Nucleolin-Mediated RNA Localization Regulates Neuron Growth and Cycling Cell Size

Abstract

How can cells sense their own size to coordinate biosynthesis and metabolism with their growth needs? We recently proposed a motor-dependent bidirectional transport mechanism for axon length and cell size sensing, but the nature of the motor-transported size signals remained elusive. Here, we show that motor-dependent mRNA localization regulates neuronal growth and cycling cell size. We found that the RNA-binding protein nucleolin is associated with importin β1 mRNA in axons. Perturbation of nucleolin association with kinesins reduces its levels in axons, with a concomitant reduction in axonal importin β1 mRNA and protein levels. Strikingly, subcellular sequestration of nucleolin or importin β1 enhances axonal growth and causes a subcellular shift in protein synthesis. Similar findings were obtained in fibroblasts. Thus, subcellular mRNA localization regulates size and growth in both neurons and cycling cells.

Graphical abstract

Figure 1

Increased Axonal Growth Rates in Importin β1 Mutant Sensory Neurons (A) Fluorescent images show cultured YFP-expressing DRG neurons from WT versus importin β1 3′ UTR-null mice at 48 hr in vitro. Scale bar, 100 μm. (B) Quantification of time-lapse imaging of YFP-expressing DRG neurons in culture. Images were taken every hour in a Fluoview FV10i incubator microscope. 3 × 3 montages of neighboring acquisition sites were analyzed using ImageJ. Longest neurite growth rates in these experiments were 6.9 ± 0.6 μm/hr for WT versus 11.5 ± 0.9 μm/hr for importin β1 3′ UTR^−/− mice. Mean ± SEM; n ≥ 30 cells per experimental group; ^∗p < 0.05 for comparison of growth rates, one-way ANOVA. (C) Whole-mount neurofilament staining in E11.5 limbs in WT and importin β1 3′ UTR^−/− mice is shown. Scale bar, 200 μm. (D) Quantification reveals significantly longer total neurite lengths at E11.5 in importin β1 3′ UTR^−/− embryos than in WT littermates (n ≥ 7; ^∗p < 0.05, Student’s t test). (E) Western blot quantifications for importin β1 in axon versus cell body compartments of sensory neurons cultured for 48 hr in compartmentalized Boyden chambers. A representative blot with the same loading order is shown above the graph. WT versus importin β1 3′ UTR^−/− neurons are shown. Mean ± SEM; n = 3; ^∗p < 0.05, Student’s t test. (F) Electron micrographs show immunogold labeling for importin β1 on ultrathin monolayer sections of cultured DRG neurons from WT and importin β1 3′ UTR^−/− mice. Scale bar, 200 nm; gold particle diameter, 10 nm. (G) Quantification of immunogold labeling confirms reduced levels of importin β1 protein in growing sensory axons of importin β1 3′ UTR^−/− mice. Mean ± SEM; n ≥ 50; ^∗∗∗p < 0.001, Student’s t test. (H) DRG neuron cultures from Islet-Cre RiboTag mice were immunostained for the tagged ribosome epitope (HA), ribosomal RNA (Y10B), and axonal tubulin (Tubb3). A representative axon tip is shown. Scale bar, 5 μm. For additional images see Figure S1C. (I) Quantification of ribosomal 18S RNA (left) and importin β1 mRNA (right) in HA-RiboTag pull-downs from axonal and cell body compartments from Islet-Cre RiboTag DRG neurons cultured for 96 hr in compartmentalized Boyden chambers. RNA levels are quantified as fold change of levels in control pull-downs from WT cultures. Mean ± SEM; n = 4; ^∗p < 0.05 and ^∗∗p < 0.005, Student’s t test. (J) Representative PLA images to identify importin β1-dynein complexes in DRG neurons grown for 48 hr in culture. After 48 hr the neurons were fixed and stained for dynein and importin β1, followed by the PLA probes. Scale bar, 50 μm. See also Figures S1D and S1E. (K) Co-immunoprecipation (coIP) of importin β1 with dynein from axoplasm. Immunoprecipitations were carried out with dynein IC74.1 intermediate-chain antibody versus non-immunized mouse IgG. See also Figure S1.

Figure 2

MAIL, A Localization Motif for Importin β1 mRNA (A) Schematic diagram of segments from the importin β1 3′ UTR (GenBank: JX096837.1) evaluated for axon-localizing activity. Regions predicted to contain stem-loop secondary structures are highlighted in red. The region between 1 and 134 nt encompasses the short form of importin β1 3′ UTR, which is restricted to the cell body. The motif for axonal importin localization (MAIL) is shown as a red stem-loop structure at 991–1,024 nt. (B) Sequences and schematic structure predictions of the MAIL motif and two derived mutants, GMAIL, with four U-G mutations in the loop region as shown, and IMAIL, which carries the GMAIL mutations together with additional mutations in the stem region, are shown. (C) Constructs containing deletions or fusions of the MAIL motif as indicated were fused with a destabilized myr-EGFP reporter and transfected to sensory neurons for FRAP analyses, with recovery monitored over 20 min. Representative images from time-lapse sequences before (−2 min) and after photobleaching (0 and 20 min) in the boxed region of interest are shown. For data from additional constructs, see Figure S2A. Scale bar, 25 μm. (D) Quantification of the FRAP analyses shown in (C). Average recoveries are shown (percentage of pre-bleach levels ± SEM). Anisomycin-treated neurons were exposed to 50 μM inhibitor prior to the imaging sequence. Time points with significant differences in axonal fluorescence compared to that observed in anisomycin-treated cultures are indicated (^∗∗∗p < 0.001 and ^∗∗p < 0.01, two-way ANOVA). For results with additional deletion constructs and anisomycin controls, see Figure S2B. (E) In situ hybridization on neurons transfected with the indicated constructs. Exposure-matched images show that only GFP mRNA with the MAIL element localizes into axons (right panel), while all reporter mRNAs are clearly expressed in corresponding cell body images (left panel). Scale bars, 25 μm (cell body) and 10 μm (axons). See also Figure S2C. See also Figure S2.

Figure 3

Axonal Nucleolin Interacts with the Importin β1 MAIL Motif (A) Bovine axoplasm (10 mg/lane) was precipitated on immobilized MAIL or GMAIL RNA motifs, and eluted proteins were separated by 10% SDS-PAGE. The gel region containing the major differential band is shown here and the complete gel is shown in Figure S3A. Mass spectrometry analyses identified nucleolin as the major unique MAIL-bound component (Figures S3A–S3C). (B) Western blot of nucleolin precipitated from rat sciatic nerve axoplasm with MAIL, GMAIL, or β-actin Zipcode RNA motifs. Precipitates were separated on 10% SDS-PAGE, blotted onto nitrocellulose, and probed with antibody against nucleolin. (C) Immunoprecipitation of 200 μg rat sciatic nerve axoplasm samples with control IgG or anti-nucleolin antibodies followed by RT-PCR for importin β1 or β-actin mRNAs. (D) Western blot of recombinant nucleolin precipitated with MAIL, IMAIL, or β-actin Zipcode RNA motifs. Input was 1 μg recombinant nucleolin per lane. (E) Primary cultured rat sensory neurons immunostained with antibodies against nucleolin (red) and NFH (green), revealing nucleolin in both neuronal cell bodies and axons. Scale bar, 20 μm; right overlay panel scale bar, 10 μm. (F) Sciatic nerve cross-sections immunostained with antibodies against nucleolin (red) and NFH (green), revealing nucleolin within sensory axons in vivo. Scale bar, 20 μm. (G) Electron micrographs showing immunogold labeling for nucleolin in axons on ultrathin monolayer sections of cultured mouse DRG neurons (left) or of sciatic nerve (right). Nucleolin is present in axons in vitro and in vivo. Scale bars, 200 nm; gold particle diameter, 10 nm. (H) Colocalization of nucleolin protein (immunostaining, green) and importin β1 mRNA (FISH, red) in sensory axons. Importin β1 mRNA colocalized with nucleolin protein (yellow) is shown in a single optical plane (scale bar, 5 μm). For cell body signal and scrambled probe control, see Figure S3D. Pearson’s correlation coefficient for importin β1 colocalization with nucleolin 0.37 ± 0.04 (n = 29) differs significantly from Pearson’s for β-actin or GAP43 (see below) (p value for importin β1 versus β-actin < 0.004, p value for importin β1 versus GAP43 < 0.0001; ANOVA with Bonferroni post hoc correction in both cases). (I) Colocalization of nucleolin protein (immunostaining, green) and β-actin mRNA (FISH, red) in sensory axons. Colocalization is shown in yellow in a single optical plane (scale bar, 5 μm). For cell body signal and scrambled probe control, see Figure S3D. Pearson’s correlation coefficient for β-actin colocalization with nucleolin 0.19 ± 0.03 (n = 20). (J) Colocalization of nucleolin protein (immunostaining, green) and GAP43 mRNA (FISH, red) in sensory axons. Colocalization is shown in yellow in a single optical plane (scale bar, 5 μm). For cell body signal and scrambled probe control, see Figure S3D. Pearson’s correlation coefficient for GAP43 colocalization with nucleolin 0.05 ± 0.01 (n = 48). See also Figure S3.

Figure 4

Depletion of Axonal Nucleolin Reduces Importin β1 Transcript in Sensory Axons (A) Cy3-labeled AS1411 and control aptamers were added to sensory neuron cultures to a final concentration of 20 μM. Neurons were fixed at the indicated time points. Scale bar, 10 μm. See also Figures S4A and S4B. (B) Western blots for nucleolin on axon versus cell body extracts from sensory neurons in compartmentalized Boyden chambers treated with AS1411 or control aptamers for 48 hr before transfer to aptamer-free medium for another 24 hr. Quantifications of blots are shown below. Mean ± SEM; n = 4; ^∗∗p < 0.01, Student’s t test. (C) Electron micrographs of cultured DRG neuron processes show immunogold labeling for nucleolin after AS1411 treatment. Scale bar, 200 nm; gold particle, 10 nm. Mean ± SEM; n ≥ 25; ^∗∗∗p < 0.001, Student’s t test. (D) Quantification of relative importin β1 transcript levels by qPCR on cell bodies and axons of cells treated with AS1411 or control DNAs. β-actin served as an internal control and did not change. Mean ± SEM; n = 3; ^∗∗p < 0.01, Student’s t test. (E) CoIP of Kif5A with nucleolin from sciatic nerve axoplasm. The control immunoprecipitation is in the presence of a blocking peptide for the nucleolin antibody. A supporting experiment is shown in Figure S4C. Quantification of the Kif5A-nucleolin coIP is shown above. Mean ± SEM; n = 5; ^∗p < 0.05, paired Student’s t test. (F) Importin β1 transcript levels co-precipitated with Kif5A. Mean ± SEM; n = 5; ^∗∗p < 0.01, ratio-paired Student’s t test. (G) Quantification of Kif5A on fluorescent Li-COR western blots of nucleolin immunoprecipitations from sciatic nerve axoplasm, after pre-incubation with AS1411 or control aptamer. Representative blots are shown below the graph. Mean ± SEM; n = 3; ^∗p < 0.05, paired Student’s t test. Similar results were obtained from neuronal cultures (Figure S4D). (H) Automated capillary electrophoresis quantification of kinesin heavy-chain (KHC) immunoreactivity co-precipitated with nucleolin from sciatic nerve axoplasm, after pre-incubation with AS1411 or control aptamer. Representative traces of the KHC immunoreactive peaks are shown on the left and quantifications are shown on the right. Mean ± SEM; n = 3; ^∗p < 0.05, paired Student’s t test. (I) Quantification of Kif5A protein pulled down by a MAIL RNA probe from sciatic nerve axoplasm pre-incubated with AS1411 or control DNA. Protein levels were quantified by automated capillary electrophoresis. Data are shown as percentage from control. Mean ± SEM; n = 9; ^∗∗p < 0.01, paired Student’s t test. See also Figure S4.

Figure 5

AS1411 Enhances Sensory Axon Growth Rates in WT, but Not in Importin β1 3′ UTR^−/− Neurons (A) Cultured DRG neurons from adult YFP/WT mice were treated with 10 μM control or AS1411 aptamer for 48 hr, and then they were replated in fresh medium without aptamer and allowed to re-grow. Representative images at three time points following replating are shown. Scale bar, 100 μm. (B) Quantification of total neurite outgrowth of sensory neurons in culture from the experiment described in (A). Total neurite growth rates in these experiments were 45.7 ± 11.2 μm/hr for control versus 90.1 ± 16.1 μm/hr for AS1411 treatment. Mean ± SEM; n ≥ 60 cells per experimental group; ^∗∗∗p < 0.001 for comparison of growth rates, one-way ANOVA. (C) Representative images of cultured WT or importin β1 3′ UTR^−/− sensory neurons treated with 10 μM control or AS1411 aptamer for 48 hr and then replated and cultured for an additional 24 hr in fresh medium without aptamer. Neurons were finally fixed, immunostained for NFH (green), and imaged. Scale bar, 200 μm. (D) Quantification of total neurite outgrowth in the experiment described in (C) reveals a significant increase in axon growth in WT neurons pretreated with AS1411, but not in importin β1 3′ UTR^−/− neurons. Mean ± SEM; n ≥ 300 cells per experimental group; ^∗∗∗p < 0.001, Student’s t test. See also Figure S5.

Figure 6

AS1411 Treatment Increases 3T3 Fibroblast Cell Size (A) Western blots showing coIP of importin β1 with dynein heavy chain 1 and nucleolin with Kif5B from confluent 3T3 cell cultures. Control immunoprecipitations were with non-immune IgG for the dynein immunoprecipitation and with blocking peptide for the precipitating antibody in the nucleolin immunoprecipitation. (B) Quantification of relative importin β1 transcript levels after pull-down for Kif5A or nucleolin is shown. Mean ± SEM; n = 4; ^∗∗∗p < 0.001, ratio-paired Student’s t test. (C) Representative images for uptake of AS1411-Cy3 into 3T3 cells are shown. Blue, DAPI; red, AS1411. Scale bar, 10 μm. (D) Representative PLA images of importin β1-dynein complexes in 3T3 cells incubated for 48 hr with AS1411 or control aptamers. After 48 hr the cells were fixed and stained with phalloidin-Cy3 and for dynein and importin β1, followed by the PLA probes. Scale bar, 50 μm. (E) Quantification of the assay shown in (D). PLA signal per cell body area was quantified using Cellprofiler software, revealing a significant reduction in signal density in 3T3 cells incubated with the AS1411 aptamer. Mean ± SEM; n = 3; ^∗∗p < 0.01, Student’s t test. (F) Representative western blots of importin β1 co-precipitated with dynein from 3T3 cells after 48 hr in culture in the presence of AS1411 or control aptamers. The quantification below shows a significant decrease in coIP of importin β1 with dynein after AS1411 treatment. Mean ± SEM; n = 3; ^∗p < 0.05, paired Student’s t test. (G) 3T3 cells were incubated with 10 μM AS1411 or control aptamer for 48 hr, after which 20,000 cells were replated for another 24 hr in fresh medium without aptamer before fixing and staining with rhodamine-phalloidin. Representative images are shown. Scale bar, 100 μm. See also Figure S6 for higher magnification images of nuclear morphology. (H) Quantification of 3T3 cell area from the experiment described in (G) reveals a significant increase upon AS1411 treatment. Mean ± SEM; n > 1,000; ^∗∗∗p < 0.001, Student’s t test. The experiment was replicated on three independent cultures. (I) 3T3 cell size at different stages of the cell cycle after 48 hr incubation with AS1411 or control aptamers at 10 μM, followed by harvesting and incubation with 10 μg/ml Hoechst 33342 and 5 μg/ml propidium iodide for live cell cycle analyses by FACS. 30,000 events were collected per sample. AS1411 treatment causes a marked increase in cell size, as shown by the right shift in population distribution in comparison to mock and control at all stages of the cell cycle. (J) Quantification of the FACS described in (I) for three independent experiments reveals a significant increase of cell size upon AS1411 treatment in all cell cycle phases. Mean ± SEM; n = 3; ^∗∗p < 0.01, Student’s t test. See also Figure S6.

Figure 7

Nucleolin and Importin β1 Localization Regulate Protein Synthesis (A) The translational activity of DRG neurons in culture was assessed by puromycin incorporation. Cultures were grown in the presence of AS1411 or control aptamer for 48 hr, and then they were replated and cultured for an additional 24 hr in fresh medium without aptamer. Neurons were then pulsed with 5 mM puromycin for 10 min at 37°C or preincubated with 40 mM anisomycin for 30 min followed by the 5 mM puromycin pulse, and then they were fixed. Fixed cultures were immunostained for NFH (green) and α-puromycin (red). Scale bar, 100 μm. For anisomycin control, see Figures S7E and S7F. (B) Representative high-sensitivity zoom images of the boxed regions in (A) reveal protein synthesis in axon tips. Scale bar, 20 μm. Quantification reveals a significant decrease in protein synthesis in axon tips of AS1411-treated WT neurons, as well as in importin β1 3′ UTR^−/− neurons. Axon tip synthesis was quantified as ratios of cell body values and then normalized to WT control. Mean ± SEM; n ≥ 80 cells from three independent cultures; ^∗∗p < 0.01 and ^∗∗∗p < 0.001, Student’s t test. (C) Representative images of cultured 3T3 cells treated with 10 μM control or AS1411 aptamer for 48 hr and then replated and cultured for an additional 24 hr in fresh medium without aptamer. The cells subsequently were incubated with puromycin with or without anisomycin as described above, and then they were fixed and stained for F-Actin, DAPI, and α-puromycin. Scale bar, 50 μm. (D) Quantification of puromycin labeling in the cytoplasm of 3T3 cells from the experiment described in (C) reveals a significant decrease in protein synthesis at the cell periphery in AS1411-treated cells. Mean ± SEM; n ≥ 200 cells from five independent cultures; ^∗p < 0.05, one-way ANOVA with Bonferroni post hoc test. (E) Schematic model of the mechanism proposed in this study. Nucleolin binds importin β1 and likely other mRNAs, and the complex is transported by a kinesin motor to the axon in a neuron or the cell cortex in cycling cells. Upon arrival at the end of the microtubules, the complex is dissassembled, with nucleolin likely docking to the plasma membrane. Local translation of the cargo RNAs generates proteins that are retrogradely transported with dynein to influence protein synthesis in the soma. The dashed line indicates a negative feedback loop postulated in the original model (Rishal et al., 2012), the details of which are still unknown. See also Figure S7.