CD44: a novel synaptic cell adhesion molecule regulating structural and functional plasticity of dendritic spines

Abstract

Synaptic cell adhesion molecules regulate signal transduction, synaptic function, and plasticity. However, their role in neuronal interactions with the extracellular matrix (ECM) is not well understood. Here we report that the CD44, a transmembrane receptor for hyaluronan, modulates synaptic plasticity. High-resolution ultrastructural analysis showed that CD44 was localized at mature synapses in the adult brain. The reduced expression of CD44 affected the synaptic excitatory transmission of primary hippocampal neurons, simultaneously modifying dendritic spine shape. The frequency of miniature excitatory postsynaptic currents decreased, accompanied by dendritic spine elongation and thinning. These structural and functional alterations went along with a decrease in the number of presynaptic Bassoon puncta, together with a reduction of PSD-95 levels at dendritic spines, suggesting a reduced number of functional synapses. Lack of CD44 also abrogated spine head enlargement upon neuronal stimulation. Moreover, our results indicate that CD44 contributes to proper dendritic spine shape and function by modulating the activity of actin cytoskeleton regulators, that is, Rho GTPases (RhoA, Rac1, and Cdc42). Thus CD44 appears to be a novel molecular player regulating functional and structural plasticity of dendritic spines.

FIGURE 1:

CD44 is expressed by neurons in adult rat brain and localizes at synapses. (A) In situ hybridization signal (CD44 mRNA, antisense probe, green) colocalizes with immunofluorescence of anti–MAP-2 (red) antibody in the CA3 field of the rat hippocampus. Hybridization with the sense probe is shown as a control. Scale bar, 15 μm. (B) Immunogold electron microscope detection (after embedding) of CD44 in the CA3 region of the hippocampus. Immunogold particles indicating CD44IR (red arrowheads) are present within the dendritic spines (SPINE, orange) and axonal boutons (B, blue). Scale bar, 250 nm. Distribution of CD44 in dendritic spines and presynaptic boutons was quantified. For χ² = 1140 or 217, respectively, and one degree of freedom, p < 000001, and so the distribution pattern of gold-labeled CD44 in both compartments is significantly different from random, in contrast to the labeling with nonimmune IgG control antibody.

FIGURE 2:

CD44 affects dendritic spine morphology. Primary hippocampal neurons transfected with pSuper or CD44 shRNA plasmids together with β-actin–GFP were subjected to immunocytochemistry using anti-CD44 antibody. (A) The effect of CD44 knockdown was estimated based on the average intensity of the CD44 immunofluorescence (IF) signal in transfected cells. AU, arbitrary units; 30 neurons per group. (B) Representative images of dendrites from dissociated hippocampal neurons (on 21 DIV) transfected with pSuper or CD44 shRNA plasmid together with a GFP-encoding vector. Scale bar, 2 μm. (C) Analysis of dendritic spine head width (left) and length (right) in all three groups. (D) Analysis of dendritic spine density. (E) Analysis of dendritic spine head width (left) and length (right) of cells cotransfected with pSuper, CD44 shRNA, or CD44 shRNA/CD44Rescue together with a GFP-encoding vector. The data were obtained from 25–30 neurons per group in three independent cultures, pSuper n_spines = 4409, CD44 shRNA n_spines = 3578. The “rescue” experiment pSuper n_spines = 450, CD44 shRNA n_spines = 500, and CD44 shRNA/CD44Rescue n_spines = 350. The data are expressed as mean ± SEM. **p < 0.01, ***p < 0.001 (Student’s t test).

FIGURE 3:

Single neurons transfected with CD44 shRNA construct in wild-type cultures exhibit a significant deficit in AMPA/kainate receptor glutamatergic transmission. (A–F) Statistics for standard parameters of AMPA/kainate mEPSCs recorded in control neurons (black), CD44 shRNA (red), and CD44Rescue neurons (white). (A) The average mEPSC frequency recorded in CD44shRNA neurons was significantly smaller than with control and CD44Rescue cells. (B) Cumulative plots of the time interval between subsequent mEPSCs recorded in various neurons. Note that CD44 silencing resulted in significant prolongation of the occurrence of mEPSCs (red line). (C, D) The manipulation of CD44 expression did not result in a significant alteration of mEPSC amplitude and 10–90 rise time. (E) Statistics of the average mEPSC monoexponential decay time constant (τ). CD44 knockdown neurons exhibited significantly larger τ values than with control and CD44Rescue cells. (F) Example of averaged mEPSC traces obtained from control neurons (black line), CD44 shRNA neurons (red line), and CD44Rescue neurons (gray line). Note the prolonged decay of mEPSCs recorded in CD44-knockdown neurons. The data were obtained from 8–13 neurons per group. The data are expressed as mean ± SEM. *p < 0.5, **p < 0.01 (Student’s t test).

FIGURE 4:

CD44 shRNA reduces the number of Bassoon-positive presynaptic puncta. (A) Representative images of dendritic segments of pSuper- or CD44 shRNA-transfected cells immunostained with anti-Bassoon and anti–PSD-95 antibodies. Arrowheads indicate dendritic spines with decreased PSD-95 level or without Bassoon signal. (B) Analysis of average PSD-95 fluorescence intensity within dendritic spines vs. dendritic shafts. (C) Quantification of average density of Bassoon-positive puncta localized within 0.2 μm of the neuronal surface apposed to the dendritic spine. The data were obtained from 8–10 neurons per group, pSuper n_spines = 600 and CD44 shRNA n_spines = 700 in three separate experiments. The data are expressed as mean ± SEM. **p < 0.01, ***p < 0.001 (Student’s t test).

FIGURE 5:

Activity-dependent structural plasticity of dendritic spines is regulated by CD44. Live-imaging sessions of neurons cotransfected with pSuper or CD44 shRNA plasmid together with a GFP-encoding vector were performed under control conditions (DMSO) and 40 min after cLTP induction. Arrowheads indicate examples of dendritic spines that increased upon cLTP induction. (A) Representative images of dendritic segments captured during the live-imaging experiment under control (DMSO) and cLTP-stimulated conditions (indicated by minus and plus signs, respectively). Open bars, cells treated with DMSO; hatched bars, cells stimulated with cLTP mixture. (B) Graph representing relative changes in the spine head width. The data were obtained from 8–10 neurons per condition, pSuper + DMSO n_spines = 420, pSuper + cLTP n_spines = 700, CD44 shRNA + DMSO n_spines = 400, and CD44 shRNA n_spines = 478. The data are expressed as mean ± SEM. ***p < 0.001 (Student’s t test).

FIGURE 6:

CD44 regulates the morphology of dendritic spines by altering the activity of small GTPases RhoA, Rac1, and Cdc42. (A–D) CD44-dependent modulation of activity of small Rho-GTPases (RhoA, Rac1, and Cdc42). (A) Structure of FRET-based biosensors of small Rho GTPase activity. (B–D) Average fluorescence lifetimes for donor (CFP) in dendritic spines of cells cotransfected with pSuper or CD44 shRNA plasmid together with Raichu-Cdc42 (B), Raichu-Rac1 (C), or Raichu-RhoA (D) FRET sensor, with representative fluorescence lifetime images of all analyzed groups. Scale bar, 5 μm. The corresponding color histograms depict the lifetime distribution in a false color scheme and visualize the lifetime variations. Warmer colors indicate shorter lifetimes and a higher level of activated protein. Cooler colors indicate longer lifetimes and a lower level of small Rho GTPase activity. (E–G) Defects in dendritic spine morphology that were induced by CD44 knockdown were rescued by the inhibition of Cdc42 activity. (E) Representative images of dendrites from dissociated hippocampal neurons (on 21 DIV) transfected with pSuper, CD44 shRNA, CD44 shRNA_DN-Cdc42, and pSuper_DN-Cdc42 plasmids together with a GFP-encoding vector. Scale bar, 5 μm. Analysis of dendritic spine length (F) and head width (G) in all four groups. The lifetime data were collected from 8–10 neurons per group and an average of 100 spines per group. The morphological data were obtained from 10–15 neurons per group and an average of 600 spines per group. The data are expressed as mean ± SEM. **p < 0.01, ***p < 0.001 (Student’s t test).