Increased activity of Piezo1 channel in red blood cells is associated with Alzheimer's disease-related dementia

Abstract

Introduction: Red blood cells (RBCs) are crucial for oxygen delivery to active tissues and endure significant mechanical forces in the microcirculatory bed. The enrichment of mechanosensitive Piezo1 channels, linked to the cytoskeleton, aids RBCs in navigating the narrow capillaries. In Alzheimer's disease (AD), impaired brain microcirculation may necessitate enhanced Piezo1 function in RBCs.

Methods: With micropipette aspiration and flow cytometry technics, we evaluated, using the specific Piezo1 agonist Yoda1, AD-related alterations in the biomechanical properties of RBCs from cognitively healthy patients (HC) and individuals with mild cognitive impairment (MCI) and AD.

Results: We show that beta-amyloid (Aβ) peptides alter the biomechanical properties of RBCs. We observed significantly higher Yoda1-induced calcium responses in RBCs in individuals with MCI and AD compared to RBCs from age-matched HC.

Conclusion: Our data suggest that Yoda1-induced Ca²⁺ flux through Piezo1 channel emerges as a measurable indicator associated with and improves the detection of AD-related dementia.

Highlights: Piezo1 channels aid the navigation of red blood cells (RBCs) through narrow capillaries. Alzheimer's disease (AD) patients show increased Yoda1-induced activation of Piezo1 in RBCs. Incorporation of Yoda1-induced Piezo1 readouts improved the detection of AD-related dementia. Investigating Yoda1-induced Piezo1 activity associated with early AD.

Keywords: Alzheimer's disease; Ca‐imaging; Piezo1; RBC.

FIGURE 1

Aβ changes biomechanical properties of RBC membrane. (A) Visual representation of the experimental technique. (B) Graphical representation of experimental technique. Pressure gradually increased by 0.5 mbar. (C) Fitting curve representing the activation probability for a given pressure in case of control (circles) and Aβ (squares) preincubated samples. (D) Boxplots representing activated pressure values for control and preincubated with 1 µM Aβ populations of RBCs. N(control) = 99, N(Aβ) = 99. (E) Boxplots representing Young's modulus values for control and preincubated with 1 µM Aβ populations of RBCs. N(control) = 73, N(Aβ) = 87. (F) Boxplots representing viscosity values for control and preincubated with 1 µM Aβ populations of RBCs. N(control) = 73, N(Aβ) = 87. For the box‐and‐whisker plots, the horizontal bar shows the median, and the upper and lower boundaries show the min and max values, respectively. Mann‐Whitney U‐test, **p < 0.01, ***p < 0.001. Aβ, beta‐amyloid protein; RBC, red blood cell.

FIGURE 2

Aβ preincubation decreasing Yoda1‐activated positive cells. (A) Visual presentation of non‐activated (a) and Yoda1 5 µM activated (b) RBCs in flow cytometry stream. (B) Box‐and‐whiskers plots of cell responded to Yoda1 5 µM application in RBCs that were incubated (or not) 20 min with Aβ protein in different concentrations for 20 min prior to flow recording. Ionomycin was used as a positive control. For the box‐and‐whisker plots, the horizontal bar shows the median, and the upper and lower boundaries show the min and max values, respectively. N (samples) = 4, Kruskal‐Wallis test, *p < 0.05, **p < 0.01. Aβ, beta‐amyloid protein; RBC, red blood cell.

FIGURE 3

Yoda1‐induced activation of Piezo1 receptors in human RBC. Activation of Piezo1 receptors presented as normalized on baseline values of Fluo4 median fluorescence intensity (MFI) induced by 5 µM Yoda1 agonist in RBCs of groups of patients: HC versus MCI patients (A), HC versus AD patients (B), and MCI versus AD patients (C) the time‐course of changes in MFI. The moment of Yoda1 application marked with an arrow. (D‐F) Box‐and‐whiskers plots showing Fluo‐4 fluorescence at 60 s in the presence of Yoda1 5 µM. For the box‐and‐whisker plots, the horizontal bar shows the median, and the upper and lower boundaries show the min and max values, respectively. *p < 0.05, ***p < 0.001 [n (control) = 31, n(MCI) = 39, n(AD) = 32, two‐way ANOVA, followed by Bonferroni's multiple comparison test]. ANOVA, analysis of variance; MCI, mild cognitive impairment; RBC, red blood cell.

FIGURE 4

Changes in optical measurements after Piezo1 receptors activation in human RBC. (A) Schematic presentation of alterations in scattered parameters of RBCs after Yoda1 application. Created with Biorender.com (B) SSC‐A and (C) FSC‐A differences in HC, MCI, and AD patients’ RBCs. Every box‐and‐whiskers plot dot is showing percentage differences between BL signal and Yoda1 treated signal within one RBC sample Ca²⁺ recording. For the box‐and‐whisker plots, the horizontal bar shows the median, and the upper and lower boundaries show the min and max values, respectively. *p < 0.05, **p < 0.01 [n (control) = 31, n(MCI) = 39, n(AD) = 32, ANOVA, followed by Bonferroni's multiple comparison test].

FIGURE 5

Plasma biomarkers characteristics of APOLLO cohort. (A) Aβ40, (B) Aβ42, (D) GFAP, (E) NfL, (F) pTau231 plasma concentrations of individuals across the AD, MCI, and HC cases. The box‐and‐whisker plots show plasma concentrations of chosen biomarkers across groups. For the box‐and‐whisker plots, the horizontal bar shows the median, and the upper and lower boundaries show the min and max values, respectively. Statistical outliers have been removed using GraphPad Prism 10. p values indicate the results of the Kruskal‐Wallis test with Dunn's multiple comparisons test at *p < 0.05, ** p <  0.01. Aβ40, beta‐amyloid peptide; GFAP, glial fibrillary acidic protein; HC, healthy controls; MCI, mild cognitive impairment; NfL, neurofilament light chain.

FIGURE 6

Results of the patient classification using a random forest model. (A) The receiver operating characteristic (ROC) curves, illustrating the model's performance in distinguishing Alzheimer's disease (AD), controls, and mild cognitive impairment (MCI) based on the area under the curve (AUC). (B) The importance of individual features in the classification of each patient class, measured by Gini decrease and p‐value. Precisely, the point size is related to the overall Gini decrease of the feature, while axis and shape are related to the Gini decrease of the feature for each specific class (i.e., more the point is at the right and more important for recognizing control patients). (C) The importance of feature interactions, represented by the frequency of co‐occurrence within the random forest's decision trees. (D) The t‐distributed Stochastic Neighbor Embedding (t‐SNE) representation of the random forest proximity matrix, revealing the similarity between individual patient samples based on their feature values and assigned class.