Introduction
Several neurodegenerative diseases, such as Alzheimer’s disease (AD), progressive supranuclear palsy (PSP), corticobasal degeneration (CBD), Pick’s disease (PiD), argyrophilic grain disease, and inherited frontotemporal dementia and parkinsonism linked to chromosome 17 (FTDP-17T) are characterized by the presence of abnormal intracellular filamentous protein inclusions that consist of hyperphosphorylated microtubule-associated protein tau and are collectively designated as tauopathies [
18,
26,
43]. The identification of mutations in the MAPT gene in FTDP-17T [
22,
43,
44] has established that dysfunction or misregulation of the tau protein is central to the neurodegenerative process in disorders with tau pathology. Furthermore, in AD it is the accumulation and dysfunction of tau that causes cell death and best correlates with the appearance of dementia [
7,
18].
Despite the knowledge that the presence of misfolded hyperphosphorylated tau is critical for development of disease and neuronal death [
15,
36], the mechanism of tau-related toxicity is still not clear. P301S tau transgenic mice (P301S mice) expressing human tau under the control of the neuronal Thy1.2 promoter develop neuronal tau aggregates in many brain areas [
1]. Tau pathology develops stereotypically between 2 and 5 months of age culminating in neuronal death most notably observed in the superficial layers of the motor and, perirhinal and piriform cortices [
1,
9,
51,
52]. To determine whether altering the environment can extend neuronal survival, we transplanted neuron precursor cell (NPC)-derived astrocytes and showed that neuronal death in the superficial layers of the motor cortex was prevented [
19], indicating a deficiency in survival support, or a gain of toxic functions, by the endogenous astrocytes. Astrocyte activation and reactive gliosis are associated with disease progression in almost all human neurodegenerative diseases [
33,
48] and astrogliosis appears to precede neuronal loss, suggesting an important causative role of astrocytes in the development of disease [
27].
Here we investigate the reasons why astrocytes from P301S mice do not prevent neuronal death whereas transplanted control astrocytes do. We show that astrocytes derived from the superficial cortex of P301S mice exhibit changes in cell specific markers that indicate astrocyte dysfunction. Moreover we demonstrate in in vitro systems that astrocytes or astrocyte conditioned medium from wild type mice have neuroprotective and synaptogenic functions that are absent in astrocytes from P301S- or P301L-tau expressing mice, which can be attributed in part to a reduction in thrombospondin-1 (TSP-1) expression in conditioned medium from P301S astrocytes.
Overall, our data demonstrate that astrocytes in the P301S tau mice are directly involved in neuronal death even though they do not express tau, highlighting a novel important contribution of astrocytes to tau-related pathogenicity, opening up new therapeutic avenues for treating diseases with tau pathology.
Materials and methods
Animals
Neurons and astrocytes were prepared from postnatal day 1–2, or 7–9 P301S tau or P301L tau female and male mice [
1,
45] along with age-matched C57BL/6 control mice. The tau mutation in the P301S mice is in the human 0N4R isoform whereas in the P301L mice, it is in the 2N4R isoform. Brain extracts were prepared from 3 to 5 month-old P301S and C57BL/6 mice. This research was conducted under the Animals (Scientific Procedures) Act 1986, Amendment Regulations 2012, following ethical review by the University of Cambridge Animal Welfare and Ethical Review Body (AWERB).
Mice were killed by cervical dislocation and brains were snap frozen on dry ice. Thick coronal slices (100 μm) extending from approximately 2.2 mm rostral to the bregma to the bregma were cut using a cryostat. The upper layers of the sensorimotor cortex were specifically dissected out using an ophthalmic blade. Dissected brain tissues were stored at −80 °C until use.
Astrocyte cultures
Primary astrocyte cultures were prepared from the cerebral cortex of 1–2 or 7–9 day-old C57 and P301S mice, or 7–8 day-old P301L mice as described previously [
42]. Briefly, mice were decapitated, the cortex was isolated and was triturated in HBSS (Hanks’ Balanced Salt Solution) by pipetting up and down. The cell suspension was incubated in 0.05% trypsin in HBSS at 37
•C to further dissociate the cells. After 30 min, fetal bovine serum (FBS) was added to a final concentration of 5% and the cell suspension was centrifuged at 1200 rpm. Pelleted cells were resuspended in DMEM with Earle’s salts supplemented with 10% FBS, 100 units/mL of penicillin and 100 μg/mL of streptomycin and plated in non-coated T75 flasks (ThermoScientific) at a density of 10
5 cells/ml. The cultures were maintained at 37 °C in 5% CO
2. Twenty-four hours after the initial plating, the medium was changed to remove non-adherent cells. When cultures reached confluence (about 1 week), non-astrocytic cells were separated from astrocytes by shaking for 15 h at 50 rpm at 37 °C (Luckham R300). Astrocyte-enriched cultures were then passaged into PDL-coated plates and maintained under the same conditions as the initial cultures. The surface-adherent monolayer cultures were > 98% positive for the astrocytic marker glial fibrillary acidic protein (GFAP). Cells were used for experiments after 5–6 days.
Neuronal cultures
Primary neuronal cultures were prepared from the cerebral cortex (3 brains per preparation) of ≥7 day-old or 1–2 day-old C57 and P301S mice. Briefly, neurons were isolated following the same protocol used for astrocytes and cultured in Neurobasal medium supplemented with 5% heat-inactivated bovine calf serum (Hyclone), B27, 1 mM L-glutamine, 100 U/mL of penicillin and 0.1 mg/mL streptomycin. Neurons were plated at a density of 105 cells/ml on 35 mm dishes coated with poly-D-lysine (10 μg/ml; Sigma). Cytosine arabinoside (2.5 μM) was added to the cultures on the second day after seeding to inhibit the proliferation of non-neuronal cells. Cells were used for experiments after 5–6 days. This protocol produced a neuron-enriched culture (95% of neurons).
Direct neuron–astrocyte co-cultures
Primary purified astrocytes from the second passage were plated at a density of 1.7 × 104 cells/cm2 on the top of AraC-treated primary neurons that had been in culture for 5–7 days. Co-cultures were fed with a mixture of one-third of astrocytic and two-thirds of neuronal medium, maintained at 37 °C in a humidified atmosphere of 5% CO2 and analyzed 4 and 8 days later. Cells were fixed and stained with the neuronal marker β-III-tubulin and the astrocytic marker GFAP to determine neuron/astrocyte number. Several fields per each experimental condition were scored for presence of neurons and astrocytes as described in the figure legends and the total number counted was used as a single value for statistical purposes. Results were obtained from 3 to 4 independent experiments (cell cultures) and each culture contained cells from the cortex of three mice.
Astrocyte conditioned medium (ACM)
Pure astrocyte cultures grown as described above for 5–6 days were passaged once. After reaching confluence, cultures were thoroughly washed to eliminate residual serum, and the cultures were maintained without fetal bovine serum for 1 day. The medium was then collected and centrifuged to remove cellular debris at 1000 rpm for 5 min and used immediately. To analyse the effect of ACM on neuronal survival, the medium in which the neurons were grown for 5–7 days was replaced with ACM and survival was analyzed after 4 and 8 days by counting β-III-tubulin positive neurons.
TSP-1 withdrawal or supplementation to ACM
TSP-1 was depleted from C57ACM by immunoprecipitation with an anti-TSP-1 antibody (Abcam, ab140250, 1:500) using magnetic Protein G Dynabeads (Invitrogen). Briefly, anti-TSP-1 antibody (Abcam 140250, 1:500) was incubated with Dynabeads with rotation for 10 min at room temperature. Then, C57ACM was added to the Dynabead-Ab complex, rotated for 10 min at room temperature, and immune complexes bound to the beads were pelleted by applying a magnetic field. The TSP-1-depleted ACM supernatant was collected and applied to neurons for 4 days. TSP-1 removal was verified by immunoblotting. For TSP-1 supplementation, ACM from P301SA was enriched with recombinant mouse TSP-1 (rTSP-1, 500 ng/ml, NovusBio) and the mixtures were added to cultured neurons for 4 days. Neuronal survival was determined by counting neurons identified by immunocytochemistry with anti-β-III-tubulin.
Proliferation capacity
Astrocytes grown to 98% confluence, were repassaged and analyzed after 2 days. Cells were incubated with the thymidine analogue 5-ethynyl-2′-deoxyuridine (EdU, 10 μM final concentration, ThermoScientific) for 2 h at 37 °C, fixed and stained using the Click-iT® EdU Alexa 488 Cell Proliferation kit (ThermoScientific).
Western blot analysis
Tissue, cultured astrocytes or neurons were lysed in RIPA buffer (150 mM NaCl, 1.0% IGEPAL® CA-630, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate and 50 mM Tris, pH 8.0) containing protease and phosphatase inhibitor cocktails (Sigma). Tissue was left in RIPA buffer for 20 min on ice before homogenisation with a teflon pestle. Homogenates were spun at 13,000 × g for 30 min and the supernatants were used for analysis. ACMs were concentrated by spinning at 3750 × g for 25 min in an Amicon centrifuge filter tubes with a 10 kDa molecular weight cut-off. Protein concentrations in the tissue extracts, cell lysates or ACM were determined with the bicinchoninic acid (BCA) protein assay kit (Pierce, ThermoScientific). An equal amount of protein from cells or ACM (15 μg) was loaded and run on a 12% SDS-PAGE and then transferred to a polyvinylidenefluoride membrane (EMDMillipore). Non specific background was blocked by a 1 h incubation at room temperature in 5% non-fat dry milk in Tris Buffered Saline with 0.1% Tween 20 (TBS-T). Incubations with primary antibodies were carried out at 4 °C for 24 h in 5% non-fat milk in TBS-T buffer at the following antibody concentrations: anti-GLAST (Abcam, ab416, 1:1000), anti-GLT1 (Abcam, ab41621, 1:1000), anti-GS (Abcam, ab49873, 1:2000), anti-GFAP (Abcam, ab10062, 1:2000), anti-S100β (Abcam, ab14688, 1:1000), anti-TSP-1 (Abcam, ab85762, 1:1000), anti-PSD-95 (Abcam, ab18258, 1:2000), anti-synaptophysin (SNP) (Abcam, ab106618, 1:1000), anti-beta actin (Sigma, A2066, 1:5000). Secondary antibody incubations were performed at room temperature for 1.5 h using HRP-conjugated anti-rabbit IgG (ThermoScientific, 1:2000) or anti-mouse IgG (Sigma, 1:4000). For ACMs, blots were visualized with Ponceau S (Sigma) and developed with Supersignal West Dura Extended Duration Chemiluminescent Substrate (Pierce, ThermoScientific).
Immunocytochemistry
Primary neuronal, astrocytes or astrocytic-neuronal co-cultures plated on glass coverslips were washed twice with TBS and fixed at room temperature for 10 min with 100% cold methanol. Cells were permeabilized with 0.1% Triton X-100 in PBS for 15 min and then incubated for 1 h in 5% goat serum to reduce nonspecific background. After an overnight incubation at 4 °C with the primary antibodies: [chicken or mouse anti-glial fibrillary acidic protein (Abcam, ab4674, 1: 200 or Dako, z0334, 1:500), anti-β-III-tubulin, (Abcam, ab18207, 1:500 or Covance, MMS-435P 1:1000), anti-synaptophysin (Abcam, ab106618, 1:500), anti-NeuN (Millipore, MAB377, 1:500)], cells were washed with TBS and incubated with secondary AlexaFluor-conjugated antibodies appropriate for the species (Molecular Probes, 1:500). To visualize cell nuclei, cultures were rinsed and then incubated in 4′,6-diamidino-2-phenylindole dihydrochloride hydrate (DAPI)/antifade (Sigma, 1:1000) diluted in TBS or Hoescht dye (Sigma, 1:5000) for 10 min at room temperature. Cover slips were mounted in FluorSave™ (EMD Millipore) and pictures were taken with a wide-field fluorescence microscope (Leica DMI 4000B microscope using a Leica DFC3000 G camera and the Leica application suite 4.0.0.11706).
Image analysis
Western blot and SNP intensity analyses were carried out using ImageJ (Rasband, W.S., ImageJ, U.S. National Institutes of Health,
http://imagej.nih.gov/ij/, 1997–2014). Quantification of SNP expression in neurons was performed by measuring mean fluorescence staining intensity within the contour drawn around the individual cells in SNP-stained neuronal cultures. At least six cells per field, and four fields per technical replicate were analyzed. Bands on blots were quantified by measuring mean gray value of individual bands using the Measure tool in ImageJ or the AlphaEaseFC Imaging System software (Alpha Innotech).
Statistical analysis
Data are expressed as mean ± SEM. Results from technical replicates or from counting several fields in each culture were pooled to give a single value for statistical purposes. Statistical analyses for significant differences were performed with unpaired t test, or one- or two-way ANOVA followed by Tukey’s posthoc test or Mann-Whitney where appropriate, using GraphPad Prism 5.0 software. The criterion for statistical significance was p < 0.05.
Discussion
Transgenic human P301S tau mice, where tau is expressed specifically in neurons under the control of the Thy1 promoter [
1], display progressive tau aggregation and neuron loss with associated astrogliosis in the superficial layers of the cerebral cortex between 2 and 5 month of age [
19]. We have previously shown that this neuronal death can be rescued by transplantation of neuronal precursor cell-derived astrocytes from wild type mice [
19], implying that the endogenous astrocytes are functionally deficient in P301S tau mice. To determine why transplanted astrocytes were protective we prepared primary co-cultures of postnatal astrocytes and neurons from the cortex of P301S tau transgenic and control mice. Our findings demonstrate that endogenous astrocytes from P301S tau mice are deficient in factors that wildtype astrocytes secrete in order to support neuronal survival and synaptogenesis. Our results thus explain the observation that wildtype astrocytes rescue transgenic P301S tau cortical neurons from death by showing that they express neurosupportive factors which are lacking in P301S-derived astrocytes.
To understand the biochemical basis for these differences we examined the expression of key proteins implicated in astrocyte function. We found an increase in the expression of GFAP and S100β, astrocytic proteins related to glial responses to injury, both in extracts from the cerebral cortex of 3 and 5 month-old P301S mice, extending previous immunohistochemical findings [
1,
19], and in primary cultures of astrocytes from P301S tau mice. Correlated with this increase, we found that cultured astrocytes from P301S mice showed enhanced proliferative capacity relative to those from control mice, indicating a cell autonomous memory of a previous injury-like state. Although this does not signify whether these changes are adaptive or maladaptive, they indicate highly coordinated changes in astrocyte behaviour [
3,
21]. Our immunoblot analyses also uncovered significant changes in the expression of proteins related to glutamate homeostasis in the superficial cerebral cortex of 3 and 5 month-old P301S tau mice and in primary cultures of astrocytes. Astrocytes secrete glutamate in response to activation, modulate glutamate receptor expression, and remove glutamate from the synaptic cleft by glutamate transporters [
2,
4,
49]. This regulation of synaptic glutamate is crucial for normal CNS function, and the sodium-dependent glutamate transport system located peri-synaptically on astrocytes contributes to the regulation of extracellular glutamate levels. Because astrocytes play a major role in control of glutamate homeostasis, we focused on three important regulatory proteins of glutamate metabolism, GS, the main glutamine metabolizing enzyme [
34], GLAST and GLT1 [
12,
34,
38], the astroglial-specific Na
+/glutamate transporters. We found a reduced expression of all three proteins in extracts from the superficial cortex of P301S mice, which were also evident in astrocytes cultured from these mice, despite their being expanded for several days ex-vivo.
Decreases in the expression of GLAST and GLT1 were previously reported in astrocytes expressing GFAP/tau mice, wildtype tau or P301L mutant tau [
14]. These mice manifested motor deficits before the development of overt tau pathology, which correlated with loss of expression and function of both glial glutamate transporters. Interestingly, there was no difference in effect between the mutant and non-mutant transgenic tau in these mouse models and since tau is not normally expressed in astrocytes, it is not clear how this pathology related to tau toxicity elicited by neurons. Notably, these models are different from our transgenic P301S mice where tau (mRNA and protein) is expressed only in neurons and is not present in astrocytes, indicating that the changes in glutamate transporters in our system must be related to neuron-astrocyte cross talk. In our model, neuronal dysfunction drives the changes similar to a reported mouse model of Parkinson’s disease where disruption of striatal glutamatergic innervation resulted in reduction in both GLT-1 and GLAST protein expression, accompanied by dysfunction of glutamate uptake [
16,
23,
31]. A study of a different P301S tau mouse model (where P301S is expressed under the prion promoter) revealed regional changes in glutamate levels that correlated with histological measurements of pathology, such as pathological tau, synapse and neuron loss [
13]. Deficits in glutamate neurotransmission and mitochondrial dysfunction were also detected in the frontal cortex and hippocampus of aged 3 × Tg AD mice, which develop beta-amyloid plaques and tau aggregates containing P301L tau [
17]. Decreased expression of glutamate-metabolizing enzymes (such as glutamate dehydrogenase and glutamine synthetase protein) in astrocytes were also found in the cerebellum of patients with Alzheimer’s disease [
8]. In the 3xTg AD mice, wildtype astrocyte transplantation was reported to improve altered behaviour and this improvement was attributed to increased expression of BDNF [
6] but we did not find a significant increase in growth factors following transplantation in our P301S tau mice [
19]. A recent study has shown that neuronal activity has a prime role in upregulating gene expression and function of glutamate transporters in astrocytes [
20]. Taken together, our results indicate that the glutamatergic system is one of the vulnerable points in the reaction between astrocytes and neurons in brain disease and injury, where astrocytes may fail to prevent glutamate excess and neuronal toxicity through loss of proper glutamate homeostasis.
Both astrocytes from P301S mice co-cultured with neurons, and P301SACM failed to protect neurons from basal cell death whereas C57A or C57ACM enhanced neuron survival. Notably, similar results were obtained using ACM from astrocytes from P301L mice, where tau is expressed under the same neuronal specific Thy1 promoter as in our P301S mice [
45]. Hence, the lack of survival support is not tau mouse model-specific nor is it related to a specific tau isoform or MAPT mutation or due to the insertion site of the transgene in the mouse genome but rather is due to the expression of mutant tau and tau pathology development. Although tau filaments and motor pathology develop consistently between 3 and 5 months in the P301S mouse, transgenic tau is expressed from postnatal day 1 and significant signs of altered behavioural function, detected by measuring ultrasound vocalisation (USV) [
39], are evident already in newborn mice 3 days postnatal with increased USV maintained up to 7 days [
40]. Our findings indicate that astrocytes develop pathological changes due to the exposure to P301S tau-expressing neurons in 7–8 day-old pups but not in 1–2 day-old mice, since we found no difference in neuron survival when neurons were exposed for 8 days to astrocytes or ACM prepared from 1 to -2 day-old P301S tau mice. Although transgenic tau is present in neurons in 1–2 day-old pups, it is possible that either it is not sufficient to induce the astrocytic reaction or that this response takes several days to develop. At both ages, in 1–2 day- or 7 day-old pups no aggregated tau is visible in neurons, indicating that toxic events precede tau filament formation. Hence the development of astrocyte dysfunction appears to relate to the earliest manifestations of neuronal tau toxicity.
Recently, IPSCs-derived astrocytes from Down syndrome (DS) patients were shown to be toxic to neurons but in this case astrocytes, like neurons, bear a trisomy of chromosome 21 [
9] whereas MAPT is located on chromosome 17. Similar to our findings, however, the study revealed that DS astroglia exhibited a higher proliferation rate, and expressed higher levels of S100β and GFAP. Furthermore, DS astrocytes contributed to the reduction of neurogenesis of DS NPCs and to the induction of DS neuron death via failure to promote maturation and synapse formation in these cells. Loss of functional synapses is a major neuropathological feature that is well defined in many AD and FTD mice models [
10,
32,
41]. In keeping with these results, we observed a significant decline in expression of the synaptic markers PSD95 and synaptophysin upon exposure of neurons to P301SACM. In contrast, exposure to C57ACM enhanced both neuron survival and expression of the two synaptic markers we investigated.
To determine the possible factors involved in astrocyte dysfunction, we sought proteins that are differentially expressed in the ≥10 kDa fractions of the ACM that may be associated with a neuroprotective effect, and thereby focused on TSP-1. TSP-1 is a well defined molecule expressed in postnatal and young adult animal brains and in human cortical astrocytes where it has been shown to promote neuroprotection [
5,
30,
54], to increase the number of synapses [
11,
24], as well as to accelerate synaptogenesis [
50]. Furthermore, TSP-1 has been implicated in neurodegenerative diseases in that addition of amyloid-β peptides, the main components of the
amyloid plaques found in the brains of Alzheimer’s patients, caused a significant decline in the release of TSP-1 from primary cultures of astrocytes [
37]. We found that P301S astrocytes
in vivo and
in vitro produce, and,
in vitro, release significantly less TSP-1. A similar decline of TSP-1 expression was described in Down Syndrome astroglia pathogenesis [
9]. To demonstrate that TSP-1 is a limiting factor in the P301SACM, immune depletion of TSP-1 from C57ACM significantly reduced neuronal survival of C57N and P301SN, whereas supplementation of TSP-1 to P301SACM restored viability, especially that of P301SN. Although we focused on TSP-1, a preliminary analysis of ACMs indicates that it is unlikely that TSP-1 is the only factor that is limiting in P301SACM. A proteomic study of adult symptomatic prion promoter-driven P301S mouse brains identified some differentially expressed proteins in astrocytes, which they propose to have neuroprotective functions [
53]. However, the prion promoter may drive expression of tau in astrocytes [
29] whereas in our model no tau is expressed in astrocytes. A key question that remains is to find out why and how the expression of these proteins is differentially regulated through neuron-astrocyte interactions.