Domain architectures and characterization of an RNA-binding protein, TLS.

Translocated in liposarcoma (TLS) is an important protein component of the heterogeneous nuclear ribonucleoprotein complex involved in the splicing of pre-mRNA and the export of fully processed mRNA to the cytoplasm. We examined the domain organization of human TLS by a combined approach using limited proteolysis, matrix-assisted laser desorption ionization time-of-flight mass spectrometry, circular dichroism, inductively coupled plasma atomic emission spectroscopy, and NMR spectroscopy. We found that the RNA recognition motif (RRM) and zinc finger-like domains exclusively form protease-resistant core structures within the isolated TLS protein fragments, while the remaining regions, including the Arg-Gly-Gly repeats, appear to be completely unstructured. Thus, TLS contains the unstructured N-terminal half followed by the RRM and zinc finger-like domains, which are connected to each other by a flexible linker. We also carried out NMR analyses to obtain more detailed insights into the individual RRM and zinc finger-like domains. The 113Cd NMR analysis of the zinc finger-like domain verified that zinc is coordinated with four cysteines in the C4 type scheme. We also investigated the interaction of each domain with an oligo-RNA containing the GGUG sequence, which appears to be critical for the TLS function in splicing. The backbone amide NMR chemical shift perturbation analyses indicated that the zinc finger domain binds GGUG-containing RNA with a dissociation constant of about 1.0 x 10(-5) m, whereas the RRM domain showed no observable interaction with this RNA. This surprising result implies that the zinc finger domain plays a more predominant role in RNA recognition than the RRM domain.

The translocated in liposarcoma (TLS) 1 protein, also termed FUS, was first identified in human myxoid and round cell liposarcomas as an oncogenic fusion protein with a stressinduced DNA-binding transcription factor, CCAAT enhancerbinding homologous protein (CHOP, also known as GADD153 or DDIT3) (1,2). The resultant fusion protein (TLS-CHOP), consisting of the N-terminal half of TLS and the full-length CHOP, appears to act as a potent transcription factor possibly by combining the TLS transactivation activity and the CHOP DNA binding activity. A different type of fusion protein, TLS-ERG (a member of the erythroblast transformation-specific (ETS) family of transcription factors), was subsequently detected in human acute myeloid leukemia (3).
The normal TLS, consisting of 526 amino acids with a calculated molecular mass of 53 kDa, belongs to a family including the closely related proteins Ewing's sarcoma (EWS) (4) and TAF II 68 (TATA-binding protein-associated factor) (5). Thus, they are collectively called the TET (TLS, EWS, TAF II 68) family. EWS and TAF II 68 interact with components of the RNA polymerase II complex (5,6). Moreover sarcoma-associated RNA-binding fly homologue (SARFH), a Drosophila homologue of TLS, is colocalized with the polymerase on active chromatin (7). The N-terminal domain of TLS is involved in transcriptional activation (8) and interacts with the polymerase (9). Indeed the N-terminal halves of the TET family members are rich in glutamine, serine, proline, and tyrosine, which are frequently found in transcriptional activation proteins such as SP1 and Krü ppel-like transcription factors (10). The several cellular partners of TLS identified thus far function as transcription factors. For example, the N-terminal half of TLS shows high affinity for the thyroid hormone receptor, which binds to certain DNA target sites (11). Moreover TLS interacts with Spi-1/PU.1, and this association impedes the transcriptional activity of Spi-1 (12).
TLS was identified as the P2 component of the heterogeneous nuclear ribonucleoprotein (hnRNP) complex by proteomic analyses using electrospray ionization mass spectrometry (MS) (13). The hnRNP complex, consisting of more than 30 different proteins, is involved in pre-mRNA splicing and transporting fully processed mRNA to the cytoplasm. Indeed many members of the hnRNP family contain multiple RNA recognition motifs (RRMs) and shuttle between the nucleus and the cytoplasm (14). The C-terminal half of TLS contains several structural motifs involved in RNA binding. A sequence analysis suggested that the C-terminal half contains an RRM flanked by multiple Arg-Gly-Gly (RGG) repeats on both the amino and carboxyl sides in addition to a small, putative zinc finger domain with four cysteines (15). The TLS protein interacts with hnRNP A1 and hnRNP C1/C2 in the nucleus and binds mRNA in vivo (8). Thus, TLS is likely to play critical roles in pre-mRNA splicing and exporting fully processed mRNA to the cytoplasm. Indeed TLS exists not only in the nucleus but also in the cytoplasm, and this shuttling depends on its C-terminal domain (16), which preferably recognizes RNA containing a GGUG sequence (17). In addition, TLS interacts with the serine-arginine family members of splicing factors through its C-terminal domain (18). Moreover TLS is colocalized specifically to spreading initiation center, which exists only in early stages of cell spreading, with ribosomal RNA and other hnRNP complex components (hnRNP K and hnRNP E1) and functionally involved in controlling the rate of cell spreading (19). Taken together, TLS appears to be a multifunctional protein responsible for a variety of regulatory processes, including transcription, mRNA splicing, mRNA transport from the nucleus to the cytoplasm, and the initiation of cell spreading. However, despite its physiological importance, little biochemical and structural data about TLS have been available thus far.
In the present study, we characterized the tertiary structure of TLS by a combined approach that comprised limited proteolysis, matrix-assisted laser desorption ionization time-offlight (MALDI-TOF) MS, and CD to identify rigid structural cores and inductively coupled plasma atomic emission spectroscopy (ICP-AES) and NMR spectroscopy for structural and functional characterization. We found that only the RRM and C4 type zinc finger domains fold into the defined core architectures in the isolated TLS protein. The TLS zinc finger domain strongly binds an oligo-RNA, whereas tight interaction of the RRM domain with the RNA was not detected. Furthermore we identified several zinc finger domain residues that contribute to the interaction with RNA. Our present results suggest that other RNA-binding proteins, such as ZNF265, which contains the same type of zinc finger (20), may also associate with RNAs in a manner similar to that of the TLS zinc finger domain.

EXPERIMENTAL PROCEDURES
Expression and Purification of TLS-The human TLS cDNA was cloned into the pGEX6P-1 vector between the BamHI and XhoI sites for expression as an N-terminal GST fusion protein. The DNA fragment encoding residues 171-526 of TLS was also cloned into the vector for an N-terminal truncated mutant protein. In this study, we refer to the TLS full-length fusion protein as GST-TLS-F and the construct containing residues 171-526 of TLS as GST-TLS-C. Escherichia coli strain BL21(DE3) competent cells were transformed with the vectors, and the transformants were grown at 20°C in LB medium containing ampicillin (0.1 mg/ml). Protein expression was induced at an A 600 ϭ 0.8 by the addition of isopropyl-␤-D-thiogalactopyranoside to a final concentration of 0.15 mM. Then the cells were grown further for 22 h at 20°C and harvested by centrifugation (6,000 ϫ g for 15 min). The cells were suspended in a buffer containing 50 mM Tris-HCl, pH 8.0, 1 mM dithiothreitol, 10 g/ml RNase A, and protease inhibitor mixture (Complete, Roche Applied Science) and lysed by sonication (ASTRASON) at 4°C. The supernatants containing the expressed proteins were clarified by ultracentrifugation (150,000 ϫ g for 30 min), and then the proteins were purified on a GSTrap FF affinity column (Amersham Biosciences).
Limited Proteolysis and Fragment Purification-Proteolysis was carried out at 20°C in a solution containing 50 mM Tris-HCl, pH 8.0, 1 mM dithiothreitol, 1 mM CaCl 2 , 3.0 mg/ml GST-TLS-F or GST-TLS-C, and various concentrations of chymotrypsin. After a 20-min incubation, the reaction was quenched by the addition of SDS-PAGE sample buffer. The reactants were analyzed by SDS-PAGE using a 15-25% gradient gel. For CD measurements, the digestion was carried out on the GSTrap FF columns, and the fragment solution was subjected to HiTrap Benzamidine FF (Amersham Biosciences) to remove the chymotrypsin from the solution followed by a treatment with 2 mM 4-amidinophenylmeth-anesulfonylfluoride to inactivate the remaining protease. Fragments were separated by HiLoad 16/60 Superdex 75 pg gel filtration chromatography (fast protein liquid chromatography system, Amersham Biosciences) in a buffer containing 20 mM potassium phosphate, pH 7.0, and 300 mM NaCl. Several fragments were further isolated by cation exchange column chromatography (HiTrap SP HP, Amersham Biosciences) in a buffer containing 20 mM potassium phosphate, pH 7.0, and 100 mM NaCl. Fragments absorbed on the SP column were eluted with a buffer containing 20 mM potassium phosphate, pH 7.0, and 1 M NaCl. The protein concentrations were determined by using a colorimetric kit (Protein Assay, Bio-Rad) and also by absorbance at 280 nm (see also Table I). All of the chemicals used were of the highest grade commercially available.
MALDI-TOF MS-Spectra were acquired on a MALDI-TOF mass spectrometer (Voyager, Applied Biosystems) equipped with a 337 nm nitrogen laser, a 20-kV extraction voltage, and a time-delayed extraction. A mixture of saturated 3,5-dimethoxy-4-hydroxycinnamic acid (Aldrich) in 33% acetonitrile and 0.1% trifluoroacetic acid was used as the matrix. The GST-TLS-C digestion was stopped by the addition of acetonitrile containing 0.1% trifluoroacetic acid. The resulting solution was dried and dissolved in 5 l of 80% acetonitrile and 0.1% trifluoroacetic acid. The fragment solutions were mixed on a plate with the matrix in a 1:1 ratio and dried at room temperature prior to mass analysis. Insulin chain B and myoglobin were used as external standards for the molecular mass calibration.
CD Spectrometry-CD spectra of GST-TLS-F and the fragments were measured on a J-720 spectropolarimeter (Jasco) at 20°C. Samples (10 M) were placed in a fused silica cell (1-mm path length), and 25 scans were averaged. Secondary structure contents of GST-TLS-F and the fragments were calculated by the program CONTIN (21) using the CD spectra in the wavelength range of 195-250 nm.
N-terminal Sequencing-The N-terminal amino acid sequences of the fragments were determined by automated Edman degradation using an Applied Biosystems 492 pulsed liquid phase sequencer equipped with an on-line 785A phenylthiohydantoin derivative analyzer.
ICP-AES-The zinc content of the TLS zinc finger-like domain (Fr3-ZnF) was determined with an Optima3000 ICP-AES instrument (PerkinElmer Life Sciences) at the Advanced Instrumentation Center for Chemical Analysis, Ehime University (22).
NMR Spectroscopy-E. coli cells harboring a TLS protein expression plasmid were cultured in M9 minimal medium containing 15  Two dimensional 1 H-15 N heteronuclear single quantum coherence (HSQC) spectra and a set of three-dimensional triple resonance NMR spectra were acquired at 30°C on Bruker 750 or 600 MHz NMR spectrometers. To obtain the resonance assignments for the backbone 1 H, 15 N, 13 C␣, 13 C␤, 13 CЈ nuclei, the following spectra were used: HNCO, HNCA, HN(CO)CA, HNCACB, and CBCA(CO)NH (23). All data were processed with the program NMRPipe (24). The program PIPP (25) was used to pick peaks for each spectrum. The resultant peak tables were subjected to an in-house, semiautomatic backbone resonance assignment program, JASS (Java script-based graphical backbone resonance assignment tool), 2 to accelerate the backbone assignment process. The 1 H chemical shifts were referenced indirectly to external 2,2-dimethyl-2-silapentane-5-sulphonic acid (DSS) as 0.0 ppm. Zero frequencies for 13 C and 15 N chemical shifts were derived from the experimentally obtained 1 H frequency for DSS by multiplying by the relative frequencies of 0.251449530 and 0.101329118, respectively (26).
Chemical shift perturbations according to the RNA titration for the 1 H and 15 N resonances of the TLS zinc finger fragment were collected from a series of 1 H-15 N HSQC spectra measured on a 750 MHz spectrometer at 30°C. The molar ratios of RNA to TLS in the titration experiments were set to 0.5, 1.0, and 2.0. The 15 N-labeled TLS protein concentration was kept at 0.1 mM in all titration experiments. Peak positions were elucidated by using contour simulation in the program PIPP, which calculates the peak position as an average of centers of simulated contour circles for all displayed contour levels (25). Onedimensional 113 Cd 2ϩ NMR spectra were collected at 30°C on a Bruker 500 MHz NMR spectrometer equipped with BBO z-axis gradient 1 H-BB probes.
Molecular Modeling of the TLS Domains-The search for homologous sequences was performed within the Protein Data Bank (27) using National Center for Biotechnology Information BLAST (28). Sequences were aligned by using the program ClustalW (29) and were amended manually based on the secondary structure prediction results generated by the programs 3D-PSSM (30) and PredictProtein (31). The structural model of the TLS domain (residues 422-453) was built by comparative modeling from the human ZNF265 (Protein Data Bank entry 1NOZ) by using a combination of Swiss-PdbViewer (32) and SWISS-MODEL (33).

Limited Proteolysis of TLS and Fragment Assignments-A
limited proteolytic analysis was carried out to identify the domains with defined architectures and unstructured regions using two purified proteins, GST-TLS-F (full-length TLS fused to GST) and GST-TLS-C (C-terminal TLS fused to GST). Cleavage by chymotrypsin yielded clear bands by SDS-PAGE (Fig.  1A), whereas the trypsin treatment produced many bands with various sizes (data not shown). The generation of these discrete bands was independent of the chymotrypsin concentration, implying that the bands are derived from protease-resistant fragments (Fig. 1A). The SDS-polyacrylamide gel patterns of the products were similar between the GST-TLS-F and GST-TLS-C digestions, indicating that the chymotrypsin-resistant core regions are located at the C terminus of TLS. To assign these fragments, we fractionated three fragments from GST-TLS-C (Fr1, Fr2, and Fr3 with 14-, 12-, and 8-kDa masses, respectively) that were produced by the on-column cleavage using the GSTrap FF column. The 8-kDa Fr3 fragment was isolated on the gel filtration column, whereas Fr1 and Fr2 were  separated by the subsequent cation exchange SP column (Fig.  1B). The 12-kDa Fr2 fragment passed through the column, whereas the absorbed 14-kDa Fr1 was eluted with buffer containing 20 mM potassium phosphate, pH 7.0, and 1 M NaCl. Next we examined the N-terminal sequences of these three fragments. The analyses revealed that both Fr1 and Fr2 have a GGPRDQG N-terminal sequence, and Fr3 has GGGSGG (Fig.  1C). The molecular masses of these fragments were then measured by MALDI-TOF MS (Fig. 1C and Table I). Taken together, these results demonstrate that Fr1 and Fr2 contain the RRM domain, while Fr3 bears the zinc finger-like domain. Thus, in agreement with the sequence analysis, we can conclude that only the RRM and zinc finger-like domains form rigid core structures in TLS.
The RRM and Zinc Finger-like Domains Exclusively Form Rigid Core Structures in TLS-We measured the CD spectra of the two fragments containing the RRM (Fr2-RRM) and putative zinc finger (Fr3-ZnF) regions (Fig. 2, inset) and the purified GST-TLS-F (Fig. 2) in addition to purified GST alone. The difference spectrum, obtained by subtracting the summed spectra of the other three components from GST-TLS-F itself, exhibited a random coil-like pattern (Fig. 2, broken curve), suggesting that the regions other than the RRM and zinc fingerlike domains are unstructured. The calculation of secondary structure contents, using the deconvolution program CONTIN (21), provided 15% ␣-helix and 25% ␤-strand for the RRM domain and 5% ␣-helix and 17% ␤-strand for the zinc fingerlike domain. These values for the RRM domain are essentially consistent with those estimated from the RRM secondary structure prediction (discussed below). In the case of the zinc fingerlike domain, the values agree well with those determined by the three-dimensional triple resonance NMR spectrum (discussed below). These results additionally support the proposal that the defined structures in TLS are confined to the RRM and zinc finger-like domains.
Structural Characterization of the Zinc Finger-like Domain-We used ICP-AES to analyze the zinc content in Fr3-ZnF, which is actually bound to the zinc finger-like domain, and found that the fragment contains the zinc atom in a 0.7 stoichiometry, which is somewhat lower than the expected value. This could be due to a slightly inaccurate protein concentration and/or zinc content determination.
To characterize the coordination scheme of the zinc atom, we next measured the 113 Cd 2ϩ NMR spectra of Fr3-ZnF in which the zinc atoms were replaced by cadmium, since 113 Cd 2ϩ NMR is the most powerful method to reveal a Zn 2ϩ coordination. First we confirmed that the 113 Cd 2ϩ -substituted zinc finger domain retains the same main chain folding as the native zinc finger as determined from the 1 H-15 N HSQC spectra with almost constant chemical shifts (data not shown). The 113 Cd 2ϩ one-dimensional NMR spectrum showed a clear peak at 678 ppm (Fig. 3). This 113 Cd chemical shift value is consistent with those observed in other tetrahedral tetrathiolate cadmium coordination sites (34). Taken together, these results prove that the core region, from residues 398 to 468, forms the C4 type zinc finger.
Interaction of the RRM and Zinc Finger Domains with RNA-It was reported that the C-terminal half of TLS with the RRM and zinc finger domains binds RNA containing a preferred RNA sequence (GGUG) with a submicromolar order dissociation constant (17). The RRM and the RGG repeats were also considered to be responsible for RNA binding. On the other hand, the ZNF265 zinc finger domain, which shares sequence similarity with the TLS zinc finger, can bind mRNA (20). Therefore, it has remained unclear whether either or both domains are responsible for RNA binding in TLS. Thus, we attempted to clarify which domain is more important for the binding of an RNA oligomer (5Ј-UAGUUUGGUGAU-3Ј) using a backbone amide NMR chemical shift perturbation analysis (Fig. 4, A and B). Upon the titration of 15 N-labeled Fr2-RRM with 0 -2 equimolar amounts of the unlabeled RNA, no significant spectral changes were observed in the RRM, indicating that a tight interaction does not occur between the RRM and the GGUG-containing RNA. A series of 1 H-15 N HSQC spectra of 15 N-labeled Fr3-ZnF were also collected upon titration from 0 to 2 equimolar amounts of RNA. In contrast to the results with RRM, 23 of the 25 signals from the amide residues in the zinc finger domain showed remarkable chemical shift changes or broadening (Fig. 4B). Most of the NMR signals displayed in Fig. 4B are derived from the structured zinc finger domain because the residues from unstructured G-rich portions yield broadened signals under these conditions. We assigned the backbone resonances from all of the residues in the zinc finger region, including amide 1 H, 13 C␣, 13 C␤, and carbonyl carbon, although Asn-442 was not assigned due to the severe overlap of its 1 H-15 N correlation signal with unassigned signals derived from residues outside of the zinc finger domain (see Supplemental Data). The signals are shown in the spectrum with their assigned residue numbers in Fig. 4B. The assigned backbone 13 C chemical shifts predict that the zinc finger consists of two ␤-strands and one ␣-helix (35). This result is essentially consistent with the homology model from the alignment between the TLS and ZNF265 sequences. Fig. 4C shows the weighted possibility of a functional contact between the RRM and zinc finger domains in TLS, we examined their interactions by backbone amide NMR chemical shift perturbation and chemical cross-linking analyses. A series of 1 H-15 N HSQC spectra of 15 N-labeled Fr3-ZnF obtained through the titration from 0 to 2 equimolar amounts of an unlabeled fragment with Fr2-RRM revealed no significant interaction between the two domains in either the presence or absence of RNA (data not shown). Similarly no significant complex consisting of a single RRM domain and a single zinc finger domain was detected in the crosslinking experiments (data not shown). These results suggest that these two domains lack specific interactions.

DISCUSSION
Functional Implications of the RRM Domain-In general, RRM domains share the same topology (␤1-␣1-␤2-␤3-␣2-␤4), and the nucleic acid binding sites in the individual RRMs are located on the concave face of the ␤-sheet. Several atomic structures of RRMs have been determined in complexes with nucleic acids (37)(38)(39)(40)(41)(42). In these complex structures, the binding modes are diverse because of the large variety of nucleic acids in DNA or RNA sequences, whereas the structural aspects of the binding share common features, i.e. the utilization of the conserved ribonucleoprotein (RNP) motifs 1 (on ␤-strand 3) and 2 (␤-strand 1) in contact with dinucleotides. Recent structural studies have revealed that a number of RRMs recognize and bind specific proteins predominantly by using RNP motifs with slightly modified signature sequences (43)(44)(45)(46).
Although it was previously reported that oligo-RNAs were bound to a TLS RRM construct containing a C-terminal RGG box (17), we did not observe a significant interaction between the isolated RRM with the GGUG-containing RNA. This discrepancy could be attributed to the absence of the RGG repeats in our TLS RRM fragment obtained from the proteolytic digestion. RGG repeats may contribute toward enhancing the RRM binding affinity for RNA. We next carried out homology modeling based on the known atomic structures of RRMs and a multiple sequence alignment (data not shown). Our model showed reasonable features in terms of the steric hindrance and the amino acid distributions in the core and surface regions and exhibited secondary structure contents consistent with those estimated from CD spectroscopy. However, the amino acid distribution of this model did not allow us to conclude whether the TLS RRM indeed binds RNA.
Functional Implications of the Zinc Finger Domain-The TLS zinc finger domain was predicted from its primary sequence, which is categorized as a Ran-binding protein zinc finger type. The Ran-binding protein zinc finger itself binds the GDP-bound form of the Ran GTPase and participates in trafficking through the nuclear membrane (47). This type of zinc finger domain is present in more than 200 proteins with various domain organizations and functions. Among them, the three-dimensional structures of the ZNF265 and Npl4 zinc fingers have been determined by NMR (20,48). Although their binding targets differ from each other, they share essentially the same architecture consisting of two distorted ␤-hairpins, which each devote two cysteines to the zinc coordination. The ZNF265 zinc finger domain is able to bind cyclin mRNA (20), while that of Npl4 specifically binds to ubiquitin (48). In the ZNF265 zinc finger domain, however, the amino acid residues responsible for the protein-RNA interaction have not been identified. Our present study demonstrated that the TLS zinc finger domain in fact coordinates the zinc ion through the four conserved cysteines. Moreover our NMR study showed that the zinc finger is able to bind a GGUG-containing RNA, and the first and second ␤-strands are mainly involved in RNA binding. Therefore, we carried out model building based on the homology with the ZNF265 zinc finger domain structure, which shows a 44% identity to that of TLS, to investigate which surface in the zinc finger domain is involved in the RNA binding (Fig. 5A).
The legitimacy of the model building is strengthened from the nearly identical positioning of one tryptophan, four cysteines, and one asparagine, which are highly conserved among more than 200 putative Ran-binding protein type zinc finger sequences. The distribution of the regions with chemical shift perturbations is depicted on the tertiary structure model of the TLS zinc finger (Fig. 5B). The residues that exhibit large chemical shift changes or broadening are characteristically localized in the first and second ␤-strands on the opposite side of the zinc-coordinated position. Consistent with this interpretation, the binding sites identified by chemical shift mapping form a basic region on the electrostatic surface potential map of the domain (Fig. 5C). The amino acids that are highly conserved among the TLS-related proteins (TET family and SARFH) and human ZNF265 zinc fingers are shown in Fig. 5D. Several residues that contribute to RNA binding are conserved in these zinc fingers, suggesting that the corresponding residues in other proteins may be involved in RNA binding. Indeed some residues lying in the corresponding region of the ZNF265 zinc finger have been suggested to participate in RNA binding (20).
In contrast, the Npl4 zinc finger domain specifically binds to ubiquitin through the Thr-Phe dipeptide adjacent to the second cysteine (48). In the TLS zinc finger domain, this dipeptide is replaced by Glu-Asn, and in the homology model, both of them are exposed to the solvent. Therefore, it is unlikely that these residues within the TLS zinc finger contribute to ubiquitin binding. The zinc binding center of the Npl4 zinc finger is also structurally related to the Fe 2ϩ binding site of the rubredoxin protein family, particularly to rubrerythrin. We examined this possibility by expressing and purifying GST-TLS-C from a Fe 2ϩ -enriched (50 M) solution. This TLS zinc finger domain was highly sensitive to proteolysis presumably because of misfolding and/or instability by the coordination with Fe 2ϩ (data not shown). Thus, we concluded that the Fe 2ϩ ion is not functional for the zinc finger domain of TLS.
Functional Implications of TLS-No interdomain interactions between the RRM and zinc finger domains were detected in the isolated TLS by our biochemical and structural experiments. These two domains are connected via a flexible, glycinerich linker, suggesting that the domains can alter their relative locations. This flexible linker may facilitate the targeting and trapping of other distantly located proteins and/or RNA modules.
Interestingly TLS was reported to bind the N-methyl-D-aspartate receptor multiprotein complex isolated from the rat brain (49). It has been reported that a subset of mRNAs colocalizes with ribosomes and other translation apparatus components in the dendritic spines of neuron cells (50). The localization of mRNAs and the subsequent protein synthesis in close vicinity of the functional point would have crucial roles in regulating synaptic function. Considering the strong coupling between transcription and mRNA transport, it is not surprising that TLS, which plays major roles in these cellular processes, is broadly distributed from the nucleus to the cytoplasm. Indeed recent cell biological analyses using primary neurons have revealed that TLS is involved in mRNA transport within neuron dendrites and is translocated to dendritic spines by mGluR5 activation. 3 It will be intriguing to see whether TLS is directly involved in the mechanism of protein synthesis as RNA cargo at the synapse.