of the neuronal MAPs, MAP2 and tau (Lewis et pMAP4.220 encodes a fusion protein ideally suited for our al. 1988; Lee et al. 1988; Himmler et al. 1989; Lee et al. investigations of presumptive MAP4 MT-binding domains 1989). Moreover, the first repeat is virtually identical to a (see below). short sequence that was identified by protein sequencing The derived amino acid sequence of pMAP4.245 is of a MT-binding fragment of a 190 kD MAP derived from shown in Fig. 4. The most provocative region of the MAP4 bovine adrenal medulla (Aizawa et al. 1989). In MAP4 the sequence is an 89-amino acid region (indicated by small distance from the end of the repeated motif to the C broken underline in Fig. 4) containing three 18-amino terminus (68 residues) is similar to that in MAP2 and acid imperfect repeats (indicated by bold underline in some forms of tau (73 residues). MAP2 and tau share Fig. 4). As demonstrated in Fig. 5, this region is strikingly further homology with each other outside of this 89-amino MT-binding domain of MAP4 31 190 kD _ N V GSTENIKHQPGGGR MAP 4 L J 101 A P D L K N V GSTENIKHQPGGG tau Q T A P V I M P D L K N V G S T E N [I] K H Q P G G G IK MAP 2 R L I fhT] Q | L P D L K N v s TIDIN I K[7IQ pnriG Gla

MAP 4 Q K L F K ,B K A Q ft K y G S L D N GfiriLrpiAlG GlA tau E K L D F K IP R V 0 K |i G S L D N THVPGGG N| MAP 2 V K L D F K E K A 9 A K V G S L D N H |H V P G G G N |

MAP 4 K G S E A P P G P P A G E E P A JTls E A tau I R E D H G A E IV y K MAP 2 I V K R E v[D H G A E I ll T Q Fig. 5. Homology between human MAP4, MAP2, tau and bovine adrenal 190 kD MAP. A segment of MAP4 sequence is aligned with homologous sequences in munne tau and murine MAP2, corresponding to amino acid positions 182-300 for tau (Lee et al. 1988) and 1669-1787 for MAP2 (Lewis et al. 1988) and a sequence obtained from the MT-binding domain of bovine adrenal 190 kD MAP (Aizawa et al. 1989). Amino acid identities are enclosed in boxes, and the three 18-amino acid repeats are indicated with bold underlining. Note the region following the third repeat, which is homologous only between tau and MAP2. This region includes a stretch of 14 amino acids that is similar to the sequences intervening between the repeats, and a stretch of five amino acids that is similar to the first five amino acids of repeat number three. acid MT-binding motif; some of this homology is shown in domain repeats, as well as the repeat-bearing motif Fig. 5. Note that the sequence that follows the third homologous to MAP2 and tau, might constitute MT- repeats of the MT-binding domains of MAP2 and tau is binding domains. similar to the sequence intervening between repeats one and two and between repeats two and three. However, this sequence is not similar to the analogous region within the Microtubule binding assay of MAP4 fusion proteins MAP4 sequence. In fact, the 89-amino acid motif is the In order to address the function of MAP4's two repeat only region in the portion of MAP4 that we have cloned domains, we made use of two of our cDNA clones, that bears obvious homology to MAP2 and tau. pMAP4.246 and pMAP4.220. Because our pMAP4.246 Another potentially interesting feature of the sequence cDNA encodes a repeated motif homologous to that of the of pMAP4.245 is a series of repeats situated closer to the N MT-binding domain of MAP2 and tau, we predicted that terminus than the presumptive MT-binding repeats its fusion protein would be capable of binding to MTs. In (Fig. 4). These repeats, which we have called the projec- contrast, clone pMAP4.220 encodes a fusion protein that tion domain repeats (see Discussion), are not related to the shares with pMAP4.246 its first 556 amino acids, contains presumptive MT-binding repeats described above; instead, six additional amino acids (LFLQQN) at its C terminus, these consist of two sets of 14-amino acid imperfect repeats and lacks the C-terminal 426 amino acids of pMAP4.245. (Fig. 4, thin underline) separated by a set of two 26-amino Thus, both fusion proteins contain the projection domain acid perfect repeats (Fig. 4, broken underline). In Fig. 6 repeats, while pMAP4.220 lacks the region with homology these repeats are aligned to show their homology, with a to MAP2 and tau. consensus sequence for the 14-amino acid repeats indi- Accordingly, extracts from lysogenic bacteria express- cated at the bottom. Since repeats are contained in MT- ing pMAP4.245 and pMAP4.220 fusion proteins were binding domains of all structural MAPs studied to date prepared and tested for binding to taxol-stabilized MTs. (Vallee, 1990), we wished to determine if the projection Fig. 7 shows the results of such an experiment in which, as

T D H A P S K E T E M A L A K D M A L A T K T E V A L A K D M E S P T K L D V T L A KDHQ P S M E S DMA L V K D M E LPIEK E V A L V K D V R W P T E T D V S S A K N V V L P T E T E V A P A K D V T L L K E T E R A S P I K M D L A PSKDHGPPKENK K E T E R A S P I K H D L A PSKDMGPPKENK I V - P A K D L V L L S E I E V A Q A N D I I S S T E I - - S S A E K V A L S S E T E V ALA R D M T L P P E T N V HI I K D K A L P L E A EVA P V Fig. 6. Projection domain repeats of K D M A QLPEI E I A P A MAP4. A region of MAP4 (amino acid K D V A P S T V K E V G L L residues 147-404) is shown with the 14- K D M S P L S E T E M A L G amino acid repeats (indicated by thin K D V T P P P E T E V V L I underline in Fig. 4) aligned to show K N V C L P P E M EVA L T homology. A consensus sequence for these repeats is shown at the bottom. The 26- amino acid repeats are aligned with each Consensus: K D M X L P X E T E V A L A other. V 32 S. J. Chapin and J. C. Bulinski B ••B-Gal Tubulin r^++ - + + I NaCI -_ + __ + 1S 1P2S 2P3S 3P4S4P5S 5P 6S 6P 2S2P

Fig. 7. MT-binding assay of MAP4 fusion proteins. Extracts from bacterial lysogens of pMAP4.245 (lanes 1-3) or pMAP4.220 (lanes 4-6) were incubated in MT assembly buffer alone (lanes 1 and 4), assembly buffer containing taxol-stabilized microtubules (lanes 2 and 5), or assembly buffer containing taxol- stabilized microtubules and 1 0.35 M NaCI (lanes 3 and 6). The mixtures were centrifuged and supernatants (S) and pellets (P) were analyzed by SDS-PAGE and immunoblotting with MAP4 antibody (>• MAP4) or /S-Gal antibody (> /3-Gal). Presence (+) or absence (-) of MTs (tubulin) or NaCI in the incubations is indicated at the top of each electropheretogram. Closed arrowhead indicates fragments 66- that bind to MTs and open arrowhead indicates fragments that do not bind to MTs (see text for details). The positions of MT standards (xlO ), are shown at the left. expected, the full-length pMAP4.245 fusion protein (filled SDS-PAGE (e.g. see Bulinski and Borisy, 1980a,6). We do arrowhead) bound to and sedimented with MTs quantitat- not yet know if the multiple transcripts arise from ively. This binding was specific, resembling the known differential RNA splicing or from transcription of multiple ionic interaction of MAPs with MTs (Vallee, 1982) as genes. However, both tau and MAP2 mRNAs are subject to judged by salt inhibition of the binding of pMAP4.245 to alternate splicing, which gives rise to multiple hetero- MTs. Minimally proteolyzed fragments of pMAP4.245 also geneous protein products (Himmler, 1989; Garner and exhibited MT-binding activity, while shorter, Matus, 1988). The fact that one of our clones contains an /J-Gal-immunoreactive fragments, which presumably insert resembling the insert found in an alternatively lacked C-terminal segments, did not. Note that while spliced tau mRNA (not shown) suggests that similar 3 major fragments of Mr ~220-230 x 10 (open arrowhead) mechanisms may generate diversity in MAP4. were detectable in MT pellets, these species were far more abundant in the supernatants, and their sedimentation with MTs was not inhibited by salt. In addition, the Discussion pMAP4.220 fusion protein failed to sediment with MTs in a salt-sensitive manner. These results demonstrate that In this paper, we have reported the isolation and the C-terminal region is a MAP4 MT-binding domain, and characterization of cDNA clones encoding most of the that this domain is the sole MT-binding domain present in human non-neuronal MAP, MAP4. Two types of exper- the portion of MAP4 we have cloned. imental data support the identity of our MAP4 clones: (1) the fusion proteins they encode share at least two non- overlapping epitopes with MAP4 protein purified from Northern analysis of MAP4 RNA from HeLa cells HeLa cells; and (2) two peptide sequences obtained from HeLa cell RNA was analyzed by Northern blotting to HeLa MAP4 verify the amino acid sequence deduced from determine the size of full-length MAP4 mRNA. The these clones. pMAP4.245 probe hybridized to multiple bands; hetero- We determined the nucleotide sequence of our longest geneous bands centered around 6 kb, and one or more MAP4 cDNA clone, pMAP4.245, obtaining the intriguing bands of 9-10 kb (Fig. 8). Each of these transcripts is result that the derived amino acid sequence of a portion of sufficiently large to encode MAP4. These RNA species are the MAP4 molecule is homologous to the MT-binding reasonably abundant, as we were able to detect them in domain of two neuron-specific MAPs, MAP2 and tau electropheretograms of 0.5 ng of total cellular RNA. The (Lewis et al. 1988; Lee et al. 1988; Himmler et al. 1989; Lee multiple MAP4 transcripts that we have observed may et al. 1989). The homologous region comprised an 89-amino contribute to the diversity in MAP4 proteins observed by acid domain containing three 18-amino acid repeats; this

MT-binding domain of MAP4 33 shown to contribute to MT binding of MAP2 and tau in in A B vivo transfection assays (Lewis et al. 1990); our data suggest that conservation of the sequence of this region is 1 2 1 2 not essential for the MT-binding and, presumably, assembly-promoting capacities of MAP4 (Fig. 7; Bulinski and Borisy, 1980a,6). Besides the MT-binding repeats, which lie near the C terminus of MAP4, we have noted another series of repeats in the N-terminal half of the molecule (see Figs 4 and 6). Both our MT-binding experiments and the overall nega- tive charge of these repeats make them inappropriate 9.5 — candidates for interaction with the acidic carboxyl ter- 7.5 — minus of tubulin. We have called these N-terminal repeats the projection domain repeats, by analogy with the two- domain structure of other fibrous MAPs. For example, 4.4 — MAP2 and tau have been shown to contain two distinct structural domains; a C-terminal MT-binding domain and an N-terminal projection domain that extends from the wall of the MT (Vallee, 1990; Lee, 1990). Secondary 2.4 — structure predictions suggest that the repetitive portion of the projection domain is composed largely of n--helices interrupted by /3-turns, in contrast to the remainder of the 1.35 — MAP4 molecule, for which little secondary structure can be predicted. The role of the projection domain repeats is unknown. To our knowledge, the only specific interaction yet attribu- Fig. 8. Northern blot of HeLa RNA. Total RNA from HeLa table to the projection domain of any of the MAPs is the in cells was electrophoresed on a 0.8 % formaldehyde-agarose gel, transferred to a Genescreen membrane, and probed with vitro binding of the N-terminal region of MAP2 to the radiolabelled insert from pMAP4.245 as described in Materials regulatory subunit of cyclic AMP-dependent protein and methods. The membrane was washed (with final kinase (Theurkauf and Vallee, 1982), a property not stringency of 0.2xSSC at 68°C). Lanes 1, 0.5/ig; lanes 2, 1.0 /Tau protein: An update on structure and function. Cell (CA 09056) during the initial stages of this work. The authors also Motil. Cytoskel. 15, 199-203. LEE, G., COWAN, N. AND KIRSCHNER, M. (1988). The primary structure thank Drs Ron Liem, Steve Chin and Gregg Gundersen, for advice and heterogeneity of tau protein from mouse brain Science 239, and assistance at each stage of the work, Dr Rachel Neve for 286-288. providing the cDNA library, and Dr Dan Webster and Mr Ricky LEE, G., NEVE, R. AND KOSIK, K. S. (1989). The microtubule binding Sucgang for providing materials and encouragement. domain of tau protein Neuron 2, 1615-1624 LEWIS, S. A., IVANOV, I. E., LEE, G.-H. AND COWAN, N. J. (1990). Organization of microtubules in dendntes and axons is determined by a short hydrophobic zipper in microtubule-associated proteins MAP 2 References and tau. Nature 342, 498-505 (and Lewis S. A. and Cowan, N. J. (correction) (1990) Nature 345, 674). AIZAWA, H., KAWASAKI, H., MUROFUSHI, H , KOTANI, S., SUZUKI, K. AND LEWIS, S. A., VILLASANTE, A , SHKRLJNE, P. AND COWAN, N J. (1986). SAKAI, H. (1989). A common amino acid sequence in 190-kDa Brain specific expression of MAP 2 detected using a cloned cDNA microtubule-associated protein and tau for the promotion of probe. J. Cell Biol 102, 2098-2105. microtubule assembly. J. bwl. Chem 264, 5885-6890. LEWIS, S. A., WANG, D. AND COWAN, N. J (1988). Microtubule- BLOOM, G S., LUCA, F. C. AND VALLBE, R. B. (1984). Widespread cellular associated protein 2 (MAP2) shares a microtubule binding motif with distribution of MAP 1A (microtubule-associated protein 1A) in the tau protein. Science 242, 936-939. mitotic spindle and on interphase microtubules. J. Cell Bwl. 98, MALLER, J. L. (1990). Xenopus oocytes and the biochemistry of cell 331-340. division. Biochemistry 29, 3166-3173. BULJNSKI, J. C. AND BORISY, G. G. (1979). Self-assembly of microtubules MATSUDAIRA, P (1987). Sequence from picomole quantities of proteins in extracts of cultured HeLa cells and the identification of HeLa electroblotted onto polyvinyhdene difluoride membranes J. biol. microtubule-associated proteins. Proc. natn. Acad. Sci. U S.A. 76, Chem 262, 10035-10038. 293-297. MORENO, S. AND NURSE, P. (1990) Substrates for p34edc2: in vivo ventas? BULINSKI, J. C. AND BORISY, G G. (1980a). Microtubule-associated Cell 61, 549-551 proteins from cultured HeLa cells: Analysis of molecular properties MURPHY, D. B AND BORISY, G. G. (1976). Association of high molecular and effects on microtubule polymerization. J bwl. Chem 255, weight proteins with microtubules and their role in microtubule 11570-11576. assembly in vitro. Proc. natn. Acad. Sci. U.S.A. 72, 2696-2700. BULINSKI, J. C. AND BORISY, G. G. (19806) Immunofluorescence NEVE, R L., HARRIS, P., KOSIK, K. S., KURNIT, D. M. AND DONLON, T. A. localization of HeLa cell micotubule-associated proteins on (1986). Identification of cDNA clones for the human microtubule- microtubules in vitro and in vivo. J. Cell Bwl. 87, 792-801. aasociated protein, tau, and chromosomal localization of the genes for BULINSKI, J. C. AND BORISY, G. G. (1980C). Widespread distribution of a tau and microtubule-associated protein 2. Molec. Brain Res. 1, 210000mol wt microtubule-associated protein in cells and tissues of 271-280. primates. J. Cell Biol. 87, 802-808. OBAR, R. A., DINGUS, J., BAYLKY, H. AND VALLEE, R B. (1989). The RTI BULINSKI, J. C. AND GUNDERSEN, G. G (1986). Preparation of antibodies subunit of cAMP-dependent protein kinase binds to a common amino- reactive with specific regions of cytoskeletal proteins. Meth. Enzym. 134, 453-460. terminal domain in microtubule-associated proteins 2A, 2B, and 2C. Neuron 3, 639-646. CATHALA, G , SAVOURBT, J.-F., MENDEZ, B., WEST, B. L , KARIN, M., MARIAL, J. A. AND BAXTER, J. D. (1983). A method for isolation of OLMSTED, J B. (1986). Microtubule-associated proteins. A. Rev Cell intact, translationally active ribonucleic acid. DNA 2, 329-335. Bwl. 2, 421-467. CHAPIN, S. J. AND BULINSKI, J. C. (1991). Preparation and functional PADGETT, R. A., GRABOWSKI, P. J., KONARSKA, M. M., SEILER, S. AND assay of pure populations of tyrosinated and detryosinated tubulin. SHARP, P. A. (1986). Splicing of messenger RNA precursors. A. Rev. Meth. Enxym. (in press). Bwchem. 55, 1119-1150. CLEVELAND, D. W., FISCHER, S. G , KIRSCHNER, M. W. AND LAEMMLI, U. PAPASOZOMENOS, S. C. AND BINDER, L I (1987). Phosphorylation K. (1977). Peptide mapping by limited proteolysis in sodium dodecyl determines two distinct species of tau in the central nervous system. sulfate and analysis by gel electrophoresis. J. bwl. Chem. 252, Cell MoUl Cytoskel. 8, 210-226. 1102-1106 PARYSEK, L. M., ASNES, C. F. AND OLMSTEAD, J. B. (1984). MAP 4: DAVIS, F. M., TSAO, T. H., FOWLER, S. K. AND RAO, P. N. (1983). Occurrence in mouse tissues. J Cell Biol 99, 1309-1315. Monoclonal antibodies to mitotic cells. Proc. natn Acad. Sci. U.S A SLOBODA, R. D., RUDOLPH, S. A., ROSENBAUM, J. L. AND GREENOART, P 80, 2926-2930. (1975). Cyclic AMP-dependent endogenous phosphorylation of a DEBRABANDER, M., BULINSKI, J. C., GEUENS, G., DEMEY, J. AND BORISY, microtubule-associated protein. Proc. natn. Acad. Sci. U.S.A. 72, G. G. (1981) Immunoelectron microscopic localization of the 210,000 177-181 mol wt microtubule-associated protein in cultured cells of primates. THEURKAUF, W E. AND VALLEE, R. B. (1982) Molecular characterization J. Cell Bwl. 91, 438-445. of the cAMP-dependent protein kinase bound to microtubule- DECAMILLJ, P , MILLER, P. E., NAVOME, F., THEURKAUF, W. E. AND associated protein 2. J. bwl. Chem. 257, 3284-3290. VALLBE, R. B. (1984). Distribution of microtubule-associated protein 2 VALE, R D. AND GOLDSTEIN, L. S. B (1990). One motor, many tails: an in the nervous system of the rat studied by immunofluorescence. expanding repertoire of force-generating enzymes Cell 60, 883-886. Neuroscience 11, 819-846. VALLEE, R. B. (1982). A taxol-dependent procedure for the isolation of GARD, D L. AND KIRSCHNER, M. W (1987). A microtubule-associated microtubules and microtubule-associated proteins J. Cell Bwl 92, protein from Xenopus eggs that specifically promotes assembly at the 436-442. plus-end. J. Cell Biol. 105, 2203-2216. VAII.FR R. B. (1990) Molecular characterization of high molecular GARNER, C. C. AND MATUS, A (1988). Different forms of microtubule- weight microtubule-associated proteins: Some answers, many associated protein 2 are encoded by separate mRNA transcripts. J. questions. Cell Motil. Cytoskel. 15, 204-209. Cell Bwl. 106, 779-783. VANDRK, D. D., CENTONZE, V. E., PELOQUIN, J., TOMBES, R. M AND GREER, K. AND ROSENBAUM, J. L. (1989) Post-translational BORISY, G. G. (1991). Proteins of the mammalian mitotic spindle1 modifications of tubulin. In Cell Movement, Vol. 2: Kinesin, Dynein, Phosphorylation and dephosphorylation of MAP4 during mitosis. J. and Microtubule Dynamics (ed. Warner, F. D. and Mclntosh, J. R.), Cell Sci. (in press). pp 47-66 New York: Alan R. Liss, Inc. VBRDE, F., LABBE, J. C, DOREE, M. AND KARSENTI, E. (1990) Regulation HENiKorr, S. (1984). Unidirectional digestion with exonuclease III of microtubule dynamics by cdc2 protein kinase in cell-free extracts of creates targeted breakpoints for DNA sequencing. Gene 28, 351-359. Xenopus eggs. Nature 34, 233-238. HIMMLER, A. (1989). Structure of the bovine tau gene: alternatively WEATHERBEE, J A., LUFTIG, R B AND WEIHING, R. R. (1980). MT-binding domain of MAP4 35 Purification and reconstitution of HeLa microtubules. Biochemistry Note added in proof 19, 4116-4123. WEATHERBBE, J. A , SHERUNE, P., MASCARDO, R N., IZANT, J. G., While this article was under review, Aizawa et al. (1990) 3 LUFTIG, R. B. AND WEIHING, R. R. (1982). Microtubule-associated published a sequence from a bovine adrenal 190xl0 Afr proteins of HeLa cells: Heat stability of the 200,000 mol wt HeLa MAPs and detection of the presence of MAP 2 in HeLa cells extracts MAP (MAPU). The striking similarity between the and cycled microtubules. J. Cell Bwl. 92, 155-163. primary structure of bovine MAPU and human MAP4 YOUNG, R. A. AND DAVIS, R. W. (1983). Efficient isolation of genes by supports the idea that MAPU is actually the bovine using antibody probes. Proc. natn Acad. Sci U.S-A. 80, 1194-1198.27. counterpart of human MAP4 (Aizawa, H., Emori, Y., WEINGAHTEN, M. D , LOCKWOOD, A. H , Hwo, S. AND KIRSCHER, M. W. Murofushi, H., Kawaski, H., Sakai, H. and Suzuki, K. (1975). A protein factor essential for microtubule assembly. Proc. (1990). Molecular cloning of a ubiquitously distributed natn. Acad. Sci. U.S.A. 72, 1858-1862. microtubule-associated protein with Mr 190 000. J. biol. (Received 24 August 1990 - Accepted 26 September 1990) Chem. 265, 13 849-13 855.

36 S. J. Chapin and J. C. Bulinski