Identification of mRNA/Protein (mRNP) Complexes Containing Purα, mStaufen, Fragile X Protein, and Myosin Va and their Association with Rough Endoplasmic Reticulum Equipped with a Kinesin Motor*

Purα, which is involved in diverse aspects of cellular functions, is strongly expressed in neuronal cytoplasm. Previously, we have reported that this protein controls BC1 RNA expression and its subsequent distribution within dendrites and that Purα is associated with polyribosomes. Here, we report that, following treatment with EDTA, Purα was released from polyribosomes in mRNA/protein complexes (mRNPs), which also contained mStaufen, Fragile X Mental Retardation Protein (FMRP), myosin Va, and other proteins with unknown functions. As the coimmunoprecipitation of these proteins by an anti-Purα antibody was abolished by RNase treatment, Purα may assist mRNP assembly in an RNA-dependent manner and be involved in targeting mRNPs to polyribosomes in cooperation with other RNA-binding proteins. The immunoprecipitation of mStaufen- and FMRP-containing mRNPs provided additional evidence that the anti-Purα detected structurally or functionally related mRNA subsets, which are distributed in the somatodendritic compartment. Furthermore, mRNPs appear to reside on rough endoplasmic reticulum equipped with a kinesin motor. Based on our present findings, we propose that this rough endoplasmic reticulum structure may form the molecular machinery that mediates and regulates multistep transport of polyribosomes along microtubules and actin filaments, as well as localized translation in the somatodendritic compartment.

Recently, we reported that the single-stranded DNA-and RNA-binding protein Pur␣, which is involved in the control of DNA replication, transcription, and translation (1), has the dual ability to up-regulate transcription of the BC1 RNA gene (2) and to mediate subsequent distribution of BC1 RNA within the somatodendritic compartment (3). Two distinct regions of the BC1 RNA gene are required for transcriptional activation in vitro (2). One of these corresponds to the 5Ј-proximal region of BC1 RNA, which overlaps with the dendrite-targeting motifs of BC1 RNA (4) and, together with Pur␣, acts as a cis-RNA element that links the BC1 RNA molecule to microtubules (MTs) 1 (3). Pur proteins are predominantly expressed in neuronal cytoplasm and are translocated into the dendrites (5). We have also reported the association of Pur␣ with free and membrane-bound polyribosomes via ribosomal subunits (5), suggesting that this transcriptional regulator may also play a role in cytoplasmic mRNA transport and translation. Within dendrites, polyribosomes preferentially congregate near postsynaptic spines (6); this is the first clue suggesting that localized mRNAs are translated into proteins on demand (7). Although little is known about when and where these polyribosomes are formed and how they emerge at postsynaptic sites, RNA-binding proteins that associate with polyribosomes appear to be involved. These proteins include mammalian Staufen (mStaufen) (8,9), Embryonic Lethal Abnormal Vision (ELAV/ Hu) (10), and Fragile X Mental Retardation Protein (FMRP) (11,12). mStaufen is involved in the delivery of RNA to dendrites (13) and has also been suggested to play a role in the transport of neuronal RNA granules (14), which may form a link between RNA localization and activity-dependent translation within dendrites (15). FMRP has been postulated to regulate protein synthesis at the synapse (16,17). However, the specific RNA sequences recognized by these trans-acting proteins differ from those recognized by Pur␣: mStaufen preferentially binds to double-stranded RNA (18,19), while ELAV/Hu has an AU-preference (20) and FMRP is reported to bind to a G quartet structure (21,22). In contrast, Pur␣ binds to (GGN) n consensus sequences (1). It is therefore possible that Pur␣ may serve a distinct function in gene expression at the cytoplasmic level via RNAs that contain such consensus sequences. In this context, it would be intriguing to characterize the proteins that interact with Pur␣-polyribosome complexes.
In the present study, we demonstrate that, following treatment with EDTA, Pur␣ is released from polyribosomes in mRNA/protein complexes (mRNPs), which also contain mStaufen, FMRP, myosin Va, and several other proteins with unknown functions. Furthermore, our results also suggest that mRNPs may reside on rough endoplasmic reticulum (rER) structures equipped with a kinesin motor. We propose that these structures may form the molecular machinery that me-* This work was supported by Grant 13680853 from the Ministry of Education, Culture, Sports, Science and Technology of Japan (to K. A.), by a Joint Research grant from Nihon University (to K. A.), and by a grant from the Ministry of Education, Culture, Sports, Science, and Technology of Japan to promote multidisciplinary research projects. 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.

MATERIALS AND METHODS
Sources of Antibodies-The antibodies used in the present study were obtained from the following sources: the rabbit anti-Pur␣ antibody was produced as described previously (5); the mouse anti-kinesin heavy chain (KHC) monoclonal antibody (clone H2; see Ref. 23), and mouse anti-FMRP monoclonal antibody (24) were purchased from Chemicon (Temecula, CA); the rabbit anti-GRP78 (BiP) antibody was obtained from Affinity Bioreagents, Inc. (Golden, CO); and rabbit antibodies against ribosomal proteins L8 and S14 (25) were kindly provided by Dr. Taka-Aki Sato of Columbia University. The rabbit anti-mStaufen antibody was a generous gift from Dr. Juan Ortin of Centro Nacional de Biotecnologia, Madrid. The rabbit anti-myosin Va antibody was raised specifically for this study as described in the following section. The specificity of the antibodies was checked by Western blot analysis.
Production of Anti-myosin Va Antibody-An anti-myosin Va antibody was raised in rabbits against a synthetic multiple antigenic peptide corresponding to amino acids 197-210 (LASNPIMESIGNAK) of myosin Va. The antibody was then affinity-purified to the peptide coupled to aminoalkyl agarose (Bio-Rad; Hercules, CA) (5). Briefly, whole immunoglobulin G (IgG) was extracted from rabbit sera by precipitation with 40% ammonium sulfate. Following centrifugation at 10,000 ϫ g for 10 min, the IgG-rich pellet was reprecipitated, dissolved in phosphate-buffered saline (PBS), dialyzed against PBS, and affinitypurified. The purified IgG was then precipitated by adding solid ammonium sulfate to 75% saturation. The pellet thus obtained was dissolved in PBS and dialyzed against PBS.
Protein Preparation-All of the following steps were performed at 0 -4°C unless otherwise stated. Mouse brains were homogenized in TKM buffer (20 mM Tris-HCl, pH 7.5, 2 mM MgCl 2 , 50 mM KCl, 250 mM sucrose, 1 mM phenylmethylsulfonyl fluoride, and 100 g/ml cycloheximide), and the homogenate was centrifuged at 1000 ϫ g for 10 min. The supernatant was recentrifuged at 10,000 ϫ g for 15 min to yield postmitochondrial supernatant (PMS). The PMS was then further centrifuged at 130,000 ϫ g for 1 h, resulting in supernatant (S100) and pellet (P100) fractions. The P100 fraction was resuspended in TKM buffer, and solid KCl was added to a final concentration of 150 mM. The sample was kept on ice for at least 15 min and then centrifuged at 10,000 ϫ g for 5 min to remove insoluble materials. The supernatant was used as the P100 extract.
Western Blot Analysis-Western blot analysis was performed essentially as described previously (3). The proteins were separated by SDS-PAGE and transferred onto polyvinylidene difluoride membranes (Immobilon-P; Millipore; Bedford, MA). The membrane was blocked overnight with 5% skim milk in TBST (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 0.05% Tween 20) at 4°C, rinsed with TBST, and reacted with the appropriate primary antibody, which was diluted according to the supplier's instructions. The primary antibody was then removed by washing with TBST, and the membrane was incubated with an appropriate alkaline phosphatase-conjugated secondary antibody. Finally, after the membrane had been thoroughly washed with TBST, the protein signals were detected using the ProtBlot Western blot AP System (Promega; Madison, WI).
Polyribosome Analysis-Polyribosomes were isolated as described previously (5). In brief, mouse brains were homogenized in TKM buffer and centrifuged at 10,000 ϫ g for 30 min at 4°C. The PMS fraction was treated with 0.5% Nonidet P-40 in the presence or absence of 25 mM EDTA for 30 min and then centrifuged on a linear sucrose gradient (8.5-45% (w/v) in TKM buffer) at 160,000 ϫ g for 2.5 h. After centrifugation, fractions were collected from the bottom of the gradient and stored at Ϫ80°C until required for analysis of their polyribosome contents.
Immunoprecipitation-P100 extracts (150 g of protein in 250 l) were incubated with 5 g (in 5 l) of either control antibody or anti-Pur␣ antibody for 2 h at 4°C in TKM buffer with or without 0.1% Nonidet P-40. For P100 extracts subjected to treatment with EDTA, the antibodies were added after the extracts had been treated with or without 25 mM EDTA on ice for 30 min. When the extracts were treated with RNase A in the presence of EDTA, 0.2 g/ml RNase A was additionally included. Subsequently, 40 l of Dynabeads M-280 sheep antirabbit IgG (Dynal; Oslo, Norway) was added, and the reaction mixtures were incubated for an additional 4 h at 4°C. The beads were then collected with a magnet, and the supernatant was saved as "unbound fraction" for subsequent Western blot analysis. The collected beads were washed five times (5 min each) in 200 l of TKM buffer containing 0.1% Nonidet P-40 or 0.1% Nonidet P-40 plus 25 mM EDTA, as appropriate. The beads were then washed for an additional 15 min, after which the immune complex was eluted in 40 l of SDS-containing sample buffer (50 mM Tris-HCl, pH 6.8, 2% SDS, 5% ␤-mercaptoethanol, and 10% glycerol). The whole eluted materials (40 l), one-tenth volume of the unbound fraction (ϳ300 l), and the last wash fraction (200 l) were analyzed by SDS-PAGE, followed by Western blot analysis. For RNA extraction, the beads were suspended in RNP buffer (50 mM Tris-HCl, pH 7.5, 25 mM KCl, 5 mM MgCl 2 , 5% glycerol, and 0.25 mM dithiothreitol) and then purified by the SDS-phenol-chloroform method, followed by ethanol precipitation as described previously (26).
Northern Blot Analysis-Immunoprecipitated RNAs were resolved on a 1.5% morpholinopropanesulfonic acid-formaldehyde-agarose gel, and the RNAs were transferred onto a nitrocellulose membrane. The membrane was prehybridized in a hybridization solution containing 5ϫ saline/sodium phosphate/EDTA, 5ϫ Denhardt's reagent, 50% formamide, 1% SDS, and 0.1 mg/ml salmon sperm DNA for 2 h at 42°C and then hybridized with 32 P-end-labeled probes overnight. The membrane was washed with 2ϫ standard saline citrate and 0.1% SDS and then with 0.2ϫ standard saline citrate and 0.1% SDS. For probe removal, the membrane was heated in boiling water for 2 min, cooled to room

FIG. 1. Immunoprecipitation of Pur␣/polyribosome complex.
A, P100 extracts of the brain treated with 0.1% Nonidet P-40 were incubated with either control antibody or anti-Pur␣ antibody. Following an additional incubation with sheep anti-rabbit antibody-coupled Dynabeads, The beads were collected with a magnet and washed, and bound protein was eluted with SDS and separated by SDS-PAGE, followed by Western blot analysis using anti-Pur␣ antibody (Pur␣), anti-ribosomal protein L8 antibody (L8), or anti-BiP antibody (BiP). The whole eluted materials, one-tenth volume of the unbound fraction, and the last wash fraction were subjected to analysis. Lane numbers correspond to those in A.

FIG. 2. Coimmunoprecipitation of myosin Va with Pur␣/polyribosome complex.
Immunoprecipitates were isolated using anti-Pur␣ antibody (lanes 1, 2, and 4) or control rabbit antibody (lane 3) from P100 extract of the brain (lanes 1, 3), S100 extract of the brain (lane 2), and P100 extract of the liver (lane 4) followed by SDS-PAGE. Open triangle, myosin Va; closed triangle, myosin II type B, molecular mass standards are indicated in kDa at the right.
Cell Culture and Immunocytochemistry-Primary cultures of hippocampal neurons from embryonic rats (day 19) were grown on polyethylenimine-coated coverslips at a density of ϳ200 cells/mm 2 . On the 6th day of culture, the cells were fixed in 3.7% formaldehyde for 10 min, rinsed with PBS, and permeabilized with 0.5% Triton X-100 in PBS for 10 min. The cells were then blocked with Dulbecco's modified Eagle's medium containing 10% fetal bovine serum for 30 min and incubated with the appropriate primary antibody for 2 h. After washing with PBS, the cells were incubated with Alexa 488-conjugated goat anti-rabbit antibody (Molecular Probes; Eugene, OR) or Cy3-conjugated donkey anti-mouse antibody (Jackson ImmunoResearch Laboratories; West Grove, PA) for 1.5 h. For double staining with a combination of the rabbit primary antibodies, a biotinylated anti-Pur␣ antibody was used. The biotinylated antibody was detected with Cy3-streptavidin (Molecular Probes) according to the manufacturer's instructions. The Biotin Labeling Kit (Roche Diagnostics; Mannheim, Germany) was used for labeling of antibodies with biotin in accordance with the supplier's instructions. The immunostained cells were viewed using a Zeiss LSM-510 fluorescence laser-scanning confocal microscope.

Coimmunoprecipitation of Polyribosomes with Pur␣
Protein-We have previously reported that the single-stranded DNA-and RNA-binding protein Pur␣ associates with polyribosomes via the ribosomal subunits (5). However, as these findings came mainly from cosedimentation studies, in the present study we performed initial immunoprecipitation experiments using an anti-Pur␣ antibody to obtain more direct data on Pur␣/polyribosome interactions. For this analysis, protein ex-tracts were prepared from the P100 fraction of brain cytoplasm using Nonidet P-40 because Pur␣ is concentrated to the greatest extent in this fraction. Fig. 1A shows that ribosomes were pulled down by the anti-Pur␣ antibody but not by the control antibody and that immunoprecipitation of the ribosomes was completely blocked by the Pur␣ peptide (Fig. 1B), confirming our previous observations. It is important to emphasize that the association of Pur␣ with the polyribosomes was not mediated by rERs; the rER-marker protein BiP was not immunoprecipitated but remained in the unbound fraction in the presence of Nonidet P-40 (Fig. 1A).
An Anti-Pur␣ Antibody Immunoprecipitates Two Distinct Nonmuscular Myosin Proteins, Myosin Va from the P100 Fraction and Myosin II Type B from the S100 Fraction-The immunoprecipitates obtained above were subjected to SDS-PAGE. Fig. 2 and show that one particular high molecular mass protein (200 kDa) was predominantly detected in the P100 brain fraction (lane 1). This protein was much less apparent in the S100 fraction and was not detected in a P100 fraction prepared from the liver. An additional protein (220 kDa) was immunoprecipitated from the S100 brain fraction (lane 2). On microsequencing of the 200 kDa protein (3), amino acid sequences were obtained for two peptides containing ten (LTNLEXVYNS) and seven (IMQLQRK) amino acids, respectively. A computer search revealed that these two peptides were identical to amino acids 950 -959 and 929 -935 of the mouse nonmuscle myosin heavy chain Va (myosin Va), respectively (27). The 220 kDa protein also yielded two peptides, which contained ten (EQAD-FAVEAL) and eight (REQEVAEL) amino acids, respectively. These peptides were identical to amino acids 415-424 and 1172-1179 of the human nonmuscle myosin heavy chain type B (28). Myosin Va is known to be an unconventional molecular motor for the transport of certain vesicles (29) and mRNAs (30),

FIG. 3. Immunochemical analyses of various protein extracts.
Pur␣/polyribosome complex-bound proteins were isolated using anti-Pur␣ antibody (Pur IgG) or control rabbit antibody (CR IgG) from the P100 extract of the brain in the absence (A) or presence (B) of Nonidet P-40 as described in the legend for Fig. 1 followed by Western blot analysis using the appropriate antibody for the indicated proteins. Only the eluted fractions (see Fig. 1) are shown. A, RNA purified from the immunoprecipitate was also analyzed by Northern blotting for the presence of 7SL RNA using a 32 P-labeled 7SL RNA-specific oligonucleotide (53) as a probe (7SL; bottom panel). B, total RNA was also purified from each immunoprecipitate (bottom panel); only the region containing 28 S rRNA and 18 S rRNA is shown. C, Western blot analysis of immunoisolated FMRP-bound proteins. FMRP-bound proteins were isolated with anti-FMRP antibody or control mouse antibody from the P100 extract of the brain treated with 0.1% Nonidet P-40, followed by Western blot analysis using antibodies indicated. D, Western blot analysis of the postmitochondrial supernatant fraction of the brains (20 g of protein) from wild type (WT) or dilute-lethal (d l ) mice. Antibodies used are indicated at the right.
FIG. 4. Velocity gradient analysis of the PMS fraction of the brain. A, the PMS was centrifuged on a sucrose gradient and fractionated. An aliquot of each fraction was subjected to Western blot analysis. B, the PMS treated with 0.5% Nonidet P-40 was separated on a sucrose gradient and analyzed as described in A. C, the PMS simultaneously treated with 0.5% Nonidet P-40 and 25 mM EDTA was separated on a sucrose gradient, followed by Western blot analysis. Antibodies used in the Western blotting are indicated at the right. Centrifugation was from right to left. The positions of 40, 60, and 80 S ribosomal particles are indicated.
while myosin type B is an isoform of nonmuscular myosin II and is reportedly involved in the transport of ␤-actin mRNA in fibroblasts (31). In the present experiments, myosin II type B, which was more concentrated in the S100 fraction, was not studied further.
The Association of Kinesin Motors, but Not Myosin Va, with the Pur␣/Polyribosome Complex Is Mediated by Rough Endoplasmic Reticulum-To confirm that the 200 kDa protein is perceptible by the anti-Pur␣ antibody, we raised an antibody against a myosin Va peptide to act as a probe. Fig. 3, A and B show that the anti-myosin Va antibody detected a protein band with the expected molecular mass (200 kDa) in the immunoprecipitate. This 200-kDa protein was confirmed to be myosin Va because it was not detected by the anti-myosin Va antibody when brain tissue from dilute-lethal mice (a myosin Va-null mutant mouse strain; Ref. 27) obtained from Jackson Laboratory (Bar Harbor, ME) was similarly analyzed (Fig. 3D). Fig. 3 again indicates that rERs are not relevant to the immunoprecipitation of myosin Va because, in the presence of Nonidet P-40, the myosin Va level remained unaffected, whereas BiP protein became undetectable (compare Fig. 3A with B). These observations suggest that myosin Va powers the translocation of Pur␣/polyribosome complexes along actin filaments via protein-protein interactions. Although this may at first seem incompatible with our previous observations that Pur␣ may play a role in MT-based RNA transport (3), a kinesin motor is reported to be in direct contact with myosin Va (32). This type of motor is known to be present in both axonal and dendritic processes (33) and to be involved in the anterograde translocation of vesicles and dendritic mRNAs such as CaMKII␣ mRNA (34). We therefore determined whether the KHC was immunoprecipitated by the anti-Pur␣ antibody. Unlike myosin Va, but similar to BiP proteins, KHC was immunoprecipitated only in the absence of Nonidet P-40 (Fig. 3, A and B), indicating that the interaction of the kinesin motor with the Pur␣/polyribosome/myosin Va complex is indirect and is mediated by rER. In support of the relevance of rER, the anti-Pur␣ antibody pulled down 7SL RNA (the RNA component of signal recognition particles) as well as BiP in the absence of Nonidet P-40 (see Fig.  3A). Thus, we suggest that polyribosomes containing Pur␣ and myosin Va could be translocated along MTs in the form of rERs equipped with a kinesin motor. Note that the molecular isoform of the kinesin motor detected here is likely to be either KIF5A or KIF5C, as the anti-KHC antibody H2 used in this study reportedly shows a preference for KIF5A and KIF5C to KIF5B during Western blot analysis (35).
Coimmunoprecipitation of Fragile X Mental Retardation Protein and mStaufen with Pur␣/polyribosome/myosin Va complexes-Pur␣ has properties very similar to those of mStaufen and FMRP, both of which bind to polyribosomes, are localized to rER, and are distributed to the somatodendritic compartment (5,9,19,36,37). We therefore investigated whether these proteins are also incorporated into polyribosomes together with Pur␣ and myosin Va. As shown in Fig. 3B, FMRP and mStaufen were specifically coimmunoprecipitated by the anti-Pur␣ antibody in the presence of Nonidet P-40. Note that the anti-FMRP antibody used in the Western blot analysis as a probe appears cross-reactive with two FMRP-related proteins, FXR1P and FXR2P as well as FMRP, as reported by others (24), but in the present study we collectively refer to them as FMRP. Furthermore, Pur␣, myosin Va, and polyribosomes were reverse-immunoprecipitated by the anti-FMRP antibody (Fig. 3C). Consistent with these results, Pur␣, FMRP, and mStaufen cosedimented with polyribosomes on a velocity gradient whose banding profiles were essentially identical to those previously reported ( Fig. 4A; see Fig. 4B for mStaufen; see also Refs. 5,9,19,36,37). Fig. 4A also shows that, in the absence of Nonidet P-40, BiP, KHC, and myosin Va were distributed throughout the gradient, although most of the KHC was localized in the top fractions. Nonidet P-40-treated PMS yielded similar profiles, except for BiP and KHC (Fig. 4B), both of which were converted to slower sedimenting materials as expected from Fig. 3, A and B. In addition, myosin Va as well as Pur␣, FMRP, and mStaufen were shifted to the top fractions of the gradient following treatment with EDTA (Fig. 4C). Other differences in the shift profiles were also noticed after EDTA treatment: FMRP and myosin Va sedimented ahead of ribosomal subunits with a peak at about 100 S, suggesting that these two proteins were released from the ribosomes upon EDTA treatment. Similar observations, in which FMRP was released as large mRNPs (ranging from 30 to 100 S) from EDTA-treated polyribosomes, have also been reported (38,39). On the other hand, Pur␣ and mStaufen exhibited somewhat broader distributions than FMRP and myosin Va (Fig. 4C); some sedimented in concert with the FMRP and myosin Va, while some sedimented with the ribosomal subunit fractions. In the former situation, Pur␣ and mStaufen appeared to be released from the polyribosome as a complex with FMRP and myosin Va (see Fig. 5B).
Pur␣, mStaufen, FMRP, Myosin Va, and Many Other Unidentified Proteins Are Released from Polyribosomes as mRNPs following Treatment with EDTA-Next we determined whether Pur␣, mStaufen, FMRP, and myosin Va are freed from polyribosomes by EDTA treatment but still held in a protein complex as suggested in Fig. 4C. Fig. 5A shows that they were all immunoprecipitated together from the P100 fraction, irrespective of whether or not the P100 fraction was pretreated with EDTA. Furthermore, no significant changes in the levels of the precipitated proteins were observed, except for myosin Va, which appeared at a slightly lower level in the presence of EDTA. In contrast, rRNAs as well as the ribosomal proteins L8 and S14 almost disappeared. These findings indicate that all four proteins are released as a complex from the polyribosome and that physical interactions with the ribosome are not required for their coimmunoprecipitation. Identical results were obtained when high molecular mass polyribosomes containing FMRP (see Fig. 4B, fractions 1-3 of the gradient) were pooled and analyzed in a pull-down assay (Fig. 5B). Taken together, our results indicate that FMRP, Pur␣, mStaufen, and myosin Va dissociate from high molecular mass polyribosomes following treatment with EDTA and shift together to the lighter fractions of the gradient (see Fig. 4, B and C). In fact, the proteins present in fractions 6 -8 of the gradient shown in Fig.  4C were coimmunoprecipitated, but no L8 was detected (Fig.  5B). We also analyzed the immunoprecipitate by SDS-PAGE followed by silver staining. As shown in Fig. 5C, as well as the ribosomal proteins (Ϫ), at least 30 nonribosomal proteins including Pur␣ were detected in the absence of EDTA. In the presence of EDTA, the ribosomal proteins were selectively lost and there were some changes in the molecular species of the other nonribosomal proteins. The nonribosomal proteins whose levels remained unaffected were marked by dots. These results indicate that Pur␣, mStaufen, FMRP, and myosin Va are held within larger protein complexes.
We carried out further examinations to determine whether the large multiprotein complexes also contain RNA. When both RNase A and EDTA were simultaneously included in the pulldown assay using the anti-Pur␣ antibody, immunoprecipitation of FMRP and myosin Va (Fig. 6A) was abolished. Similarly, almost all of the less well characterized proteins detected by the anti-Pur␣ antibody also disappeared (Fig. 6B), although some proteins remained unaffected (indicated by dots). Taken together, it is suggested that some RNAs are required to hold the proteins together. The RNAs most likely to be involved are mRNAs as the levels of the mRNAs recovered from the immunoprecipitate remained unaffected in the presence of EDTA. In contrast, the rRNAs became greatly reduced in amount ( Fig. 7; see also Figs. 5 and 6), indicating that rRNAs are not responsible for the coimmunoprecipitation of the proteins. Thus, it is reasonable to conclude that EDTA treatment causes large multiprotein-mRNA complexes containing Pur␣, mStaufen, FMRP, myosin Va, and many other proteins with unknown functions to be released from polyribosomes. However, the exact combinations of protein components needed to form mRNPs with distinct functions remain to be determined. All of these proteins may assist in the regulation of post-transcriptional events, such as transport and translation of mRNAs, by incorporation into polyribosomes within the somatodendritic compartments of neuronal cells.
Finally, the intracellular distributions of endogenous Pur␣, ribosomes, FMRP, and mStaufen were examined in cultured primary neurons by a laser-scanning confocal microscope. Consistent with biochemical data presented above, Fig. 8, A-I show that staining patterns of Pur protein were almost indistinguishable by immunofluorescence from those of ribosomes, FMRP, or mStaufen proteins. In addition, we also found that FMRP was colocalized with mStaufen in the neuronal cytoplasm (Fig. 8, J-L). Therefore, these observations provided further supporting evidence that the three proteins are held together within protein complexes containing mRNAs, although they appeared closely apposed but not fully congruent with each other in the neuronal processes under the culture conditions we employed. DISCUSSION In the present study, we demonstrated that mStaufen, FMRP, and myosin Va, which play a role in mRNA transport and translation, are coimmunoprecipitated with Pur␣/polyribosome complexes (Fig. 3B). When their banding profiles were compared on the same velocity gradient, FMRP tended to associate with relatively fast sedimenting polyribosomes, whereas Pur␣, mStaufen, and myosin Va exhibited more uniform distributions across the gradient (Fig. 4, A and B). Following treatment with EDTA, mRNPs containing Pur␣, mStaufen, FMRP, myosin Va (Fig. 5B), and other proteins with unknown functions (Fig. 5C) were released from the high molecular mass polyribosomes. As the immunoprecipitation of these proteins by an anti-Pur␣ antibody was abolished by treatment with RNase A (Fig. 6), Pur␣ may play a role in the assembly of these large mRNPs in an RNA-dependent manner. Similar RNA dependence has been observed in the assembly of nucleolin-binding RNP complexes during the biogenesis of preribosomal particles (40). FMRP-containing mRNPs are also known to contain nucleolin (41). Hence, there is a possibility that nucleolin as well as Pur␣ may be involved in the assembly process, although our mRNP samples were not analyzed for its presence. FMRP has been postulated to play a key role in targeting mRNAs to translating polyribosomes, thereby regulating the translation of its target mRNAs as the association of FMRP-containing mRNPs with polyribosomes is reportedly altered in cells carrying the severe Fragile X syndrome-causing mutation (17,39). In this respect, it is conceivable that both Pur␣ and mStaufen may mediate the association of FMRPcontaining mRNPs with polyribosomes via rRNA binding. As supporting evidence for this, we have recently observed that both Pur␣ and mStaufen protected certain regions of rRNA (average length: 350 nucleotides) from micrococcal nuclease action when polyribosomes, but not ribosomal subunits, were subjected to digestion. Furthermore, Pur␣, mStaufen, and fragmented rRNAs were coimmunoprecipitated with FMRP by an anti-Pur␣ antibody. 2 It is noteworthy that FMRP and Pur␣ are known to function as a translational repressor in vitro (42)(43)(44).
As shown in Figs. 3 and 4, myosin Va appears to associate with polyribosomes via protein-protein interactions. A further example of a protein tethering a myosin motor to RNA-containing structures has been reported in yeast; a yeast homolog of myosin Va, Myo4p, is linked to the ASH1 mRNA-She2p complex via the She3p adapter protein (30). Similarly, it is possible that Pur␣ itself links myosin Va to polyribosomes. This is supported by our observation that another isoform of myosin, myosin II type B, was immunoprecipitated from the S100 fraction by the anti-Pur␣ antibody (Fig. 2). It is noteworthy that the computer program COILS (45) 1, 2) of 25 mM EDTA. RNase A (0.2 g/ml) was additionally included during immunoprecipitation experiment (lane 4). Then RNA was purified from the immunoprecipitate and used for Northern blot analysis. 32 P-end-labeled DNA probe used was: upper panels, 28 S rRNA-specific oligonucleotide (5Ј-TGACGAG-GATTTGGCTACCTTAAGAGAGTCATA-3Ј); lower panels, oligo(dT) 25 . The same RNA blot was hybridized with 32 P-end-labeled oligo(dT) 25 probe to detect mRNAs after removal of the 32 P-labeled 28 S rRNA probe. Molecular size standards in kb and the position of 28 S rRNA are indicated at the right.  K). In A and G, a biotinylated anti-Pur␣ antibody was used. The rabbit antibodies against S14 protein (B), Pur␣ (D), or mStaufen (H, K) were detected with an Alexa 488conjugated goat anti-rabbit IgG antibody, and the mouse anti-FMRP antibody (E, J) was detected with a Cy3-conjugated donkey anti-mouse IgG antibody. The biotinylated antibody was detected with Cy3-streptavidin. Panels C, F, I, and L are superimpositions of panels A plus B, panels D plus E, panels G plus H, and panels J plus K, respectively. For controls the fluorescent-labeled secondary antibodies and Cy3-streptavidin gave no signals in the absence of their corresponding primary antibodies (data not shown). tures in Pur␣ (amino acids 271-304), myosin Va (amino acids 914 -1237 and 1314 -1443), and myosin II type B (amino acids . It would be interesting to determine whether Pur␣ interacts with myosin Va or myosin II type B via such coiledcoil structures. Figs. 3 and 4 also show that the Pur␣/polyribosome/ mStaufen/FMRP/myosin Va complex may form rER structures equipped with a kinesin motor (most likely KIF) (Fig. 5A or C). This suggests that the motor protein may power the translocation of the rER structures along MTs. It has been proposed that vesiculated rERs pinched off from the perinuclear region may be involved in mRNA trafficking and localization in neurons (46). In addition, ER transport may also play a role in the localization of Vg1 mRNA in Xenopus oocytes, which is mediated by an ER-associated Vera protein (47,48). Hence, the rER-localized polyribosome-associated proteins Pur␣, mStaufen, and FMRP may help to determine the final distribution of polyribosomes carried by rER structures along MTs. On the other hand, myosin Va is likely to play a role in polyribosomal transport processes in the peripheral regions of neuronal cells, which are rich in actin fibers. In this respect, the incidence of spine-associated polyribosomes is known to decline with the induction of long term potentiation of synaptic transmission (49); this has been regarded as a consequence of their migration into the intraspinal space (50) (which is rich in actin fibers (51)) to reach the postsynaptic density. As the postsynaptic density is reportedly rich in myosin Va (52), it could be that polyribosomes orientated toward the dendrites are equipped with myosin Va for use in actin-based transport near synapses. mStaufen has recently been reported to play a major role in the dendritic trafficking of mRNA (13). Furthermore, this protein has also been reported to form a link between the localization of RNA granules containing clusters of ribosomes and mRNAs and activity-dependent translation at synapses (15). However, it is not known how the RNA granules are transported or how the mRNAs stored in the granules are recruited to polyribosomes. Nevertheless, it would be intriguing to determine the relationship between such RNA granules and our putative polyribosomal transport complex within the somatodendritic compartment. As both structures are rich in mStaufen (Figs. 3, 4; see also Refs. 8,9,15,19), the polyribosomes might be converted into more compact structures, like RNA granules, which would be more amenable to transport through the dendritic processes, which have very small diameters in some cases.
Finally, it is possible that the less well characterized mRNP complex proteins detected by the anti-Pur␣ antibody (Fig. 5C) may play similar but distinct roles in mRNA transport and translation within the somatodendritic compartments of neurons. It would be intriguing to characterize these proteins at the molecular level and to analyze the aspects of gene expression in both the nucleus and the cytoplasm in which they are involved.