Tau mislocalization to dendritic spines mediates synaptic dysfunction independently of neurodegeneration

Abstract

The microtubule-associated protein tau accumulates in Alzheimer's and other fatal dementias, which manifest when forebrain neurons die. Recent advances in understanding these disorders indicate that brain dysfunction precedes neurodegeneration, but the role of tau is unclear. Here, we show that early tau-related deficits develop not from the loss of synapses or neurons, but rather as a result of synaptic abnormalities caused by the accumulation of hyperphosphorylated tau within intact dendritic spines, where it disrupts synaptic function by impairing glutamate receptor trafficking or synaptic anchoring. Mutagenesis of 14 disease-associated serine and threonine amino acid residues to create pseudohyperphosphorylated tau caused tau mislocalization while creation of phosphorylation-deficient tau blocked the mistargeting of tau to dendritic spines. Thus, tau phosphorylation plays a critical role in mediating tau mislocalization and subsequent synaptic impairment. These data establish that the locus of early synaptic malfunction caused by tau resides in dendritic spines.

Figure 1. Memory and synaptic plasticity in rTgP301L mice

(A) Timeline of sequential training and testing in the Morris water maze as previously described (Westerman et al., 2002; Ramsden et al., 2005). (B) No deficits were observed in locating the visible platform at any age. Only 4.5-month old rTgP301L mice showed a significant deficit in learning to locate the hidden platform. (C) rTgP301L mice at 4.5 but not 1.3 months (M) showed impaired retention of spatial reference memory as assayed by the platform crossing index. n = 12 rTg mice at each age and n = 4 mice for each TgNeg group at each age (B) for a total of 12 mice for pooled TgNeg group (C). (D) There was no difference in theta burst stimuli (TBS)-induced long term potentiation (LTP) of field excitatory postsynaptic potentials (fEPSPs) in the CA1 hippocampal subfield between rTgP301L and TgNeg littermates at 1.3 months. (E) LTP was impaired in 4.5-month old rTgP301L mice. n = 10 brain slices from each group in 4.5-month old mice (n = 6 animals/genotype); n = 14 rTg brain slices and n = 12 control slices from 1.3-month old mice (n = 6 animals/genotype). +/+, rTgP301L mice; +/−, −/+, or −/−, TgNeg control mice that express the activator, responder, or neither transgene, respectively. For experiments in panels C–E, the TgNegcontrolmicewerecombinedintoonegroupdesignatedasTgNegsincetherewasno difference between TgNeg subpopulations in cognitive performance (B). M, months; m, meters. Repeated-measures ANOVA, *p < 0.05, **p < 0.01.

Figure 2. Lack of age-dependent memory deficits and neurodegeneration in rTgWT mice

(A) rTgWT mice were longitudinally tested in the Morris water maze and compared to previously published data from the rTgP301L mice (Ramsden et al., 2005) [2.5M–9.5M n = 52 TgNeg, n = 17 rTgWT and n = 9 rTgP301L]. rTgWT mice showed deficits in performance in the Morris water maze compared to TgNeg mice. Deficits in rTgWT mice remained stable for several months, while deficits in rTgP301L mice became more severe with increasing age. Although rTgWT spent less time swimming in the target quadrant during probe trials compared to TgNeg mice, only the rTgP301L mice exhibited an age-related decline. Repeated measures ANOVA data are as follows: transgene: F(2,225) = 34.09, p < 0.0001; age versus transgene: F(6,225) = 3.61, p < 0.01 ; age, TgNeg mice: F(3,153) = 0.345, p = 0.79; age, rTgWT mice: F(3,48) = 1.518, p = 0.22; age, rTgP301L mice: F(3,24) = 5.82. (B) While both rTgWT and rTgP301L mice exhibited a significant decrease in whole brain weight early in life (1.3–4.5 months of age), suggestive of a developmental delay, the rTgP301L mice also exhibited neurodegeneration as evidenced by a progressive decrease in whole brain weight later in life (7–10 months of age). Due to small sample numbers at some ages, mice were binned into P1-P28, 1.3M-4.5M, and 7M-10M. Although there was a trend towards decreased brain weights at P1-P28, this was not statistically different. Data for transgene ANOVA are as follows: P1-P28: F(2,82) = 0.524, p = 0.5939; 1.3M-4.5M: F(2,156) = 14.564, p < 0.0001; 7M-10M: F(2,141) = 66.957, p < 0.0001. Sample sizes for TgNeg, rTgWT, rTgP301L, respectively, are as follows: P1, n = 13, n = 2, n = 2; P7, n = 20, n = 3, n = 2; P14, n = 17, n = 4, n = 4; P28, n = 9, n = 5, n = 4; 1.3M, n = 18, n = 5, n = 12; 3M, n = 35, n = 15, n = 14; 4.5M, n = 26, n = 21, n = 13; 7M, n = 19, n = 14, n = 19; 9.5M, n = 12, n = 9, n = 12; 10M, n = 36, n = 9, n = 14. M, months; **p < 0.01, ***p < 0.001, ****p < 0.0001. See also Figure S1 for details on the creation and characterization of rTgWT mice.

Figure 3. Human tau levels in the forebrains of rTgP301L and rTgWT mice

(A) Tubulin and htau proteins in the S2 fraction of forebrains from 4.5-month old (4.5M) TgNeg, rTgWT and rTgP301L mice. (B) PSD95 and htau proteins in the PSD-1 fractions of forebrains from the animals in (A). (C) There was no difference in the normalized levels of htau expression between rTgWT and rTgP301L mice in the S2 fraction. (D) There was more htau, normalized to PSD95, in the PSD-1 fractions of rTgP301L mice. (E) There was no difference in the levels of PSD95 between TgNeg, rTgWT and rTgP301L mice. (F) Levels of PSD95 did not change in relation to a-tubulin between TgNeg, rTgWT and rTgP301L mice. n = 9–10 mice per group. M, male; F, female. t-test (C, D), ANOVA followed by Bonferroni post-hoc analysis (E), *p < 0.05.

Figure 4. Effects of the FTDP-17-linked P301L mutation on tau accumulation in dendritic spines and excitatory synaptic transmission in transfected rat neurons

(A) Overlaid images of living dissociated rat hippocampal neurons (21–28 DIV) co-expressing DsRed and GFP-tagged htau proteins. (B) Cropped images from (A) including DsRed, GFP and overlaid images. Triangles denote dendritic spines devoid of WT htau and arrows denote spines containing P301L htau. See also Figure S2 for unprocessed raw images of cultured rat neurons co-transfected with DsRed and GFP-WT htau or GFP-P301L htau. (C) Quantification of total spines and htau-containing spines in neurons expressing GFP or co-expressing DsRed and GFP-tagged htau. There was more htau in the dendritic spines of neurons expressing P301L htau. n = 27 (GFP), 26 (WT) and 25 (P301L) neurons per group. (D) A higher percentage of spines contained htau in neurons expressing P301L htau. (E) Representative mEPSCs in neurons expressing GFP (E1), GFP-tagged WT htau (E2) and GFP-tagged P301L htau (E3). Large mEPSCs occurred frequently in neurons expressing GFP or WT htau, but rarely in neurons expressing P301L htau. (F) Cumulative frequency distributions of mEPSC amplitudes of the neurons in (E). There were markedly more small mEPSCs in neurons expressing P301L htau than those expressing GFP. There was also a slight, but significant, decrease in the proportion of large events in neurons expressing WT htau. (Bin size = 1 pA) (G) Mean mEPSC amplitudes and frequencies of the neurons in (E). Neurons expressing P301L htau exhibited smaller and fewer mEPSCs. n = 14 (GFP), 14 (WT) and 16 (P301L) neurons per group. ANOVA followed by Bonferroni (C) or Fisher’s PLSD (G) post-hoc analysis, t-test (D), Kolmogorov-Smirnov (F), *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 5. Excitatory synaptic transmission and surface AMPA receptors in rTgP301L and rTgWT neurons

(A) Representative averaged mEPSCs. (B) Cumulative frequency distribution of mEPSC amplitudes. There was a marked shift toward smaller amplitudes in rTgP301L neurons, and a slight shift toward smaller amplitudes in rTgWT neurons. (Bin size = 1 pA) (C) Mean mEPSC amplitudes and firing frequencies of the neurons in (A). rTgP301L neurons exhibited smaller and fewer mEPSCs. n = 15 (TgNeg), 15 (rTgWT) and 14 (rTgP301L) neurons per group. (D) PSD95 and surface AMPAR immunoreactivity. PSD95 (red) detected using a monoclonal antibody and surface AMPARs (green) detected using a polyclonal antibody against the N-terminus of GluR1 (N-GluR1) co-localized in TgNeg and rTgWT neurons, but rarely in rTgP301L neurons. (E) Quantification of spine N-GluR1 immunoreactivity co-localizing with PSD95 immunoreactivity, normalized to N-GluR1 immunoreactivity on adjacent dendritic shafts. n = 10 neurons per group. (F) Density of dendritic spines labeled with a PSD95 antibody. There were no differences in the density of PSD95 immunoreactive clusters along the dendrites. n = 10 neurons per group. Kolmogorov-Smirnov (B), ANOVA followed by Fisher’s PLSD post-hoc analysis (C, E, F), *p < 0.05, ***p < 0.001.

Figure 6. The P301L htau mutation disrupts clustering of GluR1, GluR2/3 and NR1 receptors at dendritic spines in cultured mouse neurons

(A) The PSD95 protein (left column) and the C-terminus of GluR1 subunits (middle; overlay, right) were co-stained with a mouse monoclonal antibody (red) and rabbit polyclonal antibody (green), respectively. Neurons (21–28 DIV) were cultured from TgNeg, rTgWT and rTgP301L mice (from top to bottom). Note that both surface and intracellular receptors were detected as all primary antibodies were added after the fixation and permeabilization of neurons. (B) Co-staining of PSD95 and GluR2/3 subunits (detected by an antibody that recognizes the C-termini of both GluR2 and GluR3). (C). Co-staining of PSD95 and NR1 (a mandatory subunit for functional NMDARs). (D) Quantification of spine GluR1, GluR2/3 and NR1 immunoreactivity identified by co-localization with PSD95 immunoreactivity, normalized to GluR1, GluR2/3 and NR1 immunoreactivity on adjacent dendritic shafts, respectively. In all images, small arrows denote glutamate receptor clusters that co-localize with PSD95 clusters whereas large arrows denote diffuse staining of glutamate receptors in dendritic shafts. n = 10 neurons per group. ANOVA followed by Fisher’s PLSD post-hoc analysis (D), *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 7. Pathological tau species do not accumulate in rTgWT mice

At 5 months of age (approximately when memory impairments begin, but before neuron loss), immunohistochemical studies revealed that cortical and hippocampal neurons in rTgWT mice were rarely immunoreactive for hyperphosphorylated tau species associated with disease. Detection of total tau (human and mouse) with the Tau-5 antibody in hippocampal and cortical neurons of rTgWT and rTgP301L mice. At 5 months, an immunoreactive signal for tau hyperphosphorylated at residue S202 (CP-13) was observed in the cell bodies of a small number of rTgWT hippocampal and cortical neurons. In contrast, CP-13 labeled-cell bodies and neurites were observed in a majority of rTgP301L hippocampal and cortical neurons. Strikingly, positive immunoreactive labeling for tau hyperphosphorylated at S409 (PG-5) and S396/S404 (PHF-1) was observed in hippocampal and cortical cell bodies and neurites of 5 month-old rTgP301L, but not rTgWT, mice. Nuclei were stained with DAPI. For each antibody panel, magnification was x5 (top row), x10 (middle row) and x40 (bottom row). Hipp, hippocampus; Ctx, cortex.

Figure 8. Hyperphosphorylation of P301L htau and effects of proline-directed phosphorylation on tau accumulation in dendritic spines of transfected rat hippocampal neurons

(A) Levels of phosphorylated htau in rat neurons expressing WT or P301L htau. Expression of total htau and phosphorylated S199 (pS199) htau proteins after immunoprecipitation of total htau with the Tau-13 antibody from 21–28 DIV primary rat hippocampal neurons expressing WT or P301L htau. Changes in tau conformation and phosphorylation states associated with late tau pathology were also examined with the Alz-50 antibody. In the middle panel, omission of the Tau-13 antibody (no Ab) or tissue lysate (no Lys) revealed no specific immunoblot signal for total htau. (B) There was more pS199, normalized to total tau, in the P301L htau-expressing neurons. n = 3 sample preparations. (C) Images of living rat neurons co-expressing DsRed and six different GFP-tagged htau proteins (WT, P301L, AP, AP/P301L, E14 or E14/P301L). In neurons expressing WT, AP and AP/301L htau most DsRed-labeled spines were devoid of htau. In neurons expressing P301L, E14 and E14/P301L htau most spines contained htau. (D) Quantification of total spines and htau-containing spines in neurons co-expressing DsRed and GFP-tagged htau. A higher percentage of spines contained htau in neurons expressing P301L, E14 and E14/P301L htau (compared to WT htau), and a lower percentage of spines contained htau in neurons expressing AP and AP/P301L (compared to WT htau). n = 25–28 neurons per group. t-test (B), ANOVA followed by Bonferroni post-hoc analysis (D), *p < 0.05, ***p < 0.001.

Figure 9. Quantitative measurement of GFP-htau expression levels in transfected rat hippocampal neurons

(A) Flow cytometry was used to measure GFP fluorescence intensity levels in hippocampal neurons transfected with WT and mutant GFP-htau. Untransfected hippocampal neurons served as a negative control (left panel, representative dot blot analysis) and neurons transfected with GFP alone were the positive control (right panel, representative dot blot analysis). The readout of GFP fluorescence in the right panel was used to determine the uniform gating parameters for counting GFP-positive events. Numbers in each quadrant are the percentage of cells in the respective quadrant. GFP-positive events are in the lower right quadrant. Dead neurons are in the upper left quadrant, as identified by increased fluorescence intensity of the DNA marker 7-aminoactinomycin D (7-AAD). Cellular debris is in the lower left and upper right quadrants.(B)Flow cytometry demonstrated that mean GFP fluorescence intensity did not differ significantly between different populations of hippocampal neurons transfected with GFP-htau constructs. n = 50,000 events sampled for each GFP-htau construct (10,000 events/GFP-htau construct × 5 different preparations). (C) Fluorescent pixel intensity of GFP-tagged htau constructs in dendritic shafts. n = 19 neurons per group.

Figure 10. Effects of proline-directed phosphorylation of tau on excitatory synaptic transmission in transfected rat hippocampal neurons and a hypothetical model of tau-mediated synaptic dysfunction preceding neurodegeneration

(A) Representative averaged mEPSCs. (B) The cumulative frequency distribution of mEPSC amplitudes of neurons in (A). There was a shift toward smaller amplitudes in neurons expressing P301L, E14 and E14/P301L htau. (C) Mean mEPSC amplitudes and (D) frequencies of the neurons in (A). Neurons expressing P301L, E14 and E14/P301L htau exhibited smaller and fewer mEPSCs. n = 14 (WT), 16 (P301L), 15 (AP), 15 (AP/P301L), 15 (E14) and 15 (E14/P301L) neurons per group. (E) A diagram describing the hypothetical cellular processes underlying how hyperphosphorylated tau impairs excitatory synaptic transmission through a loss of functional surface glutamate receptors. Since Aβ and genetic mutations in tau induce hyperphosphorylation of tau, we hypothesize that, following tau hyperphosphorylation, tau dissociates from the microtubules and is aberrantly targeted to dendritic spines, possibly through an actin-based process. Once mislocalized to the spines, tau impairs glutamatergic synaptic transmission by reducing the number of functional AMPA and NMDA receptors on the surface of the neuronal membrane through impaired anchoring of the glutamate receptors to the PSD complex, decreased exocytosis or increased endocytosis of the receptors. ANOVA followed by Fisher’s PLSD post-hoc analysis (C, D), Kolmogorov-Smirnov (B), **p < 0.01, ***p < 0.001.