Platelet factors are induced by longevity factor klotho and enhance cognition in young and aging mice

Abstract

Platelet factors regulate wound healing and can signal from the blood to the brain^1,2. However, whether platelet factors modulate cognition, a highly valued and central manifestation of brain function, is unknown. Here we show that systemic platelet factor 4 (PF4) permeates the brain and enhances cognition. We found that, in mice, peripheral administration of klotho, a longevity and cognition-enhancing protein^3-7, increased the levels of multiple platelet factors in plasma, including PF4. A pharmacologic intervention that inhibits platelet activation blocked klotho-mediated cognitive enhancement, indicating that klotho may require platelets to enhance cognition. To directly test the effects of platelet factors on the brain, we treated mice with vehicle or systemic PF4. In young mice, PF4 enhanced synaptic plasticity and cognition. In old mice, PF4 decreased cognitive deficits and restored aging-induced increases of select factors associated with cognitive performance in the hippocampus. The effects of klotho on cognition were still present in mice lacking PF4, suggesting this platelet factor is sufficient to enhance cognition but not necessary for the effects of klotho-and that other unidentified factors probably contribute. Augmenting platelet factors, possible messengers of klotho, may enhance cognition in the young brain and decrease cognitive deficits in the aging brain.

Fig. 1. Klotho induces platelet activation in the blood and increases circulating platelet factors.

a, Paradigm for plasma proteomics profiling. Young mice (male, age 4 months) were treated with either vehicle (Veh) (n = 10 mice) or klotho (n = 9 mice) (s.c., 10 μg kg⁻¹) followed by plasma proteomics analysis. b, Percentage of alternations among arms during exploration of the Y maze. Young mice (male, age 4 months, n = 9 mice per group) were treated with either Veh or klotho. *P = 0.014 (two-tailed t-test). c, Plasma proteomics by mass spectrometry analysis 4 h after treatment with Veh or klotho identified six differentially expressed proteins (Q < 0.05, dashed horizontal line; and fold change >2, dashed vertical line). d, Canonical functions of top six differentially expressed plasma proteins following klotho treatment (Q < 0.05 and fold change >2). e, Paradigm for measuring platelet activation. Young mice (male, age 5 months, n = 8–9 mice per group) were treated with either Veh or klotho (s.c., 10 μg kg⁻¹) followed by platelet isolation from whole blood and then platelet activation analysis by FACS with markers CD61 and CD62P. f, Flow cytometry plots from FACS showing platelet populations. The upper graphs show density plots of the platelets, gated by side scatter (SSC) (for granularity) and CD61 positivity. The lower graphs show dot plots of the percentage activated (CD61 and CD62P-positive) and resting (CD61-positive only) platelets. g, Quantification of activated platelets in young mice following treatment with Veh (n = 9 mice) or klotho (n = 8 mice). *P = 0.014 (two-tailed t-test). h, Quantification of platelets counts in young mice following treatment with Veh (n = 9 mice) or klotho (n = 8 mice). Data are presented as mean ± s.e.m. Source data

Fig. 2. ASA/CPG treatment inhibits platelet activation and blocks klotho-mediated cognitive enhancement.

a, Paradigm of treatment by aspirin (ASA, 0.4 mg ml⁻¹ in drinking water) and clopidogrel (CPG, 0.15 mg ml⁻¹ in drinking water) followed by treatment with vehicle (Veh) or klotho (s.c., 10 μg kg⁻¹) in young mice (male, age 6 months, n = 4–5 mice per group). b, Flow cytometry plots from FACS showing platelet (PLT) populations in the experimental groups. Graphs show dot plots of the percentage activated (CD61 and CD62P-positive) and resting (CD61-positive only) platelets. c, Quantification of activated platelets in young mice with (ASA/CPG) or without (control, CTL) platelet inhibition following Veh or klotho treatment (n = 4 mice for CTL/Veh; n = 5 mice for CTL/klotho; n = 5 mice for ASA/CPG/Veh; n = 4 mice for ASA/CGP/klotho). Two-way ANOVA: interaction P = 0.092. *P = 0.028 (two-tailed t-test) (Benjamini–Hochberg). d, Change in activated platelets with klotho treatment (n = 5 mice for Veh; n = 4 mice for klotho). *P = 0.025 (two-tailed t-test). e, Paradigm of ASA/CPG administration and Veh or klotho treatment (s.c. 10 μg kg⁻¹, daily) followed by cognitive testing (male, age 4 months, n = 9–14 mice per group) (elevated plus maze, EPM). f, Spatial learning curves (platform hidden) of mice treated with Veh or klotho, with or without ASA/CPG, in the Morris water maze (n = 14 mice for CTL/Veh; n = 14 mice for CTL/klotho; n = 13 mice for ASA/CPG/Veh: n = 12 mice for ASA/CPG/klotho). Two-way ANOVA: interaction klotho by time ; (CTL) P = 0.006; ***P < 0.001 (two-tailed, paired t-test: days 5, 6); two-way ANOVA: klotho (CTL versus ASA/CPG) P = 0.029. g, Probe trial conducted 1 h after hidden platform training and removal of the escape platform (n = 13 mice for CTL/Veh; n = 13 mice for CTL/klotho; n = 10 mice for ASA/CPG/Veh; n = 10 mice for ASA/CPG/klotho). The latency to the target indicates memory. Two-way ANOVA: interaction P = 0.057. *P = 0.022 (two-tailed t-test). h, Spatial and working memory of young mice treated with Veh or klotho, with or without ASA/CPG, was assessed by the two-trail Y maze (n = 14 mice for CTL/Veh; n = 12 mice for CTL/klotho; n = 12 mice for ASA/CPG/Veh; and n = 9 mice for ASA/CPG/klotho). The ratio of duration spent in novel and familiar arms during testing was measured 16 h after training. Two-way ANOVA: interaction P = 0.010; *P = 0.026 (Veh versus klotho; CTL) (two-tailed t-test) (Benjamini–Hochberg); and *P = 0.018 (CTL versus ASA/CPG; klotho) (two-tailed t-test) (Benjamini–Hochberg). i, Quantification of mouse PF4 level by ELISA of plasma from young mice 4 h (male, age 4 months) following treatment with Veh (n = 10 mice) or klotho (n = 9 mice) (10 μg kg⁻¹, s.c). *P = 0.016 (two-tailed t-test). j, Paradigm of plasma collection for PF4 ELISA from young mice (male, age 4 months, n = 7–11 mice per group) with either drinking water or ASA/CPG (in drinking water) and treatment with Veh or klotho (s.c., 10 μg kg⁻¹), followed by testing in the Morris water maze, two-trial Y maze, elevated plus maze and open field. k, Quantification of mouse PF4 level by ELISA of plasma from young mice following platelet inhibition with or without treatment of Veh or klotho daily (s.c. 10 μg kg⁻¹) (n = 9 mice for CTL/Veh; n = 11 mice for CTL/klotho; n = 7 mice for ASA/CPG/Veh; n = 7 mice for ASA/CPG/klotho). Two-way ANOVA: platelet inhibition **P = 0.007; *P = 0.042 (Veh versus klotho; CTL) (one-tailed t-test since replication) (Benjamini–Hochberg); *P = 0.017 (CTL versus ASA/CPG; klotho) (two-tailed t-test)(Benjamini–Hochberg). Data are presented as mean ± s.e.m. Source data

Fig. 3. Peripherally injected PF4 is detected in the brain and enhances synaptic plasticity through NMDA receptor-dependent mechanisms.

a, Experimental paradigm of brain tissue collection following vehicle (Veh) or HIS-tagged mPF4 injection (500 μg kg⁻¹, i.p.) in young and aging mice (male, age 4 months and 18 months, n = 3–8 mice per group). b, Quantification of HIS levels in the brain and plasma of young mice, 10 min following peripheral HIS–mPF4 injection. *P = 0.030 (brain, n = 6 mice for Veh; n = 7 mice for HIS–mPF4) (two-tailed t-test); *P = 0.013 (plasma, n = 7 mice for Veh; n = 7 mice for HIS–mPF4) (two-tailed t-test). c, Quantification of HIS levels in the brain and plasma of old mice, 10 min following peripheral HIS–mPF4 injection. **P = 0.008 (brain, n = 8 mice for Veh; n = 8 mice for HIS–mPF4) (two-tailed t-test); **P = 0.001 (plasma, n = 3 mice for Veh; n = 3 mice for HIS–mPF4) (two-tailed t-test). d, Representative coronal brain tile image of young mice following peripheral HIS–mPF4 treatment. Scale bar, 1,000 μm. e, Representative hippocampal dentate gyrus images of young mice following Veh, HIS–klotho (500 μg kg⁻¹, i.p.), or HIS–mPF4 treatment (500 μg kg⁻¹, i.p.) (n = 3 mice as independent biological replicates per group). DAPI (blue, nuclei), lectin (green, blood vessels), HIS–klotho and HIS–mPF4 (red). White arrows indicate labeling for HIS–PF4. HIS–PF4 and HIS–klotho brightness and contrast were applied equally to their own respective controls. Scale bar, 10 μm. f, Quantification of mPF4 levels by PF4 ELISA in young brain following peripheral HIS–mPF4 treatment (male, age 4 months, n = 5 mice per group). **P = 0.005 (two-tailed t-test). g, Paradigm of hippocampal LTP recordings following mPF4 bath application. h, Hippocampal LTP recording of fEPSP following Veh or mPF4 bath application (male, age 3 months, n = 6 mice per group). i, Average of fEPSP over the last 10 min of recordings in young mice treated with Veh or mPF4 (n = 6 mice per group). *P = 0.043 (two-tailed t-test). j, Paradigm of peripheral injection of Veh or mPF4 (i.p., 20 μg kg⁻¹) and plasma sample collection (male, age 4 months, n = 5–9 mice per group). k, Quantification of mPF4 levels by ELISA from plasma of young mice following its peripheral injection with either Veh (n = 4 mice) or mPF4 (n = 9 mice at 5 min, n = 9 mice at 10 min and n = 5 mice at 60 min). *P = 0.03 (two-tailed, one sample t-test compared with 1). l, Experimental paradigm of hippocampal LTP recordings from young mice treated daily with Veh or mPF4 (i.p., 20 μg kg⁻¹). m, fEPSP recordings from acute hippocampal slices of young mice (male, age 3 months; n = 7 mice per group) treated with either Veh or mPF4. n, Average fEPSP slope over the last 10 min of recordings in young mice treated with Veh or mPF4 (n = 7 mice per group). *P = 0.011 (two-tailed t-test). o, Relative fEPSP recordings of acute hippocampal slices treated with Sal or Ro 25 from young mice treated with either Veh or mPF4 (male, age 3–4 months, n = 8 mice for Veh/Sal; n = 8 mice for Veh/Ro 25; n = 9 mice for mPF4/Sal; n = 9 mice for mPF4/Ro 25). p, Relative fEPSP slope over the last 10 min of recordings in mouse slices treated with Sal or Ro 25 from young mice peripherally injected with either Veh or mPF4 (n = 8 mice for Veh/Sal; n = 8 mice for Veh/Ro 25; n = 9 mice for mPF4/Sal; n = 9 mice for mPF4/Ro 25). Two-way ANOVA: Ro 25 P = 0.006; interaction PF4 by Ro 25 P = 0.051. *P = 0.035 (one-tailed t-test since replication) (Benjamini–Hochberg); **P = 0.004 (two-tailed t-test) (Benjamini–Hochberg). Data are presented as mean ± s.e.m. Source data

Fig. 4. PF4 treatment enhances cognition in young mice and aging mice.

a, Diagram of the experimental paradigm of vehicle (Veh) or mPF4 injection (i.p. 20 μg kg⁻¹, daily) followed by testing in the elevated plus maze, open field testing, Morris water maze and the two-trial Y maze in young (male, age 3–5 months) and aging (male, age 17–20 months). b, Anxiety-like behavior was measured by percentage of time spent in the open arms of the elevated plus maze during a 10 min exploration by young (n = 17 mice per group) and aging (n = 15 mice per group) mice treated with Veh or mPF4. Two-way ANOVA: age ***P < 0.001. c, Total activity was measured by movements during exploration of the open field for 10 min of young (n = 18 mice per group) and aging (n = 14 mice per group) mice treated with Veh or mPF4. Two-way ANOVA: age ***P < 0.001. d, Spatial learning curves (platform hidden) of young mice treated with Veh (n = 13 mice) or mPF4 (n = 12 mice) in the Morris water maze. Data represent the daily average of latency to find the hidden platform over two trials. Mixed-model ANOVA for hidden training: mPF4 versus Veh, **P = 0.004 (two-tailed). e, Probe trial conducted 1 h after hidden platform training and removal of the escape platform (n = 13 mice for Veh; n = 12 mice for mPF4). Percentage of time the mice spent in the target quadrant of the maze, compared with the average of the other three quadrants, is shown to indicate the memory of the platform location. The dashed line represents chance performance at 25%. Two-way ANOVA: interaction P = 0.037; **P = 0.004 (two-tailed t-test) (Benjamini–Hochberg). f, Ratio of time young mice spent in the target quadrant relative to other quadrants of the maze (n = 13 mice for Veh; n = 12 mice for mPF4). *P = 0.024 (two-tailed t-test). g, Spatial learning curves (platform hidden) of aging mice treated with Veh or mPF4 in the water maze (n = 12 mice per group). Data represent the daily average of latency to find the hidden platform over four trials. Mixed-model ANOVA for hidden training: mPF4 versus Veh, *P = 0.048 (two-tailed). h, Probe trial conducted 1 h after hidden platform training and removal of the escape platform in aging mice (n = 11 mice for Veh; n = 12 mice for mPF4). Percentage of time mice spent in the target quadrant, compared with the average of the other three quadrants, is shown to indicate memory of the platform location. The dashed line represents chance performance at 25%. Two-way ANOVA: interaction P = 0.038; **P = 0.006 (two-tailed t-test) (Benjamini–Hochberg). i, Ratio of time aging mice spent in the target quadrant relative to other quadrants of the maze (n = 12 mice per group). **P = 0.009 (two-tailed t-test). j,k, Spatial and working memory of young and aging mice treated with Veh or mPF4 was assessed by the two-trial Y maze: ratio of distance traveled (n = 12 mice for young/Veh; n = 11 mice for young/mPF4; n = 12 mice for aging/Veh; n = 12 mice for aging/mPF4) (j) and duration in novel and familiar arms during testing was measured 16 h after training (n = 13 mice for young/Veh; n = 12 mice for young/mPF4; n = 12 mice for aging/Veh; n = 12 mice for aging/mPF4) (k). Two-way ANOVA: age ***P < 0.001; *P = 0.012 (two-tailed t-test) (Benjamini–Hochberg) (j), two-way ANOVA: age ***P < 0.001; *P = 0.012 (two-tailed t-test) (Benjamini–Hochberg) (k). l, Paradigm of Veh or mPF4 injection (i.p. 20 μg kg⁻¹, daily for 7 days) in aging mice, followed by testing in the two-trail Y maze every 2 weeks. m, Spatial and working memory of aging mice treated with Veh or mPF4 for 7 days was assessed by the two-trial Y maze (male, age 22 months, n = 15 mice for Veh; n = 14 mice for mPF4). The ratio of the percentage of duration in the novel to the familiar arm during testing was measured 16 h after training. Two-way repeated measures ANOVA: interaction PF4 by time, P = 0.001; *P = 0.042 (day 7, one-tailed t-test since replication) (Benjamini–Hochberg); *P = 0.037 (day 21, two-tailed t-test) (Benjamini–Hochberg). n, Paradigm of Veh or klotho injection in young PF4 KO mice (males and females, age 5–7 months, n = 11–12 mice per group) followed by testing in water maze. o, Probe trial conducted 24 h after hidden platform training and removal of the escape platform (n = 11 mice for wild type (WT)/Veh; n = 11 mice for WT/klotho; n = 12 mice for PF4KO/Veh; n = 11 mice for PF4KO/klotho). The percentage time of mice spent in the target quadrant of the maze is shown to indicate their memory. Two-way ANOVA: klotho P < 0.001; *P = 0.050 (two-tailed t-test) (Benjamini–Hochberg); **P = 0.001 (two-tailed t-test) (Benjamini–Hochberg). Data are presented as mean ± s.e.m. Source data

Extended Data Fig. 1. Representative FACS gating strategy for identifying platelet populations from a young mouse (linked to Fig. 1f).

(a) Isolated platelets unstimulated and stained CD61 (platelet marker) and CD61P (platelet activation marker). (b) Resting platelet population (PLT + , green) shows a shift of population to the CD61-positive (PLT + , green) compared to unstained platelets (a). The activated platelet population (Activated PLT, cyan) is set to around 1% by adjusting CD62P fluorescence intensity. Isolated platelets stained with CD61 and CD62P but not stimulated with any platelet agonists. (c,d) ADP treatment or (e,f) Thrombin treatment results in a shift of the platelet-positive population (PLT + , green) to the CD62P-positive (platelet activation positive, Activated PLT, cyan) platelets in a dose-dependent manner. Isolated platelets were stimulated with two different doses of platelet agonists, ADP or Thrombin, prior to co-labeling with CD61 and CD62P.

Extended Data Fig. 2. Acute klotho treatment does not alter the FGF23-klotho endocrine system, renal phosphate, or vitamin D (linked to Fig. 1f,g).

(a) Paradigm for serum and urine profiling. Young mice (male, age 4 months; n = 4-5 mice per group) were treated with either Veh or Klotho (s.c. 10 μg/kg) followed by serum and urine collection and profiling. (b) Urine ratio of phosphorus to creatinine in urine between groups (n = 4 mice for Veh group; n = 5 mice for Klotho group). (c) Serum Phosphorus levels between groups (n = 5 mice per group). (d) Serum 25-Hydroxyvitamin D levels between groups (n = 5 mice per group). (e) Serum FGF23 levels between groups (n = 5 mice per group). Data are presented as means ± SEM. Source data

Extended Data Fig. 3. Klotho modestly but directly activates platelets in a potentially ADP-dependent pathway (linked to Fig. 1f,g).

(a) Experimental paradigm of measurement of platelet activation with in vitro klotho treatment. Platelets were isolated from young mice (male, age 6 months, n = 5-12 mice per group) and were treated with klotho. Platelet activation levels were measured by FACS with markers CD61 and CD62P. (b) Representative flow cytometry plots from FACS showing platelet populations in the experimental groups. Graphs show dot plots of the percentage activated (CD61 and CD62P-positive) and resting (CD61-positive only) platelets. (c) Quantification of activated platelets following in vitro klotho treatment of isolated platelets (n = 12 mice for Resting; n = 5 mice for 12.5pM Klotho; n = 12 mice for 25pM Klotho; n = 12 mice for 50pM Klotho; n = 5 mice for 1uM ADP; n = 5 mice for 1uM ADP +50pM Klotho). *P = 0.041 (Resting vs 25pM of Klotho), *P = 0.021 (Resting vs 50pM of Klotho), ***P < 0.001 (Resting vs ADP) (two-tailed t-tests)(Benjamini & Hochberg). Data are presented as means ± SEM. Source data

Extended Data Fig. 4. ASA/CPG administration with or without Klotho treatment does not alter time spent in target quadrant (linked to Fig. 2f).

Probe trial conducted 1 hr after hidden platform training and removal of the escape platform (n = 13 mice for CTL/Veh; n = 14 mice for CTL/Klotho; n = 11 mice for ASA/CPG/Veh; n = 11 mice for ASA/CPG/Klotho). Percentage of time mice spent in the target quadrant of the maze is shown to indicate the memory of the platform location. The dashed line represents chance performance at 25%. Data are presented as means ± SEM. Source data

Extended Data Fig. 5. ASA/CPG administration with or without Klotho treatment does not alter swim speed, ability to find the target platform, anxiety-like behavior, or total activity (linked to Fig. 2e–h).

(a) Swim speed of mice was measured in the water maze. Mice were treated with Veh or Klotho, with or without platelet inhibition (n = 14 mice for CTL/Veh; n = 13 mice for CTL/Klotho; n = 11 mice for ASA/CPG/Veh; n = 10 mice for ASA/CPG/Klotho). (b) Ability to find the target platform, identified by a visual cue was measured by the time spent in the water maze and averaged over two trials. Mice were treated with Veh or Klotho, with or without platelet inhibition (n = 13 mice for CTL/Veh; n = 13 mice for CTL/Klotho; n = 10 mice for ASA/CPG/Veh; n = 10 mice for ASA/CPG/Klotho). (c) Anxiety-like behavior of mice was measured by the percentage of time spent in the open arms of the elevated plus maze during a 10 min exploration period of mice (n = 10-14 mice per group) treated with Veh or Klotho, with or without platelet inhibition (n = 14 mice for CTL/Veh; n = 14 mice for CTL/Klotho; n = 12 mice for ASA/CPG/Veh; n = 10 mice for ASA/CPG/Klotho). (d) Hyperactivity was measured by the total movements during exploration of the open field for 10 min of mice treated with Veh or Klotho, with or without platelet inhibition (n = 13 mice for CTL/Veh; n = 14 mice for CTL/Klotho; n = 10 mice for ASA/CPG/Veh; and n = 10 mice for ASA/CPG/Klotho). Data are presented as means ± SEM. Source data

Extended Data Fig. 6. PF4 levels correlate with platelet activation (linked to Fig. 2k).

Correlation between released PF4 levels and platelet activation following 0.1 mM and 10 mM ADP treatment in vitro onto isolated platelets. (male, age 4 months, n = 4 mice) *P = 0.013 by Linear regression. Source data

Extended Data Fig. 7. Additional mice that did not undergo watermaze were added to two-trial Y maze data shown in Fig. 4j,k.

(a) and (b) With the addition of mice (watermaze naïve) to the two-trial Y maze, (final n = 14-18 mice per group), an mPF4 effect on increasing spatial and working memory in young mice became observable, in addition to its clear effect in aging mice (a) n = 18 mice for Young/Veh; n = 18 mice for Young/mPF4; n = 15 mice for Aging/Veh; n = 16 mice for Aging/mPF4. (b) n = 17 mice for Young/Veh; n = 18 mice for Young/mPF4; n = 14 mice for Aging/Veh; n = 16 mice for Aging/mPF4. Two-way ANOVA: age ***P < 0.001 (a,b); **P = 0.002 (two-tailed t-tests)(Benjamini & Hochberg)(young and aging); *P = 0.017 (two-tailed t-tests)(Benjamini & Hochberg)(young); *P = 0.020 (two-tailed t-tests)(Benjamini & Hochberg)(aging). Data are shown as means ± SEM. Source data

Extended Data Fig. 8. hPF4 treatment enhances cognition in mice (linked to Fig. 4).

(a) Experimental paradigm of human PF4 treatment. Young mice (male, age 4 months, n = 12-14 mice per group) were treated with either Veh, hPF4 HEK293 (derived from a cell line), or hPF4 hPlatelet (derived from platelets) (20 μg/kg, i.p., daily) followed by testing in the watermaze. (b) Latency to target platform of mice (n = 12-13 mice per group) treated with Veh, hPF4 HEK23, or hPF4 hPlatelet after hidden training and removal of the escape platform. Mixed model ANOVA: *P = 0.028 (hPF4 HEK293 vs Veh)(Benjamini & Hochberg) and *P = 0.036 (hPF4 hPlatelet vs Veh)(Benjamini & Hochberg) (two-tailed) (c) Percentage of time the mice spent in the target quadrant 1 day after hidden training and removal of the escape platform (n = 14 mice for Veh; n = 12 mice for hPF4 HEK293; n = 13 mice for hPF4 hPlatelet). hPF4 HEK293 (Target vs Other) ***P < 0.001 (two-tailed t-test)(Benjamini & Hochberg); hPF4 hPlatelet ***P < 0.001 (two-tailed t-test) (Benjamini & Hochberg). (d) Ratio of time the mice spent in the target quadrant relative to other quadrants of the maze (n = 14 mice for Veh; n = 12 mice for hPF4 HEK293; n = 13 mice for hPF4 hPlatelet). One-Way ANOVA: P < 0.001, ***P < 0.001 (Dunnett’s test). Data are presented as means ± SEM. Source data

Extended Data Fig. 9. PF4 treatment restores the age-induced increase of cognition-associated factors in the aging hippocampus.

(a) PCA analysis of experimental groups for bulk RNA sequencing shows PF4-mediated alterations in aging, but not young, hippocampal differential gene expression (DEG). Analyzed hippocampus are of young (male, 3-5 months) and aging mice (male, 17-20 months) (n = 10-13 mice per group) which were treated with Veh or mPF4 injection (i.p. 20 μg/kg, daily) and underwent behavioral experiments. (b) Heatmap representing age-induced alterations in the 670 DEGs. Each row represents a gene, and each column represents a biological sample. (c) Volcano plot showing age-induced alterations in DEGs in the hippocampus. Significant genes are labeled as increased (red) or decreased (blue). Top 10 significant genes are labeled with their respective names. Adjusted P value obtained by Wald test in DESeq2. (d) Heatmap representing PF4-induced alterations in the 24 DEGs in the aging hippocampus. Each row represents a gene, and each column represents a biological sample. (e) Volcano plot showing PF4-induced alterations in the aging hippocampus. Significant genes are labeled as increased (red) or decreased (blue). Top 10 significant genes are labeled with their respective names. Adjusted P value obtained by Wald test in DESeq2. (f) Functional prediction of PF4 effects on DEG of aging hippocampus by pathway analysis predicted learning. (g) Diagram mapping cognitive performance to hippocampal gene expression to generate a molecular signature of cognition. (h) Heat map of genes that comprise the molecular signature of cognition in aging hippocampus, analyzed by age, treatment, memory score, and RNA signature score. (i) The composite RNA score, representing the cognitive molecular signature in each mouse, correlated closely with memory scores. Correlation plot with 95% confidence interval (gray band) Two-sided Pearson’s R = 0.66, ***P < 0.001. (j) Aging decreased the composite RNA score represented in box and whisker plots. mPF4 treatment reversed the effect of aging on the RNA score, or the cognitive molecular signature. **P = 0.004 (Young Veh vs Aging Veh), ***P < 0.001 (Young Veh vs Aging Veh), Wilcoxon rank-sum two-sided test (Benjamini & Hochberg). (k) Overlap between genes significantly altered by aging, PF4, and also associated with cognition. (l-m) PF4 treatment restores the age-induced increase of cognition-associated genes measured by mRNA expression (RNA-seq) (male, age 3-5 months and 17-20 months, n = 10-13 mice per group). Box plots show the median and the 25–75th percentiles, and the whiskers indicate values up to 1.5-times the interquartile range. (l) Quantification of AKAP11 by RNA sequencing shows aging increased expression and PF4 treatment decreased age-induced increase (n = 13 mice for Young/Veh; n = 12 mice for Young/mPF4; n = 12 mice for Aging/Veh: n = 10 mice for Aging/mPF4). Two-way ANOVA: age, treatment, and interaction P < 0.001; ***P < 0.001, two-tailed t-tests (Benjamini & Hochberg). (m) Quantification of MeCP2 by RNA sequencing shows aging increased expression and PF4 treatment decreased age-induced increase (n = 13 mice for Young/Veh; n = 12 mice for Young/mPF4; n = 12 mice for Aging/Veh; n = 10 mice for Aging/mPF4). Two-way ANOVA: age, treatment and interaction P < 0.001. ***P < 0.001, two-tailed t-tests (Benjamini & Hochberg). (n) Quantification of Arhgef9 by RNA sequencing shows aging increased expression and PF4 treatment decreased age-induced increase (n = 13 mice for Young/Veh; n = 12 mice for Young/mPF4; n = 12 mice for Aging/Veh; n = 10 mice for Aging/mPF4). Two-way ANOVA: treatment P < 0.01; age and interaction P = 0.001. ***P < 0.001, two-tailed t-tests (Benjamini & Hochberg). Source data