Pumilio differentially binds to mRNA 3' UTR isoforms to regulate localization of synaptic proteins

Abstract

In neuronal cells, the regulation of RNA is crucial for the spatiotemporal control of gene expression, but how the correct localization, levels, and function of synaptic proteins are achieved is not well understood. In this study, we globally investigate the role of alternative 3' UTRs in regulating RNA localization in the synaptic regions of the Drosophila brain. We identify direct mRNA targets of the translational repressor Pumilio, finding that mRNAs bound by Pumilio encode proteins enriched in synaptosomes. Pumilio differentially binds to RNA isoforms of the same gene, favoring long, neuronal 3' UTRs. These longer 3' UTRs tend to remain in the neuronal soma, whereas shorter UTR isoforms localize to the synapse. In cultured pumilio mutant neurons, axon outgrowth defects are accompanied by mRNA isoform mislocalization, and proteins encoded by these Pumilio target mRNAs display excessive abundance at synaptic boutons. Our study identifies an important mechanism for the spatiotemporal regulation of protein function in neurons.

Keywords: Neuronal 3′ UTR; Pumilio; Synaptic Proteins; Synaptosome; mRNA Localization.

Figure 1. Pumilio directly binds mRNAs encoding synaptic proteins.

(A) xRIP-3′-seq experimental and data analysis workflow. Flag-HA-tagged Pum protein was immuno-purified together with UV-cross-linked target mRNAs from adult fly heads, using beads conjugated with an anti-Flag antibody. RNA was subjected to 3′-end sequencing. To determine the exact 3′-end position, sequencing reads were mapped and unreliable clusters were filtered out (see methods). Pum target mRNAs were identified by enrichment of 3′-seq signal in the xRIP sample over input. (B) Visualization of 3′-seq signal for an exemplary Pum target mRNA (GluRIA) and the neighboring two non-target genes. Control xRIP represents samples from flies not expressing Flag-tagged Pum (w¹¹¹⁸). (C) Differential RNA expression in the Pum xRIP sample compared to input (head homogenate), represented as a function of mean expression levels. Dark blue represents: |log₂ fold change (Pum xRIP/input)| >1 with p-value < 0.05 and base mean >10, and |log₂ fold change (Pum xRIP/control xRIP)| >1. (D) Gene ontology analysis of 460 Pum target genes. The top five terms are shown. See Dataset EV1 for all significant terms (p < 0.05; one-sided EASE score adjusted using the Benjamini-Hochberg method). (E) Overlap between mouse (Zhang et al, 2017) (Zhang et al, 2017) and Drosophila (this study) Pum target genes. 118 genes represent Drosophila homologs of mouse Pum targets that also figure among the 460 Drosophila Pum targets. (F) Gene ontology analysis of 118 Pum target genes shared between fly and mouse. Significant terms related to nervous system function are shown. See Dataset EV2 for all significant terms (p < 0.05; one-sided EASE score adjusted using the Benjamini-Hochberg method).

Figure 2. Pumilio target mRNAs encode proteins enriched in synaptosome fractions.

(A) Workflow of Drosophila synaptosome isolation. Fresh head tissue from adult Drosophila was dissociated and shear force was used to separate pre- and postsynaptic membranes from nuclei, cytoplasm, and cellular debris. Synaptosome fractions (F3 and F4, red) were enriched by centrifugation on a percoll gradient. (B) Electron micrograph of the synaptosome fraction. The magnification shows an intact synaptosome (indicated by the red border) with its typical components. (C) Western blot showing the expression of the indicated protein markers in the individual fractions. Mb: membranes. Cysteine string protein (CSP), Synapsin (Syn), Syntaxin 1A (Syx1A), and Discs Large (Dlg) represent synaptic markers (in red); ELAV and Lamin C are markers of the nucleosol and nuclear lamina, respectively. (D) Gene ontology analysis of 989 proteins enriched in the synaptosome fraction. The top ten terms are shown. See Dataset EV3 for all significant terms (p < 0.05; one-sided EASE score adjusted using the Benjamini-Hochberg method). (E) Proportion of mRNAs that encode synaptic proteins in each category; “other RNAs” refers to mRNAs that are not bound by Pum and that encode proteins detected by proteomics. ***p < 0.001 (p = 4e−23, two-tailed Fisher’s exact test). Only RNAs that encode proteins detected by proteomics in the synaptosome isolation experiment were considered (see Fig. EV2B). (F) Gene ontology analysis of 116 Pum target mRNAs that encode a protein found enriched in the synaptosome fraction. The top eight terms are shown. See Dataset EV3 for all significant terms (p < 0.05; one-sided EASE score adjusted using the Benjamini-Hochberg method). (G) Proportion of Pum target mRNAs in each category of mRNA subcellular localization, represented as enrichment over non-localized genes. *p < 0.1, **p < 0.05 (two-tailed Fisher’s exact test). Source data are available online for this figure.

Figure 3. Synaptic localization of short 3′ UTR isoforms of Pumilio target mRNAs.

(A) Workflow of differential 3′-end isoform analysis. After mapping and clustering, 3′-seq read clusters were filtered against 3′-ends validated by long-read direct-RNA sequencing (see methods). For each gene, 3′-seq signal was scored for individual clusters in synaptosome fractions and compared to input to identify synaptosome-enriched 3′ UTR isoforms of each gene. (B) Differential 3′ UTR lengths in synaptosomes compared to input. For each gene, the length of each expressed 3′ UTR isoform was represented as a % of the length of the longest 3′ UTR, weighed by expression of each isoform, and combined into one % 3′ UTR length value for each gene. Genes with significant 3′ UTR length differences between synaptosome and input fractions are depicted in orange (longer 3′ UTR in synaptosomes) and blue (shorter 3′ UTR in synaptosomes) (p < 0.05, 1-tailed Z-score). Pum target genes are marked in the darker shade of color. (C) Representative examples (3′-seq tracks and 3′-end isoform models) of genes with differentially localized 3′ UTR isoforms. The upper two traces in each panel show enrichment of the shorter Beadex (Bx) 3′ UTR isoform (blue), of the longer Octopamine receptor in mushroom bodies (Oamb) 3′ UTR isoform (orange), or no change for Semaphorin 2b (Sema2b) in synaptosomes compared to input. The lower three traces show Pum preferential binding to the long (Bx, OamB) or short (Sema2b) 3′ UTR isoforms, respectively. (D) Proportion of Pum target genes in each category of 3′ UTR isoform subcellular localization, represented as enrichment over all expressed genes. ns, non-significant, ***p < 0.001 (two-tailed Fisher’s exact test). (E) Enrichment of long vs. short 3′ UTR isoform in synaptosomes compared to input. Shown are Pum target mRNAs from Fig. 3B that display differentially localized 3′ UTR isoforms. 10 and 43 genes display longer 3′ UTR isoforms in synaptosomes (orange) and input (blue), respectively. (F) Pum xRIP-3′-seq signal in long vs. short 3′ UTR isoforms, for the 43 genes in Fig. 3E that display longer 3′ UTR isoforms in the input sample compared to synaptosome sample. ns, non-significant, *p < 0.05; p(Pum Input vs. Pum xRIP) = 0.042; p(control Input vs. control xRIP) = 0.632 (two-tailed Student’s t-test). Boxes indicate range between minimum and maximum, the central line depicts the median, lower and upper bounds represent the first and third quartiles, respectively. Control xRIP represents samples from flies not expressing Flag-tagged Pum (w¹¹¹⁸). Biological replicates n = 3.

Figure 4. Pumilio binds to soma-localized neuronal 3′ UTRs of synaptic genes.

(A) Pum binding motif found enriched in neuronal 3′ UTRs. (B) Proportion of nUTR-containing genes in the indicated gene categories, represented as enrichment over all expressed genes. ***p < 0.001 (two-tailed Fisher’s exact test). (C) Gene ontology analysis of 71 Pum target genes that possess a neuronal 3′ UTR. The top three terms are shown but only one is significant. **p < 0.01 (one-sided EASE score adjusted using the Benjamini-Hochberg method). ns, non-significant. See Dataset EV4 for all terms. (D) Number and proportion of nUTR-containing genes among mRNAs with a shorter 3′ UTR in synaptosomes, in each gene category. *p < 0.01 (p = 0.0012, two-tailed Fisher’s exact test). Only mRNAs expressed in both the synaptosome and input samples were considered.

Figure 5. Impaired neurite outgrowth, mRNA delocalization, and synaptic protein overexpression in neurons of Δpum flies.

(A) Schematic of primary neuronal cell culture procedure. Nervous systems are dissected from third-instar Drosophila larvae and enzymatically and mechanically dissociated; the cells in the resulting suspension are allowed to divide and differentiate either on a coverslip for imaging, or on a microporous membrane for separation of neurite and soma compartments. DIV, days in vitro. (B) Confocal imaging of neurite phenotypes in cultured neurons of control (w¹¹¹⁸) and Δpum (pum^ET7/pum^ET9) flies, at the indicated days after plating. ELAV and HRP antibodies mark neuronal nuclei and neuronal cell membranes, respectively. Scale bars: 50 μm. (C) Quantification of neurite length in cultured neurons of control (w¹¹¹⁸) and Δpum (pum^ET7/pum^ET9) flies at the indicated days after plating. Neurites with a length >70 µm are represented as triangles at the top of the plot; neurites with a length <10 μm were excluded from the analysis. Error bars represent the mean ± SD of at least 119 neurites for each genotype and time point. **p < 0.01; p(5DIV control vs. 5DIV Δpum) = 1.071e−3, p(7DIV control vs. 7DIV Δpum) = 1.543e−3 (two-tailed Student’s t-test). ns, not significant. Total number of neurites quantified n = 1365. (D) RT-qPCR quantification of the expression of nUTR-containing (long) relative to total (short) mRNA isoforms for the indicated genes, in isolated cell bodies of control (w¹¹¹⁸) and Δpum (pum^ET7/pum^ET9) flies, at seven days in vitro. Error bars represent the mean ± SD of three biological replicates for each genotype. ***p < 0.001, **p < 0.01, *p < 0.05; p(gro control vs. gro Δpum) = 2.166e−4, p(Oamb control vs. Oamb Δpum) = 1.299e−2, p(nej control vs. nej Δpum) = 1.611e−3, p(rut control vs. rut Δpum) = 1.975e−2, p(Khc-73 control vs. Khc-73 Δpum) = 7.861e−3, p(CG31688 control vs. CG31688 Δpum) = 1.603e−2 (two-tailed Student’s t-test). (E) Confocal imaging (left) and quantification (right) of Syt1 and Brp expression in neuromuscular junctions of control (w¹¹¹⁸) and Δpum (pum^ET7/pum^ET9) third-instar larvae. HRP marks presynaptic membranes. Scale bars: 50 μm. Number of synapses scored: n = 156 (Syt1) and n = 62 (Brp) in control, n = 178 (Syt1) and n = 88 (Brp) in Δpum. Synapses with a diameter <4 µm were excluded from the analysis. p-values (two-tailed Student’s t-test) are indicated. Boxes indicate range between minimum and maximum, the central line depicts the median, lower and upper bounds represent the first and third quartiles, respectively. Source data are available online for this figure.

Figure 6. Model of 3′ UTR-dependent localization of Pum target mRNAs.

In neuronal cell bodies, Pum binds to mRNAs encoding synaptic proteins, with a preference for long 3′ UTR isoforms. This leads to the enriched localization of translationally competent short isoforms in synaptic compartments.

Figure EV1. xRIP-3′-seq identifies mRNAs directly bound by Drosophila Pumilio.

(A–C) Western blots showing Pum expression in adult fly heads. (A) Detection with an anti-Pum antibody shows the molecular weight difference between wild-type Pum (untagged) compared to Flag-HA-tagged Pum protein in three independent pum^Flag transformant flies. ELAV serves as loading control. (B) Detection with an anti-Flag antibody shows specific detection of Pum-Flag in pum^Flag flies. Histone H3 serves as a loading control. Five fly heads were used for protein preparation for each genotype. (C) Eluates from an anti-Flag antibody immunopurification of Pum from head extract of pum^Flag and control (w¹¹¹⁸) flies. (D) The Pumilio Response Element (PRE) constitutes the most enriched motif in the 3′ UTRs of Pum target mRNAs (p = 1.14*10⁻⁴³; E-value). Motifs for the top five RBPs are shown. (E) Venn diagram showing the intersection between Pum mRNA targets in Drosophila (this study) and for mammalian Pum 1 and Pum 2 identified in Zhang et al, (Zhang et al, 2017).

Figure EV2. Pum target mRNAs encode proteins enriched in synaptosome fractions.

(A) RT-qPCR quantification of the indicated transcript regions in synaptosome fractions relative to input fraction. For each gene, intron levels were normalized to coding exon mRNA levels, which were set to the value 1. Ratios represent the average of two biological replicates. (B) Proportions of proteins encoded by Pum target mRNAs (left) and all RNAs identified in Pum xRIP-3′-seq experiment (right) that were detected by proteomics in the synaptosome isolation experiment. (C) Average length of CDS, 5′ UTR and 3′ UTR, of mRNAs encoding proteins of the indicated categories. Number of genes analyzed: n = 4843 (Proteomics background), n = 989 (enriched in synaptosomes) and n = 116 (enriched in synaptosomes and encoded by a Pum target gene). Boxes indicate range between minimum and maximum, the central line depicts the median, lower and upper bounds represent the first and third quartiles, respectively. (D) Differential mRNA expression (by 3′-seq cluster expression) in synaptosome fractions compared to input. The p-value of the enrichment is represented as a function of log₂ fold change. Dark blue represents |log₂ fold change (synaptosome/input)| >0 and p-value < 0.05 (Wald test). (E) Gene ontology analysis of 1195 mRNAs depleted in synaptosome fractions compared to input. The top ten terms are shown (p < 0.01; one-sided EASE score adjusted using the Benjamini-Hochberg method). See Dataset EV3 for all significant terms. (F) Venn diagram showing the intersection between Pum target mRNAs and mRNAs enriched (left) or depleted (right) in the synaptosome fraction. (G) Number and proportion of Pum target mRNAs in each category of mRNA subcellular localization. *p = 0.1, **p < 0.05 (p = 4e−9) (two-tailed Fisher’s exact test). Only Pum target mRNAs expressed in both synaptosome and input (306 genes) were considered.

Figure EV3. Synaptic localization of short 3′ UTR isoforms of Pumilio target mRNAs.

(A) Number and proportion of Pum target genes in each category of 3′ UTR isoform subcellular localization. ns, non-significant, ***p < 0.001 (p = 4e−10, two-tailed Fisher’s exact test). Only Pum target mRNAs expressed in both synaptosome and input (306 genes) were considered. (B) Heatmaps showing Pum binding and the number of Pum binding motifs in distal 3′ UTR regions of Pum target mRNAs that display longer 3′ UTR isoforms in synaptosomes (10 genes), ranked by Pum xRIP-3′-seq signal compared to input.

Figure EV4. Pumilio binds to soma-localized long neuronal 3′ UTRs of synaptic genes.

(A) Gene ontology analysis of 271 genes that possess a neuronal 3′ UTR. All significant terms are shown (p < 0.05, one-sided EASE score adjusted using the Benjamini-Hochberg method). See also Dataset EV4. (B) Number and proportion of nUTR-containing genes in each gene category. ***p < 0.001 (p = 2e−44, two-tailed Fisher’s exact test). Only mRNAs expressed in both the synaptosome and input samples (199 nUTR-containing genes) were considered. (C) Number and proportion of nUTR-containing genes in each category of 3′ UTR isoform subcellular localization. ***p < 0.001 (p = 2e−10, two-tailed Fisher’s exact test). Only mRNAs expressed in both the synaptosome and input samples (199 nUTR-containing genes) were considered. (D) Heatmaps showing Pum binding and the number of Pum binding motifs in the nUTR of Pum target mRNAs in genes whose nUTR-containing 3′ UTR isoform is depleted in synaptosome fractions (14 genes), ranked by Pum xRIP-3′-seq signal compared to input.

Figure EV5. Impaired neurite outgrowth, mRNA delocalization, and synaptic protein overexpression in neurons of Δpum flies.

(A) Light micrographs of Drosophila primary neuronal cells cultured on a microporous membrane for separation of neurite and soma compartments. Before removal, neurites are visible as dark protrusions (white arrowheads) below the cell bodies. After soma removal, only cell debris and micropores are visible on the membrane. The three images represent three distinct cultures at 7 days in vitro. Scale bars: 200 μm. (B) Assessment of soma/neurite separation by RT-qPCR quantification of two RNAs well-known to localize to distinct neuronal compartments (neurite/synapse and soma for Arc1 and mimi, respectively). Shown is the RNA expression in neurites relative to cell bodies, in 7 days in vitro cultured neurons of control (genotype: w¹¹¹⁸) and Δpum (genotype: pum^ET7/pum^ET9) flies. For each gene, RNA levels were normalized pairwise, first to soma, and second to Arc1. Error bars represent the mean ± SD of three biological replicates for each genotype. (C) Quantification of the proportion of neurons in primary cultures from control (w¹¹¹⁸) and Δpum (pum^ET7/pum^ET9) dissected larval brains at the indicated days in vitro. The number of cells with ELAV staining (a marker of neuronal nuclei) was counted relative to the total number of cells with DAPI staining. Error bars represent the mean ± SD of at least 175 DAPI stained cells for each genotype and time point. *p < 0.05, p(3DIV control vs. 3DIV Δpum) = 0.03, p(7DIV control vs. 7DIV Δpum) = 0.04 (one-tailed Student’s t-test). ns, not significant. Total number of cells quantified n = 2100. (D) Number of neurites assessed in each genotype and time point for the quantifications shown in Figs. 5C and EV5E. Total number of neurites quantified n = 1365. (E) Quantification of average number of branch points per neurite in cultured neurons of control (w¹¹¹⁸) and Δpum (pum^ET7/pum^ET9) flies at the indicated days after plating. Error bars represent the mean ± SD of at least 119 neurites for each genotype and time point. **p < 0.01; p(5DIV control vs. 5DIV Δpum) = 0.003, p(7DIV control vs. 7DIV Δpum) = 0.008 (two-tailed Student’s t-test). ns, not significant. Total number of neurites quantified n = 1365. (F) Confocal imaging of C-terminally Flag-HA tagged flies (pum^Flag) flies at 7 days in vitro. HRP marks neuronal membranes. In merged images: HRP (magenta), HA (green) and DAPI (white). Yellow rectangles demarcate the region shown magnified in the lower panels. Scale bars: 50 µm (upper panel), 10 µm (lower panel). (G) Total RNA-seq quantification of long (dark gray) and short (light gray) 3′ UTR isoforms in seven days in vitro cultured neurons of control (w¹¹¹⁸) flies. 40 genes displaying shorter 3′ UTR isoforms in synaptosomes (longer 3′ UTR isoforms in input) from Fig. 3E are shown (3 genes not detected). Error bars represent the mean ± SD of seven biological replicates. (H) RT-qPCR quantification of the indicated transcripts in neurites relative to cell bodies, in 7 days in vitro cultured neurons of control (w¹¹¹⁸) and Δpum (pum^ET7/pum^ET9) flies. For each gene, RNA levels were normalized to those in control flies (in which log₂ fold change = 0). Error bars represent the mean ± SD of three biological replicates for each genotype. p-values are indicated; **p < 0.01, *p < 0.05; p(gro control vs. gro Δpum) = 0.02, p(ksr control vs. ksr Δpum) = 0.004, p(Bx control vs. Bx Δpum) = 0.012, p(Khc-73 control vs. Khc-73 Δpum) = 0.045, p(krz control vs. krz Δpum) = 0.002, p(CG31688 control vs. CG31688 Δpum) = 0.02, p(gug control vs. gug Δpum) = 0.004, p(gro control vs. gro Δpum) = 2.166e−4, p(gol control vs. gol Δpum) = 0.047, p(gro control vs. gro Δpum) = 2.166e−4, p(Sema1a control vs. Sema1a Δpum) = 0.025, p(orb2 control vs. orb2 Δpum) = 0.007, p(Arc1 control vs. Arc1 Δpum) =0.015 (two-tailed Student’s t-test).