The Microtubule Binding Domain of Microtubule-associated Protein MAPIB Contains a Repeated Sequence Motif Unrelated to that of MAP2 and Tau

We report the complete sequence of the microtubule-associated protein MAP1B, deduced from a series of overlapping genomic and cDNA clones. The encoded protein has a predicted molecular mass of 255,534 D and contains two unusual sequences. The first is a highly basic region that includes multiple copies of a short motif of the form KKEE or KKF_~ that are repeated, but not at exact intervals. The second is a set of 12 imperfect repeats, each of 15 amino acids and each spaced by two amino acids. Subcloned fragments spanning these two distinctive regions were expressed as labeled polypeptides by translation in a cell-free system in vitro. These polypeptides were tested for their ability to copurify with unlabeled brain microtubules through successive cycles of polymerization and depolymerization. The peptide corresponding to the region containing the KKEE and KKE~ motifs cycled with brain microtubules, whereas the peptide corresponding to the set of 12 imperfect repeats did not. To define the microtubule binding domain in vivo, full-length and deletion constructs encoding MAP1B were assembled and introduced into cultured cells by transfection. The expression of transfected polypeptides was monitored by indirect immunofluorescence using anti-MAP1B-specific antisera. These experiments showed that the basic region containing the KKEE and KKF_~ motifs is responsible for the interaction between MAP1B and microtubules in vivo. This region bears no sequence relationship to the microtubule binding domains of kinesin, MAP2, or tau. M ICROTUBULES prepared in vitro by successive cycles of assembly and disassembly consist largely of aand/3-tubulin, together with a number of other proteins that are collectively defined as microtubuleassociated proteins (MAPs) ~ (for reviews, see 32, 35, 52). Because neuronal cells are rich in microtubules, the best characterized MAPs are those from brain. Conventionally, these MAPs have been classified according to their molecular mass: the high molecular mass MAPs, MAFIA, MAPIB, MAFIC, and MAP2 (in the range 200-350 kD), and the low molecular mass MAPs, or tau proteins, a heterogeneous group of proteins in the range 35--40 kD (51). MAP3 and MAP4 are less abundant than these; MAP3 is primarily expressed in astroglia and some neurons (21), while MAP4 is expressed in glia, vascular, and some other tissues (37). With the exception of MAP1C, which has been identified as cytoplasmic dynein (38, 49) the function of brain MAPs is essentially unknown, though many have been shown to promote microtuhule assembly upon addition to purified tubulin in vitro. Among the high molecular mass MAPs, MAP1B (7) (which is referred to variously as MAP1.2 [16], MAPI(X) [9],and MAP5 [40]) differs from the others in its high abundance, its prominence both in association with microtubules 1. Abbreviation used in this paper: MAP, microtubule-associated protein. and in soluble form, and the relative inefficiency with which it cycles with tubulin, particularly during the first cycle. Nonetheless, MAP1B is generally regarded as an authentic MAP because it does cocycle with purified tubulin and promote microtubule assembly in vitro (40) and because antisera specific for MAP1B (both polyclonal and monoclonal) identify characteristic microtubule networks in neuronal (and other) cells (6, 12). MAP1B is expressed in the axons and dendrites of neurons, as well as in glia and other cells. Because it is especially prominent in axons during their initial outgrowth (40, 43), it has been suggested that MAP1B plays a role in neurogenesis. Several years ago, we isolated a set of cDNA clones encoding a portion of a MAP1 subspecies (28). Subsequently, an antiserum (3d2) raised against a fusion protein derived from one of these clones was shown to uniquely recognize MAPIB on Western blots of brain microtubules and of total protein from PC12 cells (1). The same antiserum also gives a staining pattern identical to that of other anti-MAPlB antisera on brain sections (A. Matus and N. J. Cowan, unpublished observations), thereby proving that our original set of cDNA clones were derived from MAFIB mRNAs. We have now obtained two genomic clones and further overlapping cDNA clones that together encode the entirety of MAPIB. Here we report the complete sequence of MAP1B, deduced from these clones. The encoded protein has a molecular mass of 255, © The Rockefeller University Press, 0021-9525/89/12/3367/10 $2.00 The Journal of Cell Biology, Volume 109 (No. 6, Pt. 2), Dec. 1989 3367-3376 3367 on July 7, 2017 jcb.rress.org D ow nladed fom 534 D and contains two unusual sequences. One is a set of 12 imperfect repeats that do not occur in any other sequenced protein, each 15 amino acids long, and spaced by 2 amino acids. The other is a highly basic region with many copies of the sequence KKEE and KK~, repeated but not at fixed intervals. We show that it is this latter region which is responsible for the binding of MAPlB to micrombules both in vitro and in vivo. Materials and Methods cDNA and Genomic Clones eDNA from 5-d-old mouse brain mRNA was prepared as described (26), except that half was ligated to Eco RI linkers (Boehringer Mannheim Biochemicals, Inc., Indianapolis, IN) and half to Xba I linkers, and each portion cloned into the vector ~Gem (Promega Biotec, Madison, WI). The resulting libraries were screened (4) as described in the text, using purified eDNA restriction fragments 32p-labeled by nick translation (41). Hybridizing clones were subeloned into pUC, restriction-mapped, and, in some cases, sequenced. Sequencing of the complete set of overlapping clones on both strands was accomplished by subeloning bal31-generated fragments (27) into M13, and using the dideoxy chain terminator method (42). Four overlapping genomic clones containing the 5' end of the MAPlB gene were obtained by screening a mouse genomic library in the EMBIA vector, ldndiy provided by P. WEustachio (New York Medical Center). The cosmid clone C (Fig. 1), including the 6-kb MAPIB exon, was isolated by screening a library (provided by S. Tonegawa, Massachusetts Institute of Technology) with done 141A (Fig. 1). Constructs for Translation, Fusion Proteins, and Transfection Experiments Two constructs encompassing repeated sequences were made for in vitro transcription, translation, and cycling experiments as follows: (a) clone 72 was digested with Xho I, blunt-ended with the Klenow fragment of DNA polymerase I, cut with Bgl II, and the coding fragment (Fig. 1, 5' arrow) cloned into the transcription vector pGEM3 (Promega Biotec) cut with Hinc II and Barn HI; (b) clone 141A was digested with Pst I and Eco RI, and the purified fragment (Fig. 1, 3' arrow) cloned into pGEM3 cut with the same restriction enzymes. For both constructs, an in-frame translational start signal was provided by the pGEM vector. The insert from the first of these pGem clones, including the initiator AUG, was cloned into a pSV vector (34) for expression in cultured cells (Fig. 1, construct X). In addition, constructs were made from the 5' and 3' end of the set of MAPIB cDNAs to place the translational start and stop signals: (a) the insert from clone 12X was subcloned into the Xba I site of pGEM1; and (b) the Sac I/Eco RI fragment was cloned into pGEM3 cut with Arc I, blunt ended with Klenow and then digested with Eco RI; this procedure supplied an AUG codon for translational initiation in frame with the MAPIB coding S~luence. The full coding sequence of MAPIB was assembled, first by joining clones 12X and IR at an internal Ptim I site, then by digesting the resulting construct with Xho I and F, co RI, and finally by adding both the Xho I/Ssp I fragment from cosmid clone C and the Ssp I/F, co RI fragment from clone 166 in a three-way ligation. The clones and cloning sites are depicted in Fig. 1. The full cloning region was subcloned into a pSV vector using a unique Sal I site in the 5' flanking pUC polylinker, and a unique Dra I site in the 3' untranslated region. Deleted versions were obtained by (a) joining a Msc I site to a Pvu I site blunt ended with I"4 DNA polymerase (Fig. 1, 1B-H); (b) deleting an '~l-kb Msc I fragment by self-ligating a gel purified fragment from a partial Msc I digest (Fig. 1, 1B-K); and (c) by joining the Xho I and Bg ]1 sites, by blunt ending each with the Klenow fragment of DNA polymerase I ( F i g , 1, 1B-X). To raise an antiserum specific for the second set of repeats, a Pst I/Eco RI fragment (Fig. 1, 3' arrow) from clone 141A was subcloned into a pATH vector (29) that was used to express a fusion protein whose amino terminal part is Escherichia coil TrpE, and whose carboxyterminal part is derived from MAPIB. Fusion protein excised from SDS polyacrylamide gels was used to immunize rabbits. The resulting serum is called YXY; its specificity is demonstrated in Fig. 7. Transcription, Translation, and Microtubule Cycling Experiments pGEM subclones were linearized at the 3' end and transcribed in vitro using SP6 polymerase as described (33). The capped mRNAs were translated in rabbit reticulocyte lysate (Promnga Biotec) supplemented with either [35S]methionine or [3H]lysine. Translation products were resolved on SDS polyacrylamide gels, which were fluorographed, dried, and exposed to film. Micmtubule cycling experiments were performed as described (30). Hybrid Selection and Primer Extension Clone 5X (Fig. I) was used to select MAPIB-specific mRNA from total mouse brain RNA (100#gm/2.l-cm nitrocellulose filter) by methods described previously (17). The selected RNA was used in a primer extension reaction containing an antisense oligonucleotide, 5'CCTGAGAGAAGTGTTCCTY (corresponding to nucleotides 17-34 of the sequence shown in Fig. 4), 32p-end labeled with polynucleotide kinase. Reaction conditions and analysis of the extended product were as described (3). Transfection Experiments and Immunofluorescence Hela cells were grown in DME supplemented with 5% FCS (Hydone P Pf M X B M M S D V I ! I . . . .~ . . . ) . . . , I I I V I 5 X 8X, 166 6X 34, 137 1R 20 13X 12X 72 32X 2X 141A