Microtubule-associated Protein 1S, a Short and Ubiquitously Expressed Member of the Microtubule-associated Protein 1 Family*

The related high molecular mass microtubule-associated proteins (MAPs) MAP1A and MAP1B are predominantly expressed in the nervous system and are involved in axon guidance and synaptic function. MAP1B is implicated in fragile X mental retardation, giant axonal neuropathy, and ataxia type 1. We report the functional characterization of a novel member of the microtubule-associated protein 1 family, which we termed MAP1S (corresponding to sequence data bank entries for VCY2IP1 and C19ORF5). MAP1S contains the three hallmark domains of the microtubule-associated protein 1 family but hardly any additional sequences. It decorates neuronal microtubules and copurifies with tubulin from brain. MAP1S is synthesized as a precursor protein that is partially cleaved into heavy and light chains in a tissue-specific manner. Heavy and light chains interact to form the MAP1S complex. The light chain binds, bundles, and stabilizes microtubules and binds to actin. The heavy chain appears to regulate light chain activity. In contrast to MAP1A and MAP1B, MAP1S is expressed in a wide range of tissues in addition to neurons and represents the non-neuronal counterpart of this cytolinker family. The classical microtubule-associated proteins (MAPs) 1 MAP1, MAP2, and tau were discovered more than 20 ago the of of all confirmed sequenc-ing and/or reaction of encoded proteins with MAP1S-specific antisera.

The classical microtubule-associated proteins (MAPs) 1 MAP1, MAP2, and tau were discovered more than 20 years ago by virtue of the fact that they copurified with microtubules from vertebrate brain (1). Because of the developmental regulation of expression of these proteins, their restricted localization in the axonal or somato-dendritic compartment of neurons, and the differential phosphorylation depending on the developmental stage and subcellular localization, it was soon proposed that MAPs are important regulators of neuronal microtubules during differentiation.
The MAP1 family has two members: MAP1A and MAP1B. Both proteins are of high molecular mass (ϳ300 kDa, 2500 aa), are expressed predominantly in the nervous system, and consist of several subunits, one heavy chain (HC) and at least one light chain (LC) (1). In each case, heavy and light chains are the products of proteolytic cleavage of a common polyprotein precursor. MAP1A and MAP1B share three substantial regions of sequence homology (2), one in the NH 2 terminus of the heavy chains, one in the COOH terminus of the heavy chains, and one in the COOH-terminal half of the light chains. We termed these homologous hallmark domains of the MAP1 family MH1, MH2, and MH3, respectively. In MAP1B, the MH1 and MH3 domains mediate the interaction between heavy and light chains (3), and the MH3 domain of both proteins contains an actin binding site (4). MAP1A and MAP1B are conserved in vertebrates. A MAP1 ortholog termed Futsch has been identified in Drosophila (5).
MAP1B function has been investigated by gene targeting in the mouse, and the original contention that it is important for neuronal differentiation and development of the nervous system has been confirmed (6 -9). Mice homozygous for hypomorphic or null alleles of MAP1B display defects in axonal guidance, neuronal migration, axon diameter, and myelination. Work on cultured neurons from MAP1B mutant mice indicated that MAP1B contributes to neuronal differentiation and axon extension and guidance by regulating microtubule and actin dynamics in the growth cone (10 -12). Together, these observations suggest that MAP1B and perhaps MAP1A regulate the cytoskeleton in response to extracellular guidance cues to permit axon extension and growth cone turning. This view is supported by the finding that changes in the activity of serinethreonine kinases casein kinase II, glycogen synthase kinase-3␤, and cyclin-dependent kinase 5 lead to altered MAP1B phosphorylation (13)(14)(15) and by the fact that local inactivation of MAP1B on one side of the growth cone induces growth cone turning (16). On the other hand, there is evidence that the function of MAP1B and perhaps MAP1A goes beyond regulation of the neuronal cytoskeleton. MAP1B interacts with and changes the ligand affinity of specific ␥-amino butyric acid receptor subunits (17,18). There is also evidence that a fraction of MAP1B is inserted in the axonal plasma membrane, with its extracellular domains binding to myelin-associated glycoprotein expressed on the surface of glial cells (19).
Here we report the discovery and functional characterization of a third member of the MAP1 family encoded by the genomes of mouse, rat, and human that has thus far escaped detection. Because the novel MAP1 protein is unusually short compared with high molecular mass MAP1A and MAP1B, we propose to term it MAP1S.

cDNA Constructs
All amino acid numbering refers to the murine MAP1S protein sequence denoted in GenBank TM accession number AK032300. The fulllength MAP1S cDNA encoding aa 1-973 was obtained by replacing a 693-base pair BstEII fragment of RIKEN clone ID 6430517J16 * This research was supported by a grant from the Austrian Science Fund (Project No. P14977). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
(GenBank TM accession number BB622636) with the corresponding fragment of IMAGE clone ID 4481035 (GenBank TM accession number BG261707). This was necessary because the RIKEN clone encoded a threonine instead of the wild-type proline at position 680. MAP1S-LC cDNA (aa 755-973) was obtained by PCR using primers 5Ј-GAGAAT-TCACGATGGTGGACCCGGAGGCG-3Ј and 5Ј-ATACCTCGAGCTAGA-ACTCCACCTTGCAGG-3Ј. MAP1S ⌬645-875 cDNA was obtained by deletion of the by BstEII fragment from full-length cDNA. A construct encoding the MH3 domain (human aa 935-1059 equivalent to murine aa 852-973) was obtained by PCR using clone DKFZp434G0516 (Gen-Bank TM accession number AL041174) as template and primers 5Ј-GA-TCGAATTCCCCCGGCCCGGGGTGTCAGC-3Ј and 5Ј-GATCGTCGAC-CTAGAACTCCACCTTGCAGG-3Ј. All constructs were fused in-frame to NH 2 -or COOH-terminal myc tags (aa sequence, EQKLISEEDLN) and cloned into the mammalian Tet-Off expression vector pUHD10 -3 (20). In addition, MAP1S-LC cDNA was cloned into a pET15b (Novagen, Inc., Madison, WI) derivative for the expression of NH 2 -terminal His 6tagged protein in Escherichia coli. A fragment encoding the unique domain of the murine MAP1S heavy chain (aa 430 -495) was amplified using primers 5Ј-GATAGGATCCCGCCTGCTGGATGGGCTACA-3Ј and 5Ј-ACGCGTCGACTCAGGCAGGCTCCCTTCGCACTG-3Ј, cut with BamHI and SalI, and cloned into the bacterial fusion vectors pGEX 4T-1 (Amersham Biosciences) and pMal-c2 (New England Biolabs) to obtain plasmids for the expression of GST and maltose-binding protein fusion proteins (GST-HC and maltose-binding protein-HC, respectively) in E. coli. The authenticity of all constructs was confirmed by sequencing and/or reaction of encoded proteins with MAP1S-specific antisera.
Anti-LC-Anti-LC was used at a concentration of 2.5 g/ml. The serum was raised in rabbits against the synthetic peptide CPLNTT-NPSRSRKAPARP (MAP1S aa 768 -784 linked to an NH 2 -terminal cysteine for convenient linkage to the affinity matrix; Gramsch Laboratories) and affinity-purified as described previously (3).
Coimmunoprecipitation-Whole brains from 3-week-old C57BL/6J mice were homogenized in 1 ml of ice-cold TEN buffer (100 mM Tris-HCl, pH 7.5, 100 mM NaCl, 10 mM EDTA, and 0.1 mM dithiothreitol) containing a protease inhibitor mixture (Roche Diagnostics) with a motor-driven glass/Teflon homogenizer. All steps were carried out at 4°C or on ice. After centrifugation, supernatants were precleared by incubation with 0.2 volume of protein A-Sepharose (Amersham Biosciences) for 3 h. After centrifugation for 10 min at 5000 rpm, the supernatants were incubated overnight with anti-HC (1 g/ml), anti-LC (1 g/ml), or no antibodies (negative control), followed by an additional 3-h incubation with 0.2 volume of protein A-Sepharose. Immune complexes were collected by centrifugation at 1000 rpm, washed three times with TEN buffer containing a freshly added protease inhibitor mixture, and eluted by boiling in 50 l of SDS sample buffer. Samples were analyzed by immunoblotting.

Purification of Microtubule Proteins and Cosedimentation Analysis
Purification of microtubules from 8-day-old C57BL/6J mice by assembly and disassembly cycles was performed as described previously (22) using Pipes as a buffer system, and microtubules were tested for copurification of MAP1S by immunoblot. For in vitro cosedimentation, recombinant MAP1S light chain was expressed in E. coli BL21-Codon-Plus-RIL and purified by affinity chromatography on Ni 2ϩ columns according to the manufacturers' protocols (Novagen, Inc. and Qiagen) in the presence of 6 M urea, which was removed by dialysis before use in the respective assay. Cosedimentation with MAP-free bovine brain tubulin (#TL238; Cytoskeleton, Inc.) or rabbit skeletal muscle actin (Cytoskeleton, Inc.) was carried out as described previously (4), except that 0.1% Tween and 2 mM dithiothreitol were added, and cosedimentation with actin was done in PEM buffer (100 mM Pipes, pH 6.8, 2 mM EGTA, and 1 mM MgCl 2 ). Equal volumes of supernatants and pellets were analyzed by SDS-PAGE and Coomassie Blue staining.

Cell Culture, Transfection, Immunofluorescence Microscopy, and Immunohistochemistry
PtK2 cells and murine neuroblastoma N2a cells were grown as described previously (4). Neurite outgrowth was promoted by incubating N2a cells in Dulbecco's modified Eagle's medium without fetal calf serum supplemented with 0.3 mM dibutyryl-cAMP (Sigma-Aldrich) for 48 h. Transient transfection was performed using Lipofectamine (Invitrogen) according to the manufacturer's protocol. For immunofluorescence microscopy, PtK2 or N2a cells grown on glass coverslips were analyzed as described previously (4). For immunohistochemistry, anesthetized mice were perfused transcardially with 4% paraformaldehyde in phosphate-buffered saline (PBS). Brains were dissected out, sliced sagittally, post-fixed for 1 day in 4% paraformaldehyde in PBS, and cryoprotected in 30% sucrose solutions at 4°C overnight. After freezing in liquid nitrogen, serial coronal cryosections (5 m) of the cerebellum were cut on a cryomicrotome, transferred onto chrome-alum-gelatin-coated glass slides, and dried overnight at room temperature. Slides were pretreated with Peroxidase Block (DAKO, Carpinteria, CA) for 10 min, washed twice in PBS, incubated for 1 h at room temperature with anti-HC, washed twice in PBS, and incubated in biotinylated anti-rabbit IgG (1:500; Vector Laboratories, Burlingame, CA) for 30 min, followed by two washes in PBS. Slides were then incubated in streptavidin/horseradish peroxidase (1:500; DAKO) for 30 min, washed twice in PBS, developed with diaminobenzidine DABϩ chromogen kit (DAKO) according to the manufacturer's instructions, and counterstained with hematoxylin. The slides were then dehydrated, mounted, and examined under a Zeiss Axiophot microscope using Axiovision software.

Rodent and Human Genomes Encode One Additional Member of the MAP1 Family-
The sequence comparison of MAP1A and MAP1B revealed that the two proteins share three extended regions of homology (2): one region of about 500 aa in the NH 2 terminus of the heavy chains (MH1), one region in the COOH terminus of the heavy chains preceding the proteolytic cleavage site of the precursor (MH2; 120 aa), and one region in the COOH-terminal half of the light chains (MH3; 125 aa; Fig.  1). Using the amino acid sequence of MH3 in a BLAST search (23) through available nucleic acid sequence databases, we identified numerous human, mouse, and rat cDNAs and expressed sequence tags. The majority of these sequences encoded either MAP1A or MAP1B MH3 domains. A third group of sequences showed consistent differences from either of the two known MAP1 proteins. Using these sequences in a renewed search on the nucleotide level, we identified full-length human and mouse cDNAs (representative GenBank TM accession numbers, for example, AJ440784 and AK032300, respectively) encoding a novel protein related to MAP1A and MAP1B that we termed MAP1S for reasons mentioned under "Introduction." Analysis of human and mouse genomic sequences revealed that the MAP1S gene, like the genes for MAP1A and MAP1B, consists of seven exons and that exon-intron boundaries are conserved in MAP1A, MAP1B, and MAP1S. The considerable difference in size among the three proteins is due to differences in the length of internal sequences in exon 5, whereas the respective length of the other six exons is very similar or identical in all three genes. The MAP1S gene is located on human chromosome 19p13.12 (LocusID 55201) and on mouse chromosome 8 (LocusID 270058; MGI 2443304). No other MAP1-related genes or sequences were found, indicating that MAP1S is the only other member of the MAP1 family. A cDNA fragment encoding part of MAP1S has been identified previously in a yeast twohybrid screen for potential interaction partners of the fibroblast growth factor-associated protein LRPPRC and was termed C19ORF5 (GenBank TM accession number XP_038600; Ref. 24). In a similar search for potential interaction partners of a testicular protein of unknown function termed VCY2, the full-length cDNA of MAP1S was described (GenBank TM accession number AJ440784; Ref. 25). Neither study provided data on the MAP1S protein, nor was the putative interaction with LRPPRC or VCY2 confirmed.
The amino acid sequence comparison of MAP1S, MAP1B, and MAP1A revealed that MAP1S contains all three hallmark domains of the MAP1 family of proteins (Fig. 1). The MH1 domain appears to be the most highly conserved domain of the three proteins. In MH2 and MH3, MAP1A and MAP1B are more closely related to each other compared with MAP1S. Two additional short stretches of amino acid sequence homology between MAP1B and MAP1S that are not present in MAP1A were found at positions 1568 -1592 in MAP1B corresponding to 615-639 in MAP1S (86% identity) and 2075-2090 in MAP1B corresponding to 663-678 in MAP1S (81% identity). Mouse and human MAP1S amino acid sequences are conserved at 63% identity, but the human protein contains an additional 86 aa encoded by several small insertions, the majority of which are located in human exon 5.

MAP1S Is Widely Expressed in Mouse Tissues and Slightly Up-regulated during Postnatal Brain Development-Reverse
transcription-PCR of a variety of mouse tissues showed that MAP1S mRNA was widely expressed and present in all tissues examined ( Fig. 2A). As judged from this semiquantitative analysis, levels of expression were highest in testis, brain, heart, lung, and kidney, confirming an earlier report (25). We obtained no evidence for alternative splicing by Northern blot analysis (a single 3.2-kb transcript was found; data not shown) or by comprehensive analysis of more than 250 human and mouse expressed sequence tag and cDNA sequences.
Using a specific, affinity-purified antibody against the heavy chain of MAP1S termed anti-HC, we detected MAP1S protein expression in all mouse tissues examined ( Fig. 2B; for specificity of the antibody, see Fig. 3). Remarkably, the heavy chainspecific antibody detected bands of 120 and 100 kDa that we subsequently showed to be the uncleaved full-length polyprotein precursor and the heavy chain, respectively (see Fig. 3B). MAP1S protein levels were highest in testis and brain. The amount of cleaved versus uncleaved form varied. In brain, the majority of the protein was cleaved. Heart, lung, and spleen expressed more of the uncleaved form, and in testis, kidney, and liver, the cleaved and uncleaved forms were found at approximately equal levels.
MAP1S expression during postnatal brain development was compared with the expression of MAP1A and MAP1B (Fig. 2C). In accordance with previously published reports, we found that the expression patterns of MAP1A and MAP1B were reciprocal: MAP1A is expressed at low levels at birth and then increases over the next 3 weeks, whereas MAP1B expression is high at birth, increases slightly over the next few days, and then drops to lower levels (1). The expression pattern of MAP1S resembled that of MAP1A in that there is an increase in MAP1S levels during the postnatal period. However, unlike MAP1A, MAP1S levels were already high at birth.
MAP1S Copurifies with Microtubules from Murine Brain and Is Partially Cleaved into Heavy and Light Chains-We next investigated whether MAP1S copurified with microtubules through repeated cycles of temperature-induced polymerization and depolymerization, as would be expected for a true MAP. For this analysis, we used anti-HC and anti-LC, two affinity-purified antibodies directed against unique amino acid sequences located in the presumptive heavy and light chains of MAP1S (Fig. 1). To demonstrate the specificity of these antibodies, we first conducted an immunoblot analysis of total brain homogenates (Fig.  3A). Anti-HC detected bands at 100 and 120 kDa, and anti-LC detected one band at 26 kDa. The 100-kDa band corresponds to the MAP1S heavy chain (calculated molecular mass, 80 kDa), and the 26-kDa band corresponds to the light chain (calculated molecular mass, 23 kDa). The identity of the remaining 120-kDa band became clear after copurification of MAP1S with microtubules from brain by temperature-dependent microtubule polymerization/depolymerization. After two cycles of polymerization, microtubule pellets contained MAP1S as revealed by immunoblot analysis (Fig. 3B). Treatment of the polymerized microtubules with a high concentration of salt (350 mM) led to the release of MAP1S from microtubules (data not shown), another feature reminiscent of the behavior of classical MAPs. In the immunoblot analysis, anti-HC again detected the band at 120 kDa and the heavy chain at 100 kDa, whereas anti-LC detected the light chain at 26 kDa and an additional band comigrating with the 120-kDa band detected by the heavy chain-specific antibody. This analysis identified the 120-kDa band as the uncleaved MAP1S polyprotein precursor (calculated molecular mass, 103 kDa) and showed that the MAP1S precursor is partially cleaved into heavy and light chains. It further demonstrated the specificity of the antibodies used and showed that MAP1S behaved like a classical MAP in temperature-dependent copurification with microtubules from brain. The failure of anti-LC to detect the precursor in total homogenates (Fig. 3A) is due to the lesser abundance of this protein and the weaker affinity of anti-LC compared with anti-HC. The discrepancies in observed and calculated molecular mass of precursor, heavy chain, and light chain might be due to slightly aberrant migration during electrophoresis and/or as yet unidentified posttranslational modifications.
The MAP1S Light Chain Binds to Microtubules and Actin Filaments in Vitro-Previous studies have shown that the light chains of MAP1A and MAP1B contain binding sites for microtubules and actin filaments (3,4). To investigate whether association of MAP1S with microtubules might be due to binding sites in the light chain, we tested the microtubule binding properties of MAP1S in vitro. A recombinant protein consisting of the light chain fused to six histidines was synthesized in and purified from E. coli. Purified MAP-free tubulin was polymerized in vitro in the presence of purified MAP1S light chain. The MAP1S light chain bound to microtubules and was pelleted after centrifugation (Fig. 4A). In the absence of tubulin or in the presence of a control protein (BSA), the light chain remained in the supernatant, demonstrating that its interaction with microtubules was specific. In an additional control, we ruled out that this interaction was due to the His tag fused to the light chain. Thus, a His-tagged fragment of the protein plectin, which does not contain microtubule or actin binding sites, did not bind to microtubules, and only a trace amount was found in the pellet (Fig. 4A, right panel).
To test for a potential actin binding activity in the light chain, a similar analysis was carried out with actin (Fig. 4B). Again, the MAP1S light chain bound specifically to actin and remained in the supernatant in the absence of actin, and the interaction with actin was not due to the His tag.
The MAP1S Heavy and Light Chains Associate in Vivo-To determine whether heavy and light chains of MAP1S can interact with each other in vivo, we carried out immunoprecipitation experiments using either the MAP1S heavy or light chain antibody and analyzed the resulting precipitates by im- FIG. 2. MAP1S expression is widespread in mouse tissues. A, reverse transcription-PCR analysis was carried out on 5 g of total RNA isolated from the indicated tissues of an adult male mouse using primers corresponding to sequences in exon 1 and 5, and aliquots corresponding to 0.1 g of the original RNA sample were fractionated on an agarose gel. A single amplification product of the predicted size of 493 nucleotides was detected when the reaction was carried out in the presence (ϩ), but not in the absence (Ϫ), of reverse transcriptase, demonstrating that the product was derived from RNA transcripts. sg, salivary gland. B, 120 g of protein (total lysate) from the indicated mouse tissues (postnatal day 17) were analyzed by SDS-PAGE and immunoblotting with an affinity-purified antibody against the heavy chain. Protein bands of 120 (corresponding to the uncleaved full-length MAP1S polyprotein precursor; MAP1S-FL) and 100 kDa (corresponding to the MAP1S heavy chain; MAP1S-HC) were detected in a number of mouse tissues. C, 50 g of brain protein (total lysate) prepared at the indicated days after birth were fractionated by SDS-PAGE and analyzed by immunoblotting with an affinity-purified antibody against the heavy chain, an anti-tubulin antibody (internal loading control), and antibodies directed against the MAP1A and MAP1B light chains. munoblot. When whole cell lysates of brain were precipitated with anti-LC, the precipitates contained not only the light chain but also the heavy chain (Fig. 5). Vice versa, precipitation with anti-HC led to precipitation of the heavy chain and coprecipitation of the light chain. In the absence of any antibody, neither the light chain nor the heavy chain was precipitated (data not shown). Because the heavy chain does not contain its own microtubule binding site (Fig. 7), this result suggested that the heavy chain copurified with microtubules from brain ( Fig.  3) through binding to the microtubule-bound light chain. A potential caveat of this experiment is that anti-LC and anti-HC also precipitate the uncleaved full-length MAP1S. Subsequent proteolytic cleavage of full-length MAP1S at the correct position would mimic non-covalent association of heavy and light chains. However, this is highly unlikely because this proteolytic cleavage would have to occur during or after elution of the immunoprecipitate from the protein A-Sepharose matrix by incubation with a strong denaturing buffer containing 4% SDS and 5% ␤-mercaptoethanol. Moreover, we found that immuno-precipitation of a myc-tagged light chain ectopically expressed in N2a cells from a cDNA construct encoding only the light chain coprecipitates endogenous heavy chain (data not shown).
The Light Chain of MAP1S Contains an NH 2 -terminal Microtubule-binding Domain and Binds, Bundles, and Stabilizes Microtubules in Vivo-To investigate what effects binding of MAP1S might have on microtubules, we ectopically expressed MAP1S in epithelioid PtK2 cells in which we did not detect endogenous MAP1S expression. Ectopic expression of fulllength myc-tagged MAP1S showed that the protein was able to bind to cellular microtubules without inducing recognizable changes in microtubule organization (Fig. 6, AϪC). Treatment of cells with the microtubule depolymerizing drug colchicine led to destruction of microtubules and diffuse localization of tubulin and MAP1S (Fig. 6, DϪF). In contrast, when the MAP1S light chain was expressed, it not only decorated microtubules but also induced the formation of microtubule bundles (Fig. 6, GϪI) that were stabilized and protected from the effects of colchicine (and nocodazole; data not shown), whereas the microtubules were destroyed in neighboring untransfected cells The His-tagged control protein (contr; a fragment of plectin) was not found in the pellet fraction (right panel). Thus, interaction of the light chain with actin was specific and was not due to the His tag. The positions on the gel of BSA, actin, the MAP1S light chain (MAP1S-LC), the plectin fragment (contr), and insufficiently denatured MAP1S light chain dimers and trimers (asterisks) are indicated. (Fig. 6, JϪL). Because microtubule bundling by the light chain was observed already in the absence of colchicine (Fig. 6, GϪI) and microtubule protection was observed only in light chainexpressing cells and not in cells expressing full-length MAP1S or in untransfected cells, we can rule out that the fibrillar structures co-staining with anti-tubulin and anti-myc (Fig. 6, JϪL) are non-microtubule tubulin aggregates induced by treatment with colchicine (26). Instead, these structures represent bona fide microtubule bundles stabilized by the light chain. These results demonstrated that the light chain, in addition to binding to microtubules in vitro, is capable of binding to microtubules in vivo, altering microtubule organization and stability. In addition, the strong effects of the light chain are masked by the presence of the heavy chain (Fig. 6, compare GϪL with AϪF).
To locate the microtubule-binding domain(s) in MAP1S, we expressed a MAP1S mutant lacking aa 645-875. This protein displayed diffuse distribution in the cytoplasm and was not bound to microtubules (Fig. 7, AϪC). Together with our finding that the light chain (comprising aa 755-973) did bind to microtubules, this result showed that MAP1S contains just one microtubule-binding domain and that this binding domain is located in the NH 2 terminus of the light chain spanning aa 755-875. Fig. 4 revealed that the light chain can bind to actin filaments as well as to microtubules. Because the NH 2 terminus of the light chain was shown to contain the microtubule-binding domain, we tested whether the MH3 domain comprising the COOHterminal half of the light chain has actin binding activity. Indeed, when the MH3 domain was ectopically expressed in PtK2 cells, it was found to decorate cellular stress fibers (Fig. 7,  DϪF) and not microtubules. The results shown in Fig. 7, DϪF, were obtained with the human MAP1S MH3 domain. Similar results were obtained with the corresponding mouse MAP1S MH3 domain (data not shown). These results showed that the COOH-terminal MH3 domain has actin binding activity and confirmed that the microtubule-binding domain of the light chain is located in the NH 2 terminus.

The COOH Terminus of the Light Chain Can Interact with Cellular Actin Fibers-The in vitro analysis presented in
Endogenous MAP1S Decorates Microtubules in N2a Cells and Is Found in the Soma and Dendrites of Cerebellar Purkinje Cells-A crucial requirement to establish MAP1S as a bona fide microtubule-associated protein is to demonstrate that endogenous MAP1S is indeed found on microtubules. We chose N2a mouse neuroblastoma cells for this experiment because they can be induced to differentiate in culture and immunoblot analysis revealed that these cells express considerable levels of MAP1S (data not shown). Confocal double immunofluorescence microscopy clearly showed that the microtubules in these cells are decorated with MAP1S (Fig. 8, AϪC). To confirm that MAP1S is also expressed in neurons of the brain, we stained sections of the cerebellum. MAP1S was found to be expressed in dendrites and somata of Purkinje cells (Fig. 8, D and E). Staining of Purkinje cells was not observed when sections were incubated with secondary antibody alone (Fig. 8E). In addition, staining of Purkinje cells by the anti-HC antibody was blocked by preincubation of the antibody with excess antigen, but not by preincubation with an unrelated control antigen (data not shown). These results demonstrated that endogenous MAP1S binds to microtubules in cell bodies and neurites of cultured neuroblastoma cells and is expressed in neurons of the brain.

DISCUSSION
Here we report the discovery and functional characterization of the third and final member of the MAP1 family of proteins, MAP1S. It is somewhat surprising that this protein was not discovered along with other prominent MAPs about three decades ago and has escaped detection despite its structural relatedness to MAP1A and MAP1B, its expression in the brain, and its behavior as a MAP in the classical purification protocol. The main reason for MAP1S having been missed is probably its relatively low level of expression compared with MAP1A, MAP1B, MAP2, and tau.
What makes MAP1S unique among the members of the MAP1 family is its small size. In contrast to the other members of the family, MAP1S hardly contains any extra sequences in addition to the conserved hallmark domains MH1, MH2, and MH3. Most notably, MAP1S does not contain heavy chain domains that were implicated in microtubule binding of MAP1A (27,28) and MAP1B (29). Consistent with this observation, we found only one microtubule-binding domain in MAP1S that is located in the NH 2 terminus of the light chain (Fig. 6), corresponding in position but not in sequence to the microtubulebinding domains of the light chains of MAP1A and MAP1B (3,4). Thus, the light chain microtubule-binding domains might represent the main microtubule-binding domains of MAP1 proteins, with the additional domains found in the heavy chains of MAP1A and MAP1B serving accessory functions. The comparison of the three MAP1 genes further showed that exon-intron boundaries and exon sizes are conserved, with the exception of exon 5, suggesting that all three genes were derived from a common MAP1 ancestor.
Another feature of MAP1 proteins highlighted by the characterization of MAP1S is that the hallmark domains are conserved not only in sequence but also in their position at either end of the heavy chains (MH1 and MH2) and at the COOH terminus of the light chains (MH3). This positional conservation might be necessary for the formation of a basic structural module common to all MAP1 proteins, in which the hallmark domains occupy pivotal areas or interact with each other. For example, in all MAP1 proteins, the light and heavy chains form a complex (Fig. 5) (1), in which the light chains are bound to the NH 2 terminus of the heavy chains through interaction of the MH1 and MH3 domains (3). The sequences unique in each of the three MAP1 proteins might form additional structures embedded in the basic module and might confer specific individual properties.
Our analysis of endogenous MAP1S protein expression is based on two affinity-purified antibodies, the epitopes of which FIG. 5. In vivo interaction of endogenous MAP1S heavy and light chains of mouse brain. Mouse brain homogenates (brain) were immunoprecipitated (IP) with either anti-LC or anti-HC antibodies as indicated. The immunoprecipitates (P) and corresponding supernatants (S) were fractionated by SDS-PAGE and analyzed by immunoblotting using anti-HC or anti-LC. The heavy chain was detected in anti-LC precipitates, and the light chain was detected in anti-HC precipitates, demonstrating the presence of MAP1S heavy chain/light chain complexes in the brain.
were predicted from the cDNA. We were able to show that MAP1S expression is widespread in murine tissues. In addition, we found that the full-length polyprotein precursor is expressed together with heavy and light chains. The latter are most likely proteolytic cleavage products of the polyprotein precursor because both heavy and light chains are encoded by a single contiguous reading frame in the cDNA. Alternative explanations, such as internal translation initiation on the full-length or truncated MAP1S mRNA to translate only the light chain, are not supported by our findings. We observed only one species of MAP1S mRNA, and in transfection studies, no evidence of internal translation initiation was obtained (data not shown). MAP1S shares this peculiar biosynthetic pathway with MAP1A (2) and MAP1B (30), and the proteolytic cleavage site that was identified for MAP1A and MAP1B (31) is conserved in MAP1S (data not shown). However, in contrast to MAP1B, which has not been detected in its uncleaved form, the MAP1S precursor is only partially cleaved into heavy and light chains, and the percentage of cleaved versus full-length precursor varied between tissues. This raises the possibility that proteolytic cleavage of the MAP1S precursor is regulated in a tissue-specific manner and perhaps represents a form of irreversible activation or maturation of the protein. Our finding of partial cleavage provides an opportunity to investigate the regulation and functional significance of precursor cleavage for MAP1S and the other members of the MAP1 family. Ectopic expression of MAP1S in PtK2 cells demonstrates that the light chain per se can have dramatic effects on microtubules. It transforms the cellular microtubule network; induces the formation of long, wavy microtubule bundles; and stabilizes microtubules against the effects of colchicine (Fig. 6) and nocodazole (data not shown). In contrast, when the light chain is expressed together with the heavy chain, microtubule binding is still observed, but microtubule bundling and stabilization are absent. For one, these results demonstrate another difference between MAP1S and MAP1B. The light chain of the latter is prevented even from microtubule binding in the presence of the heavy chain (3). On the other hand, these results suggest that the heavy chain has a regulatory function in the heavy chain/light chain complex. Conformational changes of the heavy chain perhaps triggered by phosphorylation or other posttranslational modifications or binding of additional regulatory proteins to either the heavy or the light chain might change light chain activity. A potential candidate for such a regulatory protein is the tumor suppressor protein RASSF1A, which has been reported to interact with MAP1S (24,32). Moreover, overexpression of RASSF1A can induce bundling and stabilization of microtubules reminiscent of the effects of the MAP1S light chain. Thus, it is conceivable that RASSF1A, by binding to endogenous MAP1S, triggers conformational changes necessary for light chain activation.
One feature that MAP1S shares with the other members of the MAP1 family is the potential of the light chain to act as a cytolinker, cross-linking microtubules and actin. Like MAP1A and MAP1B (4), the MAP1S light chain contains a microtubulebinding domain in its NH 2 terminus (which is not related in sequence to either MAP1A or MAP1B) and an actin-binding domain in the conserved MH3 domain in the COOH-terminal half. In vitro, the light chain can bind to both microtubules and microfilaments. In vivo, the MH3 domain displays actin binding activity, but the entire light chain (containing microtubulebinding domain and MH3) binds to microtubules. This paradox was observed previously with MAP1A and MAP1B as well and attributed to posttranslational modification in mammalian cells (but not in E. coli-produced recombinant light chain used for in vitro assays). This hypothesis is supported by findings that actin binding of the native MAP1B complex is inhibited by phosphorylation (33).
Perhaps the most significant difference of MAP1S compared with the other members of the MAP1 family is that MAP1S protein expression is readily detected not only in neurons of the brain but also in a wide range of other tissues. Thus, whereas MAP1S is still expressed in neurons, it also qualifies as the non-neuronal and hence generic member of this MAP and cytolinker family, whereas MAP1A and MAP1B are expressed predominantly in neurons. This finding has implications for the neuronal MAP hypothesis formulated many years ago. According to this hypothesis, neuronal MAPs confer special properties to an otherwise generic microtubule system to enable it to meet the special neuron-specific demands. From our characterization of MAP1S, it would appear that neither microtubule binding, bundling, stabilization, nor the putative cytolinker function of the MAP1 family of proteins is exclusively expressed in neurons. In a refinement of the above-mentioned hypothesis, we therefore propose that, not these functions per se, but perhaps the regulation of these activities by increased MAP1 protein expression or by the far larger and more complex neuron-specific heavy chains of MAP1A and MAP1B is the crucial difference.