Background
Microglia are dynamic immune cells that chemically and mechanically interact with their environment in the brain. They continuously survey the brain parenchyma, provide trophic support for neurons, remove unnecessary synapses and clear foreign materials through phagocytic and cytotoxic mechanisms [
1,
2]. Microglia act as antigen-presenting cells and contribute to brain homeostasis by secreting a plethora of pro- and/or anti-inflammatory cytokines and other signaling molecules depending on the situational cues [
1,
2]. Genome-wide association studies (GWAS) implicate dysfunction in several microglial innate immunity genes to increase the risk for Alzheimer’s disease (AD). In addition, late-onset AD (LOAD), covering the majority of AD cases, is associated with impairment in the clearance of amyloid beta (Aβ) [
3,
4], whereas rare genetic early-onset AD (EOAD) is more evidently caused by increased production of Aβ by neurons. The inability of microglia to clear accumulating toxic Aβ together with their proinflammatory functions are thought to be a significant contributor to LOAD pathology [
5,
6]. Thus, reshaping microglial functions represents a promising strategy for clearing accumulating deposits in AD and other neurodegenerative diseases with aberrant aggregates.
With the progression of AD pathology, aggregating Aβ form local stiff plaque deposits, thereby creating a stark contrast for the soft brain tissue (Aβ∼3 × 109 Pa vs normal brain ∼200–500 Pa) [
7‐
10]. In vitro, microglia are attracted to stiff regions, and they upregulate inflammatory mediators and change their morphology on stiffer substrates [
11,
12]. In vivo, implantation of stiff foreign bodies enhances microglial activation, ultimately leading into the encapsulation of the foreign body [
12], resembling the manner in which microglia envelope the amyloid plaques [
13]. This adaptation to mechanical stimuli suggests the presence of specific mechanosensors in microglia. The recently described PIEZO ion channels are among the most specific and sensitive mechanotransducers that translate extracellular mechanical forces to intracellular molecular signaling cascades [
14,
15]. PIEZO2 channels are mainly expressed in the nociceptive system [
16‐
18], while PIEZO1 channels are expressed in neurons and in non-neuronal cell types in various regions of the brain [
19,
20]. It has been demonstrated that soluble Aβ prevents PIEZO1-mediated Ca
2+ influx in HEK293 cells [
21], and that astroglia upregulate PIEZO1 around extracellular Aβ plaques [
19], suggesting a link between AD pathology and the function of brain cells that express these channels. However, the functional role of PIEZO channels in microglia remains unexplored.
Here we report that PIEZO1 is expressed in mouse microglia and human induced pluripotent stem cell (iPSC)-derived microglia-like cells (iMGLs). The activation of PIEZO1 with Yoda1, a small molecule agonist of the PIEZO1 channel, induces Aβ clearing functions in human iMGLs. Moreover, treatment of the 5xFAD mouse model of AD with Yoda1 recruits microglia towards Aβ plagues and leads to Aβ clearance in the hippocampus and cortex. In agreement with a previous study in HEK293 cells [
21], we confirm that Aβ inhibits PIEZO1 and demonstrate this for the first time in human iMGLs. Supporting our findings, existing transcriptional datasets indicate that the expression of
PIEZO1 is altered in specific disease-related subpopulations of microglia in the human AD and 5xFAD brains. The main findings are illustrated in a graphical abstract (Additional file
1: Fig S8).
Discussion
Here we show a novel approach to enhance the microglial Aβ clearing function through activation of mechanotransducing PIEZO1 channels with a small molecule agonist. As we demonstrate the beneficial effects of PIEZO1 activation both in human microglia in vitro and in an animal model of AD in vivo, this calcium-permeable mechanoreceptor could represent a novel translational target for LOAD patients covering the majority of AD cases associated with impairment in the clearance of Aβ [
3,
4] We show that functional PIEZO1 channels are expressed in human and mouse microglia and that Aβ1-42 compromises PIEZO1-mediated Ca
2+ signaling in human iMGLs. Activation of PIEZO1 elicits a unique functional microglial state with increased phagocytosis, survival and motility with both classical pro- and anti-inflammatory metabolic features. Our in-depth analysis of single-cell sequencing datasets show that
PIEZO1 is enriched in microglial subclusters in AD patients. This correlates with the expression of certain Trem2-dependent DAM signature genes [
41,
42] that take part in the Aβ clearing functions in a mouse model of AD. The function of microglial cells surrounding the plaques, particularly their motility and phagocytic clearance of Aβ, may be the key determinants governing the pathological processes in AD [
2,
76]. Importantly, it is shown here that PIEZO1 activation reduced existing Aβ plaques concomitantly with the increase in microglial marker Iba1 around the plaques in 5xFAD mice in vivo.
Research over the last two decades has attempted to find the determinants and outcomes of the microglial behavior but their beneficial vs. detrimental contribution in AD is still not fully clear. Sustained exposure to Aβ, cytokines and other inflammatory mediators appear to cause permanent impairment in microglial function at the plaque sites [
77] manifested by impairment in motility and phagocytosis in mice with AD phenotype [
78]. This could lead to inactivation of the protective function of microglia to clear Aβ [
79]. Our observation that Aβ inhibits native PIEZO1 in human microglia supports the theory that the function of microglia is compromised in AD by completing the vicious circle exaggerating AD pathology. Our experimental paradigm aimed to include various forms of Aβ as demonstrated earlier [
80] mimicking the various forms of Aβ microglia encounter in vivo in AD brain. This, together with our observations that brain Aβ burden contributes to the expression of
PIEZO1 in microglia, indicates that dysfunction or blockage of PIEZO1 could be involved in the early mechanisms triggering microglial malfunctions even before the diagnosed symptoms. Similar to the in vivo situation, in vitro Aβ preparations contain heterogenous Aβ species ranging from monomeric to fibril forms. Even though our study does not specify which Aβ species mediate the inhibitory effects or through which exact mechanism, it is likely that Aβ can modulate cell membrane mechanical sensitivity to control PIEZO1-triggered Ca
2+ influx. This could happen similarly as in human HEK293 cells rather than Aβ directly interacting with the PIEZO1 receptor [
21]. Thus, alleviating the Aβ-causing inhibitory effect on PIEZO1 offers a promising, early target for treatment paradigms. This view is further supported by the fact that the inhibition of PIEZO1 by the antagonists was toxic for iMGLs, although the toxicity could have also been a result of the unspecific off-target effects of the available inhibitors. Importantly, PIEZO1 activation in microglia provides a driver for multiple calcium-dependent mechanisms important for ‘waking up’ the protective function of these cells, including the immunological surveillance of the brain. This could be critically important for conditions such as AD where the proper function of microglia is compromised [
81]. This novel concept is in line with data showing that mechanosensitive stimuli, such as ultrasound, activates PIEZO1 channels [
82] and leads to clearance of brain Aβ by microglia [
83].
By using distinct mechanical cues and a selective molecular agonist we demonstrate the presence of abundant functional PIEZO1 mediated mechanotransduction in human iMGLs and mouse microglia. All the stimulation paradigms elicited prolonged functional responses lasting minutes and distinct from earlier observations of PIEZO1 mediated membrane currents in other cell types which were limited to ms range [
84] due to the ion channel inactivation mechanism. The prolonged activity could arise from microglia-specific PIEZO1 organization, and in particular, the unusual organization of cytoskeletal structures in these mobile cells, and/or changes to membrane lipid organization shaping the global properties such as tension and fluidity. The PIEZO1 channel can switch permanently from a transient to a sustained gating mode by strong mechanical stimulation [
85‐
87] and by repeated stimulation [
84,
88]. In particular, membrane cholesterol which is required for establishing a mature microglial phenotype [
89]
, could affect the activity of PIEZO1 channel clusters [
54,
89,
90].
Interestingly, we discovered that human iMGLs were highly sensitive to the PIEZO1 agonist Yoda1. Specific high sensitivity could stem from the species-specific structures of the PIEZO1 channel. Indeed, human and mouse share only 84% similarity in the base pair sequence of the respective gene and 81% similarity in amino acids of the protein (
https://blast.ncbi.nlm.nih.gov/). The semi-bell-shaped DRC of human cells contrasting to mouse microglia presented classical sigmoidal DRC could be due to specific PIEZO1 inactivation characteristics, analogous to similar contribution of desensitization to responses of capsaicin-activated transient receptor potential vanilloid (TRPV) receptors [
91]. Also differences in plasma membrane lipid and cholesterol composition could partly explain the differences between species since PIEZO1 activity appears to be concentrated in cholesterol-rich lipid raft domains [
92,
93] and human microglia, but not mouse microglia, show dysregulation of cholesterol homeostasis in sporadic AD models [
94]. In mice, global knock-out of
PIEZO is embryonically lethal, while in humans a PIEZO1 loss-of-function mutation has been reported to cause mainly a loss of lymphatic function [
95], whereas a gain-of-function mutation results in a red blood cell dehydration [
96] and is associated with protection from severe malaria in humans [
97]. All in all, the bell-shaped concentration dependence in iMGLs may have beneficial functional significance limiting potentially damaging cell Ca
2+ overload at high Yoda1 concentrations [
98].
Our calcium imaging experiments demonstrated that not all iMGLs responded similarly to Yoda1, suggesting that microglia cultures may have, similar to astrocytes [
19], subpopulations that mediate the varieties in PIEZO1 responses. This indicates that bulk analysis of microglia could mask the altered pathways that are affected within specific subpopulations. Our reanalysis of single-cell transcriptomic datasets of AD patients [
40] and DAM microglia in 5xFAD mice [
41,
42] showed apparently contradicting
PIEZO1 upregulation in a human AD-associated microglial subpopulation and, on the other hand, downregulation of
Piezo1 in mouse DAM. The original publications already stated that the human and mouse AD-associated genes differ, and some genes can be even up- or downregulated to opposing directions, most likely reflecting species, brain region and/or disease stage differences. Despite the differences, the datasets demonstrated that
PIEZO1-positive microglia highly express some AD signature genes and correlate with some of the homeostatic genes that are downregulated in mouse DAM. This was in line with our pro- and anti-inflammatory metabolic profile induced by Yoda1 in human iMGLs suggesting that PIEZO1 microglia have a unique gene expression pattern reflecting their Aβ clearing function. To fully elucidate the role of microglia in PIEZO1-mediated clearance of Aβ at subpopulation level over the time-course of AD pathology in relation to the contribution of other cell types, a longitudinal single-cell analysis of subcellular clusters would be required.
Even though we are the first to focus on PIEZO1 in microglia specifically, we cannot rule out that Yoda1 may exert its beneficial actions also through other brain cell types such as astrocytes [
19], oligodendrocytes [
51], and neuronal stem cells [
99] in AD [
19,
21,
100]. Being in line with a previous study of transgenic AD rats [
19], our analysis shows a higher number of
PIEZO1-positive astrocytes in human AD brain. However, even though prior studies demonstrate that astrocytes express functional PIEZO1 channels [
19,
101], in our study Yoda1 did not alter GFAP immunoreactivity in vivo. In contrast, it specifically increased microglial IBA1 immunoreactivity and its localization around Aβ deposits. Moreover, not all astrocytes express PIEZO1 and only a portion of them upregulate it in response to certain proinflammatory stimuli and aging [
19,
51]. Indeed, the lack of astrocytic contribution in our model could stem from young 5-month-old 5xFAD mice compared to 18-month-old rats in Velasco et al. 2018 study. These observations suggest that PIEZO1 channels may be implicated in both microglia and astrocytes in AD. However, microglial responses seem to precede and overwhelm astrocytic contribution [
19], following the paradigm that microglia are needed for astrocytic reactivation [
102]. Thus, we suggest that PIEZO1 in microglia plays a leading and mainly protective role in the context of AD pathology.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.