Abstract
Intracellular RNA localization is a widespread and dynamic phenomenon that compartmentalizes gene expression and contributes to the functional polarization of cells. Thus far, mechanisms of RNA localization identified in Drosophila have been based on a few RNAs in different tissues, and a comprehensive mechanistic analysis of RNA localization in a single tissue is lacking. Here, by subcellular spatial transcriptomics we identify RNAs localized in the apical and basal domains of the columnar follicular epithelium (FE) and we analyze the mechanisms mediating their localization. Whereas the dynein/BicD/Egl machinery controls apical RNA localization, basally-targeted RNAs require kinesin-1 to overcome a default dynein-mediated transport. Moreover, a non-canonical, translation- and dynein-dependent mechanism mediates apical localization of a subgroup of dynein-activating adaptor-encoding RNAs (BicD, Bsg25D, hook). Altogether, our study identifies at least three mechanisms underlying RNA localization in the FE, and suggests a possible link between RNA localization and dynein/dynactin/adaptor complex formation in vivo.
Similar content being viewed by others
Introduction
RNA localization allows the precise compartmentalization of gene expression in space and time, and is a widespread phenomenon in many different cell types and organisms1,2,3,4,5. Three main mechanisms have been described to account for RNA localization: (1) active transport on cytoskeletal tracks, (2) localized protection from degradation, or (3) facilitated diffusion and entrapment6. Recently, several novel mechanisms have been reported to mediate RNA localization, such as hitch-hiking on other RNAs or organelles and co-translational RNA transport7,8,9,10,11,12,13,14,15. Active transport is the best-characterized mode of RNA localization and consists of the transport of ribonucleoprotein particles by motor proteins on cytoskeletal tracks. Localizing RNAs are typically transported in a translationally silent state and contain cis-acting localization elements (LEs) that are recognized and bound by trans-acting RNA-binding proteins (RBPs) mediating motor recruitment16.
Kinesin motor proteins mostly mediate microtubule (MT) plus end-directed transport. Kinesin-1 (Khc) has been shown to mediate oskar (osk) RNA localization to the posterior pole of the Drosophila oocyte17,18. Whereas Tropomyosin-1 isoform I/C (atypical Tm1, aTm1) regulates osk posterior localization by directly stabilizing Khc interaction with the RNA19,20,21, the Exon Junction Complex (EJC) deposited upon splicing is thought to activate kinesin-1 transport of the RNA20. Little is known about MT plus end-directed RNA transport in other tissues. Interestingly, aTm1 is also important for coracle RNA localization at Drosophila neuromuscular junctions22 and the EJC has been shown to mediate NIN RNA localization in human RPE1 cells23.
Cytoplasmic dynein and its accessory complex dynactin direct trafficking of cargoes towards MT minus ends. In Drosophila, dynein-mediated RNA transport is accomplished by the dynein-activating adaptor Bicaudal-D (BicD) and the RNA binding protein Egalitarian (Egl)24,25,26. The dynein/BicD/Egl complex is thought to mediate nurse cell-to-oocyte transport of maternal RNAs, and was shown to direct apical RNA localization in the early embryo, neuroblasts, and polar cells27,28,29,30,31. The dynein/dynactin/BicD (DDB) motor complex is highly conserved and participates in the transport of different cargoes, with BicD (and its mammalian ortholog BICD2) linking the dynein motor to specific cargoes. While proteins binding to the BicD C-terminal domain (CTD), such as Egl or Rab6, impart cargo specificity26,32,33,34,35, the BicD N-terminal domain (corresponding to coiled-coil 1/2, CC1/2) binds to dynein/dynactin33,36 and activates dynein processivity26,37,38,39.
Although much of what is known about RNA localization comes from studies of maternally inherited RNAs in the Drosophila germline, several examples of localizing RNAs have been also reported in the follicular epithelium (FE) that envelops the germline cyst4,40,41,42,43,44. The FE is composed of highly polarized secretory follicle cells (FCs) belonging to the somatic lineage, with minus ends of non-centrosomal microtubules (ncMTs) anchored at the apical cell cortex facing the oocyte45. The FE is an easily manipulatable and powerful genetic system that, through the generation of mosaics, allows the dissection of the effect of mutations without disrupting developmental processes. Several lines of evidence indicate that the dynein/BicD/Egl RNA transport complex active in nurse cell-to-oocyte transport is also responsible for the apical localization of a handful of RNAs in the FE31,40,42,46,47,48. However, a comprehensive overview of RNA localization in the FE and its underlying mechanisms are lacking.
Here, we apply subcellular spatial transcriptomics to first identify the landscape of apically- and basally-localizing RNAs in the columnar FE. By screening a subset of apical and basal RNAs identified in this way, we find that the dynein/BicD/Egl machinery acts by default in directing apical RNA localization, and that an additional kinesin-1-dependent layer of regulation must be applied to direct basal RNA localization. Moreover, we identify a third, translation- and dynein-dependent mechanism that underlies the apical localization of transcripts encoding dynein-activating adaptors, providing a possible link between RNA localization and dynein/dynactin/adaptor complex formation in vivo.
Results
Identification of apical and basal RNAs in columnar follicle cells
To identify RNAs that localize apically or basally in Drosophila FE transcriptome-wide, we applied laser-capture microdissection (LCM) to isolate fragments of tissue that consisted of either the apical half (“apical domain”) or basal half (“basal domain”) of adjacent columnar follicle cells (Fig. 1a and Supplementary Movie 1). Differential gene expression analysis of apical vs. basal LCM-derived RNA-seq samples yielded 306 RNAs enriched in the apical samples and 249 RNAs enriched in the basal samples (false discovery rate [FDR] < 0.1) (Fig. 1b, Supplementary Data 1). Since LCM is highly susceptible to tissue contamination, we first aimed to identify those RNAs whose significant enrichment was a result of contamination by other cell types, such as the oocyte on the apical side or the circular muscles on the basal side (Supplementary Fig. 1a). To do so, we analyzed those RNAs characterized by high absolute log2-transformed fold change (|log2FC | ) values of apical over basal abundance that might result from contamination of neighboring tissues expressing a different set of hallmark genes. By setting an arbitrary threshold of |log2FC | > 3 as indicative of contaminant RNA identity, we found 33 putative basal contaminants of muscle origin (log2FC < −3) and 2 putative apical contaminants of oocyte origin (log2FC > 3) (Fig. 1b and Supplementary Fig. 1b). 2/3 (n = 22) of basal genes with log2FC < −3 were annotated as being expressed or having a function in muscle tissues (FlyBase) and their mapped reads were often absent or in very low number in the apical fragments (Supplementary Fig. 1c, d). Moreover, we validated through single molecule Fluorescence In Situ Hybridization (smFISH) 3 putative basal contaminants (Mhc, Act57B, wupA) as being enriched in circular muscles with little or no expression in the FE (Supplementary Fig. 1e). This analysis resulted in 304 bona fide apical RNAs and 216 bona fide basal RNAs localizing in the columnar FE (Fig. 1b, Supplementary Data 1). Finally, 16 RNAs were randomly chosen from the computationally established list of significantly enriched bona fide apical or basal RNAs and were validated as true localizing RNAs through smFISH (Fig. 1c, d).
Basal RNA localization depends on kinesin-1 and aTm1
Basal RNA localization is a largely uncharacterized phenomenon. Previous reports have identified a limited number of basally-localizing RNAs in the FE4,43,44, with little mechanistic insight. For this reason, we sought to elucidate the mechanisms behind basal RNA localization. Early reporter-based studies on the polarity of Drosophila tissues have shown that the basal domain of the FE is functionally similar to the posterior pole of the oocyte, as both compartments accumulate the MT plus end marker Kin:βgal45. Therefore, we hypothesized that the regulators of oskar posterior RNA transport might also be responsible for basal RNA localization in the FE. To test this hypothesis, we disrupted known components of the osk RNP transport machinery, such as kinesin-1 (Khc) and atypical Tropomyosin-1 (aTm1) in the FE and analyzed the localization pattern of 4 validated basal RNAs (Fkbp14, CG3308, Rtnl1, zip) (Fig. 2). In all cells lacking either Khc (Khc RNAi cells) (Fig. 2a) or aTm1 (Tm1NULL,21) (Fig. 2b), basal RNA localization was severely disrupted, with all basal RNAs analyzed becoming apically localized. To have a quantitative overview of changes in RNA localization, we considered the ratio between the apical and the basal smFISH signal intensity in either wild-type (wt) or knock-down (KD) cells, and called this parameter Degree of Apicality (DoA), as values > 1 indicate an apical localization bias (Supplementary Fig. 2a). Then, we tested whether the DoA values of each RNA analyzed significantly differ in KD vs. wt cells by calculating the ratio between the DoA(KD) and the DoA(wt) for each RNA in each of the 3 conditions (see Materials and Methods and Fig. 2 for statistical testing). With this analysis, we confirmed that all basal RNAs analyzed were affected by lack of Khc or aTm1 (Fig. 2d). To check whether the observed changes in RNA localization were specific to basal RNAs, we analyzed the localization pattern of four previously validated apical RNAs (crb, msps, qtc, CG33129) in the same mutant backgrounds. In contrast to basal RNAs, none of the apical RNAs analyzed was affected by disruption of kinesin-1-mediated RNA transport (Supplementary Fig. 3 and Fig. 2d), indicating that regulators of MT plus end-directed RNA transport specifically control basal RNA localization. These results show that kinesin-1 and aTm1 are specifically responsible for basal RNA localization in the FE.
Basal RNA localization depends on the EJC
In addition to Khc and aTm1, nuclear events such as splicing and deposition of the Exon Junction Complex (EJC) play a role in the localization of oskar RNA at MT plus ends49. To test whether the EJC is involved in basal RNA localization, we disrupted the complex by overexpressing ΔC-Pym50,51. ΔC-Pym overexpression in FC clones impaired basal RNA localization (Fig. 2c–d), whereas apical RNA localization was unaffected (Supplementary Fig. 4a; Fig. 2d). This indicates that, in addition to Khc and aTm1, the EJC is also involved in the regulation of basal RNA localization.
Surprisingly, basal RNAs are less represented than apical RNAs among RNAs previously shown to be preferentially bound by the EJC52 (Supplementary Fig. 4b). Specifically, the basal RNAs we identified are on average EJC-depleted, while apical RNAs are on average EJC-enriched. This result is consistent with a slight apical enrichment of cytoplasmic GFP-Mago, as seen by GFP-Mago fluorescence quantification in the FE (Supplementary Fig. 4c–d). Since the EJC is specifically involved in basal RNA localization (Fig. 2c–d), these results were puzzling. However, considering that the EJC is displaced from RNAs upon translation53, we further tested whether this discrepancy might be due to different translation rates of apical versus basal RNAs. To do so, we analyzed two different Drosophila ribo-seq datasets, from wild-type ovaries54 and 0–2 h embryos55 and extracted the values of translation efficiency of the apical and basal RNAs identified in our study. This analysis shows that basal RNAs are more translated than apical RNAs in both ovaries and 0–2 h embryos (Supplementary Fig. 4e), suggesting that basal RNAs are on average EJC-depleted due to their higher translation rate.
Clonal analysis allows one to control for a change in the overall abundance of a given RNA in mutant cells by comparison of the total smFISH signal intensity with that in the adjacent wild-type cells. In both Khc RNAi and ΔC-Pym cells, the total smFISH signal intensity of basal RNAs was unchanged compared to neighboring wild-type cells, showing that observed changes in RNA localization are not due to RNA degradation (Supplementary Fig. 2c).
Mislocalization of the basal RNA zip in the absence of kinesin-1 depends on Egalitarian
Interestingly, disruption of MT plus end-directed RNA transport caused mislocalization of all analyzed basal RNAs to the apical domain. Several studies reported that apical RNA localization depends on the BicD/Egl machinery, a dynein-dependent complex that localizes RNAs apically in the blastoderm embryo and is thought to be responsible for nurse cell-to-oocyte transport of maternal RNAs. Therefore, the apical mislocalization of basal RNAs observed upon knock-down of kinesin-1 regulators might be due to apical RNA transport by the dynein/BicD/Egl machinery. To test this, we generated FC clones lacking either Egl (egl RNAi) or Khc (Khc RNAi), or both Egl and Khc [(egl + Khc) RNAi] and evaluated changes in the RNA localization of zip, one of the most striking examples of the apical mislocalization phenomenon (see Fig. 2a-c). Whereas zip RNA was unaffected upon egl RNAi and strongly apically mislocalized in Khc RNAi conditions as also highlighted by our previous experiments, (egl + Khc) RNAi caused zip to assume a ubiquitous localization that would be consistent with a failure of both kinesin-1-and dynein-mediated transport (Fig. 3a). zip DoA measurements in wt and RNAi cells in each of the three conditions provided a quantitative evaluation of the changes observed in smFISH experiments (Fig. 3b), with a significant decrease in KD/wt DoA in double (egl + Khc) RNAi cells (KD/wt DoA = 1.61) compared to Khc RNAi cells (KD/wt DoA = 2.49) (Fig. 3c). Therefore, despite being dispensable in basal RNA localization under normal conditions, the dynein/BicD/Egl complex is responsible for the apical mislocalization of a basal RNA (and possibly more) when kinesin-1 activity is lacking.
Two different dynein-dependent mechanisms control apical RNA localization
As mentioned previously, several reports have identified the dynein/BicD/Egl machinery as responsible for the apical localization of a subset of RNAs in the FE, such as crumbs (crb)40,46. To test in an unbiased way the degree of involvement of the dynein/BicD/Egl machinery in the localization of apical RNAs in the FE, we generated FC mutant clones in which either cytoplasmic dynein (Dhc64C, hereafter called Dhc) or Egalitarian (egl) were knocked-down by RNAi. We then analyzed the localization pattern of 5 validated apical RNAs (crb, msps, qtc, CG33129, BicD, with crb RNA as a positive control) by smFISH, along with the quantification of RNA localization by measuring the KD/wt DoA. The localization of all apical RNAs analyzed was completely abolished when Dhc was knocked down by RNAi, with the RNAs becoming ubiquitously distributed (Fig. 4a, c, d). egl RNAi caused all apical RNAs to lose their apical localization, with the surprising exception of BicD (Fig. 4b–d; see below). For the majority of apical RNAs analyzed, the total smFISH signal intensity was similar in mutant and wild-type cells (Supplementary Fig. 2d); only CG33129 RNA displayed some degree of RNA degradation upon Dhc RNAi. In spite of this, CG33129 signal was increased in the basal domain of RNAi cells with respect to wild-type cells, indicating that RNA degradation followed transcript mislocalization. Therefore, these results suggest that the changes observed in the localization of apical RNAs are not due to RNA degradation. In contrast to apical RNAs, basal RNAs largely maintained their basal localization pattern upon either Dhc RNAi or egl RNAi treatment (Supplementary Fig. 5a, b and Fig. 4d). Basal RNA localization was only mildly affected in a subset of Dhc RNAi cells, likely as a consequence of the emergence of polarity defects in cells lacking Dhc41,56(see Fig. 4a and Supplementary Fig. 5a).
The maintenance of BicD RNA localization in egl RNAi cells (Fig. 4c) was not due to a low efficiency of the RNAi, since both egl RNA and Egl protein were significantly reduced in egl KD cells (Supplementary Fig. 5c, d). Moreover, in egg chambers entirely lacking Egl throughout the FE (eglNULLFC, see Materials and Methods), BicD RNA was still apically localized, whereas localization of CG33129 RNA, previously found to be Egl-dependent (see Fig. 4b), was disrupted (Supplementary Fig. 5e). Altogether, these results show that the dynein/BicD/Egl complex is largely responsible for apical RNA localization, but a different dynein-dependent mechanism underlies the apical localization of BicD RNA. Considering that the Egl-independent targeting of BicD RNA represents a novel mechanism of apical RNA localization, we sought to gain more insight into the mechanisms regulating its RNA transport.
BicD RNA localization requires an intact translation machinery
Localization of BICD2/BicD RNA at centrosomes in cultured cells is translation-dependent57. To test whether BicD RNA localization in the FE involves the same mechanism, we treated egg chambers ex vivo with the translation inhibitors puromycin (Puro) and cycloheximide (CHX) and analyzed the distribution of BicD RNA under these two conditions compared to control ovaries incubated in Schneider’s medium only (Fig. 5a). To assess tissue integrity, in parallel we visualized osk RNA, whose localization during the middle stages of oogenesis should not be affected by translation inhibitors. Whereas the localization pattern of BicD RNA in CHX-treated egg chambers was similar to controls (Fig. 5b, d), Puro treatment clearly impaired BicD RNA localization in the FE (Fig. 5c). The distribution of BicD signal intensity along the A-B axis of mid-stage follicle cells shows that BicD enrichment at the apical cortex of the FE was severely reduced upon Puro treatment (Fig. 5e). As a proxy for the degree of signal mislocalization, we calculated the value corresponding to 50% of the cumulative area under the curve (a.u.c.) in Puro- or CHX-treated egg chambers and compared it with untreated controls. The results of this analysis show that the BicD RNA signal shifted significantly towards the basal domain in Puro-treated egg chambers, whereas CHX had no effect on BicD RNA localization (Fig. 5f). The fact that freezing elongating ribosomes (CHX condition) does not affect BicD RNA localization, whereas blocking translation by releasing the nascent peptide (Puro condition) does, indicates that an intact translation machinery and the presence of a nascent peptide may be required for BicD RNA localization in FCs.
BicD RNA is co-translationally localized
To understand whether the localization of BicD depends on translation of its own RNA (in cis) or of other factors (in trans), we designed a series of transgenic constructs consisting of a BicD-GFP cassette inserted downstream of an 18-bp linker in which we could introduce the desired frameshift mutations without disrupting any unknown RNA localization element in the BicD CDS (Supplementary Fig. 6a). Each of these transgenes was expressed in FC clones in a BicD wild-type background and the transgenic BicD-GFP RNA was specifically detected by smFISH using antisense GFP probes. GFP RNA carrying the same 3’ untranslated region (UTR) as BicD-GFP constructs failed to localize when expressed in the germline or in the FE (Supplementary Fig. 6b), showing that this sequence alone is not sufficient to drive RNA localization. In contrast, 0BicD-GFP RNA (“In-frame”) showed a strong apical localization in FCs (Fig. 5g–i), similarly to the endogenous BicD RNA (see Supplementary Fig. 7a). Moreover, the expression of full-length BicD-GFP was validated by the presence of GFP fluorescence in CD8-mCherry+ cells expressing the transgene (Fig. 5g). Disruption of the BicD-GFP reading frame by either +1 or −1 frameshift (“Frameshift”), verified by the absence of GFP signal in CD8-mCherry+ cells, was sufficient to impair apical RNA localization (Fig. 5g–i). Importantly, the change in localization observed for Frameshift BicD-GFP RNAs was not due to RNA degradation, as the smFISH signal intensity of all RNA constructs was comparable (Fig. 5j). Consistent with the puromycin-induced impairment of RNA localization in the FE, these results show that BicD RNA is co-translationally localized at the apical cortex.
BicD and Dhc RNAs decorate dynein particles at the apical cortex
As in BicD the first peptide emerging from the ribosome is the dynein-binding domain33, the co-translational localization of BicD RNA might depend on association of nascent BicD protein with dynein. To have an indication whether this might be the case, we imaged BicD RNA by smFISH in ovaries expressing endogenously tagged Dhc-GFP58. Furthermore, hypothesizing that nascent BicD might assemble co-translationally with nascent dynein, we included smFISH probes to also detect Dhc RNA. Although the Dhc-GFP signal was diffuse in the ovary, distinct Dhc-GFP foci were detected at the apical cortex of columnar FCs (Fig. 5k) and elsewhere in the germline (see below). These foci also contain Dhc RNA, indicating that these might be sites of Dhc RNA translation. BicD RNA showed a partial co-localization with Dhc-GFP/Dhc RNA foci, consistent with the hypothesis of its co-translational association with newly synthesized Dhc protein at the apical cortex.
The first step of BicD RNA localization in the early cyst is translation-independent
In the germline, BicD has an instructive role in oocyte specification24,59,60. Importantly, BicD RNA localization reflects MT minus end enrichment45,61 in both the germline and FE (Supplementary Fig. 7a). We noticed that, as in the FE, BicD RNA localization to the anterior of the oocyte (stages 9–10) was impaired in Puro-treated ovaries, whereas CHX had no effect (Fig. 5c, d). Interestingly, Puro treatment impaired the dynein-dependent posterior localization of BicD RNA in early stage oocytes (stages 4-5), without significantly affecting nurse cell-to-oocyte transport, a process that is also dynein-dependent (Supplementary Fig. 7b). Similarly, germline-expressed Frameshift BicD-GFP RNA enriched in the oocyte, but lost the posterior localization displayed by both endogenous BicD RNA and In-frame BicD-GFP RNA (Supplementary Fig. 7c). In addition, we found that BicD RNA decorates Dhc/Dhc RNA foci in both the FE and the oocyte, but not in the nurse cells (Supplementary Fig. 7d). Taken together, these results indicate that the mid-oogenesis oocyte and the columnar FCs share a similar co-translational mechanism for BicD RNA localization. In contrast, BicD RNA nurse cell-to-oocyte localization appears to be mediated by a translation-independent mechanism that does not involve the association with Dhc/Dhc RNA particles.
A subset of dynein adaptor-encoding RNAs are also co-translationally localized in the FE
BicD belongs to the class of dynein-activating adaptors, linking cargoes to the dynein motor complex62. We found that the RNA encoding all Drosophila orthologs of the currently known or putative dynein activating adaptors (hereafter collectively called “adaptor-encoding RNAs”), namely hook (HOOK1-3), Bsg25D (NIN/NINL), Nuf (RAB11FIP3), and Milton (TRAK1-2), were significantly enriched apically in our list of localizing transcripts (Supplementary Data 1), with the exception of Spindly (SPDL1) which was below the detection threshold. By hypothesizing that the same dynein-dependent co-translational process that drives BicD RNA localization would also be responsible for the apical localization of adaptor-encoding RNAs, we tested whether the localization of adaptor-encoding RNAs was affected by either Dhc or egl RNAi. Interestingly, the apical localization of Bsg25D and hook was significantly disrupted in Dhc RNAi cells (Fig. 6a, c), but not in egl RNAi cells (Fig. 6b, c). Moreover, the apical localization of both adaptor-encoding RNAs showed sensitivity to Puro but not CHX, as seen by smFISH (Fig. 6d). In particular, apical-cortical Bsg25D signal was lost upon Puro treatment, with the signal re-distributing more basally. hook, which appeared more abundant than Bsg25D, became unlocalized upon Puro treatment. As Bsg25D smFISH signal was characterized by a low signal-to-noise ratio, we analyzed the effect of translational inhibitors on the localization of the adaptor-encoding RNAs by measuring the integrated density after signal thresholding in three subdomains: apical-cortical (apical-most 1/3 of the apical area of FCs), subapical (the remaining 2/3), and basal (Supplementary Fig. 2b). Then, we calculated the percentage of signal in each subdomain, for each RNA (including BicD) in Control, Puro, and CHX conditions. This analysis revealed that, similar to BicD RNA, the localization of hook and Bsg25D RNAs was significantly affected by Puro treatment, whereas CHX had no effect (Fig. 6e).
To further investigate whether hook and Bsg25D use the same localization mechanism as BicD, we analyzed their spatial relationship with Dhc-GFP/Dhc RNA particles. As for BicD, both Bsg25D (Fig. 6f) and hook (Fig. 6g) were shown to partially co-localize with, thus decorate, Dhc-GFP foci containing Dhc RNA. Overall, these results suggest that the RNAs encoding the dynein activating adaptors BicD, hook, and Bsg25D represent a subgroup of apical RNAs that share the same co-translational, dynein-dependent mechanism that ensures their localization at cortical dynein foci also containing Dhc RNA.
Discussion
Only few examples of localizing RNAs in the FE have been described to date, with little mechanistic insight4,40,41,42,43,44. To explore the extent of RNA localization in a somatic tissue in vivo and gain insight into the mechanisms underlying the phenomenon, we have used laser-capture microdissection of apical and basal subcellular fragments of columnar follicle cells coupled with RNA-seq to identify localizing RNAs in this tissue. This allowed us to investigate in detail the landscape of mechanisms that mediate both apical and basal RNA localization in the FE (Fig. 7a). In our study, we found that basal RNA localization is mechanistically analogous to posterior RNA localization in the oocyte (represented by osk), reflecting MT plus end enrichment45. Khc, aTm1, and the EJC appear to be core components of a general basal RNA localization machinery. These results are in line with previous findings on osk RNA indicating that Khc/aTm1 bind to the 3’UTR20 and the EJC activates kinesin-1 transport through association with the coding sequence63.
According to our analysis, deposition of the EJC is necessary but not sufficient to determine RNA localization, as we found that the EJC is deposited on both apically- and basally-directed RNAs. Interestingly, Kwon et al.23 found that the EJC specifically localizes to the basal body of the primary cilium in mono-ciliated cells, where it controls the centrosomal localization of NIN RNA towards MT minus ends. In contrast, our data suggest that in columnar follicle cells, which are characterized by non-centrosomal MTOCs64,65,66, the EJC may play a role in MT plus end-directed (basal) RNA transport by acting synergistically with kinesin-1 and aTm1. Strikingly, the localization of osk RNA to the posterior pole of the oocyte also relies on the presence of MTs generated from non-centrosomal MTOCs65.Therefore, the mammalian EJC might have acquired a specific role in the localization of NIN RNA at basal bodies of mono-ciliated cells, while the Drosophila EJC appears to contribute to the MT plus end-directed localization of several RNAs through a centrosome-independent mechanism both in the somatic follicular epithelium (basal RNAs) and in the germline (osk RNA).
Interestingly, when either component of the kinesin-1 transport complex was lacking, basal RNAs were mislocalized to the apical domain in a dynein-dependent process. Therefore, dynein-mediated apical localization represents a default mechanism that must be overcome by kinesin-1 to drive basal RNA localization. Two possible scenarios could explain dynein-mediated apical mislocalization upon kinesin inhibition. Dynein and kinesin-1 could be engaged in a tug-of-war, pulling the RNAs in opposing directions, a phenomenon observed in the transport of vesicles and lipid droplets67. Alternatively, the dynein complex could be kept in an inhibited state and activated upon disruption of kinesin-1 and its regulators. If the tug-of-war scenario were correct, we would have expected a change in zip RNA localization in all RNAi conditions including egl RNAi alone, namely a shift to a more basal localization due to the enhanced Khc-dependent motility. However, since we did not see a significant change in zip localization when only Egl was knocked down, the tug-of-war hypothesis appears to be less likely than the inhibition hypothesis. In addition, this phenomenon recalls osk RNA mislocalization to the oocyte anterior upon disruption of kinesin-1, aTm1 or EJC components17,18,21,68,69,70,71,72 which was hypothesized to occur due to a failure to inactivate dynein-mediated RNA transport18.
Apical RNA localization, on the other hand, can be divided into two mechanistically distinct categories, both based on dynein-mediated transport. The first category includes those RNAs that are transported apically by the dynein/BicD/Egl machinery, a well characterized RNA transport complex that directs RNAs towards MT minus ends in a variety of tissues29. Our data suggest that the majority of apically localizing RNAs may belong to this class, as the localization of most of our randomly chosen apical RNAs was affected in both Dhc RNAi and egl RNAi conditions. This hypothesis is consistent with previous studies that identified several apical RNAs as BicD/Egl cargoes, in a variety of Drosophila tissues31,40,42,46,47,48. The BicD/Egl machinery has been hypothesized to be part of a larger RNP complex that ensures a tight translational control of the transported RNA73. Our finding that basal RNAs are on average more translated than apical RNAs suggests that RNAs transported apically in the FE by the dynein/BicD/Egl transport complex might indeed be kept in a translationally silent state until they have reached their final destination.
The second category of dynein-dependent apical RNAs does not involve Egalitarian activity for their localization. This includes a subgroup of dynein-activating adaptors, namely BicD, hook, and Bsg25D (BICD2, HOOK1-3, and NIN/NINL in mammals). Common features of their apical RNA localization include sensitivity to puromycin and partial co-localization with cortical dynein foci containing also Dhc RNA. Moreover, both Bsg25D74 and BicD (this study) RNA constructs containing the CDS alone are sufficient for the accumulation of their encoded protein at MT minus ends. Puromycin causes the disassembly of the translational machinery and the release of the N-terminal peptides emerging from ribosomes. As the N-terminal portion of these adaptors binds dynein or dynactin subunits33,75,76,77,78, we propose that the apical localization of BicD, hook, and Bsg25D depends on the co-translational association between dynein components and nascent adaptors at cortical dynein foci (Fig. 7b). This process might also be conserved in mammals, since the localization of both BICD2 and NIN RNA was shown to be puromycin-sensitive57. Previous studies have shown that the presence of either BICD2, HOOK3 or NIN/NINL promotes the formation of highly processive dynein/dynactin complexes37,38,79. Therefore, it is possible that co-translational assembly of components of the dynein-adaptor complexes is necessary to overcome dynein auto-inhibition80,81. BicD, hook, and Bsg25D may co-translationally associate with dynein soon after nuclear export of the RNA, promoting its apical transport in a manner similar to what has been proposed for PCNT RNA targeting at centrosomes14. Alternatively, since dynein can also function as a MT-tethered static anchor in mid-oogenesis oocytes and follicle cells82,83, the interaction between dynein and nascent adaptor proteins could occur after the RNA has reached the cell cortex by dynein-mediated transport. Indeed, puromycin treatment did not completely abolish the apical enrichment of adaptor-encoding RNAs, despite causing a marked decrease in their signal close to the apical cortex, where they decorate dynein cortical foci.
In vitro studies have shown that full-length BicD/BICD2 adopts an autoinhibitory conformation resulting from CC1/2 folding onto the CTD-containing CC326,36,84. Although the leading hypothesis in the field is that cargo binding to the CTD is responsible for the alleviation of auto-inhibition by freeing up the N-terminal dynein-binding domain26,32,33,36, it is possible that in vivo both nascent BicD interaction with dynein and cargo binding to the CTD might cooperate in preventing BicD intramolecular inhibition in the cellular environment. Strikingly, whereas the mechanism underlying oocyte localization of BicD RNA during mid-oogenesis resembles that observed in follicle cells, the nurse cell-to-oocyte transport of BicD RNA appears to be governed by a different, translation-independent mechanism that may not involve interaction with Dhc/Dhc RNA particles, consistent with a previous study indicating that BicD RNA is translationally inhibited by Me31B in the nurse cells85. In contrast to early egg chambers in which the MT network emanates from a posteriorly-positioned microtubule organizing center in the oocyte, mid-stage oocytes and columnar follicle cells are both characterized by non-centrosomal MTs (ncMTs) tethered to the cell cortex86. Therefore, the establishment of ncMTs could be at the basis of the mechanistic switch from translation-independent to co-translational BicD RNA localization in these compartments. A recent report has shown that NIN RNA (the mammalian ortholog of Bsg25D) localizes at ncMTs and its expression is essential for apico-basal MT formation and columnar epithelial shape87. Therefore, it is possible that the co-translational transport of adaptor-encoding RNAs may be important for correct ncMT nucleation at the apical cortex of the follicular epithelium.
Methods
LCM sample preparation
w1118 virgin females were kept with males for 24 h at 25 °C on yeast-supplemented cornmeal food.
Ovaries were dissected in PBS, transferred to a cryomold and snap-frozen in cold 2-Methylbutane after removal of excess PBS. Frozen ovaries were immediately covered with OCT cryoembedding compound (Sakura) and snap-frozen again. Before cryostat sectioning, each block was equilibrated at −20 °C for 1 h. 10 μm cryosections of OCT-embedded ovaries were carefully placed on a MembraneSlide NF 1.0 PEN (Zeiss), briefly thawed at RT and immediately fixed in 75% RNase-free (RF) ethanol for 30 s. Excess OCT was removed with ddH2O RF, and slides were stained in 100 μl Histogene staining solution (Arcturus) according to the manufacturer’s instructions. Finally, sections were dehydrated in increasing ethanol concentrations (75%, 95%, 100%), and briefly air-dried before LCM.
LCM and RNA-seq
LCM was performed with a Zeiss PALM MicroBeam and visualized under a 63X objective. Sectioned mid-oogenesis egg chambers were staged according to morphological criteria. Once stage 9–10 egg chambers had been identified, either the apical half (“apical fragment”) or the basal half (“basal fragment”) of 5-10 contiguous columnar follicle cells was microdissected and collected into the cap of an AdhesiveCap tube (Zeiss). 10 fragments of either apical or basal sample type from different egg chambers were pooled for each replicate, with a total microdissected area of ~3000–4000 μm2/replicate. LCM samples were processed according to Chen et al.88 to produce high-quality Illumina sequencing libraries. Samples were multiplexed and simultaneously sequenced in a single lane using the NextSeq500 system according to the manufacturer’s instructions.
RNA-seq analysis
Pre-processing of demultiplexed raw reads was performed on EMBL’s instance of Galaxy platform. Raw reads were trimmed to remove low-quality bases, filtered from rRNA reads, and mapped against D. melanogaster Release 6 (dm6) reference genome. To control for RNA degradation that might have occurred during LCM, the normalized transcript coverage of the uniquely mapping reads was calculated with CollectRNAseqMetrics (part of Picard tools, http://broadinstitute.github.io/picard/). Uniquely mapped read counts were normalized with DESeq289. Differential gene expression analysis was performed with DESeq2 by comparing the mean read counts of the Apical (4 replicates, A1-A4) and Basal (4 replicates, B1-B4) samples. Replicates A5 and B5 were excluded from further analysis due to their high degree of dissimilarity with replicates of the same sample type as shown by PCA and Euclidean distance analysis, probably due to a high degree of contamination from neighboring tissues. Statistical significance was set to an FDR-adjusted p value <0.1 (Benjamini-Hochberg correction for multiple testing). The R package ComplexHeatmap90 was used to generate the heatmap in Fig. 2b.
Identification of contaminant reads
Identification of contaminant RNAs was performed with R Studio. Among the RNAs that were significantly enriched in either the apical (log2FC > 0) or the basal (log2FC < 0) domain, were considered contaminants those RNAs displaying high absolute log2FoldChange (|log2FC | ), indicating that they were probably originating from neighboring tissues. A threshold of log2FC > 3 and log2FC < −3 was arbitrarily set to identify putative apical and basal contaminants, respectively. The functional annotation of each contaminant candidate was retrieved on FlyBase91 (release FB2020_6) and their read distribution among apical and basal replicates analyzed through Integrative Genomics Viewer (IGV)92.
Fly stocks and genetics
All fly stocks were maintained at 18 °C on standard fly food. For crosses, virgin females were mated with w1118 males at 25 °C on cornmeal food supplemented with yeast. Female offspring of the desired genotype were incubated with w1118 males on a yeast-supplemented medium for 24 h at 25 °C to stimulate the development of vitellogenic stage egg-chambers before ovary dissection.
The following stocks were obtained from the Bloomington Drosophila Stock Center (BDSC): w1118 (wild-type; #3605), DhcRNAi (#36698), eglRNAi (#28969), KhcRNAi (#35409), UAS-NLS-mCherry (#38425), osk-Gal4 (#44242), VK33 (#9750). Other stocks used were: HsFLP; arm > f + >Gal4; UAS-CD8-mCherry and tj-Gal4/CyO (gifts of Juan Manuel Gomez Elliff), Tm1eg1/TM3Sb,Ser and Tm1eg9/TM3Sb,Ser21, GFP-Mago93, Dhc64C-GFP58, vasa-Gal4/TM3Sb (gift of Jean Rene Huynh), UAS-ΔC-Pym-GFP51, UAS-Egl94, eglWU50/SM1 and eglPR29/SM6A24. Transgenic flies carrying UAS-GFP, UAS-0BicD-GFP, UAS-(+1)BicD-GFP, and UAS-(−1)BicD-GFP were generated in this study by phiC31 integrase-mediated recombination using the VK33 line, which carries an attP site on the third chromosome.
For the generation of eglNULLFC flies, eglWU50/CyO; osk-Gal4/TM3Ser were crossed with eglPR29/CyO; UAS-Egl/TM3Ser to generate eglWU50/eglPR29; osk-Gal4/UAS-Egl, expressing Egl only in the germline lineage to rescue the formation of rudimentary ovaries. tj-Gal4 and vasa-Gal4 drivers were used to express UAS-containing transgenes in the whole follicular epithelium and in the germline, respectively.
To generate flies for FC mutant clone induction, male flies carrying a UAS-containing transgene were crossed with hsFlp; arm > f + >Gal4; UAS-CD8-mCherry virgins, and F1 females were subjected to heat-shock as described below.
To generate flies for induction of FC mutant clones in the experiment illustrated in Fig. 3, HsFLP; arm > f + >Gal4/CyO; KhcRNAi/TM6B,Tb flies were crossed with + ; UAS-NLS-mCherry/CyO; eglRNAi/TM3Ser. F1 Female flies the desired genotypes [eglRNAi/TM6B,Tb for the egl RNAi condition; KhcRNAi/TM3Ser for the Khc RNAi condition; eglRNAi/KhcRNAi for the (egl + Khc) RNAi condition] were collected and subjected to heat-shock as described below.
Generation of follicle cell clones
The UAS-Gal4 “flip-out” system was used to generate marked mutant clones in a wild-type background95,96. Freshly eclosed females resulting from each cross were collected and mated with w1118 males for 24 h at 25 °C on food supplemented with yeast. Flies were heat-shocked for 1 h in a water bath heated at 37 °C. According to Gonzalez-Reyes & St Johnston97, heat-shocked females were kept for 39 h at 25 °C with males on yeast before dissection, thus allowing follicle cells that induced the expression of the transgene at stage ~5 to develop into stage 10 follicle cells.
Ex vivo pharmacological treatment
Young w1118 female flies were incubated with males for 24 h at 25 °C on fly food supplemented with yeast. Ovaries were dissected in PBS and immediately incubated in Schneider’s medium (Gibco) supplemented with 15% FBS (Gibco), 0.6X penicillin/streptomycin (Invitrogen), 200 μg/ml insulin (Sigma). For translation inhibitor treatment, either 200 μg/ml puromycin (Gibco) or 200 μg/ml cycloheximide (Sigma) or no compound (control) was added fresh to the medium and ovaries were incubated for 30 min at RT before fixation.
Generation of BicD-GFP constructs and transgenic fly lines
AttB-pUASp-BicD-GFP-K10 or AttB-pUASp-GFP-K10 plasmids carrying a w + cassette, a TLS-deficient version of the K10 3’UTR, and attB sites for phiC31 integrase-mediated recombination into the VK33 line were generated as follows.
To generate plasmid vectors carrying the BicD-GFP gene cassettes (0BicD-GFP, (-1)BicD-GFP, (+1)BicD-GFP), BicD and GFP CDS were amplified by PCR and the two fragments were combined into AttB-pUASp-K10 vector by InFusion cloning (Clontech) according to the manufacturer’s instructions. pBS-BicD (BicD-RA, FlyBase ID: FBpp0080555) plasmid (a kind gift from Jean-Baptiste Coutelis) was used as template to generate BicD CDS PCR amplicons. The Fw primer used to amplify BicD CDS was designed in order to include, in addition to a 20 nt-homology with AttB-pUASp-K10 vector, the Drosophila Kozak sequence98 in frame with a linker sequence where frameshift mutations could be generated, and a region annealing to nt 4-29 of BicD CDS. To generate 0BicD-GFP construct, the 18-bp linker containing the ATG (5’- ATGATCCTAGGCGCGCGG- 3’) was inserted in frame with nt 4-2346 of BicD-RA. To generate (+1)BicD-GFP construct, a C was inserted at position 4 in the N-terminal 18-bp linker (5’- ATGCATCCTAGGCGCGCGG- 3’). To generate (−1)BicD-GFP construct, a G was deleted at position 10 in the N-terminal 18-bp linker (5’- ATGATCCTA_GCGCGCGG- 3’). 0BicD-GFP, (−1)BicD-GFP, and (+1)BicD-GFP full insert sequences with the respective predicted translated ORF are listed in Supplementary Data 2.
To generate UAS-GFP construct, GFP ORF was amplified with a Fw primer containing KpnI restriction site upstream of GFP ATG and with a Rev primer containing NotI restriction site and the stop codon. The amplified fragment was gel purified, digested with KpnI and NotI and ligated into a AttB-pUASp-K10 vector digested with the same enzymes.
Each AttB-containing plasmid was purified and sequenced before injection into VK33 embryos carrying an attP site on the 3rd chromosome. Injected flies were crossed with If/CyO; Sb/TM3Ser individuals and transgenic F1 flies were identified by appearance of red eye color.
Immunostaining
5–10 pairs of ovaries were dissected in PBS and immediately fixed in 2% PFA in PBSTX(0.1%) (PBS + 0.1% Triton-X100) on a Nutator for 20 min at RT, followed by two washes of 15 min each with PBSTX(0.1%) shaking at RT. Ovaries were then blocked in 1X casein/PBSTX(0.1%) (stock: 10X casein blocking buffer, Sigma) for 30 min and incubated with rabbit anti-Egl primary antibody24 (kind gift from R. Lehmann) diluted 1:1000 in blocking buffer o/n at 4 °C. Alexa fluor 647 goat anti Rabbit (Jackson Immuno Research) secondary antibody was added at 1:750 dilution in blocking buffer for 2 h at RT. Samples were washed 3 × 10 min with 1X casein/PBSTX(0.1%), 1 × 10 min with PBSTX(0.1%) + 1:15,000 DAPI and kept o/n in 100 μl of 80% TDE/PBS before mounting on microscope slides.
Single molecule in situ Fluorescence Hybridization (smFISH)
smFISH antisense oligonucleotides (listed in Supplementary Data 3) were designed and labelled with dye-conjugated ddUTPs according to the protocol described by Gáspár et al.99 to generate oligonucleotides labelled at their 3’ and with ATTO-633-NHS ester (ATTO-TEC). When dual-color smFISH experiments were performed, each probe set was labelled with either ATTO-633 or ATTO-565. The degree of labelling (DOL, % of labelled oligos) and concentration of the labelled probe sets was measured according to the published algorithm.
Dissected ovaries were immediately fixed in 2% PFA/PBSTX(0.1%) gently shaking for 20 min at RT. In case of ex vivo ovary incubation, dissected ovaries were incubated in Schneider’s medium supplemented with the respective pharmacological treatment before proceeding with fixation, as described above. Fixed ovaries were rinsed and washed twice with PBSTX(0.1%) for 10 min before dehydrating them by replacing PBSTX(0.1%) with increasing concentrations of ethanol/PBSTX(0.1%). Fixed and dehydrated ovaries were kept in 100% ethanol at −20 °C for up to 10 days until the day of the experiment.
An optimized version of the smFISH protocol described in Hampoelz et al.100 was followed with minor modifications. All steps were performed at RT unless specified otherwise. Dehydrated ovaries were first rinsed with PBSTX(0.1%), followed by 2 × 15 min washes with PBSTX(0.1%), and incubated in Pre-hybridization Buffer (2 x SSC, 10% deionized formamide, 0.1% Tween-20) gently shaking for 30 min. The Pre-hybridization Buffer was replaced with 250 μl of Hybridization Buffer (2 x SSC, 10% deionized formamide, 0.1% Tween-20, 2 mM vanadyl ribonucleoside complex (New England Biolabs), 100 μg/mL salmon sperm DNA (Invitrogen), 10% dextran sulfate, 20 μg/mL BSA) pre-warmed at 37 °C in which smFISH probes were added to a final concentration of 1 nM/probe. Ovaries were kept hybridizing in the dark for 16-17 h on a heat block set at 37 °C shaking at 1000 rpm. To remove the excess probes, ovaries where washed 3 × 10 min at 37 °C with Washing Buffer (2 x SSC, 10% deionized formamide, 0.1% Tween-20). 1:15,000 DAPI was added to the second wash. Finally, samples were rinsed 4x in PBST(0.1%) (PBS + 0.1% Tween20) and kept in in 100 μl of 80% TDE/PBS for at least 1 h before mounting on microscope slides.
Z-stacks of images were acquired on a Leica TCS SP8 confocal microscope with 405 nm, 488 nm, 552 nm and 640 nm fixed excitation laser lines using a 63 × 1.3 NA glycerol immersion objective. A suitable range for spectral detection was carefully chosen for each channel to avoid cross-talk of fluorescence emission. Images were automatically restored by deconvolution with the Lightning module.
Quantification of smFISH signal
To quantify smFISH fluorescence of localizing RNAs, average Z-projections of deconvolved confocal image stacks were analyzed with Fiji101. In mosaic FE, for each wild-type (wt, unmarked) and mutant (mCherry-marked) group of cells within the same Z-stack, a region of interest (ROI) was drawn encompassing the apical and the basal cytoplasm of 5-10 adjacent follicle cells (with the exclusion of nuclei); in addition, a ROI was drawn in an area of the image were no signal was present (background, bg). The mean fluorescence intensity (m.f.i.) was measured for each ROI.
The degree of apicality (DoA) of a given RNA in each cell type (t) (wild-type or mutant) and each experimental condition c, was measured as follows:
To quantitatively analyze changes in RNA localization in each experimental condition c, the DoA measured in mutant (KD) cells was divided by the DoA measured in neighboring wild-type (wt) cells within the same Z-stack:
Only in Tm1NULL condition, due to the impossibility to obtain a mosaic tissue, the DoA of a given RNA in each cell type (t) (wild-type or Tm1NULL), was measured as follows:
To calculate the change in DoA, the DoA measured in single Tm1NULL egg chambers was divided by the average DoA measured in n wild-type egg chambers (wt):
Bioinformatic analysis of apical and basal RNAs
To analyze EJC enrichment on apical and basal RNAs, we annotated differentially EJC-enriched genes from Obrdlik et al.52 as apical or basal according to our RNA-seq analysis results, excluding putative oocyte and muscle contaminants, and plotted their log2FoldChange(EJC vs. RBP). The p-value was estimated by double-sided Wilcoxon rank sum test.
To calculate translation efficiency of apical and basal RNAs, ribosome profiling (RFP) TPM values as well as RNA-seq TPM values were extracted from published ribo-seq datasets of 0-2 h embryos55 and control RNAi ovaries54. Translation efficiency was calculated as RFP[TPM]/RNA-seq [TPM] for apical and basal RNAs identified by our study. The p-value was estimated by double-sided Wilcoxon rank sum test.
Statistics and reproducibility
Immunostaining and/or smFISH experiments were performed on ovaries from at least 3 flies (hundreds of egg chambers) and repeated at least twice at different times.
In Figs. 2d, 4d, and 6c, KD/wt DoA values of each RNA and experimental condition was measured across at least 3 different Z-projections, and the average value obtained was compared to the null hypothesis H0: KD/wt(DoA) = 1, corresponding to no change in RNA localization bias following KD treatment [DoA(KD)=DoA(wt)]. One-sample two-sided Student’s t-test was used to compare means to a reference value of mu = 1 in each experimental condition. In Supplementary Figure 2c-d, we calculated the total smFISH signal intensity as the sum of apical and basal signals in mutant and wild-type cells, computed the KD/wt total signal intensity and tested against a null hypothesis of H0: KD/wt(total signal intensity)=1, corresponding to no change in total signal intensity in KD cells. One-sample one-sided Student’s t-test (alternative=“less”) was used to compare means to a reference value of mu = 1 in each experimental condition. In Supplementary Fig. 4d, GFP-Mago fluorescence intensity was measured in apical and basal domains. The A/B ratio of GFP intensity was tested against a null hypothesis of H0: A/B=1, corresponding to an evenly distributed signal long the A-B axis. One-sample two-sided Student’s t-test was used to compare mean GFP-Mago fluorescence to a reference value of mu = 1.
In Fig. 3b, independent two-sided Student’s t-test was used to compare mean DoA(wt) and DoA(RNAi) values in each condition. In Fig. 3c, mean KD/wt DoA values across conditions were compared by one-way ANOVA followed by Tukey’s post-hoc tests.
Fluorescence intensity along lines was measured with Fiji on average Z-projections of confocal images and plotted with R Studio. Intensity values from each channel were normalized to 0-1 range.
BicD or GFP mean fluorescence intensity along the A-B axis of the epithelium was measured in groups of 5–10 adjacent follicle cells as line plots. At least 3 line plots were generated for each RNA measured in each condition. The value corresponding to 50% of the cumulative area under the curve (a.u.c.) of each plot was considered as the variation of the respective RNA localization along the A-B axis of the epithelium. Welch two sample t-test was used to compare mean values of the 50% of the a.u.c with respect to untreated controls (pharmacological experiments) or In-frame BicD-GFP (Frameshift vs. In-frame variation).
Quantification of BicD, hook, and Bsg25D signal following translation inhibition
smFISH signal in Control, CHX, and Puro conditions was quantified using Fiji. Due to low signal-to-noise ratio of Bsg25D and, to a lesser extent, hook, smFISH signal threshold was first manually adjusted. Then, the integrated density (IntDen) of background-subtracted signal was measured in the apical-cortical, subapical, and basal domains of adjacent FCs. To minimize sample-to-sample variability due to differences in threshold adjustment, the percentage of signal present in each subdomain within single egg chambers was measured, and average percentages were plotted for each RNA in each condition. Statistical testing was performed by independent two-sided Student’s t-test (n = 4 replicates per RNA analyzed).
Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
Data availability
The data supporting the findings of this study are available within the article and its supplementary information files. Raw microscopy images are available upon request from the corresponding author. The D. melanogaster genome release 6 (dm6) data used in this study are available in the NCBI GenBank assembly database under accession code GCA_000001215.4. The raw RNA-seq data generated in this study have been deposited in the ArrayExpress database under accession code E-MTAB-9127. The processed RNA-seq data are provided in the Supplementary Files (Supplementary Data 1). Source data are provided with this paper.
References
Shepard, K. A. et al. Widespread cytoplasmic mRNA transport in yeast: identification of 22 bud-localized transcripts using DNA microarray analysis. Proc. Natl Acad. Sci. U.S.A. 100, 11429–11434 (2003).
Lécuyer, E. et al. Global analysis of mRNA localization reveals a prominent role in organizing cellular architecture and function. Cell 131, 174–187 (2007).
Mili, S., Moissoglu, K. & Macara, I. G. Genome-wide screen reveals APC-associated RNAs enriched in cell protrusions. Nature 453, 115–119 (2008).
Jambor, H. et al. Systematic imaging reveals features and changing localization of mRNAs in Drosophila development. Elife 4, (2015). https://doi.org/10.7554/eLife.05003
Moor, A. E. et al. Global mRNA polarization regulates translation efficiency in the intestinal epithelium. Science 357, 1299–1303 (2017).
Medioni, C., Mowry, K. & Besse, F. Principles and roles of mRNA localization in animal development. Development 139, 3263–3276 (2012).
Ferrandon, D., Koch, I., Westhof, E. & Nüsslein-Volhard, C. RNA-RNA interaction is required for the formation of specific bicoid mRNA 3’ UTR-STAUFEN ribonucleoprotein particles. EMBO J. 16, 1751–1758 (1997).
Jambor, H., Brunel, C. & Ephrussi, A. Dimerization of oskar 3’ UTRs promotes hitchhiking for RNA localization in the Drosophila oocyte. RNA 17, 2049–2057 (2011).
Corradi, E. et al. Axonal precursor miRNAs hitchhike on endosomes and locally regulate the development of neural circuits. EMBO J. 39, e102513 (2020).
Cioni, J.-M. et al. Late Endosomes Act as mRNA Translation Platforms and Sustain Mitochondria in Axons. Cell 176, 56–72.e15 (2019).
Liao, Y.-C. et al. RNA Granules Hitchhike on Lysosomes for Long-Distance Transport, Using Annexin A11 as a Molecular Tether. Cell 179, 147–164.e20 (2019).
Harbauer, A. B. et al. Neuronal mitochondria transport Pink1 mRNA via synaptojanin 2 to support local mitophagy. Neuron 110, 1516–1531.e9 (2022).
Cohen, B. et al. Mitochondria serve as axonal shuttle for Cox7c mRNA through mechanism that involves its mitochondrial targeting signal. bioRxiv 2021.05.19.444640 (2021). https://doi.org/10.1101/2021.05.19.444640
Sepulveda, G. et al. Co-translational protein targeting facilitates centrosomal recruitment of PCNT during centrosome maturation in vertebrates. Elife 7, (2018).
Bergalet, J. et al. Inter-dependent Centrosomal Co-localization of the cen and ik2 cis-Natural Antisense mRNAs in Drosophila. Cell Rep. 30, 3339–3352.e6 (2020).
Xing, L. & Bassell, G. J. mRNA localization: an orchestration of assembly, traffic and synthesis. Traffic 14, 2–14 (2013).
Brendza, R. P., Serbus, L. R., Duffy, J. B. & Saxton, W. M. A function for kinesin I in the posterior transport of oskar mRNA and Staufen protein. Science 289, 2120–2122 (2000).
Zimyanin, V. L. et al. In vivo imaging of oskar mRNA transport reveals the mechanism of posterior localization. Cell 134, 843–853 (2008).
Dimitrova-Paternoga, L. et al. Molecular basis of mRNA transport by a kinesin-1-atypical tropomyosin complex. Genes Dev. 35, 976–991 (2021).
Gáspár, I., Sysoev, V., Komissarov, A. & Ephrussi, A. An RNA‐binding atypical tropomyosin recruits kinesin‐1 dynamically to oskar mRNPs. EMBO J. 36, 319–333 (2016).
Erdélyi, M., Michon, A. M., Guichet, A., Glotzer, J. B. & Ephrussi, A. Requirement for Drosophila cytoplasmic tropomyosin in oskar mRNA localization. Nature 377, 524–527 (1995).
Gardiol, A. & St Johnston, D. Staufen targets coracle mRNA to Drosophila neuromuscular junctions and regulates GluRIIA synaptic accumulation and bouton number. Dev. Biol. 392, 153–167 (2014).
Kwon, O. S. et al. Exon junction complex dependent mRNA localization is linked to centrosome organization during ciliogenesis. Nat. Commun. 12, 1351 (2021).
Mach, J. M. & Lehmann, R. An Egalitarian-BicaudalD complex is essential for oocyte specification and axis determination in Drosophila. Genes Dev. 11, 423–435 (1997).
Navarro, C., Puthalakath, H., Adams, J. M., Strasser, A. & Lehmann, R. Egalitarian binds dynein light chain to establish oocyte polarity and maintain oocyte fate. Nat. Cell Biol. 6, 427–435 (2004).
Dienstbier, M., Boehl, F., Li, X. & Bullock, S. L. Egalitarian is a selective RNA-binding protein linking mRNA localization signals to the dynein motor. Genes Dev. 23, 1546–1558 (2009).
Clark, A., Meignin, C. & Davis, I. A Dynein-dependent shortcut rapidly delivers axis determination transcripts into the Drosophila oocyte. Development 134, 1955–1965 (2007).
Wilkie, G. S. & Davis, I. Drosophila wingless and pair-rule transcripts localize apically by dynein-mediated transport of RNA particles. Cell 105, 209–219 (2001).
Bullock, S. L. & Ish-Horowicz, D. Conserved signals and machinery for RNA transport in Drosophila oogenesis and embryogenesis. Nature 414, 611–616 (2001).
Hughes, J. R., Bullock, S. L. & Ish-Horowicz, D. Inscuteable mRNA localization is dynein-dependent and regulates apicobasal polarity and spindle length in Drosophila neuroblasts. Curr. Biol. 14, 1950–1956 (2004).
Van de Bor, V., Zimniak, G., Cérézo, D., Schaub, S. & Noselli, S. Asymmetric localisation of cytokine mRNA is essential for JAK/STAT activation during cell invasiveness. Development 138, 1383–1393 (2011).
Matanis, T. et al. Bicaudal-D regulates COPI-independent Golgi-ER transport by recruiting the dynein-dynactin motor complex. Nat. Cell Biol. 4, 986–992 (2002).
Hoogenraad, C. C. et al. Bicaudal D induces selective dynein-mediated microtubule minus end-directed transport. EMBO J. 22, 6004–6015 (2003).
Coutelis, J.-B. & Ephrussi, A. Rab6 mediates membrane organization and determinant localization during Drosophila oogenesis. Development 134, 1419–1430 (2007).
Januschke, J. et al. Rab6 and the secretory pathway affect oocyte polarity in Drosophila. Development 134, 3419–3425 (2007).
Hoogenraad, C. C. et al. Mammalian Golgi-associated Bicaudal-D2 functions in the dynein-dynactin pathway by interacting with these complexes. EMBO J. 20, 4041–4054 (2001).
McKenney, R. J., Huynh, W., Tanenbaum, M. E., Bhabha, G. & Vale, R. D. Activation of cytoplasmic dynein motility by dynactin-cargo adapter complexes. Science 345, 337–341 (2014).
Schlager, M. A. et al. Bicaudal d family adaptor proteins control the velocity of Dynein-based movements. Cell Rep. 8, 1248–1256 (2014).
Sladewski, T. E. et al. Recruitment of two dyneins to an mRNA-dependent Bicaudal D transport complex. Elife 7, (2018). https://doi.org/10.7554/eLife.36306
Li, Z., Wang, L., Hays, T. S. & Cai, Y. Dynein-mediated apical localization of crumbs transcripts is required for Crumbs activity in epithelial polarity. J. Cell Biol. 180, 31–38 (2008).
Horne-Badovinac, S. & Bilder, D. Dynein regulates epithelial polarity and the apical localization of stardust A mRNA. PLoS Genet. 4, e8 (2008).
Vazquez-Pianzola, P. et al. The mRNA transportome of the BicD/Egl transport machinery. RNA Biol. 14, 73–89 (2017).
Schotman, H., Karhinen, L. & Rabouille, C. dGRASP-mediated noncanonical integrin secretion is required for Drosophila epithelial remodeling. Dev. Cell 14, 171–182 (2008).
Serano, J. & Rubin, G. M. The Drosophila synaptotagmin-like protein bitesize is required for growth and has mRNA localization sequences within its open reading frame. Proc. Natl Acad. Sci. U.S.A. 100, 13368–13373 (2003).
Clark, I. E., Jan, L. Y. & Jan, Y. N. Reciprocal localization of nod and kinesin fusion proteins indicates microtubule polarity in the Drosophila oocyte, epithelium, neuron and muscle. Development 124, 461–470 (1997).
Bhagavatula, S. & Knust, E. A putative stem-loop structure in Drosophila crumbs is required for mRNA localisation in epithelia and germline cells. J. Cell Sci. 134, (2021). https://doi.org/10.1242/jcs.236497
Karlin-Mcginness, M., Serano, T. L. & Cohen, R. S. Comparative analysis of the kinetics and dynamics of K10, bicoid, and oskar mRNA localization in the Drosophila oocyte. Dev. Genet. 19, 238–248 (1996).
Jambor, H., Mueller, S., Bullock, S. L. & Ephrussi, A. A stem-loop structure directs oskar mRNA to microtubule minus ends. RNA 20, 429–439 (2014).
Hachet, O. & Ephrussi, A. Splicing of oskar RNA in the nucleus is coupled to its cytoplasmic localization. Nature 428, 959–963 (2004).
Gehring, N. H., Lamprinaki, S., Kulozik, A. E. & Hentze, M. W. Disassembly of exon junction complexes by PYM. Cell 137, 536–548 (2009).
Ghosh, S., Obrdlik, A., Marchand, V. & Ephrussi, A. The EJC binding and dissociating activity of PYM is regulated in Drosophila. PLoS Genet. 10, e1004455 (2014).
Obrdlik, A., Lin, G., Haberman, N., Ule, J. & Ephrussi, A. The transcriptome-wide landscape and modalities of ejc binding in adult drosophila. Cell Rep. 28, 1219–1236.e11 (2019).
Dostie, J. & Dreyfuss, G. Translation Is Required to Remove Y14 from mRNAs in the Cytoplasm. Curr. Biol. 12, 1060–1067 (2002).
Greenblatt, E. J. & Spradling, A. C. Fragile X mental retardation 1 gene enhances the translation of large autism-related proteins. Science 361, 709–712 (2018).
Greenblatt, E. J., Obniski, R., Mical, C. & Spradling, A. C. Prolonged ovarian storage of mature Drosophila oocytes dramatically increases meiotic spindle instability. Elife 8, https://doi.org/10.7554/eLife.49455 (2019).
Ronchi, P. et al. High-precision targeting workflow for volume electron microscopy. J. Cell Biol. 220, e202104069 (2021).
Safieddine, A. et al. A choreography of centrosomal mRNAs reveals a conserved localization mechanism involving active polysome transport. Nat. Commun. 12, 1352 (2021).
Gáspár, I., Phea, L. J. & Ephrussi, A. Egalitarian feeds forward to Staufen to inhibit Dynein during mRNP transport. bioRxiv 2021.04.24.441269 (2021). https://doi.org/10.1101/2021.04.24.441269
Wharton, R. P. & Struhl, G. Structure of the Drosophila BicaudalD protein and its role in localizing the posterior determinant nanos. Cell 59, 881–892 (1989).
Suter, B. & Steward, R. Requirement for phosphorylation and localization of the Bicaudal-D protein in Drosophila oocyte differentiation. Cell 67, 917–926 (1991).
Steinhauer, J. & Kalderon, D. Microtubule polarity and axis formation in the Drosophila oocyte. Dev. Dyn. Publ. Am. Assoc. Anat. 235, 1455–1468 (2006).
Olenick, M. A. & Holzbaur, E. L. F. Dynein activators and adaptors at a glance. J. Cell Sci. 132, https://doi.org/10.1242/jcs.227132 (2019).
Ghosh, S., Marchand, V., Gáspár, I. & Ephrussi, A. Control of RNP motility and localization by a splicing-dependent structure in oskar mRNA. Nat. Struct. Mol. Biol. 19, 441–449 (2012).
Mahowald, A. P., Caulton, J. H., Edwards, M. K. & Floyd, A. D. Loss of centrioles and polyploidization in follicle cells of Drosophila melanogaster. Exp. Cell Res. 118, 404–410 (1979).
Nashchekin, D., Fernandes, A. R. & St Johnston, D. Patronin/shot cortical foci assemble the noncentrosomal microtubule array that specifies the drosophila anterior-posterior axis. Dev. Cell 38, 61–72 (2016).
Wu, X., Tanwar, P. S. & Raftery, L. A. Drosophila follicle cells: morphogenesis in an eggshell. Semin. Cell Dev. Biol. 19, 271–282 (2008).
Hancock, W. O. Bidirectional cargo transport: moving beyond tug of war. Nat. Rev. Mol. Cell Biol. 15, 615–628 (2014).
Cha, B.-J., Serbus, L. R., Koppetsch, B. S. & Theurkauf, W. E. Kinesin I-dependent cortical exclusion restricts pole plasm to the oocyte posterior. Nat. Cell Biol. 4, 592–598 (2002).
Hachet, O. & Ephrussi, A. Drosophila Y14 shuttles to the posterior of the oocyte and is required for oskar mRNA transport. Curr. Biol. 11, 1666–1674 (2001).
Mohr, S. E., Dillon, S. T. & Boswell, R. E. The RNA-binding protein Tsunagi interacts with Mago Nashi to establish polarity and localize oskar mRNA during Drosophila oogenesis. Genes Dev. 15, 2886–2899 (2001).
Newmark, P. A. & Boswell, R. E. The mago nashi locus encodes an essential product required for germ plasm assembly in Drosophila. Development 120, 1303–1313 (1994).
Palacios, I. M., Gatfield, D., St Johnston, D. & Izaurralde, E. An eIF4AIII-containing complex required for mRNA localization and nonsense-mediated mRNA decay. Nature 427, 753–757 (2004).
Vazquez-Pianzola, P. & Suter, B. Conservation of the RNA Transport Machineries and Their Coupling to Translation Control across Eukaryotes. Comp. Funct. Genomics 2012, 287852 (2012).
Kowanda, M. et al. Loss of function of the Drosophila Ninein-related centrosomal protein Bsg25D causes mitotic defects and impairs embryonic development. Biol. Open 5, 1040–1051 (2016).
Urnavicius, L. et al. The structure of the dynactin complex and its interaction with dynein. Science 347, 1441–1446 (2015).
Schroeder, C. M. & Vale, R. D. Assembly and activation of dynein-dynactin by the cargo adaptor protein Hook3. J. Cell Biol. 214, 309–318 (2016).
Lee, I.-G. et al. A conserved interaction of the dynein light intermediate chain with dynein-dynactin effectors necessary for processivity. Nat. Commun. 9, 986 (2018).
Lee, I.-G., Cason, S. E., Alqassim, S. S., Holzbaur, E. L. F. & Dominguez, R. A tunable LIC1-adaptor interaction modulates dynein activity in a cargo-specific manner. Nat. Commun. 11, 5695 (2020).
Redwine, W. B. et al. The human cytoplasmic dynein interactome reveals novel activators of motility. Elife 6, https://doi.org/10.7554/eLife.28257 (2017).
Torisawa, T. et al. Autoinhibition and cooperative activation mechanisms of cytoplasmic dynein. Nat. Cell Biol. 16, 1118–1124 (2014).
Zhang, K. et al. Cryo-EM Reveals How Human Cytoplasmic Dynein Is Auto-inhibited and Activated. Cell 169, 1303–1314.e18 (2017).
Delanoue, R. & Davis, I. Dynein anchors its mRNA cargo after apical transport in the Drosophila blastoderm embryo. Cell 122, 97–106 (2005).
Delanoue, R., Herpers, B., Soetaert, J., Davis, I. & Rabouille, C. Drosophila Squid/hnRNP helps Dynein switch from a gurken mRNA transport motor to an ultrastructural static anchor in sponge bodies. Dev. Cell 13, 523–538 (2007).
Stuurman, N. et al. Interactions between coiled-coil proteins: Drosophila lamin Dm0 binds to the bicaudal-D protein. Eur. J. Cell Biol. 78, 278–287 (1999).
Nakamura, A., Amikura, R., Hanyu, K. & Kobayashi, S. Me31B silences translation of oocyte-localizing RNAs through the formation of cytoplasmic RNP complex during Drosophila oogenesis. Development 128, 3233–3242 (2001).
Tillery, M. M. L., Blake-Hedges, C., Zheng, Y., Buchwalter, R. A. & Megraw, T. L. Centrosomal and Non-Centrosomal Microtubule-Organizing Centers (MTOCs) in Drosophila melanogaster. Cells 7, (2018).
Goldspink, D. A. et al. Ninein is essential for apico-basal microtubule formation and CLIP-170 facilitates its redeployment to non-centrosomal microtubule organizing centres. Open Biol. 7, (2017).
Chen, J. et al. Spatial transcriptomic analysis of cryosectioned tissue samples with Geo-seq. Nat. Protoc. 12, 566–580 (2017).
Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014).
Gu, Z., Eils, R. & Schlesner, M. Complex heatmaps reveal patterns and correlations in multidimensional genomic data. Bioinformatics 32, 2847–2849 (2016).
Larkin, A. et al. FlyBase: updates to the Drosophila melanogaster knowledge base. Nucleic Acids Res. 49, D899–D907 (2021).
Robinson, J. T. et al. Integrative genomics viewer. Nat. Biotechnol. 29, 24–26 (2011).
Newmark, P. A., Mohr, S. E., Gong, L. & Boswell, R. E. mago nashi mediates the posterior follicle cell-to-oocyte signal to organize axis formation in Drosophila. Development 124, 3197–3207 (1997).
Bullock, S. L., Nicol, A., Gross, S. P. & Zicha, D. Guidance of bidirectional motor complexes by mRNA cargoes through control of dynein number and activity. Curr. Biol. 16, 1447–1452 (2006).
Struhl, G. & Basler, K. Organizing activity of wingless protein in Drosophila. Cell 72, 527–540 (1993).
Pignoni, F. & Zipursky, S. L. Induction of Drosophila eye development by decapentaplegic. Development 124, 271–278 (1997).
González-Reyes, A. & St Johnston, D. Patterning of the follicle cell epithelium along the anterior-posterior axis during Drosophila oogenesis. Development 125, 2837–2846 (1998).
Cavener, D. R. Comparison of the consensus sequence flanking translational start sites in Drosophila and vertebrates. Nucleic Acids Res. 15, 1353–1361 (1987).
Gaspar, I., Wippich, F. & Ephrussi, A. Enzymatic production of single-molecule FISH and RNA capture probes. RNA 23, 1582–1591 (2017).
Hampoelz, B. et al. Nuclear Pores Assemble from Nucleoporin Condensates During Oogenesis. Cell 179, 671–686.e17 (2019).
Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Nat. Methods 9, 676–682 (2012).
Acknowledgements
We thank the Bloomington Drosophila Stock Center (BDSC), J. M. Gomez Elliff, A. Debec, I. Gaspar, S. Bullock, J. R. Huynh and R. Lehmann for providing fly lines and reagents. We are grateful to I. Gaspar and K. Zarnack for helpful discussions and training. We are grateful to A. Obrdlik and Lin Gen for discussion of the IpaRT dataset. We thank the EMBL GeneCore Facility (D. Pavlinic, V. Benes), Advanced Light Microscopy Facility (S. Terjung), Drosophila Injection Service (A. Reversi), Centre for Bioimage Analysis (C. Tischer), Centre for Statistical Data Analysis (B. Klaus), Genome Biology Computational Support (C. Girardot), and W. Huber for their support. We are grateful to K. Zarnack, P. Ronchi, L. Russo, A. Reversi and members of the Ephrussi lab for critically reading the manuscript. This work was supported by EMBL. L.C. was supported by DFG-FOR 2333 grants EP 37/2-1 and EP 37/4-1 from the Deutsche Forschungsgemeinschaft (Germany) to A.E.
Funding
Open Access funding enabled and organized by Projekt DEAL.
Author information
Authors and Affiliations
Contributions
Conceptualization: L.C. and A.E.; Investigation: L.C.; Data analysis: L.C; Writing – original draft: L.C.; Writing – review & editing: L.C. and A.E; Supervision: A.E.; Funding acquisition: A.E.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. Peer reviewer reports are available
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Source data
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Cassella, L., Ephrussi, A. Subcellular spatial transcriptomics identifies three mechanistically different classes of localizing RNAs. Nat Commun 13, 6355 (2022). https://doi.org/10.1038/s41467-022-34004-2
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467-022-34004-2
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.