Single-Molecule Imaging of PSD-95 mRNA Translation in Dendrites and Its Dysregulation in a Mouse Model of Fragile X Syndrome

Abstract

Fragile X syndrome (FXS) is caused by the loss of the fragile X mental retardation protein (FMRP), an RNA binding protein that regulates translation of numerous target mRNAs, some of which are dendritically localized. Our previous biochemical studies using synaptoneurosomes demonstrate a role for FMRP and miR-125a in regulating the translation of PSD-95 mRNA. However, the local translation of PSD-95 mRNA within dendrites and spines, as well as the roles of FMRP or miR-125a, have not been directly studied. Herein, local synthesis of a Venus-PSD-95 fusion protein was directly visualized in dendrites and spines using single-molecule imaging of a diffusion-restricted Venus-PSD-95 reporter under control of the PSD-95 3'UTR. The basal translation rates of Venus-PSD-95 mRNA was increased in cultured hippocampal neurons from Fmr1 KO mice compared with WT neurons, which correlated with a transient elevation of endogenous PSD-95 within dendrites. Following mGluR stimulation with (S)-3,5-dihydroxyphenylglycine, the rate of Venus-PSD-95 mRNA translation increased rapidly in dendrites of WT hippocampal neurons, but not in those of Fmr1 KO neurons or when the binding site of miR125a, previously shown to bind PSD-95 3'UTR, was mutated. This study provides direct support for the hypothesis that local translation within dendrites and spines is dysregulated in FXS. Impairments in the regulated local synthesis of PSD-95, a critical regulator of synaptic structure and function, may affect the spatiotemporal control of PSD-95 levels and affect dendritic spine development and synaptic plasticity in FXS.

Keywords: FMRP; Fragile X syndrome; PSD-95; PSD-95 mRNA; local translation; single-molecule imaging.

Figure 1.

PSD-95 mRNA translation occurs in dendrites and dendritic spines. A, Venus-PSD-95 translation reporter: Venus-PSD-95 fusion protein with PSD-95 ORF, 5′UTR, and 3′UTR. Blue circle represents miR-125a binding site. Ten amino acids representing the N terminus of PSD-95, which contains two cysteine residues shown to be palmitoylated and necessary for membrane targeting of PSD-95 (Topinka and Bredt, 1998), were inserted at the N terminus of Venus (red) to favor membrane targeting and thus limit diffusion. In the proximal 3′UTR, there are four copies of boxB phage sequence inserted for mRNA labeling by λN22 phage protein fused to three CFP and nuclear localization signal (M9) (see Fig. 4). B, Illustration depicting visualization of local translation of Venus-PSD-95 reporter within a dendrite. While PSD-95 ORF is still being translated, Venus ORF has time to start the maturation process to become fluorescent and thus increase the chances of Venus-PSD-95 reporter to be detected near its translation site. In addition, the N terminus of PSD-95, which contains two cysteine residues where PSD-95 is palmitoylated, could help to anchor Venus-PSD-95 to the cell membrane in the proximity of the translation site. C, Diagram depicting the experimental setup, with the illumination area (large outer circle) situated in an apical dendrite of a neuron expressing Venus-PSD-95. White interior circle represents the ROI used for analysis. To filter out detection of possible mobile Venus-PSD-95 molecules that might have been synthesized nearby, all Venus-PSD-95 molecules detected within the periphery of the illumination area (red ring) were excluded from analysis. D, Frame 1 of a representative time-lapse imaging of a dendrite of a 14 DIV hippocampal neuron transfected with Venus-PSD-95 translation reporter was used to determine the dendrite contour, before gradual photo-bleaching of preexisting Venus protein. A dendritic spine (white rectangle) was analyzed in panels E, G. Scale bar, 5 μm. E, Montage of the time-lapse imaging corresponding to the white rectangle in D. The four successive frames of time-lapse imaging were collected at ∼20 min after Frame 1 and are displayed in pseudo-color within a 36,300–35,000 pixel intensity value range, which allowed visualization of individual Venus-PSD-95 molecules as light flashes on a black background after the preexisting signal was photo-bleached (E, arrowheads). The full arrowheads indicate the appearance of three molecules in the same frame. Observe the single-molecule behavior of the particles: single-step appearance and single-step disappearance. Empty arrowhead at 300 ms indicates one of the molecules that appeared at 150 ms, which is assigned as being the same molecule even with a small displacement. The montage shows raw data images, without any processing, aside from contrast adjustment. Scale bar, 2 μm. F, Event map corresponding to the dendrite ROI shown in D. The positions of the first appearance of all Venus-PSD-95 molecules detected over a time interval (15 s here) were recorded and plotted generating an event map (Tatavarty et al., 2012). Each green dot indicates the location where each newly translated Venus-PSD-95 molecule was detected. Red dots (white arrows) indicate the individual Venus-PSD-95 molecules indicated by white arrows in E. The dendrite contour is outlined in white, based on the image shown in D. G, Enlarged inset of the event map corresponding to the white rectangle in D. H, Event schedule showing the detection of events in the dendrite ROI shown in D, F, over a 15 s period. Red bar represents the three molecules shown in E–G. I, Montage showing an ROI (1.5 μm/1.5 μm) from time-lapse imaging of a dendrite expressing Venus-PSD-95 over 100 frames (∼15 s at 0.15 ms interval) advances from left to right and top to bottom. The four discrete flashes of light detected represent single-Venus-PSD-95 molecules. Green squares represent the frame where the single molecule (light flash) was first detected, which was used to establish the position and the time of appearance of each Venus-PSD-95 molecule. The numbers on top and left side indicate the frame number within the montage. The montage shows raw data images, without any processing, aside from contrast adjustment. Scale bar, 1.5 μm.

Figure 2.

Venus-PSD-95 translation events are detected near synapses. A, Montage showing raw time-lapse images of a dendrite segment of a 14 DIV hippocampal neuron expressing Venus-PSD-95 reporter. Individual Venus-PSD-95 molecules are indicated by white arrowheads, in the first appearance frame, and by open arrowheads subsequently. The montage shows raw data images, without any processing, aside from contrast adjustment. Scale bar, 2.5 μm. B, Single-molecule analysis of Venus-PSD-95 translation was first performed in live cells to generate an event map (green) followed by immunostaining for synapsin (red) and Venus (blue). Venus signal in the first frame of the time-lapse and the Venus signal from the immunostaining image were used to determine the cell contour and to align and merge the event map (green) to the synapsin (red). Scale bar, 2.5 μm. Translation event clusters are visualized in dendritic shaft and in spines and near synapsin puncta. White rectangles labeled 1, 2, and 3 represent ROIs shown at higher magnification at B (bottom). Scale bar, 0.5 μm. C, Quantification of the spatial relationship between Venus-PSD-95 translation events and synapsin puncta. Venus-PSD-95 translation events were classified based on whether they overlap with synapsin puncta or not, and the distance from an event to the periphery of the closest synapsin punctum. The vast majority of the events is detected at <0.5 μm from a synapsin punctum. N = 5 dendrites, 3 different cultures, 2333 translation events quantified. Error bars indicate SEM.

Figure 3.

Venus-PSD-95 translation event rate decreases following treatment with anisomycin or TTX. A, Event maps (shown as heatmap) of events detected over 3 min, before addition of anisomycin (top) or 30 min after addition of anisomycin (bottom). Scale bar, 5 μm. B, Quantification of events (normalized) before addition of anisomycin (before anis) or 30 min after addition of anisomycin (after anis). N = 4. *p < 0.05 (t test). C, Quantification of extracellular noise (normalized) before addition of anisomycin (before anis) or 30 min after addition of anisomycin (after anis). N = 3. ^#p > 0.1 (t test). D, Event maps of events detected over 3 min, at 30 min after addition of anisomycin (+ anisomycin; top) or at 45 min after anisomycin was washed out (− anisomycin; bottom). Scale bar, 2 μm. E, Quantification of events (normalized) at 30 min after addition of anisomycin (+ anisomycin) and at 45 min after anisomycin was removed (− anisomycin). N = 3. *p < 0.05 (t test). F, Event maps of events detected over 3 min were created at time 0 (0 min) and after 30 min. DIC images were also collected at 0 and 30 min. Overlay of the first time-lapse frame (showing the dendrite outline; blue), the DIC image, and the corresponding event map, reveals similar event rate at 0 and 30 min (quantified in G), and no changes in the dendrite morphology as viewed in DIC. G, Quantification of events (normalized) detected in dendrites at time 0 (0 min) and after 30 min. N = 4. ^#p > 0.1 (t test). H, Event maps of events detected over 3 min were created before TTX and at 15 min after adding TTX to the medium (after TTX). DIC images were also collected before and after TTX. Overlay of the first time-lapse frame (showing the dendrite outline; blue), the DIC image, and the corresponding event map reveals decreased event rate after TTX (quantified in I), and no changes in the dendrite morphology as viewed in DIC. I, Quantification of events (normalized) before addition of TTX (before TTX) or 15 min after addition of TTX (after TTX). N = 3. *p < 0.05 (t test). J, Fluorescence intensity of individual particles detected in neurons was similar to that of purified individual Venus molecules imaged in vitro using identical parameters. K, Particles detected in neurons showed fast photo-bleaching kinetics, similar to that previously reported for Venus (Yu et al., 2006).

Figure 4.

Expression and localization of Venus-PSD-95 in dendrites relative to endogenous PSD-95. A, Representative images of 14 DIV hippocampal neurons mock-transfected or transfected with Venus-PSD-95 reporter on which immunocytochemistry for PSD-95 and Venus (GFP) was performed. PSD-95 puncta intensity is quantified in B. Arrowheads indicate PSD-95 puncta where PSD-95 and Venus (GFP) signal overlap. Raw data epifluorescence images are shown, without any processing aside from contrast adjustment. Scale bars, 20 μm. B, Quantification of PSD-95 puncta intensity in mock- and Venus-PSD-95 reporter-transfected 14 DIV neurons. Three independent experiments done in three different cultures were quantified. N (Venus-PSD-95/mock) = 14/12 (neurons), 766/526 (PSD-95 puncta). *p < 0.05 (t test). C, Representative images of 14 DIV hippocampal neurons mock-transfected or transfected with Venus-PSD-95 reporter on which FISH for PSD-95 mRNA and Venus (GFP) mRNA was performed using Stellaris probes. PSD-95 mRNA FISH signal dendritic intensity at >50 μm from the cell body is quantified for both conditions in D. Raw data epifluorescence images are shown, without any processing, aside from contrast adjustment. Scale bars, 20 μm. D, Quantification of PSD-95 FISH signal intensity in mock- and Venus-PSD-95 reporter-transfected 14 DIV neurons. Three independent experiments done in three different cultures were quantified. N (Venus-PSD-95/mock) = 6/23 (neurons), 29/38 (dendrites). *p < 0.05 (t test). E, Diagrams of Venus-PSD-95 (top), λN22-CFP (middle), and mCherry-FMRP (bottom) constructs. Venus-PSD-95 mRNA contains four boxB sequences, which are bound by λN22-CFP (containing M9 nuclear localization signal). F, Dendrite of a 14 DIV neuron cotransfected with the constructs shown in E. Green represents Venus-PSD-95 mRNA labeled by λN22-CFP. Red represents mCherry-FMRP. Arrowheads indicate the Venus-PSD-95 mRNA granules that colocalize with mCherry-FMRP.

Figure 5.

Developmental pattern of endogenous PSD-95 protein expression is perturbed in Fmr1 KO hippocampal neurons. Hippocampal neurons at 7, 14, or 21 DIV were fixed and immunocytochemistry was performed as indicated. To determine the developmental level of expression of the indicated proteins, the dendritic average fluorescence intensity was measured along the proximal 50 μm of dendrites using a line scan. All the experiments were repeated at least three times using at least three different cultures. N represents the total number of neurons quantified per condition. A, Representative images of 7, 14, and 21 DIV WT and Fmr1 KO cultured hippocampal neurons from PSD-95 immunofluorescence. Scale bar, 10 μm. PSD-95 fluorescence intensity is shown as a heat map. B, Quantification of a representative PSD-95 immunofluorescence experiment. The experiments were repeated three times using three different cultures and compared by two-way ANOVA (p < 0.05). N_WT (7/14/21 DIV) = 60/77/47; N_KO (7/14/21 DIV) = 63/87/55. C, Dendritic FMRP levels in 7, 14, and 21 DIV WT hippocampal neurons. Representative images are shown. Scale bar, 10 μm. FMRP immunofluorescence intensity is shown as a heat map. D, Quantification of a representative FMRP immunofluorescence experiment is shown. The experiments were repeated three times, using different cell cultures. N (7/14/21 DIV) = 37/28/37 neurons in total. E, Dendritic synapsin (red) and PSD-95 (green) levels in 7, 14, and 21 DIV WT hippocampal neurons. Representative images are shown. Scale bar, 25 μm. F, Average of three synapsin immunofluorescence experiments from three different cell cultures. N (7/14/21 DIV) = 26/41/33 neurons in total. G, Higher magnification of the white rectangles shown in E, depicting synapsin (red) and PSD-95 (green) expression levels at 7, 14, and 21 DIV, respectively. Scale bar, 10 μm.

Figure 6.

Basal translation of Venus-PSD-95 is increased in Fmr1 KO hippocampal neurons. Representative event maps showing the Venus-PSD-95 translation events detected over 5 min in a dendrite of a 12–14 DIV WT (A) and Fmr1 KO (B) hippocampal neuron. Each green dot represents the location where each newly translated Venus-PSD-95 molecule was first detected. The dendrite contour is outlined in white. Red circles represent the ROI where the translation rate was quantified in spines or in dendritic shaft. Scale bar, 1 μm. Representative event schedules showing Venus-PSD-95 translation in WT (C) and Fmr1 KO neurons (D). Spine (top) and base of spine (bottom). E, Quantification of Venus-PSD-95 translation rates (events/min) measured in spine and dendritic shaft for WT and Fmr1 KO hippocampal neurons at 13–15 DIV. *p < 0.05 (t test). WT events/min: spine, 0.6; spine base, 1.2; N = 13. KO events/min: spine, 1.4; spine base, 4.7; N = 14.

Figure 7.

Activity-regulated translation of PSD-95 mRNA is occluded in Fmr1 KO mouse hippocampal neurons. A, WT and miR125aMut constructs. The miR125a binding site in the PSD-95 3′UTR was mutated as previously described (Muddashetty et al., 2011). B, Dendrite of a 14 DIV neuron expressing Venus-PSD-95 translation reporter. Single-molecule imaging was performed before and after DHPG treatment. Translational events detected under control conditions (green) and after DHPG treatment (red) were overlaid on the first image of the time-lapse. The dendrite countour (blue) is delineated by the Venus-PSD-95 protein accumulated in the cells before photo-bleaching. After mGluR stimulation with DHPG, new translation events are detected in proximity of sites where translation events were detected before DHPG treatment (open arrowheads) but also at new sites (white arrowheads). C, Event schedule corresponding to the dendrite shown in B. Green represents events detected under control conditions. Red represents events detected after DHPG treatment. D–L, Single-molecule imaging was performed in WT and Fmr1 KO neurons expressing Venus-PSD-95 reporter, and in WT neurons expressing miR125aMut reporter, before and after adding DHPG. Representative event schedules for each condition, which were not normalized are shown in D, G, J. The event rate before DHPG was used as an internal control, and the rates after DHPG were normalized to the rate before DHPG for each cell (E, F, H, I, K, L). D, Representative event schedule of a WT neuron showing the translation rate before and after DHPG treatment. E, Treatment with DHPG induces a rapid increase in normalized translation rate of PSD-95 mRNA in dendrites of WT neurons (13–15 DIV). N = 7. F, Data as in E, binned at 3 min intervals. N = 7. *p < 0.05 (t test). G, Representative event schedule of a Fmr1 KO neuron showing the translation rate before and after DHPG treatment. H, Treatment with DHPG did not change normalized translation rate of PSD-95 mRNA in dendrites of Fmr1 KO neurons (13–15 DIV). N = 4. I, Data as in H, binned at 3 min intervals. N = 4. *p < 0.05 (Student's t test). J, Representative events schedule of a WT neuron transfected with the miR125aMut construct showing the translation rate before and after DHPG treatment. K, Treatment with DHPG does not enhance translation rate of miR125aMut Venus-PSD-95 mRNA reporter in dendrites of WT neurons (13–15 DIV). N = 7. L, Data as in K, binned at 3 min intervals. N = 7. NS In all panels, error bars indicate SEM. *p < 0.05 (t test). N is shown for each experiment. Repeated-measures analysis shows that DHPG response in WT neurons (E) is significantly different from that in Fmr1 KO neurons (H) or when miR125a mutant construct is used (K). *p < 0.05.

Figure 8.

Model for regulation of PSD-95 mRNA translation at the synapse. At the synapse, PSD-95 mRNA is in two distinct states: (1) At resting state, PSD-95 mRNA is bound by FMRP, which represses its translation in conjunction with RISC complex (which contains AGO2 and miR-125a–5p). (2) Upon activation of mGluRs, FMRP is dephosphorylated and subsequently ubiquitinated and degraded. Dephosphorylation of FMRP leads to disassembly of RISC complex and derepression of PSD-95 mRNA translation. Translation of Fmr1 mRNA present at the synapse replenishes the pool of FMRP at the synapse, leading to return to resting state (1).