|Home | About | Journals | Submit | Contact Us | Français|
Extracellular signals regulate protein translation in many cell functions. A key advantage of control at the translational level is the opportunity to regulate protein synthesis within specific cellular subregions. However, little is known about mechanisms that may link extracellular cues to translation with spatial precision. Here we show that a transmembrane receptor, DCC, forms a binding complex containing multiple translation components, including eukaryotic initiation factors, ribosomal large and small subunits, and monosomes. In neuronal axons and dendrites DCC colocalizes in particles with translation machinery, and newly synthesized protein. The extracellular ligand netrin promoted DCC-mediated translation and disassociation of translation components. The functional and physical association of a cell surface receptor with the translation machinery leads to a generalizable model for localization and extracellular regulation of protein synthesis, based on a transmembrane translation regulation complex.
Transmembrane receptors can provide a direct link between extracellular signals and intracellular machinery. It has long been known that receptor cytoplasmic domains can associate with two types of intracellular machinery: the cytoskeleton, and cytoplasmic signaling proteins (Flanagan and Koch, 1978; Rodbell, 1980; Ullrich and Schlessinger, 1990). From there, regulatory information can be conveyed by signaling pathways to regulate functional outputs throughout the cell, including gene expression at both transcriptional and translational levels (Pawson and Nash, 2003).
One key advantage of controlling gene expression at the translational level is that this allows protein synthesis to be spatially localized to specific subregions of the cytoplasm. This localization of protein synthesis is thought to be important in diverse biological functions, such as setting up the primary axes of the Drosophila embryo, cell migration and adhesion, axon guidance, and regulation of the synapse (St Johnston, 2005; Sutton and Schuman, 2006; Lin and Holt, 2008; Rodriguez et al., 2008). Moreover, genome scale analysis in the early Drosophila embryo indicates that most mRNAs show localization to diverse and specific subcellular sites, including at the cell membrane, implying that mechanisms to regulate localized translation participate in many, if not most, cell functions (Lecuyer et al., 2007). Regarding mechanisms for translational regulation, a great deal of progress has been made in identifying ribonucleoprotein complexes that mediate mRNA localization and translational control (St Johnston, 2005; Kiebler and Bassell, 2006; Rodriguez et al., 2008), and molecular pathways that transmit signals from cell surface receptors and regulate the translational machinery (Tee and Blenis, 2005; Sonenberg and Hinnebusch, 2007). However, it is still not well understood how extracellular signals may regulate translation with spatial precision.
Translation can be divided into three main stages: initiation, elongation and termination. The primary site for translational regulation in most systems is the initiation step, and therefore key roles in translational control are played by components involved in initiation (Sonenberg and Hinnebusch, 2007). During initiation, the ribosomal 40S and 60S subunits join the mRNA in a series of steps mediated by eukaryotic initiation factors (eIFs). The small subunit forms the 43S preinitiation complex by recruiting the initiator tRNA, bound to eIF2, a protein that is an important target of translational control. This complex then recruits mRNAs with a 5' cap through the action of eIF4E, a cap-binding protein that is a key mediator of multiple pathways of translational control. After scanning to the initiator AUG codon, the small subunit is joined by the large subunit to form the monosome, and elongation can then proceed on polysomes.
In neurons, roles for protein translation have been identified in both dendrites and axons. Developing axons can translate protein, and roles for protein synthesis have been proposed in axon growth, guidance, and regeneration. Within the axon, there can be further localization of newly synthesized proteins, preferentially within one segment along the length of an axon, or on one side of a growth cone, exposed locally to an extracellular cue (Campbell and Holt, 2001; Zhang et al., 2001; Brittis et al., 2002; Ming et al., 2002; Leung et al., 2006; Yao et al., 2006; Willis et al., 2007). In dendrites, protein synthesis is well established to have roles in synaptic plasticity, including learning and memory. Proteins can be synthesized locally within the dendrite in response to extracellular signals, and localization of translation at or near individual synapses may play a role in plasticity that is synapse-specific (Steward and Schuman, 2003; Sutton and Schuman, 2006).
To explain how protein synthesis could occur preferentially in the subcellular region of a neuron receiving an extracellular signal, one model is that transmembrane receptors would operate through signaling pathways that act at a distance, influencing translation with a range extending over part of the cell. We set out here to test an alternative, although not mutually exclusive, model based on the precedent of transmembrane receptor association with other types of machinery inside the cell. In this model, transmembrane receptors that regulate protein synthesis would colocalize with translation machinery in a particle where they would be integrated in a molecular binding complex. To test this second model, we focused on DCC, a well characterized receptor in axon growth and guidance (Keino-Masu et al., 1996; Fazeli et al., 1997), and also found postsynaptically in dendrites (Parent et al., 2005). DCC is one of the receptors for the extracellular factor netrin (Keino-Masu et al., 1996; Fazeli et al., 1997), and netrin is known to promote protein synthesis in axons (Campbell and Holt, 2001).
The results here show a physical complex of DCC with components involved in translation initiation, including eIFs, ribosomal subunits, and monosomes. Furthermore, DCC functionally mediated translational regulation in response to its ligand netrin-1. These studies lead to a generalizable model with transmembrane association of cell surface receptors and the translation machinery contributing to the specificity, efficiency, and spatially precise control of translation, based on a novel type of transmembrane complex regulated by extracellular cues.
In initial experiments to explore a potential association of DCC and translation machinery, we tested for colocalization in spinal commissural axon growth cones, where DCC promotes growth and chemoattraction in response to netrin (Keino-Masu et al., 1996; Fazeli et al., 1997). In particular, colocalization would be of interest at filopodial tips, where DCC is known to be located, and which are highly motile structures that explore the environment and respond to extracellular cues (Drees and Gertler, 2008). Immunofluorescence localization showed punctate labeling of both DCC and ribosomal markers, and overlap was seen in a subset of puncta, with a diameter of approximately 0.2 to 1 μm (Figures 1A-1D, S1A-S1C and S1H). Notably, colocalization was consistently seen at the tips of filopodia. Quantitation of the overlap within the growth cone indicated approximately 35% of DCC puncta overlapping with ribosomal puncta, and approximately 83% of ribosomal puncta overlapping with DCC puncta. When Pearson's correlation coefficient was calculated for individual DCC puncta that showed detectable ribosome labeling, in >90% of cases the value was between 0.5 and 0.95, confirming true colocalization of puncta rather than fortuitous overlap (see Experimental Procedures).
Colocalization was also tested in hippocampal neuron dendrites. DCC puncta colocalized with markers of the translational machinery, including at synapses identified by triple-labeling with the postsynaptic marker PSD-95 (Figures 1E-1H, S1D-S1G and data not shown). Approximately 35% of PSD-95 puncta overlapped with DCC puncta, while most but not all DCC puncta overlapped with markers of translational machinery; for example 89% of DCC puncta overlapped with eIF4E puncta. These results appear broadly consistent with electron microscopic studies reporting that polysomes are present in approximately 10-40% of synaptic spines (Ostroff et al., 2002; Steward and Schuman, 2003).
To complement the overall view of subcellular distribution provided by fluorescence imaging, electron microscopy was used for resolution on a molecular scale. Immunogold labeling showed colocalization of DCC and ribosome markers in axonal growth cones, axon shafts, near the tips of axon filopodia, and in dendritic synaptic spines (Figures 1I-1V). As expected for ultrathin sections (80 nm), labeling density was moderate, and some particles did not show colocalization (Figure S1I), also consistent with the fluorescence results showing colocalization of some but not all puncta. While the immunolocalization procedure does not allow detailed visualization of organelles, counterstaining of the cytoplasm showed that many of the colocalized particles were near the cell surface, while others were in intracellular locations where DCC may be associated with intracellular membranes, perhaps reflecting transport to or from the cell surface. It should be noted, however, that it is not possible here to accurately quantitate the proportion of surface versus intracellular location due to the sectioning geometry; or to determine the precise location of molecules relative to one another or the cell surface, since the gold particles are on a similar scale to the molecules and are tethered via antibodies that can span up to ~30 nm. Addressing colocalization, however, distances among clustered particles were generally 0-30 nm, well within the range for immunogold detection of a direct molecular interaction.
We next tested whether DCC and translational machinery might show physical association. Initially, to provide an unbiased sampling of candidate interactors, full length DCC or vector control were transfected into 293 cells, then DCC was immunoprecipitated, and coprecipitated proteins were identified by tandem mass spectrometry. As shown in Figure 2A, translational components were among the coprecipitated proteins, including translation initiation factors (polypeptides of eIFs 1, 2, 3, 4 and 5) and proteins of the ribosomal small subunit (S10, S13 and S24) and large subunit (L5, L10a, L13, L23, L28 and L38). Most of the proteins from this list were then tested by Western blot of transfected 293 cells, confirming specific co-precipitation in every case tested (Figures 2B and S2). In addition, coprecipitation of selected proteins was confirmed with embryonic spinal cord, providing a native developmental context where the proteins are expressed at endogenous levels (Figure 3B). Importantly, however, no co-precipitation was seen with a DCC protein lacking the cytoplasmic domain (DCC-Δcyto; Figure 3C), indicating that DCC is not associated with the translation machinery merely because the DCC polypeptide is itself being synthesized there, and indicating that these associations are mediated by the DCC cytoplasmic domain.
To investigate the interaction further, we performed sucrose gradient velocity sedimentation of ribosomes. DCC was found to co-sediment with both small and large ribosomal subunits, as well as monosomes, while little or no obvious DCC signal was seen in the polysome fraction under these conditions (Figures 3A and S3). In addition to 293 cells, a comparable distribution was obtained with native embryonic spinal cord (Figure 3B). On a per-subunit basis the association of DCC with the large and small subunits appeared approximately comparable (Figure 3A). Sedimentation profiles of the key translational regulators eIF4E and eIF2α are also shown for comparison (Figure 3A), with eIF2α cosedimenting with subunits, monosomes and polysomes (Ramaiah et al., 1992), while eIF4E cosedimented primarily with the 40S fraction as in previous studies (Kedersha et al., 2002; Napoli et al., 2008). The lack of prominent association of DCC with polysomes provided further confirmation that the association with ribosomes is not simply attributable to the DCC polypeptide itself being translated there. The results, however, appeared very consistent with a role in translational regulation, since the initiation stage involves recruitment of 40S and 60S subunits to form the 80S ribosome, and is the major site of translational control in most systems.
Netrin-1 promotion of axon growth and chemoattraction can be mediated not only through DCC, but also two other receptors, neogenin and DSCAM. To assess whether DCC can mediate effects of netrin-1 on translation, DCC was transfected into 293 cells, which did not contain detectable levels of endogenous netrin receptors. Translation was assessed using a dual-luciferase reporter system for cap-dependent translation with a cap-independent cistron as internal control (Figure S4A)(Kruger et al., 2001). The results indeed showed a DCC-dependent effect of netrin-1 on translation in this system (Figures 4A and S4B-S4F). Likewise, an increase in total cell protein synthesis was seen by radiolabeled amino acid incorporation (Figure S4G), although as in previous studies (for example Pfeifer et al., 2008) the signal-to-background ratio was lower than for reporter translation. The effect of netrin-1 was also tested on the association of DCC with translational components, showing that it promoted dissociation (Figures 4B and S4H). (Note, however, that an association was still detectable in 293-net cells, consistent with the results in Figure 2.) Finding that netrin promoted dissociation of DCC from translational machinery appeared very consistent with our earlier experiments which had shown an association of DCC with components involved in translation initiation, but not with polysomes (Figures 2 and and3).3). These results all appear consistent with a model where full length DCC associates with translation initiation machinery, and signaling by DCC within this complex promotes formation of actively elongating polysomes which are no longer associated with DCC.
Since our earlier results had indicated that the DCC cytoplasmic domain was required for association with translation machinery, we assessed the role of this domain in translation. The DCC-Δcyto mutant had no significant effect in mediating the effect of netrin-1, showing a requirement for the cytoplasmic domain in regulating translation (Figure 4A).
As an additional test, a truncated receptor was made containing the DCC cytoplasmic domain only (DCCcyto). Expressing a truncated cell surface receptor is a common strategy that can provide information about binding and functional properties of a specific receptor domain. Functional effects of truncated receptors are often interfering rather than activating, since they may retain the ability to bind and occupy some signaling components, while lacking other domains required for productive signaling. For DCCcyto in particular, functional effects on translation, if any, were expected to be inhibitory, since the DCC extracellular domain is required for downstream signaling (Round and Stein, 2007). When DCCcyto was tested by transfecting it into 293 cells, this caused cell death (data not shown). While this could be consistent with translation inhibition, it did not provide a good system to study translation further. When DCCcyto was instead tested in a cell-free system, the reticulocyte lysate, it was indeed found to inhibit translation, as indicated both by reduced reporter translation, and reduced polysome levels (Figures 4C and 4D). The reduction in polysomes was accompanied by an increase in the 80S fraction, indicating that a late step after 80S monosome formation became rate limiting under these conditions. Cosedimentation of the DCCcyto polypeptide in these experiments was seen with ribosomal 40S, 60S and 80S fractions (Figure 4D). Thus DCCcyto, like full length DCC, associated physically with machinery involved in translation initiation, and produced functional effects on translation, although in the case of DCCcyto these effects were inhibitory not promoting, consistent with this mutant lacking an extracellular domain. These results provided further support for an overall model where the DCC cytoplasmic domain interacts both physically and functionally with the translation initiation machinery.
The DCC intracellular domain contains three motifs, P1, P2 and P3, which are conserved from C. elegans to mammals, and are therefore likely to mediate downstream functions (Figure 5A)(Garbe and Bashaw, 2004; Round and Stein, 2007). When DCC mutants were tested in the reporter assay for cellular translation, P3 deletion caused some reduction of translation (DCC-ΔP3 mutant; Figure 5B). This is expected, since P3 enhances DCC dimerization, which is believed to mediate signal transduction across the membrane, and it also binds known signaling proteins (Round and Stein, 2007) which might in principle modulate translation. A more drastic effect was seen upon deletion of the P1 motif (DCC-ΔP1 mutant), which completely blocked the ability of DCC to promote translation (Figure 5B). Correspondingly, the P1 deletion reduced the association of DCC with translational components close to background levels (Figure S5A).
Among the translational components that coprecipitated with DCC in the mass spectrometry analysis, the most prominent was ribosomal protein L5, with 11 peptides (Figure 2A). L5 also appeared interesting because it is thought to be located at or near the ribosomal P site, and L5 mutations can affect peptidyl-tRNA binding (Fabijanski and Pellegrini, 1981; Meskauskas and Dinman, 2001; Spahn et al., 2001; Takahashi et al., 2005). Purified recombinant L5 was therefore tested for binding to purified recombinant DCC cytoplasmic domain, and the results showed direct binding (Figure 5C). While this direct association of DCC with a ribosomal protein is striking, it should be emphasized that this is unlikely to be the only interaction of DCC relevant to translation (see Discussion).
A common strategy to test the biological function of protein associations is to mutate the interaction site. We were therefore interested to test whether the association of DCC with L5 could be mapped to a specific site within the DCC primary sequence. When nested deletions were examined removing successively larger regions from the C-terminus (Figure 5A), a region containing the P1 motif was found to be sufficient for L5 binding (Figure 5E). The P1 motif was found also to be necessary for L5 binding, as shown by testing the DCC-ΔP1 mutant (Figure 5D). It may be worth noting that although the P1 motif can also interact with co-receptors in the UNC5 family, they mediate repellent actions of netrin whereas commissural neurons are attracted, and they are not expressed in commissural neurons or the cell lines used in this study (Round and Stein, 2007; and unpublished data). The experiments here show that the P1 motif, which we had already found to be required to promote translation, was also the sequence necessary and sufficient for association of DCC with ribosomal protein L5.
DCC and L5 were also examined for overlapping subcellular expression. Since none of the L5 antibodies tested gave a signal in immunocytochemistry, a GFP-L5 fusion protein was co-expressed with DCC in 293 cells, where DCC promotes neurite-like extensions. Colocalization was seen in the neurite-like extensions, notably at their tips (Figures 6F-6H). Similarly, GFP-L5 expressed in commissural neurons colocalized with DCC at filopodial tips in axon growth cones (Figure S5B).
Having found that the P1 motif of DCC is required both for regulation of translation and for association with L5, the other side of the interaction was addressed, by testing whether perturbations of L5 might affect actions of the DCC cytoplasmic domain. Disruption of L5 within cells is problematic due to the key roles of L5 in ribosome biogenesis and function. Therefore, we used the cell-free reticulocyte system which we had already shown to be inhibited by DCCcyto. A functional interaction between L5 and DCCcyto was tested by the logic of dominant interference: if L5 provides a bridge between DCC and the ribosome, then addition of sufficient L5 should functionally occupy both its binding partners and therefore block the effects of the DCC cytoplasmic domain on translation. Accordingly, L5 addition was found to cause a dose-dependent and highly effective reversal of the DCCcyto effect on translation (Figure 5I). These experiments demonstrate a functional interaction between the DCC cytoplasmic tail and L5 in translation.
As a further test of our models, the effect of the P1 motif was tested on commissural axon behavior. Previous studies have implicated the P1 motif in axon attraction and growth promotion in C. elegans (Gitai et al., 2003; Garbe and Bashaw, 2004), and in UNC-5H dependent axon repulsion (Garbe and Bashaw, 2004; Round and Stein, 2007), and also have shown a role for the P3 motif in commissural axon guidance (Finger et al., 2002). Here, to assess P1 functionally in commissural axons, we used DCC-ΔP1 as a dominant interfering construct. The rationale for this type of dominant interfering experiment is that a mutant receptor can block the actions of endogenous wild type receptor by heterodimerization or by binding unproductively to upstream or downstream components. The DCC–ΔP1 construct was confirmed to have an interfering effect on translation, and to be expressed in neuronal growth cones (Figures S5C-S5E). To test commissural axon behavior, DCC was electroporated into chick spinal cord open book explants, using constructs which also expressed a fluorescent protein reporter, and 36 hours later axon trajectory was traced (Figures 5J-5P). Previous work has shown that DCC promotes axon growth and chemoattraction in response to floor plate derived netrin: following DCC gene disruption axons initially grow in a ventral direction but many fail to reach the floor plate (Fazeli et al., 1997). In control experiments here, full length DCC, or the use of GFP versus RFP markers, did not impair the ability of axons to reach the floor plate (Figures 6L, 6M, 6P, and data not shown). When DCC-ΔP1 was introduced, however, axons grew out in a ventral direction, but a greatly reduced proportion reached the midline (p<0.0001; Figures 6N to 6P), producing a phenotype comparable to DCC mutant mice (Fazeli et al., 1997). While the P1 motif might in principle affect other downstream processes in addition to translation, these results in spinal cord appear consistent with all our other data, and indicate that P1, a short motif that mediated translational regulation, and was necessary and sufficient for associations with ribosomal components, also affected DCC function in promoting outgrowth or attraction of spinal commissural axons toward the midline.
Finding that DCC colocalizes and physically associates with the translation machinery, and regulates translation, implied that DCC should be capable of regulating protein synthesis in a localized manner within the cell. To investigate this, we used a tracer of newly synthesized protein, azidohomoalanine (AHA), which acts as an amino acid analog incorporated in place of methionine, and can be visualized after cell fixation by covalent cycloaddition with a tagged alkyne (Dieterich et al., 2007; Beatty and Tirrell, 2008). AHA labeling was found to overlap with DCC in axons of commissural neurons, notably in puncta at the tips of filopodia (Figures 6A, 6B, S6A and S6B). We also used an independent tracer of newly synthesized protein which operates by a different mechanism, Cy3-puromycin, a fluorescently tagged derivative of the aminoacyl-tRNA analog puromycin (Smith et al., 2005) and obtained comparable results (Figures S6C-S6I). In hippocampal neurons too, AHA showed punctate labeling that overlapped with DCC puncta (Figure 6C). When the effect of netrin was tested, DCC puncta in commissural axon filopodia showed increased labeling of newly synthesized protein (Figures 6D and S6G). Thus, the accumulation of newly synthesized protein within DCC microdomains was increased by netrin transmembrane signaling.
Physical association of cellular components plays a central role in establishing both the spatial organization of the cell, and the specificity of cellular pathways. Cell surface receptors and the translation machinery are two of the basic classes of cellular components. The results here lead to a generalizable model where a transmembrane receptor interacts both extracellularly with signaling cues, and intracellularly with the translation machinery, to form a transmembrane translation regulation particle.
The results here showed that DCC associated with eIFs, ribosomal 40S and 60S subunits, and 80S ribosomes, though not prominently with polysomes. Taken together with finding that netrin-1 promoted translation in a DCC dependent manner, and reduced the association of DCC with translation components, these results lead to the following model. DCC forms a complex with translation initiation machinery. Signaling by DCC within this complex is enhanced by netrin, and promotes the formation of actively elongating polysomes which are no longer associated with DCC (Figure 7A). Signaling by DCC is believed to be mediated by dimerization or higher-order clustering, and in the absence of netrin, spontaneous dimerization and signaling by DCC may promote some translation, although at a lower level than with netrin. These interactions with the translation machinery are mediated via the DCC cytoplasmic domain, which was found to be necessary and sufficient for interactions with translational components (Figures 7B and 7C).
The association of DCC with translation initiation machinery, but not prominently with the polysome fraction, is reminiscent of the key translational regulator eIF4E, which cosedimented mainly with the ribosomal small subunit and not prominently with polysomes, as confirmed under the conditions here and seen in previous results (Kedersha et al., 2002; Napoli et al., 2008). Why DCC, or eIF4E, would not show more prominent association with the polysome fraction is an interesting question, since polysomes undergo repeated rounds of initiation as successive ribosomes are added. One potential explanation is suggested by observations that the initial recruitment of translation machinery to an mRNA involves formation of specialized ribonucleoprotein structures where the mRNA is looped to place the 5’ cap next to the 3’ polyA tail, whereas subsequently added ribosomes join an mRNA loop that is already undergoing active translation and may no longer require all the same components (Amrani et al., 2008). Although eIF4E and DCC shared the property of promoting translation yet not cosedimenting prominently with polysomes, they are very different in other respects. Whereas eIF4E initially recruits mRNA to the small subunit, we found that the truncated DCCcyto fragment affected a late stage after 80S ribosome assembly. While it is possible that this late step is the primary site of regulation by full length DCC, additional stages of translation initiation are likely to be regulated by full length DCC, as discussed further below.
Regarding protein interactions linking DCC to translational control, our studies implicated the conserved cytoplasmic DCC P1 motif, and ribosomal protein L5. Previous studies of eukaryotic L5 indicate that it is located at or near the ribosomal P site, and showed that mutations in L5 can influence binding of the peptidyl-tRNA to the ribosome (Fabijanski and Pellegrini, 1981; Meskauskas and Dinman, 2001; Spahn et al., 2001; Takahashi et al., 2005). Therefore, one model could be that effects of DCC on translation might be mediated via structural changes in L5. An alternative model would be that L5 serves more passively as a tether helping link DCC to the ribosome, while translational regulation would be mediated via other DCC interactors. The second of these models may be favored by our finding that DCC preferentially affected cap-dependent rather than cap-independent translation, since functional effects transmitted by a general translation component such as L5 might be expected to influence both types of translation equally. In either model, the location of L5 is notable because it could provide a mechanism to position DCC, together with other associated proteins, near the ribosomal P site, where the ribosomal large and small subunits, the initiator tRNA, and the initiator AUG codon all come together during translation initiation.
Another protein that may be involved in linking DCC to translation is Nck-1, which has been found to bind directly to the DCC cytoplasmic domain (Li et al., 2002). In a separate study, Nck-1 was shown to bind directly to eIF2β, and to regulate translation (Kebache et al., 2002). Since eIF2β is part of the 43S pre-initiation complex, Nck-1 therefore has the necessary binding specificities to act as a bridge between DCC and the ribosomal small subunit. Thus, together, L5 and Nck-1 can provide a model for DCC to be linked to both the ribosomal large and small subunits.
DCC associates with several other signaling proteins, some of which have been implicated in translational control in other systems (Garbe and Bashaw, 2004; Round and Stein, 2007). Also, eIF4E and its upstream regulator 4E-BP1 are thought to mediate netrin signaling (Campbell and Holt, 2003; Leung et al., 2006), and we find here that eIF4E associates in a complex with DCC. We therefore favor an overall model where DCC would act as a platform for the assembly of a large multicomponent complex, containing not only ribosomes and eIFs, but also multiple signal transduction components, providing more than one pathway structurally and functionally linking DCC to translational regulation within an integrated molecular apparatus.
In relation to the lack of prominent DCC association with polysomes, it is interesting that ribosomes have long been known to associate with the cytoskeleton, and in contrast to DCC, this cytoskeletal association has been reported to involve few or no monosomes or subunits, but most or all polysomes (Lenk et al., 1977; Hesketh and Pryme, 1991). This suggests a model where DCC forms a complex with translation initiation components, whereas actively translating polysomes would instead become associated with the cytoskeleton. Cytoskeletal attachment of polysomes could provide mechanisms either for directed movement of translating polysomes, or to retain them within specific subcellular sites close to the receptor (Figure 7A).
The identification of a complex of a transmembrane receptor with translation machinery opens up many possibilities for functions of this complex in ribonucleoprotein transport, localization, metabolism and regulation. In this regard, it would be interesting to know whether the DCC-containing particles identified here have any relationship with previously described ribonucleoprotein particles (Kiebler and Bassell, 2006) or mRNA binding proteins (St Johnston, 2005; Kiebler and Bassell, 2006; Lin and Holt, 2008; Rodriguez et al., 2008). Also, while DCC may regulate general translation of all mRNAs that are present locally, an appealing model is that its association with translation components may provide a mechanism to preferentially localize and regulate a specific subset of mRNAs.
Our observations on the association of DCC with translational machinery in neurons appear consistent with previous work showing functional roles for local protein translation in chemoattractant responses to netrin (Campbell and Holt, 2001; Ming et al., 2002). Meanwhile, guidance receptors also interact directly with the cytoskeleton, and accordingly translation inhibitors do not block all axon morphological responses, as reported for example by Roche et al. (2009) for cues other than netrin, although their study like others confirmed an involvement of translation in chemoattraction. The association of DCC with sites of protein synthesis within micron scale subcellular structures such as the filopodial tip and synaptic spine, as seen here, can provide a mechanism to help regulate translation with spatial precision in processes such as axon guidance and synaptic plasticity.
While the study here focuses on DCC, the model of receptor association with translation machinery may be more broadly applicable. Many imaging studies have found ribosomes, or newly synthesized protein, in the same vicinity as cell surface receptors, for example in fibroblast lamellipodia, near synapses or other sites of cell-cell contact, and under the surface membrane of axons (Koenig and Martin, 1996; Steward and Schuman, 2003; Rodriguez et al., 2006; Sutton and Schuman, 2006). Although it is not clear whether this localization places ribosomes in direct association with the membrane, it is worth noting that the electron microscopic approaches that are generally used identify polysomes but not subunits or monosomes (Sutton and Schuman, 2006), whereas we found here an association of DCC with subunits and monosomes. Other potentially consistent observations include the presence of a Wnt receptor in particles that sediment with approximately the velocity of a ribosome (Bilic et al., 2007), and the association of focal adhesion proteins with initiation factors and ribosomal proteins (de Hoog et al., 2004). We find that a subset of receptors other than DCC show physical and functional interactions with the translation machinery (unpublished data). While specific molecular mechanisms are likely to vary with different receptors, the concept of a transmembrane translation regulation particle, where cell surface receptors form a physical complex with protein synthesis machinery and regulate translation, could be widespread in cell biology.
Regulated protein synthesis is implicated in a wide variety of cellular functions controlled by extracellular cues in both normal biology and pathology (St Johnston, 2005; Tee and Blenis, 2005; Sutton and Schuman, 2006; Lecuyer et al., 2007; Lin and Holt, 2008; Rodriguez et al., 2008). Linking a cell surface receptor in a complex together with the translation machinery could have several potential advantages. One may be to enhance the efficiency or speed of regulation by bringing components together in an integrated molecular machine. Another is the potential for each receptor to associate with a unique repertoire of mRNAs, providing a novel layer of specificity in biological regulation. A third advantage may be to allow spatial precision of translation, which may be crucial for cellular functions such as motility, adhesion, guidance, and synaptic plasticity. Association of DCC or other cell surface receptors with translation machinery in a transmembrane translation regulation particle could contribute to varied developmental, physiological and pathological processes.
Commissural neuron cultures were prepared from E10.5-11.5 mouse spinal cords by trituration and plated on polylysine; >95% of cells were positive for commissural markers TAG-1 and DCC. Hippocampal neuron cultures from E18-19 rat brain were cultured for three weeks on laminin polylysine.
Cell lysates were prepared in Triton-X100 buffer. For spinal cord lysates, approximately 15 mouse E11.5 cords were homogenized. Immunoprecipitation used protein G–Sepharose beads washed with RIPA buffer. Far Western binding analysis used affinity purified recombinant GST fusion proteins probed with recombinant purified DCC cytoplasmic domain, followed by anti-DCC antibody and detection by chemiluminescence. For tandem mass spectrometry, after gel purification and trypsin digestion, peptides were analyzed by LC-MS/MS. All major peaks in the spectrum were assigned to predicted peptide ions, peptide representation of 138 proteins was present in immunoprecipitates from cells with DCC and not control cells, and peptides for proteins with two or less peptide hits were individually analyzed to confirm tryptic peptide identity. The entire immunoprecipitation and mass spectrometry analysis was performed twice; L5 and other ribosomal proteins and initiation factors were identified in both experimental series, and aggregate peptide counts are shown in Figure 2A.
293-net cells transiently transfected with DCC were treated with cycloheximide, and lysed in 1% Triton X-100. For each spinal cord gradient, twelve cords were homogenized in low salt buffer with digitonin. Lysates were precleared, centrifuged on a sucrose velocity gradient, and eluted on a fractionator monitored by 254 nm UV.
Cellular cap-dependent translation assays used bicistronic reporter plasmid pRL-5'-IRES-FL (Kruger et al., 2001), which directs cap-dependent translation of Renilla luciferase (RL), and also cap-independent IRES-mediated translation of firefly luciferase (FL) which was used as an internal control. In vitro translation experiments used Retic Lysate IVT Kit (Ambion), with Xef-1 RNA template, adding purified recombinant GST-DCCcyto, GST-L5 or GST control. Translated Xef-1 protein was resolved on SDS-PAGE, transferred to nitrocellulose and quantitated by phosphoimager. For AHA labeling, cells were incubated in methionine-free medium without serum containing 100 μg/ml AHA for the times indicated, then fixed and permeabilized with Triton-X100 and incubated in Click-iT tetramethylrhodamine (Invitrogen).
For electron microscopy, cultured neurons were permeabilized in saponin or Triton-X100, glutaraldehyde post-fixed, embedded in resin, and 80 nm sectioned. Non-specific distribution of particles in areas devoid of cells was negligible, and controls showed no gold particle clustering. Fluorescence imaging used an inverted laser confocal at 0.25-2.0 micron per z-series slice. All fluorescence quantitation used original unprocessed image data, with no pixels at zero intensity or saturated. In panels displayed in the Figures, for consistent visibility across the intensity range, contrast and brightness were adjusted uniformly within each experimental series. To confirm that individual yellow puncta reflected true colocalization, rather than fortuitous overlap, ROIs for individual puncta were selected by automatic thresholding in one channel, then Pearson's correlation coefficient (r) was measured for the two channels. This tests all pixels within an ROI for correlation over the full range of intensity values: chance overlap corresponds to r=0; identical patterns produce r=1, which would suggest an artifact; and co-localization of two molecules in a common structure should produce an intermediate value. For the growth cone as a whole, r values up to 0.8 were obtained. As a control, a circular ROI was selected containing a central part of the growth cone, and one channel was gradually rotated; although the density of yellow pixels remained similar, the r value for correlation of intensities was confirmed to fall to approximately zero. To quantify numbers of overlapping puncta, ROIs were selected in each channel independently by automatic thresholding to select 0.2-1 μm puncta, and the number of puncta that overlapped was then counted. To quantify AHA intensity, each ROI for quantitation was selected automatically using the green channel, with a constant threshold above background, to select 0.2-1 μm puncta from twenty different filopodia in twenty different growth cones, chosen at random and blind to the red channel. Mean pixel intensity was then measured within each ROI in the red channel.
We thank Marc Kirschner, Dietmar Schmucker, Melissa Hancock, Marc Tessier-Lavigne, Nancy Kadersha, Steven Gygi, Ross Tomaino, Tom Walz, Maria Ericsson, Natalie Farny, Pam Silver, Joachim Hauber, Taka-Aki Sato, Nathalie Lamarche-Vane, John Blenis, Davie Van Vactor, Judy Glaven, Yao Chen, Dan Nowakowski, Miriam Osterfield, and Nicolas Preitner, for help, reagents, advice, and comments on the manuscript. This work was supported by grants from the NIH (J.G.F.) and NCIC (P.P.R.), an HHMI fellowship (P.A.B.), a CIHR fellowship (J.T.), and a Canada Research Chair and HFSP Career Development Award (P.P.R.)
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.