Antigen-specific age-related memory CD8 T cells induce and track Alzheimer's-like neurodegeneration

Abstract

Cerebral (Aβ) plaque and (pTau) tangle deposition are hallmarks of Alzheimer's disease (AD), yet are insufficient to confer complete AD-like neurodegeneration experimentally. Factors acting upstream of Aβ/pTau in AD remain unknown, but their identification could enable earlier diagnosis and more effective treatments. T cell abnormalities are emerging AD hallmarks, and CD8 T cells were recently found to mediate neurodegeneration downstream of tangle deposition in hereditary neurodegeneration models. The precise impact of T cells downstream of Aβ/pTau, however, appears to vary depending on the animal model. Our prior work suggested that antigen-specific memory CD8 T ("^hiT") cells act upstream of Aβ/pTau after brain injury. Here, we examine whether ^hiT cells influence sporadic AD-like pathophysiology upstream of Aβ/pTau. Examining neuropathology, gene expression, and behavior in our ^hiT mouse model we show that CD8 T cells induce plaque and tangle-like deposition, modulate AD-related genes, and ultimately result in progressive neurodegeneration with both gross and fine features of sporadic human AD. T cells required Perforin to initiate this pathophysiology, and IFNγ for most gene expression changes and progression to more widespread neurodegenerative disease. Analogous antigen-specific memory CD8 T cells were significantly elevated in the brains of human AD patients, and their loss from blood corresponded to sporadic AD and related cognitive decline better than plasma pTau-217, a promising AD biomarker candidate. We identify an age-related factor acting upstream of Aβ/pTau to initiate AD-like pathophysiology, the mechanisms promoting its pathogenicity, and its relevance to human sporadic AD.

Keywords: Alzheimer’s disease; T cell; biomarker; mouse model; neuroscience.

Fig. 1.

Amyloid and Tau pathology. WBs of APP (100 kDa) and its cleavage products (APP^Cl; <100 kDa) in the brain using antibody 4G8 (upper blot; 10 wk postinjection) and ab14220 (lower blot; 3 wk postinjection) after T cell injection (→) into recipients (A). Expanded versions of these WBs are shown in SI Appendix, Fig. S2 A and B. Forebrain Aβ1-40 ELISA in B6.Foxn1 recipients of PBS of wt-CD8 T cells 15 mo postinjection (B). 4G8-positive Plaques ± pTau/curcumin staining in brains of the above mice, and in 18-mo-old Tg2576 (AD-Tg) brain (C). ThioS staining of vasculature in the wt-CD8 group cortex (D), and 4G8 staining in the periventricular region (E) 6 mo postinjection. Representative plaque morphology and size in the wt-CD8 brain 15 mo postinjection (F). Forebrain pTau and Tau PHF WBs 10 wk postinjection (G), and compiled pTau and PHF values (H). Gallyas/silver-stained cells in ^hiT mouse groups 6 mo after i.v. control/cell injection, and in 14-mo-old Tg2576 (AD-Tg) mice (I). Hippocampal sections from the indicated groups (all B6.Foxn1 recipients, except AD-Tg = Tg2576 mice), were stained with 4G8 (Aβ) and curcumin 6 mo postinjection, or at 14 mo of age for AD-Tg (J). Right-facing arrows highlight Aβ deposits with no curcumin costaining. Up-facing arrows depict colocalized Aβ and curcumin deposits. Down-facing arrows depict curcumin⁺ structures with no Aβ costaining, i.e., nonamyloid fibrillar deposits. No DAPI was used; blue background is provided for anatomical context. ThioS staining of dentate gyrus in the same ^hiT mice (PBS and wt-CD8 group mice 6 mo after control/cell injection), and in 20-mo-old AD-Tg rats (K). Individual and overlaid images of sequential 4G8 (green)/pTau (red) → Gallyas (black) stains in the wt-CD8 group hippocampus 15 mo postinjection (L). Plots depict averages ± SEM. *P < 0.05, **P < 0.01, ***P < 0.005 by the two-sided T-test, relative to the PBS group.

Fig. 2.

Amyloid and Tau pathology in ^hiT mice with T cell functional inhibition. Compiled Aβ ELISA (A), 4G8 plaque burden (B) pTau and PHF signal (pS199/202 antibody [Invitrogen] used for pTau in WBs and tissue staining; Phospho-PHF-tau pSer202+Thr205 Antibody [AT8] used for Tau PHF in WBs) (C). Gallyas silver-stained cells in PBS, wt-CD8, PrfKO-CD8, and IfnγKO-CD8 ^hiT recipients 15 mo after control/cell injection and in 18-mo-old Tg2576 (AD-Tg) brain regions (D). Gallyas⁺ neurons from the IfnγKO-CD8 group cortex (E). Representative plaque morphology and size (F and G) in IfnγKO-CD8 brain 15 mo postinjection, and comparison to plaque size in the wt-CD8 group (H). Plots depict averages ± SEM. *P < 0.05, **P < 0.01, ***P < 0.005 by the two-sided T-test, relative to the PBS group.

Fig. 3.

Neurodegeneration and cognition in nude mice harboring ^hiT cells. Cell/control recipients in all panels are B6.Foxn1 exclusively. NeuN and GFAP staining (A and B), and cell counts in the hippocampus, 15 mo after cell/control injection (C). Brain atrophy over time in PBS and wt-CD8 groups (mass normalized to PBS controls at each time point; D). Representative forebrain westerns (E), and GAPDH-normalized NeuN, drebrin, and synaptophysin (SYNP) western signals (F). Correlation of NeuN with brain weight (G). Plots depict averages ± SEM. *P < 0.05, **P < 0.01, ***P < 0.005 by the two-sided T-test, relative to the PBS group. Representative open-field test at 13 mo (H). Fear conditioning (FC) over time (I), and spontaneous alternation behavior (SAB) at 12 mo (J). Barnes maze (BM) learning/training (K), retention (L), and reversal (M and N) phases, at 14 mo (black, colored symbols = P relative to PBS, wt-CD8, respectively). Plots depict averages ± SEM. *P < 0.05, **P < 0.01, ***P < 0.005 by two-sided ANOVA (panel K) or two-sided T-test (all others), relative to the PBS group unless otherwise indicated.

Fig. 4.

Gene expression changes in the ^hiT mouse brain. RNAseq analysis on ^hiT and control forebrains. Most cell type-specific genes were down-regulated with the exception of neuron-specific genes (A). Differentially regulated genes among 84 AD-associated loci from GWAS; bolded black font depicts genes uniquely regulated in each group (B). Mechanistic model of gene, pathway, and disease induction by ^hiT cells based on combined pathological, knockout, and gene expression analysis (C). Abbreviations: CD8, cluster of differentiation antigen 8; Peptide Ag, peptide antigen; APP, Amyloid Precursor Protein; MHC I (HLA), major histocompatibility complex class I protein (Human Leukocyte Antigen); PRF1, Perforin 1 protein; IFNγ, Interferon-gamma (proinflammatory) cytokine; GZMB, Granzyme B (lytic) protein.

Fig. 5.

^hiT parameters in human Alzheimer’s. Patient cohorts: University of Antwerp = “UA”; Cedars-Sinai Medical Center = “CSMC”; University of California, Davis = “UCD”; West Los Angeles Veteran’s Administration Hospital = “WLAVA” (A). KLRG1⁺ (B) and APP_(471–479)/HLA-A2-reactive KLRG1⁺ (C) CD8 T cells in CTRL, MCI ± CSF AD biomarkers (MCI, MCI–AD), and verified Alzheimer’s (AD) blood. T cell subpopulations vs. MoCA score (D), and correlation of APP_(471–479)/HLA-A2-reactive KLRG1⁺ CD8 with Mini Mental State Exam (MMSE) score and age (with negative pHLA multimer staining subtracted; significance of both APP-specific T cells and age with MMSE score was increased without such subtraction, with P = 0.01 and 0.02, respectively) (E and F). PRF1 WB and IF representative examples of PRF1 staining in presumptive vesicles with or without (arrows) costained Aβ (G), with compiled IF and WB quantifications in age-matched CTRL and AD brains (H). Representative APP_(471-479)/HLA-A2-reactive CD8 staining (I) and compiled quantification (J) in the brain. Plots depict averages ± SEM. *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001 by the two-sided T-test, relative to CTRL unless otherwise indicated.

Fig. 6.

CD8 T_RM biomarker potential in human Alzheimer’s. Correlation of APP_(471–479)/HLA-A2-reactive KLRG1⁺ CD8 levels with Aβ1-42, total-tau, and P-tau181 in CSF, and with MMSE score in all AU patients combined (A), and correlation of parental KLRG⁺CD8⁺ T cells with these parameters in AU Alzheimer’s patients only (note: several higher APP-specific T cell levels were in control patients that were not subjected to MMSE testing) (B). Combined variance of T_RM markers, CD8A, CD44, and CD103 yielded the indicated areas under the curve (AUC) in Receiver Operating Characteristic (ROC) plots of T cell-high and T cell-low samples from a publicly available patient dataset (C). AUC was not significantly altered when CD103 variance was used alone in this cohort (0.781, T cell-high; 0.548, T cell-low). ROC plots of APP_(471–479)/HLA-A2 multimer-reactive KLRG1⁺ CD8 T cells in blood relative to normal aging controls, both from UA cohort (D). Mild Cognitive Impairment without (MCI-normal bio) and with (MCI-AD bio) CSF biomarkers consistent with AD, and confirmed AD patients ages 57 to 84 (AD-all). AD-age-matched indicates ROC analysis on 10 AD patients for whom precisely age-matched controls were available (±1 y; n = 10). P < 0.001 for all curves except MCI – normal bio (P = 0.003).