PSD-95 Alters Microtubule Dynamics via an Association With EB3

The SH3 domain of PSD-95 directly binds to a polyproline-rich region of EB3. A, EB3 directly interacts with the SH3 domain of PSD-95. Lysates from COS-7 cells expressing EB3-GFP were resolved by SDS-PAGE and transferred to PVDF membrane. Membrane strips were overlayed with the indicated GST fusion proteins of PSD-95 domains followed by immunoblotting with an antibody to GST. B, Purified EB3 and PSD-95 interact directly. Equimolar amounts of purified, bacterially expressed GST-EB3 and GST-PSD-95 were incubated at 4°C overnight followed by immunoprecipitation with an antibody to PSD-95. Precipitates were subjected to immunoblotting with an antibody to EB3. IP, Immunoprecipitation; WB, Western blot. C, Sequence alignment of the EB microtubule-associated protein family members. EB3 contains a polyproline region. Conserved residues are highlighted in yellow and boxed. The gaps are indicated with dashes. An EB3-specific proline-rich residue motif is highlighted in red from amino acids 136 to 141. D, Display of the best EB3 proline-rich peptide conformation and its accommodation in the SH3-binding groove of PSD-95 after docking using GOLD. The peptide was rendered in element-coded sticks (red, oxygen; blue, nitrogen; white, hydrogen; gray, carbon) and the N and C termini are indicated. The PSD-95 structure is shown using a gray-coded solvent accessible surface (SAS), and residues in the SH3 domain in PSD-95 involved in the peptide binding are shown in cyan.

Deletion of the SH3 binding region in PSD-95 rescues PSD-95-induced branching deficits. A, Representative inverted GFP images of hippocampal neurons transfected with the cDNA constructs for the indicated proteins. Neurons were transfected on DIV 7 and analyzed for dendrite number at DIV 12. Scale bars: 50 μm. B, Sholl analysis of neurons transfected with cDNA constructs encoding PSD-95-GFP, PSD-95 ΔSH3-GFP, or GFP. Neurons expressing PSD-95 ΔSH3-GFP do not show PSD-95-promoted decreases in dendritic arborization but instead show similar levels of branching to neurons expressing GFP up to a distance of 100 μm from the cell body. ***p < 0.001 by ANOVA followed by Bonferroni multiple comparisons test. n = 42 neurons for each condition. Error bars indicate SEM.

PSD-95 decreases the binding of EB3 to microtubules in heterologous cells and inhibits microtubule assembly in a cell-free system. A, Microtubules were immunoprecipitated from lysates of COS-7 cells expressing EB3-GFP and exogenously added GST-PSD-95 or GST-PSD-95ΔSH3. Immunoprecipitates were subjected to SDS-PAGE and Western blotting for the presence of EB3. B, Light scattering of microtubules at OD₃₅₅ in the presence of 1 mm GTP was used to detect the effect of PSD-95 on microtubule polymerization. Addition of 4 m GST-PSD-95, but not GST-PSD-95ΔSH3, slows the polymerization of purified tubulin compared to the addition of equimolar concentrations of GST alone. Inset, Western blots of purified tubulin fractions (right lane), used in the above polymerization assay, show detectable amounts of EB3. The immunoreactive protein band comigrates with that from a rat brain homogenate (left lane). C, Quantification of total tubulin polymerization shows that GST-PSD-95, but not GST-PSD-95ΔSH3, lowers the total amount of polymerization that occurred as measured by V_max of the curve. *p < 0.05 by two-way repeated measures ANOVA followed by a Bonferroni multiple comparisons test to compare each point on the curve. n = 3. Error bars indicate SEM.

PSD-95 increases microtubule crossing at dendrite branch points. A–C, EM images of microtubules of hippocampal neurons transfected at DIV 10 and fixed on DIV 12. Asterisks indicate a branch point from a dendrite. A, Control neurons that express GFP have organized microtubules at dendrite branch points. B, Neurons that overexpress PSD-95-GFP show disorganized microtubules at that cross at dendrite branch points when compared to control GFP expressing neurons. C, Neurons that express PSD-95ΔSH3-GFP show organized microtubules at dendrite branch points, similar to control neurons expressing GFP. D, Quantitation of the number of microtubules that cross at dendrite branch points in transfected hippocampal neurons represented in A–C. Microtubules cross dendrite branch points more often in neurons that overexpress PSD-95-GFP when compared to microtubules of neurons that overexpress GFP or PSD-95ΔSH3 GFP. **p < 0.01 by Kruskal–Wallis test followed by Dunn's multiple comparison test. n = 5 neurons for each condition. E, Example of how microtubule crossings were counted. A 0.5 μm line (black box) was drawn and any microtubules crossing that line were counted (arrows). Scale bars: A–C, E, 0.5 μm. Error bars indicate SEM.

PSD-95 decreases the lifetime of EB3 comets in hippocampal neurons. A, Diagram of a mutant form of EB3 with key amino acids of the proline-rich region changed to threonines. B, Wild-type, but not EB3-THR, coimmunoprecipitates with PSD-95 from extracts of COS-7 cells coexpressing the indicated proteins. The anti-EB3 antibody used for Western blotting recognizes a nonspecific (ns) ∼50 kDa protein band in COS-7 cell lysates. IP, Immunoprecipitation; WB, Western blot. C, Low-magnification images of GFP-tagged EB3 comets in a growing neurite. Asterisks indicate representative comets. D, Representative time series images of GFP-tagged EB3 and EB3-THR comets in the dendrites of hippocampal neurons and GFP-tagged EB3 and EB3-THR comets in hippocampal neurons also expressing PSD-95-mRFP or PSD-95ΔSH3-mRFP. Images were taken at 18–24 h posttransfection, which was performed at the time of plating. Scale bars: 1 μm. Arrows indicate first and last appearances of comets. Asterisks track with tip of the comets. E, Quantitation of the effect of PSD-95-mRFP overexpression on EB3-GFP and EB3-THR-GFP comet lifetime. Overexpression of PSD-95-mRFP decreases the lifetime of EB3-GFP comets but does not decrease the lifetime of EB3-THR-GFP comets. *p < 0.05; **p < 0.01 by Kruskal–Wallis test followed by Dunn's multiple comparison test compared to EB3-GFP. n ≥ 15 comets for each condition. F, Quantitation of the effect of PSD-95 overexpression on EB3-GFP and EB3-THR-GFP comet velocity. Overexpression of PSD-95-GFP decreases the velocity of both EB3-GFP and EB3-THR-GFP comets, suggesting that this effect is not specific to the direct interaction between PSD-95 and EB3. *p < 0.05; **p < 0.01 by Kruskal–Wallis test followed by Dunn's multiple comparison test compared to EB3-GFP. n ≥ 15 comets for each condition.

Overexpression of PSD-95 decreases branching even in the presence of a nonbinding mutant of EB3. A, Representative inverted GFP images of hippocampal neurons transfected with the indicated cDNA constructs. Neurons were transfected on DIV 7 and analyzed for dendrite number at DIV 12. Scale bars: 50 μm. B, Sholl analysis of neurons transfected with cDNA constructs for the indicated proteins. Analysis of neurons represented in A shows that overexpression of EB3-THR-GFP results in decreased dendritic complexity when compared to dendrite numbers of control neurons expressing GFP or neurons expressing EB3-GFP. ***p < 0.01 by two-way repeated measures ANOVA followed by a Bonferroni multiple comparisons test to compare each point on the curve. n = 24 neurons for each condition. C, Sholl analysis of neurons expressing EB3-GFP or EB3-THR-GFP alone or with PSD-95-mRFP. Neurons expressing EB3-THR-GFP show a decreased dendritic complexity when compared to control neurons. Expression of PSD-95-mRFP decreases dendrite number. Coexpression of either EB3-GFP or EB3-THR-GFP with PSD-95-mRFP results in a decrease in dendritic complexity that is equal to the decrease mediated by PSD-95 alone. ***p < 0.01 by two-way RM ANOVA followed by a Bonferroni multiple comparisons test to compare each point on the curve. n = 24 neurons for each condition. Error bars indicate SEM.