Pathologically phosphorylated tau at S396/404 (PHF-1) is accumulated inside of hippocampal synaptic mitochondria of aged Wild-type mice

Brain aging is a natural process characterized by cognitive decline and memory loss. This impairment is related to mitochondrial dysfunction and has recently been linked to the accumulation of abnormal proteins in the hippocampus. Age-related mitochondrial dysfunction could be induced by modified forms of tau. Here, we demonstrated that phosphorylated tau at Ser 396/404 sites, epitope known as PHF-1, is increased in the hippocampus of aged mice at the same time that oxidative damage and mitochondrial dysfunction are observed. Most importantly, we showed that tau PHF-1 is located in hippocampal mitochondria and accumulates in the mitochondria of old mice. Finally, since two mitochondrial populations were found in neurons, we evaluated tau PHF-1 levels in both non-synaptic and synaptic mitochondria. Interestingly, our results revealed that tau PHF-1 accumulates primarily in synaptic mitochondria during aging, and immunogold electron microscopy and Proteinase K protection assays demonstrated that tau PHF-1 is located inside mitochondria. These results demonstrated the presence of phosphorylated tau at PHF-1 commonly related to tauopathy, inside the mitochondria from the hippocampus of healthy aged mice for the first time. Thus, this study strongly suggests that synaptic mitochondria could be damaged by tau PHF-1 accumulation inside this organelle, which in turn could result in synaptic mitochondrial dysfunction, contributing to synaptic failure and memory loss at an advanced age.


Estimation of mitochondrial complex activity. The activity of the mitochondrial complex I and III
were estimated by measuring the produced amount of ROS and ATP production in hippocampal mitochondrial enriched preparations. Mitochondrial ROS production was measured using 25 μM of CM-H2DCFDA (DCF) (485 nm, 530 nm) in the Biotek Synergy HT plate reader as previously described 31,43,44 . Isolated mitochondria (25 μg of protein) were added to 100 μl of KCl respiration buffer with 5 mM pyruvate and 2.5 mM malate as oxidative substrates and incubated at 37 °C. ROS production was calculated as the maximum DCF fluorescence following 30 min of incubation, expressed in arbitrary fluorescence units. After ROS measurements, KCl respiration buffer containing mitochondria was centrifuged at 8000 g for 10 min at 4 °C and then the ATP concentration was measured in the supernatant using the luciferin/luciferase bioluminescence assay kit (ATP determination kit no. A22066, Molecular Probes, Invitrogen) 31,43-45 . Measurement of ROS content. ROS production was measured using the fluorescent dye CM-H2DCFDA (catalog number C6827, Thermo Fisher Scientific). Briefly, hippocampal samples, diluted in Respiration Buffer, were added to a black 96-well plate in duplicate followed by the addition of 25 μM DCF. Then, the plate was incubated for 5 min and read in BioTek Synergy HT (485 nm, 530 nm).
Measurement of ATP concentration. The ATP concentration was measured in the obtained hippocampal tissue lysates (ATP content in the hippocampus) with Triton buffer (5 mM Tris, 150 mM NaCl, 1 mM EDTA, 1% (v/v) Triton X-100, pH = 7.4). ATP produced by isolated mitochondria incubated with oxidative substrates was evaluated in the supernatant of isolated mitochondrial fractions. In both cases, ATP was measured using a luciferin/luciferase bioluminescence assay kit (ATP determination kit no. A22066, Molecular Probes, Invitrogen) 31,[43][44][45] . The amount of ATP in each sample was calculated from standard curves and normalized to the total protein concentration.
Hippocampal slices, staining with mitochondrial fluorescent dyes and immunofluorescence. The brains of 3 and 18 month-old mice were dissected and immediately frozen at − 150 °C. The frozen brains were mounted using Optimal cutting temperature compound (OCT compound) in a cryostat at − 22 °C, then coronal 25-µm-thick slices of unfixed hippocampal tissue were obtained. Hippocampal slices were mounted on glass slides and incubated as previously described with mitochondrial fluorescent dyes 43,46,47 . First, the slices were washed three times for 5 min in PBS and then incubated with MitoTracker Green FM to measure mitochondrial mass 43,46,47 and MitoTracker Red CM-H2Xros to determine mitochondrial membrane potential 43,46,47 . All these dyes were diluted in Krebs-Ringer-Hepes-bicarbonate (KRH) buffer (136 mM NaCl, 20 mM HEPES, 4.7 mM KCl, 1.5 mM MgSO 4 , 1.25 mM CaCl 2 , 5 mM glucose; pH = 7.4) and incubated for 45 min at 37 °C. After incubation, slices were washed three times for 5 min in PBS and mounted with fluorescent mounting media with DAPI (Vector Laboratories Inc, CA, USA). For immunofluorescence, the slices were fixed with Paraformaldehyde (PFA) 4% for 10 min. The primary antibody was incubated overnight (O.N) at 4 °C while the incubation with the secondary antibody was for 1 h at room temperature. Images were acquired on a TCS SP8 laser-scanning confocal microscope (Leica Microsystems, Wetzlar, Germany).
Synaptic and non-synaptic mitochondria isolation. Hippocampi from 3 and 18 month-old mice were dissected on ice in Isolation Medium (IM: 0,4 M sucrose, 150 mM mannitol, 2 mM EGTA, 10 mM HEPES, pH = 7.4) as previously described 31,48 . Hippocampi were lysed in a glass homogenizer in 7.5 ml of IM buffer. Proteinase K protection assay. Mitochondrial fractions of 18 month-old mice were treated with Proteinase K (0.5 or 0.25 μg/ml) in the presence of buffer, 0.1% digitonin, or 1% Triton X-100 for 15 min at room temperature. We observed the same result with both concentrations. The control condition was performed in the absence of Proteinase K. Later, the protease inhibitor was added to inactivate Proteinase K for 5 min, and then 1 × loading buffer was added before boiling the samples to perform an SDS-PAGE.
Image analysis. All slides were photographed and scanned under the same magnification, laser intensity, brightness, and gain. Images were processed using the Fiji software (NIH Image) 50 , adjusting the fluorescence threshold intensity in every picture.  Statistical significance was determined using one-way ANOVA with Bonferroni's post-test. p-values > 0.05 and ≤ 0.05 were regarded, respectively, as not statistically significant and as statistically significant. In the figures, p-values between 0.01 and 0.05 are marked with one asterisk, p-values between 0.001 and 0.01 with two asterisks, and p-values less than 0.001 are shown with three asterisks. All statistical analyses were performed using Prism software (GraphPad Software, Inc.).
Ethical approval and consent to participate. The experimental procedures were approved by the Bioethical and Biosafety Committee of Universidad San Sebastián.

Results
Oxidative balance and mitochondrial bioenergetics function are impaired during aging. The free radical theory of aging relies on the fact that biomolecules can be damaged by ROS accumulation, leading to functional cell loss 51 . In the brain, mitochondria are one of the main producers of ROS through complexes I and III 11 therefore, an age-associated mitochondrial dysfunction can lead to increased ROS production and cellular impairment. Here, we evaluated the oxidative damage in the hippocampus of adult mice (3 month-old) and old mice (18 month-old). We assessed the levels of a known oxidative marker 4-Hydroxynonenal (4HNE), which is the product of lipid peroxidation of lipoproteins, by immunofluorescence ( Fig. 1a) and western blot (Fig. 1b).
In both cases, we observed increased 4HNE levels in the hippocampus of 18 month-old (mo) mice ( Fig. 1a,b). Specifically, immunofluorescence assays revealed an increased 4HNE signal in the three studied hippocampal regions: dentate gyrus (DG), cornu ammonis 1 (CA1), and cornu ammonis 3 (CA3) (Fig. 1a). We measured ROS content in the hippocampal lysate, using the fluorescent dye CM-H2DCFDA, and we also observed increased ROS levels in aged mice (Fig. 1c), indicating oxidative stress in the hippocampus during normal aging. We then evaluated the bioenergetic mitochondrial function in the hippocampus of 3 and 18 mo mice. We measured the mitochondrial membrane potential in non-fixed hippocampal slices using the fluorescent dye Mitotracker Red CM-H2Xros, a highly sensitive dye to changes in mitochondrial potential 43,46,47 . We observed decreased fluorescence in DG, CA1, and CA3 hippocampal regions of 18 mo mice (Fig. 1d), indicating depolarization of the mitochondria in the hippocampus of mice at an advanced age. In addition, we evaluated the expression of the oxidative phosphorylation complexes in samples of 3 and 18 mo mice by western blot. Our results reveal significantly decreased levels of complex I and complex IV, accompanied by increased levels of complex V in hippocampal samples of aged mice (Fig. 1e). Finally, we evaluated the ATP content in the hippocampal lysate of 3 and 18 mo mice and observed a significant decrease in ATP in old mice (Fig. 1f). Altogether, these results suggest impaired bioenergetic function of hippocampal mitochondria during aging, which could at least partially explain the oxidative damage present in the aged hippocampus.
Severe structural alterations in the mitochondria during normal aging. Mitochondria are double-membrane organelles organized as an interconnected network 52 . The mitochondrial bioenergetic function is related to mitochondrial morphology 31,53 ; and therefore impaired mitochondrial integrity can, in turn, affect their activity; leading to degenerative processes 31,54 . Since old mice have impaired mitochondrial function, we evaluated the mitochondrial structure in the CA1 hippocampus of 3 and 18 mo mice by transmission electron microscopy (TEM) (Fig. 2). We evaluated mitochondrial swelling, mitochondrial intact membranes, average mitochondrial area, and the number of synaptic mitochondria (Fig. 2), using different parameters as previously described 49 . Our results showed an increased percentage of swollen mitochondria (Fig. 2ai,bi) and a decreased percentage of mitochondria with an intact membrane in 18 mo mice compared with 3 mo mice (Fig. 2aii,bii). We also observed increased mitochondrial area in aged mice (Fig. 2aiii,biii). Interestingly, when we evaluated the number of synapses containing both pre-synaptic and post-synaptic mitochondria, we observed no significant difference between 3 and 18 mo mice (Fig. 2aiv,biv): we obtained a similar observation when we compared the number of synapses containing only mitochondria in either the pre-synaptic or post-synaptic region (Fig. 2bv,bvi). Taken together, these results indicate that mitochondria of the CA1 hippocampus from 18mo mice lose their structure with no changes in their distribution throughout the synapses. Therefore, this could be directly related to the negative changes observed in the bioenergetic function.
Phosphorylated tau at the PHF-1 epitope is increased in the hippocampus of aged mice. Considering that abnormal forms of tau produce mitochondrial dysfunction 55 , we evaluated phosphorylated tau at the PHF-1 epitope (Fig. 3). This phosphorylation is present in different tauopathies 21 . Therefore, we evalu-  www.nature.com/scientificreports/ ated PHF-1 expression using a specific antibody that simultaneously recognizes both phosphorylated residues (Fig. 3a) in the hippocampus of 3 and 18 mo mice by western blot in a whole lysate (Fig. 3b) and through immunofluorescence (Fig. 3c). Interestingly, we observed a drastic increase in the tau PHF-1 levels, with no changes in total tau levels in old mice (18 m) compared with 3 mo adult mice (Fig. 3b). Consistently, we found increased tau PHF-1 positive signals in the three analyzed hippocampal regions (DG, CA1, and CA3) of aged mice (Fig. 3c), suggesting that this modified form of tau accumulates in the hippocampus during aging. Next, we evaluated the possible localization of tau PHF-1 in the hippocampus. For this, we performed a co-immunofluorescence for Tom70 (mitochondrial marker) and tau PHF-1 in the hippocampal slices. Surprisingly, we observed co-localization of tau PHF-1 with the mitochondria, as indicated by white arrows in 63 × images in Fig. 3d. This last result suggests that tau PHF-1 accumulation during aging could occur in the hippocampal mitochondria.

Tau phosphorylation at Ser396/404 (PHF-1) is increased in a mitochondrial fraction from the hippocampus of 18 month-old mice. Different forms of tau induce mitochondrial dysfunction 35 and
in Alzheimer's Disease tau overexpression has been observed in the mitochondria 56 . However, it is currently unknown if this protein is located in the mitochondria under physiological conditions in vivo during normal aging. To corroborate that tau PHF-1 accumulates in hippocampal mitochondria of old mice, we obtained a fraction enriched in hippocampal mitochondria of 3 and 18 mo mice, using a sucrose buffer and differential centrifugation, as indicated in the scheme of Fig. 4a. To assess if the mitochondrial fraction was correctly isolated, and considering that tau protein has also been described in the nucleus 57 , we performed an immunoblot assay for a nuclear protein (Lamin A) and for mitochondrial resident proteins such as OXPHOS complexes (Fig. 4b), confirming the purity of this mitochondrial fraction. In this mitochondrial fraction, we evaluated the tau PHF-1 and total tau expression (Fig. 4c). We observed that tau PHF-1 levels are significantly increased in the mitochondrial fraction of aged mice compared to 3mo mice (Fig. 4c,d), without significant differences in total tau levels between both ages (Fig. 4c,d). This supports our idea that this phosphorylated form of tau accumulates in the mitochondria during normal aging. Then, we evaluated the bioenergetic function of isolated mitochondria from the hippocampus of 3 and 18 mo mice, measuring ROS and ATP production after the addition of the oxidative substrates pyruvate and malate (Fig. 4e). We observed no significant differences in ROS levels in both age groups; but interestingly our results revealed a significantly reduced ATP production (Fig. 4f), indicating deficient mitochondrial oxidative phosphorylation during aging, which could be mediated, almost in part, by the presence of tau PHF-1 in the mitochondria. In neurons, two mitochondrial populations with different susceptibility to aging-dependent dysfunction have been described 33 , known as synaptic and non-synaptic mitochondria according to their localization. We evaluated if the accumulation of phosphorylated tau PHF-1 is differential in these two populations. For this, we used a Percoll gradient to isolate non-synaptic mitochondria and synaptic mitochondria from synaptosomes of the hippocampus of adult and old mice (Fig. 4g). Interestingly, when we evaluated the tau PHF-1 expression in both mitochondrial populations, we observed that tau PHF-1 is increased mostly in the synaptic mitochondria of aged mice, with a weak phosphorylated tau signal in the non-synaptic mitochondria, which became more evident with longer exposure time (Exp. 2) (Fig. 4h). Also, we observed that the total tau signal was similar between 3 and 18 month-old mice in both synaptic and non-synaptic mitochondria. Altogether, these results strongly suggest that tau PHF-1 accumulation in the hippocampus of aged mice occurs mainly in synaptic mitochondria.
Tau PHF-1 accumulates inside synaptic mitochondria of aged wild-type mice. As mentioned previously, tau PHF-1 appears to accumulate in synaptic hippocampal mitochondria during normal aging. To corroborate these findings, we performed an immunogold electron microscopy (IEM) assay in samples of the CA1 region from the hippocampus of 3 and 18 mo mice (Fig. 5). First, we evaluated the total tau PHF-1 positive signal within one electron microscopy grid (7 mm 2 ) (Fig. 5ai,bi). Consistent with our prior results, we observed a drastic increase in the tau PHF-1 signal in old mice (Fig. 5ai,bi). Next, we analyzed the tau PHF-1 positive signal located in the mitochondria (Fig. 5aii,bii). In fact, our results revealed that tau PHF-1 was increased in the mitochondria during aging (Fig. 5aii,bii), supporting our biochemical assays. Finally, we analyzed the amount of tau PHF-1 positive signal in synaptic mitochondria and as expected, aged mice showed higher levels of tau PHF-1 in synaptic mitochondria compared with 3 mo mice (Fig. 5aiii-iv and 5biii-v). Specifically, we observed an increase in the number of both pre-synaptic and post-synaptic mitochondria that were positive for tau PHF-1 in old animals compared with 3 mo animals ( Fig. 5biv and 5bv). Interestingly, the tau PHF-1 signal in the mitochondria was observed in the periphery and in the center of this organelle, strongly suggesting that PHF-1 can enter the mitochondria, locating in the intramembranous space or the mitochondrial matrix (Fig. 5aiii-iv). Finally, to validate that tau protein PHF-1 localizes inside the mitochondria during normal aging, we performed . Phosphorylated tau at Ser396/404 (PHF-1) epitopes increased during aging in the hippocampus. (a) Schematic representation of tau protein and the phosphorylation at Ser404/396 residues, epitope known as PHF-1. Created with BioRender.com, (b) Western blot of tau PHF-1 in hippocampal tissue of 3 and 18-month-old (mo) mice. Densitometric analysis is shown as relative values to the control. (c) Representative immunofluorescence images for PHF-1 tau DG, CA1, and CA3 hippocampal regions, with its quantitative fluorescence analysis. (d). Double staining for Tom70 (mitochondrial marker; green) and PHF-1 (Red) to analyze the colocalization of phosphorylated tau with mitochondria, in DG, CA1 and CA3 hippocampal regions. White arrows show co-localization between PHF-1 tau and mitochondria.The images were taken using a (c) 20 × objective and (d) 63 × objective. n = 3 different animals of each age. Graph bars represent means ± SEM. *p < 0.05. **p < 0.01; ***p < 0.001. DG Dentate Gyrus, CA1 Cornu ammonis 1 and CA3 Cornu Ammonis 3.  www.nature.com/scientificreports/ a proteinase K protection assay in a mitochondrial fraction of 18 mo mice (Fig. 5c,d). Briefly, we used Digitonin for solubilizing the outer mitochondrial membrane (OMM) and Triton X-100 for solubilizing both the inner (IMM) and outer mitochondrial membrane (OMM). Thus, we analyzed the mitochondrial internal localization of tau PHF-1. In a first assay, we charged the full volume of the obtained mitochondrial fraction and we observed a positive signal for tau PHF-1 only with proteinase K (line 2), an effect that was not observed in presence of Proteinase K plus Digitonin (line 3) or Triton X 100 (line 4) (Fig. 5c). These results suggest that tau PHF-1 is located mainly in the intermembranous space (IMS) of the mitochondria. To corroborate that the absence of signal in line 3 is not because the amount of PHF-1 tau in the mitochondrial matrix is too low, we performed a second experiment where we charged 1/3 of the full volume in the 2 first lines, and the full volume in the last 2 lines (Fig. 5d). Interestingly, and as expected, we observed tau PHF-1 signal in line 3 (digitonin + proteinase K), indicating that PHF-1 is also present in the mitochondrial matrix but in a much smaller proportion. This result is important because it demonstrates that tau PHF-1 is localized inside the hippocampal mitochondria of aged mice. Taken together, the results of this study showed that phosphorylated tau at the PHF-1 epitope is increased in the hippocampus of mice during normal aging. Even more, we demonstrated for the first time that tau PHF-1 accumulates in the mitochondria of aged wild-type mice, localizing specifically inside of the synaptic mitochondria from the hippocampus at an advanced age.

Discussion
In the present study, we demonstrate that oxidative stress and mitochondrial alterations occur simultaneously to increased phosphorylated tau at PHF-1 in the hippocampus of aged WT mice. We observed that tau PHF-1 accumulates in a mitochondrial fraction, mostly into synaptic mitochondria, demonstrating that this pathological form of tau PHF1 is located inside of the synaptic mitochondria during normal aging. These results strongly suggested that age-related synaptic impairment could be due to the accumulation of phosphorylated tau PHF-1 in the synaptic mitochondria and propose a mechanism that explains why synaptic mitochondria are more vulnerable to damage during aging 31 .
Aging is the primary risk factor for neurodegenerative diseases that currently lack early diagnosis and cures 58 . This process leads to the decline of cognitive abilities 59 , reducing the quality of life of elderly persons, whose population has increased over the last years. However, the cellular and molecular mechanisms leading to ageassociated alterations are not completely clear 60 . Increasing evidence shows that synaptic dysfunction 61 and mitochondrial impairment 62-64 are considered early events during aging and in the pathogenesis of neurodegenerative disorders 63 . Also, mitochondrial dysfunction is considered a hallmark of aging 6,63 , because its functional deterioration contributes to the aged phenotype 63,65 . Considering the high-energy requirements of neurons to adequately perform all their functions in the brain, it is logical to hypothesize that an energy imbalance could lead to neurodegenerative pathologies. The main energy producers (ATP) in cells are mitochondria, which maintain the bioenergetic homeostasis of neurons 66 . Also, mitochondria form ROS, which damages neurons when produced in excess 67,68 leading to the "free-radicals theory of aging" 69 ; but this could be a direct effect of mitochondrial dysfunction. Defects in mitochondrial bioenergetics, as well as disturbances in redox homeostasis, are characteristic signs of neuronal damage in the aged brain. Thus, increased ROS production may trigger a "vicious cycle" of oxidative stress, leading to more severe mitochondrial dysfunction, which contributes to aging 10 . In this context, in this study we showed several impairments in hippocampal mitochondria of aged WT mice, such as increased oxidative stress and bioenergetic impairment, where the increased oxidative stress observed in aged mice could contribute to the impairment of the ETC functioning. We observed a decrease in the expression of the complex I and IV, which are the two complexes that have more subunit encoded in the mtDNA 70 . Thus, the increase in oxidative stress present in the aged hippocampus could induce mtDNA damage, decreasing the expression of these two complexes and leading to the impairment in the ATP production of the aged mitochondria that we also reported. Moreover, we show structural alterations in aged mitochondria of WT mice (increased mitochondrial area and swelling, and decreased membrane integrity), corroborating that aged hippocampus from C57BL/6 mice manifest mitochondrial alterations related to dysfunction in the absence of pathology. The reasons that lead to this dysfunction are still a matter of study, but one possibility is the accumulation of abnormal proteins in the neurons.
Effectively, during aging, accumulations of specific proteins are observed in the brain and can be drastically increased in neurodegenerative disorders 15,71 . Among these accumulations are those formed by tau, a neuronal protein that is involved in tubulin polymerization and microtubule stabilization 72,73 . Tau can be phosphorylated at serine 85 or different threonine sites, many of which are important for regulating its function, and others are associated with pathology 74 . Here, we showed that the levels of a modified form of tau protein are increased in the normal brain. Specifically, we demonstrated that phosphorylated tau PHF-1 accumulates in the hippocampus of aged WT mice. Interestingly, it is reported that oxidative stress contributes to tau phosphorylation in PHF-1 and other epitopes in a mechanism that is still unknown 75,76 . It is reported that oxidative stress can induce tau phosphorylation through the activation of the glycogen synthase kinase-3β (GSK-3β) 77 and the activation of p38 MAPK 78 . Since PHF-1 is an epitope that is reported to be phosphorylated by GSK-3β 79 , the age-related increased oxidative stress reported here may activate GSK-3β kinase, contributing to higher tau PHF-1 levels in the hippocampus, nevertheless, this hypothesis remains to be determined. In agreement with our results, tau phosphorylation at PHF-1 also has been shown as a characteristic of healthy aging in the brain, where the presence of phosphorylated tau increases with age in animals and non-demented patients, but this change is even greater in AD 80,81 .
When tau is abnormally phosphorylated it dissociates microtubules, losing its normal function and accumulating in neurons 60,82,83 . Tau hyperphosphorylation leads to aberrant self-assembly in insoluble aggregates, www.nature.com/scientificreports/ accompanied by synaptic dysfunction and neuronal death in a series of neurodegenerative diseases known as tauopathies 19 . In particular, phosphorylation of tau PHF-1 is sufficient to promote microtubule dissociation 84 and it can accumulate in other intracellular structures, contributing to its dysfunction. Interestingly, this is the first report that demonstrates that phosphorylated tau (PHF-1) can localize into hippocampal mitochondria of aged mice, through biochemical and electron microscopy analyses and the first to propose a possible mechanism to explain, almost in part, the mitochondrial dysfunction observed in the aged hippocampus. Consistent with this observation, diverse studies suggest that the accumulation of modified forms of tau in neurons disrupts mitochondrial function by an unknown mechanism 85,86 . Also, it was recently shown that tau interacts with several mitochondrial membrane-bound proteins including ATP synthase, mitochondrial creatine kinase U-type, and Drp1 41 . However, no study had shown direct evidence that tau, and specifically its phosphorylated form at the PHF-1 site, could be found in the aged mitochondria or pathological conditions. This is a novel and surprising finding that opens the possibility that phosphorylated tau (PHF-1) could be in some way responsible for the mitochondrial dysfunction reported here such as the increase in the redox imbalance and the bioenergetic deficits. This is in agreement with studies that showed that tau PHF-1 induces the Aβ-mediated loss of mitochondrial membrane potential 39 and that neuronal cultures from KO tau mice prevent this reduction in mitochondrial membrane potential 87 . In addition, tau deletion in WT mice improves mitochondrial function in non-aged mice 43 . Nevertheless, is not clear whether this disruption in mitochondrial function is a direct or indirect effect of tau. Considering that there is a close relationship between mitochondrial structure and function, another possibility is that the accumulation of tau PHF-1 promotes structural alterations reported in hippocampal mitochondria, favoring the presence of disrupted mitochondria. Thus, tau PHF-1 also could induce functional failure of synaptic mitochondria, mainly considering that if IMM and cristae are disrupted the ETC assembly can be affected; which in turn could lead to bioenergetic impairment. It is reported that in AD tau induces defective mitophagy through the decrease of the translocation of Parkin to mitochondria 88 , leading to the accumulation of defective mitochondria. Also, it is reported that phosphorylated tau alters the mitochondrial dynamics in AD, increasing the fission proteins by their interaction with Drp-1, and decreasing the fusion proteins 37 . These results suggest that tau affect mitochondrial morphology in pathological conditions and a similar effect could be occurring in the aging; however, whether tau PHF-1 induces structural changes by any of these mechanisms or it is directly responsible for mitochondrial dysfunction at an advanced age needs to be determined. Although in neurons tau is mainly located in axons, it is expressed in a lower amount in somatodendritic compartments, including the plasma membrane, nucleus, and significantly lower amounts in dendrites 74 . Tau mainly interacts with cytoskeletal proteins but also communicates either directly or indirectly with other protein types 41 . These include kinases and phosphatases, extracellular proteins, and membrane proteins 41 . Tau directly binds to Fyn kinase and PSD95 in the post-synaptic region, and this interaction is dependent on Tyr18 phosphorylation 74 . Also, tau can interact with synaptophysin, suggesting a role in the pre-synaptic region 41 . Therefore, it is evident that tau plays a role in the synapses, which is in agreement with our findings that tau PHF-1 is located in synaptic mitochondria.
Synapses are sites of high energy demand and calcium variations; therefore synaptic mitochondria are fundamental to maintain bioenergetic homeostasis at synapses, synaptic function, and memory formation 89 . Interestingly, hippocampal synaptic mitochondria are more sensitive to cumulative damage, and their dysfunction occurs previous to non-synaptic mitochondrial failure 31 . Synaptic mitochondria, that are obtained from a synaptosomal extract, come exclusively from neurons 28 ; then, alterations in these mitochondria are strictly restricted to neurons and not to another cellular type 10 . Synaptic mitochondria are usually punctuated and isolated, and they are synthesized in the neuronal soma and transported to the axon or dendrite 90 . As a result of this process, these mitochondria are considered older than mitochondria present in the soma of neurons, and they exhibit greater sensitivity to the damage caused by oxidative stress and bioenergetic failure 28 . This sensitivity occurs since synaptic mitochondria possess functional differences compared to non-synaptic mitochondria, such as lower calcium buffer capacity, high ROS production, and lower expression levels of mitochondrial respiratory complexes 91 . Interestingly, aging seems to enhance the differences between these two mitochondrial populations. For example, synaptic mitochondrial extracts of old rats 32 or mice 33 presented a significant reduction in mitochondrial respiration, and they were more susceptible to calcium stress 32 compared to mitochondria of young animals. However, why hippocampal synaptic mitochondria are more vulnerable to cumulative damage is still unknown. Interestingly, our results reveal that phosphorylated tau at PHF-1 is observed mainly in synaptic mitochondria in both non-aged and aged mice, an effect that is drastically increased in aged synaptic mitochondria. Thus, we demonstrated that increased tau PHF-1 levels at an advanced age are located more specifically into synaptic mitochondria and could be involved in the dysfunction of this mitochondrial pool. This is particularly important since the effect of tau PHF-1 accumulation in these mitochondrial population could impact directly Figure 5. Tau PHF-1 accumulates inside hippocampal synaptic mitochondria at an advanced age. (a) Immunogold for PHF-1 tau in the hippocampal CA1 region of 3 and 18 month-old mice. Representative images of immuno-gold electron microscopy (IEM) (43,000x) showing (i) the amount of PHF-1 positive signal within an area of the grid of 7 mm 2 ; (ii) amount of PHF-1 in the mitochondria, (iii) the number of PHF-1 tau in pre and (iv) post-synaptic mitochondria. (b) Quantitative analysis of IEM. (c). Western Blot of the proteinase K protection assay in a mitochondrial fraction of 18 mo mice. The mitochondrial fraction was treated with Proteinase K with or without 0.1% Digitonin or 1% Triton X-100. The mitochondrial fraction without Proteinase K was used as a control. (left, Assay 1) Western Blot of the total volume of mitochondrial samples. (Right, Assay 2) Western Blot of 1/3 of the volume of samples 1 and 2. n = 4 different animals. Graph bars represent means ± SEM. *p < 0.05, **p < 0.01.

Scientific Reports
| (2021) 11:4448 | https://doi.org/10.1038/s41598-021-83910-w www.nature.com/scientificreports/ in the synapse functioning. Here we show new evidence that is helpful to understand part of the mechanism underlying the previously reported age-related synapse impairment 92 . Tau PHF-1 accumulation inside synaptic mitochondria may play a crucial role in the morphological alterations in these population, which could explain its functional impairment during aging 31 , suggesting that this structural and functional impairment associated with phosphorylated tau accumulation in synaptic mitochondria is a key issue involved in memory loss during aging. More studies are necessary to validate this hypothesis.

Conclusion
Interestingly, our results reveal that tau PHF-1 is higher located in the periphery of the mitochondria, possibly bound to proteins of the OMM as suggested by interaction assays 41 , in the intermembrane space or bound to proteins of the IMM, with a minor proportion of tau PHF-1 in the mitochondrial matrix. This is particularly relevant because it reveals for the first time that tau PHF-1 can enter synaptic mitochondria during normal aging. Also, this study proposes that tau PHF-1 is located inside the synaptic mitochondria and could trigger the synaptic failure observed in the hippocampus during the early stages of AD and in pathological mouse models and patients. www.nature.com/scientificreports/ Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.