A novel isoform of sarcolemmal membrane-associated protein (SLMAP) is a component of the microtubule organizing centre

The microtubule organizing centre (MTOC) or the centrosome serves a crucial role in the establishment of cellular polarity, organization of interphase microtubules and the formation of the bipolar mitotic spindle. We have elucidated the genomic structure of a gene encoding the sarcolemmal membrane-associated protein (SLMAP), which encodes a 91 kDa polypeptide with a previously uncharacterized N-terminal sequence encompassing a forkhead-associated (FHA) domain that resides at the centrosome. Anti-peptide antibodies directed against SLMAP N-terminal sequences showed colocalization with γ-tubulin at the centrosomes at all phases of the cell cycle. Agents that specifically disrupt microtubules did not affect SLMAP association with centrosomes. Furthermore, SLMAP sequences directed a reporter green fluorescent protein (GFP) to the centrosome, and deletions of the newly identified N-terminal sequence from SLMAP prevented the centrosomal targeting. Deletion-mutant analysis concluded that overall, structural determinants in SLMAP were responsible for centrosomal targeting. Elevated levels of centrosomal SLMAP were found to be lethal, whereas mutants that lacked centrosomal targeting inhibited cell growth accompanied by an accumulation of cells at the G2/M phase of the cell cycle.

we elucidated the genomic organization of the entire SLMAP gene, which was found to comprise 24 exons spread over ~122 kb of DNA. The genomic structure of the SLMAP gene predicted the presence of several splice variants, one of which was found to be a component of the MTOC.

Materials and Methods
Isolation and sequence analysis of genomic clones. Genomic clones corresponding to the SLMAP gene were isolated by direct screening of two mouse genomic libraries (λFIXII and λDASH II; Stratagene). SLMAP cDNA probes used for library screening were PCR generated (Table 1) and digoxigenin labeled, according to the manufacturer's directions (Roche Applied Science) for subsequent use in Southern blotting experiments. Positive phage DNA clones were digested and resolved by electrophoresis. Restriction enzyme digest fragments of the phage genomic DNA clones were analyzed by Southern blotting to identify SLMAP exons. Genomic DNA for each of the eight clones was isolated (Table 2), subcloned into pBlueScript KS (Stratagene) and subsequently sequenced using AmpliTaw sequencing methodology (ABI). Sequencing of the identified exons enabled the identification of exon-intron junctions by PCR using forward and reverse primers designed to span successive exons (Table  1). PCR conditions consisted of the following steps: (1) 3 minutes at 94°C; (2) 30 cycles of denaturation at 94°C for 30 seconds, annealing at 50-55°C for 30 seconds, extension at 72°C for 1-3 minutes; and (3) 10 minutes at 72°C. Amplicons were resolved by electrophoresis, and the isolated DNA was cloned into the pCR4-TOPO cloning vector (Invitrogen) for subsequent sequence analysis. Computer-assisted alignment of SLMAP cDNA sequence (Accession #AF304451) with genomic sequences facilitated the identification of putative exons. The exon-intron organization of SLMAP3 was verified by aligning the mouse cDNA sequence of SLMAP3 (Accession AF304451) with mouse genomic sequences (NW 000090.1) deposited in GenBank.
Consensus phosphorylation, N-glycosylation and N-myristoylation sites were identified using PROSITE database available on the EMBL server (Falquet et al., 2002).
Cell culture, transfections and drug treatments NIH 3T3 fibroblast cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum, 50 units/ml penicillin, 50 µg/ml streptomycin and gentamycin at 37°C in a humidified 5% CO2 atmosphere. Transient transfection experiments were performed using the LIPOFECTAMINE PLUS TM (Gibco BRL) transfection reagent according to the manufacturers' specifications. For experiments where microtubules were disrupted, NIH 3T3 cells grown on sterile glass coverslips were treated with either nocodazole (Sigma Chemical Co., 6 µg/ml) or paclitaxel (Sigma Chemical Co., 4 µM) in growth media (DMEM) for 3 hours at 37°C.

Antibodies
Two polyclonal antibodies against SLMAP were generated by injecting rabbits with two synthetic SLMAP-specific peptides. Anti-SLMAP(C) rabbit antisera was raised against the carboxyl 370 amino acids of SLAP, as previously described (Wigle et al., 1997). A second antibody, designated anti-SLMAP(N), was raised against the peptide RLSRGEESPPCEI, which corresponds to the extreme N-terminus of SLMAP3, within the region separating two different initiating methionines M1 and M2 ( Fig. 2A). Monoclonal anti-γ-tubulin (clone GTU-88, Sigma Chemical Co.) was used to identify centrosomes, and monoclonal anti-α-tubulin (clone DM 1A, Sigma Chemical Co.) was used to visualize cytoplasmic microtubules in immunocytochemical studies.

Isolation of centrosomes
Centrosomes were isolated from NIH 3T3 cells using sucrose density gradients according to the method of Moudjou and Bornens (Moudjou and Bornens, 1998).

Immunocytochemistry
Cells grown on sterile glass coverslips were fixed by two methods. The first method involved a 10 minute exposure to 4% paraformaldehyde (PFA) in phosphate buffer at room temperature. The second method consisted of a 1 minute exposure to microtubule stabilization buffer (MTSB: 4 M glycerol, 100 mM PIPES pH 6.9, 1 mM EGTA, 5 mM MgCl2), followed by a 2 minute exposure to MTSB containing 0.5% Triton-X-100 to extract soluble cellular components, then an additional 2 minute exposure to MTSB. Detergent extracted cells were then fixed with 4% PFA as described. Following either fixation method, coverslips were mounted onto glass slides and cells were incubated with relevant antibody(s) diluted in PBS containing 0.3% Triton-X-100 (PBS-T) for 3 hours at room temperature. After several washes in PBS, cells were incubated in the appropriate fluorochrome-conjugated secondary antibody(s) for 45 minutes at 37°C. DNA was stained with 4′,6-diamidino-2-phenylindole dihydrochloride (DAP1, 1 µg/ml) for 15 minutes at room temperature.

Microscopy and image analysis
Live cells were visualized with Axiovert S100 TV (Carl Zeiss) microscope equipped with SensiCam digital camera (PCO CCD Imaging). For immunocytochemistry studies, cells were visualized using Axiophot (Carl Zeiss) microscope equipped with a 3CCD colour video camera. Acquired images were digitally processed using Northern Eclipse (Version 5.0, Empix Imaging) acquisition software. Images were further processed using Adobe Photoshop TM 5.0 (Adobe Systems).
BrdU incorporation GFP-pcDNA3 or the SLMAP expression plasmids (GFP-SLMAP3M1; GFP-SLMAP3M1∆C; GFP-SLMAP3M2∆C) were transiently transfected into NIH 3T3 cells grown on glass coverslips. The exon-intron junctions of SLMAP3 and sizes of the respective exons and introns. ‡ λ DASH II mouse genomic library (Stratagene); see Table 3. 5′-Bromo-2′-deoxyuridine (BrdU, 10 µmol/l, Boehringer Mannheim) was added to the culture media at 36 hours post-transfection for 2, 8, 12 or 24 hours. At the end of each labeling period, cells were washed in PBS and fixed in 70% ethanol (in 50 mM glycine, pH 2) for 30 minutes at -20°C. Cells were then covered with anti-BrdU monoclonal antibody (1:10, Boehringer Mannheim) and anti-GFP rabbit antibody (1:10, Clontech) in incubation buffer (66 mM TRIS buffer, 0.66 mM MgCl2, 1 mM β-mercaptoethanol) for 30 minutes at 37°C. Following several washes in PBS, cells were incubated in secondary antibodies (in PBS) for 30 minutes at 37°C. Nuclei were stained with DAP1 as previously described. Cells were then covered with mounting media and processed for immunofluorescence microscopy.

Fluorescence activated cell sorting (FACS analysis)
Transfected cells were harvested and analyzed 48 hours after removal of DNA precipitates and prepared for FACS analysis as described by Pestov et al. (Pestov et al., 1999). The DNA content of propidium iodide (Molecular Probes, 10 µg/ml) stained GFP positive cells was measured by a Becton Dickinson FACScan cytometer using an argon laser (488 nm) and analyzed with MulticycleAV (Phoenix Flow System) program.

Genomic organization of the SLMAP gene
We previously reported that the 3′ region of the SLMAP gene is composed of 11 exons and encodes a 37 kDa SLMAP polypeptide, which is expressed in a tissue-specific manner (Wielowieyski et al., 2000). The elucidation of the genomic structure of the complete SLMAP gene may offer valuable insights into gene function. In the current study SLMAP genomic sequences were analyzed by a combination of approaches involving direct screening of mouse genomic libraries using SLMAP cDNAs, a PCR-based cloning strategy and the alignment of SLMAP cDNA with GenBank deposited genomic sequences. By direct screening of the genomic library or the PCR-based cloning strategy, eight genomic clones were isolated as outlined in Table 2. The genomic clones ranged in size from 2 kb to 20 kb. Characterization of these genomic clones revealed that the SLMAP gene spans over 122 kb of genomic DNA and consists of 24 exons (I-XXIV), including five alternative exons (XI, XII, XIII, XVII and XXIII) ( Table  3, Fig. 1). Sequence data for each intron-exon boundary conformed to the canonic consensus donor (GT) and acceptor (AG) splice sites described by Mount (Mount, 1982) and is depicted in Fig. 1A. Intron sizes varied from 87 bp to over 50.5 kb of genomic DNA, whereas SLMAP exons ranged in size from 51 bp to over 1.8 kb (Table 3). Furthermore, each class of intron (0, 1 and 2) was represented within SLMAP genomic sequences (Mount, 1982).
As a result of our analysis, three additional alternative exons were identified in SLMAP. These novel alternative exons are exons XI (51 bp), XII (51 bp) and XIII (63 bp) ( Table 3, Fig. 1). Each alternative exon is flanked by either class 1 introns (introns 10, 11, 13), which interrupt the coding between the first and second base of the codon, or a class 0 intron (intron 12), which occurs between codons (Mount, 1982). Notably, splicing of any of the alternative exons maintains the open reading frame of the predicted polypeptides. A PROSITE sequence scan did not predict the presence of conserved functional domains in alternative exons XI, XII or XIII; however, consensus sites for casein kinase 2 phosphorylation, protein kinase C phosphorylation, N-glycosylation and N-myristoylation were identified (Fig. 1B) (Falquet et al., 2002).
Leucine zipper motifs, two regions of specialized coiled-coil structure in the common carboxyl portion of SLMAPs, are encoded by exons XX and XXI, which correspond to exons VII and VIII in SLMAP1 (Wielowieyski et al., 2000). Two distinct transmembrane domains, TM1 and TM2, were previously identified within the extreme C-terminus of SLMAPs and are encoded by alternative exon XXIII and exon XXIV of SLMAP3, known as alternative exon X and exon XI in SLMAP1 (Wielowieyski et al., 2000). The expression of TM1 and TM2 has previously been described as mutually exclusive. Splicing of alternative exon TM1 introduces an inframe stop codon which renders the second transmembrane domain (TM2) nonfunctional.
Two in-frame start codons (ATG1; M1 and ATG2; M2) were previously reported in the full-length cDNA sequence of the rabbit SLMAP3 isoform (Accession #U21157) (Wigle et al., 1997) and have since been found to be conserved in mouse SLMAP cDNA. Analysis of the mouse SLMAP genomic sequences revealed that the first initiating methionine (ATG1; M1) is encoded by exon I, whereas the second initiating methionine (ATG2; M2) resides in exon II. Computer-assisted sequence analysis of exon I resulted in the identification of a consensus forkhead associated domain (FHA) corresponding to amino acid 24-58 within the region spanning M1 and M2 ( Fig. 2A; bold print). FHA domains exist in proteins of diverse function and have been recognized as phosphoserine/ threonine-specific protein-protein interaction motifs involved Journal of Cell Science 117 (11) Exon XIII: in phosphopeptide recognition (Durocher et al., 2000;Sun et al., 1998).

Endogenous expression and subcellular distribution of SLMAP
In vitro transcription-translation experiments revealed the presence of two SLMAP products whose molecular masses were similar to those predicted by the utilization of M1 (91 kDa) or M2 (80 kDa) (Wigle et al., 1997). To test whether a 91 kDa protein is produced in vivo, implying the utilization of M1 as the initiating methionine, antibodies were generated against a peptide sequence in the M1-M2 region at the N-terminus of SLMAP (anti-SLMAP-N) or against a fusion protein encompassing the C-terminal region of SLMAP (anti-SLMAP-C). Both antibodies recognized a 91 kDa protein in NIH 3T3 cell extracts (Fig. 2B, lanes 1 and 2). Immunoadsorption of the anti-SLMAP(N) antibodies with purified SLMAP fusion protein confirmed the specificity of the antiserum (Fig. 2B, lane 3). These results suggest that a 91 kDa polypeptide is produced from M1 of SLMAP3 mRNA in NIH 3T3 cells.
The use of M1 as a translation initiation codon is predicted to add 132 residues to the N-terminal region of the previously identified SLMAP3 isoform. We then examined if these residues confer different biochemical and subcellular properties to SLMAP3M1 not previously described for other SLMAP isoforms. Immunocytochemical analysis of NIH 3T3  cells that exclusively express the 91 kDa isoform revealed that SLMAP3M1 resides in several subcellular compartments. While diffuse cytoplasmic SLMAP staining was evident, SLMAP proteins could also be detected in one or two foci generally situated adjacent to the nuclear membrane ( Fig.  3Aa,b). The foci staining pattern, but not the diffuse cytoplasmic staining pattern, was resistant to detergent extraction, suggesting that SLMAP3M1 is tightly assembled with detergent-insoluble structures (Fig.  3Ab). The N-terminal directed peptide antibody (anti-SLMAP(N); Fig. 3Ac), but not the control pre-immune serum (Fig. 3Ad), detected SLMAP proteins at perinuclear foci in detergent extracted COS-7 cells as well. Peptide competition experiments abolished the staining pattern observed with anti-SLMAP(C) (Fig.  3Ae) and incubation with the pre-immune antisera also abolished staining (Fig. 3Af), thus confirming the antibody specificity for SLMAP proteins. The subcellular distribution of SLMAP was similar in other cells that express the 91 kDa SLMAP isoform, including Journal of Cell Science 117 (11) murine myoblasts (C2C12) and embryonic stem cells (data not shown).

SLMAP is a novel component of the centrosome
The perinuclear foci staining of SLMAP is reminiscent of proteins that localize to the centrosome. To determine if SLMAP is associated with centrosomes, colocalization experiments were performed with γ-tubulin, a well-characterized core centrosomal component (Stearns et al., 1991). As illustrated in Fig. 3B, SLMAP labeling was coincident with the γ-tubulin signal at each stage of the cell cycle as determined by DAPI staining. Interphase (G1) cells (Fig.  3Ba,f,k) contained one centrosome stained by both anti-SLMAP and anti-γ-tubulin. In mitotic cells SLMAP (Fig. 3Bbe) colocalized with γ-tubulin ( Fig. 3Bg-j) from prophase to anaphase (Fig. 3Bl-o). The SLMAP protein appears to have a symmetrical distribution during centrosome division given that anti-SLMAP appeared to stain each centrosome with equal intensity.
Many proteins reside at the minus ends of microtubules and are thus dependent on microtubules for their associations with the MTOC. These proteins can be distinguished from integral centrosomal proteins by treatments with agents that specifically disrupt microtubules. The fungal toxin nocodazole causes depolymerization of the cytoplasmic microtubule network, whereas taxol (paclitaxel) acts to stabilize cytoplasmic microtubules by binding tubulin directly along the  (a,d,g,j) or treated with either taxol (5 µM, 4 hours; b,e,h,k) to stabilize microtubules or nocodazole (6 µg/ml, 4 hours; c,f,i,l) to depolymerize microtubules. Cells were prepared for immunofluorescence microscopy and costained with anti-SLMAP (a,b,c), anti-γ-tubulin (d,e,f) and DAP1 (g,h,i). To show that the drugs affected cytoplasmic microtubules, cells were stained with anti-α-tubulin monoclonal antibodies (j,k,l). Bar, 25 µm. (B) Centrosomes were isolated from exponentially growing NIH 3T3 cells by fractionation on a sucrose gradient as described previously (Moudjou and Bornens, 1998). Equivalent amount (15 µg) of each protein fraction was analyzed by SDS-PAGE followed by western blotting with anti-SLMAP or anti-γ-tubulin. Lane 1 (lysed extract), lane 2 (crude centrosomes) and lanes 3-6 represent fractions obtained from sucrose density gradients. length of the microtubule (De Brabander et al., 1986). On treatment with either drug, integral centrosomal proteins remain bound to centrosomal components, whereas microtubule-dependent centrosome-associated proteins are no longer retained at the MTOC (De Brabander et al., 1986). Under control conditions (untreated) SLMAP3M1 was found to colocalize with γ-tubulin (Fig. 4Aa,d). The effects of taxol (Fig. 4Ak) and nocodazole (Fig. 4Al) on microtubules were confirmed by staining with α-tubulin, which illustrated a distinct redistribution of cytoplasmic microtubules. Treatment with taxol (Fig. 4Ab,e,h,k) or nocodazole (Fig. 4Ac,f,i,l) did not alter the localization of SLMAPs at centrosomes. These findings confirm that SLMAP localization at the centrosome is independent of microtubule assembly.
To show biochemically that SLMAP is a component of centrosomes, these subcellular structures were purified from NIH 3T3 cell extracts by fractionation on a sucrose density gradient (Moudjou and Bornens, 1998). The sucrose gradient fractions were resolved by SDS-PAGE and examined by immunoblotting with anti-γ-tubulin and anti-SLMAP (Fig.  4B). The centrosomal marker γ-tubulin was present in the cell lysate (lane 1), enriched in the crude centrosome preparation (lane 2) and even further concentrated in fraction 2 (lane 5) from the sucrose density gradient. The 91 kDa SLMAP3M1 protein was found to co-enrich with the γ-tubulin peak in fraction 2 (lane 5) of the purified centrosome preparation.

Centrosomal targeting is mediated by the novel Nterminal region of SLMAP
The data presented indicates that SLMAP3M1 is an integral component of centrosomes. These observations led us to hypothesize that the newly identified sequences from the M1 initiating codon might play a role in targeting SLMAP to the MTOC. To test this hypothesis SLMAP3M1 cDNA (Fig. 5A) and a series of SLMAP deletion mutants (Fig. 6A-D) were fused in frame to GFP to produce GFP-SLMAP fusion proteins. The GFP-SLMAP fusion constructs were used for transient expression studies in NIH 3T3 cells. Fig. 5B (a) shows that SLMAP3M1 cDNA encodes information that is sufficient to target the heterologous reporter GFP to the centrosome in live cells. This localization pattern was not Journal of Cell Science 117 (11) observed in live mitotic cells expressing GFP alone (Fig. 5Bc). GFP-SLMAP3M1 colocalized with γ-tubulin at the centrosome (Fig. 5Bb,d), consistent with the localization of endogenous SLMAP3M1 (Figs 3, 4). Deletion of the two leucine zipper regions did not affect GFP targeting to the centrosome (Fig.  6Aa), and the C-terminal truncation mutant (SLMAP3M1∆C) also retained the ability to target GFP to centrosomes (Fig.  6Ab). A cDNA encoding SLMAP initiated from the second methionine (GFP-SLMAP3M2∆C) lacking the transmembrane domain was unable to target GFP to centrosomes (Fig. 6Ac). It is notable that C-terminal mutants lacking the putative transmembrane domain resulted in the exclusion of SLMAP from the perinuclear membranes and reticular formations (Fig.  6Cb,c), which is consistent with the observation that Cterminal sequences play a role in membrane targeting of SLMAP (Wielowieyski et al., 2000). These results suggest that the newly described N-terminal sequences encompassing the FHA domain of SLMAP contain information required for its centrosomal localization.
To identify the minimal sequences present in SLMAP sufficient for targeting to the MTOC, a series of SLMAP Nterminal fragments fused to GFP were utilized in transient expression studies in NIH 3T3 fibroblast cells. Diffuse cytosolic and nuclear localization was observed in cells expressing the GFP-SLMAP mutant expressing amino acids 1-163 (Fig. 6Ba), while GFP-positive detergent-insoluble aggregates were detected in cells transfected with SLMAP mutants expressing amino acids 1-183 and 1-206 (Fig. 6Bb,c). On the basis of observations by live microscopy and costaining with the centrosome marker protein γ-tubulin (data not shown), GFP fluorescence was not detected at centrosomes in cells overexpressing these extreme N-terminal SLMAP3M1 mutants. These observations suggest that N-terminal sequences, in addition to preservation of the overall structure of SLMAP or SLMAP-mediated protein interactions, may be important determinants for targeting SLMAPs to centrosomes.

SLMAP overexpression affects cell proliferation
The finding that SLMAP3M1 was localized at the centrosome raised the possibility as to whether SLMAPs serve a role in mitosis and cell growth, and we examined this by ectopic expression studies. We observed that a large proportion of cells expressing the full-length GFP-SLMAP3M1 were nonviable and rounded in appearance (Fig. 6Db) and detached from the coverslips within 36 hours following transfection. By contrast, cell viability did not appear to be affected by the expression of either the C-terminal mutant (GFP-SLMAP3M1∆C; Fig. 6Dc) or the GFP-tagged SLMAP mutant lacking both the N-and Cterminal sequences (GFP-SLMAP3M2∆C; Fig. 6Dd). To address this issue further we examined the effect of ectopic expression of SLMAP3M1, SLMAP3M2 and SLMAP3M2∆C on cell proliferation. A reduction in the level of BrdU incorporation by more than twofold was observed in cells expressing GFP-SLMAP3M1 fusion protein relative to GFP transfected control cells for labeling periods of 2 to 8 hours (Fig. 7A). Approximately 60% of cells expressing either SLMAP3M1, SLMAP3M2 or SLMAP3M2∆C displayed BrdU incorporation after a 24 hour pulse, whereas essentially all of the control cells were positive for BrdU. These results indicate that regulated levels of SLMAPs are important for normal cell growth.
We sought to determine whether overexpression of SLMAPs altered cell proliferation by interfering in a cell cycle phasespecific manner. Fluorescence activated cell sorting (FACS) was used to identify GFP-expressing cells and thereby identify changes in cell cycle progression induced by SLMAP (Fig.  7B). Overproduction of GFP-SLMAP3M1 was toxic to NIH 3T3 cells as the majority of cells died within 36 hours posttransfection. However, cells expressing the deletion mutant GFP-SLMAP3M2 remained viable for the time required for FACS analysis, and overproduction of this deletion mutant caused about a fourfold increase in G2/M (4N DNA; Fig. 7Bii) content compared with control cells (Fig. 7Bi). These results suggest that deregulated levels of SLMAP3 in NIH 3T3 cells alter proliferation by interfering at the G2/M phase of the cell cycle.  (Pestov et al., 1999).

Discussion
In the present study, we show that a novel SLMAP isoform is a component of the MTOC and may serve a role in centrosomal functions. The SLMAP gene is encoded by 24 exons that span over ~122 kb of continuous DNA sequence. Analysis of the new sequences defined the presence of three additional regions of alternative splicing (exons XI, XII, XIII) in SLMAP compared with our previous results. A new SLMAP variant (SLMAP3M1) containing an N-terminal extension of 132 amino acid residues that are encoded in exon I was identified to be a component of the MTOC. Using anti-peptide antibodies directed toward the N-terminal sequences, a SLMAP protein of ~91 kDa was identified in various cell extracts. These findings confirm that SLMAP3M1 represents a ubiquitously expressed SLMAP isoform. Indirect immunofluorescence studies using fixation procedures known to aid visualization of the cytoskeleton localized SLMAP3M1 to the centrosome. SLMAP3M1 was detected at the centrosome at all stages of the cell cycle (therefore not a transient association) and colocalized with γ-tubulin. Treatments with reagents that specifically disrupt cytoplasmic microtubules did not affect SLMAP association at centrosomes, indicating that SLMAPcentrosome association occurs independently of microtubules. This new SLMAP isoform may be a core component of the MTOC. SLMAP3M1 sequences were found to contain the information required to target the heterologous protein GFP to centrosomes. Large C-terminal deletions eliminating much of the coiled-coil structure including the leucine zipper motifs did not affect targeting of GFP to centrosomes. However, removal of the newly identified N-terminal residues from SLMAP abolished the targeting of GFP to the centrosome, indicating that the 132 N-terminal amino acid residues are required for centrosomal targeting. Further analysis of SLMAP deletion mutants encoding N-terminal sequences revealed that this region of SLMAP is not alone sufficient for centrosome targeting. Although the N-terminus of SLMAP may serve a crucial role in mediating SLMAP-centrosome associations, the preservation of the overall structure of SLMAP may constitute an additional determinant for ensuring faithful targeting at the MTOC. Conserved motifs for protein targeting to and retention within organelles such as the ER and Golgi have been characterized (Pelham, 1988;Jackson et al., 1990;Kjer-Nielsen et al., 1999;Munro and Nichols, 1999;Brown et al., 2001), yet little is known of conserved sequences directing proteins to centrosomes. Recent studies have highlighted a conserved region of approximately 90 amino acids encoded in pericentrin and AKAP450 that confers targeting to centrosomes (Gillingham and Munro, 2000). This motif, known as the PACT domain, was not identified within the N-terminal sequences of SLMAP or within any other region of this molecule. Thus, our results indicate that structural motifs, in addition to motifs such as the PACT domain, are crucial in targeting components to centrosomes. In this regard, recent studies by Petit et al. (Petit et al., 2003) indicate that coiled-coil motifs play important roles in targeting molecules to the centrosome.
Amino acid sequence analysis of the new SLMAP Nterminal revealed the presence of a conserved motif known as the forkhead associated (FHA) domain. FHA domains have been characterized in several proteins of diverse function, including kinases, phosphatases, kinesins, transcription factors, and DNA-and RNA-binding proteins, as well as metabolic enzymes (Hofmann and Bucher, 1995;Durocher and Jackson, 2002), and are known to serve as phosphoserine/ threonine-specific protein-protein interaction motifs (Durocher et al., 2000;Sun et al., 1998). In this regard, FHA-interacting proteins such as Hklp2, a human homolog of Xklp2, a Xenopus kinesin-like motor protein, which serve roles in centrosome separation and spindle bipolarity, are known to reside at the MTOC (Boleti et al., 1996;Sueishi et al., 2000). The FHA domain in SLMAP may thus mediate associations with phosphoproteins present at MTOC. In this regard, the Aurora family of mitotic serine-threonine kinases is required for MTOC functions (Kimura et al., 1997;Kimura et al., 1999;Giet and Prigent, 1999). The presence of the FHA domains in SLMAP raises the possibility that it may interact with phosphorylated substrates of this kinase to affect centrosome activity and cell viability.
SLMAP shares several features inherent to other centrosomal proteins, including a net negative charge (pI 4.9) and extensive regions of coiled-coil structure (Doxsey et al., 1994;Boukson-Castaing et al., 1996;Errabolu et al., 1994). The alternative exons XI, XII and XIII were predicted to introduce post-translational modification sites including phosphorylation for casein kinase 2 and protein kinase C, as well as N-glycosylation and N-myristoylation. Such modifications may impose significant consequences with respect to the biological function and regulation of SLMAPs. In view of the extensive splicing predicted in SLMAP, we speculate that alternative splicing may represent a fundamental mechanism that confers functional diversity among SLMAP variants as has been reported for other coiled-coil proteins. Our recent unpublished data indicate that SLMAP isoforms can homodimerize via the specialized leucine-rich coiled-coil regions and reside in different cellular organelles, including the Golgi apparatus. In this regard the SLMAPs are similar to Golgin 97 and TGN 38, which can also reside in the Golgi and the centrosome (Takatsuki et al., 2002). Studies indicate that the spatial arrangement of the Golgi with respect to the centrosome is important for Golgi organization following mitosis, and tethering these two organelles may be a critical event mediated by as-yet-unidentified molecules (Sutterlin et al., 2002;Takatsuki et al., 2002). The SLMAPs share an overall structural homology with docking molecules such as the syntaxins and may therefore contribute to tethering function for the centrosome and the Golgi, possibly via homodimer formation. Although we do not know the precise function of SLMAP at centrosomes, the deregulation of SLMAP expression had a marked effect on cell viability and cell cycle progression. It is notable that recent data has implicated a crucial role for the centrosome in cell cycle control, although the identity of the molecular components involved remain to be defined (Rieder et al., 2001). Overall, our findings imply that the new SLMAP isoform is a component of the centrosome whose targeting is mediated by its unique structural determinants and as such may participate in the important functional roles being ascribed to this organelle.