The ER proteostasis network is altered in elderly mammals during normal brain aging or dementia

Initially, to study possible contributions of the UPR to age‐associated cognitive decline in humans, we analyzed RNA sequencing data from the hippocampus of normal subjects as well as those diagnosed with dementia in a unique aged population‐based cohort from the Adult Changes in Thought (ACT) study (http://aging.brain‐map.org). Remarkably, an unbiased functional enrichment analysis of differentially expressed genes (DEGs) in demented patients revealed that “unfolded protein binding and processing” are the most altered biological functions (Fig 1A). Moreover, XBP1 was predicted as one of the possible transcription factors driving such alterations in the proteostasis network (Fig 1A; Dataset EV1 for a complete analysis). In addition, chaperones accounted for the most altered genes in patients with dementia in the dataset analyzed (Fig 1B, Table 1), while some classical UPR mediators were not altered (Fig EV1A).

To determine the age‐related capacity of the mammalian brain to engage the UPR, we treated young, middle‐aged, and aged mice with intraperitoneal injection of tunicamycin, a well‐established pharmacological inducer of ER stress (Medinas et al, 2018; Fig 1C). This regimen was used to test whether activation of the UPR is altered by stress at various time points during aging. Tunicamycin is known to induce a high level of ER stress, thus facilitating measurement of UPR signaling. We monitored mRNA levels of Xbp1s, Hspa5 (Bip/Grp78), and Ddit3 (Chop) in the hippocampus, in addition to cerebral cortex and cerebellum. Remarkably, the capacity to induce Xbp1s was significantly attenuated in the hippocampus of middle‐aged and aged animals compared with young animals following experimental ER stress (Fig 1D, left). The UPR mediators Ddit3 and Hspa5 manifested similar findings (Fig 1D, middle and right), whereas other markers such as Atf3 and Pdia3 did not show significant alterations (Fig EV1D,E). A similar tendency was observed in the cerebrocortex of these animals (Fig EV1B), whereas no differences were detected in the cerebellum (Fig EV1C). These results suggest that tunicamycin can exert effects on UPR signaling in the brain differentially with age, particularly impacting the aged hippocampus. Finally, to exclude possible systemic effects of tunicamycin treatment, we injected the ER stressor directly into the hippocampus of young, middle‐aged, or aged animals (Fig 1C). We successfully recapitulated the results obtained following intraperitoneal injections, observing decreased levels of Xbp1s in the hippocampus of middle‐aged and aged animals 24 h after exposure when compared with young mice (Fig 1E). As a control, this ER stress‐triggering paradigm did not result in differential neurotoxicity during aging as measured by FluoroJade‐C staining in the site of injection (Fig EV1F,G).

Increased generation of reactive oxygen and nitrogen species, such as nitric oxide (NO), has been observed during aging (Go & Jones, 2017). Previous reports indicated that S‐nitrosylation of IRE1 inhibits its ribonuclease activity (Yang et al, 2015), resulting from posttranslational modification of cysteine thiol (or more properly thiolate anion) groups at Cys931 and Cys951 by NO‐related species (Nakato et al, 2015). Accordingly, we evaluated the levels of S‐nitrosylation of IRE1 (abbreviated SNO‐IRE1) during normal brain aging. We measured the ratio of SNO‐IRE1 to input IRE1 by biotin‐switch assay in the brains of young and old animals and found a significant increase in this ratio in aged samples (Fig 1F, Figure Source Data 1–2). This result may potentially account, at least in part, for the decrease in IRE1 activity and thus compromised ability to adapt to ER stress with age. Notably, when we also performed standard immunoblots of IRE1 to quantify levels by densitometry in young and aged animals, we found that total IRE1 decreased with aging (Fig 1G, Figure Source Data 3), as suggested previously based on mRNA levels (Gavilán et al, 2006). Thus, while the absolute level of S‐nitrosylated IRE1 may not be very different in young versus aged brain, the proportion of SNO‐IRE1 significantly increases with age (Fig 1F,G, Figure Source Data 1–3), suggesting overall inhibition of its enzyme activity. These results suggest that the occurrence of specific molecular alterations in the aging hippocampus interfere with the capacity to adapt to ER stress.

Genetic disruption of IRE1‐XBP1s accelerates or exacerbates aging‐associated cognitive decline

To evaluate and quantify cognitive decline associated with normal aging in rodents, we assessed young (3–6 months old), middle‐aged (12–14 months old), and aged (18–20 months old) mice using the novel object recognition (NOR), novel object location (NOL), contextual fear conditioning (CFC), and Barnes maze tests; all of which have been previously reported to be sensitive to age‐dependent alterations (Shoji et al, 2016). Importantly, aged mice interacted with objects in the same amount of time as young animals (Figs EV1H and 1J), thus excluding age‐related motor or sensory alterations that could impact the evaluation of the tests. Moreover, animals at all ages did not exhibit intrinsic preferences for the randomized objects (Fig EV1I) used in the tests nor the quadrants of the arena (Fig EV1K), and their performance was not impacted when evaluated at middle age and again at advanced age (Fig EV1L).

To begin to assess the role of the UPR in brain health span, we conditionally ablated the RNase domain of IRE1 in the nervous system using a CRE transgenic mouse line driven by the Nestin promoter (IRE1^cKO) for general deletion in nervous tissue (Fig EV2A) at the stage of neural progenitor cells (Duran‐Aniotz et al, 2017; Appendix Figs S1A and S2 for detailed workflow). Remarkably, middle‐aged IRE1^cKO animals lost their capacity to discriminate between novel objects in the NOR test when compared with Cre‐negative/wild‐type (WT) littermates (IRE1^WT). Note that control animals only displayed poor performance at advanced age (Fig 2A). The NOL test highlighted the impaired capacity of middle‐aged or older WT animals to discriminate newly located objects (Fig EV2B). Importantly, aged IRE1^cKO animals also showed this deficit, while younger IRE1^cKO mice exhibited normal behavior (Fig EV2B). The apparent absence of phenotypes in IRE1^cKO during youth on these two tests suggests the occurrence of age‐dependent deficits after IRE1 ablation. To further examine this hypothesis, we evaluated an additional cohort of mutant mice with a deletion of IRE1 mediated by Cre driven by the Camk2a promoter (IRE1^cKO/CaK; Appendix Fig S1B). Notably, we could recapitulate the same findings observed in IRE1^cKO mice in the NOR test, with young animals exhibiting no alterations upon IRE1 ablation but middle‐aged IRE1^cKO/Cak displaying impaired capacity to discriminate novel objects when compared with age‐matched controls (Fig EV2C).

IRE1 signaling has various outputs in addition to controlling XBP1 mRNA splicing, including degradation of mRNAs and regulation of cell signaling (Hetz et al, 2020). Accordingly, to further validate our results, we generated a small cohort of XBP1 conditional knockout animals (XBP1^cKO; Appendix Fig S1C). This model manifests normal IRE1 expression but lacks XBP1 in the hippocampus. Using this additional mouse model, we recapitulated the deficits observed in middle‐aged IRE1^cKO and IRE1^cKO/Cak mice on the NOR test (Fig EV2D). Since XBP1 ablation has been reported in some cases to trigger aberrant phenotypes because of compensatory IRE1 overactivation (Lee et al, 2008, 2011; Hur et al, 2012), we focused our analysis on IRE1‐deficient animals. In addition, we have previously reported that young XBP1^cKO animals displayed diminished performance on fear conditioning and maze‐based tests (Martínez et al, 2016), thus obfuscating interpretation of results when age‐dependent cognitive decline is being assessed. Hence, we evaluated a cohort of IRE1^cKO animals and their control littermates at different ages using the CFC test that evaluates freezing responses before and after presentation of a tone. On the day of testing, our analysis showed that young IRE1^cKO mice displayed increased freezing before presentation of the tone, indicative of altered contextual fear memory acquisition. With aging, however, freezing on this test significantly decreased (Fig 2B). Importantly, these results were not observed in WT littermates, which presented similar responses throughout aging (Fig 2B). Collectively, assessment of NOR and CFC data throughout aging suggested that genetic ablation of IRE1 in the nervous system may accelerate the emergence of age‐associated alterations in cognitive function.

IRE1^cKO displays early appearance of senescent cells in the hippocampus

Previous findings suggested that an increased number of senescent cells in brain tissue correlates with age‐dependent cognitive decline (Baker & Petersen, 2018). To determine whether disruption in the IRE1 pathway in the brain resulted in altered senescence, we evaluated the number of senescent cells using β‐galactosidase and p‐γ‐H2AX staining in hippocampal tissue. Histological analysis indicated progressive accumulation of senescent cells in the hippocampus as animals aged (Figs 2C,D and EV2E–G). Notably, an increase in the number of senescent cells comparable with that found in aged controls was observed in the brains of middle‐aged IRE1^cKO animals (Figs 2C,D and EV2E–G). These results suggest that genetic disruption of the IRE1 pathway in the hippocampus accelerates the appearance of senescent cells in the hippocampus, and this finding correlates with the development of cognitive impairment.

IRE1 deletion in the adult hippocampus accelerates age‐dependent cognitive decline

Genetic manipulations during development could result in compensatory mechanisms that do not necessarily reflect the direct contribution of a gene to the observed phenotype in the adult animal. To determine whether ablation of IRE1 expression in the adult could alter the cognitive capacity of mice in an age‐dependent manner, we locally delivered adeno‐associated viral vectors (AAV) to express CRE recombinase or control vector expressing GFP (AAV‐Mock) into the hippocampus of IRE1 floxed animals (Appendix Fig S1D). This strategy significantly reduced the levels of IRE1 mRNA in the hippocampus (Fig EV2H,I) and allowed us to test animals before and after AAV injections for comparison. Middle‐aged or young mice were tested on the NOR assay before AAV‐CRE injection and then monitored again 4 weeks after brain surgery. Remarkably, targeting IRE1 in the hippocampus of middle‐aged mice impaired the capacity to discriminate novel objects when compared with the group that received empty vector (Fig 2E). Importantly, this experimental setup did not impact the performance of young animals (Fig 2E), and thus recapitulated our previous findings in IRE1^cKO animals regarding the occurrence of an age‐dependent effect of IRE1 ablation that accelerated cognitive decline.

Previous reports indicated that normal aging results in a decrease in the number of dendritic spines, correlating with impaired cognition (Dickstein et al, 2013). To determine possible morphological alterations triggered by IRE1 deficiency during aging, we evaluated dendritic spine density on pyramidal neurons in the CA1 region following AAV‐GFP delivery to label neuronal morphology and dendritic arbors. Morphological assessment using confocal microscopy indicated that middle‐aged IRE1^flox/flox mice displayed a decrease in the density of dendritic spines in the CA1 region following AAV‐Cre/GFP injection when compared with age‐matched littermates injected with control vector expressing only GFP (Fig 2F,G). Importantly, as a control, expression of Cre in the brain of young IRE1^flox/flox animals did not alter the distribution of dendritic spines (Fig 2F,G), confirming the occurrence of age‐related alterations. Taken together, our results indicate that targeting IRE1 expression in the adult brain results in age‐dependent cognitive decline and morphological alterations, supporting our previous observations on IRE1^cKO, IRE1^cKO/Cak, and XBP1^cKO mice.

Aged Tg^XBP1s animals are less susceptible to age‐dependent cognitive decline and display reduced senescent cells in the hippocampus

To test the consequences of artificially bolstering an adaptive UPR during aging, we developed strategies to increase the levels of the spliced and active form of XBP1 in the brain (Fig EV3A). For this purpose, we initially evaluated cognitive performance during aging of transgenic mice that overexpressed XBP1s under the control of the PrP promoter (Martínez et al, 2016; referred to here as Tg^XBP1s, Appendix Figs S1E and S3, for detailed workflow).

Notably, XBP1s overexpression prevented the appearance of age‐related deterioration in brain function with middle‐aged or advanced‐age mice retaining their ability to discriminate between newly located objects and novel objects compared with WT littermates (Fig 3A,B). We have previously shown that young Tg^XBP1s perform better than young controls in a memory flexibility paradigm based on spatial testing (Martínez et al, 2016), indicating that Tg^XBP1s mice exhibit increased learning capacity in both spatial memory acquisition and flexibility. Here, we aimed to evaluate whether this transgenic line can maintain such capacity throughout aging using the Barnes Maze test, focusing mainly on spatial memory acquisition. Importantly, young Tg^XBP1s performed similarly to young Non‐Tg littermates, as indicated by the latency to find the target during 4 days of training (Fig EV3B). Remarkably, middle‐aged and aged Tg^XBP1s mice performed better than age‐matched control littermates, displaying decreased latency to find the target hole during both the skill acquisition days and on the test day (Fig 3C,D). Finally, using the CFC test, we were not able to detect spontaneous age‐related phenotypes in WT animals, and we also could not detect significant alterations in freezing responses between Tg^XBP1s and age‐matched littermates before tone presentation (Fig EV3C).

Next, we measured the content of senescent cells in the hippocampus of Tg^XBP1s animals during aging. Remarkably, histological analysis of aged animals indicated a significant decrease in the number of senescent cells in aged Tg^XBP1s compared with age‐matched WT littermates (Fig 3E,F). Collectively, our behavioral and morphological assessments indicate that forced expression of XBP1s in the brain over the life span of an animal prevents age‐associated cognitive decline.

XBP1s overexpression in Tg^XBP1s animals impacts age‐related changes in the hippocampal proteome, affecting proteins involved in neuronal function and neurodegenerative diseases

To further evaluate the effects of XBP1s overexpression on aging at the protein level, we performed an unbiased proteomic analysis of hippocampal tissue derived from animals at different ages during XBP1s overexpression (Fig 4A). We observed a strong positive correlation between the proteomic changes in middle‐aged and aged mice compared with young mice (r = 0.53) (Fig 4B). Notably, these age‐related alterations were largely restored by XBP1s overexpression in middle‐aged (r = −0.49) and aged mice (r = −0.46), with nearly 70% of the proteins reversing their pattern of change (Fig 4B). By contrast, changes induced in the proteome by XBP1s overexpression in early life were unrelated to the changes observed during aging (Fig 4B). Next, we identified the proteins downregulated during mouse aging and upregulated upon XBP1s overexpression, or vice versa, in middle‐aged and aged mice (Fig EV4A). The overlap between proteins regulated in opposite directions was statistically significant in all these analyses (Fig EV4A).

To investigate possible molecular pathways that may explain the protective effects of XBP1s overexpression during normal brain aging in mice, we performed a gene set enrichment analysis (GSEA), comparing significant proteomic alterations between young and aged WT animals (Fig 4A). This analysis revealed clusters of proteins involved in long‐term potentiation, calcium signaling, and metabolic control (Fig EV4B; Table 2; please refer to Dataset EV2 for detailed analysis). The comparison between middle‐aged with aged WT animals revealed additional clusters of proteins related to synaptic vesicle recycling, endocytosis, and cytoskeletal dynamics, among other processes (Fig EV4B; Table 2).

Next, we compared Tg^XBP1s animals with age‐matched littermates (Fig 4A) and found similar enriched pathways related to aging in WT animals (Figs 4C and EV4C,D; Table 3; Dataset EV2 for detailed analysis). This analysis highlighted proteins involved in neurotransmission (long‐term potentiation, neurofilaments, glutamatergic synapses, exocytosis, and myelin sheath), potentially underlying the functional improvements observed in these animals at the cognitive level. Interestingly, enriched terms were also associated with several age‐related neurodegenerative diseases (amyotrophic lateral sclerosis [ALS], prion diseases, and AD; Figs 4C and EV4C,D, Table 3).

To evaluate the potential relevance of the effect of XBP1s overexpression in humans, we identified the top‐ranking proteins that were downregulated during mouse aging and upregulated upon XBP1s overexpression, or vice versa, in middle‐aged and aged mice (Fig 4D, left), and then determined whether the expression of these proteins changed with age in the human hippocampus (Fig 4D, right). In aged mice, proteins downregulated during aging and upregulated by XBP1s overexpression (Fig 4D, dark green/blue dots) tended to be negatively correlated with human aging (manifesting decreased expression with age). Additionally, proteins upregulated during aging and downregulated by XBP1s (Fig 4D, light green/blue dots) showed a trend of positive correlation in the aged human brain. Those findings did not reach statistical significance in two other human datasets analyzed (Fig EV4E). Overall, our analysis indicated that overexpression of XBP1s in the aging brain reversed changes in expression of a cluster of proteins that partially superimposed with previously observed protein alterations in the human brain with aging.

XBP1s gene delivery by AAV into the hippocampus of aged animals reverses age‐associated cognitive decline

To assess whether aberrant proteostasis and its impact on brain function during aging can be restored, we used a gene transfer approach to deliver XBP1s into the hippocampus. For this purpose, we performed bilateral injections of AAV expressing XBP1s into the hippocampus of aged mice that already manifested cognitive decline (Figs 5A and EV5A). Remarkably, the administration of AAV‐XBP1s to middle‐aged and aged animal brains improved performance in various cognitive tests compared with age‐matched animals injected with control virus (please refer to Appendix Figs S1F and S4 for detailed workflow). Middle‐aged and aged animals injected with AAV‐XBP1s manifested the ability to discriminate novel objects or newly located objects (Fig 5B,C). Additionally, evaluation in the Barnes maze indicated that aged mice receiving AAV‐XBP1s displayed a significant decrease in latency to find the target on the second day of memory acquisition (Fig 5D), although no significant differences were found on the test day (Fig EV5B). Moreover, analysis of the CFC indicated that aged mice injected with AAV‐XBP1s into the hippocampus showed an increased freezing response following tone presentation compared to age‐matched controls, although no significant alterations were found before tone presentation (Fig 5E). This result suggests that aged animals treated with AAV‐XBP1s have improved fear conditioning memory, although contextual memory was not impacted. Taken together, our behavioral analyses indicated that XBP1s overexpression in the hippocampus was sufficient to reverse several aging‐associated cognitive alterations.

Next, to confirm that the age‐associated phenotypes observed in IRE1‐deficient animals were dependent on XBP1 expression, we treated middle‐aged IRE1^cKO mice with AAV‐XBP1s. We performed the NOR test before and after bilateral intrahippocampal injections of AAV‐XBP1s or control viral constructs (Appendix Figs S1G and S5). As expected from our results reported in Fig 2A, the middle‐aged IRE1^cKO were unable to differentiate novel objects (Fig 5F). Notably, however, treatment with AAV‐XBP1s, but not control AAV, restored their ability to discriminate between novel objects within 4 weeks of the injection (Fig 5F). These results strongly support the notion that XBP1s is a major effector of IRE1‐dependent preservation of cognitive function during aging.

XBP1s gene delivery by AAV into the hippocampus of aged mice improves electrophysiological and morphological parameters

Age‐associated cognitive impairment involves alterations in the circuitry of the hippocampus, affecting its electrical activity (Dickstein et al, 2013). Therefore, we evaluated electrophysiological parameters of hippocampal slices derived from aged animals injected with AAV‐XBP1s. Firing rates of CA1 neurons of aged animals (22–24 months old) injected with AAV‐Mock or AAV‐XBP1s were recorded during spontaneous activity or following picrotoxin (PTX) treatment to antagonize GABAergic inhibitory interneurons in order to foster excitatory activity in the hippocampal circuit. We observed a decrease in basal firing rates (Fig 6A) and bursting activity (Fig 6B) in hippocampal CA1 neurons after treating aged animals with AAV‐XBP1s. Additionally, increased firing rates and busting activity were observed in CA1 neurons following PTX treatment in XBP1s‐overexpressing animals (Fig 6A,B). Importantly, induction of long‐term potentiation (LTP), thought to represent an electrical correlate of learning and memory, was significantly improved in hippocampal slices derived from aged mice treated with AAV‐XBP1s compared with controls (Fig 6C; see fiber volley amplitudes in Fig EV5C). Correlating with these findings, we observed a significant increase in dendritic spine density in CA1 neurons compared with age‐matched control animals (Fig 6D). Finally, we measured the distribution of senescent cells in the hippocampus and found decreased accumulation of β‐galactosidase activity (Fig 6E) and p‐γ‐H2AX immunolabeling (Fig 6F) in the hippocampus following XBP1s overexpression. Taken together, these results suggest that XBP1 expression in the aged hippocampus improves behavioral, morphological, and electrophysiological parameters associated with normal brain aging.

XBP1s gene delivery modifies pathways related to synaptic function and neurodegenerative disease

Finally, we evaluated proteomic changes in the hippocampus of aged animals treated with AAV‐XBP1s. GSEA indicated that the administration of AAV‐XBP1s into the brain modified the expression of proteins involved in synaptic function, neurofilaments, and vesicle cycle dynamics, in addition to proteins associated with neurodegenerative diseases (Figs 6G,H and EV5D, Table 4; please refer to Dataset EV2, for full analysis). These findings are similar to those we obtained in Tg^XBP1s animals. We also detected important changes in proteins related to the extracellular matrix (Table 4), including collagens, laminin subunits, and heparan sulfate proteoglycan 2. Other changes included glycolysis/gluconeogenesis‐related proteins, suggesting an impact on metabolic pathways. Overall, our proteomic profiling suggests that forced expression of XBP1s influences age‐associated protein alterations related to synaptic function and are consistent with the positive effect of XBP1s administration in prolonging brain health span in terms of cognitive function.

Experimental design

This study aimed to test the role of the IRE1‐XBP1s pathway in mediating mammalian brain aging. We first identified this pathway to be correlated with dementia emergence in a human cohort population. Next, we evaluate different cohorts of transgenic mice in gain and loss‐of function mutations impacting IRE1‐XBP1s pathway in the brain using behavioral, morphological, and proteomic approaches during aging. Finally, we treated aged WT animals with adeno‐associated virus to manipulate XBP1s levels in the brain to test possible improvements in cognition and the reversal of age‐associated phenotypes.

Human study population

RNA‐seq human data were extracted from The Aging, Dementia and Traumatic Brain Injury Study (https://aging.brain‐map.org/overview/home) which is a detailed neuropathologic, molecular, and transcriptomic characterization of brains of control and TBI exposure cases from a unique aged population‐based cohort from the Adult Changes in Thought (ACT) study. Differential gene expression in RNA‐seq was performed directly by the website comparing all donors clinically diagnosed as “Non demented” (n = 49) versus “Dementia, type unknown,” “Possible Alzheimer's disease,” “Probable Alzheimer's disease” (n = 42) irrespective to sex, years of education, TBI or pathological markers of AD pathology. Only minimum of 30 M 50 bp paired‐end reads per sample were considered. A fold change cutoff > 1.2‐fold was implemented in the gene list (FDR < 0.5) to further functional enrichment analysis evaluation. Values are plotted as Fragments Per Kilobase Million (FPKM).

Animals

C57BL/6 mice were employed, maintained in a facility with 12 h light/dark cycle at 25°C with food and water provided ad libitum. Cohorts of aged, middle‐aged, and young animals were directly obtained from the Jackson's Laboratory (USA). Different colonies of transgenic mice were generated in house (all in the C57BL/6 background) and are indicated in Appendix Fig S1. All animal procedures were approved by the Bioethics Committee of the Faculty of Medicine, University of Chile (protocol number 18166‐MED‐UCH).

Stereotaxic injections

Young (3‐month‐old), middle‐aged (12 month‐old), and aged (18 month‐old) mice received bilateral stereotaxic injections of AAV (1 × 10⁹ viral genomes (VGs)/μl) in the hippocampi using the coordinates AP: −1.9, DV: +1.7, and ML: ±1.0. Male mice were deeply anesthetized using isoflurane (4%), and a stereotaxic apparatus coupled to a Hamilton microsyringe was used for the procedure. Cranium was exposed though a skin incision, and bilaterally symmetrical holes were opened using a dental drill. Injections were performed at approximately 0.5 μl/min in a total of 1 μl. Mice were returned to their home cages and kept under close monitoring until they were awake. Mice were injected with either control AAV serotype 2 vector (Mock) expressing eGFP (Addgene, #49055‐AAV2), AAV2‐CRE (Addgene, #105545‐AAV2), or AAV2‐XBP1s (produced at Genzyme by Pablo Sardi) under the control of the CMV promoter, in addition to an eGFP cassette to monitor transduction efficiency, as we previously reported (Valdés et al, 2014). For intracerebral injections with tunicamycin or vehicle, the same protocol was applied.

Behavioral assessment design

Behavioral experiments were performed in a blinded manner, both for genetically modified animals and AAV‐injected mice, using groups of age‐matched controls. Both males and females were evaluated for all tests in transgenic colonies in a balanced fashion. For AAV‐based experiments, only males were used due to logistic and experimental limitations. Injected mice underwent behavioral evaluation 3–4 weeks following injections. Assessments were designed in line with bioethical guidelines restricting the total number of animals used in the study, thus prohibiting us to use separated cohorts for each behavioral task in this study and are indicated as follows.

Behavioral tests description

Tissue collection and processing

Briefly, mice were deeply anesthetized with ketamine/xylazine and perfused with ice‐cold saline. The brain was removed from the skull and separated into two hemispheres. The hippocampus, cerebral cortex, and cerebellum of the left hemisphere were dissected out and stored frozen at −80°C until analysis. The right hemisphere was postfixed in 4% paraformaldehyde in PBS overnight at 4°C, followed by cryopreservation in 30% sucrose in PBS and freezing medium (OCT, TissueTek). Subsequently, 40 μm‐thick sagittal sections were obtained free‐floating on a Leica cryostat for immunostaining using anti‐NeuN (Millipore, MAB377, 1:100) to label neurons. Five serial sections every 200 μm were stained per animal. Fluorescence images were acquired using a confocal microscope (Nikon C²⁺).

Dendritic spine imaging

Brain slices were cut at 40‐μm thickness on a cryostat. AAV2‐GFP fluorescence was previously confirmed in injected animals to validate viral transduction using an inverted epifluorescence microscope and then imaged on a confocal microscope, Nikon Eclipse T1, at 60× magnification with additional digital zoom of 3×. Similar regions were compared among animals (CA1 region, spines in primary and secondary dendrites between the stratum radiata and the pyramidal layer, AP: −1.9 to −2.1 from the bregma). Z‐stacks were acquired in 0.5‐μm slices, laser intensity at 0.5–1% and 12.5 μs/pixel at 1024 × 1024 resolution. Z‐Stacks were then summed using ImageJ software for total maximum intensity to generate one single stacked 8‐bit image. The number of spines was manually quantified in scaled images and divided by the length of the dendrite analyzed; 5–10 dendrites per animal were used for analysis.

Immunofluorescence

Brain slices were incubated in citrate buffer at 96°C for 30 min for epitope exposition and washed in PBS. After this, slices were washed in TBS and incubated in blocking solution (3% BSA and 0.05% Triton X‐100) for 60 min at room temperature. Slices were then incubated overnight at 4°C with primary antibody anti‐phospho‐Histone H2AX Ser139 (Sigma‐Aldrich, 05‐636) diluted 1:1,000 in 3% BSA in TBS. Secondary antibody was Alexa Fluor 568 (Invitrogen A‐11031) diluted 1:2,000 in 3% BSA in TBS. The samples were washed three times using TBS, and in the final wash, DAPI was added and incubated for 5 min. Images were taken with Leica TCS SP8 confocal microscope with a 40× objective magnification. ImageJ and LAS X software were used to process the stacked images. The percentage of positive cells for p‐H2AX was graphed. Representative images are shown.

Senescence‐associated beta‐galactosidase (SA‐βgal) staining

Histochemical detection of SA‐βgal activity was performed as previously reported (Debacq‐Chainiaux et al, 2009). Briefly, slices were incubated in a 1 mg/ml of solution of 5‐bromo‐4‐chloro‐3‐indolyl β‐d‐galactopyranoside in 0.04 M citric acid/sodium, 0.005 M K₃FeCN₆, 0.005 M K₄FeCN₆, 0.15 M NaCl and 0.002 M MgCl₂, and diluted in phosphate‐buffered saline (pH 6) for 16 h. After incubation, slices were washed with TBS and mounted on superfrost microscope slides (ThermoFisher 6776214) using Fluoromount‐G (ThermoFisher, 00‐4958‐02). One slice per animal was used, and all slices were processed at the same time. Images were acquired on a Leica DM500 binocular microscope equipped with a Leica ICC50 W camera using 4× and 10× objective magnifications. ImageJ software was used to process the images. Positive area for SA‐βgal activity was measured and representative images are shown.

FluoroJade‐C staining

Brain sections were incubated in 0.06% potassium permanganate solution for 10 min. Next, sections were incubated in 0.0001% of Fluoro‐Jade C solution and washed three times with distilled water and incubated with DAPI for 5 min. Images were acquired using Olympus XI spinning disk microscope (10× magnification). ImageJ software was used to process images and evaluate integrated density of hippocampal area.

Electrophysiological measurements

Biochemical and molecular analysis

Total tissue RNA was extracted using Trizol™ Reagent (Invitrogen) according to the manufacturer's instruction. cDNA was synthesized with random primers using a High‐Capacity cDNA Reverse Transcription KIT (Applied Biosystems) and subsequently subjected to quantitative PCR analysis using HOT FIREPol® EvaGreen® qPCR Mix plus (ROX) (Solis Bio Dyne) on a Stratagene Mx3000P machine (Agilent Technologies). Actin mRNA expression was used to normalize all samples. The sequences of primers used were as follows:

Detection of S‐nitrosylated (SNO‐)IRE1 and total IRE1 levels

We performed biotin‐switch assays using whole‐brain tissue samples as previously described with minor modifications (Uehara et al, 2006). Briefly, brain tissue extracts were prepared in 400 μl HEN‐RIPA buffer (100 mM Hepes pH 7.5, 1 mM EDTA, 0.1 mM neocuproine, 150 mM NaCl, 1% NP‐40, 0.5% sodium deoxycholate, and 0.1% SDS) containing a thiol blocking reagent (20 mM methyl methanethiosulfonate [MMTS]) in a 3 ml homogenizer. After centrifuging to remove tissue debris, SDS was added to each sample to a final concentration of 1%. Samples were then incubated for 30 min at 42°C to block free thiol groups. After removing excess MMTS by acetone precipitation, S‐nitrosothiols were reduced to thiols with 10 mM ascorbate. The newly formed thiols were then linked with the sulfhydryl‐specific biotinylating reagent N‐[6‐(biotinamido)hexyl]‐3′‐(2′‐pyridyldithio)propionamide (Biotin‐HPDP). The biotinylated proteins were pulled down with Streptavidin‐agarose beads, and Western blot analysis performed to detect SNO‐IRE1 (1:1,000, Cell Signaling 3294S). To monitor the amount of “input” protein by immunoblot, a 20 μl aliquot of the sample was saved prior to the step of Streptavidin‐agarose bead addition in the biotin‐switch assay. Total IRE1 was monitored by quantitative densitometry of standard immunoblot assays. GAPDH was monitored as a control to ensure equal loading (1:1,000, Millipore MAB374).

Quantitative proteomic analysis

The hippocampi of male mice transgenic, nontransgenic, or transduced with AAV2 (Mock or XBP1s) were homogenized in TEN buffer (10 mM Tris–HCl pH 8.0, 1 mM EDTA, 100 mM NaCl, 1% NP‐40) and protein inhibitor cocktail (Roche), and then sonicated for 15 s at 30% amplitude (QSonica). For each sample lysate, 20 μg was precipitated with chloroform/methanol. Samples for mass spectrometry analysis were prepared as described (Plate et al, 2016). Air‐dried pellets were resuspended in 1% RapiGest SF (Waters) and diluted to final volume in 100 mM HEPES (pH 8.0). Proteins were reduced with 5 mM Tris(2‐carboxyethyl)phosphine hydrochloride (Thermo Fisher) for 30 min and alkylated with 10 mM iodoacetamide (Sigma‐Aldrich) for 30 min at room temperature in the dark. Proteins were then digested for 18 h at 37°C with 0.5 μg trypsin (Promega). After digestion, the peptides from each sample were reacted for 1 h with the appropriate tandem mass tag (TMT) isobaric reagent (Thermo Fisher) in 40% (v/v) anhydrous acetonitrile and quenched with 0.4% ammonium bicarbonate for 1 h. Samples with different TMT labels were pooled and acidified with 5% formic acid. Acetonitrile was evaporated on a SpeedVac and debris removed by centrifugation for 30 min at 18,000 g. MudPIT microcolumns were prepared as described (Ryno et al, 2014). Liquid chromatography–tandem mass spectrometry (LC‐MS/MS) analysis was performed using a Q‐Exactive HF mass spectrometer equipped with an Ultimate 3000 nLC 1000 (Thermo Fisher). MudPIT experiments were performed by 10 μl sequential injections of 0, 10, 20, 30, …, 100% buffer C (500 mM ammonium acetate in buffer A) and a final step of 90% buffer C/10% buffer B (100% acetonitrile, 0.1% formic acid, v/v/v), with each step followed by a gradient from buffer A (95% water, 5% acetonitrile, 0.1% formic acid) to buffer B. Electrospray was performed directly from the analytical column by applying a voltage of 2.2 kV with an inlet capillary temperature of 275°C. Data‐dependent acquisition of MS/MS spectra was performed with the following settings: Eluted peptides were scanned from 300 to 1,800 m/z with a resolution of 120,000. The top 15 peaks for each full scan were fragmented by higher energy collisional dissociation (HCD) using a normalized collision energy of 38%, isolation window of 0.7 m/z, a resolution of 45,000, AGC target 1e5, maximum IT 60 ms, and scanned from 100 to 1,800 m/z. Dynamic exclusion was set to 10 s. Peptide identification and protein quantification was performed using Proteome Discoverer 2.4 (ThermoFisher). Spectra were searched using SEQUEST against a UniProt mouse proteome database. The database was curated to remove redundant protein and splice‐isoforms, and common contaminants were added. Searches were carried out using a decoy database of reversed peptide sequences using Percolator node for filtering and the following settings: 10 ppm peptide precursor tolerance, 6 amino acid minimum peptide length, trypsin cleavage (maximum 2 missed cleavage events), static Cys modification of 57.021517 (carbamidomethylation), and static N‐terminal and Lys modification of 229.1629 (TMT‐sixplex), FDR 0.01, 2 peptide IDs per protein. Normalization of TMT reporter ion intensities was carried out based on total peptide abundance in each channel, and subsequently, TMT ratios for each identified protein were calculated in reference to a common pooled sample. Finally, the scaled, reference‐normalized TMT intensities were compared between young WT and XBP1s Tg (n = 4, 4), middle‐aged WT and XBP1s Tg (n = 3, 4) and aged WT and Tg^XBP1s (n = 3–4) or aged injected with AAV‐Mock (n = 4) or AAV‐XBP1s (n = 5) and significance was assessed using a background based t‐test with multiple testing correction in the Reporter Ions Quantifier node (Benjamini et al, 2006). Significantly altered genes were filtered by considering a cutoff of FDR < 0.1 irrespective to fold change between groups (Dalman et al, 2012). Keratins and other common contaminants were excluded from further analyses (Hodge et al, 2013). Proteins where TMT intensities were absent in more than one sample per group were also excluded. Gene set enrichment analysis was performed in EnrichR or STRING platforms using Kyoto Encyclopedia for Genes and Genomes (KEGG) and Gene Ontology annotations (Kuleshov et al, 2016; Szklarczyk et al, 2019). Heatmaps and hierarchical clustering were generated using Morpheus (https://software.broadinstitute.org/morpheus).

The correlation between the results from the differential proteomics analysis was calculated using Spearman's method on log‐fold changes. To identify proteins leading the reversal of the aging‐related changes by XBP1s, we ranked the proteins from the most downregulated to the most upregulated during aging, and from the most up‐regulated to the most downregulated by XBP1s. Then we calculated the rank product of each protein based both ranking lists. Top 50 proteins with the highest and lowest rank product were selected for the downstream comparison with human aging. To identify proteins changing with aging in humans, we used transcriptomic data from human hippocampus from Kang et al (2011), Berchtold et al (2008) and GTEX v8 (https://gtexportal.org/home). We calculated Spearman's correlation between the log‐transformed normalized expression values and the age of the samples. Mouse orthologs of the human genes were obtained using the WORMHOLE ortholog prediction web tool with default parameters (Sutphin et al, 2016).

Statistical analysis

Statistical significance was queried with Student's t‐test or Kolmogorov–Smirnov test for single comparisons and for multiple comparisons with simple or repeated measurements one‐way or two‐way ANOVA, followed by Holm–Sidak's post hoc test, depending on experimental design. P‐values were considered significant when they were < 0.05. Number of animals and P‐values are discriminated in each graph legend. P‐values computed by two‐way ANOVA indicating source of variation between variables are reported in figure legends.