YAP prevents premature senescence of astrocytes and cognitive decline of Alzheimer's disease through regulating CDK6 signaling

Abstract Senescent astrocytes accumulate with aging and contribute to brain dysfunction and diseases such as Alzheimer's disease (AD), however, the mechanisms underlying the senescence of astrocytes during aging remain unclear. In the present study, we found that Yes‐associated Protein (YAP) was downregulated and inactivated in hippocampal astrocytes of aging mice and AD model mice, as well as in D‐galactose and paraquat‐induced senescent astrocytes, in a Hippo pathway‐dependent manner. Conditional knockout of YAP in astrocytes significantly promoted premature senescence of astrocytes, including reduction of cell proliferation, hypertrophic morphology, increase in senescence‐associated β‐galactosidase activity, and upregulation of several senescence‐associated genes such as p16, p53 and NF‐κB, and downregulation of Lamin B1. Further exploration of the underlying mechanism revealed that the expression of cyclin‐dependent kinase 6 (CDK6) was decreased in YAP knockout astrocytes in vivo and in vitro, and ectopic overexpression of CDK6 partially rescued YAP knockout‐induced senescence of astrocytes. Finally, activation of YAP signaling by XMU‐MP‐1 (an inhibitor of Hippo kinase MST1/2) partially rescued the senescence of astrocytes and improved the cognitive function of AD model mice and aging mice. Taken together, our studies identified unrecognized functions of YAP‐CDK6 pathway in preventing astrocytic senescence in vitro and in vivo, which may provide further insights and new targets for delaying brain aging and aging‐related neurodegenerative diseases such as AD.


| INTRODUC TI ON
As one of the most important body organs, the brain is particularly sensitive to aging (Yankner et al., 2008). Brain aging is accompanied by a decline in various cognitive functions (Yankner et al., 2008).
In addition, the risk for aging-related neurodegenerative diseases, such as Alzheimer's disease (AD) (Ziegler-Graham et al., 2008) and Parkinson's disease (PD) (Myall et al., 2017;Reeve et al., 2014), exponentially increase with age. Therefore, how to delay brain aging and prevent aging-related diseases is critical for human health.
Brain aging begins with the senescence of cells in the brain.
Recent studies have shown that astrocytic specific genes alter their expression patterns in human hippocampus during aging (Soreq et al., 2017). Interestingly, astrocytic senescence has been proposed as a component of AD (Bhat et al., 2012;Garwood et al., 2017).
The Hippo signaling pathway is a key regulator of stem cell selfrenewal, tissue regeneration and organ size, and is a conserved classical pathway formed by the kinase cascade (F. X. Yu et al., 2013; F. X. Yu et al., 2015;F. X. Yu et al., 2012). As a key effector of the Hippo signaling pathway, YAP has been shown to play an important roles in cell proliferation, differentiation and tissue regeneration Rosado-Olivieri et al., 2019; C. Xie et al., 2020). Recent evidence has shown that YAP signaling is closely related to individual aging and cellular senescence (Fu et al., 2019;Iwasa et al., 2013;Jia et al., 2018;Jin et al., 2017;Santinon et al., 2018;Q. Xie et al., 2013;. YAP-1 deficiency promotes health aging of Caenorhabditis elegans (Iwasa et al., 2013), and YAP inhibits the senescence of human fibroblast cells (Santinon et al., 2018;Q. Xie et al., 2013), human mesenchymal stem cells (Fu et al., 2019), human periodontal ligament stem cells (Jia et al., 2018), glioma cells (X. Xu et al., 2020), and hepatic stellate cells (Jin et al., 2017). However, it remains unclear whether YAP signaling regulates the senescence of astrocytes in aged brain and AD.
In the present study, we found that YAP was downregulated and inactivated in senescent astrocytes in vitro and in vivo, and YAP prevented the senescence of astrocytes through the CDK6 signaling.
Our study provides a new molecular mechanism for regulating astrocytic senescence, which may provide new insights and targets for delaying brain aging and aging-related neurodegenerative diseases.

| YAP is downregulated and inactivated in hippocampal astrocytes of aged mice and AD model mice
To explore the potential functions of YAP in the aging brain, hippocampal tissues of 2 M and 24 M old mice were collected. SAβ-gal staining, which is a classical method for testing cell or tissue senescence (Dimri et al., 1995), showed the increased β-gal intensity in hippocampus of 24 M old mice, compared with that in 2 M old mice ( Figure S1a). In these hippocampus of 24 M old mice, the protein levels of YAP were significantly decreased, as well as the decrease in Lamin B1 (a senescent marker) levels, which is a common feature of senescent cells (Freund et al., 2012;Shimi et al., 2011), however, both p-YAP and p-YAP/YAP ratio were significantly increased (Figure 1ad). Moreover, the phosphorylation levels of the Hippo kinases such as LATS1, MST1, MOB1 and YAP were increased significantly in old mice, compared with that in young mice (Figure S1b-h). These results suggest that YAP is downregulated and inactivated in aged brains in a Hippo pathway-dependent manner. Our previous studies have shown that YAP is mainly expressed in astrocytes and neural stem cells of the brain (Z. . Indeed, as shown in To further confirm whether YAP is also downregulated and inactivated in aging-related diseases such as AD, YAP expression was detected in the hippocampal tissues of APP/PS1 mice ( Figure   S1i), a transgenic mouse model of AD (Lok et al., 2013). As shown in Figure 1g-j, again, both YAP and Lamin B1 expression were decreased, however, p-YAP levels and the p-YAP/YAP ratios were increased in the hippocampus tissues of AD model mice. The phosphorylation levels of the Hippo kinases such as LATS1, MST1, MOB1 and YAP were elevated significantly in the AD model mice, compared with that in wild-type (WT) mice (Figure S1j-p), implying that the downregulation and inactivation of YAP in AD model mice was in preventing astrocytic senescence in vitro and in vivo, which may provide further insights and new targets for delaying brain aging and aging-related neurodegenerative diseases such as AD.

K E Y W O R D S
alzheimer's disease, astrocytes, CDK6, senescence, YAP dependent on the Hippo pathway. Moreover, immunohistochemical staining also showed that YAP expression was suppressed in the hippocampal astrocytes of AD model mice (Figure 1k,l). Taken together, these results indicate that YAP is downregulated and inactivated in the hippocampal astrocytes of aged mice and AD model mice, which may contribute to brain aging.

| YAP is downregulated and inactivated in the naturally senescent astrocytes and D-galactose (Dgal) or paraquat (PQ)-induced senescent astrocytes in vitro
To further examine the expression pattern of YAP in senescent astrocytes, the primary astrocytes were cultured to DIV 90, which is considered as naturally senescent astrocytes (Pertusa et al., 2007), interestingly, in these naturally senescent astrocytes, both Lamin B1 and YAP were significantly decreased, however, p-YAP and p-YAP/YAP levels were increased . To further investigate the functions of YAP in pathological senescent astrocytes, D-galinduced senescent astrocyte models (Y. Shen et al., 2014) were used.
In our system, primary cultured astrocytes were treated with D-gal at 111 mM for 5 days, and then SAβ-gal staining confirmed the senescence of astrocytes ( Figure S2a-c). Moreover, mRNA levels of p16 (a senescent marker) were significantly increased in these senescent astrocytes ( Figure S2d). In addition, these astrocytes showed To further confirm our results, another senescent astrocyte model was established as reported previously (Chinta et al., 2018).
Young astrocytes were treated with PQ for 24 h, then, PQ was removed from the culture medium, and the astrocytes were recovered for another 8 days ( Figure S2i). As shown in Figure S2j-l, SAβ-gal staining revealed that the percentage of β-galactosidase positive astrocytes was elevated significantly after PQ treatment, while the expression of Lamin B1 was significantly decreased, which indicates that PQ could induce the senescence of astrocytes. Similarly, in these senescent astrocytes, YAP expression was significantly reduced, whereas p-YAP level and p-YAP/YAP ratio were increased, as well as the increase in p53 level and decrease in Lamin B1 level To determine whether the inactivation of YAP in senescent astrocytes in vitro is dependent on activation of the Hippo pathway, the protein levels of Hippo kinases were detected. As shown in Figure S3a-g, the protein levels of LATS1, MST1, and SAV1 were decreased in senescent astrocytes, whereas the levels of p-LATS1/ LATS1, p-MST1/MST1, and p-MOB1 were increased in senescent astrocytes, indicating that inactivation of YAP signaling in astrocytic senescence is dependent on Hippo pathway. Moreover, mRNA levels of LATS1, MST1, and SAV1 were significantly decreased ( Figure   S3h-j) in senescent astrocytes, which indicated that the reduction of these proteins after senescence was probably due to suppressed transcription of these genes. Taken together, these results suggest that YAP is downregulated and inactivated both in the naturally senescent astrocytes and D-gal-or PQ-induced senescent astrocytes in a Hippo pathway-dependent manner.

| YAP deletion in astrocytes promotes the premature senescence of astrocytes in vitro and in vivo
To examine the roles of YAP in senescent astrocytes, YAP GFAP -CKO mice (conditionally knockout YAP in astrocytes) were generated, by crossing YAP f/f mice with GFAP-cre transgenic mice. Both western blot and immunostaining showed that YAP was efficiently knockout in YAP GFAP -CKO mice and cultured YAP −/− astrocytes ( Figure S4a Moreover, more significant increase in β-galactosidase positive cells and more potent reduction of Lamin B1 were found in YAP −/− astrocytes, which suggest that YAP −/− astrocytes was more vulnerable to senescence induction, compared with YAP +/+ astrocytes (Figure 3af). These results suggest that YAP deletion promotes the premature senescence of astrocytes in vitro.
Again, in the hippocampus of old YAP f/f mice, the expression of YAP and Lamin B1 was significantly decreased, and p53 expression was significantly increased, compared with that in young YAP f/f mice. As expected, in YAP GFAP -CKO mice, p53 expression was significantly increased than that in young YAP f/f mice and old YAP f/f mice, however, Lamin B1 expression was significantly decreased

| YAP prevents premature senescence of astrocytes through the CDK6 pathway
How does YAP prevent the senescence of astrocytes? Previous studies have reported that knockdown of YAP promotes the senescence of IMR-90 cells through negatively regulating CDK6, which is a cyclin-dependent kinase (Q. Xie et al., 2013), thus we tested whether YAP prevents the senescence of astrocytes through CDK6 signaling.
In the hippocampus of old YAP f/f mice, the expression of YAP and CDK6 were significantly decreased, compared with that in young YAP f/f mice, interestingly, YAP and CDK6 expression were decreased in old YAP GFAP -CKO mice, compared with that in old YAP f/f mice To further examine whether CDK6 is a target gene of YAP downstream to regulate the senescence of astrocytes, YAP +/+ and YAP −/− astrocytes were cultured. Real-time PCR and western blot showed that both mRNA and protein level of CDK6 were significantly decreased in YAP −/− astrocytes, respectively, not only in control astrocytes, but also in D-gal-induced senescent astrocytes (Figure 4e-g), which suggest that YAP is required for the expression of CDK6. Furthermore, overexpression of CDK6 partially rescued D-gal-induced senescence of YAP −/− astrocytes (Figure 4h-j). Taken together, these results suggest that YAP prevents the senescence of astrocytes through the CDK6 pathway.

| Activation of YAP by suppressing the Hippo pathway delays the senescence of astrocytes and improves the cognitive function of aged mice and AD model mice
To F I G U R E 1 YAP is downregulated and inactivated in hippocampal astrocytes of aged mice and AD model mice. (a) Western blot analysis of p-YAP, YAP, and Lamin B1 expression in the hippocampus of 2 M and 24 M mice. (b-d) Quantification of relative expression of p-YAP/ YAP (b), YAP (c) and Lamin B1 (d) as shown in (a) (n = 3 per group, normalized to 2 M mice). (e) Double immunostaining analysis of YAP (green) and GFAP (red) in the hippocampus of 2 M and 24 M mice. (f) Quantitative analysis of YAP + cells over total astrocytes as shown in (e) (n = 3, normalized to 2 M mice). (g) Western blot analysis of p-YAP, YAP, and Lamin B1 expression in the hippocampus of WT and AD model mice (6 M). (h-j) Quantification of relative expression of p-YAP/YAP (h), YAP (i) and Lamin B1 (j) as shown in (g) (n = 3, normalized to WT mice). (k) Double immunostaining analysis of YAP (green) and GFAP (red) in the hippocampus of WT and AD model mice (6 M). (l) Quantitative analysis of YAP + cells over total astrocytes as shown in (k) (n = 3, normalized to WT mice). Scale bar, 100 μm. Data were mean ± s.e.m. * p < 0.05, ** p < 0.01, *** p < 0.001 F I G U R E 2 YAP is downregulated and inactivated both in naturally senescent astrocytes and D-gal or PQ-induced senescent astrocytes.

| DISCUSS ION
In the present study, we found that YAP signaling was downregulated and inactivated through activation of Hippo kinases in senescent astrocytes in vitro and in vivo. Mechanistically, YAP prevented premature senescence of astrocytes through the CDK6 signaling.
Activation of YAP signaling by XMU-MP-1 treatment delayed the senescence of astrocytes in vitro and in vivo, and thereby improving the cognitive function of aged mice and AD model mice (see Graphical abstracts). Our study provides a new molecular mechanism for regulating astrocytes senescence, which may provide new targets and directions for delaying brain aging and neurodegenerative diseases.
Previous studies have shown that YAP is downregulated in IMR90 cells during senescence and knockdown of YAP promotes premature senescence of IMR90 cells (Q. Xie et al., 2013), and accelerates the senescence of human mesenchymal stem cells (Fu et al., 2019) and hepatocytes (Jin et al., 2017). Consistent with these previous results, we found that YAP was downregulated and inactivated in senescent astrocytes in vivo and in vitro in a Hippo pathway-dependent manner. We also note that one previous study has shown that the nuclear YAP was remarkably decreased in neurons under AD pathology due to the sequestration into cytoplasmic amyloid beta aggregates, which indicates YAP-dependent neuronal necrosis in AD (Tanaka et al., 2020). However, in our studies, we found that YAP was mainly  (c) (n = 9, normalized to WT mice). (e) qPCR analysis of relative CDK6 mRNA levels in YAP +/+ and YAP −/− control and D-gal-induced senescent astrocytes (n = 3 per group, normalized to control YAP +/+ astrocytes). (f) Western blot analysis of YAP and CDK6 expression in YAP +/+ and YAP −/− control and D-gal-induced senescent astrocytes. (g) Quantification of relative CDK6 expression as shown in (f) (n = 3 per group, normalized to control YAP +/+ astrocytes). (h) Protocol for CDK6 transfection in D-gal-treated astrocytes. (i) Representative images of SAβgal staining of YAP +/+ and YAP −/− astrocytes without transfection, or D-gal-induced senescent YAP +/+ and YAP −/− astrocytes transfected with control vector or CDK6 plasmid (CDK6-Res) for 2 days. (j) Quantification of the percentage of β-galactosidase + cells over total cells as shown in (i) (n = 15 per group). Scale bar, 20 μm. Data were mean ± s.e.m. * p < 0.05, ** p < 0.01, *** p < 0.001 tau aggregates, and drive neurodegenerative disease (Bussian et al., 2018). Interestingly, beta-amyloid triggered senescence of astrocytes in vitro, which implied mutual promotion of astrocyte senescence and beta-amyloid deposition (Bhat et al., 2012). Senescent astrocytes express a characteristic of SASP, which in turn could elicit deleterious effects on the surrounding neurons, disturb the maintenance of homeostasis in brain, in large part, contribute to age-related inflammation and chronic neurodegenerative diseases (Bhat et al., 2012;Han et al., 2020;Hou et al., 2018;Hou et al., 2017;Hou et al., 2019;Salminen et al., 2011;C. Yu et al., 2017). In our study, we found that astrocytic YAP was downregulated in aged mice and AD model mice in a Hippo kinases dependent manner, and deletion of YAP or activation of YAP in astrocytes promotes or delays the aging of the brain, respectively, indicating that senescent astrocytes by YAP downregulation may contribute to aging of the brain and AD.
Several previous studies have revealed that YAP-CDK6 pathway inhibits cellular senescence (Q. Xie et al., 2013;Yang et al., 2021). Consistent with previous studies, in our studies, several lines of evidences suggest that YAP-CDK6 signaling mediates D-gal-induced senescence of astrocytes. Firstly, YAP was F I G U R E 5 Activation of YAP by suppressing the Hippo pathway delays D-gal-induced senescence of astrocytes in vitro. (a) Representative images of SAβ-gal staining of control YAP +/+ and YAP −/− astrocytes under various conditions. (b) Quantification of the percentage of β-galactosidase + astrocytes over total astrocytes as shown in (a) (n = 20 per group). (c) Western blot analysis of p-YAP, YAP, Lamin B1 and CDK6 protein expression in control astrocytes, D-gal-induced senescent astrocytes with or without XMU-MP-1 treatment. (d) Quantification of relative expression of p-YAP/YAP (n = 7), YAP (n = 7), Lamin B1 (n = 4) and CDK6 (n = 5) (normalized to control astrocytes) as shown in (c). (e) Double immunostaining analysis of Ki67 (green) and GFAP (red) in control astrocytes, D-gal-induced senescent astrocytes with or without XMU-MP-1 treatment. (f) Quantitative analysis of the percentage of Ki67 + astrocytes over total astrocytes as shown in (e) (n = 45 per group). (g) Representative images of SAβ-gal staining of cultured astrocytes with or without CP-10 treatment. (h) Quantification of the percentage of β-galactosidase + astrocytes over total astrocytes as shown in (g) (n = 15). (i) Representative images of SAβ-gal staining of cultured astrocytes under various conditions. (j) Quantification of the percentage of β-galactosidase + astrocytes over total astrocytes as shown in (i) (n = 15 per group). ns, not significant. Scale bar, 20 μm. Data were mean ± s.e.m. * p < 0.05, ** p < 0.01, *** p < 0.001 downregulated not only in the hippocampal astrocytes of aging mice and AD model mice, but also in D-gal and PQ-induced senescent astrocytes. Secondly, YAP deletion in astrocytes significantly reduced CDK6 expression, aggravated cell senescence, and impaired cognitive function. Thirdly, overexpression of CDK6 partially rescued the senescence of astrocytes, and inhibition of CDK6 abolished XMU-MP-1's preventive effects on senescence of astrocytes. Moreover, as shown in Figure S7a-c, the transcription factor FOXM1, which is the substrate of CDK6 and protects cancer cells from senescence (Anders et al., 2011;Rader et al., 2013), showed decreased expression in senescent astrocytes, and another substrate of CDK6, the oxygen sensor PHD1 (Kennel et al., 2018;Ortmann et al., 2016), showed increased expression, indicating that FOXM1 and PHD1 might be the downstream targets of CDK6 during astrocytic senescence. In future, how YAP-CDK6 signaling prevents astrocytic senescence requires further study.
Some strategies are developed to remove senescent cells relatively specifically and ameliorate the condition of the aging persons, such as the senolytic therapy (combined use of dasatinib and quercetin) (M. Xu et al., 2018;. In our study, we found that activation of YAP by XMU-MP-1 partially improved the cognitive function of AD model mice and old mice, and inhibition of CDK6 promoted astrocytic senescence. These evidences indicated that activation of the YAP-CDK6 pathway may delay brain aging and aging-related neurodegenerative diseases. However, a number of studies have demonstrated that YAP and CDK6 promote tumorigenesis Takeuchi et al., 2017). It is possible that the YAP-CDK6 pathway may promote tumor progression under excessive activation, and may rejuvenate senescent astrocytes with moderate activation, thereby playing dual roles under cancer or cellular senescence conditions, depending on the activation degree of this pathway, although this remains to be studied further. Therefore, in future, to delay brain aging and aging-related neurodegenerative diseases, appropriate activation of YAP should be considered carefully, and further investigations are required.
In summary, our studies identified unrecognized functions of YAP in preventing astrocytic senescence in vitro and in vivo, revealed the YAP-CDK6 pathway in the negative control of astrocytic senescence, and discovered the new function of XMU-MP-1 in antiaging, which may provide new insights and targets for delaying brain aging and aging-related neurodegenerative diseases.

| cDNA constructs
pcDNA3.1-CDK6 (http://www.addge ne.org/75170/) was purchased from Addgene and its sequence was verified by sequencing. intraperitoneally at a dose of 1 mg/kg (Fan et al., 2016), given every 2 days, and lasted for 2 weeks. All the concentrations of these reagents are indicated above, if there is no special instruction.

| Cell culture and transfection
Primary astrocyte cultures were prepared from the cerebral cortex of P1-P3 mice as described previously (Z. . The purity of GFAP positive (a marker of astrocytes) cells in our culture system was more than 94%. Appropriate plasmids (2 μg per 35mm dish) were transfected into the cells using the Lipofectamine™ 3000 Transfection Reagent (L3000-015, Invitrogen) as the manufacturer's instructions.

| Immunostaining
For cultured cells, immunostaining was performed as previously described (X. Xu et al., 2020). Mounting was done after another three washes. Images were acquired by using a fluorescence microscopy or a confocal microscopy (Zeiss) and analyzed by Image J software.

| Quantitative real-time PCR (qRT-PCR)
Total RNA extraction, reverse transcription and the quantification of mRNA levels were done as previously described (X. Xu et al., 2020). β-actin or GAPDH was chosen as the endogenous con-

| SAβ-gal staining
SAβ-gal staining kit (G1580, Solarbio) was adopted to assess cellular senescence according to the protocol clearly provided by the manufacturer, as previously described (X. Xu et al., 2020).

| Spatial reference memory on the elevated Y-maze
The Y-maze task for hippocampus-dependent spatial reference memory was performed as previously described , with slight modifications. Briefly, mice were habituated in the testing room overnight before the spontaneous alternation test by using an opaque perspex Y-maze (40 cm long and 9 cm wide with 21 cm high walls). Each animal was placed in turn in arm A of the Y-maze and allowed to explore for 8 min, and the arm entries and walking distance were recorded. All equipment was cleaned between each trial with 75% ethanol. Spontaneous alternation was defined as a successive

| Barnes circular maze
The Barnes circular maze test was performed on a white circular surface, 92 cm in diameter, with 20 holes equally spaced around the perimeter as previously described (Nanou et al., 2016), with slight modifications. A black Plexiglas escape box (15 × 7 × 7 cm) containing paper cage bedding on its floor was located under one of the holes (defined as the target hole) in a room with a light intensity of 1200 lux. The location of the target was consistent for a given mouse. Firstly, adaptation was accomplished by each mouse. For each mouse, it was guided into the hole twice within 3 min, at an experiment interval of 15 min. Training lasted for 4 successive days (day 1-4) were performed, each mouse for 2 trials, with a 15 min interval. To start a trial, a test mouse was placed in the center of the maze, held there for 10 s by using a dark box, and then the mouse was allowed to explore freely. Once the mouse entered the escape box, it was remained undisturbed for 1 min with the light off. Mice that failed to find the target hole were gently guided to the escape box. On the fifth day (day 5), a probe trial was conducted. The escape box was changed to another hole and the maze was rotated to keep the spatial position of the target hole unchanged, in order to avoid odor guidance. All equipment was cleaned between each trial with 75% ethanol. Time latency to reach the target hole and the search errors were recorded.

| Statistical analysis
All data values were expressed as mean ± SEM derived from at least three independent experiments. GraphPad Prism software was used for statistical analysis. Student's t-test and ANOVA analysis were applied in this study. A p value of <0.05 was considered to be statistically significant.

CO N FLI C T O F I NTE R E S T
The authors declare no competing interests. and J.J.W also contributed to mouse breeding and genotyping. X.X.X and W.J.F wrote the manuscript. X.X.X, W.J.F, Y.W, and Z.Z.H contributed to revising the original text.

DATA AVA I L A B I L I T Y S TAT E M E N T
The authors declare that all data supporting the conclusions of this study are presented within the paper and the supplementary information files and are available from the authors.