Cellular Senescence Is a Central Driver of Cognitive Disparities in Aging

Abstract

Cognitive function in aging is heterogeneous: while some older individuals develop significant impairments and dementia, others remain resilient and retain cognitive function throughout their lifespan. The molecular mechanisms that underlie these divergent cognitive trajectories, however, remain largely unresolved. Here, we utilized a high-resolution home-cage-based cognitive testing paradigm to delineate mechanisms that contribute to age-related cognitive heterogeneity. We cognitively stratified aged C57Bl/6N male mice by cognitive performance into intact (resilient) or impaired subgroups based on young performance benchmarks. Cognitively impaired males exhibited marked reactive gliosis in the hippocampus, characterized by microglial activation, increased astrocyte arborization, and elevated transcriptional expression of reactivity markers. These changes were accompanied by increased markers of cellular senescence and the associated senescence-associated secretory phenotype (SASP) in impaired animals, including p16^INK4a, SASP factors (e.g., Il-6, Il-1b, Mmp3), and SA-β-gal staining in the hippocampus. Notably, clearance of senescent cells using senolytic agents dasatinib and quercetin ameliorated the heterogeneity in cognitive performance observed with age and attenuated impairment-associated gliosis, senescence markers, and mitochondrial dysfunction. Aged female mice could not be stratified into subgroups yet showed increased neuroinflammation with age that was not resolved with senolytics. Collectively, our findings implicate cellular senescence as a central driver of sex-specific neuroinflammation that drives divergent cognitive trajectories in aging. Thus, we demonstrate that senolytic treatment is an effective therapeutic strategy to mitigate cognitive impairment by reducing neuroinflammation and associated metabolic disturbances.

Keywords: cellular senescence; cognitive heterogeneity; dementia; neuroinflammation; reactive gliosis; senolytic.

FIGURE 1

High‐resolution performance‐based spatial working memory paradigm facilitates cognitive stratification of aged male mice. (A) Illustration of the PhenoTyper with CognitionWall used to assess cognitive performance over a 90‐h period. Initial discrimination learning occurs over the first 2 days of testing (hours 1–50), where the mouse learns to enter the left‐most entry of the cognition wall five times to receive a food pellet. On days 3–4 (hours 51–90), the mouse undergoes a reversal learning task, relearning to enter the right‐most entry to receive a food pellet. Behavior is assessed during both dark and light periods using an infrared camera, with movements tracked using Noldus Ethovision software. (B) Entries needed to reach the 80% criterion (based off the success rate of the trailing 30 entries) in the initial learning and reversal learning tasks of young (6 months; n = 14) and aged (22–24 months; n = 35) mice. (C) Circadian activity assessed via the onset of activity during the dark phase was comparable among young and cognitively stratified aged groups. (D) Total distance moved during the dark phases of the day/light cycle was unchanged between groups. (E) Cumulative learning index depicts separations between aged cognitively intact and impaired animals during the reversal phase. (F) Cognitive flexibility was assessed during the first 10 h of the reversal phase and was significantly reduced in the aged cognitively impaired group. (G) Maximal learning of aged cognitively impaired mice was significantly decreased at the end of the reversal phase. For graphs C–N, colors represent the following: Young (black), aged intact (blue), aged impaired (red). Shaded bars in graphs C and E represent the dark periods of the L:D cycle. Error bars: Mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

FIGURE 2

Cognitively impaired aged male mice show increased markers of reactive gliosis within the hippocampus. (A) Representative confocal images of immunostained hippocampal sections within the CA1 region labeled for GFAP (green; astrocytes), Iba1 (red; microglia), and DAPI (blue; nuclei) from cognitively stratified mice. White arrows specify the cells depicted in each insert. Scale bar, 50 μm. (B) Proportion of Iba1+ cell number normalized to total cell number (determined via DAPI⁺ nuclei count) in the CA1 region of the hippocampus. 3–4 sections were analyzed per n. (C) Transcriptional expression of the microglial marker Aif1 in the hippocampus. n = 5 mice/group. (D) Transcriptional expression of the astrocyte cytoskeletal element Gfap and astrocyte reactivity markers (S1pr3, Serpina3n, H2‐D1, Timp1) within the hippocampus. n = 5 mice/group. (E) Quantification of the number of GFAP⁺ astrocyte projections plotted against the radial distance from the soma obtained via Sholl analysis in the CA1 region of the hippocampus. (F, G) Violin plots depicting maximal branch length and maximal number of intersections of GFAP+ astrocytes assessed by Sholl analysis. For graphs B–G, colors represent the following: Young (black), aged intact (blue), aged impaired (red). Error bars: Mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

FIGURE 3

Cognitively impaired aged male mice display increased signatures of cellular senescence within the hippocampus. (A–H) Transcriptional expression of cell cycle arrest markers (p16 ^INK4a and p21 ^WAF1 ) and senescence‐associated markers, including secreted cytokines (Il‐6, Il‐1b, Tnf‐ɑ), nuclear lamin component (Lmnb1), and matrix metalloproteases (Mmp3, Mmp12) within the hippocampus. n = 5–9 animals/group. (I) Quantification of senescence‐associated beta‐galactosidase (SA‐β‐gal) staining within the hippocampus (representative images, panel J; quantification from 3 to 4 sections per animal, n = 5 animals/group). (J) Representative sagittal images of senescence‐associated beta‐galactosidase (SA‐β‐gal) and p16^INK4a immunolabeling co‐localizing with microglial and astrocyte markers Iba1 and GFAP within the dentate gyrus of the hippocampus. Scale bars, 200 μm (SA‐β‐gal), 10 μm (p16INK4a). DG denotes the granule cell layer of the dentate gyrus. For graphs A–I, colors represent the following: Young (black), aged intact (blue), aged impaired (red). Error bars: Mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 4

Senolytic treatment ameliorates cognitive heterogeneity and hippocampal senescence markers in aged male mice. (A) Entries needed to reach 80% criterion in the initial and reversal learning phases of young, vehicle‐ and senolytic‐treated male mice. (B) Cumulative learning index depicts separations between vehicle‐treated cognitively impaired mice and vehicle‐treated cognitively intact mice and senolytic‐treated mice during the reversal phase. (C) Maximal learning of aged cognitively impaired mice at the end of the reversal phase was significantly decreased compared to vehicle‐treated cognitively intact and senolytic‐treated mice. (D–G) Transcriptional expression of cell cycle arrest markers (p16 ^INK4a , p21 ^WAF1 ) and SASP factors (IL‐6, IL‐1b) within the hippocampus of vehicle‐treated cognitively stratified and senolytic‐treated aged male mice. (H, I) Quantification and representative images of senescence‐associated beta‐galactosidase staining within the hippocampus. DG denotes the granule cell layer of the dentate gyrus. Scale bar, 200 μm; 3–4 sections per animal, n = 5 mice/group. (J–L) Volcano plots depicting the fold change of cytokines and chemokines assessed via the RT² profiler between (J) vehicle‐treated intact mice and young, (K) vehicle‐treated impaired and intact mice, and (L) senolytic‐treated and vehicle‐treated impaired aged mice. Points above the dotted line denote significantly different genes (p < 0.05). For all histograms, colors represent the following: Young (black), vehicle‐treated intact (blue), vehicle‐treated impaired (red), and senolytic‐treated (pink). Shaded bars in graphs B represent the dark periods of the L:D cycle. Error bars: Mean ± SEM. *p < 0.05, ***p < 0.001, ****p < 0.0001.

FIGURE 5

Reactive gliosis and deficits in mitochondrial respiration in the hippocampus associated with cognitive impairment are ameliorated by senolytic therapy. (A) Representative confocal images of immunostained hippocampal sections within the CA1 region labeled for GFAP (green; astrocytes), Iba1 (red; microglia), and DAPI (blue; nuclei) from young, vehicle‐treated, cognitively stratified, and senolytic‐treated male mice. White arrows specify the cells depicted in each insert. n = 5 animals/group. Scale bar, 50 μm. (B) Proportion of Iba1+ cell number normalized to total cell number (determined via DAPI⁺ nuclei count) in the CA1 region of the hippocampus. 3–4 sections were analyzed per n. (C) Transcriptional expression of the microglial marker Aif1 of the hippocampus of young, vehicle‐ and senolytic‐treated aged male mice. n = 6 mice/group. (D) Expression heatmap of reactive astrocyte‐associated genes (expressed as fold change over young), including Gfap, Serpina3n, S1pr3, H2‐D1, and Lcn2. * and # denote significance between intact/impaired and impaired/senolytic‐treated groups, respectively. n = 6 animals/group. (E) Quantification of the number of astrocyte projections plotted against the radial distance from the soma obtained via Sholl analysis. (F, G) Maximal branch length and maximal number of intersections of GFAP+ astrocytes were quantified using Sholl analysis. (H) Oxygen consumption rate in permeabilized hippocampi following sequential addition of the substrates glutamate/malate (GM), ADP (2.5 mM), succinate (Suc), rotenone (Rot), and ascorbate/TMPD (Asc/TMPD). n = 7–11 animals/group. For graphs B, C and E–H, colors represent the following: Young (black), vehicle‐treated aged intact (blue), vehicle‐treated aged impaired (red), senolytic‐treated (pink). Error bars: Mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.