The DEAD-box RNA helicase DDX3 associates with export messenger ribonucleoproteins as well as tip-associated protein and participates in translational control

Abstract

Nuclear export of mRNA is tightly linked to transcription, nuclear mRNA processing, and subsequent maturation in the cytoplasm. Tip-associated protein (TAP) is the major nuclear mRNA export receptor, and it acts coordinately with various factors involved in mRNA expression. We screened for protein factors that associate with TAP and identified several candidates, including RNA helicase DDX3. We demonstrate that DDX3 directly interacts with TAP and that its association with TAP as well as mRNA ribonucleoprotein complexes may occur in the nucleus. Depletion of TAP resulted in nuclear accumulation of DDX3, suggesting that DDX3 is, at least in part, exported along with messenger ribonucleoproteins to the cytoplasm via the TAP-mediated pathway. Moreover, the observation that DDX3 localizes transiently in cytoplasmic stress granules under cell stress conditions suggests a role for DDX3 in translational control. Indeed, DDX3 associates with translation initiation complexes. However, DDX3 is probably not critical for general mRNA translation but may instead promote efficient translation of mRNAs containing a long or structured 5' untranslated region. Given that the DDX3 RNA helicase activity is essential for its involvement in translation, we suggest that DDX3 facilitates translation by resolving secondary structures of the 5'-untranslated region in mRNAs during ribosome scanning.

Figure 1.

DDX3 interacts with TAP in vivo and in vitro. (A) HeLa cells that transiently expressed HA-tagged TAP and TAPΔC proteins for 24 h were subjected to immunostaining with anti-HA. (B) In situ hybridization of poly(A)⁺ RNA was performed with HeLa cells that transiently expressed HA-tagged TAPΔC protein (arrowhead). (C) HEK293 cells were transiently transfected with empty vector (−) or the vector encoding FLAG-tagged TAP or TAPΔC (ΔC). Immunoprecipitation was performed at 24 h after transfection by using anti-FLAG. Immunoblotting was performed using polyclonal anti-DDX3 (top) and anti-FLAG (bottom). (D) Mock-transfected HEK293 cell lysate (−) or lysate containing transiently expressed FLAG-tagged DDX3 were subjected to immunoprecipitation as in C. Endogenous TAP was detected by anti-TAP. (E) Schematic diagram shows DDX3 fragments that were used for the pull-down assay. The assay using recombinant GST, GST-DDX3, or GST-DDX3 fragments as bait to pull-down ³⁵S-labeled TAP protein (autoradiogram). Coomassie Blue staining of the GST-fusion proteins is shown at bottom.

Figure 2.

DDX3 associates with TAP and mRNPs in both the nucleus and the cytoplasm. (A) Immunostaining of HeLa cells was performed using affinity-purified anti-DDX3. (B) HEK293 cells were transiently transfected with the vector expressing FLAG-tagged TAP or with empty vector (−). Immunoprecipitation of the nuclear (N) and cytoplasmic (C) fractions of the cell lysate was performed using monoclonal anti-FLAG, followed by immunoblotting to detect endogenous DDX3 and FLAG-TAP (top). The subcellular fractions were also subjected to immunoblotting using antibodies against nucleolin, lamin A/C, and α-tubulin (bottom). (C) Cell fractionation was performed in UV-irradiated (UV) or mock-treated (−) HEK293 cells as in B. The lysate fractions were then subjected to chromatography over poly(U)-Sepharose. Proteins cross-linked to poly(A)⁺ RNAs were detected by anti-DDX3 (top). The efficiency of cell fractionation was also determined as in B (bottom). (D) The TAP-targeting siRNA was transiently transfected into HEK293 cells. Immunoblotting of the cell lysate was performed using anti-TAP (left). UV-cross-linking and poly(U)-Sepharose-affinity selection were performed as described in C, except that total cell lysate was used.

Figure 3.

DDX3 may be in part exported to the cytoplasm via the TAP-mediated pathway. (A) HeLa cells were treated with 40 nM leptomycin B (LMB) or mock-treated (mock) for 3 h and subsequently subjected to immunostaining by using anti-DDX3 and anti-cyclin B1. DNA staining with Hoechst 33258 indicates the cell nucleus. (B) Knockdown of TAP expression and cell fractionation were performed in HEK293 cells as described in Figure 2, D and B, respectively. Immunoblotting was carried out using antibodies against DDX3, lamin A/C, and α-tubulin.

Figure 4.

DDX3 is a component of stress granules. (A) HeLa cells were transiently transfected with a vector encoding GFP-DDX3 for 24 h. Subcellular localization of GFP-DDX3 was detected by fluorescence microscopy. (B) HeLa cells were treated with 0.5 mM sodium arsenite (arsenite) or mock-treated (mock) for 1 h. Cells were subsequently subjected to immunostaining using anti-DDX3. (C) HA-tagged TIA-1, HuR, and TAP were each transiently coexpressed with GFP-DDX3 in HeLa cells. Immunostaining was carried out using monoclonal anti- HA to detect HA-tagged proteins. B and C were obtained using a laser confocal microscope.

Figure 5.

Association of DDX3 with translation initiation complexes. (A) FLAG-tagged TAP, DDX3 or Dbp5 was transiently coexpressed with HA- or myc-tagged eIF4A, eIF2α, PABP1, and TIA-1 in HEK293 cells. Immunoprecipitation was performed using anti-FLAG, followed by immunoblotting using antibodies against the HA, myc, or FLAG epitope. Ig H represents the immunoglobulin heavy chain. The input of FLAG-tagged TAP and DDX3 was visible upon longer exposure (lanes 1 and 2). (B) HEK293 cytoplasmic extract was centrifuged through a continuous 15–40% sucrose gradient. The polysome profile was plotted by A₂₅₄ values (top). RNAs and proteins were recovered from 22 fractions for analysis (middle). RNAs were resolved on a 1% formaldehyde/agarose gel and rRNAs were visualized by ethidium bromide staining. Immunoblotting was performed using anti-DDX3. Immunoblotting of the first 12 fractions is shown at the bottom using antibodies against eIF4G, DDX3, eIF5, eIF2α, and eIF4E. (C) GFP-DDX3 was transiently expressed in HeLa cells. Immunostaining was carried out using antibodies against eIF4A, eIF4E, or PABP1.

Figure 6.

DDX3 is dispensable for general translation. (A) HeLa cells were cotransfected with the Renilla luciferase vector (the scheme) and the empty pSilencer vector (mock) or the vector expressing DDX3 shRNA1 or shRNA2 or TNPO3 shRNA (a control shRNA, lane 4) for 48 h. The bar graph shows the average Renilla luciferase activity in each transfection (lanes 2–4) relative to that of the mock transfectant (lane 1). Average values with SD were obtained from three independent experiments. Immunoblotting was performed using anti-DDX3 and anti-α-tubulin. (B) HeLa cells were transfected with pSilencer (mock) or the vector expressing DDX3-targeting shRNA1. DDX3 and α-tubulin were detected by corresponding antibodies (top). The transfectants were subjected to the pulse-labeling assay. Autoradiography shows radiolabeled proteins of the cell lysates (bottom).

Figure 7.

DDX3 is required for efficient translation initiation of mRNAs harboring a long or structured 5′-UTR. (A) Schematics illustrate the luciferase reporters and the possible secondary structures for the 5′-UTR in the engineered pFL-SV40 reporters. (B) Transient transfection was performed essentially similar to Figure 6A, except that an FL reporter (hairpin, TGFβ, or ODC) and pRL-SV40 were cotransfected. The averaged FL activity was normalized to that of the RL activity in individual transfectants. The bar graph shows the average FL/RL value of each transfection relative to that of the mock transfectant. The data were obtained from three independent experiments. A representative immunoblot is shown at the bottom. (C) The ODC-containing FL reporter was used. Lane 1 is similar to lane 1 of B. Lanes 2–4, transfection was performed similarly to lane 2 of B, except that each transfection contained one additional vector (lane 2, empty pcDNA3.1 vector; lane 3, vector encoding shRNA-resistant wild-type DDX3; lane 4, vector encoding a DDX3 helicase mutant, S382L). The bar graph and immunoblot are presented similarly to B. (D) The in vitro translation was performed in the reticulocyte lysate using in vitro transcribed RL control mRNA from pRL-SV40 (Figure 6A) and TGFβ1 5′-UTR–containing FL mRNA as reporter in the presence of recombinant GST-DDX3. Luciferase activity of each reaction was normalized to the control reaction without DDX3.