Tau is a microtubule‐associated neuronal protein found mainly in axons. However, its presence in dendrites and dendritic spines is particularly relevant due to its involvement in synaptic plasticity and neurodegeneration. Here, we show that Tau plays a novel in vivo role in the morphological and synaptic maturation of newborn hippocampal granule neurons under basal conditions. Furthermore, we reveal that Tau is involved in the selective cell death of immature granule neurons caused by acute stress. Also, Tau deficiency protects newborn neurons from the stress‐induced dendritic atrophy and loss of postsynaptic densities (PSDs). Strikingly, we also demonstrate that Tau regulates the increase in newborn neuron survival triggered by environmental enrichment (EE). Moreover, newborn granule neurons from Tau−/− mice did not show any stimulatory effect of EE on dendritic development or on PSD generation. Thus, our data demonstrate that Tau−/− mice show impairments in the maturation of newborn granule neurons under basal conditions and that they are insensitive to the modulation of adult hippocampal neurogenesis exerted by both stimulatory and detrimental stimuli.
Tau regulates the in vivo maturation of newborn granule neurons under physiological conditions. In addition, Tau deficiency prevents both the detrimental and stimulatory actions of acute stress and environmental enrichment, respectively, on the functional maturation of these cells. Altogether, our data shed light on novel functions of Tau protein related to the plastic modulation of adult hippocampal neurogenesis by external stimuli.
Tau plays a novel in vivo role in the synaptic maturation of newborn granule neurons under basal conditions.
Tau is involved in immature granule neuron cell death caused by acute stress.
Tau regulates the stress‐induced impairment of newborn granule neuron functional maturation.
Tau regulates the increase in newborn neuron survival triggered by environmental enrichment.
Tau modulates the stimulatory effects of environmental enrichment on the functional maturation of newborn granule neurons.
Tau is a neuronal microtubule‐associated protein (MAP) that promotes microtubule assembly and stabilization (Weingarten et al, 1975). It plays key roles in the establishment of neuronal polarity and migration during embryonic development (Caceres & Kosik, 1990; Dawson et al, 2001; Sapir et al, 2012; Sayas et al, 2015), in axonal transport (Ballatore et al, 2007; Hernandez & Avila, 2010) and in intracellular trafficking (Hernandez & Avila, 2010). Under physiological conditions, Tau is located mainly in axonal microtubules (Hirokawa et al, 1996; Aronov et al, 2001), although increasing evidence supports its presence also in dendrites (Ittner et al, 2010) and dendritic spines (Ittner et al, 2010; Mondragon‐Rodriguez et al, 2012; Kimura et al, 2014). The affinity of Tau to bind microtubules is finely regulated, depending mainly on the selective expression of various isoforms and post‐translational modifications. Tau protein is encoded by mapt gene. Human mapt contains 16 exons, from which several Tau isoforms are generated by alternative splicing (Goedert et al, 1989; Andreadis et al, 1992). Exon 10 encodes one of the four repeat sequences that form the microtubule‐binding domain. The presence or absence of exon 10 results in Tau isoforms with four (Tau4R) or three (Tau3R) repeat sequences, Tau4R showing a higher affinity to bind microtubules than Tau3R ones (Lu & Kosik, 2001; Avila et al, 2004). At early developmental stages, Tau3R predominates (Lu & Kosik, 2001; Avila et al, 2004), conferring a lower stability of the cytoskeleton and allowing the morphological differentiation and migration of developing neurons. In contrast, in the adult murine brain, Tau4R is the predominant isoform (Lu & Kosik, 2001; Avila et al, 2004), thereby guaranteeing the stability of the cytoskeleton required to maintain neuronal integrity. It is important to note an exception that occurs during adult hippocampal neurogenesis (AHN), when Tau3R isoforms can be found (Bullmann et al, 2007; Llorens‐Martin et al, 2012).
Adult neurogenesis occurs in discrete brain regions during adulthood. Among these regions, the hippocampal dentate gyrus (DG) has attracted increasing attention due to the functional relevance of AHN, a process related to hippocampal‐dependent learning and mood regulation (Sahay & Hen, 2008; Aimone et al, 2011). AHN encompasses the proliferation of adult neural stem cells, their differentiation into mature neurons, and their incorporation into the hippocampal circuitry (Ming & Song, 2005). Furthermore, the addition of new neurons to the hippocampal circuit is regulated by numerous external factors such as physical activity, environmental enrichment (EE), and stress (Gould et al, 1997; Kempermann et al, 1997), thus conferring an outstanding degree of plasticity to the network.
By using a Tau knockout mouse model (Dawson et al, 2001), here we demonstrate that Tau is involved in the morphological differentiation and synaptic integration of newborn hippocampal granule neurons in vivo. Moreover, we report a novel function of Tau in the regulation of the negative consequences of acute stress on AHN. We also demonstrate, for the first time, that Tau regulates the stimulatory effects of EE on AHN.
Tau protein is not involved in the regulation of the basal rate of adult hippocampal neurogenesis (AHN)
We first examined whether Tau protein is involved in the regulation of AHN under basal conditions. Figure 1A–D shows that the absence of Tau did not affect the number of proliferative phospho‐histone 3 (PH3)+ cells (t = 0.757; P = 0.471) or the number of apoptotic caspase‐cleaved actin (fractin)+ cells (t = −0.771; P = 0.452) in the DG. However, as proliferation and death rates are general parameters that may not reflect changes in specific cell subpopulations, we also analyzed the survival of 1‐, 2‐, 4‐, 6‐, and 8‐week‐old cells labeled with the thymidine analogs 5‐iodo‐2′‐deoxyuridine (IdU) or 5‐chloro‐2′‐deoxyuridine (CldU). Figure 1E shows the experimental design. Tau depletion did not alter cell survival at any of the ages studied (Fig 1F). Hence, we conclude that Tau protein is not involved in the regulation of newborn neuron survival rate. Figure 1G summarizes the main cell markers expressed during the sequential differentiation stages that newborn neurons go through before they become fully mature. We addressed whether Tau deficiency affects the differentiation of newborn granule neurons and their commitment to neural lineage. To explore this notion, the percentage of cells that expressed doublecortin (DCX) and neuronal nuclei (NeuN) markers was evaluated in 1‐, 4‐, and 8‐week‐old CldU+ or IdU+ newborn neurons. As expected, DCX expression decreased with time, whereas the opposite effect was observed for NeuN (Fig 1H). No differences in the expression of either of these markers were found in response to Tau deficiency. Thus, it can be concluded that Tau is not involved in newborn neuron differentiation under basal conditions. Moreover, the absence of Tau did not alter the number of Sex determining region Y‐box 2 (Sox2)+ (t = −1.180; P = 0.261) or brain lipid‐binding protein (BLBP)+ (t = −1.015; P = 0.330) neural progenitor cells. In addition, no differences in the number of DCX+ neuroblasts (t = −1.201; P = 0.253) or in the number of calretinin+ immature neurons (t = −1.427; P = 0.179) were found in Tau−/− mice compared to WT (Fig 1I–N). All together, these data suggest that neither is Tau involved in the regulation of AHN rate under basal conditions.
Tau protein is necessary for the dendritic maturation of newborn granule neurons
To study the role of Tau in the morphological maturation of newborn neurons, we analyzed the morphology of the dendritic tree of 4‐ and 8‐week‐old newborn granule neurons labeled with PSD95‐GFP‐expressing retroviruses. Figure 2A shows representative images of 4‐week‐old newborn granule neurons of WT and Tau−/− mice. In these neurons, the absence of Tau decreased the total dendritic length (t = 4.963; P < 0.001) (Fig 2B). Moreover, Sholl's analysis revealed alterations in the complexity of the dendritic tree (Fig 2C). In particular, Tau deficiency decreased dendritic branching in 100–150 μm (t = 2.397; P = 0.020), 150–200 μm (U = 143; P < 0.001), and 200–250 μm (U = 354; P = 0.037) from the cell soma. Figure 2D shows representative images of 8‐week‐old newborn granule neurons of WT and Tau−/− mice. However, at this cell age, the absence of Tau did not lead to differences in the total dendritic length (t = −1.266; P = 0.210) (Fig 2E) or in the complexity of the dendritic tree (Fig 2F).
Taken together, these results show that Tau is involved in the morphological maturation of newborn granule neurons, since its absence causes a transient alteration of the dendritic arborization of these cells.
Tau protein is necessary for the formation of postsynaptic densities (PSDs), dendritic spines, and mossy fiber terminals (MFTs) in granule neurons
In addition to the morphology of newborn neurons, we examined the participation of Tau in the functional maturation of these cells. Figure 2G shows representative images of 4‐week‐old WT and Tau−/− newborn granule neurons infected with PSD95‐GFP‐expressing retroviruses and their respective high‐power magnifications, in which PSDs (afferent synapses) can be observed. No differences were found in PSD density along the dendritic tree (Fig 2H). However, the average PSD area was higher in the 2nd (U = 56,105; P < 0.001) and 4th (U = 992,492; P < 0.001) branching orders of Tau−/− dendrites compared to WT ones (Fig 2I). The absence of Tau in 8‐week‐old WT and Tau−/− newborn neurons (Fig 2J), fully integrated into the circuitry and infected with PSD95‐GFP‐expressing retroviruses, decreased the density (t = 2.034; P = 0.047) (Fig 2K) and area (U = 1,282,519; P < 0.001) (Fig 2L) of PSDs in the 5th dendritic branching order. Conversely, the PSD area was higher in the remaining branching orders (2nd (U = 81,074; P = 0.069); 3rd (U = 769,008; P < 0.001), and 4th (U = 1,304,378; P = 0.001)).
In light of these results, we examined whether the Tau deficiency has additional effects on the morphology of the dendritic spines of 8‐week‐old newborn neurons, which had shown the most remarkably alterations in PSDs. It should be noted that each dendritic spine can contain one, none, or more than one PSD (Fig EV1A). The absence of Tau increased the density of spines in the 4th branching order dendrites (U = 232; P = 0.022), whereas a trend to decrease this parameter was observed in 5th branching order dendrites (t = 1.815; P = 0.077) (Fig EV1B). Moreover, the diameter of the head of the spines located in 5th branching order dendrites was reduced due to Tau deficiency (U = 177,552; P < 0.001) (Fig EV1C). In addition, we classified the dendritic spines into three categories (stubby, thin, and mushroom) and quantified the percentages of each type of spine (Fig EV1D). Tau deficiency reduced the percentage of mushroom spines in 3rd (χ2 = 6.109; P = 0.014) and 5th (χ2 = 12.92; P < 0.001) branching order dendrites. Furthermore, it increased the percentage of stubby spines in the 3rd (χ2 = 7.076; P = 0.008) and that of thin spines in the 5th (χ2 = 8.173; P = 0.004), respectively.
In order to further characterize the morphological alterations caused by Tau deficiency on the postsynaptic elements of granule neurons, we next analyzed the ultrastructure of the afferent synapses in their dendritic spines by electron microscopy (Fig EV2A). This analysis was performed in three separate regions of the molecular layer (ML) of the DG: external (EML), medial (MML), and inner (IML). We found that Tau deficiency reduced the density of synapses (number of synapses per mm2) in the whole ML ((EML: t = 2.363; P = 0.0019); (MML: t = 9.282; P < 0.001); (IML: t = 2.371; P = 0.019)) (Fig EV2B). Moreover, it increased the size of the synaptic cleft ((EML: t = −11.216; P < 0.001); (MML: t = −12.385; P < 0.001); (IML: t = −8.343; P < 0.001)) (Fig EV2C) and caused alterations in PSDs, which showed a reduced area ((EML: t = 2.64; P = 0.0087); (MML: t = 2.322; P = 0.021); (IML: t = 2.641; P = 0.0087)) (Fig EV2D) and length ((EML: t = 2.668; P = 0.0079); (MML: t = 0.818; P = 0.413); (IML: t = 4.036; P < 0.001)) (Fig EV2E), and an increased depth ((EML: t = −2.669; P = 0.0079); (MML: t = −0.488; P = 0.625); (IML: t = −3.051; P = 0.0024)) (Fig EV2F) than WT ones.
In order to evaluate whether Tau deficiency also has an impact on the morphology of newborn neuron MFTs (efferent synapses), we quantified the area of each individual MFT in 8‐week‐old retrovirally labeled newborn granule neurons. Figure EV3A and B shows representative images of the whole hippocampus (A), as well as high‐power magnifications of MFTs in the CA3 field (B) of WT and Tau−/− mice. No differences were found between WT and Tau−/− animals (U = 17,886; P = 0.166) (Fig EV3C). In addition, and given the high variability in size of the MFT population, we plotted the percentages of MFT grouped by size in WT and Tau−/− mice (Fig EV3D). As shown, no differences in the distribution of MFT sizes were observed to be caused by the absence of Tau (K–S Z = 1.089; P = 0.187).
Finally, we analyzed the effects of Tau deficiency on the ultrastructure of the MFTs under an electron microscope (Fig EV3E). In this regard, we studied the stratum lucidum and stratum pyramidale of the CA3 region. The density of presynaptic vesicles (number of vesicles per μm2) (t = 3.548; P < 0.001) (Fig EV3F) and the length of the presynaptic active zone (t = 5.589; P < 0.001) (Fig EV3G) were reduced, whereas the size of the synaptic cleft (t = −7.409; P < 0.001) (Fig EV3H) was increased in Tau−/− animals compared to WT ones.
In summary, these results suggest that Tau participates in the functional maturation of granule neurons, since its absence alters PSD, dendritic spine, and MFT morphology.
The absence of Tau modifies the electrophysiological properties of granule neurons
In order to determine whether the morphological alterations caused by Tau deficiency affect the basal electrophysiological properties of granule neurons, we recorded the membrane properties and synaptic activity of these cells under whole‐cell configuration. Figure 3A and B shows representative traces of miniature excitatory postsynaptic currents (mEPSCs) recorded at a holding potential of −65 mV from WT (black traces) and Tau−/− (red traces) mice. The resting membrane potential of Tau−/− cells was hyperpolarized compared to that of WT cells (t = 2.217; P = 0.038) (Fig 3C).
We focused our analyses on the mEPSCs to assess whether synaptic inputs were altered by the absence of Tau. Granule cells of the DG of Tau−/− mice did not exhibit alterations in the amplitude (t = −0.256; P = 0.800) (Fig 3D) neither in the frequency (U = 53,000; P = 0.483) (Fig 3E) of mEPSCs compared to WT ones. However, examination of the cumulative probability distribution of amplitudes of mEPSCs revealed two distinct populations of mEPSCs altered in opposite directions in Tau−/− mice (Fig 3F). Thus, the amplitudes of small mEPSCs of Tau−/− mice were statistically larger than those of WT animals (K–S Z = 6.306; P < 0.001) (Fig 3F (i)). In contrast, when comparing the mEPSCs of larger amplitudes, those of Tau−/− mice were smaller than those of WT animals (K–S Z = 9.820; P < 0.001) (Fig 3F (ii)). This result is indicative of a redistribution of excitatory synaptic weights in Tau−/− mice.
Tau protein is required for stress‐induced death of newborn granule neurons
Given that AHN is regulated by external stimuli, we next examined whether Tau protein is involved in the negative regulation of AHN exerted by acute stress. We administered CldU thymidine analog to WT and Tau−/− animals in order to label dividing newborn neurons, and then split these animals into two experimental conditions: mice subjected to the Porsolt test (P), which is considered an acute stress model; and control mice not subjected to this test (CNP). Animals were sacrificed 1 week after CldU administration. Mice were subjected to the Porsolt test on the 2 days prior to sacrifice (Fig 4A and B). Figure 4C shows representative images of 1‐week‐old CldU‐labeled hippocampal newborn neurons of WT CNP, WT P, Tau−/− CNP, and Tau−/− P mice. The Porsolt test decreased 1‐week‐old cell survival in both genotypes (F1,52 = 39.14; P < 0.001), but this decrease was higher in WT than in Tau−/− mice (interaction F1,52 = 11.007; P = 0.002) (Fig 4D). These data suggest that the absence of Tau partially protects immature newborn neurons from acute stress‐induced cell death. It is important to note that, as previously described (Llorens‐Martin & Trejo, 2011), the Porsolt test did not increase the number of fractin+ apoptotic cells in either of the genotypes (F1,23 = 0.085; P = 0.774) (Fig 4E).
Tau protein mediates the deleterious effect of stress on newborn granule neuron morphology
With the aim to study the involvement of Tau protein in the morphological alterations triggered by the Porsolt test on newborn granule neurons, we analyzed 8‐week‐old newborn neurons labeled with PSD95‐GFP‐expressing retroviruses in WT and Tau−/− mice subjected or not to this test (Fig 5A). Figure 5B shows representative images of neurons in each experimental condition. The Porsolt test reduced the total dendritic length of WT (P < 0.001) but not of Tau−/− neurons (P = 0.623) (interaction F1,131 = 4.701; P = 0.032) (Fig 5C). Indeed, Tau depletion also partially prevented the alterations in the dendritic branching promoted by the Porsolt test, as revealed by Sholl's analysis. Although acute stress reduced branching between 50 and 100 μm from the soma in both genotypes (F1,133 = 26.19; P < 0.001), it did not cause alterations in any other point of the dendritic tree of Tau−/− neurons (Fig 5D). In contrast, while WT neurons also showed reduced dendritic branching between 100 and 150 μm from the soma (P = 0.002), they showed increased dendritic branching between 200 and 250 μm (P = 0.017) and between 250 and 300 μm (P = 0.008) from the soma.
Tau protein mediates the deleterious effect of stress on the PSDs and MFTs of newborn granule neurons
Figure 5E and F shows representative images of PSDs located in the proximal dendrites of WT CNP, WT P, Tau−/− CNP, and Tau−/− P newborn neurons labeled with PSD95‐GFP‐expressing retroviruses. In addition to causing alterations in the dendritic morphology of newborn neurons, acute stress also promoted a general decrease in PSD density throughout their whole dendritic tree in WT animals (2nd (P = 0.019), 3rd (P = 0.003), 4th (P = 0.005), and 5th (P = 0.003) branching orders). However, in Tau−/− animals, this decrease occurred only in 4th branching order dendrites (P = 0.007) (Fig 5G). The average PSD area was not affected by the Porsolt test in WT or in Tau−/− neurons (Fig 5H).
Interestingly, the levels of AMPA receptor GluR1 subunit were decreased in the hippocampus of WT animals after the Porsolt test (P = 0.006). However, this decrease was not observed in Tau−/− animals (P = 0.213) (Fig 5I and J).
Figure 5K and L shows representative images of the whole hippocampus, together with high‐power magnifications of mossy fiber terminals (MFTs) in the CA3 field of WT CNP, WT P, Tau−/− CNP, and Tau−/− P animals. Analysis of the average area of MFTs showed that the effect of acute stress differed in WT and Tau−/− neurons (interaction F1,693 = 7,06; P = 0.008). While the Porsolt test reduced the area of MFTs in WT neurons (P = 0.038), this effect was not observed in Tau−/− ones (Fig 5M).
Tau protein mediates the pro‐neurogenic effects of environmental enrichment (EE)
To test whether Tau protein is also involved in the positive regulation of AHN exerted by EE, we analyzed the survival of 8‐week‐old adult‐generated neurons in WT and Tau−/− animals in control housing (CH) or EE conditions (Fig 6A and B). EE increased the number of DCX+ neuroblasts in both genotypes (F1,33 = 73.12; P < 0.001) (Fig 6C and D). However, analysis of the survival of 8‐week‐old newborn neuron revealed that the effect was greater in WT than in Tau−/− animals (interaction F1,30 = 7.15; P = 0.013) (Fig 6E and F). Consistent with these data, EE induced a decrease in the number of fractin+ apoptotic cells in both genotypes (F1,32 = 98.10; P < 0.001), but this effect was less pronounced in Tau−/− animals (Fig 6G).
It is known that GABA promotes cell survival during AHN (Song et al, 2013). In order to study whether GABAergic innervation is related to the pro‐neurogenic effect of EE, we analyzed GAD‐65+ GABAergic terminals in the ML of mice in each experimental condition (Fig 7A). The results showed that EE promoted different effects in WT and Tau−/− animals in both the density (interaction F1,308 = 12.064; P = 0.021) (Fig 7B) and the average area (interaction F1,58527 = 62.406; P < 0.001) (Fig 7C) of GABAergic terminals. While EE promoted an increase in both parameters in WT mice (P = 0.009 and P < 0.001, respectively), it caused a reduction in both in Tau−/− animals (P = 0.023 and P < 0.001, respectively). Figure EV4A–D shows the regional changes observed in these parameters in the three subregions of the ML: EML (B), MML (C), and IML (D).
Tau protein mediates the stimulatory effects of EE on newborn neuron morphology
To study whether Tau protein participates in the adaptation of newborn granule neurons to EE, we injected PSD95‐GFP‐expressing retroviruses into the hippocampus of WT and Tau−/− animals before they were housed under CH or EE conditions for 8 weeks (Fig 8A). EE increased total dendritic length in WT mice (P = 0.001), whereas it had no effect on Tau−/− animals (P = 0.629) (interaction F1,131 = 4.26; P = 0.041) (Fig 8B and C). Indeed, EE caused an increase in the dendritic branching of WT neurons—an effect not observed in Tau−/− neurons—as revealed by Sholl's analysis (Fig 8D). In particular, this increment occurred between 50 and 100 μm (P = 0.004) and between 100 and 150 μm (P = 0.011) from the cell soma in WT mice. Thus, EE increased the dendritic length and branching in newborn neurons of WT but not of Tau−/− animals.
Tau protein mediates the stimulatory effects of EE on the PSDs and MFTs of newborn neurons
Figure 8E and F shows representative images of WT and Tau−/− PSDs in distal and proximal dendrites of 8‐week‐old newborn granule neurons labeled with PSD95‐GFP‐expressing retroviruses in CH or EE conditions. EE increased PSD density in proximal dendrites of WT neurons compared to WT CH neurons (2nd branching order dendrites P < 0.001) (Fig 8G). However, this effect was not observed in Tau−/− neurons (2nd branching order dendrites P = 0.235) (interaction F1,132 = 6.839; P = 0.010). Moreover, EE further decreased the density of PSDs in distal dendrites of Tau−/− mice compared to Tau−/− CH mice (5th branching order dendrites P = 0.006). In addition, EE decreased the average area of the PSDs in distal dendrites of both genotypes (4th (F1,7109 = 25.98; P < 0.001) and 5th (F1,6960 = 25.27; P < 0.001) branching orders). However, in enriched Tau−/− animals, this decrease also took place in proximal dendrites (2nd (P < 0.001) and 3rd (P < 0.001) branching orders) (Fig 8H).
Figure 8I and J shows representative images of MFTs in the CA3 region of WT CH, WT EE, Tau−/− CH, and Tau−/− EE animals. EE increased the average area of MFTs in WT (P = 0.002) but not in Tau−/− newborn neurons (P = 0.440) (Fig 8K).
In summary, EE increased PSD density in proximal dendrites and decreased PSD area in distal dendrites of WT newborn neurons, whereas it decreased PSD density in distal dendrites and reduced PSD area throughout the dendritic tree in Tau−/− newborn neurons. Moreover, EE caused an increment in the area of MFTs of WT newborn neurons. This observation was not made in the absence of Tau.
Finally, a schematic model illustrating the effects of the absence of Tau on the modulation of AHN by external stimuli is shown in Fig 9.
Tau is involved in the morphological and functional maturation of newborn granule neurons under basal conditions
Tau protein is crucial for the establishment of neuronal polarity during embryonic development, for axonal transport, and for the intracellular trafficking of various molecules (Caceres & Kosik, 1990; Dawson et al, 2001; de Barreda et al, 2010; Hernandez & Avila, 2010; Llorens‐Martin et al, 2011; Sapir et al, 2012). Given the tight control of Tau expression during AHN, it is reasonable to assume that this protein plays a regulatory role in this process. However, despite the widespread expression of Tau during AHN (Fuster‐Matanzo et al, 2009), none of the cellular subpopulations studied was affected by the absence of Tau. Strikingly, here we demonstrate that Tau depletion causes a transient alteration in the morphology of adult newborn neurons. Our data are consistent with those of previous reports showing a delayed morphological maturation of embryonic neurons both in vitro (Dawson et al, 2001) and in vivo (Sapir et al, 2012). Our results show that Tau knockout reduced the total dendritic length and branching of 4‐week‐old newborn granule neurons, a phenomenon not observed in 8‐week‐old newborn neurons. These data suggest that other microtubule‐associated proteins play compensatory roles at later stages of newborn neuron maturation. In line with the former notion, a compensatory role of MAP1A has been described in these mice (Harada et al, 1994).
In addition to the dendritic alterations, the absence of Tau affected newborn granule neuron PSDs. Tau knockout specifically reduced not only the number but also the area of most distal PSDs (located on 5th branching order dendrites) in these neurons. In addition, it decreased the percentage of mushroom spines and the spine head diameter in the aforementioned dendrites. These data might be relevant given that the main afferent pathway in the DG is composed by the axons of the pyramidal neurons of the entorhinal cortex (EC) (the perforant pathway). This pathway preferentially ends in the two outer thirds of the ML, where the 5th branching order dendrites of the granule neurons are located. Moreover, Tau deficiency also triggered ultrastructural alterations in the afferent synapses of granule neurons. An enlargement in both the synaptic cleft and the depth of the PSDs was detected. This observation may indicate the destabilization of these elements caused by Tau deficiency. In addition, we showed that absence of this protein also alters the ultrastructure of MFTs (efferent synapses), which may have important consequences for granule neuron functionality.
Indeed, electrophysiological recordings showed that the granule neurons of Tau−/− mice exhibit a hyperpolarized resting membrane potential. Moreover, with respect to synaptic strength, we observed a bidirectional effect in Tau‐deficient neurons, in which smaller mEPSCs got larger, and conversely, the larger mEPSCs got smaller. The molecular basis for these alterations is unclear, but they are indicative of a redistribution of synaptic weights, which is consistent with the bidirectional effect of PSD size that we also observed in these neurons. In fact, Tau is present in dendritic spines (Ittner et al, 2010; Mondragon‐Rodriguez et al, 2012; Frandemiche et al, 2014). In these structures, it has been shown that activation of NMDA glutamate receptor triggers Tau phosphorylation, thus regulating the interaction of Tau with the actin cytoskeleton, PSD95, and Fyn kinase (Ittner et al, 2010; Mondragon‐Rodriguez et al, 2012; Frandemiche et al, 2014). Interestingly, the basal subtle electrophysiological alterations described in this work seem to occur specifically in granule neurons, as it has been demonstrated that Tau deficiency does not affect basic synaptic currents in other regions of the brain (Ittner et al, 2010). Thus, it can be hypothesized that granule neurons are selectively affected by Tau deficiency in terms of their basal synaptic activity. In addition, alterations that may be related to deficient synaptic plasticity (LTP and LTD) have been reported in Tau−/− mice (Ahmed et al, 2014; Kimura et al, 2014; Regan et al, 2015). In fact, hippocampal LTD is considered necessary for clearing old memories, and Tau−/− mice show impairments in learning flexibility and various types of hippocampal‐dependent memory (Ikegami et al, 2000; Ahmed et al, 2014; Lei et al, 2014; Ma et al, 2014; Regan et al, 2015). Remarkably, it has been described that granule neurons are important for flexibility of learning strategies (Garthe et al, 2009).
Tau mediates the deleterious effects of acute stress on newborn neurons
Tau absence confers neuroprotection in several models of neuronal damage, such as traumatic brain injury (Cheng et al, 2014), neuroinflammation (Maphis et al, 2015), amyloid β‐mediated excitotoxicity (Roberson et al, 2007, 2011; Ittner et al, 2010; Vossel et al, 2010), and epilepsy (Holth et al, 2013), among others. Given that acute stress negatively affects AHN (Gould et al, 1992) and considering the neuroprotection exerted by the lack of Tau in the aforementioned models, we addressed whether the Porsolt test would cause similar detrimental effects in Tau−/− and WT mice.
The Porsolt test causes selective cell death of 1‐week‐old newborn neurons but does not cause a net increase in general apoptosis in the DG (Llorens‐Martin & Trejo, 2011). We found that the reduction of newborn neuron survival was less pronounced in Tau−/− compared to WT mice, thereby suggesting a certain degree of neuroprotection against neuronal death triggered by forced swimming in the former. Of note, this phenomenon is not attributable to an ineffectiveness of the Porsolt test on Tau−/− mice, as immobility time scores were similar in both genotypes. Interestingly, cell death triggered by forced swimming is mediated by glutamate excitotoxicity (Moghaddam et al, 1994; Nair & Bonneau, 2006), a phenomenon that Tau−/− neurons are protected against (Ittner et al, 2010).
In addition, we also found that while the Porsolt test decreased the total dendritic length and branching of WT neurons, it caused no alterations to the morphology of Tau−/− neurons. Moreover, forced swimming reduced the number of PSDs in WT mice, while it did not cause remarkable variations in Tau−/− animals. Indeed, although forced swimming reduced the average area of the MFTs in WT neurons, it did not cause this effect in Tau−/− ones. These data are in agreement with the relevant role played by Tau at the presynaptic compartment (Decker et al, 2015; Sokolow et al, 2015). Altogether, our findings support the above‐indicated notion that newborn granule neurons are, to some extent, protected from the detrimental consequences of the Porsolt test.
One of the mechanisms triggered by this test is the decrease in the levels of GluR1 subunit of AMPA receptors (Groc et al, 2008; Llorens‐Martin & Trejo, 2011). Given the growing relevance of AMPA receptor‐modulating compounds (known as AMPAkines) in the treatment of major depression (Bai et al, 2003; Arai & Kessler, 2007), we addressed whether the reduction in AMPA receptor expression caused by acute stress occurs in both genotypes in the same manner. Indeed, this effect was not detected in Tau−/− animals. In this regard, Regan et al (2015) demonstrated that Tau−/− mice present basal deficits in AMPA internalization. This observation is in agreement with our results, since we observed unchanged levels of GluR1 in these mice after the Porsolt test. Given that AMPA receptor internalization triggers dendritic spine disappearance (Collingridge et al, 2004), it can be hypothesized that the impairment in this mechanism explains the unchanged number of PSDs in Tau−/− mice. Furthermore, it could be related to the neuroprotection shown in newborn neurons against forced swimming.
Tau regulates the effects of EE on adult hippocampal neurogenesis
Environmental enrichment is one of the strongest positive regulators of AHN (Kempermann et al, 1997). Given the alterations observed in newborn neurons of Tau−/− mice, we examined whether EE would trigger similar stimulatory effects on AHN in both genotypes. EE triggered a similar increase in the number of DCX+ neuroblasts in WT and Tau−/− mice, thereby demonstrating that EE causes a general stimulatory effect in the hippocampus. However, the stimulatory effect of EE in 8‐week‐old newborn neuron survival in Tau−/− mice was less pronounced than in WT animals. In this regard, it has been demonstrated that some of the neuroprotective actions of EE are exerted through an increase in BDNF (Hu et al, 2013), and the actions of which are mediated by GABA (Waterhouse et al, 2012). Interestingly, we observed increased GABAergic innervation in the ML of the WT hippocampus in response to EE but not in that of Tau−/− animals.
Furthermore, analysis of the morphology of Tau−/− newborn neurons suggested an impaired capacity to detect the increased demand for information processing caused by EE. This stimulus increased total dendritic length and branching in WT newborn neurons, a phenomenon previously described (Llorens‐Martin et al, 2013). However, no increase in these parameters was observed in enriched Tau−/− mice. Thus, although fully mature newborn granule neurons did not differ morphologically between the two genotypes under CH conditions, we propose that the cytoskeleton of Tau−/− neurons lack, to some extent, the plasticity needed to adapt to the stimulatory actions of EE.
It is noteworthy that one of the most challenging features of EE is that both permanent elements (such as the size and shape of the cage) and similar but changing ones (interchangeable objects and toys) must be constantly remembered and or/newly memorized by the animal. Given that the whole environment is changed periodically during the enrichment period, both the formation and clearance of new memories are probably required. In line with this, the lifetime of hippocampal synapses matches the longevity of hippocampal memories (Attardo et al, 2015), and EE increases both LTD and LTP (Artola et al, 2006).
Importantly, EE resulted in a net increase in the density and a reduced area of PSDs in WT newborn neurons. It has been suggested that this phenomenon is caused by the resulting addition of newly generated, small, synaptic contacts (Llorens‐Martin et al, 2013). In contrast, EE reduced the number and size of the PSDs located on the 5th branching order dendrites of Tau−/− neurons, which were already reduced under CH conditions in comparison with those of WT mice. Given that Tau−/− mice present impairments in both LTP (Ahmed et al, 2014) and LTD (Kimura et al, 2014), the reduction in the number and size of 5th branching order PSDs in response to EE may be the consequence of two events. On one hand, due to inefficient LTP generation, EE might not be able to trigger a net increase in the formation of new PSDs. On the other hand, due to deficient LTD induction, the reduced clearance of old synaptic contacts may harness their replacement by newly generated synaptic contacts, a phenomenon that controls newborn neuron integration into the hippocampal circuit (Toni & Sultan, 2011).
Tau is translocated to the postsynaptic compartment of dendrites in response to synaptic stimulation (Mondragon‐Rodriguez et al, 2012; Frandemiche et al, 2014). Moreover, it has been suggested that the presence and interaction of microtubules at synaptic spines is crucial for the BDNF‐triggered addition of new synaptic contacts during development (Gu et al, 2008) and that microtubule dynamics control the shape and maturation of synaptic spines (Hu et al, 2008; Jaworski et al, 2009). Thus, it is reasonable to assume that the putative alterations of microtubule cytoskeleton plasticity, as well as the weakness of PSDs caused by the absence of Tau, may be related to the impaired enhancement of synaptic plasticity triggered by EE in Tau−/− mice. In addition, analysis of newborn neuron MFTs revealed that while EE increased the average area of MFTs in WT mice, it did not cause any change in Tau−/− neurons. These observations would suggest that Tau−/− presynaptic structures also show impaired plasticity.
Taken together, our data suggest that Tau plays a novel role in the maturation of newborn granule neurons in vivo under basal conditions. Furthermore, we provide evidence that Tau regulates the effects of external stimuli on AHN. In addition to other forms of neuroprotection caused by the absence of Tau, here we show that this protein mediates the deleterious effects of acute stress on AHN. Furthermore, we demonstrate, for the first time, that Tau modulates the stimulatory actions of EE on newborn neurons. All together, these data shed light on novel functions of Tau protein related to the plastic modulation of AHN by external stimuli, which, to the best of our knowledge, have not been demonstrated for any other protein in the AHN model.
Materials and Methods
Tau−/− mice were generated as previously described (Dawson et al, 2001). Heterozygous (Tau+/−) mice were crossed in order to obtain homozygous Tau knockout mice (Tau−/−) and control littermates (WT). Mice were housed at the Centro de Biología Molecular “Severo Ochoa” animal facility under standard laboratory conditions, in accordance with European Community Guidelines (directive 86/609/EEC), and were handled complying with European and local animal care protocols. Six to ten animals per experimental condition and time point were used for histological and biochemical determinations. Three animals per experimental condition and time point were used for retroviral labeling, electrophysiological recordings, and electron microscopy analyses. Animals were sacrificed at 4 months of age.
Administration of thymidine analogs
The thymidine analogs CldU and IdU (Sigma‐Aldrich, St. Louis, MO) were used to analyze the survival of 1‐, 2‐, 4‐, 6‐, and 8‐week‐old newborn granule neurons. CldU was administered in a single intraperitoneal injection of 42.75 mg/kg. IdU was administered over 24 h diluted in drinking water at 0.92 mg/ml. These doses were based on equimolar doses of 50 mg/kg (Llorens‐Martin et al, 2010) and 0.8 mg/ml BrdU (Lugert et al, 2010), respectively.
Retroviral stock preparation
We used a retroviral stock (PSD95‐GFP) encoding for PSD95 fused to GFP (Kelsch et al, 2008). The PSD95‐GFP retrovirus allowed PSD visualization (green channel). Moreover, anti‐GFP immunohistochemistry (red channel) allowed visualization of the whole neuronal morphology (Kelsch et al, 2008). The plasmids used to produce the PSD95‐GFP retrovirus were kindly provided by Prof. Carlos Lois (University of Massachusetts). Retroviral stocks were concentrated to working titers of 1 × 107–2 × 108 pfu/ml by ultracentrifugation (Zhao et al, 2006). Since the retroviruses used are engineered to be replication incompetent, only dividing cells at the time of surgery can be infected (Zhao et al, 2006; Kelsch et al, 2008).
Mice were anesthetized with isoflurane and placed in a stereotaxic frame. Coordinates (mm) relative to bregma in the anteroposterior, mediolateral, and dorsoventral axes were as follows: dentate gyrus (DG) [−2.0, 1.4, 2.2]. Two μl/DG of virus solution was infused at 0.2 μl/min via a glass micropipette. Mice were 8 weeks old at the time of retroviral injections.
Environmental enrichment (EE)
We used a previously described EE protocol (Llorens‐Martin et al, 2010). Enriched mice were housed in groups of 10 animals for 8 weeks in large transparent polycarbonate cages (55_33_20 cm, Plexx Ref. 13005). All enriched cages were equipped with various types of running wheels. The mice had free access to toys of different shapes, sizes, materials, and surface texture. Every other day, a set of 10–15 different toys and new bedding were placed in the cages in order to change the whole environment, as previously described (Llorens‐Martin et al, 2010).
Forced swimming (the Porsolt test)
Animals were placed in a 12‐cm‐diameter and 29‐cm‐tall cylinder filled with water at 23°C for 6 min on two consecutive days (Detke et al, 1995). Acute stress induced by forced swimming led to depressive‐like behavior in mice.
Morphometric analysis of newborn granule neurons
Three series of sections from each PSD95‐GFP retrovirus‐injected animal were used for the immunohistochemical detection of GFP. Thirty randomly selected neurons from each experimental condition were reconstructed under a LSM710 Zeiss confocal microscope (25× oil immersion objective). Confocal stacks of images were obtained (Z‐axis interval: 2 μm), and Z‐projections were analyzed for the determination of total dendritic length and Sholl's analysis. All cells were traced using NeuronJ plugin for ImageJ software. Sholl's analysis was performed using the plugin ShollAnalysis for ImageJ. This analysis consists on placing a central point on the cell soma and tracing concentric circles (separated by a 10‐μm distance interval). The number of times that the dendritic tree intersects each circle is represented in the graphs (number of crossings).
Number and area of PSDs of newborn granule neurons
The number and area of PSD95‐GFP+ clusters were analyzed separately for each branching order in the dendritic tree of retrovirus‐labeled newborn neurons. A minimum of 30 segments belonging to each experimental condition were analyzed for each branching order. Confocal stacks of images were obtained in a LSM710 Zeiss confocal microscope (63× oil immersion objective; XY dimensions: 67.4 μm; z‐axis interval: 0.2 μm). Two‐channel stack Z‐projections were obtained. The dendritic length of each segment was measured (red channel, GFP), and the number and area of PSD‐GFP+ clusters were analyzed using the semi‐automatic Particle Analyzer plugin for ImageJ (green channel, PSD95).
Area of MFTs of newborn granule neurons
The area of individual MFTs labeled with GFP was measured in the CA3 region of PSD95‐GFP retrovirus‐injected animals. A minimum of 20 stacks of images per experimental condition were obtained in a LSM710 Zeiss confocal microscope (63× oil immersion objective; XY dimensions: 100 μm; Z‐interval: 0.5 μm). Stacks were randomly obtained among the different sections composing the series. GFP‐Z‐projections were obtained, and the area of each MFT was measured manually in ImageJ software, as previously described (Toni et al, 2008). The following criteria were selected for MFT quantification: (i) the diameter of the MFT is more than threefold greater than the diameter of the axon; (ii) the MFT is connected to the axon on at least one end; and (iii) the MFT is relatively isolated from other MFTs for accuracy of tracing (Toni et al, 2008). Given that the MFTs of the CA2 region are significantly smaller than those of the CA3 (Llorens‐Martin et al, 2015), we avoided the CA2 region when obtaining the images. All the measurements were made in equivalent rostrocaudal positions of the CA3 field that were always closer to the hilus than to the CA2 region. A minimum of 100 MFTs per experimental condition were measured.
Statistical analysis was performed using the SPSS 22 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. For comparisons between 2 experimental groups, date from cell counts (PH3, fractin, Sox2, DCX, thymidine analog survival, BLBP, and calretinin), morphometric analysis (total dendritic length and Sholl's analysis), analysis of the PSDs in newborn neurons under a confocal microscope (density and area of PSDs), analysis of dendritic spines under a confocal microscope (density and head diameter of dendritic spines), electron microscopy analysis (number of presynaptic vesicles, length of the active zone, synaptic cleft size, PSD length, depth, and area), and electrophysiological recordings were analyzed by a Student's t‐test in the case of normal sample distribution or by a nonparametric test (Mann–Whitney U‐test) in those cases in which normality could not be assumed. For comparisons between more than 2 experimental groups (Western blot quantifications, thymidine analog experiments, cell counts, GABAergic innervation, morphometric analysis, and analysis of the PSDs when analyzed in either stressed or enriched mice), data were analyzed by a two‐way ANOVA test. In those cases in which the two‐way ANOVA interaction was statistically significant, a one‐way ANOVA followed by a Fischer LSD post hoc analysis was used to compare the differences between individual groups. Graphs represent mean values ± SEM. The percentage of the different types of dendritic spines of newborn neurons, as well as the percentage of cells expressing DCX and NeuN, were analyzed by a chi‐squared (χ2) test. The comparison between the cumulative distributions of amplitudes of mEPSCs, as well as the comparison between the distributions of MFT sizes, was performed by a Kolmogorov–Smirnov Z‐test.
A detailed methodological description of the sacrifice and tissue processing, immunohistochemistry, cell counting, analysis of GABAergic innervation, morphometric analysis of the dendritic spines of newborn granule neurons, electron microscopy, Western blot, and electrophysiological recordings are provided in Appendix Supplementary Methods.
NPB performed and supervised all the experiments. MLLM performed the electron microscopy experiments. JJA performed the retroviral injections. MN performed and analyzed the electrophysiological experiments. JAE supervised the electrophysiological experiments. MLLM, JA, FH, MN, and JAE obtained funding. NPB, MLLM, and JA wrote the manuscript.
Conflict of interest
The authors declare that they have no conflict of interest.
Expanded View Figures PDF
This work was supported in part by grants from the Alzheimer's Association (2015‐NIRG‐340709) (M. Llorens‐Martín), the Ministerio de Economia y Competitividad (SAF‐2014‐5040‐P, J. Ávila; BUF2013‐40664‐P, F. Hernández; SAF2014‐58598‐JIN, M. Navarrete; SAF2014‐57233‐R, Jose A. Esteban), and from the Centro de Investigación Biomédica en Red sobre Enfermedades Neurodegenerativas (CIBERNED, ISCIII) (J. Ávila). The authors thank E. García for help producing retroviral vectors, C. Lois for providing the plasmids used for the production of the PSD95‐GFP virus, and R. Cuadros for the technical assistance with WBs and electron microscopy. The authors also thank the “Servicio de Microscopía Óptica y Confocal” of the CBMSO for their help with confocal microscopy.
FundingAlzheimer's Associationhttp://dx.doi.org/10.13039/100000957 2015‐NIRG‐340709
[The copyright line of this article was changed on 17 June 2016 after original online publication.]
This is an open access article under the terms of the Creative Commons Attribution‐NonCommercial‐NoDerivs 4.0 License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made.
- © 2016 The Authors. Published under the terms of the CC BY NC ND 4.0 license