Tau is required for the function of extrasynaptic NMDA receptors

Tau is a microtubule-associated neuronal protein found mainly in axons. However, increasing evidence indicates that it is also present in dendrites, where it serves as an essential mediator of synaptic NMDA (N-methyl-D-aspartate) receptor-dependent excitotoxicity. Of note, NMDA receptors can also be found outside synapses in the plasma membrane, and activation of extrasynaptic NMDA receptors has been shown to be more linked to excitotoxicity than the activation of synaptic ones. Little is known about the role of Tau in the activity of extrasynaptic NMDA receptors. Thus, we have used a Tau knockout mouse model (Tau−/− mice) to analyze the consequences of Tau absence in extrasynaptic NMDA receptor activity. We demonstrate that absence of Tau leads to a decrease in functional extrasynaptic NMDA receptors in the hippocampus. We propose that this impairment in extrasynaptic NMDA receptor activity may contribute to the well-known neuroprotective effect associated with Tau deficiency under pathological conditions.


Results
Dendritic Tau has been indirectly associated to synaptic NMDA receptor function 4 . However, little is known about its contribution to the function of extrasynaptic NMDA receptors. Thus, we analyzed possible differences in the composition and/or in the function of NMDA receptors in WT and Tau −/− mice.
Tau protein is not involved in the regulation of the levels of NMDA receptors in the hippocampus in vivo. Specific NMDA receptor subunits are not confined to specific localizations in the plasma membrane. However, it has been proposed that GluN2B-containing receptors are more mobile than those containing GluN2A, thus contributing to an enrichment in the former at extrasynaptic sites and in the latter at synaptic sites 12 . Thus, in order to examine a possible role of Tau protein in the expression and/or distribution of NMDA receptors, we analyzed the levels of GluN2 subunits in the hippocampi of WT and Tau −/− mice. Western blot analysis revealed no differences in the levels of GluN2B subunits recognized with an antibody directed against the N-terminal domain of the protein (GluN2B N-ter) (t = 0.294; p = 0.774) (Fig. 1A). Of note, this N-terminal domain is always present in GluN2B-containing NMDA receptors. In contrast, the C-terminal domain of GluN2B subunits can be cleaved in extrasynaptic pools of the receptor 12 . In such cases, GluN2B protein would not be expected to be recognized by an antibody directed against this C-terminus (GluN2B C-ter), a feature that could help to distinguish between different pools of NMDA receptors. However, no differences were found either between WT and Tau −/− mice when the GluN2B C-ter antibody was used (t = 1.181; p = 0.260) (Fig. 1B). Similarly, no differences were found in the levels of GluN2A subunits (t = −0.889; p = 0.391) (Fig. 1C).

Absence of Tau protein does not alter post-translational GluN2B modifications.
With the aim to fine-tune the analysis of the effects of Tau absence on NMDA receptor distribution, we next examined synaptosome-enriched fractions obtained from the hippocampi of WT and Tau −/− mice. The absence of Tau did not alter the levels of PSD95 (U = 7.000; p = 0.556), the main scaffold synaptic protein ( Fig. 2A). Likewise, no differences were found in the levels of GIPC (U = 10.000; p = 1.000) (Fig. 2B), an alternative scaffolding protein that preferentially stabilizes extrasynaptic NMDA receptors 13 . Similarly, Tau absence did not lead to differences in the levels of GluN2B recognized either in its C-terminus (U = 8.000; p = 0.730) or N-terminus domain (U = 8.000; p = 0.730) ( Fig. 2A,C). GluN2B are thought to be the most mobile and widespread distributed subunits of NMDA receptors 12 . In this regard, posttranslational modifications, such as tyrosine phosphorylation, have been related to GluN2B cell surface mobility. Thus, we next analyzed the levels of pGluN2B(Y1472), a phosphorylation found mainly on synaptic GluN2B subunits, and pGluN2B(Y1336), a phosphorylation found mainly on extrasynaptic GluN2B subunits 14 , in synaptosome-enriched fractions of WT and Tau −/− mice. No differences were detected in the levels of either of these phosphorylated forms relative to total GluN2B levels, although pGluN2B(Y1472) tended to decrease in Tau −/− mice (U pGluN2B(Y1472) = 2.000; p = 0.063; U pGluN2B(Y1336) = 9.000; p = 0.905) (Fig. 2D,E). Moreover, since phosphorylation at S1480 of GluN2B subunits mediates NMDA receptor internalization from synapses by disrupting their interaction with PSD95 15 , we evaluated the levels of pGluN2B(S1480) in WT and Tau −/− synaptosome-enriched fractions. No differences were found either in this parameter, although it tends to be increased in Tau −/− mice (U = 0.000; p = 0.05) ( Supplementary Fig. 1).
To further evaluate the involvement of Tau in extrasynaptic NMDA receptor distribution, we analyzed by immunohistochemistry the area occupied by pGluN2B(Y1336) in the hippocampi of WT and Tau − / mice. Figure 3A shows representative images of pGluN2B(Y1336) and PSD95, the latter a marker of synaptic sites, in the CA1 region of the hippocampi of WT animals. Consistent with the association of Y1336 phosphorylation mainly with extrasynaptic GluN2B subunits, the percentage of colocalization between pGluN2B(Y1336) and PSD95 was relatively low: 16.82 ± 1.43% (Fig. 3B). Figure 3C shows representative images of the area occupied by pGluN2B(Y1336) in the CA1 region of WT and Tau −/− mice. No significant differences caused by the absence of Tau were found in this parameter (U = 1181.000; p = 0.758) (Fig. 3D). Strikingly, the percentage of colocalization between pGluN2B(Y1336) and PSD95 in Tau −/− animals was reduced as compared to WT ones (U = 828.000; p = 0.004). However, this effect was found to be the consequence of a decrease in the area occupied by PSD95 in Tau −/− mice (U = 729.000; p = 0.001) ( Supplementary Fig. 2).
In summary, these results suggest that Tau protein does not affect the total levels or the subcellular distribution of NMDA receptor subunits in the hippocampus. tau −/− neurons lack extrasynaptic NMDA receptor currents. Alterations in the dynamics of postsynaptic densities related to impaired structural plasticity were recently described in Tau −/− granule neurons 10 . Importantly, this lack of plasticity could involve decreased mobility of NMDA receptors at synapses, which in turn has been related to defective long-term depression (LTD) 16 . Actually, Tau −/− neurons show impaired LTD 17 , and Tau phosphorylation at S396 has been suggested to be involved in such type of synaptic plasticity 18 . According to that, we found pTau(S396) in synaptosome-enriched fractions of WT animals ( Supplementary Fig. 3). Thus, it is possible that absence of Tau results in a decrease in the amount of extrasynaptic NMDA receptors, arising from a deficient dendritic spine turnover 12 , which cannot be detected by biochemical approaches due to the low percentage of such receptors compared to synaptic ones 19,20 . In order to test this hypothesis, we evaluated next potential functional differences in WT and Tau −/− neurons by using electrophysiological recordings.
To isolate responses from extrasynaptic receptors, we used a strategy based on blocking synaptic responses with the irreversible, use-dependent NMDA receptor blocker MK-801 during synaptic stimulation 19,20 . Essentially, baseline NMDA receptor responses from hippocampal slices are recorded, alternating Schaffer collateral synaptic stimulation and local NMDA puffs on the stratum radiatum (see scheme in Fig. 4A). The NMDA puff is expected to elicit a combined response from both synaptic and extrasynaptic receptors. The contribution from synaptic receptors is then removed by continuing Schaffer collateral stimulation in the presence of MK-801, while puffed stimulation is stopped. Once the inhibition of synaptic responses is stabilized, the puffed stimulation is resumed. The relative inhibition of the new puffed response is expected to be lower than that of the synaptic response, as extrasynaptic receptors should not have been blocked during the MK801 incubation (see result from WT mice in Fig. 4B). By comparing the extent of inhibition of the electrical and puffed responses, the contribution of the extrasynaptic receptors can be calculated (Fig. 4D, WT) 19 . This contribution was about 30% in WT mice (Fig. 4E), which is in good agreement with previous calculations using this method 19,20 , and also with morphological calculations using immunogold electron microscopy 21 . Strikingly, the extent of MK-801 inhibition was virtually identical for the synaptic and puffed responses in the case of Tau −/− slices (Fig. 4C, points on the diagonal line in Fig. 4D). These numbers indicate a close to null contribution of extrasynaptic receptors to the puffed response in the absence of Tau (Fig. 4E).
Therefore, these electrophysiological recordings reveal the virtual absence of functional extrasynaptic NMDA receptors in CA1 neurons of Tau −/− mice.

Discussion
Amyloid-induced toxicity has been proposed to occur through the interaction of the peptide with NMDA receptors. Compared to AMPA receptors, NMDA receptors show a more stable presence at the cell membrane, thereby facilitating the entry of calcium, which can be toxic over certain levels 22 . Interestingly, NMDA receptors can be found both inside and outside synapses 21 , the latter being more associated with the activation of cell death pathways 5 .
Here, we report for the first time electrophysiological recordings showing a decreased functionality of extrasynaptic NMDA receptors in the hippocampus of Tau −/− mice. Of note, absence of Tau has been extensively to WT ones. Quantification is shown on the right. Graphs show mean ± SEM. N WT = 5 mice; N Tau −/− = 4 mice. The levels of pGluN2B(Y1472) and pGluN2B(Y1336) were calculated related to the levels of GluN2B N-ter. Actin was used as loading control. Blots were cropped to improve the clarity of the presentation. Fulllength blots are presented in Supplementary Fig. 5.  (2019)  www.nature.com/scientificreports www.nature.com/scientificreports/ associated to neuroprotection against NMDA receptor-dependent excitotoxicity 23 . Traditionally, this neuroprotection has been related to a dendritic role of Tau in the functionality of synaptic NMDA receptors 4 . However, given the strong association of extrasynaptic NMDA receptors with cell death pathways, we propose the impairment in extrasynaptic NMDA receptor functionality as an additional neuroprotective mechanism led by Tau absence.
NMDA receptors can laterally diffuse between synaptic and extrasynaptic sites, although the mechanisms underlying this mobility are not fully understood. In this regard, GluN2B subunits are thought to be more mobile, widespread and less confined to synapses than GluN2A subunits 5 . Indeed, posttranslational modifications of GluN2B have been related to specific NMDA receptor localizations 14 . Phosphorylation of GluN2B at Y1472 promotes the interaction of the subunit with scaffold proteins like PSD95 4 , thus stabilizing NMDA receptors in the postsynaptic density 24 . Accordingly, phosphorylation of GluN2B at Y1472 is decreased in extrasynaptic NMDA receptors, the phosphorylation at GluN2B Y1336 being predominant in these receptors 14,21 . Thus, the levels of GluN2B phosphorylation at Y1472 and Y1336 could serve as indirect markers of synaptic and extrasynaptic NMDA receptors, respectively. However, we did not find differences in Y1472 or Y1336 GluN2B phosphorylation in response to the absence of Tau. Thus, it cannot be assumed a decrease in the levels of extrasynaptic receptors www.nature.com/scientificreports www.nature.com/scientificreports/ in Tau −/− animals based on this criteria. Nevertheless, the mechanisms that regulate NMDA receptor subcellular localization and function includes a wide range of protein-protein interactions, phosphorylation, palmitoylation, ubiquitination and receptor proteolytic cleavage 12 . Thus, additional experiments would be required to further analyze the involvement of Tau in the localization of NMDA receptors at synaptic and extrasynaptic sites. Noteworthy, there is a well-known association of Tau with actin 16,18,25 , and actin is closely related to NMDA receptor localization and function: depolymerization of actin filaments decrease NMDA receptor signaling 26 ; changes in actin dynamics are required for long-term plasticity and synaptic function 27,28 ; and alterations of the actin cytoskeleton may facilitate the lateral diffusion of NMDA receptors from synaptic to extrasynaptic sites 29 . In line with this, Tau phosphorylated at serine 396 has been related to the induction of long-term depression (LTD) 17 , and as a result of LTD actin cytoskeleton ultimately undergoes depolymerization 30 . As a consequence, synaptic NMDA receptors could be moved to extrasynaptic regions 12 (Fig. 5). Thus, Tau may be required to facilitate changes in the actin cytoskeleton surrounding NMDA receptors to facilitate their lateral diffusion, and in the absence of Tau, less functional extrasynaptic NMDA receptors could be found.
Altogether, our results reveal that lack of Tau protein impairs the functionality of extrasynaptic NMDA receptors. Given the pivotal role of this type of NMDA receptors in the activation of cell death pathways 5,6 , this decreased functionality may contribute to the well-known neuroprotection conferred by the absence of Tau 1,2,4,10,31 . However, it is important to note that Tau deletion is not always neuroprotective. For example, acute knockdown of Tau at the hippocampus may cause memory deficits 32 , Tau deletion can promote brain insulin resistance 33 , and absence of Tau results in impaired plasticity of newborn granule cells 10 . In line with this, absence or pathological modifications of Tau have been associated with impaired LTP [34][35][36][37] , which in turn is closely related www.nature.com/scientificreports www.nature.com/scientificreports/ to synaptic functionality. Thus, absence of Tau can be beneficial or detrimental depending on the nature of the stimuli and the cell processes affected. In this regard, the novel effect of Tau on extrasynaptic NMDA receptor functionality reported here could represent a more specific target for therapies based on the neuroprotection conferred by the removal of Tau.
In summary, we have used biochemical and electrophysiological analyses to study the involvement of Tau protein in the functionality of extrasynaptic NMDA receptors. Based on electrophysiological analysis, we found that Tau is required for the function of these receptors. Since that function is related to Aβ neuronal toxicity, we propose that such impairment in extrasynaptic NMDA receptor activity may contribute to the neuroprotective effect of Tau deficiency found in the presence of Aβ.

Methods
Animals. Tau −/− mice were generated as previously described 38 . Heterozygous (Tau +/− ) mice were crossed in order to obtain homozygous Tau knockout mice (Tau −/− ) and control littermates (WT). Mice were housed under standard laboratory conditions in a specific pathogen-free colony facility at the Centro de Biología Molecular "Severo Ochoa" (CBMSO). Housing conditions were in accordance with European Community Guidelines (directive 86/609/EEC), and animals were handled complying with European and local animal care protocols. Animal experiments received the approval of the CBMSO Ethics Committee and the National Ethics Committee (PROEX 291/15). Both male and female mice were included in this study. Depending on the experiment, 4-7 animals per genotype were used. They were 4 months old at the time of sacrifice for biochemical and immunohistochemical analyses, and in a range of 7-9 months old for electrophysiological recordings.
Sacrifice and tissue processing. Animals were fully anesthetized by an intraperitoneal injection of pentobarbital (EutaLender, 60 mg/kg) and perfused intracardially with 0.9% saline. For immunohistochemical analysis, brains were removed and fixed overnight in 4% paraformaldehyde (PFA). Sagittal sections (50 µm thick) were obtained on a Leica VT1200S vibratome, and series of sections comprising every 8th section were generated. For biochemical analysis of tissue extracts, hippocampi were dissected and preserved at −80 °C until processing. For biochemical analysis of synaptosome-enriched fractions, hippocampi were dissected, and immediately subjected to subcellular fractionation.
For electrophysiological experiments, coronal acute hippocampal slices (300 µm thick) were prepared. Mice were briefly anesthetized using dry ice sublimated with water, and the brains were removed and placed in partially frozen Ca 2+ -free dissection medium (10 mM D-Glucose, 4 mM KCl, 26 mM NaHCO 3 , 234 mM sucrose, 5 mM MgCl 2 , 1:1000 Phenol Red) saturated with 5% CO 2 /95% O 2 . Coronal slices were cut using a Leica VT1200S vibratome and recovered for 1 h at 32 °C in a submersion-type incubation chamber containing artificial cerebrospinal fluid (ACSF; 119 mM NaCl, 2.5 mM KCl, 2.5 mM CaCl 2 , 1.2 mM MgCl 2 , 26 mM NaHCO 3 , 1 mM NaH 2 PO 4 , 11 mM glucose; pH 7.4). Subsequently, the incubation temperature was lowered to 25 °C, at which point the slices were considered ready for experimentation. subcellular fractionation. Subcellular fractionation of hippocampal tissue was performed following the protocol described in 39 , with minor modifications. Briefly, hippocampi were dissected and homogenized in ice-cold buffer containing 0.32 M sucrose, 1 mM HEPES, 1 mM MgCl 2 , 1 mM EDTA and 1 mM NaHCO 3 . The homogenized tissue was centrifuged at 1000 g, 4 °C for 10 min, and the resulting supernatant was centrifuged www.nature.com/scientificreports www.nature.com/scientificreports/ again at 13000 g, 4 °C for 15 min to obtain a crude synaptosome-enriched fraction in the pellet. This pellet was then resuspended in buffer containing 75 mM KCl and 1% Triton X-100 and centrifuged at 100000 g, 4 °C for 90 min. The resulting pellet, enriched in membrane-associated proteins, was resuspended in 20 mM HEPES to obtain a final synaptosome-enriched fraction. Protein content was determined by the Pierce ® BCA Protein Assay kit (Thermo Scientific), and samples were processed for Western blot analysis. All buffers used in this protocol contained phosphatase inhibitor cocktail (PhosSTOP, Sigma-Aldrich) and protease inhibitor cocktail (Complete, Roche).
Analysis of the fluorescence intensity of extrasynaptic-enriched GluN2B subunits (pGluN2B Y1336) in the CA1 region of the hippocampus. Single plane images were acquired in a LSM710 Zeiss confocal microscope (63x oil immersion objective; XY dimensions: 24.1 μm). Five animals per genotype were used in this analysis, and ten images of randomly selected regions of the CA1 region of the hippocampus per animal were obtained from the sections comprising the series. To analyze all the images, an invariant threshold was applied to the green (pGluN2B Y1336) and red (PSD95) channels using Metamorph software (Molecular Devices), and the area over the threshold was automatically calculated. The percentage of colocalization between pGluN2B(Y1336) and PSD95 was also calculated by the software.
Electrophysiological recordings. Electrophysiological experiments to determine the extrasynaptic proportion of functional NMDA receptors were loosely based on the procedure described in 19 . Briefly, electrically evoked field excitatory postsynaptic potentials (fEPSPs) and NMDA-evoked puff fEPSPs were recorded in the stratum radiatum of CA1 of acute hippocampal slices at 25 °C. The recording chamber was perfused with low-magnesium ACSF containing 0.2 mM MgCl 2 , 100 µM picrotoxin and 10 µM CNQX, and gassed with 5% CO 2 /95% O 2 . Bipolar stimulation electrodes were placed among Schaffer collaterals, and glass recording electrodes (1-2 MΩ; filled with ACSF) were placed ~200 µm away in the direction of the fiber projection. A glass puff pipette (1-2 MΩ) containing NMDA (1 mM in ACSF) was positioned adjacent to the recording electrode, immediately above the surface of the slice. Responses mediated by both synaptic and extrasynaptic NMDA receptors were elicited with NMDA puffs, delivered using pressurized nitrogen (4.5 bars) gated by a Picospritzer III unit (200-500 ms pulse duration). Electrically evoked and puff fEPSPs were collected at interleaved 4-min intervals during baseline acquisition. Once a stable baseline had been recorded for 20 min, puff delivery was stopped, and MK-801 (40 µM) was infused into the bath. Stimulation frequency was subsequently increased to 0.066 Hz, and evoked responses were allowed to run down until they reached a stable nadir, at which point a final puff was delivered. Stimulation-evoked and puff fEPSPs were quantified as their peak amplitude. statistical analysis. Statistical analysis was performed using the SPSS 24 software (SPSS, 1989; Apache Software Foundation, Chicago, IL, USA). The Kolmogorov-Smirnov test was used to test the normality of the sample distribution. Atypical data were detected with box-plots and eliminated when necessary. In the case of normal sample distribution, data were analyzed by a t Student test. In those cases in which normality could not be assumed, data were analyzed by the non-parametric Mann-Whitney U test. Graphs represent mean values ± SEM.

Data Availability
The data generated during this study is included in this published article and its Supplementary File, Figs 1-5.