Chimeric antigen receptors discriminate between tau and distinct amyloid-beta species

Abstract

Background: The lack of a definitive cure for Alzheimer's disease (AD) is fueling the search for innovative therapeutic strategies. Having revolutionized cancer immunotherapy, immune cell engineering with chimeric antigen receptors (CAR) is being explored to target AD. Whether CARs can recognize distinct amyloid-β (Aβ) species and tau neurofibrillary tangles (NFTs)-hallmark pathologies of AD-remains unclear.

Methods: To investigate this, we engineered a series of CARs using single-chain fragment variable (scFv) derived from the variable light and heavy chains of antibodies tested in AD clinical trials. These included E2814 (E2814-CAR), targeting tau; Lecanemab (Lec-CAR) and Aducanumab (Adu-CAR), targeting Aβ; and Donanemab (Don-CAR) and Remternetug (Rem-CAR), targeting the truncated pyroglutamated Aβ species Aβp3-42. To evaluate CAR function, we utilized the murine DO11.10 CD4⁺ T-cell hybridoma line as a scalable and reproducible platform. CAR activation was assessed in response to tau preformed fibrils (PFFs), Aβ_1-42 oligomer-enriched aggregates, and Aβp3-42 aggregates, using flow cytometry for CD69 expression and ELISA for IL-2 secretion. To validate this platform, we tested Adu-CAR in primary mouse CD4⁺ T cells treated with Aβ_1-42 aggregates and assessed activation via flow cytometry for CD69 and CD25 expression.

Results: DO11.10 cells expressing E2814-CAR-but not Lec-CAR-responded to tau PFFs. In contrast, cells expressing Adu-CAR, and to a lesser extent Lec-CAR-but not E2814-CAR-responded to Aβ_1-42 aggregates. For Aβp3-42 aggregates, Rem-CAR elicited the strongest response, followed by Adu-CAR, while E2814-CAR and Don-CAR showed no activation. The activation of Adu-CAR by Aβ_1-42 aggregates was recapitulated in primary CD4⁺ T cells, as measured by CD69 expression.

Conclusions: Our findings demonstrate that CARs can detect and discriminate between tau PFFs, Aβ1-42, and Aβp3-42 aggregates. This highlights the potential of repurposing AD antibodies for CAR-based therapies to selectively target tau NFTs and distinct forms of Aβ senile plaques.

Keywords: Aducanumab; Alzheimer’s disease; Amyloid beta; Chimeric antigen receptor; Donanemab; E2814; Lecanemab; Pyroglutamated amyloid beta; Remternetug; Tau.

Fig. 1

Engineering and evaluation of CAR constructs based on Alzheimer’s disease-targeting antibodies. a Detailed schematic of the CAR constructs targeting tau PFFs, Aβ_1-42, and Aβp3-42. Each construct features a human CD8α signal peptide (green) followed by an antigen-binding domain consisting of an scFv with the VLC (purple) followed by the VHC (purple) from antibodies E2814 (E2814-CAR), Lecanemab (Lec-CAR), Aducanumab (Adu-CAR), Donanemab (Don-CAR), or Remternetug (Rem-CAR), linked via a G4S [3] flexible linker. The scFv is fused to a murine CD28-based hinge (grey), transmembrane (TM) domain (blue), costimulatory domain (orange), and a mouse CD3ζ signaling domain (teal). Component sequences for each CAR construct are detailed in Table 1a–f. b Overview of the multicistronic retroviral vector system used for the experiments containing an MSCV backbone, the CAR construct, a murine Foxp3 (mFoxp3), and EGFP reporter gene, separated by T2A and P2A sequences. An assessment of the Foxp3 is not incorporated in the current study (see Materials and Methods for additional information) c Representative flow cytometry plots of EGFP and viability dye staining (Live/Dead Fixable Violet) used to sort transduced cell clones (Q3) via FACS. d Subsequent FACS-based selection strategy after expansion using a standardized EGFP signal intensity range (black region) to ensure homogeneous CAR-expression in DO11.10 cell clones used for the experiments

Fig. 2

Tau PFFs selectively activate E2814-CAR-expressing cell clones without inducing cell death. a Schematic overview of the experiments. After tau PFFs treatments for 45 h, the media was collected for analysis of IL-2 by ELISA and the cells were harvested for analysis of EGFP expression, viability, and CD69 expression by flow cytometry. b–d Total number of live cells remaining after tau PFFs treatment for (b) untransduced, (c) E2814-CAR, and (d) Lec-CAR clones. e–g Percentage of CD69-positive (CD69⁺) live cells for (e) untransduced, (f) E2814-CAR, and (g) Lec-CAR DO11.10 cell clones. Only relevant statistical p-values are shown; full statistical analyses available in Tables 2a-c & Tables 3a-c. Statistical comparisons: b Dunnett, c, d Dunn’s, and e–g Tukey multiple comparisons. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Graphs display mean ± S.E.M. Three biological replicates

Fig. 3

E2814-CAR-expressing cell clones secrete IL-2 in response to tau PFF treatment. a Bar chart of IL-2 concentrations measured in the media of tau PFF-treated CAR clones. IL-2 levels in untransduced cells at all doses and in E2814-CAR at 3.125 and 6.25 nM are below the assay’s 3.1 pg/mL detection threshold. Treatments with IL-2 levels that remain below VEH (0 nM) are below the black dashed arrows marking mean VEH levels. E2814-CAR IL-2 levels that remain below 0 nM are 3.125–100 nM tau PFFs, while Lec-CAR-expressing cells show no IL-2 increase at any concentration. b Statistical analysis comparing IL-2 production at 0 nM (VEH) and 200 nM tau PFFs in E2814-CAR-expressing cells shows a significant increase. Statistical test: independent two-tailed t-test. * p < 0.05. Graphs display mean ± S.E.M. Three biological replicates

Fig. 4

Cytotoxic Aβ_1-42 treatments selectively activate Adu-CAR and Lec-CAR-expressing cell clones. a Schematic overview of the experiments. After Aβ_1-42 treatments for 48 h, the media was collected for analysis of IL-2 by ELISA and the cells were harvested for analysis of EGFP expression, viability, and CD69 expression by flow cytometry. b–e Total number of live cells remaining after Aβ_1-42 treatment for b untransduced, c E2814-CAR, d Lec-CAR, and e Adu-CAR clones. f–i Percentage of CD69-positive (CD69⁺) live cells analyzed for f untransduced, g E2814-CAR, h Lec-CAR, and i Adu-CAR clones. Statistical comparisons: b, g Dunn’s, c–f Games-Howell, and h, i Tukey multiple comparisons. Only relevant statistical p-values are shown; full statistical analyses available in Tables 5a, c-e & Tables 6a-d. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Graphs display mean ± S.E.M. Three biological replicates

Fig. 5

IL-2 secretion by Adu-CAR clones shows the strongest response to Aβ_1-42 treatments. a Bar chart of IL-2 concentrations measured in the media of Aβ_1-42-treated CAR clones. IL-2 levels in untransduced cells at all doses remain below the assay’s 3.1 pg/mL detection threshold. Treatments with IL-2 levels that remain below VEH (0 nM) are below the black dashed arrows marking mean VEH levels. For E2814-CAR-expressing cells, IL-2 levels remain below 0 μM at 0.1–5 μM Aβ_1-42. For Lec-CAR and Adu-CAR, IL-2 levels remain below VEH at 0.1 and 1 μM Aβ_1-42. b–d Statistical analyses of IL-2 production in response to 0–20 μM Aβ_1-42 for b E2814-CAR, c Lec-CAR, and d Adu-CAR. e Signal-to-noise ratio analysis comparing E2814-CAR, Lec-CAR, and Adu-CAR at each Aβ_1-42 concentration. The black dotted arrow indicates VEH levels. Statistical comparisons: b Games-Howell, c, d Tukey, e and Bonferroni multiple comparisons. Only relevant statistical p-values are shown; full statistical analyses available in Tables 7a-d. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Graphs display mean ± S.E.M. Three biological replicates

Fig. 6

Aβp3–42 treatments increase Rem-CAR, Adu-CAR and Lec-CAR-expressing CD69 positive populations. a Schematic overview of the experiments. After Aβp3–42 treatments for 48 h, the media was collected for analysis of IL-2 by ELISA and the cells for analysis of EGFP expression, viability, and CD69 expression by flow cytometry. b–g Total number of live cells remaining after Aβp3–42 treatments at indicated concentrations for b untransduced, c E2814-CAR, d Rem-CAR, e Don-CAR, f Lec-CAR, and g Adu-CAR clones. h–m Percentage of CD69-positive (CD69⁺) live cells across Aβp3–42 treatments for h untransduced, i E2814-CAR, j Rem-CAR, k Don-CAR, l Lec-CAR, and m Adu-CAR clones. m Adu-CAR clones are missing the 5 μM treatment condition because it violated the assumption of normality. Non-parametric test available in Table 9 g. Statistical comparisons: b, d, e, f Dunnett, g Dunn, c, h Games-Howell, and i, j, k, l, m Tukey multiple comparisons. h, i, k Only relevant statistical p-values are shown; full statistical analyses are available in Tables 8, 9. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Graphs display mean ± S.E.M. Three biological replicates

Fig. 7

IL-2 secretion by Rem-CAR and Adu-CAR clones ratifies response to Aβp3–42. a IL-2 concentrations measured in the culture media of CAR-expressing DO11.10 cells treated with 1–20 μM Aβp3–42 in 0–900 (left) and 0–12 (right) pg/ml ranges. IL-2 levels in untransduced cells (Untrnsd) remained below the assay’s detection threshold of 3.1 pg/mL across all conditions (black dashed arrow, 1–12 pg/mL range). Similarly, E2814-CAR-expressing cells showed IL-2 levels below VEH for all Aβp3–42 treatments. Untransduced cells and E2814-CAR clones were not further analyzed. Rem-CAR, Don-CAR, Lec-CAR, and Adu-CAR clones exhibited IL-2 levels above the detection threshold and were further analyzed. b–e Statistical analysis of IL-2 secretion in response to Aβp3–42 treatment for individual CAR constructs: b Rem-CAR, c Don-CAR, d Lec-CAR, and e Adu-CAR. f Signal-to-noise ratio comparison across Rem-CAR, Lec-CAR, and Adu-CAR clones. Statistical tests: b Tukey; c–e Dunn’s test; (f) Bonferroni-corrected comparisons. Only relevant p-values are shown; complete statistical results are available in Tables 10a-d, g. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Bar graphs represent mean ± S.E.M. from three biological replicates

Fig. 8

Aβ1-42 treatment and CD3/CD28 beads increase CD69 expression in Adu-CAR-expressing CD4⁺ T cells. a–c Schematic representations of the multicistronic constructs: a Adu-CAR:mFoxp3:EGFP, b mFoxp3:EGFP, and c MIGR:EGFP. All constructs utilize the MIGR1 retroviral transfer vector and incorporate 2A sequences to enable polycistronic expression. The term “MIGR” in MIGR:EGFP denotes the vector backbone and is used to distinguish the construct from endogenous EGFP positivity. d–f Mean-normalized percentages of CD69-positive transduced (EGFP⁺) cells following treatment, analyzed separately for d MIGR:EGFP, e mFoxp3:EGFP, and f Adu-CAR:mFoxp3:EGFP. g–i Mean-normalized percentages of CD25-positive transduced (EGFP⁺) cells for the same constructs: g MIGR:EGFP, h mFoxp3:EGFP, and i Adu-CAR:mFoxp3:EGFP. Statistical comparisons: d–f, i Games–Howell and g, h Tukey’s multiple comparisons. Full statistical details are provided in Tables 12a–f. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Bar graphs represent mean ± S.E.M. Two biological replicates, each with two technical replicates