KIF13A drives AMPA receptor synaptic delivery for long-term potentiation via endosomal remodeling - PubMed (original) (raw)

Figure 1.

Figure 1.

KIF13A is required for LTP. (A) Quantification of KIF13A mRNA levels from dissociated hippocampal neurons infected with the shKIF13A lentivirus (red) or control lentivirus (lacking the shRNA sequence; black) relative to the uninfected condition. Cultures were lysed after 15 DIV and 7 d after infection. Bars show mean ± SEM, together with individual values from each experiment. n = 9 independent experiments; P = 0.008, Wilcoxon test. (B) Left: Representative Western blot result from protein lysates from conditions described in A. Apparent protein size was lower than expected, perhaps as a consequence of proteolytic degradation. Right: Quantification of KIF13A protein levels relative to uninfected condition. Bars show mean ± SEM, together with individual values from each experiment; n = 4 independent experiments. (C) Left: Time-course of the averaged AMPAR-mediated evoked synaptic currents (mean ± SEM per condition). LTP was induced (pairing protocol of 300 pulses at 3 Hz) in organotypic hippocampal slices (7‒10 DIV) infected in the CA1 layer with a lentivirus expressing shKIF13A and analyzed at 7‒10 d after infection. Uninfected (black) and infected (red) cells were recorded. Amplitude of the synaptic responses is normalized to a 3-min baseline. Right: Average of AMPAR-mediated responses from the last 5 min of the recording and normalized to the baseline. Bar plots show mean ± SEM, together with individual values for each experiment. Left bars (LTP, paired) correspond to the stimulation pathway in which postsynaptic depolarization (0 mV) was paired to presynaptic stimulation (3 Hz, 300 pulses). Right (shaded) bars (control, unpaired) correspond to the pathway that was not stimulated during depolarization. Inset: Representative traces from uninfected (black lines) and shKIF13A (red lines), averaged from baseline (dark lines) or from the last 5 min of the recording (light lines). Scale bars: vertical, 10 pA; horizontal, 10 ms. n = 10 uninfected cells, and n = 11 shKIF13A-expressing cells. Uninfected neurons are significantly potentiated from baseline (P = 0.01, Wilcoxon test). Uninfected and shKIF13A-expressing neurons were significantly different (P = 0.03, Mann–Whitney test). (D) Basal synaptic responses mediated by AMPARs (left) and NMDARs (right) from adjacent uninfected and shKIF13A-infected CA1 pyramidal neurons at −60 mV and at +40 mV (at 65 ms), respectively. Gray symbols represent each individual pair while red symbols show the mean ± SEM of the measured currents. n = 26 cell pairs for AMPAR currents and n = 19 cell pairs for NMDAR currents; P value according to the Wilcoxon test; n.s., not significantly different. Insets: Representative traces for uninfected (black lines) and shKIF13A (red lines) cells; scale bars: vertical, 50 and 10 pA respectively; horizontal, 10 ms. (E) RT-qPCR quantification of KIF13B mRNA levels from dissociated hippocampal neurons infected with shKIF13B (burgundy) or control (lacking shRNA sequence; black) lentivirus, relative to uninfected condition. Bars show mean ± SEM, together with individual values for each experiment. n = 6 independent experiments; P = 0.03, Wilcoxon test. (F) Similar to C, but with shKIF13B-expressing (burgundy) neurons. Inset: scale bars: vertical, 10 pA; horizontal, 20 ms. n = 9 uninfected cells and n = 11 shKIF13B-expressing cells for the LTP (paired) pathway, and n = 5 uninfected cells and n = 8 shKIF13B-expressing cells for the control (Unpaired) pathway. Uninfected and shKIF13B-expressing neurons are significantly different from baseline (P = 0.02 and P = 0.03, respectively) according to the Wilcoxon test.

Figure S1.

Figure S1.

Basal physiological properties of neurons expressing shKIF13A and KIF13A-ST. (A–C) Capacitance (A), holding current (B), and input resistance (C) were measured from whole-cell patch-clamp experiments of uninfected (black) or shKIF13A- (red) and KIF13-ST–expressing (green) CA1 pyramidal neurons. Whole-cell capacitance was calculated from the exponential decay of the current in response to a square voltage step. Input resistance is calculated from Ohm's law using the same voltage step. Bars represent the mean ± SEM, together with individual values for each cell. n = 39, 26, or 13 cells for uninfected, shKIF13A, or KIF13-ST neurons, respectively.

Figure S2.

Figure S2.

Specificity of shKIF13A and shKIF13B knockdown, and effect on axonal polarization. (A) RT-qPCR quantification of KIF13B mRNA levels from dissociated hippocampal neurons infected with the shKIF13A lentivirus (red) or control lentivirus (lacking the shRNA sequence; black) relative to the uninfected condition. Cultures were lysed after 15 DIV and 7 d after infection. Bars show mean ± SEM, together with individual values from each experiment. n, number of independent experiments; not significantly different from uninfected (Wilcoxon test). (B) RT-qPCR quantification of KIF13A mRNA levels from dissociated hippocampal neurons infected with shKIF13B (burgundy) or control (lacking shRNA sequence; black) lentivirus relative to the uninfected condition. Bars show mean ± SEM, together with individual values for each experiment. n = 3 (skKIF13A) and n = 6 (shKIF13B) independent experiments; not significantly different from uninfected (Wilcoxon test). (C) Left: Representative confocal images of 3 DIV dissociated hippocampal neurons infected with a control lentivirus (control; top panels) or lentiviruses expressing shKIF13A (middle panels) and shKIF13B (bottom panels), all expressing mCherry (red, left panels). Tau immunocytochemistry (ICC) was performed to assess neurite length (right panels) in all conditions. Scale bars in right panels are identical to those in left panels. Right: Quantification of axonal polarization was calculated from the ratio between the length of the major neurite compared with the length of the secondary neurite. Bars represent the mean ± SEM, together with individual values for each experiment. shKIF13B neurons (burgundy) show significantly less axonal polarization with respect to control (black) and shKIF13A (red) neurons (P = 0.0001 and P < 0.0001, respectively, Mann–Whitney test). n = 51, 72, or 105 cells for control, shKIF13A- or shKIF13B-expressing neurons, respectively.

Figure 2.

Figure 2.

Activity-dependent transport of AMPARs is mediated by KIF13A. (A) Left: Representative images of hippocampal neuronal cultures after 7 d in vitro, infected either with a control (top row) or the shKIF13A-expressing lentivirus (bottom row; red channel) for 7 d. Cultures were fixed and stained for surface GluA1 (nonpermeabilizing conditions, magenta) and total GluA1 (after permeabilization, cyan). Scale bars, 20 µm. Right: Quantifications of fluorescence intensity, measured separately in the somatic and dendritic compartments for both surface (magenta) and total (cyan) GluA1 channels. Plots show mean fluorescence intensity for the individual experiments and mean ± SEM for each pool of receptors in the different compartments. n = 32 for both control neurons and shKIF13A-expressing neurons from six independent experiments. (B) Left: Representative images for dissociated hippocampal neurons infected with lentivirus, as described for A. Immunocytochemistry was performed after baseline or 15 min of cLTP induction. Scale bars, 2.5 µm. Dashed lines outline neuron morphology (from mask of the mCherry channel). Arrowheads indicate individual spines used for the quantification of both surface (left) and total (right) GluA1 immunostaining. Plots on the right show cumulative probability distributions of fluorescence intensity from individual spines for each condition described above. Gray traces represent spines from neurons infected with a control lentivirus, and red traces show spines from neurons infected with a lentivirus expressing shKIF13A. Dashed lines represent baseline values, while continuous lines represent values after cLTP induction. P values compare cLTP versus baseline condition (black for control and red for shKIF13A), according to the Kolmogorov–Smirnov test (left graph, surface GluA1) or control versus shKIF13A in baseline condition (right graph, total GluA1). n = 1,229 spines (24 neurons) for baseline conditions and n = 1,183 spines (26 neurons) for cLTP conditions, from control neurons. n = 750 spines (25 neurons) for baseline conditions and n = 608 spines (24 neurons) for cLTP conditions from shKIF13A-expressing neurons, analyzed from three independent experiments. (C) Quantifications of fluorescence intensity in the adjacent dendritic shaft from each condition described in B. Plots show mean fluorescence intensity after cLTP normalized to mean baseline intensity for each condition for the individual experiments and mean ± SEM for each pool of receptors (surface GluA1, magenta; total GluA1, cyan). n = 26 control neurons and n = 24 shKIF13A-expressing neurons from three independent experiments. (D) Left: Representative images of spine morphology from mCherry channel of the experiments described in B. Scale bars, 2.5 µm. Right, cumulative probability distribution of fluorescence intensity at spines over the value at the adjacent dendrite from the different conditions described for B. Dashed lines represent baseline values, while continuous lines represent values after cLTP induction. P values compare cLTP versus baseline condition (black for control and red for shKIF13A) according to the Kolmogorov–Smirnov test.

Figure 3.

Figure 3.

KIF13A domain analysis for LTP. (A) Time course (mean of the normalized excitatory postsynaptic currents [EPSCs] ± SEM per condition) of the LTP induction in organotypic hippocampal slices (7‒10 DIV) expressing GFP-KIF13A-ST (green) or mRFP-KIF13A-tail (orange) at 24 h after infection. Top left: Inset shows schematic representation of the GFP-KIF13A-ST and mRFP-KIF13A-tail recombinant proteins with known domains (coiled-coil [C-C] and FHA). Top right: Insets show representative traces for uninfected (black lines), GFP-KIF13A-ST (green lines), and mRFP-KIF13A-tail (orange lines) neurons, averaged from baseline (dark lines) or from the last 5 min of the recording (light lines). Scale bars: vertical, 20 pA; horizontal, 10 ms. Bars show mean ± SEM, together with individual values for each experiment, from the last 5 min of the recording and normalized to the baseline. n = 19, 11, or 9 cells for control, GFP-KIF13A-ST– or mRFP-KIF13A-tail–expressing neurons, respectively, for the LTP (paired) pathway. n = 12, 5, or 5 cells for control, GFP-KIF13A-ST–,or mRFP-KIF13A-tail–expressing neurons, respectively, for the control (unpaired) pathway. Uninfected and KIF13A-tail–expressing neurons are significantly potentiated (P = 0.002 and P = 0.01, respectively) with respect to baseline (Wilcoxon test). P values displayed in the panel represent KIF13A-ST significant difference from uninfected (P = 0.001) and KIF13A-tail (P = 0.002), respectively (Mann–Whitney test). (B and C) Basal synaptic responses mediated by AMPARs (B) and NMDARs (C) from organotypic hippocampal slices (7‒10 DIV) infected with a Sindbis virus expressing GFP-KIF13A-ST. Synaptic responses were recorded from pairs of uninfected and infected neighboring CA1 neurons. Gray circles represent values from each individual pair, while green symbols show the mean ± SEM of the measured currents. n = 15 cell pairs for AMPAR currents and n = 12 cell pairs for NMDAR currents; n.s., not significantly different (Wilcoxon test). Insets: Representative traces for uninfected (black lines) and GFP-KIF13A-ST (green lines) cells; scale bars: vertical, 10 pA; horizontal, 10 ms. (D) Time course (mean of the normalized EPSCs ± SEM per condition) of the LTP induction in organotypic hippocampal slices (7‒9 DIV) transfected with a biolistic approach to allow simultaneous expression of shKIF13A and KIF13A-MS-YFP (pink; KIF13A-MS-rescue) for 7–9 d. Top left: Inset shows schematic representation of the KIF13A-MS-YFP recombinant protein. Top right: Insets show representative traces for uninfected (black lines) and KIF13A-MS-rescue (pink lines) neurons, averaged from baseline (dark lines) or from the last 5 min of the recording (light lines). Scale bars: vertical, 10 pA; horizontal, 10 ms. Bars show mean ± SEM, together with individual values for each experiment, from the last 10 min of the recording and normalized to the baseline. n = 9 uninfected cells and n = 10 KIF13A-MS rescue neurons for the LTP (paired) pathway and n = 6 uninfected cells and n = 5 KIF13A-MS rescue neurons for the control (unpaired) pathway. Uninfected and KIF13A-MS-rescue neurons are significantly potentiated (P = 0.008 and P = 0.02, respectively) with respect to baseline (Wilcoxon test).

Figure S3.

Figure S3.

Expression of the recombinant proteins GFP-KIF13A-ST, mRFP-KIF13A-tail, and KIF13A-MS-YFP. (A) Representative confocal images of GFP-KIF13A-ST (green), mRFP-KIF13A-tail (orange), and KIF13A-MS-YFP (pink) expressed in organotypic hippocampal slices. (B) Western blot analysis of GFP-KIF13A-ST expression with antibodies against GFP and KIF13A. Infection with a GFP virus is used as control. Actin is used as loading control. (C) Similar Western blot analysis with slices expressing mRFP-KIF13A-tail or tdTomato, as control. (D) Western blot analysis of dissociated hippocampal neurons showing the reduction on the endogenous KIF13A protein levels (shKIF13A and KIF13A-MS-rescue conditions; KIF13A antibody) together with the expression of the recombinant and shKIF13A-resistant KIF13A-MS-YFP protein (KIF13A-MS-rescue condition; GFP antibody). Recombinant KIF13A-MS-YFP is not recognized by the KIF13A antibody, as the epitope is located in the absent globular tail domain. Both control (lacking the shRNA sequence) and shKIF13A lentiviruses express mCherry. Actin is used as loading control.

Figure 4.

Figure 4.

Protein interactions between GluA1, KIF13A and centaurin-α1 upon LTP induction. (A) GFP-immunoprecipitation from protein extracts of 7‒10 DIV organotypic hippocampal slices expressing GFP-KIF13A-ST protein or GFP as control. Extracts were obtained from slices maintained at basal conditions (aCSF) or from slices undergoing cLTP induction for different times (7.5 min, 15 min) or from slices recovered for 15 min following cLTP (Recovery), as indicated. Coimmunoprecipitated proteins (“Bound”) and input fractions were analyzed by Western blot using antibodies for GluA1 and centaurin-α1. IP, immunoprecipitation. Left, representative Western blots. Right, quantification of the amount of coprecipitated GluA1 normalized to baseline (dashed line). Dark symbols represent mean ± SEM; gray symbols represent individual values from each experiment. n = 5, 4, 5, and 4 experiments for baseline, 7.5-min cLTP, 15-min cLTP, and recovery, respectively. (B) Left: Similar to A, with the immunoprecipitation of endogenous GluA1. Nonimmune rabbit IgGs were used as immunoprecipitation control. GluA1, centaurin-α1, and KIF13A were detected by Western blot. Right: Quantification of the coprecipitated amount of KIF13A (green circles) and centaurin-α1 (purple circles). P values (P = 0.009 and P = 0.003) represent significant difference from baseline (Friedman test). n = 4, 4, 4, and 3 experiments for baseline, 7.5-min cLTP, 15-min cLTP, and recovery, respectively, in the case of KIF13A coimmunoprecipitation. n = 5, 5, 5, and 4 experiments for baseline, 7.5-min cLTP, 15-min cLTP, and recovery, respectively, in the case of centaurin-α1 coimmunoprecipitation. (C) Left: Similar to B, with the immunoprecipitation of endogenous centaurin-α1 and Western blot detection of endogenous KIF13A, GluA1, and centaurin-α1. Nonimmune goat IgGs were used as immunoprecipitation control. Right: Quantifications of the coprecipitated KIF13A (green circles) and GluA1 (black circles). P value represents significant difference from baseline (P = 0.009, Friedman test). n = 4, 4, 4, and 3 experiments for baseline, 7.5-min cLTP, 15-min cLTP, and recovery, respectively, in the case of KIF13A coimmunoprecipitation. n = 5, 5, 5, and 4 experiments for baseline, 7.5-min cLTP, 15-min cLTP, and recovery, respectively, in the case of GluA1 coimmunoprecipitation.

Figure S4.

Figure S4.

Immunoprecipitation of GFP-KIF13A-ST and effect of phosphatase inhibitors on centaurin-α1 coprecipitation. (A) GFP-immunoprecipitation from protein extracts of 7–10 DIV organotypic hippocampal slices expressing GFP-KIF13A-ST protein, or GFP as control, as described for Fig. 4 A of the main text. The immunoprecipitation of the recombinant protein was detected by Western blot with an anti-GFP antibody. GFP-KIF13A-ST appears as multiple bands, probably due to proteolytic degradation. Arrowhead indicates full-length recombinant protein; asterisks indicate light and heavy chains of the IgG used for immunoprecipitation. (B) Representative Western blots from experiments similar to Fig. 4 B of the main text but in the presence (+inhibitors) or absence (−inhibitors) of phosphatase inhibitors in the homogenization buffer. Nonimmune rabbit IgGs were used as immunoprecipitation control. GluA1, centaurin-α1, and pGluA1 (S845) were detected by Western blot. (C) Quantification of the amount of pGluA1 (S845) immunoprecipitated in both conditions (black circles, with phosphatase inhibitors; red squares, without phosphatase inhibitors). Darker lines represent the mean ± SEM, while lighter lines show individual values for each experiment. n = 4 experiments. (D) Quantification of the amount of centaurin-α1 protein coprecipitated in both conditions (black circles, with phosphatase inhibitors; red squares, without phosphatase inhibitors) and normalized to baseline coprecipitated protein. Darker lines represent the mean ± SEM, while lighter lines show individual values for each experiment. n = 4 experiments.

Figure S5.

Figure S5.

Subcellular fractionation of hippocampal slices. Hippocampal slices were homogenized (total fraction) and subsequently underwent different centrifugation steps to isolate different subcellular fractions: Micros., light membranes plus microsomes; PSD, postsynaptic density; TSF, Triton-soluble fraction and presynaptic plus extrasynaptic fraction. Blots show two different experiments performed in parallel. Synaptophysin and PSD-95 are used as pre and postsynaptic markers, respectively.

Figure 5.

Figure 5.

LTP requires centaurin-α1. (A) Left: Representative Western blot of protein extracts from dissociated hippocampal neurons (15 DIV) infected for 7 d with shCentaurin-α1 lentivirus (purple) or with a control lentivirus (lacking the shRNA sequence, black). Right: Quantification of centaurin-α1 protein levels relative to uninfected condition. Bars show mean ± SEM, together with individual values from each experiment; n = 6 independent experiments; shCentaurin-α1 values were significantly different from uninfected and control (P = 0.03, Wilcoxon test). (B) Western blot similar to the one shown in A, including neuronal cultures infected with the rescue GFP-centaurin-α1 lentivirus. (C) GFP-channel (top) and immunohistochemistry (IHC) against centaurin-α1 (bottom) from organotypic hippocampal slices infected with the rescue GFP-centaurin-α1 lentivirus. (D) Left: Time course of the LTP induction in organotypic hippocampal slices (7‒11 DIV) infected for 7‒10 d with lentivirus for the expression of shCentaurin-α1 (purple symbols) or shCentaurin-α1 plus shRNA-resistant GFP-centaurin-α1 (blue symbols), compared with uninfected neurons (gray symbols). The amplitude of the synaptic response is normalized to a 3-min baseline before the LTP induction (300 pulses at 3 Hz, coupled to postsynaptic depolarization at 0 mV). Each point represents the mean of the normalized EPSCs ± SEM per condition. Right: Average response from the last 5 min of the recording and normalized to the baseline. Bars represent the mean ± SEM, together with individual values for each experiment. Left bars (LTP, paired) correspond to the stimulation pathway in which LTP protocol was induced; right (shaded) bars (control, unpaired) correspond to the pathway that was not stimulated during induction. n = 20, 10, or 12 cells for control, shCentaurin-α1, or centaurin-α1 rescue neurons, respectively, for the LTP (paired) pathway. n = 20, 6, or 6 cells for control, shCentaurin-α1, or centaurin-α1 rescue neurons, respectively, for the control (unpaired) pathway. Uninfected and rescue neurons are significantly potentiated with respect to baseline (P = 0.0003 and P = 0.005, respectively, Wilcoxon test). P values displayed in the figure indicate significant difference between shCentaurin-α1 with respect to uninfected (P = 0.001) and rescue (P = 0.009) conditions, respectively (Mann–Whitney test). (E and F) Basal synaptic responses mediated by AMPARs (E) and NMDARs (F) recorded simultaneously from adjacent neurons, uninfected and infected with either shCentaurin-α1 (purple) or GFP-centaurin-α1 rescue (blue). Gray symbols represent absolute values of the responses from each individual pair. Colored symbols show the mean ± SEM of the measured currents. n = 16 cell pairs for shCentaurin-α1 and n = 15 for centaurin-α1 rescue, in the case of AMPAR currents (E). n = 12 cell pairs for shCentaurin-α1 and n = 15 for centaurin-α1 rescue in the case of NMDAR currents (F). n.s., not significantly different (Wilcoxon test). Insets: Representative traces for uninfected (black lines), shCentaurin-α1 (purple lines), and centaurin-α1 rescue (blue lines) neurons; scale bars: vertical, 10 pA; horizontal, 10 ms.

Figure 6.

Figure 6.

Role of KIF13A in LTP-induced endosomal redistribution. (A) Representative images of KIF13A-YFP (magenta) and mCherry-FIP2 (green) fluorescence in transfected HEK-293T cells 24 h after transfection. Scale bars, 10 µm. (B) Quantification of colocalization between KIF13A-YFP and mCherry-FIP2 in clustered areas compared with the expected colocalization for a random distribution of the proteins (percentage of mCherry-FIP2 and KIF13A-YFP clustered area over whole-cell area, respectively). Plots represent percentage of the area for each individual cell and show mean ± SEM for each analysis. n = 16 cells from two independent experiments; P < 0.0001, Wilcoxon test. (C) Quantification of colocalization between KIF13A-YFP and mCherry-FIP2 clusters from the experiments described in A using PCC an MOC. Individual Manders coefficients (M1 and M2, KIF13A and FIP2, respectively) are also shown. Plots represent correlation coefficient for each individual cell and show mean ± SEM for each analysis. n = 16 cells from two independent experiments. (D) Representative images of a cLTP experiment showing dendritic branches from an organotypic hippocampal slice infected with GFP-FIP2 (green) together with a control lentivirus (control; left panels) or a lentivirus expressing shKIF13A (right panels), both expressing mCherry (red). Arrows indicate coinfected dendrites. Scale bars, 10 µm. (E) Quantifications of cluster density per dendrite for each condition before and after cLTP induction. Analysis of cluster density was performed blind with respect to the lentivirus identity. Gray lines represent each analyzed dendrite, and black and red symbols show the mean ± SEM for each condition. n = 24 (control) and n = 13 (shKIF13A) dendrites from 10 and 6 independent experiments, respectively; P = 0.005, Wilcoxon test. (F) Cumulative probability distribution of cluster area of GFP-FIP2 for each condition described in D and E. Lighter lines represent baseline conditions. Darker lines represent conditions after cLTP induction. Gray lines represent data from slices infected with the control lentivirus (lacking shRNA sequence). Red lines represent data from slices expressing shKIF13A. P value < 0.0001 (Kolmogorov–Smirnov test) compares baseline and 15-min LTP in control conditions. Cluster size did not change in dendrites lacking KIF13A (shKIF13A; P = 0.92, Kolmogorov–Smirnov test).