Tau inhibits PKA by nuclear proteasome‐dependent PKAR2α elevation with suppressed CREB/GluA1 phosphorylation

Abstract Intraneuronal accumulation of wild‐type tau plays a key role in Alzheimer's disease, while the mechanisms underlying tauopathy and memory impairment remain unclear. Here, we report that overexpressing full‐length wild‐type human tau (hTau) in mouse hippocampus induces learning and memory deficits with remarkably reduced levels of multiple synapse‐ and memory‐associated proteins. Overexpressing hTau inhibits the activity of protein kinase A (PKA) and decreases the phosphorylation level of cAMP‐response element binding protein (CREB), GluA1, and TrkB with reduced BDNF mRNA and protein levels both in vitro and in vivo. Simultaneously, overexpressing hTau increased PKAR2α (an inhibitory subunit of PKA) in nuclear fraction and inactivated proteasome activity. With an increased association of PKAR2α with PA28γ (a nuclear proteasome activator), the formation of PA28γ‐20S proteasome complex remarkably decreased in the nuclear fraction, followed by a reduced interaction of PKAR2α with 20S proteasome. Both downregulating PKAR2α by shRNA and upregulating proteasome by expressing PA28γ rescued hTau‐induced PKA inhibition and CREB dephosphorylation, and upregulating PKA improved hTau‐induced cognitive deficits in mice. Together, these data reveal that intracellular tau accumulation induces synapse and memory impairments by inhibiting PKA/CREB/BDNF/TrkB and PKA/GluA1 signaling, and deficit of PA28γ‐20S proteasome complex formation contributes to PKAR2α elevation and PKA inhibition.


| INTRODUC TI ON
Intracellular accumulation of the abnormally modified microtu buleassociated protein tau is a hallmark pathology in the brain of Alzheimer's disease (AD) and the other tauopathies. However, how tau accumulation induces synapse and memory deficits is unclear. Tau is a axonallocated microtubuleassociated protein involving in microtubule assemble and stabilization; however, ab normal dendritic translocation of tau proteins induces the den dritic complex loss and synaptic neurodegeneration (Shentu et al., 2018). The increased total tau in plasma and accumulated tau in brain correlate with the severity of memory deficit in AD (Mielke et al., 2017;Thal et al., 2000). Clustering of tau fibrils in the mem brane imbalance neuronal homeostasis by the impaired synaptic composition of Na + /K + -ATPase and αamino3hydroxy5me thylisoxazole-4-propionic acid (AMPA) receptors (Shrivastava et al., 2019).
Phosphorylation of cAMP-response element binding protein (CREB) at Ser133 activates the transcription factor and promotes the synthesis of synapse-and memory-associated proteins, thus poten tiates learning and memory (Tully, Bourtchouladze, Scott, & Tallman, 2003), while CREB dephosphorylation at Ser133 has been observed in learning and memory disorders. For instance, both Aβ treatment and tau overexpression inhibit CREB phosphorylation with learn ing and memory deficits (Yin et al., 2016). As a transcription factor, CREB-mediated transcription of BDNF is sufficient to promote neu ronal dendrite outgrowth, enhance synapse formation, and improve learning and memory (Bu et al., 2017;Segarra-Mondejar et al., 2018).
CREB also mediates the transcription of AMPA receptors, which is the molecular basis of excitatory synaptic transmission (Middei et al., 2013).
Protein kinase A (PKA) is the most recognized kinase in phos phorylating CREB at Ser133 and AMPA-type glutamate receptor trafficking, by which it promotes transcription of synaptic proteins and plays a crucial role in learning and memory formation (Taylor, Ilouz, Zhang, & Kornev, 2012). β-amyloid (Aβ) can inhibit PKA (Tong, Thornton, Balazs, & Cotman, 2001), while activating PKA enhances long-term potentiation (LTP) and memory formation .
PKA activation leads to fast exocytosis of AMPA receptors through GluA1 phosphorylation at Ser845, which in turn potentiates LTP (Oh, Derkach, Guire, & Soderling, 2006). PKA is a tetrameric protein consisting of two regulatory subunits (PKA-Rs) and three catalytic subunits (PKA-Cs). When in an inactive state, the pseudo-substrate sequence of the PKA-Rs blocks the active site on the PKA-Cs. Three PKA-Cs (C-α, C-β, and C-γ) and two PKA-Rs (RI and RII) families have been discovered, and all PKA-Rs have different cAMP binding prop erties (Ilouz et al., 2017). The two R families have two isoforms: α and β (RI-α, RI-β, RII-α, and RII-β). When cAMP binds to the R sub unit, it can alleviate the self-inhibiting contact and release activated monomeric PKA-Rs (Skalhegg & Tasken, 2000). PKA-Rs released are immediately degraded by proteasome to maintain the activated state of PKA-Cs (Tai & Schuman, 2008). The liberated PKA-Cs are required to sustain PKA function.
Panneuronal overexpression of tau induces synapse and mem ory deficits in mice, and simultaneously, a remarkable CREB dephos phorylation was shown (Yin et al., 2016). However, it is not known how tau accumulation downregulates CREB phosphorylation. By neuronspecific overexpression of human fulllength wildtype tau (hTau), we demonstrate in the present study that hTau accumula tion inhibits PKA/CREB signaling with downregulation of multiple synapse-and memory-associated proteins and GluA1-containing AMPARs surface expression both in vitro and in vivo. The mech anism underlying PKA inactivation involves an impaired nuclear proteasome-mediated PKAR2α elevation. Simultaneous downregu lating PKAR2α or upregulating proteasome restored PKA activity, CREB and GluA1 phosphorylation, BDNF expression with increased spine density, dendritic complex, and membranous trafficking of GluA1. Activating PKA/CREB signaling also rescued learning and memory deficits in hTauoverexpressing mouse model.

| Overexpressing hTau in mouse hippocampal CA3 neurons induces memory deficits with PKA inhibition and CREB/GluA1 dephosphorylation
To explore the in vivo toxicity of neuronal tau accumulation, we made neuronspecific overexpression of hTau by stereotaxic infu sion of AAV-syn-hTau-EGFP (hTau) into the mouse hippocampal CA3 subset, and the same volume of AAV-syn-EGFP (Vec) was infused as control. After one month, robust expression of AAV-syn-hTau-EGFP was confirmed by direct fluorescence imaging and Tau5 staining ( Figure 1a,b). On average, the total tau level increased to ~2.1-fold of the control (Figure 1c), which may represent an early Braak stage of tau pathology in the AD patients according to the previous reports (Kurbatskaya et al., 2016). The oligomeric and fibrillary tau proteins were detected by using an anti-tau oligomer antibody T22 ( Figure   S1a) and Thioflavin-S staining ( Figure S1b), respectively. Then, the effects of hTau accumulation on learning and memory of mice were measured by novel object recognition (NOR) and Morris water maze (MWM) test. In NOR test, the recognition index was comparable be tween hTau and the control groups in the training trial ( Figure 1d). However, the recognition and discrimination indexes to the new ob ject were significantly decreased in the hTau mice during test trials measured at 24 hr after the training trial (Figure 1e,f), indicating an impaired long-term memory. In MWM test, we observed that neu ronal overexpression of hTau in hippocampal CA3 subset for one month induced spatial learning deficit during 6 days of training trial ( Figure 1g). The spatial memory was tested on day 8 by removed the platform. We observed that the hTau-overexpressing mice exhibited memory deficits shown by an increased latency to find the platform

| Overexpressing hTau in cultured hippocampal neurons decreases synaptic proteins with PKA inhibition and CREB/TrkB and GluA1 dephosphorylation
To explore the mechanisms underlying hTaurelated neurodegenera tion, we overexpressed human tau (lenti-hTau-mCherry) or the empty vector (lenti-mCherry) in cultured hippocampal neurons (5 div). After 7 days, the overexpression of tau was confirmed by Western blot ting using Tau5, and the total tau level increased to ~1.8-fold of the control level (Figure 2a (Figure 2j-l). The activity of PKA was inhibited, and the levels of phosphorylated CREB at Ser133 (pCREB), phosphorylated GluA1 at Ser845 (pGluA1), the phosphorylated TrkB at Tyr706 (pTrkB), and the protein level of BDNF were all decreased after overexpressing hTau (Figure 2m,n). These data suggest that overexpressing hTau induces synapse impairment with the mecha nisms involving inhibition of PKA/CREB/BDNF signaling.

| Upregulating PKA rescues hTau-induced CREB/GluA1 dephosphorylation and BDNF expression with improved synaptic plasticity and cognitive functions
To verify the role of PKA in hTau-induced inhibition of PKA/CREB/ BDNF signaling and the impairments of synaptic and cognitive func tions, we first infected primary cultured hippocampal neurons (5 div) with lenti-syn-hTau-mCherry or lenti-syn-mCherry for 7 days and treated the neurons with rolipram, a PKA agonist for 12 hr. Then, the activity of PKA/CREB/BDNF signaling and synaptic plasticity was measured. Rolipram treatment attenuated hTau-induced PKA inhibition with restoration of pS133-CREB, pS845-GluA1, BDNF mRNA, and protein levels (Figure 3a-c). Simultaneously, the expres sion of surface GluA1, which plays an important role in synaptic transmission and is Ser845-phosphorylation-dependent (Oh et al., 2006), was remarkably suppressed in hTau-overexpressing neurons, while rolipram treatment restored the surface GluA1 level measured in living neurons with an extracellular GluA1 antibody ( Figure 3d).
Whole-cell patch-clamp recordings were also performed to detect the synaptic transmission in primary cultured hippocampal neurons.
Overexpressing hTau decreased both frequency and the amplitude of mEPSC, which were restored by rolipram (Figure 3e,f). The spine density, dendrite length, and complexity were decreased by over expressing hTau and rolipram rescued the dendritic impairments in cultured hippocampal neurons (Figure 3g-k). These in vitro data con firm that PKA inhibition plays a crucial role in hTau-induced synapse impairments.
To validate whether activating PKA could improve cognitive deficits in mice, dorsal hippocampal CA3 of 2-month-old mice was stereotaxically infused with AAV-syn-hTau-EGFP or AAV-syn-EGFP as vector control. After four weeks, the mice were intra peritoneally infused with rolipram (0.03 mg/kg per day) or normal saline (NS) for 20 days and then learning and memory were tested and GluA1 phosphorylation (Figure 4l,m). These in vivo data fur ther confirm that PKA inhibition plays a crucial role in hTau-in duced synapse impairment and memory deficit.

| Overexpressing hTau increases nuclear PKAR2α and inhibits proteasome, the latter contributes to the hTau-induced PKAR2α elevation
To explore the mechanism underlying the hTau-induced PKA in hibition, we measured different subunits of PKA by infecting the primary hippocampal neurons (5 div) with lenti-syn-hTau-mCherry or lenti-syn-mCherry viruses for 7 days. The reduced PKA activity in hTau-overexpressing neurons was detected using a PKA activ ity assay kit ( Figure 5a). By Western blotting using PKA substrate antibody, the reduced PKA activity in total extracts and the nu clear fraction in hTauoverexpressing neurons were also detected ( Figure 5b,c). The catalytic subunits of PKA (PKA-Cs) are regulated by the regulatory subunits (PKA-Rs) and cAMP, when cAMP binds to the PKA-Rs, PKA-Rs can be removed from PKA-Cs, and activated monomeric PKA-Cs are released (Skalhegg & Tasken, 2000;Taylor et al., 2012). Therefore, we first measured cAMP level but no significant difference was shown between hTau and control groups (Figure 5a), suggesting that hTau-induced PKA inhibition is not cAMP-depend ent. Then, we measured PKA-Rs and PKA-Cs in total extracts and nuclear fractions. Overexpressing hTau increased nuclear level of PKAR2α without changing PKAR1α, PKAR1β, PKAR2β, and PKA-Cs in both fractions (Figure 5b,c, S2a-c). Significant elevation of PKAR2α in nuclear fraction was also detected in brain slices after hTau overexpression ( Figure 2d). These data suggest that hTau ac cumulation may inhibit PKA through upregulating nuclear PKAR2α.
To further explore the mechanisms underlying PKAR2α eleva tion, we measured the proteasome activity. With the upregulated PKAR2α protein level in total extract and the nuclear fraction Treating the cultured hippocampal neurons with 10 μM MG132 also induced the same changes as observed in N2a cells (Figure 5g,h).
These data indicate that proteasome inhibition at least contributes to the hTau-induced nuclear elevation of PKAR2α, which in turn in hibits PKA/CREB/BDNF signaling.

| Overexpressing hTau disrupts 20S-PA28γ complex formation and upregulating PA28γ or downregulating PKAR2α rescues PKA/CREB signaling
Inhibition of proteasome activity has been widely observed in AD and other tauopathy, and PKA-Rs are degraded in 26S proteasome (Tai & Schuman, 2008). As the hTau-induced PKAR2α elevation was largely detected in the nuclei, we measured whether PA28γ

Intracellular accumulation of hyperphosphorylated tau has been
widely observed in neurodegenerative diseases, such as AD, pro gressive supranuclear palsy (PSP), corticobasal degeneration, Pick disease, Huntington disease, and frontotemporal dementia with par kinsonism-17 (Wang & Mandelkow, 2016). Studies also show that tau accumulation induces synaptic dysfunction, neuron loss, and mem ory impairment in different mouse models (Fu et al., 2017). In AD, it F I G U R E 1 Overexpressing hTau in mouse hippocampal CA3 neurons induces memory deficits with PKA inhibition and CREB/GluA1 dephosphorylation. (a-c) AAV-syn-hTau-EGFP was infused into the hippocampal CA3 subset of 2-month-old mice, and after one month, the expression of hTau in the infused site (scale bar: left, 250 μm; right, 50 μm) was confirmed by direct fluorescence imaging (a), and Western blotting (110 kDa) using Tau5 (b and c; n = 6 mice per group). (d-f) Overexpressing hTau induces memory deficit measured by novel object recognition (NOR) test. The recognition index measured at the first training day (d), and the reduced recognition index (e) and the discrimination index (f) recorded at 24 hr after training (n = 15 mice per group) were analyzed. (g-k) Overexpressing hTau induces spatial learning and memory deficits measured by Morris water maze (MWM). The spatial learning deficit was shown by the increased latency to find the hidden platform during 6 days of training (g), while the memory deficit was shown by the increased latency to reach the platform site (h), decreased target platform crossings (i), and decreased time spent in the target zone (j) measured at day 8 by removing the platform (n = 15 mice per group). The swimming speed was not changed (k). (l) Overexpressing hTau inhibits PKA activity measured by using a PKA activity assay kit. The injected hippocampal CA3 subsets were dissected, and the tissue extracts were used. (m, n) Overexpressing hTau reduced levels of pGluA1 and pCREB measured by Western blotting (m) and the quantification (n); HT7 specifically reacts with human tau. Data were expressed as mean ± SEM, *p < .05, **p < .01 versus Vec, but *p < .05 versus object A in (e) has emerged that Aβ toxicity depends on dendritic function of tau and tau knockout mitigates Aβ toxicity (Ittner et al., 2010). PHF-tau (paired helical filament) pathology is correlated with the degree of dementia as defined by the clinical dementia rating scale (Thal et al., 2000). In sporadic AD, accumulation of wild-type tau is the major component of neurofibrillary tangles.
In AD patients, tau pathology was first appeared in the ento rhinal cortex, then spread to limbic region of the hippocampus and F I G U R E 2 Overexpressing hTau in primary cultured neurons decreases synaptic proteins with reduced phosphorylation of CREB/TrkB and GluA1. (a, b) The hippocampal neurons (5 div) were infected with lenti-syn-hTau-mCherry or the empty vector and cultured for another 7 days, and the expression of hTau (110 kDa) was confirmed by Western blotting using Tau5 (reacts with total tau). (c, d) Infection of different MOI lenti-hTau virus induced a dose-dependently increased expression of tau in cultured hippocampal neurons measured by Western blotting using Tau5 (n = 6, **p < .01 vs. MOI 1). (e) Infection of 10 MOI lenti-hTau virus significantly decreased CREB phosphorylation (n = 6, **p < .01 vs. MOI 1). (f-h) Overexpressing hTau induces reduction of multiple synaptic proteins. The hippocampal neurons (5 div) were infected with lenti-syn-hTau-mCherry or the empty vector and cultured for another 7 days, and the levels of presynaptic (g) and postsynaptic (h) proteins were measured by Western blotting. The experiments were repeated three times from nine different batches of primary cultured neurons. (i) Overexpressing hTau decreases mRNA level of SYT with no significant effect on SYN1 and SYP measured by qPCR. (j-l) Overexpressing hTau decreases dendrites branches and total dendrite length. The cultured hippocampal neurons (18 div) were stained with MAP2, and then, the dendrite arborizations and total length were analyzed by Sholl (scale bar: 50 μm). The total dendrite length and the branches at 20 to 140 μm from soma were measured (n = 30 per group from six independent cultures, black line, p < .05, vs. Vec). (m, n) Overexpressing hTau decreases CREB/TrkB/GluA1 phosphorylation and reduces BDNF measured by Western blotting. The phosphorylation levels of CREB, TrkB, and GluA1 were normalized to the total protein levels or to βactin. The experiments were repeated three times from nine different batches of primary neurons. Data were expressed as mean or mean ± SEM, *p < .05, **p < .01 versus Vec, but **p < .01 versus MOI 1 in (d) and (e) finally to the whole neocortex. CA3 is one of the critical subset of hippocampal tri-synaptic circuit, which plays a crucial role in spatial memory. In our previous studies, we consistently observed that CA3 is the most sensitive subset for accumulation of the phosphorylated tau under stresses compared with other subsets of the hippocam pus (Li et al., 2016;Yin et al., 2016). To explore the mechanism of F I G U R E 3 Upregulating PKA rescues hTau-induced synaptic impairments with increased CREB/GluA1 phosphorylation and BDNF expression. (a-c) The hippocampal neurons (5 div) were infected with lenti-syn-hTau-mCherry or the empty vector for 7 days, and then, the neurons were treated with 40 μM rolipram to activate PKA for 12 hr. The mRNA level of BDNF was measured by qPCR (a); the protein levels of pGluA1, GluA1, pCREB, CREB, and BDNF were detected by Western blotting (b); and the quantitative analysis was performed (c) (n = 9 per group). (d) Upregulating PKA restores the hTau-suppressed membrane localization of GluA1 measured by incubation the living neurons (12 div) with an extracellular N-terminal of GluA1 antibody (scale bar, 20 μm). (e, f) Upregulating PKA restores the hTau-suppressed synaptic transmission. The representative traces of mEPSCs recorded by whole-cell patch clamp in hTau-or Vec-infected primary hippocampal neurons (15 div) treated with rolipram or vehicle. Scale bars: 5 pA, 2s (e). The average frequency and amplitude of mEPSCs were collected from at least 7-12 neurons per group (f). (g, h) Upregulating PKA restores the hTau-induced spine impairments. For spine analysis in vitro, the hippocampal neurons (12 div) were infected with lenti-EGFP and lenti-syn-hTau-mCherry or the control vector. Spine density (21 div) was measured by two-photon confocal laser scanning microscopy (g). The spine number per 10 μm of dendritic length was analyzed using ImageJ software (h). At least 30 neurons from six independent cultures were collected for each group. Scale bar 2 μm. (i) Upregulating PKA restores the hTau-suppressed dendrite development in cultured hippocampal neurons (18 div) stained with MAP2. (j) The total length and arborizations of dendrites were analyzed by Sholl (scale bar: 50 μm). At least 30 neurons from six independent cultures were analyzed for each group. (k) Upregulating PKA rescues hTau-induced the decrease dendrites branches at 10 to 160 μm from soma (black line, p < .05, vs. Vec, red line, p < .05 vs. hTau). At least 30 neurons from six independent cultures were analyzed for each group. Data were expressed as mean or mean ± SEM, *p < .05, **p < .01 versus Vec. # p < .05, ## p < .01 versus hTau F I G U R E 4 Upregulating PKA rescues synaptic dysfunction and cognitive decline with restoration of CREB/GluA1 phosphorylation. (a) Schematics show the treatments of mice. AAV-syn-hTau-EGFP or the empty vector was infused stereotaxically into the hippocampal CA3 of 2-month-old mice for one month, and then, rolipram (PKA activator, 0.03 mg/kg per day) or normal saline (NS) was infused intraperitoneally for 20 days. After MWM test, the mice were sacrificed and the hippocampi were dissected for electrophysiology recording and Western blotting analysis. (b-f) Simultaneous activation of PKA attenuates hTau-induced spatial learning and memory deficits shown by the reduced latency to find the hidden platform during 6 days of training on MWM (b), and the decreased latency to find the platform site (c), the increased target platform crossing (d), and the duration in the target zone measured at day 8 after removed the platform (e). The swimming velocity was not changed (f) (n = 15 mice per group). (g-i) Simultaneous activation of PKA rescues hTau-induced inhibition of LTP induction shown by the restored fEPSP amplitudes (input-output curve) induced by minimum stimulation intensity (g), and the restored fEPSP slope induced by applying three trains of high-frequency stimulation (HFS) (h, i). (j, k) Simultaneous activation of PKA rescues hTau-induced spine impairments. The mice were infected and treated as described in panel a. Golgi staining was used to evaluate the changes of dendritic spines in the hippocampal CA3 subset of the mice, and at least 30 neurons from six mice per group were analyzed. (l, m) Simultaneous activation of PKA rescues hTau-induced dephosphorylation of GluA1 and CREB measured by Western blotting. HT7 specifically reacts with human tau. The data were expressed as mean ± SEM. *p < .05, **p < .01 versus Vec, # p < .05, ## p < .01, versus hTau F I G U R E 5 Overexpressing hTau inactivates PKA with proteasome-involved elevation of PKAR2α and inhibition of CREB/BDNF signaling. (a) Overexpressing hTau inhibits PKA without changing cAMP in cultured hippocampal neurons (5 + 7 div) measured, respectively, by using PKA activity assay kit and the cAMP ELISA kit. (b, c) Overexpressing hTau inhibits PKA activity with an elevated PKAR2α in nuclear fraction of the cultured neurons. The hippocampal neurons (5 div) were infected with lenti-syn-hTau-mCherry or the empty vector for 7 days, and then, PKA activity (using the phosphor-PKA substrate) and the level of PKAR2α (PKA regulatory subunit) in total cell lysis and the nuclear fraction were measured by Western blotting. (d) Overexpressing hTau inhibits proteasome activity. The hippocampal neuron (5 div) was infected with lenti-syn-hTau-mCherry or the empty vector for 7 days, and the proteasome activity in the cell lysis was measured using a proteasome activity kit. (e, f) Inhibiting proteasome by treating N2a cells with different concentrations of MG132 induces PKAR2α elevation with reduced CREB phosphorylation and BDNF expression measured by Western blotting. (g, h) Inhibiting proteasome by treating cultured hippocampal neurons with MG132 (10 μM) for 12 hr induces PKAR2α elevation with reduced CREB phosphorylation and BDNF expression measured by Western blotting. Data were expressed as mean ± SEM, *p < .05, **p < .01 versus Vec, but *p < .05, **p < .01 versus 0 μM in (f), **p < .01, ***p < .001 versus Vih in (h). The experiments were carried out from nine different batches of N2a cells and cultured hippocampal neurons GluA1, GluA2, PSD93, PSD95, and SYT were also detected with unchanged GluN2B, SYP, and SYN1. Decreased GluA1 phosphor ylation at Ser845 also contributes to hTau-induced synaptic trans mission efficacy. Therefore, we speculate that hTau may decrease synaptic protein via PKA/CREB-mediated transcription.
We also noticed that overexpressing hTau only selectively de creased mRNA and protein levels of SYT among several vesicles re lease-related presynaptic proteins (including SYN1, SYP and SYT). SYN1, located in the presynaptic terminal, controls the vesicle stor age and mobilization (Cesca, Baldelli, Valtorta, & Benfenati, 2010).
SYP is a membrane protein located mainly on the presynaptic ves icles with the major functions in regulating vesicle formation and release (Valtorta, Pennuto, Bonanomi, & Benfenati, 2004). SYT is a calcium sensor located on the membrane of presynaptic vesicles with the functions in calcium-dependent release of vesicles (Jahn & Fasshauer, 2012;Kochubey, Lou, & Schneggenburger, 2011). By website searching, we learn that all these three molecules have the motifs for CREB binding. Thus, the hTau-induced reduction of SYT may be independent of CREB. Previous study demonstrated that overexpressing hTau decreased sEPSC with a disrupted cal cium homeostasis (Yin et al., 2016), suggesting that hTau impairs calcium-related presynaptic release. As a calcium sensor, SYT may be more vulnerable to the hTauinduced impairments.
PKA is a crucial kinase in phosphorylating several learning and memory-associated proteins, such as CREB at Ser133 and GluA1 at Ser845 (Oh et al., 2006;Shen et al., 2016). Here, we observed that overexpression of hTau in hippocampal neurons pronouncedly impaired PKA signaling in both total lysates and the nuclear frac tion compared with the vector control. The role of PKA on memory maintenance in different brain subset and pathologies is not always consistent, some researchers show that Aβ inhibits PKA, while acti vating PKA in hippocampus can improve cognitive functions (Gong et al., 2004;You et al., 2018). Other studies show that PKA accel erates tau phosphorylation, while inhibiting PKA in the prefrontal cortex improves memory in both aged rats and monkeys (Wang et al., 2018;Xie et al., 2016). We observe that rolipram restores memory and LTP deficits in hTau infusion mice. PKA activation by rolipram also rescues CREB/GluA1 phosphorylation with BDNF level and an enhanced surface GluA1 expression, and the latter is Ser845-phosphorylation-dependent and plays an important role in synaptic transmission (Oh et al., 2006). Intriguingly, activating PKA by rolipram increased mRNA levels of BDNF with no significant change of pCREB in total cell extracts. A previous study showed that overexpressing hTau increased pCREB in the nuclear fraction (Yin et al., 2016), and the latter is the active pool of CREB binding to BDNF promoter region and thus to promote BDNF expression.
Additionally, the pCREB can recruit CREB-binding proteins, such as C/EBP, which can also regulate expression of BDNF (Alberini & Chen, 2012;Hayes, Towner, & Isackson, 1997). Furthermore, the re covered dendritic length and complexity are also seen after rolipram treatment. These data indicate that PKA may serve as a potential target of tauopathies.  (Weise et al., 2018). Aβ treatment of the cul tured hippocampal neurons leads to the inactivation of PKA with persistence of its regulatory subunit PKAR2α (Vitolo et al., 2002).
F I G U R E 6 Overexpressing hTau disrupts PA28γ-20S proteasome complex formation and downregulating PKAR2α or upregulating PA28γ attenuates hTau-induced PKA/CREB inhibition. (a) Overexpressing hTau increases association of PKAR2α with PA28γ. The cultured hippocampal neurons infected with lenti-syn-hTau-mCherry virus for 7 days, and PA28γ was co-immunoprecipitated by anti-PA28γ antibody and detected by Western blotting using anti-PKAR2α and anti-PA28γ antibody. For all co-immunoprecipitations, results are received from three independent experiments. (b) Overexpressing hTau decreases association of PKAR2α with 20S proteasome subunit. PKAR2α was co-immunoprecipitated using anti-PKAR2α antibody and detected by Western blotting using anti-20S proteasome subunit and anti-PKAR2α antibody. (c) Overexpressing hTau decreases association of PA28γ with 20S proteasome subunit. 20S proteasome subunit was coimmunoprecipitated using anti-20S protease subunit and detected by Western blotting using anti-PA28γ antibody and anti-20S proteasome subunit. (d) Tau dose not bind to PA28γ. The cultured hippocampal neuron extract was co-immunoprecipitated using anti-PA28γ, and tau was not detected in the precipitates by Tau5. (e, f) Downregulating PKAR2α by shRNA attenuates hTau-induced CREB dephosphorylation. The primary cultured hippocampal neurons (5 div) were co-infected with hTau and shPKAR2α lentivirus. After 7 days, the indicated proteins were measured by Western blotting. (g, h) Upregulating PA28γ attenuates hTau-induced PKAR2α elevation and CREB dephosphorylation. The hTau and PA28γ lentivirus were co-infected in primary cultured hippocampal neurons (5 div) for 7 days, and the indicated proteins were measured by Western blotting. Data were expressed as mean ± SEM. For Western blotting, the n number was 9 (from nine different batches of cultured hippocampal neurons). *p < .05, **p < .01, versus Vec, & p < .05, versus hTau in (f), # p < .05, ## p < .01, versus hTau in (h) A significantly elevated PKAR2α in the nuclear fraction with un changed cAMP level was also shown in the present study after over expressing hTau. Therefore, the elevation of nuclear PKAR2α may be responsible for the hTau-induced PKA inhibition. Three major intracellular clearance systems, including the autophagic-lysosomal network (ALN), chaperone-mediated autophagy (CMA), and the ubiquitin-proteasome system (UPS), have all been identified in neu rons. The ALN and CMA mainly function in the cytoplasm, while the UPS mainly operates in the nuclei (Boland et al., 2018). Studies also show that P301L-tau inhibits 26S proteasome (Myeku et al., 2016).
These findings suggest a link between nuclear-elevated PKAR2α and the impaired proteasome system. To explore the mechanism underlying hTau-induced PKAR2α elevation and proteasome inhibition in nuclear fraction, we stud ied PA28γ, the nuclear activator of 20S proteasome subunit. A significantly increased PA28γ-PKAR2α interaction and decreased complex formation of PA28γ and 20S proteasome subunit with a reduced association of 20S proteasome subunit with PKAR2α were detected in hTau-overexpressing neurons (Figure 6a-c). These data indicate that proteasome inhibition is responsible for the hTau-induced nuclear elevation of PKAR2α and the latter mediates hTau-induced PKA inhibition and CREB dephosphorylation. PKA increases 26S proteasome abundance and stability by promoting the association of 20S core with the 19S regulatory complex, facil itating protein degradation (Lokireddy et al., 2015). Activation of proteasome through PKA leads to proteasome-mediated degrada tion in APP/PS1 mice and Tg4510 mice (Myeku et al., 2016;Smith, Pozueta, Gong, Arancio, & Shelanski, 2009). We observed that the proteasome activation by PA28γ overexpression attenuated

| Animals, stereotaxic surgery, and drug treatment
Wild-type male C57BL/6J mice (2-month-old, 110 ± 10 g) were pur chased from Experimental Animal Center of Wuhan University and kept with accessible food and water under a 12hr light/dark cycle.
All animal experiments were approved by the Ethics Committee of Tongji Medical College. Mice were anesthetized with isoflurane and placed on a stereotaxic apparatus and then were sterilized with iodophor, and the scalp was incised along the midline of the head. Hole was stereotaxically drilled in the skull at posterior 2.2 mm, lateral 2.6 mm, and ventral 2.3 mm relative to the bregma.

Using a microinjection system (World Precision Instruments),
AAV-syn-hTau-EGFP or control vector (1 μl, 3.78 × 1,012 viral ge nome per milliliter) was injected into bilateral hippocampal CA3 region at a rate of 0.125 μl/min, the needle was kept for 10 min before withdrawal, the skin was sutured, and mice were placed beside on a heater for recovery. Rolipram was diluted to 3 mg/ml with sterile 0.9% saline containing 5% (vol/vol) Tween-80 and 5% (vol/vol) PEG-400. Four weeks after virus infusion, rolipram was intraperitoneally injected (0.03 mg/kg) for continuous 20 days. Hippocampus was isolated and minced into ice-cold Hank's buff ered saline solution, and incubated in 0.25% trypsin at 37°C for 15 min, then using a pipette gently triturate the tissue eight to ten times to dissociate the cells and obtain a homogenous cell sus pension, and determine the density of cells on a hemocytometer.

| Cell culture
Neurons were planted with the F-12 medium containing 10% FBS on a coverslip in a 12-well plate with 20,000 cells per well or on a six-well plate with 100,000 cells per well. 4 hr after planting, planting medium was gently removed and 1.5 ml of fresh main tenance medium containing 97% neurobasal medium, 2% B27, and 1% Glutamax warmed at 37°C was added into each well.
Lentiviruses were used with MOI of 10 in primary hippocampal neurons at 5 div.

| Western blotting and immunoprecipitation
Western blotting was performed as previously described. Briefly, N2a cells or primary neurons were collected and lysed with RIPA buffer for 10 min, and then centrifuged at 12,000 g for 10 min at 4°C. Mice were anesthetized with isoflurane and then sacri ficed. The mice brain subset hippocampal CA3 were separated with vibrating microtome (Leica, VT1000S) on ice-cold PBS ac cording to the mouse brain atlas, and then, the CA3 tissue was homogenized brain with RIPA buffer and centrifuged at 12, 000 g for 15 min at 4°C. Then, the supernatants were collected and the protein concentration was measured using the BCA method, and loading buffer was added before electrophoresis. The protein was separated by SDS/PAGE and transferred onto nitrocellulose mem branes (Whatman), then incubated with primary antibodies at 4°C overnight. Incubation of secondary antibodies was performed at room temperature for 1 hr. The membranes were developed using the ECL detection system.
For immunoprecipitation, the cultured hippocampal neurons were lysed on ice with RIPA buffer for 10 min containing 1:100 PMSF

| Measurement of cAMP
Monoclonal Anti-cAMP antibody-based direct cAMP enzyme-linked immuno-absorbent assay (ELISA) kit was used to measure the con tent of cAMP in Vec-and hTau-transfected cells. Experiment was performed as described in the instruction.

| PKA activity
PKA activity was measured by PKA kinase activity kit (Enzo Life Sciences, ADI-EKS-390A), which is based on a solid phase ELISA that utilizes a synthetic peptide as a substrate for PKA and a poly clonal antibody that recognizes the phosphorylated form of sub strate. The absorbance was measured in a microplate reader at 450 nm.

| Immunofluorescence
Neurons fixed in 4% paraformaldehyde for 15 min or frozen brain sections were permeabilized in 0.5% Triton X-100 phosphate buffer for 30 min, followed by incubation with phosphate buffer contain ing 0.1% Triton X-100 and 3% BSA to block nonspecific binding. The incubation of primary antibodies was performed at 4°C overnight.

| Dendrite and spine analysis
Dendritic length and complexity were analyzed by Sholl analysis using Image-Pro Plus 6.0 software. The hippocampal neurons (5 div) were infected with lenti-syn-mCherry-hTau or the control virus, after cul tured for another 13 days, and the neurons were fixed in 4% paraform aldehyde, and then incubated with polyclone anti-MAP2 antibody at 4°C overnight. The neurons were washed in PBS and incubated with Alexa flour 488 secondary antibody at room temperature for 1 hr.
Images were captured using Carl Zeiss LSM710 confocal microscope.
At least 15 cultured neurons from three different cultures were used for quantitative analysis per group by single-blind method.
For in vitro spine analysis, primary hippocampal neurons (12 div) were infected with lenti-EGFP and lenti-syn-hTau-mCherry or the control vector, and the fluorescent images were obtained at 21 div with a confocal microscope (LSM710, Zeiss, Germany). The spine number in the dendrites per 10 mm was analyzed using an ImageJ software.

| Novel object recognition
The mice were habituated to the box (50 cm × 50 cm × 50 cm plas tic container) for 5 min before the test. The box was cleaned with 70% ethanol after each mouse test. In the training trail, mice were granted 5 min to familiarize themselves with A object and B ob ject. The box was cleaned with 70% ethanol between each trail as well. Exactly 24 hr after the familiarization period, B object was re placed with a novel C object, and the mice were granted 5 min to explore both objects. The behavior was recorded by a video cam era positioned above the box. The recognition times (T) for A, B, and C objects were recorded and calculated. The recognition index (TA/(TA + TB), TB/(TA + TB) and TC/(TA + TC)) and discrimination index (TC − TA)/(TA + TC) were analyzed as described previously (Lueptow, 2017).

| Morris water maze
The MWM test was performed from 2:00 p.m. to 8:00 p.m. as de scribed before (Yin et al., 2016). In the learning period, three trials a day for consecutive 6 days, mice were granted in the water maze to find a hidden platform under the milky water for 60 s. In each trial, the mice started from one of the four quadrants (exclude the target quad rant) facing the wall and ended when the mice arrived on the platform.
The mice were guided onto the platform if the mice did not find the platform in 60 s. The latency and the swimming path were recorded by a video camera 1.5 m from the water surface. Two days after training, the spatial memory was measured. Mice were granted to explore the water maze for 60 s with the platform removed. The time used to enter the target platform area, time spend in the target quadrant, and the number of target platform crossings were recorded by a digital video camera connected to a computer (Chengdu Taimeng Software Co. Ltd).

| Whole-cell patch-clamp recording
Primary rat hippocampal neurons (8 div) cultured in a dish of 35 mm diameter were infected with lentisynhTaumCherry or the control vector for 7 days and then treated with rolipram for 12 hr. For wholecell patch-clamp recording, culture medium was replaced with the bath solution containing (in mM): 140 NaCl, 3 KCl, 1.5 MgCl 2 , 10 HEPES, 11 Glucose, and 2.5 CaCl 2 (pH = 7.4), which were supple mented with 1 μM tetrodotoxin and 2 μM bicuculline. The micro electrodes (3-6 MΩ) were tip-filled with internal solution composed of 128 mM Cs-methanesulfonate, 17.5 mM CsCl, 9 mM NaCl, 1 mM MgCl 2 , 0.2 mM EGTA, and 10 mM HEPES (pH 7.4). Cells were held at −70 mV in voltage-clamp mode to record mEPSC. Input and series resistances were monitored continuously, and data were discarded if either of them changed more than 20%. All data were obtained using a MultiClamp 700B patch-clamp amplifier, sampled at 10 kHz, and filtered at 2 kHz using a Digidata 1550B analog-digital inter face (molecular devices). mEPSCs were analyzed using Clamp fit 10.6 (Axon Instruments).

| Golgi staining
Golgi staining was performed by using a FD Rapid Golgi Stain Kit (FD neurotechnology, PK401). Briefly, mice were anesthetized by isoflurane and intracardially perfused with normal saline for 10 min.
Brains were dissected out and immersed in an impregnation solu tion (equal solution A and B, containing mercuric chloride, potassium dichromate, and potassium chromate) for 24 at room temperature, impregnation solution was replaced, and samples were stored for another 2 weeks. Then, the brains were transferred to solution C for 72 hr. The brain was cut into sections of 100 μm thickness by a Leica VT1000s vibration microtome, and sections were mounted on gelatin-coated microscope slides with solution C. Slides were rinsed twice in distilled water (2 min each) and then placed in a mixture of solution D:E:distilled water (1:1:2) for 10 min. After rinsing for dis tilled water, sections were dehydrated in 50%, 75%, 95%, and 100% ethanol for 4 times (5 min each). Sections were immersed in xylene (Sinopharm Chemical Reagent, 10023418) for three times (5 min each) and then mounted with cytoseal. The images were taken using Olympus BX60 (Tokyo). ImageJ software was used to analysis the spine density, which was presented as the number of spines per 10 μm of dendritic length.

| Thioflavin-S staining
To confirm the existence of tau fibrils, Thioflavin-S (Sigma) staining was performed. Mouse brain slices with CA3 injected with AAV-GFP or AAV-hTau-EGFP were incubated in 0.25% potassium permanga nate solution for 20 min and then rinsed in distilled water and incu bated in bleaching solution (containing (in distilled water): 2% oxalic acid and 1% potassium metabisulfite) for 2 min. The slices were in cubated with blocking solution (containing (in water): 1% sodium hy droxide and 0.9% hydrogen peroxide) for 20 min, 0.25% acidic acid for 5 s, then washed and stained for 5 min with 0.0125% Thioflavin-S in 50% ethanol. Then, the slices were washed with 50% ethanol for 3 times and PBS for another 3 times. Then, the slices were covered with mounting solution and inspected under a confocal microscope.

| Statistical analysis
Data were expressed as mean ± SEM and analyzed by GraphPad Prism 6 statistical software (GraphPad Prism). The Student's t test was used for the comparison between two groups, and the one-way ANOVA or two-way ANOVA was used to analyze the data among three or four groups. The statistical significance was assessed at p < .05.