Cytoplasmic cleavage of IMPA1 3' UTR is necessary for maintaining axon integrity

Abstract

The 3' untranslated regions (3' UTRs) of messenger RNAs (mRNAs) are non-coding sequences involved in many aspects of mRNA metabolism, including intracellular localization and translation. Incorrect processing and delivery of mRNA cause severe developmental defects and have been implicated in many neurological disorders. Here, we use deep sequencing to show that in sympathetic neuron axons, the 3' UTRs of many transcripts undergo cleavage, generating isoforms that express the coding sequence with a short 3' UTR and stable 3' UTR-derived fragments of unknown function. Cleavage of the long 3' UTR of Inositol Monophosphatase 1 (IMPA1) mediated by a protein complex containing the endonuclease argonaute 2 (Ago2) generates a translatable isoform that is necessary for maintaining the integrity of sympathetic neuron axons. Thus, our study provides a mechanism of mRNA metabolism that simultaneously regulates local protein synthesis and generates an additional class of 3' UTR-derived RNAs.

Keywords: 3’UTR; 3’UTR cleavage; NGF; RNA processing; alternative polyadenylation; axons; local translation; mRNA localization; neuronal development; sympathetic neurons.

Graphical abstract

Figure 1

3′ End RNA-seq on RNA isolated from axons and cell bodies (A) Accumulation of reads at the 3′ end of the transcripts. The read density (number of reads/nt divided by the total number of reads) of 4,975 transcripts between 2,000 and 3,000 nt long is shown. Dashed line indicates the 3′ end. (B) Identification of novel 3′ ends in the longest 3′ UTR of Elovl5. Raw coverage was smoothed using a running median (window width of 100 nt), and potential 3′ ends were identified by segmenting sudden transitions in read depth. (C) Maximum 3′ UTR lengths for existing annotations in Ensembl Rn5 and for those newly identified by 3′ end RNA sequencing (RNA-seq) in this study (^∗∗∗p = 1.010881e−06; two-sided Wilcoxon rank-sum test). (D) Left: single-molecule FISH (smFISH) of Nefl long 3′ UTR in sympathetic neurons cell bodies and axons. Arrowheads indicate mRNA puncta without any pixel dilation. Insets show 5× magnification of boxed area. Scale bar, 10 μm. Right: genome browser view of Nefl in axons and cell bodies. (E) Percentage of cell body and axonal transcript IDs showing multiple 3′ UTRs (left) and distribution of 3′ UTR isoforms per expressed Ensembl transcript ID (right) (^∗∗∗p = 3.116589e−72; two-sided Fisher’s exact count test). (F) Scatterplot of the relative usage of promoter-proximal and promoter-distal poly(A) sites in cell bodies and axons (FDR < 0.01 between cell body and axonal compartment; Fisher’s exact test). Distal shifts in axons compared with cell body (dark blue); proximal shifts in axons compared with cell body (light blue). (Inset) 3′ UTR isoforms with proximal or distal shift uniquely detected in axons (see STAR methods for details). (G) Statistically enriched GO terms for transcripts showing a proximal (top) or distal (bottom) shift in poly(A) site usage in axons. (H) Genome browser view of representative transcripts with a marked shift toward decreased (Atp5f1) or increased (Vps36) promoter-proximal poly(A) site usage in axons compared with cell bodies. See also Figure S1.

Figure 2

Remodeling of IMPA1 3′ UTR in axons (A) Left: genome browser view of IMPA1 transcripts in axons and cell bodies transcriptomes by 3′ end RNA-seq, Ensembl annotation, and 3′ RACE (gray, black, and red arrowheads, respectively). Dashed lines indicate 3′ end of the isoforms. Right: percentage of RACE clones (top) containing different IMPA1 3′ UTR isoforms in axons and cell bodies (cell bodies n = 15, axons n = 12), and schematics of the three IMPA1 3′ UTRs (bottom). Shadowed circle indicates axonal localization signal in IMPA1-L 3′ UTR. (B) Northern blot analysis of RNA isolated from naive (-) or NGF-differentiated PC12 cells and sympathetic neurons using probes annealing with IMPA1 CDS (left) or IMPA-L 3′ UTR (right). Note that the resolution of agarose gels and the abundance of the IMPA1-S isoform prevent the discrimination of IMPA1-S and IMPA1-L when using IMPA1 CDS probe. Asterisk (^∗) indicates a further isoform with a very short 3′ UTR detected by IMPA1 CDS probe. Representative images from 3 independent experiments. (C) Quantitative analysis of GFP protein immunofluorescence in axons of sympathetic neurons expressing either myrdEGFP-IMPA1-L, myrdEGFP-IMPA1-C, or myrdEGFP-Histone H3 (HH3). (Inset) Distribution of GFP-positive axons at the indicated length intervals (^∗p = 0.0108; chi-square = 6.494, df = 1). See also Figure S2.

Figure 3

The 3′ UTRs of many transcripts are remodeled in axons (A) Top: genome browser view of IMPA1, Sms, and Maoa transcripts in sympathetic neuron axons and cell bodies transcriptomes by 3′ end RNA-seq and Ensembl annotation (gray and black arrowhead, respectively). Dashed lines indicate 3′ end of the short isoforms. Bottom: number of clones of cleaved IMPA1-L, Sms, and Maoa 3′ UTR fragments purified from axonal RNA and grouped according to distance from the predicted cleavage site (red line). Each bin represents 50 nt. (B) Absence of cleaved fragments in Maf1, Cops3, and Fdxr transcripts (top, left lanes). The presence of corresponding cDNAs was assessed by regular RT-PCR (top, right lanes). Gray vertical line indicates samples ran on separated gel. Asterisk (^∗) represents primer dimers bands. Primers to amplify Fdxr cDNA are intra-exonic therefore they amplify also genomic DNA (gDNA). Noncontiguous lanes from the same experiment are shown side by side, as indicated by the gray line. Bottom: genome browser view of Maf1, Cops3, and Fdxr transcripts in axons and cell bodies transcriptomes by 3′ end RNA-seq and Ensembl annotation (gray and black arrowhead, respectively). See also Figure S3.

Figure 4

A complex containing Ago2, HuD, and Upf1 interacts with IMPA1 transcript (A) Partial list of Upf1 interactors identified in PC12 cells by mass spectrometry. n.d., not detected. (B) Western blots of Upf1, Ago2, and PI3K subunit p85 (as loading control) on axons and cell bodies of sympathetic neurons (n = 3). (C) Co-immunoprecipitation of Upf1 (left) or HuD (right) with the indicated proteins in sympathetic neurons (n = 3). (D) RNA immunoprecipitation (RIP) of IMPA1-L, Rpl10a (as negative control), or Maf1 mRNA, with HuD (left) or Ago2 antibody (right), and normal IgG antibody in sympathetic neuron lysates. Data are means ± SEM of ΔΔCt values between antibody or immunoglobulin G (IgG) samples and respective inputs expressed as fold of inputs (^∗p = 0.0465, paired one-tailed t test, t = 2.434, df = 3, n = 4; ^∗∗∗∗p < 0.0001 two-way ANOVA, Sidak’s multiple comparison test t = 22.11, df = 4, n = 3). n.s., not statistically significant. (E) Immuno-FISH of IMPA1-L transcript with Ago2 (top) and HuD (bottom) proteins. mRNA puncta were not subject to pixel dilation. Insets- show 5X Mmagnification of boxed area. Scale bar, 10 μm. (F) RIP of IMPA1-L transcript with Upf1 or HuD antibodies. Experiments were performed on PC12 cells either transfected with control (siRNA), HuD, or Upf1 siRNA or on wild-type (WT) or CRISPR-deleted Ago2 (Ago2^−/−) PC12 cells. (*p = 0.036, unpaired two-tail t test, t = 3.179, df = 4; only statistically significant comparisons are indicated, n as indicated.) See also Figure S4 and Tables S1–S3.

Figure 5

Ago2 mediates the cleavage of IMPA1 3′ UTR (A) RNA oligonucleotide (oligo)-mediated ligation (RML) of IMPA1-L transcript followed by qRT-PCR of WT or Ago2^−/− PC12 cells. Results are presented as fold over WT samples. Unpaired two-tailed t test, t = 9.912, df = 8 (n = 5). (B) Radioactive in vitro cleavage assay of 5′ end-labeled IMPA1-L RNA oligos using cytoplasmic lysates of sympathetic neurons and human recombinant Ago2 (left) or mouse recombinant WT or catalytic mutant (Ago2^D597A) Ago2. Control Ago2 is a commercially available human WT Ago2 (right). Arrows indicate cleaved fragment, and asterisk indicates a smaller fragment, probably due to trimming of the main fragment. Irrelevant lanes have been removed (n = 3). (C) Radioactive in vitro cleavage assay of 5′ end-labeled WT or mutant IMPA1-L RNA oligos. Left: a band corresponding to the expected size of the cleaved fragment (67 nt, arrows) is detected in WT oligos while it is absent in mutants. Noncontiguous lanes from the same experiment and autoradiography blot are shown side by side, while irrelevant lanes have been removed, as indicated by the gray lines (n = 4). Right: folding predictions of WT and mutant IMPA1-L oligos. Color-coded probability of pairing is shown as heatmap. ΔG values are indicated. Shadowed area points to the effect of the mutation on the secondary structure of the oligo. (Bottom) Sequence of the WT oligo used for folding prediction. Arrowhead indicates the point of cleavage. Boxed sequence is deleted in the Δcleavage site mutant. Nucleotides in red bold are mutated in the mutant loop oligo. Asterisk points to the truncation of the Δstem mutant. (D) HA-ms IMPA1 RIP in PC12 cells expressing either WT HA-ms IMPA1-rat L 3′ UTR or a mutant bearing a deletion of the cleavage site (Δcleavage-3′ UTR). Data were normalized by c-myc mRNA as a positive control. Unpaired two-tailed t test (n = 4). (E) IMPA1-L cleavage was assayed by RML qRT-PCR on RNA purified from PC12 cells transfected with the indicated siRNAs. Two-way ANOVA, Dunnett’s multiple comparison test, df = 16. (F) Number of clones bearing IMPA1-C, IMPA1-S, or IMPA1-L 3′ UTR in WT and Ago2^−/− PC12 cells. Thirty-five and 41 randomly selected clones obtained by 3′ RACE from WT and Ago2^−/− cells were sequenced. p = 0.0255, chi-square test, df = 2. All data in this figure are presented as mean ± SEM. Tests are indicated in the legend and p values in the figure. See also Figure S5.

Figure 6

Mutation of IMPA1 proximal PAS sites decreases translation (A) Luciferase assay of PC12 cells transfected with the indicated firefly luciferase vectors and renilla luciferase, as indicated. One-way ANOVA, Tukey’s post hoc test, at least 4 independent experiments, F = 47.66, df = 58. (B) Left: firefly luciferase northern blotting of PC12 cells transfected with either Firefly-IMPA1-L or Firefly- IMPA1-L ΔPAS expression vectors. Irrelevant lanes have been removed. Right: quantitative analysis of Firefly luciferase levels. t test Holm-Sidak method, df = 4 (n = 3). (C) Lysates of PC12 cells transfected with the indicated Firefly-IMPA1 3′ UTR vectors were separated by polysomal fractionation, RNA was isolated from each fraction and subjected to northern analysis using ³²P-labeled probes to detect Firefly (top) or GAPDH (bottom) transcripts. Bands were captured using a phosphorimager and quantified using ImageQuant v.5.2. Vertical black bar indicates separation between monosomal and polysomal fractions. (Right) Graphs of the amount of mRNA in cumulative monosomal fractions for lysates transfected with indicated vectors. One-way ANOVA, Tukey’s post hoc test, df = 6 (n = 3). All data in this figure are presented as mean ± SEM. Tests are indicated in the legend and significant p values in the figure. See also Figure S6.

Figure 7

IMPA1-C rescues axonal degeneration when targeted distally (A) Immunostaining of sympathetic neurons and axons transfected with the indicated siRNAs and the mouse IMPA1 HA-tagged vectors. siRNAs: scrm, scrambled; IMPA1 CDS, targeting IMPA1 coding sequence; IMPA1-L, targeting the unique 120 nt of IMPA1-L 3′ UTR. Vectors: control, empty vector; IMPA1-C, expressing HA-tagged ms IMPA1-rat C; IMPA1-C+120, carrying also the 120 nt axonal localization element. Neurons were stained with DAPI, anti-HA, and anti-mCherry antibodies. Scale bar, 25 μm. (B) Representative images of superior cervical ganglia explants electroporated with scrambled siRNA or IMPA1 CDS siRNA, in the presence of HA-ms IMPA1-C (left) or HA-ms IMPA1-C+120, and GFP (right). Arrows point to healthy, intact axon bundles and arrowheads to degenerating axon bundles with characteristic beads-on-string appearance. Scale bar, 75 μm. (C) Quantitative analysis of the data shown in (A), presented as mean ± SEM. One-way ANOVA, Tukey’s multi-comparison test, F = 11.46, df = 104 (only statistically significant comparisons are indicated). (D) Schematic representation summarizing the cleavage and remodeling of 3′ UTR in axons. See also Figure S7.