Design and characterizations of Rab FRET sensors.

a, Schematic diagrams illustrating the FRET sensor design for Rab proteins. RBD: Rab binding domain; GDP: guanosine diphosphate; GTP: guanosine-5’-triphosphate. b, Representative fluorescence lifetime images of HEK 293T cells transfected with Rab wild type (WT), dominant negative (DN, Rab4 [S27N], Rab5 [S34N], Rab7 [T22N], Rab8 [S22N] and Rab10 [T23N]) and constitutively active (CA, Rab4 [Q72L], Rab5 [Q79L], Rab7 [Q67L], Rab8 [Q67L] and Rab10 [Q68L]) sensors. FRET donor/acceptor pair is mEGFP/mCherry for Rab4, 5, 7 and 8; for Rab10, FRET donor/acceptor pair is mTurquoise2/mVenus. Warmer color indicates shorter lifetime and higher activity. Scale bars represent 5 µm. c, Quantification of binding fraction for experiments in b. Data represent mean ± SEM. N=5-32. One-way ANOVA followed by Bonferroni’s multiple comparison tests were performed (* p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001). d-h, Representative fluorescence lifetime images of individual Rab wild type sensor cotransfected with the corresponding GAP or GEF cDNAs. Scale bars represent 5 µm. i, Quantification of binding fraction change for experiments in d-h. N=5-20. Data represent mean ± SEM. j, Activation of Rab sensors by NMDA application in rat hippocampal CA1 pyramidal neurons. Upper: representative fluorescence lifetime images of Rab sensor-expressing neurons in response to bath application of 15 µM NMDA for 2 min. Scale bar is 1 µm. Lower: quantification of binding fraction change for upper lane experiments. Data represent mean ± SEM. N=4-11 for each group.

Spatiotemporal dynamics of Rab10 inactivation during sLTP in single spines.

a, Representative donor fluorescence intensity image (gray) and fluorescence lifetime (FLIM, colorful) images of Rab10 inactivation during sLTP induced by two-photon glutamate uncaging. Arrowheads indicate the stimulated spine. The colder color indicates longer lifetime and lower Rab10 activity. Scale bar represents 1 µm. b, Averaged time courses of Rab10 inactivation measured as binding fraction changes between mTurquoise2-Rab10 and mVenus-Rim1 [20-277]-mVenus in the stimulated spine (red), adjacent spine (black) and dendrite (blue). Grey rectangle bar indicates glutamate uncaging (0.5 Hz, 60 s). Data are presented in mean ± SEM. N=49/42 (spine/neuron), 49/42 (spine/neuron) and 49/42 (dendrite/neuron) for the stimulated spine, adjacent spine and dendrite, respectively. c,d, Quantification of Rab10 binding fraction changes in the transient phase (c, averaged over 1.3-4 min) and sustained phase (d, averaged over 19-31 min) in the stimulated spine (stim), adjacent spine (adj) and dendrite (dend) for the same experiments as b. Data represent mean ± SEM. One-way ANOVA followed by Bonferroni’s multiple comparison tests were performed for the stimulated spine, adjacent spine and dendrite (**** p<0.0001). Effects of pharmacological agents on Rab10 inactivation in the stimulated spines are also presented. All pharmacological inhibition experiments were paired with controls from the same batch of slices. Data represent mean ± SEM. Student’s t-tests were performed (n.s., not significant, ** p<0.01, *** p<0.001, **** p<0.0001). N=15/14, 13/11, 8/7, 15/12, 9/8 and 11/8 (spine/neuron) for Ctrl, AP5, scrambled, CN21, U0124 and U0126, respectively. When mTurquoise2-Rab10 was paired with a false acceptor (False), mVenus-Rabenosyn5 [439-503]-mVenus, little activity change was observed. Data represent mean ± SEM. Student’s t-tests were performed (**** p<0.0001). N=7/7 and 14/9 (spine/neuron) for Ctrl and False, respectively. e, Averaged time courses of spine volume change in the same experiment as b. Data represent mean ± SEM. f,g, Quantification of spine volume changes in the transient phase (f, averaged over 1.3-4 min) and sustained phase (g, averaged over 19-31 min) for the same experiments as c and d. Data represent mean ± SEM. Student’s t-tests were used for all groups (n.s., not significant, * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001).

Spatiotemporal dynamics of Rab4 activation during sLTP induced at single spines.

a, Representative donor fluorescence intensity image (gray) and fluorescence lifetime (FLIM, colorful) images of Rab4 activation during sLTP induced by two-photon glutamate uncaging. Arrowheads indicate the stimulated spine. The warmer color indicates shorter lifetime and higher Rab4 activity. Scale bar represents 1 µm. b, Averaged time courses of Rab4 activation measured as binding fraction changes between mEGFP-Rab4 and mCherry-Rabenosyn5 [439-503]-mCherry in the stimulated spine (red), adjacent spine (black) and dendrite (blue). Grey rectangle bar indicates glutamate uncaging (0.5 Hz, 60 s). Data are presented in mean ± SEM. N=42/34 (spine/neuron), 42/34 (spine/neuron), and 42/34 (dendrite/neuron) for the stimulated spine, adjacent spine and dendrite, respectively. c,d, Quantification of Rab4 binding fraction changes in the transient phase (c, averaged over 1.3-4 min) and sustained phase (d, averaged over 19-31 min) in the stimulated spine (stim), adjacent spine (adj) and dendrite (dend) for the same experiments as b. Data represent mean ± SEM. One-way ANOVA followed by Bonferroni’s multiple comparison tests were used for the stimulated spine, adjacent spine and dendrite (** p<0.01, *** p<0.001, **** p<0.0001). Effects of pharmacological agents on Rab4 activation in the stimulated spines are also presented. All pharmacological inhibition experiments were paired with controls from the same batch of slices. Data represent mean ± SEM. Student’s t-tests were performed (n.s., not significant, * p<0.05, ** p<0.01, **** p<0.0001). N=8/6, 12/9, 9/8, 13/11, 10/8 and 9/8 (spine/neuron) for Ctrl, AP5, scrambled, CN21, U0124 and U0126, respectively. When mEGFP-Rab4 was paired with a false acceptor (False), mCherry-Rim1 [20-227]-mCherry, little activity change was observed in the stimulated spines. Data represent mean ± SEM. Student’s t-tests were performed (n.s., not significant, **** p<0.0001). N=6/5 and 12/9 (spine/neuron) for Ctrl and False, respectively. e, Averaged time courses of spine volume change in the same experiments as b. Data represent mean ± SEM. f,g, Quantification of spine volume changes in the transient phase (f, averaged over 1.3-4 min) and sustained phase (g, averaged over 19-31 min) for the same experiments as c and d. Data represent mean ± SEM. Student’s t-tests were used for all groups (n.s., not significant, * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001).

Rab4 positively regulates transient phase of structural LTP and Rab10 negatively regulates structural and electrophysiological LTP.

a,b, Representative fluorescence images of spine volume change in the stimulated spines after manipulations of Rab10 and Rab4 signaling. Asterisks indicate the stimulated spines. Scale bars represent 1 µm. a, Left: rat CA1 pyramidal neurons were transfected with mEGFP and scrambled shRNA (Ctrl shRNA); mEGFP and shRNA against Rab10 (Rab10 shRNA); mEGFP, shRNA against Rab10, and mCherry-shRNA-resistant Rab10 (Rab10 rescue). Right: rat CA1 pyramidal neurons were transfected with mEGFP and scrambled shRNA (Ctrl shRNA); mEGFP and shRNAs against Rab4a and Rab4b (Rab4a/4b shRNA); mEGFP, shRNAs against Rab4a and Rab4b, and mCherry-shRNA-resistant Rab4a (Rab4a rescue). b, Left: rat CA1 pyramidal neurons were transfected with mEGFP (Ctrl); mEGFP and mCherry-Rab10 [T23N] (Rab10 DN); mEGFP and mCherry-Rab10 [Q68L] (Rab10 CA). Right: rat CA1 pyramidal neurons were transfected with mEGFP (Ctrl); mEGFP and mCherry-Rab4a [S27N] (Rab4a DN); mEGFP and mCherry-Rab4a [Q72L] (Rab4a CA). c,d, Averaged time course of spine volume change for experiments in a and b. Fluorescence intensity of mEGFP was used to measure the spine volume change. Data represent means ± SEM. All experiments were paired with the same day controls from the same batch of slices. c, Left: N=22/20, 22/20 and 16/13 (spine/neuron) for Ctrl shRNA (black), Rab10 shRNA (blue) and Rab10 rescue (red), respectively. Right: N=17/17, 25/21 and 10/8 (spine/neuron) for Ctrl shRNA (black), Rab4a/4b shRNA (blue) and Rab4a rescue (red), respectively. d, Left: N=22/18, 17/15 and 19/14 (spine/neuron) for Ctrl (black), Rab10 DN (blue) and Rab10 CA (red), respectively. Right: N=21/15, 19/18 and 16/11 (spine/neuron) for Ctrl (black), Rab4a DN (blue) and Rab4a CA (red), respectively. e,f, Quantitative analysis of the transient volume change (volume change averaged over 1.3-4 min, left) and the sustained volume change (volume change averaged over 20-35 min, right) for c in e and for d in f. Data represent means ± SEM. One-way ANOVA followed by Bonferroni’s multiple comparison tests were performed (n.s., not significant, * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001). g, Representative fluorescence images of stimulated spine volume change in CA1 pyramidal neurons from Rab10fl/fl mice transfected with tdTomato-Cre and mEGFP (Rab10 KO), or tdTomato and mEGFP as a control (Ctrl). Asterisks indicate the stimulated spines. Scale bars represent 1 µm. h, Averaged time courses of stimulated spine volume change for experiments in g. Data represent means ± SEM. N=12 for Ctrl (black) and 12 for Rab10 KO (blue). i, Quantification of the transient volume change (left) and the sustained volume change (right) for h. Data represent means ± SEM. Unpaired two-tailed Student’s t-tests were used (* p<0.05). j, Electrophysiological recordings showing average input/output curve of Rab10fl/fl:CaMKIIα-Cre+/- mice (Rab10 KO, blue, n=11) and littermate Rab10fl/fl:CaMKIIα-Cre-/- control mice (Ctrl, black, n=11). Data are mean ± SEM. k, Time course of extracellularly recorded excitatory postsynaptic potential (fEPSP) slope before and after LTP induction in CA1 pyramidal neurons from Rab10fl/fl:CaMKIIα-Cre+/- mice (blue, number of slices/animals = 28/7) and littermate Rab10fl/fl:CaMKIIα-Cre-/- mice (black, n = 30/8). Data are represented as mean ± SEM. Insets (top) are the representative traces of fEPSP before and after stimulation for Cre+ (blue) and Cre-(black) mice. l, Quantification of the average fEPSP slope at 40-60 min for experiments in k. Data are mean ± SEM. Unpaired two-tailed Student’s t-tests were used (* p<0.05).

Rab4 and Rab10 positively and negatively regulate activity-dependent SEP-GluA1 exocytosis in the stimulated spines during sLTP, respectively.

a, Upper panel: schematic for SEP-GluA1 FRAP and two-photon glutamate uncaging experiment. Whole dendrite photobleaching was performed from -2 min to 0 min, followed by single spine glutamate uncaging from 0 min to 1 min (0.5 Hz, 60 s). Lower panel: representative pseudo color images of SEP-GluA1 (green) FRAP after two-photon glutamate uncaging in a single spine of hippocampal CA1 pyramidal neurons coexpressing mCherry (magenta) and scrambled shRNA. White arrowheads indicated the stimulated spine. Scale bar represents 1 µm. b, Averaged time courses of SEP-GluA1 (left) and mCherry (right) fluorescence intensity (F/F0) in the stimulated spine (red), adjacent spine (black) and dendrite (blue) of neurons expressing SEP-GluA1, mCherry, and scrambled shRNA. Black arrows indicate the time points after photobleaching and glutamate uncaging, respectively. Data represent mean ± SEM. N=43/35 (spine/neuron), 43/35 (spine/neuron) and 43/35 (dendrite/neuron) for the stimulated spine, adjacent spine and dendrite, respectively. c, Averaged time courses of SEP-GluA1 FRAP in the stimulated spines of neurons expressing SEP-GluA1, mCherry, and scrambled shRNA (Ctrl shRNA, black, n=43/35); SEP-GluA1, mCherry, and TeTxLC (TeTxLC, grey, n=19/12); SEP-GluA1, mCherry, and shRNAs against Rab4a and Rab4b (Rab4a/4b shRNA, red, n=26/16); SEP-GluA1, mCherry, shRNAs against Rab4a and Rab4b, and shRNA-resistant Rab4a (Rab4a rescue, orange, n=16/12); SEP-GluA1, mCherry, and shRNA against Rab10 (Rab10 shRNA, green, n=23/15); SEP-GluA1, mCherry, shRNA against Rab10, and shRNA-resistant Rab10 (Rab10 rescue, blue, n=22/19). Data represent mean ± SEM. All experiments were paired with the same day controls from the same batch of slices. d, Quantification of SEP-GluA1 fluorescence intensity at 0 min and 2 min in the same experiments as c. Data represent mean ± SEM. One-way ANOVA followed by Bonferroni’s multiple comparison tests were performed (n.s., not significant, * p<0.05, *** p<0.001). e, Averaged time courses of mCherry fluorescence intensity in the stimulated spines of the same neurons as c. Data represent mean ± SEM. f, Quantification of mCherry fluorescence intensity at 0 min and 2 min in the same experiments as c. Data represent mean ± SEM. One-way ANOVA followed by Bonferroni’s multiple comparison tests (n.s., not significant, ** p<0.01, *** p<0.001). Please note that the Ctrl shRNA samples in c-f are the same as those in a-b. g, Proposed model for Rab4 and Rab10 mediated AMPAR trafficking and sLTP. Activation of postsynaptic NMDARs triggers Ca2+ influx (∼ms) and CaMKII activation (∼s), which is relayed by the transient activation of Rab4 (∼min) and persistent inactivation of Rab10 (∼min). Rab4 activation and Rab10 inactivation result in the potentiated AMPAR exocytosis and sLTP induction in single dendritic spines.

mTurquoise2-Rab4 and mEGFP-Rab10 FRET sensors in HEK 293T Cells, all Rab sensor activity changes upon NMDA application, and localization of endogenous Rab10.

(a) Representative fluorescence lifetime images of HEK 293T cells transfected with mTurquoise2-Rab4 sensors. Scale bars represent 5 µm. (b) Representative fluorescence lifetime images of HEK 293T cells transfected with mEGFP-Rab10 sensors. Scale bars represent 5 µm. (c) Binding fraction of mTurquoise2-Rab4 sensors. Data represent mean ± SEM (** p<0.01, one-way ANOVA followed by Bonferroni’s multiple comparison tests). N=8, 8, and 8 from left to right. (d) Binding fraction of mEGFP-Rab10 sensors. Data represent mean ± SEM (**** p<0.0001, one-way ANOVA followed by Bonferroni’s multiple comparison tests). N=32, 32, and 32 from left to right. (e) Averaged time courses of Rab sensor binding fraction change by NMDA application in rat hippocampal CA1 pyramidal neurons. Data represent mean ± SEM. N=4-11 for each group. (f) Schematics of HA-tag knockin into endogenous Rab10 by SLENDR technique. Mouse genomic loci of Rab10 shows the target sites for Cas9, sgRNA and ssODNs. The sgRNA target regions and PAM sequences are labelled with red and orange, respectively. The start codons are marked with magenta. The Cas9 cleavage sites are indicated by black arrowheads. The recombination and control primer sets are in arrows. (g) Validation of SLENDR-mediated 2XHA tag insertion into endogenous Rab10. Left: PCR genotyping in genomic DNA extracted from Neuro 2a cells electroporated with indicated sgRNA and ssODNs. Right: Sanger sequencing demonstrated the knockin of 2XHA tag to the N-terminus of endogenous Rab10. (h) Confocal microscopic images of the hippocampal CA1 region at P37 stained with DAPI (blue), HA tag fused to the N-terminus of endogenous Rab10 (green) and EGFP (magenta). Scale bar is 50 μm. (i) Representative images of the secondary apical dendrites of CA1 pyramidal neurons stained with HA tag fused to the N-terminus of endogenous Rab10 (green) and EGFP (magenta). Scales bar is 1 μm. (j) Representative images of endogenous Rab10 (green, SLENDR-mediated 2XHA tag knockin) and exogenous endosomal markers (magenta) in dendrites of CA1 pyramidal neurons. Exogenously expressed mEGFP-Rab5a, mCherry-Rab11a and mEGFP-Rab7 were used as markers for early endosome, recycling endosome and lysosome, respectively. Scales bars are 1 μm.

Relationship between initial spine volume and basal Rab GTPase activity, spine volume change or activity change during sLTP.

(a and b) Relationship between the initial spine volume and spine volume changes during the transient (a, averaged over 1.3-4 min) or sustained phase (b, averaged over 19-31 min) in neurons expressing Rab10 sensor. N=21/20 (spine/neuron). No significant correlation (p > 0.05) was found. (c) Relationship between spine volume and basal binding fraction (BBF) of Rab10 sensor. N=81/20 (spine/neuron). No significant correlation (p > 0.05) was found. (d and e) Relationship between the initial spine volume and changes in binding fraction of Rab10 sensor during the transient (d, averaged over 1.3-4 min) or sustained phase (e, averaged over 19-31 min) of sLTP in the stimulated spines. N=21/20 (spine/neuron). No significant correlation (p > 0.05) was found. (f and g) Relationship between the initial spine volume and spine volume changes during the transient (f, averaged over 1.3-4 min) or sustained phase (g, averaged over 19-31 min) in neurons expressing Rab4 sensor. N=42/34 (spine/neuron). No significant correlation (p > 0.05) was found. (h) Relationship between spine volume and basal binding fraction (BBF) of Rab4 sensor. N=90/39 (spine/neuron). No significant correlation (p > 0.05) was found. (i and j) Relationship between the initial spine volume and Rab4 activity changes during the transient (i, averaged over 1.3-4 min) or sustained phase (j, averaged over 19-31 min) of sLTP in the stimulated spines. N=42/34 (spine/neuron). No significant correlation (p > 0.05) was found.

Binding fraction changes of mTurquoise2-Rab10 paired with false acceptor during sLTP.

(a) Averaged time course of changes in binding fraction of Rab10 sensor (Ctrl, black) in the stimulated spines during sLTP. When mTurquoise2-Rab10 was paired with a false acceptor (False, red), mVenus-Rabenosyn5 [439-503]-mVenus, little activity change was observed. Data represent mean ± SEM. N=7/7 and 14/9 (spine/neuron) for Ctrl and False, respectively. (b) Quantification of changes in binding fraction during the transient phase (1.3-4 min) and sustained phase (19-31 min) for the same experiments as in (a). Data represent mean ± SEM (**** p<0.0001, Student’s t-tests). (c) Averaged time courses of changes in spine volume for the same experiments as in (a). Data represent mean ± SEM. (d) Quantification of changes in spine volume during the transient phase (1.3-4 min) and sustained phase (19-31 min) for the same experiments as in (a). Data represent mean ± SEM (n.s., not significant, Student’s t-tests). Please note the quantification data in (b) and (d) are also presented in Fig. 2c,d,f,g.

Inactivation of Rab10 during sLTP induced at near physiological temperature.

(a) Averaged time courses of changes in binding fraction of Rab10 sensor in the stimulated spines during sLTP at 25-27°C (black) and 33-35°C (red). Data represent mean ± SEM. N=49/42 and 14/12 (spine/neuron) for 25-27°C and 33-35°C, respectively. (b) Quantification of changes in binding fraction in the transient phase (1.3-4 min) and sustained phase (19-31 min) for the same experiments as in (a). Data represent mean ± SEM (n.s., not significant, Student’s t-tests). (c) Averaged time courses of changes in spine volume for the same experiments as in (a). Data represent mean ± SEM. (d) Quantification of changes in spine volume during the transient phase (1.3-4 min) and sustained phase (19-31 min) for the same experiments as in (a). Data represent mean ± SEM (n.s., not significant, Student’s t-tests). Please note that the 25-27°C (black) samples in (a-d) are the same as those in Fig. 2b, e.

Properties of basal binding fraction and binding fraction change in the stimulated spines for Rab10 and Rab4 sensors.

(a) Relationship between normalized donor intensity and spine binding fraction (BF) for Rab10 sensor. N=49/42 (spine/neuron). No significant correlation (p > 0.05) was found. (b) Frequency distribution of basal binding fraction in the stimulated spines (BFspine) for Rab10 sensor. N=49/42 (spine/neuron). (c) Frequency distribution of basal binding fraction difference between the spines and dendrite (BFspine - BFdendrite) for Rab10 sensor. N=49/42 (spine/neuron). (d) Averaged time course of binding fraction (BF) subtracted by basal binding fraction (BBF) of dendrite for the stimulated spines and dendrite in Rab10 sensor expressing neurons. For spines with a higher BF (high, blue) than the dendrite (high, gray), Rab10 activity decreased to a level lower than that of the dendrite during sLTP. Data represent mean ± SEM. N=21/18 (spine/neuron). For spines with a lower BF (low, red) than the dendrite (low, black), Rab10 activity still decreased during sLTP. Data represent mean ± SEM. N=28/24 (spine/neuron). (e) Relationship between normalized donor intensity and spine binding fraction for Rab4 sensor. N=42/34 (spine/neuron). No significant correlation (p > 0.05) was found. (f) Frequency distribution of basal binding fraction in the stimulated spines for Rab4 sensor. N=42/34 (spine/neuron). (g) Frequency distribution of basal binding fraction difference between the spines and dendrite for Rab4 sensor. N=42/34 (spine/neuron). (h) Averaged time course of binding fraction subtracted by basal binding fraction of dendrite for the stimulated spines and dendrite in Rab4 sensor expressing neurons. Data represent mean ± SEM. N=42/34 (spine/neuron).

Intensity and mean intensity of Rab10 and Rab4 sensor donors in the stimulated spines and dendrites during sLTP.

(a and b) Averaged time courses of Rab10 donor intensity (a) and mean intensity (b) in the stimulated spines (red) and dendrites (black) during sLTP. Data represent means ± SEM. N=49/42 (spine/neuron) and 49/42 (dendrite/neuron) for the stimulated spine and dendrite, respectively. (c) Quantification of Rab10 donor mean intensity in the baseline (averaged over -5-0 min), transient phase (averaged over 1.3-4 min) and sustained phase (averaged over 19-31 min) for the experiments in (b). Data represent means ± SEM. Red color statistics indicate comparisons with baseline in the stimulated spines, and black color statistics indicate comparisons with baseline in the dendrites (n.s., not significant, two-way ANOVA). Blue color statistics indicate comparisons between the stimulated spines and dendrites (n.s., not significant, Student’s t-tests). (d and e) Averaged time courses of Rab4 donor intensity (d) and mean intensity (e) in the stimulated spine (red) and dendrite (black) during sLTP. N=42/34 (spine/neuron) and 42/34 (dendrite/neuron) for the stimulated spine and dendrite, respectively. (f) Quantification of Rab4 donor mean intensity in the baseline (averaged over -5-0 min), transient phase (averaged over 1.3-4 min) and sustained phase (averaged over 19-31 min) for the experiments in (e). Data represent means ± SEM. Red color statistics indicate comparisons with baseline in the stimulated spines and black color statistics indicate comparisons with baseline in the dendrites (n.s., not significant, **** p<0.0001, two-way ANOVA). Blue color statistics indicate comparisons between the stimulated spines and dendrites (n.s., not significant, * p<0.05, ** p<0.01, Student’s t-tests).

Rab10 inactivation under manipulations of putative upstream signaling pathways.

(a and b) Averaged time courses for changes in binding fraction of Rab10 sensor (a) and volume (b) of the stimulated spines during sLTP. Black and red curves represent control (Ctrl) and AP5 (50 µM), respectively. Data represent mean ± SEM. N=15/14 and 13/11 (spine/neuron) for Ctrl and AP5, respectively. (c and d) Averaged time courses for changes in binding fraction of Rab10 sensor (c) and volume (d) of the stimulated spines during sLTP. Black and blue curves represent scrambled peptide control (10 µM) and CN21 peptide (10 µM), respectively. Data represent mean ± SEM. N=8/7 and 15/12 (spine/neuron) for scrambled peptide and CN21 peptide, respectively. (e and f) Averaged time courses for changes in binding fraction of Rab10 sensor (e) and volume (f) of the stimulated spines during sLTP. Black and green curves represent U0124 control (20 µM) and U0126 (20 µM), respectively. Data represent mean ± SEM. N=9/8 and 11/8 (spine/neuron) for U0124 and U0126, respectively. For all pharmacological inhibition experiments, hippocampal slices were incubated in the indicated drugs for 30 min before experiments. All experiments were paired with controls from neurons in the same batch of slices.

Changes in the binding fraction of mEGFP-Rab4 paired with false acceptor during sLTP.

(a) Averaged time course of changes in binding fraction of Rab4 sensor (Ctrl, black) in the stimulated spines during sLTP. When mEGFP-Rab4 was paired with a false acceptor (False, red), mCherry-Rim1 [20-227]-mCherry, little activity change was observed. Data represent mean ± SEM. N=6/5 (spine/neuron) for Ctrl, and 12/9 for False. (b) Quantification of changes in binding fraction in the transient phase (1.3-4 min) and sustained phase (19-31 min) for the same experiment as in (a). Data represent mean ± SEM. Stars denote statistical significance (n.s., not significant, **** p<0.0001, Student’s t-tests). (c) Averaged time courses of spine volume changes for the same experiments as in (a). (d) Quantification of changes in spine volume in the transient phase (1.3-4 min) and sustained phase (19-31 min) for the same experiments as in (a). Data represent mean ± SEM (n.s., not significant, Student’s t-tests). Please note the quantification data in (b) and (d) are also presented in Fig. 3c,d,f,g.

Activation of Rab4 during sLTP induction at near physiological temperature.

(a) Averaged time courses for changes in binding fraction of Rab4 sensor in the stimulated spines during sLTP at 25-27°C (black) and 33-35°C (red). Data represent mean ± SEM. N=42/34 and 16/13 (spine/neuron) for 25-27°C and 33-35°C, respectively. (b) Quantification of binding fraction changes in the transient phase (1.3-4 min) and sustained phase (19-31 min) for the same experiments as in (a). Data represent mean ± SEM (n.s., not significant, Student’s t-tests). (c) Averaged time courses of changes in spine volume for the same experiments as in (a). Data represent mean ± SEM. (d) Quantification of spine volume changes during the transient phase (1.3-4 min) and sustained phase (19-31 min) for the same experiments as in (a). Data represent mean ± SEM (n.s., not significant, Student’s t-tests). Please note that the 25-27°C (black) samples in (a-d) are the same as those in Fig. 3b, e.

Pharmacological evaluations of upstream signaling pathways mediating Rab4 activation.

(a and b) Averaged time courses for changes in binding fraction (a) of Rab4 sensor and volume (b) of the stimulated spines during sLTP. Black and red curves represent control (Ctrl) and AP5 (50 µM), respectively. Data represent mean ± SEM. N=8/6 and 12/9 (spine/neuron) for Ctrl and AP5, respectively. (c and d) Averaged time courses for changes in binding fraction (c) of Rab4 sensor and volume (d) of the stimulated spines during sLTP. Black and blue curves represent scrambled peptide control (10 µM) and CN21 peptide (10 µM), respectively. Data represent mean ± SEM. N=9/8 and 13/11 (spine/neuron) for scrambled peptide and CN21 peptide, respectively. (e and f) Averaged time courses for changes in binding fraction of Rab4 sensor (e) and volume (f) of the stimulated spines during sLTP. Black and green curves represent U0124 control (20 µM) and U0126 (20 µM), respectively. Data represent mean ± SEM. N=10/8 and 9/8 (spine/neuron) for U0124 and U0126, respectively. For all pharmacology experiments, hippocampal slices were incubated with the indicated drugs for 30 min before experiments. All experiments were paired with control neurons from the same batch of slices.

Validation of Rab GTPase shRNA and shRNA-resistant Rab GTPases, and effects of Rab knockdown on spine size and density.

(a) Schematic of dual-luciferase reporter assay. (b) Validation of knockdown by specific Rab4a, Rab4b and Rab10 shRNAs using dual-luciferase reporter assay. Individual psiCHECK-2-Rab GTPase was cotransfected into HEK 293T cells with scrambled shRNA (Ctrl shRNA), hRluc shRNA (Positive control) or individual Rab GTPase shRNA. Data represent mean ± SEM (**** p<0.0001, one-way ANOVA followed by Bonferroni’s multiple comparison tests). N=8, 8, 8, 8, 8, 8, 10, 10, 10 wells from left to right. (c) Verification of shRNA-resistant Rab4a. HEK 293T cells were transfected with psiCHECK-2-shRNA-resistant Rab4a and scrambled shRNA (Ctrl shRNA), psiCHECK-2-shRNA-resistant Rab4a and hRluc shRNA (Positive control), psiCHECK-2 Rab4a and Rab4a shRNA (Rab4a+shRNA), or psiCHECK-2-shRNA-resistant Rab4a and Rab4a shRNA (Res-Rab4a+shRNA). Data represent mean ± SEM (n.s., not significant, **** p<0.0001, one-way ANOVA followed by Bonferroni’s multiple comparison tests). N=4, 4, 4, 4 wells from left to right. (d) Verification of shRNA-resistant Rab10. HEK 293T cells were transfected with psiCHECK-2-shRNA-resistant Rab10 and scrambled shRNA (Ctrl shRNA), psiCHECK-2-shRNA-resistant Rab10 and hRluc shRNA (Positive control), psiCHECK-2 Rab10 and Rab10 shRNA (Rab10+shRNA), or psiCHECK-2-shRNA-resistant Rab10 and Rab10 shRNA (Res-Rab10+shRNA). Data represent mean ± SEM (n.s., not significant, **** p<0.0001, one-way ANOVA followed by Bonferroni’s multiple comparison tests). N=5, 5, 5, 5 wells from left to right. (e and f) Validation of shRNA by western blot. Western blot of total protein extracts from cortical neurons cultured 15-17 days in vitro infected with lentiviral vectors expressing mEGFP plus either scrambled control shRNA or Rab shRNA. Data shown are representative of two independent experiments (n=4). (g) Quantification of the western blot experiments in (e) and (f). Data represent mean ± SEM (*** p<0.001, **** p<0.0001, Student’s t-tests). (h) Knockdown of Rab4a/4b or Rab10 had no effect on basal spine size. Data represent mean ± SEM (n.s., not significant, Student’s t-tests). N=20, 24, 20 and 20 (neurons) from left to right. (i) Knockdown of Rab4a/4b had no effect on spine density while knockdown of Rab10 enhanced spine density. Data represent mean ± SEM (n.s., not significant, * p<0.05, Student’s t-tests). N=20, 24, 20 and 20 (neurons) from left to right.

Deletion of Rab10 has no effect on CA3-CA1 synaptic transmission in the hippocampus.

(a and b) Whole cell recordings of AMPAR (a) and NMDAR (b) mediated EPSCs in Rab10fl/fl:CaMKIIα-Cre+/- (Rab10 KO) mice and littermate Rab10fl/fl:CaMKIIα-Cre-/- control (Ctrl) mice. (a) Amplitude quantification of AMPAR EPSCs (at -70mV) in Ctrl (black, n=48/5 cells/animals) and Rab10 KO (blue, n=45/5 cells/animals) mice. Insets are representative evoked responses (black for Ctrl and blue for Rab10 KO). (b) Amplitude quantification of NMDAR EPSCs (at +40mV) for the same experiments in (a). To avoid contamination with residual AMPAR currents, the amplitude of NMDAR EPSCs was calculated by measuring the response amplitude at 50 ms after the peak. Insets are representative evoked responses. (c) AMPAR/NMDAR EPSC ratio for Ctrl (black) and Rab10 KO (blue) mice. (d) Input-output relationship for Ctrl (black) and Rab10 KO (blue) mice. No significant difference was detected in any of the results using unpaired t-tests.

Rab4 and Rab10 regulate activity-dependent GluA1 exocytosis in the stimulated spines during sLTP.

(a-e) Representative images of SEP-GluA1 (green) FRAP after two-photon glutamate uncaging in the stimulated spines of hippocampal neurons coexpressing mCherry and TeTxLC (a); mCherry and Rab4a and Rab4b shRNAs (b); mCherry, Rab4a and Rab4b shRNAs and shRNA-resistant Rab4a (c); mCherry and Rab10 shRNA (d) or mCherry, Rab10 shRNA and shRNA-resistant Rab10 (e). White arrowheads indicate the stimulated spine. Scale bar represents 1 µm.

Fluorescence intensity of SEP-GluA1 and mCherry in adjacent spines and dendrites.

(a and c) Averaged time courses of SEP-GluA1 (a) and mCherry (c) fluorescence intensity for adjacent spines in the same experiments as in Fig. 5c. Data represent mean ± SEM. N=43/35, 19/12, 26/16, 16/12, 23/15 and 22/19 (spine/neuron) for Ctrl, TeTxLC, Rab4a/4b shRNA, Rab4a rescue, Rab10 shRNA and Rab10 rescue, respectively. (b and d) Quantification of SEP-GluA1 (b) and mCherry (d) fluorescence intensity for adjacent spines at 0 min and 2 min. Data represent mean ± SEM (n.s., not significant, one-way ANOVA followed by Bonferroni’s multiple comparison tests). N=43/35, 19/12, 26/16, 16/12, 23/15 and 22/19 (spine/neuron) for Ctrl, TeTxLC, Rab4a/4b shRNA, Rab4a rescue, Rab10 shRNA and Rab10 rescue, respectively. (e and g) Averaged time courses of SEP-GluA1 (e) and mCherry (g) fluorescence intensity for dendrites in the same experiments as in Fig. 5c. Data represent mean ± SEM. N=43/35, 19/12, 26/16, 16/12, 23/15 and 22/19 (dendrite/neuron) for Ctrl, TeTxLC, Rab4a/4b shRNA, Rab4a rescue, Rab10 shRNA and Rab10 rescue, respectively. (f and h) Quantification of SEP-GluA1 (f) and mCherry (h) fluorescence intensity for dendrites at 0 min and 2 min. Data represent mean ± SEM (n.s., not significant, one-way ANOVA followed by Bonferroni’s multiple comparison tests). N=43/35, 19/12, 26/16, 16/12, 23/15 and 22/19 (spine/neuron) for Ctrl, TeTxLC, Rab4a/4b shRNA, Rab4a rescue, Rab10 shRNA and Rab10 rescue, respectively. (i) SEP-GluA1 FRAP after subtraction of the surface area increase in the stimulated spines for experiments in Fig. 5c. N=43/35, 19/12, 26/16, 16/12, 23/15 and 22/19 (spine/neuron) for Ctrl, TeTxLC, Rab4a/4b shRNA, Rab4a rescue, Rab10 shRNA and Rab10 rescue, respectively.