Removal of an inhibitory effect of the Drosophila BicD CTD is required for efficient transport of heterologous cargoes. (A) Images of syncytial blastoderm embryos of the indicated genotypes 10–14 min after injection with MS2x6 RNA (top panels) or following staining with anti-MS2 coat protein (CP) antibody (bottom panels; uninjected embryos). The apical cytoplasm (arrow) is a thin layer above the row of nuclei. (wt) Wild type (yw strain) in this and other figures. Bar, 50 μm. (B) Quantification of MS2x6 RNA localization efficiency. The top middle and top right panels in A are representative examples of weak and strong localization, respectively. Embryos were injected with antibodies to dynein intermediate (Dic) or heavy chain (Dhc) 2–5 min before RNA injection as indicated, or just injected with RNA (−). Injection of buffer alone or control antibodies has no effect on RNA localization efficiency (Bullock and Ish-Horowicz 2001).

The localization driven by BicDΔC∷CP is dependent on endogenous BicD and Egl. (A) The α-BicD-C antibody 1B11 does not recognize the truncated BicD in extracts of BicDΔC∷CP embryos. The expected position of BicDΔC∷CP is marked by the α-BicD-N antibody 4C2, which binds an epitope between amino acids 1 and 543. (B) Localization of MS2x6 RNA by BicDΔC∷CP is strongly reduced by preinjection of α-BicD-C, α-Egl, or by halving egl gene dosage. The data for MS2x6 localization in uninjected BicDΔC∷CP embryos is reproduced from Figure 1A for comparison. (C) Recruitment of Egl to apically enriched MS2x6 RNA in BicDΔC∷CP embryos. Bar, 50 μm.

Egl interacts directly with the BicD CTD through amino acids 1–79 and does not have a general function in BicD-mediated transport. (A) Yeast two-hybrid assays of interactions of the BicD CTD with full-length BicD, Rab6, and Egl. AD and BD, fusion to GAL4 activation, and DNA-binding domains, respectively. Interactions reconstitute the expression of Ade and His and result in growth of colonies on selective plates. The interactions between Egl and BicD were also confirmed by swapping the identity of the proteins in the AD and BD vectors (see Supplemental Table 1 for these and other interactions assayed). (B) Assessment of the binding of in vitro translated ³⁵S-methionine-labeled Egl fragments to the BicD CTD, fused to GST, or GST alone (GST). Radioactive Egl signals are shown. Load represents 10% of that used in each pull-down assay. (C) Egl is coimmunoprecipitated with BicD∷GFP, but not BicD^K730M∷GFP, from transgenic ovary extracts (using α-GFP). Load represents 2% of that used in each immunoprecipitation assay. (D) UAS-driven overexpression of BicD (with the GAL4VP16^nos driver [NGV]) prevents basal redistribution of lipid droplets, leading to an opaque cytoplasm. Overexpression of Egl leads to excessive basal movement and an almost transparent cytoplasm. Nile red staining of neutral lipids confirmed that this is due to a redistribution of droplets and not a gross change in the amount of droplets in the embryo (data not shown). The effect on Egl overexpression is suppressed by mild overexpression of full-length BicD (tub-BicD∷CP). Arrow marks the junction of the yolk (y) and cytoplasm (c) in the blastoderm embryo. Supplemental Figure 1B shows protein levels in the different scenarios. (E) Nuclei (marked by Elav) are mislocalized in the optic stalk (os) and have an irregular apico–basal distribution in the retina (left) in third instar mutants for BicD, but not egl. Zygotic mutant genotypes are shown; no maternal or zygotic Egl and BicD protein could be detected in the respective mutants at this stage of development (X Li and S Bullock, in prep.). Bars: D, 50 μm; E, 35 μm. (F) Gel filtration of wild-type embryo extract in the presence or absence of RNaseA. Data shown are fractions of total Western blot signals for Egl or BicD from individual 4-mL fractions. Numbers on the top correspond to molecular weight of globular protein standards extrapolated from column calibration. We determined that the gel filtration profile of a control protein (α-tubulin) was not altered by RNaseA treatment (data not shown). The majority of Egl, but <20% of BicD, is present in a relatively high-molecular-weight complex whose size is sensitive to RNaseA treatment.

Egl and BicD are the only detectable factors stably and specifically enriched from embryonic extracts on active minus-end-directed localization signals. (A) Quantification of apical localization activity of wild-type and mutant localization elements in the context of fusion to the streptavidin aptamer. Total number of embryos scored for each RNA is shown. (as) Antisense; (−) aptamer alone. Representative images of embryos injected with localizing and nonlocalizing fusion transcripts are shown below. (B) Sypro Ruby stained gel showing detectable enrichment of only Egl and BicD from embryonic extracts on the wild-type K10 TLS compared with on an antisense TLS (MS, mass spectrometry IDs). The RNA-binding proteins Yps and CG14648, as well as proteins that bind the streptavidin matrix (asterisks), are present in similar amounts in experiments with wild-type and mutant elements. Yps (Mansfield et al. 2002) and CG14648 (S Bullock, unpubl.) do not appear to participate in Egl and BicD-dependent mRNA transport in vivo and are therefore likely to be nonspecific RNA-binding factors. Conducting the experiment with extracts that express Egl∷GFP in addition to endogenous Egl confirms the identity of Egl as one of the factors specifically retained on localizing elements. (B′) Western blot of samples in B confirming the identity of Egl and BicD as the proteins enriched preferentially on active TLS elements. Load lane is from Egl∷GFP extracts and represents 2% of input into RNA-binding assays. Additional TLS mutants are shown in C. (D) Silver-stained gel showing that Egl and BicD are also the only factors visibly and reproducibly enriched from wild-type embryonic extracts on minimal localizing signals from h (hSL1), grk (GLS), and I factor (ILS) relative to nonlocalizing controls. Although some additional differential bands are seen on hSL1 and GLS relative to controls, these were not reproducible. Western blots are shown in D′. (E) Quantification of relative amounts of Egl and BicD captured from wild-type embryonic extracts on RNAs as indicated. Data are quantified from fluorescent Western blot signals normalized to the wild-type TLS. The number above the bars is the number of independent experiments. Error bars represent SEM.

Recombinant Egl selectively recognizes several mRNA localization elements in vitro. (A,B) In vitro translated (IVT), ³⁵S-methionine-labeled proteins, as indicated, were tested for recruitment to streptavidin/aptamer-immobilized RNAs. Images show radioactive signals. Load represents 5% of that used in each RNA-binding assay. (C) Quantification of recruitment of IVT-Egl to RNAs, normalized to the wild-type TLS. Number of independent experiments is indicated above bars; error bars represent SEM. (D) Recombinant Egl alone, Egl:BicD complex, or BicD alone, overexpressed and purified from yeast, tested for recruitment to streptavidin/aptamer-immobilized RNAs. Images show Western blots. Load represents 10% of that used in each RNA-binding assay. (E) Quantification of specific recruitment of yeast-expressed Egl–BicD complex to RNAs (using Western blot signals), normalized to the wild-type TLS. Number of independent experiments is indicated above bars; error bars represent SEM. (F) EMSA showing the increased affinity of the yeast-expressed Egl–BicD complex for a wild-type (TLS) compared with a mutant stem–loop (TLS^as) (both in the context of the streptavidin aptamer). Top band is ribonucleoprotein (RNP) complex. (G) EMSA showing a higher concentration of cold TLS^as than cold TLS is needed to compete equivalently (open arrowheads) with ³²P-labeled-TLS RNA for Egl–BicD. A 50 nM Egl-BicD was loaded in each assay. Each RNA was linked to the streptavidin aptamer.

Egl binds RNA through amino acids 1–814 independently of putative exonuclease activity. (A) Representative examples of the recruitment of ³⁵S-methionine-labeled, in vitro translated Egl constructs to streptavidin-immobilized active (TLS and ILS) and inactive (TLS^U6C and ILS^as) RNA elements. (B) Summary of mapping of Egl RNA-binding features from at least two independent experiments for each construct. (+) RNA binding; (−) no/very weak RNA binding; (nd) not determined. Numbers indicate Egl amino acid position. Mutation of two residues essential for catalytic activity of canonical 3′–5′ exonucleases (Y704 and D708) does not interfere with binding of full-length Egl to RNA. Equivalent molar concentrations of in vitro translated proteins were used in each experiment.

Model for a complete link between minus-end-directed mRNA localization signals and microtubules. See the text for details. The arrows show putative interactions of BicD with the dynein/dynactin complex (Hoogenraad et al. 2001, 2003). For simplicity, single copies of the mRNA signal, Egl and BicD, are shown, but mRNAs can be transported as particles (Delanoue et al. 2007) and dimerization or oligomerization of BicD and Egl may be significant for achieving a transport-competent complex. In addition to the factors shown, generic RNA-binding proteins and a dedicated plus-end-directed motor could also be involved. Dlc, Dic, and Dhc are components of the dynein complex implicated in the mRNA transport complex to date.