ABSTRACT

The adult Drosophila intestinal epithelium is a model system for stem cell biology, but its utility is limited by current biochemical methods that lack cell type resolution. Here, we describe a new proximity-based profiling method that relies upon a GAL4 driver, termed intestinal-kickout-GAL4 (I-KCKT-GAL4), that is exclusively expressed in intestinal progenitor cells. This method uses UV crosslinked whole animal frozen powder as its starting material to immunoprecipitate the RNA cargoes of transgenic epitope-tagged RNA binding proteins driven by I-KCKT-GAL4. When applied to the general mRNA-binder, poly(A)-binding protein, the RNA profile obtained by this method identifies 98.8% of transcripts found after progenitor cell sorting, and has low background noise despite being derived from whole animal lysate. We also mapped the targets of the more selective RNA binder, Fragile X mental retardation protein (FMRP), using enhanced crosslinking and immunoprecipitation (eCLIP), and report for the first time its binding motif in Drosophila cells. This method will therefore enable the RNA profiling of wild-type and mutant intestinal progenitor cells from intact flies exposed to normal and altered environments, as well as the identification of RNA-protein interactions crucial for stem cell function.

INTRODUCTION

The adult Drosophila intestine has become a premier model for understanding the biology and behavior of resident stem cells in their native context (Li and Jasper, 2016). One key approach has been transcript profiling that characterizes the gene expression signatures of cell types in both homeostatic and perturbed conditions (e.g. Dutta et al., 2015; Hung et al., 2020). This approach has relied on the manual dissection of intestines. Manual dissection is not only laborious and time consuming, but also precludes certain types of analysis that require large amounts of intact, undegraded starting material, such as immunoprecipitation (IP). Methods that rely upon the sequencing of immunoprecipitates associated with RNA-binding proteins (RBPs) to analyze post-transcriptional control mechanisms (e.g. RiboTag) are not currently feasible in adult Drosophila intestinal stem cells but have been crucial for characterizing the identity and differentiation mechanisms in other stem cell lineages (Baser et al., 2019; Sanz et al., 2009; Tahmasebi et al., 2018). There is therefore a need for improved methods to profile this stem cell population in Drosophila.

Manual dissection of intestines has been required because of the lack of tools to label individual intestinal cell types and no other cells in the rest of the animal. Such intestine-exclusive markers would enable use of the whole fly as starting material, allowing rapid production of large amounts of starting material. GAL4 drivers are available that can label the various individual cell types in the intestinal stem cell lineage, including intestinal stem cells (ISCs) themselves as well as their transient progenitor cell daughters (termed enteroblasts; EBs) and two differentiated cell types, enterocytes (ECs) and enteroendocrine cells (EEs) (Jiang and Edgar, 2009; Micchelli and Perrimon, 2006; Zeng et al., 2010). See Fig. 1A for a summary of intestinal cell types and their most common markers. The widely used esg-GAL4 progenitor driver labels both ISCs and EBs, which are collectively referred to as progenitor cells, whereas Dl-GAL4 and gbe-GAL4 are reported to specifically label ISCs and EBs, respectively (Jiang and Edgar, 2009; Micchelli and Perrimon, 2006; Zeng et al., 2010). However, these currently used GAL4 drivers are either known or likely to be active elsewhere in the adult (Biteau et al., 2010).

Here, we describe a method that uses a new intestinal progenitor GAL4 driver that is not expressed outside the intestine to profile the general transcriptome as well as specific RBP cargoes expressed in this cell type. We reasoned that an intersectional approach could be used to design such a GAL4 driver. Intersectional methods limit transgenic expression to cells in which two different enhancers are both active. In a recombinase-mediated intersectional method, for example, the recombinase under the control of one enhancer activates a GAL4 driver under the control of a second enhancer via recombinase-mediated removal of an intervening stop cassette (Fig. 1B). Such a recombinase-based method involving a pan-intestinal enhancer and a progenitor enhancer should limit expression to only intestinal progenitor cells. We designed our transgenes using KD recombinase (KDR), which was recently shown to mediate the excision, or ‘kick-out’, of sequences between KDR target (KDRT) sites in Drosophila cells (Nern et al., 2011). We chose to use KDR so that the resulting system could be used in tandem with other recombinases, such as FLP, for additional manipulation of intestinal cells. Because this system is designed for intestine-specific activation of a kick-out transgene, we refer to it as intestinal-kick-out, or I-KCKT.

RESULTS

I-KCKT-GAL4 labels most intestinal progenitor cells

To build the transgenes needed for our system, we first searched for defined pan-intestinal and progenitor enhancer sequences. Progenitor enhancer sequences have previously been identified in the miranda (mira) gene locus (Bardin et al., 2010), but the availability of a defined pan-intestinal enhancer was less clear. Transgenes with regulatory sequences from some intestine-specific genes have been reported (e.g. mex1-GAL4, npc1b-GAL4, etc.) (Phillips and Thomas, 2006; Voght et al., 2007), but careful inspection found that these transgenes were either not expressed throughout the intestine or not in all cell types. We therefore took a candidate gene approach to identify intestine-specific enhancers based on gene expression data reported in FlyAtlas 2 (Leader et al., 2018), testing DNA fragments from intestine-specific genes for intestinal activity. KDR transgenes containing putative enhancer sequences from βTrypsin (βTry), θTrypsin (θTry), κTrypsin (κTry), CG18404 or CG10116 were tested for intestinal activity in flies also harboring two additional transgenes, a tubulin-KDRT-stop-KDRT-GAL4.p65 kick-out transgene and a UAS-6XGFP responder (Shearin et al., 2014). GFP expression was monitored in both the larval and adult intestine for all resulting strains except for θTry; in this case, only larvae were analyzed because adults of the proper genotype failed to eclose. Although all strains displayed some GFP expression in the larval intestine, most GFP patterns were patchy and non-uniform (Fig. 1C-H). In the adult intestine, however, two KDR lines, βTry-KDR and CG10116-KDR, appeared to drive expression throughout the tissue (Fig. 1I,N). Careful inspection of these two indicated that CG10116-KDR was probably expressed in most cells throughout the midgut, raising the possibility that the associated enhancer fragment was pan-intestinal-specific and could be used to label adult intestinal progenitor cells exclusively.

Fig. 1.

KD driver analysis identifies a pan-intestinal enhancer in adults. (A) Schematic of the intestinal stem cell lineage, which includes intestinal stem cells (ISCs) and enteroblasts (EBs), collectively referred to as progenitor cells, as well as differentiated enterocytes (ECs) and enteroendocrine cells (EEs). Cell type markers are shown in gray. (B) Schematic of the I-KCKT system, which involves three transgenes: a KD-expressing transgene under the control of an intestinal enhancer, a transgene in which a KDRT-flanked stop cassette separates GAL4 from a progenitor enhancer and a UAS responder, in this case, controlling GFP expression. (C-N) KD recombinase-mediated labeling of larval (C-H) and adult (I-N) intestinal cells. KD lines contain 3.4 kb βTry (C,I), 1.6 kb θTry (D,J), 0.6 kb κTry (E,K), 2.6 kb CG18404 (F,L), 0.5 kb CG18404 (G,M) or 1 kb CG10116 (H,N) enhancer sequences. KD expression pattern was detected based on KDRT-mediated activation of a tubulin-based GAL4.p65 driver in combination with a UAS-6XGFP responder. Full genotypes are listed in Table S3.

Fig. 1.

KD driver analysis identifies a pan-intestinal enhancer in adults. (A) Schematic of the intestinal stem cell lineage, which includes intestinal stem cells (ISCs) and enteroblasts (EBs), collectively referred to as progenitor cells, as well as differentiated enterocytes (ECs) and enteroendocrine cells (EEs). Cell type markers are shown in gray. (B) Schematic of the I-KCKT system, which involves three transgenes: a KD-expressing transgene under the control of an intestinal enhancer, a transgene in which a KDRT-flanked stop cassette separates GAL4 from a progenitor enhancer and a UAS responder, in this case, controlling GFP expression. (C-N) KD recombinase-mediated labeling of larval (C-H) and adult (I-N) intestinal cells. KD lines contain 3.4 kb βTry (C,I), 1.6 kb θTry (D,J), 0.6 kb κTry (E,K), 2.6 kb CG18404 (F,L), 0.5 kb CG18404 (G,M) or 1 kb CG10116 (H,N) enhancer sequences. KD expression pattern was detected based on KDRT-mediated activation of a tubulin-based GAL4.p65 driver in combination with a UAS-6XGFP responder. Full genotypes are listed in Table S3.

To test this possibility, we evaluated CG10116-KDR activity in the intestinal progenitor cells of flies harboring a mira-containing stop cassette transgene as well as UAS-stinger-GFP, a nucleus-localized GFP reporter that is weaker than 6XGFP and therefore easier to score at single-cell resolution (Barolo et al., 2000). For these experiments, we analyzed strains containing two different mira transgenes, one encoding an enhanced version of GAL4 with the p65 transcriptional activation domain (mira-KDRT-stop-KDRT-GAL4.p65) and one with an unmodified version of GAL4 (mira-KDRT-stop-KDRT-GAL4) that could be used in conjunction with the GAL80-dependent temporal and regional gene expression targeting (TARGET) system for conditional expression (McGuire et al., 2004). For brevity, we refer to the strains combining CG10116-KDR with mira-KDRT-stop-KDRT-GAL4.p65 or mira-KDRT-stop-KDRT-GAL4 as I-KCKT-GAL4.p65 or I-KCKT-GAL4, respectively. To analyze CG10116-KDR-mediated expression in these strains, we compared GFP expression with that of a previously generated and validated progenitor reporter, mira-His2A.mCherry.HA (Miller et al., 2020), in each of the five intestinal regions (Fig. 2A-J; Fig. S1). Quantification of the percentage of mCherry+ cells that were also GFP+ indicated that ∼100% of progenitor cells were labeled in most regions of both I-KCKT strains (Fig. 2P,Q). The only two exceptions to this trend were regions 1 and 3 of I-KCKT-GAL4.p65 intestines, where only 58.7±15.5% (n=5) and 64.4±17.0% (n=5) of progenitor cells were labeled, respectively. Importantly, we also found that no GFP+ cells were mCherry in either strain, indicating that no non-progenitor cells were labeled in I-KCKT strains. To confirm this analysis, we also compared GFP expression driven by I-KCKT-GAL4 to a second validated progenitor reporter, esg-lacZ (Micchelli and Perrimon, 2006), and found similarly that almost all LacZ+ cells were also GFP+ and that no GFP+ cells were LacZ (Fig. S2). The only exception was region 3, where ∼69.9±9.0% of LacZ+ cells were GFP+. This discrepancy might indicate that esg-lacZ labels a small subset of EEs in this midgut area (Hung et al., 2020). Collectively, this analysis demonstrated that almost all progenitor cells, but few if any other cells, were labeled in the intestines of I-KCKT strains.

Fig. 2.

I-KCKT-GAL4.p65 and I-KCKT-GAL4 label most intestinal progenitor cells and ISC-KCKT-GAL4TS labels most ISCs. (A-J) Micrographs of five intestinal regions (R1-R5) stained for stinger-GFP (green), the intestinal progenitor marker mira-His2A.mCherry.HA (red) and the DAPI DNA marker (blue). GFP expression is driven by either I-KCKT-GAL4.p65 (A-E) or I-KCKT-GAL4 (F-J). (K-O) Micrographs of five intestinal regions (R1-R5) stained for stinger-GFP (green), the intestinal progenitor marker mira-His2A.mCherry.HA (red), the EB marker 3Xgbe-smGFP.V5.nls (white) and the DAPI DNA marker (blue). GFP is driven by ISC-KCKT-GAL4TS. Because of weak staining in either the red or green channel in A-O, not all cells co-expressing GFP and mCherry appear yellow. Single channel images showing this weak staining are included in Fig. S1. (P,Q) Histograms showing the percentage of mira-His2A.mCherry.HA-labeled intestinal progenitor (IP) cells per intestinal region (R1-R5) that are labeled with stinger-GFP driven by either I-KCKT-GAL4.p65 (P) or I-KCKT-GAL4 (Q). (R,S) Histograms showing the percentage of ISCs (R) and EBs (S) labeled with stinger-GFP driven by ISC-KCKT-GAL4TS. Graphs show mean±s.d. (n=5 for each intestinal region of each genotype). Full genotypes are listed in Table S3.

Fig. 2.

I-KCKT-GAL4.p65 and I-KCKT-GAL4 label most intestinal progenitor cells and ISC-KCKT-GAL4TS labels most ISCs. (A-J) Micrographs of five intestinal regions (R1-R5) stained for stinger-GFP (green), the intestinal progenitor marker mira-His2A.mCherry.HA (red) and the DAPI DNA marker (blue). GFP expression is driven by either I-KCKT-GAL4.p65 (A-E) or I-KCKT-GAL4 (F-J). (K-O) Micrographs of five intestinal regions (R1-R5) stained for stinger-GFP (green), the intestinal progenitor marker mira-His2A.mCherry.HA (red), the EB marker 3Xgbe-smGFP.V5.nls (white) and the DAPI DNA marker (blue). GFP is driven by ISC-KCKT-GAL4TS. Because of weak staining in either the red or green channel in A-O, not all cells co-expressing GFP and mCherry appear yellow. Single channel images showing this weak staining are included in Fig. S1. (P,Q) Histograms showing the percentage of mira-His2A.mCherry.HA-labeled intestinal progenitor (IP) cells per intestinal region (R1-R5) that are labeled with stinger-GFP driven by either I-KCKT-GAL4.p65 (P) or I-KCKT-GAL4 (Q). (R,S) Histograms showing the percentage of ISCs (R) and EBs (S) labeled with stinger-GFP driven by ISC-KCKT-GAL4TS. Graphs show mean±s.d. (n=5 for each intestinal region of each genotype). Full genotypes are listed in Table S3.

I-KCKT-GAL4 is not detected in non-intestinal tissue

To determine whether intestinal progenitor cells were exclusively labeled in I-KCKT adults, we performed two analyses. First, we crossed I-KCKT strains to UAS-6XGFP and visually inspected adults for GFP expression. I-KCKT strains displayed prominent 6XGFP fluorescence in the intestine but little, if any, elsewhere (Fig. 3A,B). For comparison, we similarly analyzed three other widely used drivers known to be expressed in some or all intestinal progenitor cells: a GawB P-element insertion in the escargot (esg) locus (esg-GAL4), a GawB P-element insertion in the Delta (Dl) locus (Dl-GAL4), and a Notch-responsive GAL4 reporter that contains binding sites for the Grainyhead and Suppressor of Hairless transcription factors (gbe-GAL4) (Micchelli and Perrimon, 2006; Zeng et al., 2010). All three were detected in the intestine (Fig. 3E-G); Dl-GAL4 and gbe-GAL4 also displayed prominent non-intestinal expression, whereas esg-GAL4 appeared more similar to the I-KCKT with regard to intestinal specificity. For a more rigorous analysis, we also performed comparative western blot analysis for 6XGFP expression on three different protein extracts generated from each strain: whole animal extract (total), extract from dissected gastrointestinal tracts that included the Malphigian tubules (intestine), and extract from all remaining non-intestinal tissue (carcass) (Fig. 3H). For both I-KCKT strains, GFP was detected in intestinal but not carcass extract, and the amount in the intestine was roughly similar to the amount in the total extract. In contrast, GFP was detected in both the intestine and carcass of the three comparison strains, esg-GAL4, Dl-GAL4 and gbe-GAL4. We also tested the conditional expression of mira-KDRT-stop-KDRT-GAL4 by generating an I-KCKT-GAL4TS strain that harbored an ubiquitously expressed, temperature-sensitive GAL80 transgene (tub-GAL80TS) and then performing both analyses on flies that had been incubated at the nonpermissive (18°C) and permissive (30°C) temperatures (Fig. 3H-J). This analysis detected no GFP expression at the nonpermissive temperature, and clear expression at the permissive temperature. We noted that I-KCKT-GAL4TS-driven GFP signal could also be detected in the progenitor cells of the Malphigian tubules (arrowheads in Fig. 3J), which are esg+ cells that are related to midgut progenitor cells (Singh et al., 2007). We also noted that, like esg-GAL4 but to a lesser degree, I-KCKT-GAL4 lost its progenitor-specificity in the intestinal tissue of aged animals (Fig. S3). Collectively, these results indicated that, unlike other commonly used progenitor drivers, I-KCKT-GAL4-based expression could not be detected outside the intestine.

Fig. 3.

I-KCKT-GAL4 drivers are intestine-specific. (A-G) Pictures of adult female flies showing 6XGFP expression patterns driven by I-KCKT-GAL4.p65 (A), I-KCKT-GAL4TS at 30°C (B), EB-KCKT-GAL4TS at 30°C (C), ISC-KCKT-GAL4TS at 30°C (D), esgP{GawB}NP5130-GAL4 (E), Dl05151-G-GAL4 (F) or Su(H)GBE-GAL4 (G). Insets show enlargements of abdominal regions displaying intestinal GFP expression. (H) Western blot analysis of tissue extracts from total animals (T), dissected intestines with Malphigian tubules (I), or dissected carcasses (C) probed with either anti-GFP (top) or anti-tubulin (bottom) antibodies. Lysates were generated from adults harboring UAS-6XGFP driven by I-KCKT-GAL4.p65, I-KCKT-GAL4, esgP{GawB}NP5130-GAL4, Dl05151-G-GAL4, Su(H)GBE-GAL4, I-KCKT-GAL4TS at 18°C, I-KCKT-GAL4TS at 30°C or ISC-KCKT-GAL4TS at 30°C. Longer exposures of 18°C samples were also blank, indicating low background at the nonpermissive temperature. (I,J) Micrographs of intestines showing 6XGFP expression patterns driven by I-KCKT-GAL4TS from adults kept either at 18°C (I) or 30°C (J) and counterstained for the DAPI DNA marker (blue). Yellow arrowheads indicate GFP expression at the base of the Malphigian tubules. Full genotypes are listed in Table S3.

Fig. 3.

I-KCKT-GAL4 drivers are intestine-specific. (A-G) Pictures of adult female flies showing 6XGFP expression patterns driven by I-KCKT-GAL4.p65 (A), I-KCKT-GAL4TS at 30°C (B), EB-KCKT-GAL4TS at 30°C (C), ISC-KCKT-GAL4TS at 30°C (D), esgP{GawB}NP5130-GAL4 (E), Dl05151-G-GAL4 (F) or Su(H)GBE-GAL4 (G). Insets show enlargements of abdominal regions displaying intestinal GFP expression. (H) Western blot analysis of tissue extracts from total animals (T), dissected intestines with Malphigian tubules (I), or dissected carcasses (C) probed with either anti-GFP (top) or anti-tubulin (bottom) antibodies. Lysates were generated from adults harboring UAS-6XGFP driven by I-KCKT-GAL4.p65, I-KCKT-GAL4, esgP{GawB}NP5130-GAL4, Dl05151-G-GAL4, Su(H)GBE-GAL4, I-KCKT-GAL4TS at 18°C, I-KCKT-GAL4TS at 30°C or ISC-KCKT-GAL4TS at 30°C. Longer exposures of 18°C samples were also blank, indicating low background at the nonpermissive temperature. (I,J) Micrographs of intestines showing 6XGFP expression patterns driven by I-KCKT-GAL4TS from adults kept either at 18°C (I) or 30°C (J) and counterstained for the DAPI DNA marker (blue). Yellow arrowheads indicate GFP expression at the base of the Malphigian tubules. Full genotypes are listed in Table S3.

ISC-KCKT-GAL4 labels most adult intestinal stem cells

To expand on the utility of the I-KCKT system, we investigated whether I-KCKT-based GAL4 expression could be limited to either the ISC or EB subsets of progenitor cells. To do so, we prepared strains in which the three I-KCKTTS transgenes were combined with a GAL4-silencing GAL80 transgene that contained either the EB-specific gbe synthetic enhancer (gbe-GAL80) or a fragment from the Delta locus reported to be active in ISCs (GMR24H06-GAL80) (Furriols and Bray, 2001; Guo et al., 2013). We refer to these strains as ISC-KCKT-GAL4TS and EB-KCKT-GAL4TS, respectively. Like I-KCKT-GAL4TS, both ISC-KCKT-GAL4TS and EB-KCKT-GAL4TS activities were intestine specific, based on visual and/or western blot analysis (Fig. 3C,D,H). To evaluate the cell type specificity of this activity, we drove GFP expression in both strains and compared it to a dual reporter combination that effectively distinguished ISCs and EBs (Fig. S4). The dual reporters used were the progenitor marker mira-His2A.mCherry.HA (Miller et al., 2020) and the EB-specific marker 3Xgbe-smGFP.V5.nls (Buddika et al., 2020a). For this analysis, we scored mCherry+, V5 cells as ISCs and mCherry+, V5+ cells as EBs. Most ISCs were specifically labeled in the ISC-KCKT-GAL4TS strain: 84.1±12.3% (n=5) of ISCs were labeled, whereas 11.2±10.0% (n=5) of EBs were labeled across all five regions of the intestine (Fig. 2K-O,R,S; Fig. S1). In contrast, EB-specific labeling in EB-KCKT-GAL4TS was less effective: only 30.7±20.1% (n=5) of EBs were labeled, whereas 57±19.8% (n=5) of ISCs were labeled across all five regions of the intestine (Fig. S5). Various possibilities could explain this EB-KCKT-GAL4TS result, including that the GMR24H06 enhancer fragment was not active in most ISCs and/or that GAL80 activity continued in EB cells. Nevertheless, the analysis indicated that ISC-KCKT-GAL4TS specifically labeled most ISCs.

CLIP-seq of PABP using I-KCKT-GAL4 identifies progenitor-expressed genes

We next investigated whether these I-KCKT strains could be used to streamline current methods in order to profile progenitor cells molecularly. RNA-seq profiling methods that rely on esg-GAL4, for example, require labor-intensive dissection of hundreds of intestines followed by digestion of this tissue to release labeled progenitor cells for fluorescence-activated cell sorting (FACS) (Dutta et al., 2015; Fast et al., 2020; Korzelius et al., 2019; Li et al., 2018) (Fig. 4A, left). Because I-KCKT-GAL4 activity was limited to intestinal progenitor cells, we tested whether crosslinking immunoprecipitation (CLIP) combined with sequencing (CLIP-seq) analysis of I-KCKT-GAL4-driven FLAG-tagged poly-A binding protein (PABP) that used whole animal lysate as its starting material would recover progenitor cell mRNA (Fig. 4A, right). PABP is a general mRNA-binding protein that has been used to profile the mRNA transcriptomes of other cell types (Hwang et al., 2016; Yang et al., 2005). We first verified that the FLAG-tagged version of PABP that had previously been used to profile mRNAs in Drosophila photoreceptor cells was expressed in progenitor cells when crossed to I-KCKT-GAL4TS (Fig. S6) (Yang et al., 2005). Then, we crushed frozen flies from this strain to generate a fine powder, subjected this powder to UV crosslinking to covalently link protein and RNA complexes, immunoprecipitated PABP using anti-FLAG beads, and recovered the associated RNAs. After ribosomal RNA (rRNA) depletion, independent libraries were prepared from duplicate PABP immunoprecipitates (PABP IPs). In parallel, two libraries were also prepared from RNA extracted from the starting whole animal lysate, and we used these ‘input’ libraries as our normalization controls. Differential gene expression analysis found 1661 transcripts enriched in the PABP IP samples and 3293 enriched in input (Fig. 4B; Table S1). Representative examples of these respective classes included the progenitor enriched esg transcript and the ovary enriched oskar transcript (osk) (Fig. 4C,D). The PABP IP enriched set contained many additional previously characterized progenitor genes including Delta, sox21a, sox100b and zfh2, suggesting an overall similarity to the progenitor transcriptome.

Fig. 4.

CLIP-seq of PABP driven by I-KCKT-GAL4 identifies progenitor expressed genes. (A) Representation of conventional FACS (left) and I-KCKT-based (right) intestinal progenitor RNA sequencing. (B) Volcano plot of differentially expressed genes in PABP IP versus total input. Each dot represents a single gene. Yellow indicates a false discovery rate adjusted P-value (FDR)<0.05 and a log2 fold change <1 or >−1. Green indicates log2 fold change >1 or <−1 and FDR≥0.05. Red indicates FDR<0.05 and log2 fold change >1 or <−1. A selected set of significantly changed genes are shown in blue. (C,D) Genome browser tracks of normalized total input and PABP IP at the esg (C) or osk (D) loci. Note that the two replicates of the total input (green and yellow) and the two replicates of the PABP IP (red and blue) have been overlaid. (E) Upset plot showing the overlap of genes identified by PABP CLIP-seq versus RNA-seq of FACS-isolated progenitor cells reported in Korzelius, Dutta, Buddika and Fast databases. Numbers above each bar show the size of each intersection. Only a select set of meaningful overlaps are shown. (F) Correlogram of Spearman's rho values for pairwise comparisons of progenitor expressed genes in PABP CLIP-seq; progenitor RNA-seq reported in Buddika, Dutta, Fast or Korzelius; or RNA-seq from Drosophila female head. (G) Heatmaps of odds ratio (green) and Jaccard index (gray) values for pairwise comparisons of PABP IP CLIP-seq, RNA-seq of total input or RNA-seq of dissected female heads against FACS-based progenitor RNA-seq. (H,I) Venn diagrams showing genes significantly enriched in PABP-CLIP that overlap with the top 10% of input genes based on normalized expression values (H) or genes present in input but not in any of the FACS-based RNA-seq gene lists (I). (J) Heatmaps of odds ratio (red) and Jaccard index (blue) values of pairwise comparisons of PABP IP enriched or PABP IP depleted genes with six other gene sets: fat body, head, ovary or testis enriched genes (identified relative to midgut genes) as well as midgut or progenitor enriched genes from Buddika (identified relative to whole animal input). **P<0.01, ***P<0.001, ****P<0.0001 based on Fisher's exact test.

Fig. 4.

CLIP-seq of PABP driven by I-KCKT-GAL4 identifies progenitor expressed genes. (A) Representation of conventional FACS (left) and I-KCKT-based (right) intestinal progenitor RNA sequencing. (B) Volcano plot of differentially expressed genes in PABP IP versus total input. Each dot represents a single gene. Yellow indicates a false discovery rate adjusted P-value (FDR)<0.05 and a log2 fold change <1 or >−1. Green indicates log2 fold change >1 or <−1 and FDR≥0.05. Red indicates FDR<0.05 and log2 fold change >1 or <−1. A selected set of significantly changed genes are shown in blue. (C,D) Genome browser tracks of normalized total input and PABP IP at the esg (C) or osk (D) loci. Note that the two replicates of the total input (green and yellow) and the two replicates of the PABP IP (red and blue) have been overlaid. (E) Upset plot showing the overlap of genes identified by PABP CLIP-seq versus RNA-seq of FACS-isolated progenitor cells reported in Korzelius, Dutta, Buddika and Fast databases. Numbers above each bar show the size of each intersection. Only a select set of meaningful overlaps are shown. (F) Correlogram of Spearman's rho values for pairwise comparisons of progenitor expressed genes in PABP CLIP-seq; progenitor RNA-seq reported in Buddika, Dutta, Fast or Korzelius; or RNA-seq from Drosophila female head. (G) Heatmaps of odds ratio (green) and Jaccard index (gray) values for pairwise comparisons of PABP IP CLIP-seq, RNA-seq of total input or RNA-seq of dissected female heads against FACS-based progenitor RNA-seq. (H,I) Venn diagrams showing genes significantly enriched in PABP-CLIP that overlap with the top 10% of input genes based on normalized expression values (H) or genes present in input but not in any of the FACS-based RNA-seq gene lists (I). (J) Heatmaps of odds ratio (red) and Jaccard index (blue) values of pairwise comparisons of PABP IP enriched or PABP IP depleted genes with six other gene sets: fat body, head, ovary or testis enriched genes (identified relative to midgut genes) as well as midgut or progenitor enriched genes from Buddika (identified relative to whole animal input). **P<0.01, ***P<0.001, ****P<0.0001 based on Fisher's exact test.

To address this possibility rigorously, we compared the PABP-associated transcriptome to the RNA-seq profile of FAC-sorted progenitor cells that was obtained by identifying the genes common to four previously reported RNA-seq datasets. We reasoned that a common transcriptome identified in multiple independent experiments would reduce the batch effects of individual datasets. The datasets we used were reported by Buddika et al. (2020b preprint), Dutta et al. (2015), Fast et al. (2020) and Korzelius et al. (2019), and we refer to them as Buddika, Dutta, Fast and Korzelius, respectively. All four datasets were generated from sorted esg-GAL4+ progenitor cells, but Buddika used the same rRNA-depletion library preparation method that we used for the PABP libraries, whereas Dutta, Fast and Korzelius used poly(A)-selected RNA for their libraries (see Fig. S7A for a summary of the various datasets). Read counts for the four FACS datasets as well as the PABP IP dataset were normalized in parallel to facilitate direct comparison; a transcript was considered to be expressed in a dataset if it had a normalized expression value greater than 10 counts per million (cpm) in each replicate. We chose this cutoff to eliminate background noise, probably caused by genes that were expressed at low levels. Pairwise comparisons between the genes expressed in each FACS dataset indicated that they were largely, but not completely, overlapping (Fig. S7B). Together, these four datasets shared 5027 genes in common, which we considered the core progenitor transcriptome. The PABP IP dataset contained all but 61 of these, indicating that PABP IP identified 98.8% (4966 out of 5027) of core progenitor transcripts (Fig. 4E). An additional 1839 transcripts were found in the PABP IP, and 76.2% of these were also found in at least one other FACS dataset. A total of 437 transcripts (6.4%) were unique to the PABP IP, in line with the numbers unique to each FACS dataset, which probably reflect differences caused by the technical approach. Differences could include normal deviations associated with distinct experimental settings, library preparation methods or differences related to the preparation of flash frozen versus mechanically disrupted samples. Together, these results indicate that the PABP IP method effectively identifies progenitor transcripts.

To evaluate the significance of the overlap between these five datasets, Spearman correlation matrices were generated and visualized as correlograms. As a negative control for this analysis, we also prepared a sixth dataset of head tissue data from FlyAtlas 2 (Leader et al., 2018). As expected, the PABP IP dataset correlated well with all FACS-based progenitor datasets, and all five datasets showed a similar, lower correlation with the non-intestinal dataset (Fig. 4F). We then used the GeneOverlap R package to perform Fisher's exact test to evaluate the statistical significance of the overlap between different datasets. Fisher's exact test also computes both an odds ratio and a Jaccard index, which represent the strength of association and similarity between two datasets, respectively. The PABP IP scored highly in both of these indices compared with each of the four FAC-sorted datasets and significantly higher than either the total input or head tissue datasets, further indicating the similarity between the PABP IP and FACS-based datasets (Fig. 4G).

Finally, we evaluated whether the PABP IP dataset contained progenitor enriched transcripts, which we identified from systematic re-analysis of the four FACS-based datasets. For this re-analysis, differential expression analysis was used to select for transcripts enriched in Buddika, Dutta, Fast and Korzelius compared with two sources of whole animal data, total input from the PABP experiment or a re-analyzed whole female adult dataset from FlyAtlas 2. The PABP IP dataset was then compared with each of these computed progenitor enriched datasets. As expected, PABP IP enriched genes showed significantly low P-values and high odds ratio and Jaccard index scores, strengthening the positive relationship of PABP and FACS-based datasets (Fig. S7C). This bioinformatic analysis supported the conclusion that the PABP IP dataset represents the gene expression profile of progenitor cells.

CLIP-seq of I-KCKT-GAL4-driven PABP has little background

Having established that the PABP IP dataset contained progenitor genes, we also sought to evaluate how much nonspecific background was represented in this dataset. Hypothesizing that the most abundant transcripts in the whole animal lysate were the most likely sources of possible contamination, we first identified the top 10% of genes in the input (based on the cpm values) and found 1049 genes. Only 207 of these genes were also identified in the PABP IP enriched gene set, suggesting that transcripts abundantly expressed in the starting lysate were not preferentially recovered in the IP (Fig. 4H). Supporting this observation, we identified 854 genes in the input that were not present in any of the RNA-seq profiles and therefore not expected to be present in the PABP IP (Fig. S7D) and confirmed that only four of these transcripts were present in the PABP IP enriched gene list (Fig. 4I). Finally, we again used re-analyzed FlyAtlas 2 data to identify genes enriched in the fat body, head, ovary and testis relative to midgut, and then compared those tissue enriched genes to the genes either enriched in or depleted from the PABP IP. We found that odds ratio and Jaccard index scores were low for PABP enriched genes, indicating minimal overlap, and significantly higher for PABP depleted genes (Fig. 4J). As a positive control for this analysis, we performed the same comparisons with genes enriched in the FlyAtlas 2 midgut dataset as well as genes enriched in the Buddika FACS dataset. As expected, PABP IP enriched genes show highest association with genes enriched in FACS-isolated progenitors, and significant overlap was also observed with the midgut enriched gene list (Fig. 4J). Collectively, this analysis indicates that the PABP dataset contains minimal background noise from other tissues, probably because of the stringency of the washes after the crosslinking step.

I-KCKT-based eCLIP analysis identifies FMRP-bound mRNAs in the intestine

To illustrate the breadth of its possible applications, we employed I-KCKT-GAL4 to identify the RNA cargo of a more selective RBP, Fragile X mental retardation protein (FMRP, also known as FXR1 or FMR1), specifically in progenitor cells using enhanced CLIP (eCLIP)-seq. FMRP limits ISC expansion but its mRNA targets in these cells are unknown (Luhur et al., 2017). The eCLIP method identifies RBP binding sites at single nucleotide resolution, and contains a number of key modifications in comparison to the CLIP method we used for PABP: (1) size fractionation-based purification of the RBP-RNA complex, (2) RNase treatment and adapter ligations that map the exact position of crosslinking and (3) preparation of a size-matched input (SMI) negative control sample for stringent normalization (Van Nostrand et al., 2016). We first verified that UAS-FMRP.FLAG.GFP in combination with I-KCKT-GAL4TS was detected in progenitor cells after 2 and 10 days of induction and that it caused a reduction in progenitor cell number when induced for 10 but not 2 days (Fig. S6). We then prepared FMRP IP and SMI libraries from whole animals collected at the 2 day time point (Fig. S8A,B). After removing rRNA reads, multimapping reads and duplicated reads, about 8% (8,438,543), 5% (2,487,118) and 7% (2,545,165) of total reads were recovered as usable reads from the SMI and two IP samples (UV1 and UV2), respectively (Fig. S8C), consistent with the recovery rate from other eCLIP analyses (Van Nostrand et al., 2016).

Peak calling analysis identified 13,297 reproducible FMRP binding sites across 1829 genes (Table S2) and indicated that there was high correlation between replicates (Fig. S8D), demonstrating the robustness of our modified eCLIP method. Analysis of peak distribution indicated that >85% of FMRP binding sites were in protein-coding transcripts and, within these transcripts, the majority of sites were in coding sequences (Fig. 5A,B). Both of these results were consistent with previous characterizations of FMRP distribution in multiple species, including proximity-based analyses in mice and human tissues as well as activity-based analyses in fly cells (Darnell et al., 2011; Li et al., 2020; McMahon et al., 2016). In addition, we noted that this analysis identified several confirmed direct targets of Drosophila FMRP found by genetic or targeted biochemical means, including CaMKII, chic, Dscam1, futsch, ninaE, rg and tral (Cvetkovska et al., 2013; Kim et al., 2013; Monzo et al., 2010; Reeve et al., 2005; Sears et al., 2019; Sudhakaran et al., 2014; Wang et al., 2017; Zhang et al., 2001). As the first CLIP-based assay to characterize Drosophila FMRP, this analysis also identified CAUUG(A/U) as its top binding motif (Fig. 5C), consistent with prior identification of the AUUG sequence in one of the top binding motifs of human FXR1 (Feng et al., 2019).

Fig. 5.

eCLIP of FMRP driven by I-KCKT-GAL4 identifies intestinal target mRNAs. (A,B) Pie chart representing the percentage of target gene types (A) and binding site distribution (B) in mRNAs identified in FMRP eCLIP. (C) The top FMRP binding motif identified using DREME. (D) Venn diagram showing overlap between FMRP target genes with the cumulative progenitor transcriptome (identified by PABP IP or RNA-seq of FACS isolated progenitors). (E) Genome browser tracks of normalized size-matched input (SMI) and FMRP IP at the esg locus of the Drosophila genome. Note that the two replicates of FMRP IP were merged prior to genome browser visualization. Locations of four FMRP binding regions are indicated with gray bars. (F) Bar plot showing fold enrichment of 12 mRNAs in FMRP IP compared with whole intestinal input as determined by qPCR. Negative controls are separated from the other genes by a dotted line. (G) Box and whiskers plot showing the log value of mean transcript length of FMRP-bound versus unbound transcripts that are expressed in intestinal progenitors. Note that the length (bp) of the longest transcript isoform was used for this analysis whenever a gene had multiple transcript isoforms. *P<0.05, ****P<0.0001.

Fig. 5.

eCLIP of FMRP driven by I-KCKT-GAL4 identifies intestinal target mRNAs. (A,B) Pie chart representing the percentage of target gene types (A) and binding site distribution (B) in mRNAs identified in FMRP eCLIP. (C) The top FMRP binding motif identified using DREME. (D) Venn diagram showing overlap between FMRP target genes with the cumulative progenitor transcriptome (identified by PABP IP or RNA-seq of FACS isolated progenitors). (E) Genome browser tracks of normalized size-matched input (SMI) and FMRP IP at the esg locus of the Drosophila genome. Note that the two replicates of FMRP IP were merged prior to genome browser visualization. Locations of four FMRP binding regions are indicated with gray bars. (F) Bar plot showing fold enrichment of 12 mRNAs in FMRP IP compared with whole intestinal input as determined by qPCR. Negative controls are separated from the other genes by a dotted line. (G) Box and whiskers plot showing the log value of mean transcript length of FMRP-bound versus unbound transcripts that are expressed in intestinal progenitors. Note that the length (bp) of the longest transcript isoform was used for this analysis whenever a gene had multiple transcript isoforms. *P<0.05, ****P<0.0001.

The FMRP mRNA cargo includes ∼20% of the protein-coding transcripts expressed in progenitor cells that were identified by the PABP IP analysis described above (Fig. 5D). These FMRP targets were significantly associated with processes related to stem cell proliferation, stem cell maintenance, cell differentiation and translation repressor activity (Fig. S8E), consistent with the known roles of FMRP in stem cell populations such as ISCs (Luhur et al., 2017). One representative example of these FMRP targets was esg (Fig. 5E), which plays crucial roles in progenitor cells where it is a known target of post-transcriptional control (Antonello et al., 2015; Korzelius et al., 2014). We also noted that ∼7% (111/1560) of the protein-coding FMRP cargo was made up of transcripts not identified in either the PABP IP or any of the four progenitor cell RNA-seq datasets (Fig. 5D). Of these 111, 31 had low cpm values in the progenitor datasets that were below our cutoff of 10 cpm for expression. These might represent weakly expressed progenitor transcripts that were bound by FMRP. The other 80 were transcripts that had 0 cpm in all progenitor datasets, indicating that they were not expressed in progenitor cells and therefore were most probably nonspecific background in the FMRP IP. More stringent wash steps during immunopurification might reduce this ∼5% background even more.

To verify the eCLIP-identified FMRP targets, we first performed quantitative PCR on RNA immunoprecipitated with endogenous FMRP from wild-type intestines (Fig. S8H). We chose seven target genes (apt, Egfr, esg, Mes2, shg, Tet and Unr) and five negative controls (CG15784, CG5767, gapdh1, gapdh2 and Pyk) for this analysis. As expected, the FMRP target genes showed between 2- and 18-fold enrichment in FMRP IP relative to whole intestine input, whereas the negative controls were either not significantly enriched or significantly de-enriched in the FMRP IP (Fig. 5F). In addition, we found that the length distribution of the FMRP-bound mRNAs was significantly longer than the other protein-coding transcripts expressed in progenitor cells (Fig. 5G), consistent with recent ribosome footprinting showing that FMRP preferentially regulates the translation of large proteins in Drosophila oocytes (Greenblatt and Spradling, 2018). Together, these results indicate that applying eCLIP to whole animal tissue containing progenitor cells that express FLAG-tagged FMRP effectively recovered the FMRP cargo from these cells.

DISCUSSION

Here we describe a method to purify the RNA cargoes associated with intestinal progenitor RBPs using whole animal lysate as the starting material. The method relies on a new progenitor driver, I-KCKT-GAL4, that, unlike esg-GAL4, is not expressed outside the intestine. This method has wide applicability for profiling the general PABP-bound transcriptome of wild-type progenitor cells. In addition, it opens the possibility for molecular characterization of the stem cell targets of selective RBPs, a growing number of which have been shown to play crucial roles in progenitor cells (e.g. LIN28, FMRP, SPEN, TIS11) (Andriatsilavo et al., 2018; Chen et al., 2015; Luhur et al., 2017; McClelland et al., 2017). Furthermore, this method can be used in conjunction with additional GAL4-based mRNA profiling methods such as TU-tagging and ribosome profiling (Chen and Dickman, 2017; Hida et al., 2017), as well as methods that target other classes of RNAs, including microRNAs (Luhur et al., 2013). Although the method is currently limited to progenitor cells, its applicability could be expanded by generating kick-out transgenes that drive expression in additional intestinal cell types such as EEs and ECs. Furthermore, although we focused on protein-RNA interactions in this study, we also expect that I-KCKT-GAL4 in combination with mass spectrometry can be used to identify protein-protein interactions and thereby characterize protein complexes in progenitor cells.

The I-KCKT method offers a number of advantages over current progenitor cell profiling methods, which typically involve FAC sorting of progenitor cells from dissected and then mechanically disrupted intestines. The use of whole animals rather than dissected intestines as starting material greatly expedites sample preparation, enabling the analysis of larger numbers of samples that could be used to test additional conditions and manipulations. In addition, because sample preparation is fast, this approach might effectively capture labile RNA profiles. Along these lines, elimination of the FAC sorting step should also improve the accuracy of results, as FAC sorting is known to affect gene expression profiles because of the time and mechanical disruption associated with this step (Andrä et al., 2020; Richardson et al., 2015). In addition, this method could make molecular profiling of progenitors more accessible, because cell sorting runs can be costly and the equipment is not always available.

Although the conditional expression of transgenic PABP to retrieve bulk mRNA has been used to approximate the transcriptomes of specific cell types in a variety of species (Roy et al., 2002; Tenenbaum et al., 2000; Yang et al., 2005), some issues regarding PABP should be considered when using this approach. PABP displays some differential binding affinity to poly(A) RNAs that can introduce bias (Yang et al., 2005). In addition, retrieval of RNAs with poly(A) tails misses deadenylated mRNAs as well as some noncoding RNAs, thereby biasing the resulting datasets toward actively translating coding transcripts. Finally, the cell toxicity associated with transgenic expression of PABP might alter endogenous gene expression. Nevertheless, the cargo of PABP is better reflective of cell transcriptomes than the cargos of other general RNA binders and, furthermore, can be analyzed with paperCLIP to finely map 3′-UTR ends and uncover alternative polyadenylation sites in a cell type-specific manner (Hwang et al., 2016; Tenenbaum et al., 2000). It should also be noted that PABP IP results are suitable for comparison with RNA-seq libraries that are generated via poly(A)-selection, rather than rRNA depletion, as both methods select for polyadenylated transcripts.

As the first CLIP-based analysis of Drosophila FMRP, this study reports both novel mRNA targets as well as the binding motif of this protein in intestinal progenitor cells. The set of FMRP targets identified here displays partial overlap with those identified in two recent studies that used either ribosome profiling or a proximity-based activity assay to identify FMRP targets in oocytes, cultured cells and neurons (Greenblatt and Spradling, 2018; McMahon et al., 2016). Differences in the targets reported by these studies reflect not only their distinct technical approaches but also the variety of cell types analyzed, as FMRP is known to display cell type binding preferences in other species (Maurin et al., 2018). Because loss of FMRP leads to fragile X syndrome (FXS), the leading form of intellectual disability in humans, analysis of FMRP has focused on its activity in differentiated neurons. However, FMRP also functions in stem cell populations, and its dysregulation in these cells probably contributes to understudied FXS symptoms, including elevated brain size, accelerated growth and gastrointestinal problems (Luhur et al., 2017). Thus, future analysis of the stem cell targets of FMRP identified here may characterize regulatory relationships that are relevant to therapies designed to treat the entire repertoire of FXS symptoms.

A current key limitation of the I-KCKT method is its purification of epitope-tagged UAS-based transgenic protein rather than endogenous protein. This raises the concern that the transgenic protein might be expressed either at higher-than-endogenous levels, which might cause inappropriate interactions with non-target mRNAs, or at lower-than-endogenous levels, which might cause non-representative interactions with target mRNAs. Because the TARGET system controlling transgene expression is temperature sensitive, this concern can be at least partially addressed by testing multiple temperatures in the permissive range to identify conditions supporting physiological protein expression. As an alternative approach to address this concern, our current effort is focused on modifying endogenous RBP loci to contain FRT-flanked epitope tags. I-KCKT-based expression of FLP should then lead to the production of epitope-tagged endogenous protein in progenitor cells that would be available for eCLIP-based analysis. In addition, parallel eCLIP analysis of wild-type and RNA-binding-domain mutant versions of an RBP will provide even more stringent ways of eliminating background and identifying bona fide in vivo RNA targets. Thus, we expect that I-KCKT-based methods will allow unprecedented analysis of RNA-based mechanisms of progenitor cells.

MATERIALS AND METHODS

Drosophila strains and husbandry

All fly strains were cultured on standard Bloomington Drosophila stock center media (https://bdsc.indiana.edu/information/recipes/bloomfood.html). Flies were reared in 18°C, 25°C and 30°C incubators set for a 12 h light-dark schedule and 65% humidity. The genotypes of all strains used in this study are listed in Table S3. Transgenesis to create new strains was performed by Rainbow Transgenic Flies using plasmid DNA described below. Additional stocks were obtained from the Bloomington Drosophila Stock Center ({20XUAS-6XGFP}attP2, P{UAS-Stinger}2, P{tubP-GAL80[ts]}20, P{UAS-dPF}D), Steven Hou of the National Institutes of Health, USA (P{Su(H)GBE-GAL4, P{GawB}Dl05151-G), the Kyoto Drosophila Stock Center (P{GawB}NP5130) and Steve Stowers of Montana State University, USA ({20XUAS-DSCP-6XGFP}attP2). For GAL80-dependent conditional expression experiments, flies were reared at 18°C, collected over 2 days and then shifted to 30°C for up to 10 days before analysis. The following strains have been deposited at the Bloomington Drosophila Stock Center: (1) I-KCKT-GAL4.p65 (BDSC #91526. Full genotype: mira-KDRT>-dSTOP-KDRT>-GAL4.p65}attP40 ; {CG10116-KD.PEST}attP2). (2) I-KCKT-GAL4TS (BDSC #91410 Full genotype: {mira -KDRT>-dSTOP-KDRT>-GAL4}attP40, P{tubP-GAL80[ts]}20 ; {CG10116-KD.PEST}attP2). (3) ISC- KCKT-GAL4TS (BDSC #91411 Full genotype: {gbe-GAL80}ZH-2A ; {mira -KDRT>-dSTOP-KDRT>-GAL4}attP40, P{tubP-GAL80[ts]}20 ; {CG10116-KD.PEST}attP2).

Transgenes

KD enhancer transgenes

PCR products containing enhancer fragments amplified from genomic DNA were cloned into the HindIII and AatII sites of pJFRC161 (20XUAS-IVS-KD.PEST; a gift from Gerald Rubin of Janelia Farm Research Campus, USA, Addgene plasmid #32140) using HiFi DNA Assembly Mix (New England Biolabs). Oligo pairs used to amplify enhancer fragments from βTry (3437/3438), θTry (3439/3440), κTry (3446/3447), CG18404 (3443/3444 for the 2.6 kb fragment, 3445/3443 for the 0.5 kb fragment) and CG10116 (3441/3442) are shown in parentheses; oligo sequences are reported in Table S4. Junctions of resulting plasmids were verified by sequencing prior to the preparation of plasmid DNA for transgenesis. KD enhancer transgenes were inserted into the attP2 landing site.

KDRT GAL4 transgenes

pJFRC164 (21XUAS-KDRT>-dSTOP-KDRT>-myr.RFP, a gift from Gerald Rubin; Addgene plasmid #32141) was used as a backbone for KDRT-flanked stop cassette plasmids. Four-way HiFi DNA Assembly reactions were performed with (1) 7.2 kb HindIII/XbaI-digested pJFRC164 plasmid backbone, (2) 1.4 kb AatII/XhoI-digested pJFRC164 KDRT>-dSTOP-KDRT fragment, (3) PCR-amplified GAL4-containing fragments and (4) PCR amplified enhancer-containing fragments. GAL4.p65 was amplified from pBPGAL4.2.p65Uw (a gift from Gerald Rubin; Addgene plasmid #26229) with oligo pair 3433/3434. The GAL4 sequence was amplified from pBPGAL4.1Uw (a gift from Gerald Rubin; Addgene plasmid #26226) with oligo pair 4401/4402. Enhancer fragments included a 2.6 kb tubulin fragment amplified with oligo pair 3626/3627, 2.6 kb 5′ and 1.6 kb 3′ miranda fragments amplified with oligo pairs 3489/3490 and 3491/3492, respectively, and a 3.4 kb delta fragment amplified with oligo pair 3431/3432. Plasmid junctions and open reading frames were verified by sequencing prior to the preparation of DNA for transgenesis. Oligo sequences are reported in Table S4, and complete plasmid sequences are available upon request. KDRT transgenes were inserted into the attP40 landing site.

GAL80 transgenes

For GMR24H06-GAL80, the R24H06 enhancer fragment was PCR amplified from genomic DNA and subcloned upstream of the GAL80 open reading frame in a pATTB-containing transformation plasmid. Transgene 3xgbe-GAL80 was generated in a similar manner, but using oligos encoding three tandem copies of the gbe sequence (Furriols and Bray, 2001). Both of these GAL80 transgenes were inserted into the ZH-2A landing site.

Dissections and immunostaining

Adult female flies were dissected in ice-cold 1× PBS and fixed in 4% paraformaldehyde (Electron Microscopy Sciences) in PBS for 45 min. For all samples to be stained with antibodies other than anti-Dl, tissue was then washed in 1× PBT (1× PBS, 0.1% Triton X-100) and then blocked (1× PBT, 0.5% BSA) for at least 45 min. Samples were incubated at 4°C overnight with primary antibodies, including rabbit anti-RFP (600-401-379, Rockland; 1:1000), mouse anti-V5 (MCA1360GA, Bio-Rad; 1:250), chicken anti-beta-galactosidase (ab9361, Abcam; 1:2000), and mouse anti-FLAG (F1804, Sigma; 1:1000). The following day, samples were washed in 1× PBT and incubated for 2-3 h with secondary antibodies, including Alexa Fluor 488- and 568-conjugated goat anti-rabbit, anti-mouse, anti-rat and anti-chicken antibodies (Life Technologies; 1:1000). Finally, samples were washed in 1× PBT, including one wash with DAPI (1:1000), and mounted in Vectashield mounting medium (Vector Laboratories).

For staining with anti-Dl antibodies, an alternative sample preparation scheme adapted from Lin et al. (2008) was followed. Briefly, intestines were dissected in ice-cold Grace's insect medium (Lonza Bioscience) and fixed in a 1:1 (v/v) mixture of heptane (Sigma) and 4% paraformaldehyde (Electron Microscopy Sciences) in water for 15 min. The bottom aqueous paraformaldehyde layer was removed, 500 µl of ice-cold methanol added and the mixture shaken vigorously for 30 s. The methanol-heptane mixture was removed and incubated with 1 ml ice-cold methanol for 5 min. Next, samples were gradually rehydrated with a series of 0.3% PBT (1× PBS, 0.3% Triton X-100):methanol (3:7, 1:1, 7:3) washes, then washed with 0.3% PBT alone for another 5 min and blocked (0.3% PBT, 0.5% BSA) for at least 45 min. The primary antibody was mouse anti-Delta (C594.9B, Developmental Studies Hybridoma Bank, DSHB; 1:500); secondary antibodies are described above. Samples were mounted in ProLong Diamond mounting medium (Invitrogen).

Microscopy and image processing

Images of whole flies were collected on a Zeiss Axio Zoom microscope and images of dissected intestines on a Leica SP8 Scanning Confocal microscope. Samples to be compared were collected under identical settings on the same day, image files were adjusted simultaneously using Adobe Photoshop CC and figures assembled using Adobe Illustrator CC.

Western blot analysis

Female Drosophila flies were used for protein isolation. Extracts were prepared from either whole animals or separately from the intestines and Malphigian tubules or from the remaining carcass tissue of dissected animals. Tissues were lysed in protein lysis buffer containing 25 mM Tris-HCl pH 7.5, 150 mM KCl, 5 mM MgCl2, 1% NP-40, 0.5 mM DTT and 1× Complete Protease Inhibitor Cocktail (Sigma). Protein extracts were resolved on a 4-20% gradient polyacrylamide gel (Bio-Rad), transferred to Immobilon-P membrane (Millipore) and probed with rabbit anti-GFP (ab290, Abcam; 1:10,000) or mouse anti-α-tubulin (12G10, DSHB; 1:1000) antibodies. For IP verification, blots were stained with mouse anti-FLAG (F1804, Sigma; 1:3500) or mouse anti-FMRP (5A11, DSHB; 1:500). Subsequently, blots were washed extensively with 1× TBST (1× TBS, 0.1% Tween-20) and incubated with anti-rabbit or anti-mouse conjugated HRP secondary antibodies. After extensive secondary washes with 1× TBST, blots were treated with ECL detection reagents (Thermo Scientific) and finally exposed to chemiluminescence films (GE Healthcare).

CLIP and eCLIP library construction and sequencing

PABP CLIP

Whole flies were collected and immediately snap frozen in liquid nitrogen and stored at −80°C until sample preparation. For each replicate, 200 flies (100 male and 100 female) were ground into a fine powder using a pre-cooled mortar and pestle on dry ice. The powder was irradiated three times at 120 mJ/cm2 in a UV crosslinker (Stratagene) and transferred to a 2 ml tube containing 1 ml lysis buffer (150 mM NaCl, 50 mM Tris-HCl pH7.5, 1 mM EDTA, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 5× Complete Protease Inhibitor Cocktail (Sigma, 11836145001) and 4 U/ml RNase Inhibitor Murine (BioLabs, M0314L) using a cold spatula. Any remaining fly powder was washed off the spatula with an additional 1 ml lysis buffer. Tubes were incubated on ice for 15-30 min, with mixing every 5 min. Epitope-tagged proteins were immunoprecipitated with anti-FLAG magnetic beads (Sigma, M8823) following the manufacturer's instructions. RNA was eluted from the beads with Proteinase K (Sigma, AM2546), extracted with TRIzol LS Reagent (Ambion, 10296028) and used to make libraries with the Ovation SoLo RNA-Seq System (Tecan Genomics, S02240). Purified libraries were submitted for paired-end 75 bp sequencing on the Illumina NextSeq platform at the Center for Genomics and Bioinformatics at Indiana University, Bloomington.

FMRP eCLIP

For FMRP eCLIP libraries, the protocol described by Van Nostrand et al. (2016), was followed. Briefly, lysate preparation and immunoprecipitation was performed as described in the PABP CLIP section above, except that 8 U of RNAse I (Ambion, AM2295) was used per sample. Then, whole lysate and FMRP IP samples were phosphatase-treated, ligated with a 3′ RNA adapter, size fractionated to extract the 100-150 kD fraction, reverse transcribed, phosphatase treated again and ligated with a 5′ DNA adapter. Resulting eCLIP libraries were submitted for paired-end 75 bp sequencing on the Illumina NextSeq platform at the CGB in Indiana University, Bloomington or paired-end 100 bp sequencing on the DNBseq platform at BGI Genomics in China.

RNA-seq and PABP CLIP-seq data analysis

RNA-seq datasets were obtained from the following references and are referred to in the text by the surname of the first author: Buddika et al., 2020b preprint; Dutta et al., 2015; Fast et al., 2020; Korzelius et al., 2019; Leader et al., 2018. See Fig. S7A for summaries of each dataset. Both RNA-seq and PABP CLIP-seq read files were processed and aligned to the Berkeley Drosophila Genome Project (BDGP) assembly release 6.28 (Ensembl release 99) reference genome using a Python-based in-house pipeline (https://github.com/jkkbuddika/RNA-Seq-Data-Analyzer v1.0). Briefly, the quality of raw sequencing files was assessed using FastQC version 0.11.9, low quality reads were removed using Cutadapt (Martin, 2011) version 2.9, reads aligning to rRNAs were removed using TagDust2 (Lassmann, 2015) version 2.2 and remaining reads mapped to the Berkeley Drosophila Genome Project (BDGP) assembly release 6.28 (Ensembl release 99) reference genome using STAR (Dobin et al., 2013) genome aligner version 2.7.3a and de-duplicated using SAMtools (Li et al., 2009) version 1.10. Subsequently, the aligned reads were counted to the nearest overlapping feature using the Subread (Liao et al., 2019) version 2.0.0 function featureCounts. Finally, bigWig files representing RNA-seq coverage were generated using deepTools (Ramírez et al., 2016) version 3.4.2 with the following settings: normalizeUsing CPM, binSize 1. All programs, versions and dependencies required to execute the RNA-seq data analyzer are described in the user guide and can be installed using miniconda. For the transcriptome analysis, raw read counts from all datasets (see Fig. S7A for details of datasets) were normalized simultaneously using the trimmed mean of M-values (TMM) method. Normalized expression values (counts per million) were calculated using functions implemented in the Bioconductor package edgeR (Robinson et al., 2010) version 3.28.1. For stringent elimination of background noise, we defined a gene with cpm>10 in all replicates of a dataset as an expressed gene, and the collection of such genes as the transcriptome. Differential gene expression analysis was performed with the Bioconductor package DESeq2 (Love et al., 2014) version 1.26.0. Differentially expressed genes were identified using FDR<0.05 and log2 fold change >1 or <−1, unless otherwise noted. Only protein coding genes were used for both edgeR-based transcriptome analysis and DESeq2-based differential gene expression analysis. All data visualization steps were performed using custom scripts written in R.

eCLIP data analysis

The eCLIP read file processing, alignment and processing were done using an easy-to-use Python-based in-house analysis pipeline, eCLIP data analyzer (https://github.com/jkkbuddika/eCLIP-Data-Analyzer). Although the eCLIP data analyzer pipeline permits analysis in both paired and single-end modes, we performed all our analyses using the single-end mode, which uses only R2 read for analysis. The eCLIP data analyzer pipeline performs the following steps: (1) read quality assessment using FastQC, (2) trimming of universal eCLIP adaptors using Cutadapt (Martin, 2011), (3) addition of UMI sequence (5′-NNNNNNNNNN in the R2 read) using UMI-tools (Smith et al., 2017) to read name to facilitate PCR duplicate removal, (4) removal of reads aligning to rRNAs using Tagdust2 (Lassmann, 2015), (5) alignment of remaining reads to the provided genome (i.e. BDGP assembly release 6.28, Ensembl release 100) using STAR (Dobin et al., 2013), (6) coordination of sort and index alignment outputs using SAMtools (Li et al., 2009), (7) removal of PCR duplicates using UMI-tools (Smith et al., 2017), (8) generation of bigWig files representing RNA-seq coverage tracks using deepTools (Ramírez et al., 2016), (9) quantification of nearest overlapping features using Subread (Liao et al., 2019) function featureCounts and (10) peak calling using PureCLIP (Krakau et al., 2017). All programs, versions and dependencies required to execute the eCLIP data analyzer are described in the user guide and can be installed using miniconda. Input normalized peak calling output files were processed as described (Busch et al., 2020). Subsequently, significantly enriched de novo binding motifs were identified with DREME (options: -oc, -dna, -t 18000, -e 0.05) using shuffled input sequences as a control (Bailey, 2011). Protein-coding FMRP target genes were used to identify enriched Gene Ontology (GO) terms using gProfiler (Raudvere et al., 2019). Selected significantly enriched GO categories were plotted using R. Unless otherwise noted, all data visualization steps were performed using custom scripts written in R.

CLIP-qPCR

Intestines were dissected from 100 w1118 females, and FMRP was precipitated with anti-FMRP antibody (5A11, DSHB) using the same method described above. RNA was isolated from either the starting intestinal lysate or from the IP using Trizol LS (Ambion, 10296028). Resulting RNA was treated with Turbo DNase (ThermoFisher, AM2239) and ∼200 ng of input RNA and all the RNA derived from the IP sample were used for cDNA synthesis with Superscript III (ThermoFisher, 56575). Quantitative PCR was performed using the PowerUp SYBR Green Master Mix (ThermoFisher, A25742) on a StepOnePlus machine (ThermoFisher). Primers for all targets detected are listed in Table S4. Transcript levels were quantified in duplicate and normalized to CG10116. Fold enrichment was calculated as the ratio of transcript in IP versus input.

Statistical analysis

Statistical analyses were performed using Prism (GraphPad, Version 7.0). Datasets were tested for normality using the D'Agostino–Pearson test. Comparison of two datasets was performed with an unpaired t-test, whereas qPCR samples were analyzed with a paired t-test. The statistical significance of overlaps in Venn diagrams was calculated using a hypergeometric test on R. When multiple pairwise comparisons were needed, the R package GeneOverlap was used to perform Fischer's exact test, which yields the statistical significance, odds ratio and Jaccard indices for each pairwise comparison. Significance is indicated as follows: ns, not significant; *P<0.05; **P<0.01; ***P<0.001; ****P<0.0001.

Acknowledgements

We thank Steven Hou, Gerald Rubin, Steve Stowers, the Kyoto Drosophila Stock Center, the Bloomington Drosophila Stock Center (supported by grant NIHP4OOD018537), the Drosophila Genome Resource Center (supported by grant NIH2P40OD010949) and the Developmental Studies Hybridoma Bank for reagents; and the Light Microscopy Imaging Center (supported by grant NIH1S10OD024988-01) for access to the SP8 confocal.

Footnotes

Author contributions

Conceptualization: N.S.S.; Methodology: N.S.S., K.B., J.X.; Software: K.B.; Validation: I.S.A.; Formal analysis: K.B., I.S.A.; Investigation: K.B., J.X.; Writing - original draft: N.S.S., K.B.; Writing - review & editing: N.S.S., K.B.; Visualization: K.B.; Project administration: N.S.S.; Funding acquisition: N.S.S.

Funding

We thank the National Institute of General Medical Sciences (R01GM124220) for financial support. Deposited in PMC for release after 12 months.

Data availability

The PABP CLIP-seq and FMRP eCLIP-seq datasets from this study have been deposited in GEO under accession number GSE160128.

Peer review history

References

Andrä
,
I.
,
Ulrich
,
H.
,
Dürr
,
S.
,
Soll
,
D.
,
Henkel
,
L.
,
Angerpointner
,
C.
,
Ritter
,
J.
,
Przibilla
,
S.
,
Stadler
,
H.
,
Effenberger
,
M.
et al. 
(
2020
).
An evaluation of T-cell functionality after flow cytometry sorting revealed p38 MAPK activation
.
Cytometry A
97
,
171
-
183
.
Andriatsilavo
,
M.
,
Stefanutti
,
M.
,
Siudeja
,
K.
,
Perdigoto
,
C. N.
,
Boumard
,
B.
,
Gervais
,
L.
,
Gillet-Markowska
,
A.
,
Al Zouabi
,
L.
,
Schweisguth
,
F.
and
Bardin
,
A. J.
(
2018
).
Spen limits intestinal stem cell self-renewal
.
PLoS Genet.
14
,
e1007773
.
Antonello
,
Z. A.
,
Reiff
,
T.
,
Ballesta-Illan
,
E.
and
Dominguez
,
M.
(
2015
).
Robust intestinal homeostasis relies on cellular plasticity in enteroblasts mediated by miR-8-Escargot switch
.
EMBO J.
34
,
2025
-
2041
.
Bailey
,
T. L.
(
2011
).
DREME: motif discovery in transcription factor ChIP-seq data
.
Bioinformatics
27
,
1653
-
1659
.
Bardin
,
A. J.
,
Perdigoto
,
C. N.
,
Southall
,
T. D.
,
Brand
,
A. H.
and
Schweisguth
,
F.
(
2010
).
Transcriptional control of stem cell maintenance in the Drosophila intestine
.
Development
137
,
705
-
714
.
Barolo
,
S.
,
Carver
,
L. A.
and
Posakony
,
J. W.
(
2000
).
GFP and beta-galactosidase transformation vectors for promoter/enhancer analysis in Drosophila
.
BioTechniques
29
,
726
,
728, 730, 732
.
Baser
,
A.
,
Skabkin
,
M.
,
Kleber
,
S.
,
Dang
,
Y.
,
Gülcüler Balta
,
G. S.
,
Kalamakis
,
G.
,
Göpferich
,
M.
,
Ibañez
,
D. C.
,
Schefzik
,
R.
,
Lopez
,
A. S.
et al. 
(
2019
).
Onset of differentiation is post-transcriptionally controlled in adult neural stem cells
.
Nature
566
,
100
-
104
.
Biteau
,
B.
,
Karpac
,
J.
,
Supoyo
,
S.
,
Degennaro
,
M.
,
Lehmann
,
R.
and
Jasper
,
H.
(
2010
).
Lifespan extension by preserving proliferative homeostasis in Drosophila
.
PLoS Genet.
6
,
e1001159
.
Buddika
,
K.
,
Ariyapala
,
I. S.
,
Hazuga
,
M. A.
,
Riffert
,
D.
and
Sokol
,
N. S.
(
2020a
).
Canonical nucleators are dispensable for stress granule assembly in Drosophila intestinal progenitors
.
J. Cell Sci.
133
,
jcs243451
.
Buddika
,
K.
,
Huan
,
Y.-T.
,
Butrum-Griffith
,
A.
,
Norrell
,
S. A.
,
O'Connor
,
A. M.
,
Patel
,
V. K.
,
Rector
,
S. A.
,
Slovan
,
M.
,
Sokolowsky
,
M.
,
Kato
,
Y.
et al. 
(
2020b
).
Intestinal progenitor P-bodies maintain stem cell identity by suppressing pro-differentiation factors
.
BioRxiv
/2020/175398
.
Busch
,
A.
,
Brüggemann
,
M.
,
Ebersberger
,
S.
and
Zarnack
,
K.
(
2020
).
iCLIP data analysis: A complete pipeline from sequencing reads to RBP binding sites
.
Methods
178
,
49
-
62
.
Chen
,
X.
and
Dickman
,
D.
(
2017
).
Development of a tissue-specific ribosome profiling approach in Drosophila enables genome-wide evaluation of translational adaptations
.
PLoS Genet.
13
,
e1007117
.
Chen
,
C.-H.
,
Luhur
,
A.
and
Sokol
,
N.
(
2015
).
Lin-28 promotes symmetric stem cell division and drives adaptive growth in the adult Drosophila intestine
.
Development
142
,
3478
-
3487
.
Cvetkovska
,
V.
,
Hibbert
,
A. D.
,
Emran
,
F.
and
Chen
,
B. E.
(
2013
).
Overexpression of Down syndrome cell adhesion molecule impairs precise synaptic targeting
.
Nat. Neurosci.
16
,
677
-
682
.
Darnell
,
J. C.
,
Van Driesche
,
S. J.
,
Zhang
,
C.
,
Hung
,
K. Y. S.
,
Mele
,
A.
,
Fraser
,
C. E.
,
Stone
,
E. F.
,
Chen
,
C.
,
Fak
,
J. J.
,
Chi
,
S. W.
et al. 
(
2011
).
FMRP stalls ribosomal translocation on mRNAs linked to synaptic function and autism
.
Cell
146
,
247
-
261
.
Dobin
,
A.
,
Davis
,
C. A.
,
Schlesinger
,
F.
,
Drenkow
,
J.
,
Zaleski
,
C.
,
Jha
,
S.
,
Batut
,
P.
,
Chaisson
,
M.
and
Gingeras
,
T. R.
(
2013
).
STAR: ultrafast universal RNA-seq aligner
.
Bioinformatics
29
,
15
-
21
.
Dutta
,
D.
,
Dobson
,
A. J.
,
Houtz
,
P. L.
,
Glässer
,
C.
,
Revah
,
J.
,
Korzelius
,
J.
,
Patel
,
P. H.
,
Edgar
,
B. A.
and
Buchon
,
N.
(
2015
).
Regional cell-specific transcriptome mapping reveals regulatory complexity in the adult Drosophila Midgut
.
Cell Rep.
12
,
346
-
358
.
Fast
,
D.
,
Petkau
,
K.
,
Ferguson
,
M.
,
Shin
,
M.
,
Galenza
,
A.
,
Kostiuk
,
B.
,
Pukatzki
,
S.
and
Foley
,
E.
(
2020
).
Vibrio cholerae-symbiont interactions inhibit intestinal repair in Drosophila
.
Cell Rep.
30
,
1088
-
1100.e1085
.
Feng
,
H.
,
Bao
,
S.
,
Rahman
,
M. A.
,
Weyn-Vanhentenryck
,
S. M.
,
Khan
,
A.
,
Wong
,
J.
,
Shah
,
A.
,
Flynn
,
E. D.
,
Krainer
,
A. R.
and
Zhang
,
C.
(
2019
).
Modeling RNA-binding protein specificity in vivo by precisely registering protein-RNA crosslink sites
.
Mol. Cell
74
,
1189
-
1204.e1186
.
Furriols
,
M.
and
Bray
,
S.
(
2001
).
A model Notch response element detects Suppressor of Hairless-dependent molecular switch
.
Curr. Biol.
11
,
60
-
64
.
Greenblatt
,
E. J.
and
Spradling
,
A. C.
(
2018
).
Fragile X mental retardation 1 gene enhances the translation of large autism-related proteins
.
Science
361
,
709
-
712
.
Guo
,
Z.
,
Driver
,
I.
and
Ohlstein
,
B.
(
2013
).
Injury-induced BMP signaling negatively regulates Drosophila midgut homeostasis
.
J. Cell Biol.
201
,
945
-
961
.
Hida
,
N.
,
Aboukilila
,
M. Y.
,
Burow
,
D. A.
,
Paul
,
R.
,
Greenberg
,
M. M.
,
Fazio
,
M.
,
Beasley
,
S.
,
Spitale
,
R. C.
and
Cleary
,
M. D.
(
2017
).
EC-tagging allows cell type-specific RNA analysis
.
Nucleic Acids Res.
45
,
e138
.
Hung
,
R.-J.
,
Hu
,
Y.
,
Kirchner
,
R.
,
Liu
,
Y.
,
Xu
,
C.
,
Comjean
,
A.
,
Tattikota
,
S. G.
,
Li
,
F.
,
Song
,
W.
,
Ho Sui
,
S.
et al. 
(
2020
).
A cell atlas of the adult Drosophila midgut
.
Proc. Natl. Acad. Sci. USA
117
,
1514
-
1523
.
Hwang
,
H.-W.
,
Park
,
C. Y.
,
Goodarzi
,
H.
,
Fak
,
J. J.
,
Mele
,
A.
,
Moore
,
M. J.
,
Saito
,
Y.
and
Darnell
,
R. B.
(
2016
).
PAPERCLIP identifies microRNA targets and a role of CstF64/64tau in promoting non-canonical poly(A) site usage
.
Cell Rep.
15
,
423
-
435
.
Jiang
,
H.
and
Edgar
,
B. A.
(
2009
).
EGFR signaling regulates the proliferation of Drosophila adult midgut progenitors
.
Development
136
,
483
-
493
.
Kim
,
J. H.
,
Wang
,
X.
,
Coolon
,
R.
and
Ye
,
B.
(
2013
).
Dscam expression levels determine presynaptic arbor sizes in Drosophila sensory neurons
.
Neuron
78
,
827
-
838
.
Korzelius
,
J.
,
Naumann
,
S. K.
,
Loza-Coll
,
M. A.
,
Chan
,
J. S. K.
,
Dutta
,
D.
,
Oberheim
,
J.
,
Glässer
,
C.
,
Southall
,
T. D.
,
Brand
,
A. H.
,
Jones
,
D. L.
et al. 
(
2014
).
Escargot maintains stemness and suppresses differentiation in Drosophila intestinal stem cells
.
EMBO J.
33
,
2967
-
2982
.
Korzelius
,
J.
,
Azami
,
S.
,
Ronnen-Oron
,
T.
,
Koch
,
P.
,
Baldauf
,
M.
,
Meier
,
E.
,
Rodriguez-Fernandez
,
I. A.
,
Groth
,
M.
,
Sousa-Victor
,
P.
and
Jasper
,
H.
(
2019
).
The WT1-like transcription factor Klumpfuss maintains lineage commitment of enterocyte progenitors in the Drosophila intestine
.
Nat. Commun.
10
,
4123
.
Krakau
,
S.
,
Richard
,
H.
and
Marsico
,
A.
(
2017
).
PureCLIP: capturing target-specific protein-RNA interaction footprints from single-nucleotide CLIP-seq data
.
Genome Biol.
18
,
240
.
Lassmann
,
T.
(
2015
).
TagDust2: a generic method to extract reads from sequencing data
.
BMC Bioinformatics
16
,
24
.
Leader
,
D. P.
,
Krause
,
S. A.
,
Pandit
,
A.
,
Davies
,
S. A.
and
Dow
,
J. A. T.
(
2018
).
FlyAtlas 2: a new version of the Drosophila melanogaster expression atlas with RNA-Seq, miRNA-Seq and sex-specific data
.
Nucleic Acids Res.
46
,
D809
-
D815
.
Li
,
H.
and
Jasper
,
H.
(
2016
).
Gastrointestinal stem cells in health and disease: from flies to humans
.
Disease models & mechanisms
9
,
487
-
499
.
Li
,
H.
,
Handsaker
,
B.
,
Wysoker
,
A.
,
Fennell
,
T.
,
Ruan
,
J.
,
Homer
,
N.
,
Marth
,
G.
,
Abecasis
,
G.
,
Durbin
,
R.
, and
1000 Genome Project Data Processing Subgroup
. (
2009
).
The sequence alignment/Map format and SAMtools
.
Bioinformatics
25
,
2078
-
2079
.
Li
,
Q.
,
Nirala
,
N. K.
,
Nie
,
Y.
,
Chen
,
H.-J.
,
Ostroff
,
G.
,
Mao
,
J.
,
Wang
,
Q.
,
Xu
,
L.
and
Ip
,
Y. T.
(
2018
).
Ingestion of food particles regulates the mechanosensing misshapen-yorkie pathway in Drosophila intestinal growth
.
Dev. Cell
45
,
433
-
449.e436
.
Li
,
M.
,
Shin
,
J.
,
Risgaard
,
R. D.
,
Parries
,
M. J.
,
Wang
,
J.
,
Chasman
,
D.
,
Liu
,
S.
,
Roy
,
S.
,
Bhattacharyya
,
A.
and
Zhao
,
X.
(
2020
).
Identification of FMR1-regulated molecular networks in human neurodevelopment
.
Genome Res.
30
,
361
-
374
.
Liao
,
Y.
,
Smyth
,
G. K.
and
Shi
,
W.
(
2019
).
The R package Rsubread is easier, faster, cheaper and better for alignment and quantification of RNA sequencing reads
.
Nucleic Acids Res.
47
,
e47
.
Lin
,
G.
,
Xu
,
N.
and
Xi
,
R.
(
2008
).
Paracrine Wingless signalling controls self-renewal of Drosophila intestinal stem cells
.
Nature
455
,
1119
-
1123
.
Love
,
M. I.
,
Huber
,
W.
and
Anders
,
S.
(
2014
).
Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2
.
Genome Biol.
15
,
550
.
Luhur
,
A.
,
Chawla
,
G.
and
Sokol
,
N. S.
(
2013
).
MicroRNAs as components of systemic signaling pathways in Drosophila melanogaster
.
Curr. Top. Dev. Biol.
105
,
97
-
123
.
Luhur
,
A.
,
Buddika
,
K.
,
Ariyapala
,
I. S.
,
Chen
,
S.
and
Sokol
,
N. S.
(
2017
).
Opposing post-transcriptional control of InR by FMRP and LIN-28 adjusts stem cell-based tissue growth
.
Cell Rep.
21
,
2671
-
2677
.
Martin
,
M.
(
2011
).
Cutadapt removes adapter sequences from high-throughput sequencing reads
.
EMBnet J.
17
,
3
.
Maurin
,
T.
,
Lebrigand
,
K.
,
Castagnola
,
S.
,
Paquet
,
A.
,
Jarjat
,
M.
,
Popa
,
A.
,
Grossi
,
M.
,
Rage
,
F.
and
Bardoni
,
B.
(
2018
).
HITS-CLIP in various brain areas reveals new targets and new modalities of RNA binding by fragile X mental retardation protein
.
Nucleic Acids Res.
46
,
6344
-
6355
.
McClelland
,
L.
,
Jasper
,
H.
and
Biteau
,
B.
(
2017
).
Tis11 mediated mRNA decay promotes the reacquisition of Drosophila intestinal stem cell quiescence
.
Dev. Biol.
426
,
8
-
16
.
McGuire
,
S. E.
,
Mao
,
Z.
and
Davis
,
R. L.
(
2004
).
Spatiotemporal gene expression targeting with the TARGET and gene-switch systems in Drosophila
.
Science's STKE
2004
,
pl6
.
McMahon
,
A. C.
,
Rahman
,
R.
,
Jin
,
H.
,
Shen
,
J. L.
,
Fieldsend
,
A.
,
Luo
,
W.
and
Rosbash
,
M.
(
2016
).
TRIBE: hijacking an RNA-editing enzyme to identify cell-specific targets of RNA-binding proteins
.
Cell
165
,
742
-
753
.
Micchelli
,
C. A.
and
Perrimon
,
N.
(
2006
).
Evidence that stem cells reside in the adult Drosophila midgut epithelium
.
Nature
439
,
475
-
479
.
Miller
,
D. E.
,
Kahsai
,
L.
,
Buddika
,
K.
,
Dixon
,
M. J.
,
Kim
,
B. Y.
,
Calvi
,
B. R.
,
Sokol
,
N. S.
,
Hawley
,
R. S.
and
Cook
,
K. R.
(
2020
).
Identification and characterization of breakpoints and mutations on Drosophila melanogaster balancer chromosomes
.
G3
10
,
4271
-
4285
.
Monzo
,
K.
,
Dowd
,
S. R.
,
Minden
,
J. S.
and
Sisson
,
J. C.
(
2010
).
Proteomic analysis reveals CCT is a target of Fragile X mental retardation protein regulation in Drosophila
.
Dev. Biol.
340
,
408
-
418
.
Nern
,
A.
,
Pfeiffer
,
B. D.
,
Svoboda
,
K.
and
Rubin
,
G. M.
(
2011
).
Multiple new site-specific recombinases for use in manipulating animal genomes
.
Proc. Natl. Acad. Sci. USA
108
,
14198
-
14203
.
Phillips
,
M. D.
and
Thomas
,
G. H.
(
2006
).
Brush border spectrin is required for early endosome recycling in Drosophila
.
J. Cell Sci.
119
,
1361
-
1370
.
Ramírez
,
F.
,
Ryan
,
D. P.
,
Gruning
,
B.
,
Bhardwaj
,
V.
,
Kilpert
,
F.
,
Richter
,
A. S.
,
Heyne
,
S.
,
Dündar
,
F.
and
Manke
,
T.
(
2016
).
deepTools2: a next generation web server for deep-sequencing data analysis
.
Nucleic Acids Res.
44
,
W160
-
W165
.
Raudvere
,
U.
,
Kolberg
,
L.
,
Kuzmin
,
I.
,
Arak
,
T.
,
Adler
,
P.
,
Peterson
,
H.
and
Vilo
,
J.
(
2019
).
g:Profiler: a web server for functional enrichment analysis and conversions of gene lists (2019 update)
.
Nucleic Acids Res.
47
,
W191
-
W198
.
Reeve
,
S. P.
,
Bassetto
,
L.
,
Genova
,
G. K.
,
Kleyner
,
Y.
,
Leyssen
,
M.
,
Jackson
,
F. R.
and
Hassan
,
B. A.
(
2005
).
The Drosophila fragile X mental retardation protein controls actin dynamics by directly regulating profilin in the brain
.
Curr. Biol.
15
,
1156
-
1163
.
Richardson
,
G. M.
,
Lannigan
,
J.
and
Macara
,
I. G.
(
2015
).
Does FACS perturb gene expression?
Cytometry A
87
,
166
-
175
.
Robinson
,
M. D.
,
McCarthy
,
D. J.
and
Smyth
,
G. K.
(
2010
).
edgeR: a Bioconductor package for differential expression analysis of digital gene expression data
Bioinformatics
26
,
139
-
140
.
Roy
,
P. J.
,
Stuart
,
J. M.
,
Lund
,
J.
and
Kim
,
S. K.
(
2002
).
Chromosomal clustering of muscle-expressed genes in Caenorhabditis elegans
.
Nature
418
,
975
-
979
.
Sanz
,
E.
,
Yang
,
L.
,
Su
,
T.
,
Morris
,
D. R.
,
McKnight
,
G. S.
and
Amieux
,
P. S.
(
2009
).
Cell-type-specific isolation of ribosome-associated mRNA from complex tissues
.
Proc. Natl. Acad. Sci. USA
106
,
13939
-
13944
.
Sears
,
J. C.
,
Choi
,
W. J.
and
Broadie
,
K.
(
2019
).
Fragile X Mental Retardation Protein positively regulates PKA anchor Rugose and PKA activity to control actin assembly in learning/memory circuitry
.
Neurobiol. Dis.
127
,
53
-
64
.
Shearin
,
H. K.
,
Macdonald
,
I. S.
,
Spector
,
L. P.
and
Stowers
,
R. S.
(
2014
).
Hexameric GFP and mCherry reporters for the Drosophila GAL4, Q, and LexA transcription systems
.
Genetics
196
,
951
-
960
.
Singh
,
S. R.
,
Liu
,
W.
and
Hou
,
S. X.
(
2007
).
The adult Drosophila malpighian tubules are maintained by multipotent stem cells
.
Cell Stem Cell
1
,
191
-
203
.
Smith
,
T.
,
Heger
,
A.
and
Sudbery
,
I.
(
2017
).
UMI-tools: modeling sequencing errors in Unique Molecular Identifiers to improve quantification accuracy
.
Genome Res.
27
,
491
-
499
.
Sudhakaran
,
I. P.
,
Hillebrand
,
J.
,
Dervan
,
A.
,
Das
,
S.
,
Holohan
,
E. E.
,
Hulsmeier
,
J.
,
Sarov
,
M.
,
Parker
,
R.
,
VijayRaghavan
,
K.
and
Ramaswami
,
M.
(
2014
).
FMRP and Ataxin-2 function together in long-term olfactory habituation and neuronal translational control
.
Proc. Natl. Acad. Sci. U.S.A.
111
,
E99
-
E108
.
Tahmasebi
,
S.
,
Amiri
,
M.
and
Sonenberg
,
N.
(
2018
).
Translational control in stem cells
.
Front. Genet.
9
,
709
.
Tenenbaum
,
S. A.
,
Carson
,
C. C.
,
Lager
,
P. J.
and
Keene
,
J. D.
(
2000
).
Identifying mRNA subsets in messenger ribonucleoprotein complexes by using cDNA arrays
.
Proc. Natl. Acad. Sci. USA
97
,
14085
-
14090
.
Van Nostrand
,
E. L.
,
Pratt
,
G. A.
,
Shishkin
,
A. A.
,
Gelboin-Burkhart
,
C.
,
Fang
,
M. Y.
,
Sundararaman
,
B.
,
Blue
,
S. M.
,
Nguyen
,
T. B.
,
Surka
,
C.
,
Elkins
,
K.
et al. 
(
2016
).
Robust transcriptome-wide discovery of RNA-binding protein binding sites with enhanced CLIP (eCLIP)
.
Nat. Methods
13
,
508
-
514
.
Voght
,
S. P.
,
Fluegel
,
M. L.
,
Andrews
,
L. A.
and
Pallanck
,
L. J.
(
2007
).
Drosophila NPC1b promotes an early step in sterol absorption from the midgut epithelium
.
Cell Metab.
5
,
195
-
205
.
Wang
,
X.
,
Mu
,
Y.
,
Sun
,
M.
and
Han
,
J.
(
2017
).
Bidirectional regulation of fragile X mental retardation protein phosphorylation controls rhodopsin homoeostasis
.
J Mol Cell Biol
9
,
104
-
116
.
Yang
,
Z.
,
Edenberg
,
H. J.
and
Davis
,
R. L.
(
2005
).
Isolation of mRNA from specific tissues of Drosophila by mRNA tagging
.
Nucleic Acids Res.
33
,
e148
.
Zeng
,
X.
,
Chauhan
,
C.
and
Hou
,
S. X.
(
2010
).
Characterization of midgut stem cell- and enteroblast-specific Gal4 lines in drosophila
.
Genesis
48
,
607
-
611
.
Zhang
,
Y. Q.
,
Bailey
,
A. M.
,
Matthies
,
H. J. G.
,
Renden
,
R. B.
,
Smith
,
M. A.
,
Speese
,
S. D.
,
Rubin
,
G. M.
and
Broadie
,
K.
(
2001
).
Drosophila fragile X-related gene regulates the MAP1B homolog Futsch to control synaptic structure and function
.
Cell
107
,
591
-
603
.

Competing interests

The authors declare no competing or financial interests.

Supplementary information