Presynaptic FMRP and local protein synthesis support structural and functional plasticity of glutamatergic axon terminals

Abstract

Learning and memory rely on long-lasting, synapse-specific modifications. Although postsynaptic forms of plasticity typically require local protein synthesis, whether and how local protein synthesis contributes to presynaptic changes remain unclear. Here, we examined the mouse hippocampal mossy fiber (MF)-CA3 synapse, which expresses both structural and functional presynaptic plasticity and contains presynaptic fragile X messenger ribonucleoprotein (FMRP), an RNA-binding protein involved in postsynaptic protein-synthesis-dependent plasticity. We report that MF boutons contain ribosomes and synthesize protein locally. The long-term potentiation of MF-CA3 synaptic transmission (MF-LTP) was associated with the translation-dependent enlargement of MF boutons. Remarkably, increasing in vitro or in vivo MF activity enhanced the protein synthesis in MFs. Moreover, the deletion of presynaptic FMRP blocked structural and functional MF-LTP, suggesting that FMRP is a critical regulator of presynaptic MF plasticity. Thus, presynaptic FMRP and protein synthesis dynamically control presynaptic structure and function in the mature mammalian brain.

Keywords: LTP; RNA-binding protein; enriched environment; fragile X syndrome; hippocampus; protein synthesis; structural plasticity.

Figure 1.. MFs synthesize protein locally.

A. Representative tiled fluorescence image of a hippocampal slice expressing Halo-actin labeled with JF549 HaloTag ligand after targeted stereotactic injection in dentate gyrus. B. CHX bath application blocks labeling of newly synthesized Halo-actin (JF646) in individual MF boutons. (Top) Timeline of Halo-actin pulse-chase experiment. (Bottom left) Representative images of JF549/JF646 (cyan/magenta) labeled Halo-actin in mossy fiber boutons (arrowheads) from Control or CHX-treated slices. (Bottom right) Control: 1.00 ± 0.17 vs. CHX: 0.36 ± 0.06 (Mean ± S.E.M. of slices), Mann-Whitney, U = 1, p = 0.03; n = 29, 20 boutons; 5, 4 slices respectively; 4 animals. C. Transected mossy fiber axons synthesized β-actin protein at the same levels as intact control axons. (Top) Timeline of Halo-actin pulse-chase experiment. (Bottom left) Representative images of JF549/JF646 (cyan/magenta) labeled Halo-actin in mossy fiber boutons (arrowheads) from control or transected slices. (Bottom right) Control: 1.01 ± 0.28 vs. transected: 0.99 ± 0.26 (Mean ± S.E.M. of slices), Mann-Whitney, U = 28, p > 0.999; n = 23 boutons, 8, 7 slices respectively; 4 animals. N = slices. Black line and bar represent the mean ± 95% confidence interval of boutons. Points representing individual boutons are color-coded by slice and normalized to mean of Control. See also Supp. Figure 1. n.s. = p>0.05; *p≤0.05; **p≤0.01; ***p≤0.001; ****p≤0.0001.

Figure 2.. Locally synthesized actin contributes to MFB actin pool within 1 hour.

A. (Top) Timeline of Halo-actin pulse-chase photoactivation (PA) experiment. (Left, bottom) Representative tiled images of slices from Halo-actin injected mice incubated with JF549, followed by PA-JF646, then photoactivated in stratum lucidum (pink circle) or no PA for controls. (center) MFBs from slices that underwent PA had significantly greater levels of newly synthesized Halo-actin than those with no PA (scale bar = 5 μm). B. Quantification of locally synthesized Halo-actin measured by intensity of PA-JF646 inside MFBs identified using JF549. PA: 2144.0 ± 137.2 v. no PA: 359.7 ± 29.2 (Mean ± S.E.M.); Mann-Whitney, U = 0, p < 0.0001. n = 31, 20 boutons, 5,3 slices respectively, 3 animals. C. (Left, top) Timeline of 1 hour post Halo-actin PA experiment. (Left, bottom) Representative tiled images of slices from Halo-actin injected mice photoactivated in DG (pink circle) 1 hour prior to fixation. (Right) Representative images of MFBs 1 hour after PA at different distances from the DG somas (CA3a, CA3b, CA3c) with Halo-actin labeled using Halo-Ab staining (scale bar = 5 μm). D. (Left, top) Timeline of 3.5 hour post Halo-actin PA experiment. (Left, bottom) Representative tiled images of slices from Halo-actin injected mice photoactivated in DG (pink circle) at 3.5 hours prior to fixation. (right) Representative images of MFBs with all Halo-actin labeled using Halo-Ab staining at different distances from the DG somas (CA3a, CA3b, CA3c) 3.5 hours after PA (scale bar = 5 μm). E. Quantification of PA-JF646 mean intensity in single MFBs revealed that there was no significant difference between Halo-actin levels slices with no PA vs slices that were fixed 1 hour after PA in the CA3b region where electrophysiological recordings of MF-LTP are performed. After 3.5 hours, PA-Halo-actin from DG is detectable in CA3b MFBs. No PA: 344.4 ± 78.5 v. 1 hour: 307.8 ± 55.8 v. 3.5 hours: 1582.0 ± 254.4 (Mean ± S.E.M. of slices); One-Way ANOVA with Tukey’s test for Multiple Comparisons, F [2,9] = 21.79, p < 0.001. n = 20, 24, 24 boutons, 3,5,4 slices respectively, 3–4 animals. N = slices. F. Summary data of mean intensity averaged by slice show significant increases in all CA3 subregions 3.5 hours v. 1 hour after PA. Two-Way ANOVA with Sidak’s Test for Multiple Comparisons, F[1,21] = 48.54, p < 0.0001. n = 5 slices (1 hr), 4 slices (3.5 hours) 4 animals. Black line and bar represent the mean ± 95 % confidence interval of boutons. Points representing individual boutons are color-coded by slice. n.s. = p>0.05; *p≤0.05; **p≤0.01; ***p≤0.001; ****p≤0.0001.

Figure 3.. Ribotag mouse reveals MF axons and boutons contain ribosomes

A. (Top) Representative tiled image of hippocampal slice from Ribotag transgenic mouse injected with lentivirus encoding for ChIEF-tdTomato driven by C1QL2 promoter for dentate GC-specific expression. Dashed line indicates outline of granular layer in dentate gyrus. (Bottom) Magnification of MF boutons and axons show HA labeling in Cre+ slices. White boxes indicate magnified boutons shown in panel B. B. Magnification of Cre+ MFBs with HA puncta indicated by white arrowheads. Clustering is noted in some boutons. C. HA intensity inside Cre+ MFBs is significantly increased compared to GFP Controls with no HA expression. Control: 0.020 ± 0.002 v. Cre: 1.77 ± 0.25 (Mean ± S.E.M.); Mann-Whitney, U = 264.5, p < 0.0001. n = 39, 35 boutons, 7, 5 slices respectively, 4 animals. D. Relative frequency (%) histogram of HA intensity values showed that ~ 65% of Cre+ MFBs contain ribosomes (MFBs with >10 fold Control levels) compared to 0% in Control slices. E. HA intensity was significantly greater at all distances from the GC soma and highest in the CA3b region. Two-Way ANOVA with Tukey’s test for Multiple Comparisons F[1,55] = 5.82, p < 0.0001, CA3b:Cre vs. CA3c:Cre, p < 0.01, n = 33, 28 boutons, 6, 4 slices respectively, 4 animals. n.s. = p>0.05; *p≤0.05; **p≤0.01; ***p≤0.001; ****p≤0.0001.

Figure 4.. LTP induction elicits changes in local actin synthesis in MF tract

A. (Top) Timeline of Halo-actin LTP experiment. (Bottom) Representative images of pulse-chase labeled mossy fiber boutons in Halo-actin injected slices from CTRL v. LTP slices. B. MF-LTP increased newly synthesized Halo-actin in individual MF boutons CTRL: 1.0 ± 0.10 v. LTP: 1.29 ± 0.09 (Mean ± S.E.M. of slices); Unpaired t-test, p = 0.04, n = 122, 120 boutons 16,14 slices respectively, 8 animals. Black line and bar represent the mean ± 95 % confidence interval of slices. Points representing average bouton intensity per slice are normalized to mean of Control. C. (Left) Relative frequency (%) histogram of bouton intensity values indicated LTP primarily impacts low intensity boutons, shifting translationally quiescent boutons (i.e. Bin Center 5) to a translationally active state (i.e. Bin Center ≥ 25), KS test, p = 0.04, n = 122, 120 boutons 16,14 slices respectively, 8 animals. (Right) Representative boutons from indicated Bins corresponding to histogram on left. N = slices. n.s. = p>0.05; *p≤0.05; **p≤0.01; ***p≤0.001; ****p≤0.0001.

Figure 5.. Structural and functional plasticity of MFBs requires protein synthesis

A. Representative tiled image of acute hippocampal slice from WT C57 mice injected with C1QL2-ChiEFtom2A-GFP. B. Representative optogenetically-induced, extracellular field excitatory postsynaptic potentials (o-fEPSPs) recorded in CA3 resulting from optogenetic light activation (470 nm) in the hilus of slices from A. using MF-LTP induction protocol (125 pulses, 25 Hz, 3×). C. Representative 3D reconstructions of giant MFBs from slices in A. D. Quantification of MFB volume (μm³) from 3D reconstructions revealed MF-LTP significantly increased bouton volume 1 hour post-LTP. This increase was blocked by bath application of protein synthesis inhibitor, cycloheximide (CHX, 80 μM). Veh: 14.24 ± 1.34 v. Veh + LTP: 24.31 ± 2.09 v. CHX: 14.84 ± 3.94 v. CHX + LTP: 13.91 ± 1.30; Two-Way ANOVA with Tukey’s test for Multiple Comparisons F[1,19] = 5.82, p = 0.024, n = 41, 41, 36, 54 boutons respectively, 6 slices, 3 animals. E. LTP induction resulted in significant shift in the distribution of bouton volume when protein synthesis is intact. KS test, Z = 3.11, p < 0.00001. F. Length of filopodia was not altered by LTP induction or blockade of protein synthesis. Veh: 1.10 ± 0.18 v. Veh + LTP: 0.91 ± 0.16 v. CHX: 1.10 ± 0.11 v. CHX + LTP: 1.12 ± 0.14; Two-Way ANOVA with Tukey’s test for Multiple Comparisons F[1,252] = 3.00, p = 0.08, n = 52, 57, 64, 83 filopodia respectively, 6 slices, 3 animals. G. Number of filopodia per bouton was not altered by LTP induction or blockade of protein synthesis. Veh: 1.10 ± 0.18 v. Veh + LTP: 0.92 ± 0.16 v. CHX: 1.07 ± 0.12 v. CHX + LTP: 1.12 ± 0.14; Two-Way ANOVA with Tukey’s test for Multiple Comparisons F[1,263] = 0.63, p = 0.43, n = 48, 62, 83, 74 boutons respectively, 6 slices, 3 animals. H. MF-LTP was intact in transected slices, but blocked by bath application of cycloheximide (CHX, 80 μM). Transected: 164.92 ± 13.66 v. transected + CHX: 111.32 ± 11.75; Two sample t-test, p = 0.018, n = 5s, 4a (Transected + CHX) 6s, 6a (Transected). N = slices. Individual points representing boutons are color-coded by slice. Large black circle and bar represent the mean ± 95 % confidence interval of boutons. See also Supp. Figure 2. n.s. = p>0.05; *p≤0.05; **p≤0.01; ***p≤0.001; ****p≤0.0001.

Figure 6.. Presynaptic FMRP regulates MF activity-dependent protein synthesis and structural and functional plasticity

A. (Top) Representative images of puromycin labeling (PMY) in slices from Fmr1^fl/fl mice injected with CAMKII-mCherry-Cre or CAMKII-mCherry for WT and Fmr1 cKO. Dotted lines indicate the MF tract (stratum lucidum, SL) labeled with marker ZnT3. The stratum radiatum (SR) was measured as a control. LTP was induced with electrical stimulation (3X 125 pulses, 25 Hz). B. Fmr1 cKO increased PMY labeling in SL under basal conditions and with LTP induction. WT: 1.00 ± 0.03 v. WT + LTP: 1.75 ± 0.08 v Fmr1 cKO: 1.62 ± 0.07 v. Fmr1 cKO + LTP: 2.58 ± 0.06 (Mean ± S.E.M.); Two-Way ANOVA with Tukey’s test for Multiple Comparisons, Genotype: F[1,57] = 30.94, p < 0.0001, Condition: F[1,57] = 43.5, p < 0.0001. n = 13, 15, 17, 16 slices respectively, 8 mice. Points represent individual slices and are normalized to mean of WT. C. Representative 3D reconstructions of giant MF boutons from Fmr1^fl/fl mice injected with lentivirus encoding C1QL2-ChiEFtom2A-Cre (Fmr1 cKO) or C1QL2-ChiEFtom2A-GFP (CTRL). D. MF bouton volume is increased by LTP induction in WT mice but not in FMRP cKO. CTRL: 13.64 ± 0.88 v. CTRL+ LTP: 26.43 ± 2.17 v. Fmr1 cKO: 15.59 ± 1.73 v. Fmr1 cKO + LTP: 15.46 ± 2.98 (Mean ± S.E.M. of slices). Two-Way ANOVA with Tukey’s test for Multiple Comparisons, F[1,16] = 17.85, p = 0.0069; n = 70, 32, 76, 65 boutons respectively, 5 slices, 5 animals. In rest of superplots in this figure, individual points representing boutons are color-coded by slice. Large black circle and bar represent the mean ± 95 % confidence interval of boutons. N = slices. E. LTP induction results in significant shift in the distribution of bouton volume. KS test, Z = 1.8, p = 0.002 F. Length of filopodia was not altered by LTP induction but is enhanced by presynaptic FMRP KO. CTRL: 1.96 ± 0.15 v. CTRL + LTP: 1.93 ± 0.24 v. Fmr1 cKO: 2.42 ± 0.09 v. Fmr1 cKO + LTP: 2.45 ± 0.23 (Mean ± S.E.M. of slices). Two-Way ANOVA with Tukey’s test for Multiple Comparisons, Interaction: F[1,16] = 0.026, p = 0.87, Genotype: F[1,16] = 6.42, p = 0.022; n = 77, 61, 72, 62 filopodia respectively, 5 slices, 5 animals. G. Number of filopodia per bouton was not changed. WT: 1.41 ± 0.19 v. WT-LTP: 1.26 ± 0.15 v. cKO: 1.45 ± 0.25 v. cKO-LTP: 0.85 ± 0.10 (Mean ± S.E.M. in slices). Two-Way ANOVA with Tukey’s test for Multiple Comparisons Interaction: F[1,16] = 1.5, p = 0.24, n = 101, 58, 76, 75 boutons respectively, 5 slices, 5 animals. H. (Left) Representative EPSPs from extracellular field recording pre- and post-MF-LTP from acute hippocampal slices of Fmr1^fl/fl mice injected with AAV5-CAMKII-mCherry-Cre or AAV5-CAMKII-mCherry to KO presynaptic FMRP. (Right) Summary time-course plot of MF-LTP recordings show MF-LTP was blocked by presynaptic Fmr1 KO. WT: 124.66 ± 8.47 v. cKO 99.17 ± 1.92 (Mean ± S.E.M.). Two-sample t-test, p = 0.006. See also Supp. Figures 3 and 4. n.s. = p>0.05; *p≤0.05; **p≤0.01; ***p≤0.001; ****p≤0.0001.

Figure 7.. PKA and PP2A regulate activity-dependent FMRP granule assembly, protein synthesis, FMRP phosphorylation, and MF-LTP.

A. (Left) Representative tiled image of mouse hippocampal slice expressing FMRP-Halotag fusion protein in GCs. B. Schematic of parameters defining ‘ assembled’ FMRP granules and ‘ disassembled’ granules. C. Ratio of ‘ assembled’ to ‘ disassembled’ granules was reduced by electrical MF-LTP or chemical LTP induction with forskolin (50 μM, 10 min). CTRL: 0.197 ± 0.05 v. LTP: 0.891 ± 0.3 v. FSK: 0.894 ± 0.3 (Mean ± S.E.M.); One-Way ANOVA with Tukey’s test for Multiple Comparisons, F[2,42] = 3.828, p = 0.0297; n = 18, 13, 11 slices respectively D. Representative images of MFBs expressing Halo-FMRP. High intensity large ‘ assembled’ granules in MFBs are indicated with white arrowheads and low intensity, diffuse ‘ disassembled’ granules are indicated with white arrows. Insets are granules indicated with yellow arrows. Green dotted line represents contour of ‘ assembled’ granule area quantified in panel F. E. MF-LTP induction with electrical stimulation (3X 125 pulses, 25 Hz) or forskolin (50 μM, 10 min) reduced average size of ‘ assembled’ granules after 10 minutes. CTRL: 1.00 ± 0.04 v. LTP: 0.77 ± 0.06 v. FSK: 0.84 ± 0.04 (Mean ± S.E.M.); One-Way ANOVA with Tukey’s test for Multiple Comparisons, F[2,42] = 6.65, p = 0.003; n = 18, 13, 11 slices respectively. F. (Top) Representative images of puromycin labeling (PMY) in slices from WT mice treated with PKA activator, forskolin (fSk, 50 μM) or PP2A inhibitor okadaic acid (OKA, 25 nM) or PKA inhibitor PKI (1 μM) Dotted lines indicate the MF tract (stratum lucidum, SL) labeled with marker ZnT3. LTP was induced with electrical stimulation (3X 125 pulses, 25 Hz). G. Protein synthesis in the MF induced tract was increased by PKA activation and electrical LTP induction. This increase in protein synthesis was blocked by inhibition of PP2A and PKI. CTRL: 1.0 ± 0.03 v. FSK: 1.88 ± 0.06 v. LTP: 1.54 ± 0.08 v. OKA + LTP: 1.00 ± 0.06 v. PKI + LTP: 1.01 ± 0.06 (Mean ± S.E.M. of slices, normalized to Control); One-Way ANOVA with Tukey’s test for Multiple Comparisons, F [4,45] = 45.29, p < 0.0001. n = 10 slices, 4 animals. H. (Left) Representative images of hippocampal slice stained for p-FMRP-(Ser499). (right) Quantification of pFMRP intensity reveals phosphorylation of FMRP was decreased by PKA activation in a PP2A-dependent manner. CTRL: 1.00 ± 0.10 v. FSK: 0.43 ± 0.08 v. OKA + FSK: 0.95 ± 0.10 (Mean ± S.E.M. of slices); One-Way ANOVA with Tukey’s test for Multiple Comparisons, F [2,18] = 11.47, p = 0.0006. n = 7 slices, 4 animals. I. (Top) Representative EPSPs from extracellular field recording pre- and post-MF-LTP from acute hippocampal slices. (Bottom) Summary time-course plot indicates MF-LTP was blocked by PP2A inhibition by OKA (25 nM). Control: 1.45 ± 0.08 v. +OKA: 1.17 ± 0.05 (Mean ± S.E.M. of last 20 min); Unpaired t-test, p = 0.015, 7 slices, 5 animals; 6 slices, 5 animals respectively. N=slices. See also Supp. Figure 5 n.s. = p>0.05; *p≤0.05; **p≤0.01; ***p≤0.001; ****p≤0.0001.

Figure 8.. Enriched environment alters structural properties and local protein synthesis of MFs

A. Representative 3D reconstructions of giant MF boutons. B. Quantification of MF bouton volume from 3D reconstructions reveals EE changed the overall volume of MF boutons. HC: 12.24 ± 0.66 v. EE: 15.16 ± 1.12 (Mean ± S.E.M. of slices); Two sample t-test, p = 0.04; n = 97, 158 boutons respectively, 6 slices, 5 animals. For all Superplots in this figure, individual points representing boutons are color-coded by slice. Large black circle and bar represent the mean ± 95 % confidence interval of boutons. C. EE did not alter the distribution of MF bouton volume. KS test, Z = 0.67, p = 0.79 D. EE significantly changes the length of filopodia. HC: 1.75 ± 0.11 v. EE: 2.95 ± 0.29 (Mean ± S.E.M. of slices); Two sample t-test, p = 0.002; n = 42, 80 filopodia, 7, 6 slices respectively, 5 animals. E. EE had no effect on the number of filopodia/bouton. HC: 0.85 ± 0.14 v. EE: 0.88 ± 0.08; Mann Whitney, p =0.42; n = 47, 92 filopodia, 7, 6 slices respectively, 5 animals. F. (Left) Representative images of MF boutons expressing Halo-actin virus from mice in HC or 2 weeks EE. (Right, top) Timeline of Halo-actin pulse-chase labeling. (Bottom, right) Halo-actin signal measured in MF boutons was reduced in EE mice compared to HC controls. HC: 1.00 ± 0.10 v. EE: 0.77 ± 0.05 (Mean ± S.E.M. of slices); Two sample t-test, p = 0.04. (Bottom right) Halo-actin signal measured in filopodia of MF boutons was increased in EE mice. HC: 1.00 ± 0.05 v. EE: 1.22 ± 0.08; Mann Whitney, U = 24, p = 0.030. n = 81, 91 boutons/ 100,121 filopodia, 10,11 slices respectively, 5 animals. Points representing individual boutons are normalized to mean of Control. G. Enriched environment did not occlude increased puromycin labeling (PMY) of newly synthesized proteins in MFs of presynaptic cKO of Fmr1 mice. WT HC: 1.00 ± 0.03 v. Fmr1 cKO HC: 1.90 ± 0.14 v WT EE: 1.56 ± 0.07 v. Fmr1 cKO EE: 1.96 ± 0.13 (Mean ± S.E.M.); Two-Way ANOVA with Tukey’s test for Multiple Comparisons, Interaction: F[1,59] = 6.312, p = 0.015. n = 18,14, 15, 16 slices respectively. Points represent individual slices and are normalized to mean of WT HC. N = slices. See also Supp. Figure 6. n.s. = p>0.05; *p≤0.05; **p≤0.01; ***p≤0.001; ****p≤0.0001.