Sporadic Alzheimer’s Disease- and Neurotoxicity-Related microRNAs Affecting Key Events of Tau-Driven Adverse Outcome Pathway Toward Memory Loss

Strategy for the selection of the candidate miRs implicated in sAD and/or neurotoxicity

A comprehensive search for publicly available articles referring to “microRNA”, “miRNA”, “miR” in combination with “Alzheimer’s disease”, “AD”, and “environmental neurotoxicity”, “chemical-induced neurotoxicity” in PubMed and ScienceDirect databases was conducted. First, the miR search covered implication of human miRs, detected in biofluids, such as blood, serum, plasma, cerebrospinal fluid, and in brain tissues, from patients with clinical sAD. The retrieved miR profiles of sAD patients were compared to those of healthy controls. Classification of sAD stage was based on several clinical diagnostic criteria, including Mini-Mental State Examination scoring (MMSE), Braak staging, or Clinical Dementia Rating (CDR), established by National Institute of Neurological Disorders or Stroke in collaboration with the Alzheimer’s Association (NINDS-AA), National Institute of Neurologic and Communicative Disorders and Stroke/Alzheimer’s Disease and Related Disorders Association (NINCDS–ADRDA), or US National Institute on Aging–Alzheimer’s Association (NIA-AA). Additionally, miRs related to environmental neurotoxicity were retrieved from available epidemiological and in vivo human and animal studies, relevant to the nervous system and investigating brain tissue, brain-cells and/or blood, following exposure to neurotoxicants such as industrial chemicals, drugs, pollutants, or metals. All retrieved miRs were quantitatively detected, by using either highly sensitive and specific methods such as quantitative polymerase chain reaction (qPCR), or high-throughput technologies (e.g., microarrays, next generation sequence). Next, the available molecular and cellular information related to the potential role of the miRs in AD or neurotoxicity was collected and placed in the context of the proposed tau-driven AOP toward memory loss. Search priority was given to human studies, but animal studies were considered to fill knowledge gaps when human studies were not available. Only articles written in English were included, which were published from 1999 till 2021.

Functional miR analysis

To understand the role of miRs, in silico-based functional analysis was performed. This approach typically consists of target prediction following functional analysis of identified miR targets. Human miR–target interactions validated at experimental level were retrieved by using the MiRTarBase [43] for all identified miRs implicated in sAD and neurotoxicity. The analysis resulted in a list of genes, whose expression is modulated by the investigated miRs, according to the database. The number and the reliability of hits accounting for each miR-target interaction were registered. The identified genes were organized in predictive sets, including genes that are potential targets of the miRs related to the hypothetical starting point and to the KEs defined by the AOP. In order to functionally characterize each predictive set, the list of pathways where target genes are involved was obtained by using PathDip 4, an annotated database of human signaling cascades, comprising core pathways from 24 major curated pathways databases [44]. For each event of interest, text-mining analysis superimposed specific semantic parameters to filter pertinent pathways and to select among the targets only genes involved in these pathways. The list of miRs that, according to functional analysis, potentially affect these pathways was reversed engineered and compared to the literature outputs. The results of functional analysis were visualized by using the software flourish.studio and R 3.6.0 (packages ggplot2, igraph and alluvial).

miRs in AD and neurotoxicity

All selected miRs and their detailed information related to sAD and chemical-induced neurotoxicity are provided in the Supplementary Table 1. Based on the selection criteria, 235 human (mature) miRs with sufficient evidence for their involvement in human AD pathology, were identified in cerebrospinal fluid (N = 100), serum (N = 83), plasma (N = 57), whole blood (N = 40), peripheral blood mononuclear cells (N = 16), and brain tissues (N = 89). Additionally, 115 and 87 miRs linked to human and animal chemical-induced neurotoxicity, respectively. Not all the selected neurotoxicity-related miRs were implicated in the human sAD. The numbers of miRs overlapping between sAD (human) and neurotoxicity (human and/or animal) are illustrated, while Table 1 shows common miRs in detail.

Among the miRs potentially in sAD and human chemical-induced neurotoxicity, let-7b, let-7f, let-7g, miR-10b, miR-21, miR-27a/b, miR-28, miR-30b/c/d, miR-103a, miR-107, miR-142, miR-144, miR-146a, miR-150, miR-151a, miR-181a, miR-191, miR-199a, miR-222, miR-320a/b/c, miR-378c, miR-409, miR-423, and miR-425 were shown by epidemiological studies to be dysregulated in either human plasma or whole blood after exposure to air pollution [45, 46]. In another human study, an aberrant expression of let-7g, miR-10a, miR-23a, miR-30c, miR-34c, miR-98, miR-124, miR-145, miR-195, miR-376a, miR-498, miR-532, miR-550a, and miR-647 was observed after exposure of human neural progenitor cells to paraquat [47]. Changes in miR-9, miR-125b, miR-128, miR-146a, and miR-322 expression by human brain cells after exposure to neurotoxic metal sulfates such as aluminum sulfate [48–50] or aluminum-maltolate [51], have been reported. Human neuroblastoma cell line SH-SY5Y revealed a modified miR-214 expression following treatment with isoflurane, a commonly used anesthetic [52]. Acyclovir, an antiviral drug against herpes simplex virus-1, upregulated miR-146a in mixed cultures of primary human neuronal and glial cells [53]. Moreover, dysregulation of miR-21 has been reported following treatment of human glioma cells with chemotherapeutic drugs, temozolomide and taxol [54, 55].

With regard to miRs implicated in both human AD and animal neurotoxicity, ethanol-induced neurotoxicity involved changes in let-7b expression in rat [56], in miR-10a/b in mouse brain [57], in miR-10b, miR-29c, miR-30a/e, miR-145, miR-200a, and miR-339 in fetal mouse brain [58], and in miR-9, miR-21, miR-153, and miR-335 in fetal mouse cerebral-cortex neuronal cells [59]. Short-term nicotine exposure was associated with alterations in the expression of miR-10b, miR-21, miR-26a, miR-29a, miR-29b, miR-29c, miR-30a-5p, miR-30c, miR-93, miR-98, miR-125b, miR-137, miR-152, miR-181b, miR-186, miR-188, miR-210, miR-301a, miR-335, and miR-374 in rat or mouse neuronal cultures [60, 61]. Exposure of olfactory system of zebrafish to Cu dysregulated the expression of let-7a, let-7e, let-7f, let-7g, let-7i, miR-23a, miR-27b, miR-30c, miR-31, miR-126, miR-128, miR-138, miR-146a, miR-150, miR-193b, miR-214, and miR-219 [62]. Also, exposure of mice to hexahydro-1,3,5-trinitro-1,3,5-triazine, a common environmental contaminant, induced upregulation of miR-30a-5p, miR-195, and miR-206 [63]. Exposure of mouse serotonergic cells and rat hippocampal neurons to antipsychotic or antidepressant drugs, such as lithium/valproate or fluoxetine, was associated with decreased expression of miR-34a and increased expression of miR-16, respectively [64, 65]. In adult zebrafish, exposure to fipronil or triazophos, or their mixture, induced changes in miR-30b, miR-31, miR-128, and miR-455 levels [66].

Coupling retrieved miRs from literature to the hypothetical starting point and key events of the tau-driven AOP blueprint toward memory loss

The miRs identified in the literature were linked to the hypothetical starting point and KEs described in the proposed AOP for memory loss [11]. Notably, no miRs were found to affect cytotoxic tau oligomers (KE5), dysfunctional axonal transport (KE6), and neuronal dysfunction (KE9). All retrieved miRs, which have been detected in brain (cells/tissues) or blood, with evidence for their links to the starting point and KEs of the AOP are provided below. As aforementioned, priority was given to human studies, however, when they were not available animal studies relevant to the described events were included. In addition, starting point- and KEs-related miRs found in non-brain cells are also discussed.

Glucose and cholesterol dysmetabolism (hypothetical starting point)

Mitochondrial dysfunction (KE1)

Several miRs are known to regulate mitochondrial proteins involved in mitochondrial oxidative phosphorylation (OXPHOS) and electron transport chain (ETC) Complexes I-VI, affecting ATP production, mitochondrial metabolism and eventually mitochondrial function [99], in both humans and animals.

miR-7 regulated mitochondrial permeability transition pore (MPTP) via targeting voltage-dependent anion channel 1 (VDAC1) in human neuroblastoma SH-SY5Y cells [100]. miR-34b/c modulated mitochondrial function in differentiated human SH-SY5Y dopaminergic neuronal cells [101]. Increased expression of miR-27a and miR-103 in human SH-SY5Y cells treated with tumor necrosis factor a (TNFα) suppressed the expression of functional units of Complex I, resulting in mitochondrial dysfunction [102].

Several animal studies have produced data likely to be relevant for human pathology. miR-34a negatively affected the integrity of the blood–brain barrier through reduction of mitochondrial oxidative phosphorylation and ATP production, and cytochrome c levels in murine cerebrovascular endothelial cells [103]. An in vivo study revealed that mice lacking miR-29 showed upregulation of iron responsive element binding protein 2 (Ireb2) [104], which has been implicated in intracellular iron delivery in neurons and mitochondrial damage [105]. miR-124 affected mitochondrial localization and function by targeting the intermediate filament vimentin in mouse motor neurons [106]. miR-127 targeted the mitochondrial membrane protein mitoNEET in rat primary cultured spinal neurons [107]. miR-330 has been suggested to alleviate mitochondrial dysfunction in AD by targeting vav guanine nucleotide exchange factor 1 (VAV1) via the MAPK signaling pathway in neuron cells of mice [108]. The brain-specific miR-338 regulated OXPHOS and mitochondrial metabolism via cytochrome c oxidase IV (COXIV) in neurons, involved in ETC in mitochondria of rat neuronal cells [109, 110].

miR-15b, miR-16, miR-195 and miR-424, all sharing the same “seed” sequence, were implicated in ATP production [111], while miR-23a and miR-23b targeted glutaminase, an important player in mitochondrial metabolism, and particularly in ATP production and glutathione synthesis, in different tissues or cells of human or mice [112]. Also, miR-210 was implicated in mitochondrial respiration via targeting mitochondrial iron sulphur cluster homologue in many cell types [109]. miR-15a targeted uncoupling protein 2 (in mouse β-cells), which regulates oxygen consumption and ATP production [109]. Moreover, miR-143 is involved in mitochondrial function via targeting the ERK5 pathway in human colorectal cancer DLD-1 cells [113]. In cancer cells, miR-183 targets isocitrate dehydrogenase 2, an important member of the Krebs cycle, thereby modulating mitochondrial function, as well [114].

Oxidative stress (KE2)

The miR-mediated regulation of oxidative stress has been described in respect to its involvement in several pathological conditions, including neurological disorders [115]. In human or animal studies, miR-9 [116, 117], miR-125b [116], miR-34c [118], miR-142 [119], miR-29, miR-124, miR-128 [117], and miR-210 [120] have been linked to oxidative stress triggered by reactive oxygen species (ROS) accumulation, in brain cells.

Circulating miR-124a, miR-125b, miR-142, and miR-483 have been associated with cellular oxidative stress in mild cognitive impairment (MCI) patients [121]. In serum of Parkinson’s disease patients and in human SH-SY5Y cells, miR-153 has been reported to promote oxidative stress via targeting nuclear factor erythroid 2-related factor 2 (Nrf2) [122].

Moreover, let-7a altered the levels of ROS in murine microglia under inflammatory conditions [123]. miR-132 inhibited the expression of inducible nitric oxide synthase and oxidative stress by inhibiting MAPK1 expression in hippocampus tissue of AD rats, improving their cognitive function [124]. Expression of miR-146a inversely correlated with expression of oxidative stress indicators, including malondialdehyde and p22phox, in the brain of chronic type 2 diabetes rats [125]. The miR-200 family is implicated in oxidative stress through ROS production in rat hippocampal neurons [126]. miR-486 represses neuronal differentiation 6 (NEUROD6) in mouse motor neurons due to progressive neurodegeneration mediated by ROS [127]. miR-330 suppressed oxidative stress in neuronal cells of AD mice by targeting VAV1 via the MAPK signaling pathway [74].

Other oxidative stress-related miRs, including let-7f, miR-9, miR-16, miR-21, miR-22, miR-29b, miR-99a, miR-125b, miR-128, miR-143, miR-144, miR-155, and miR-200c were reported in human cultured cells when treated with H₂O₂ [117]. Lastly, miR-23b was found to regulate ROS production in cancer cells [128], and to inhibit oxidative stress in mice with diabetic neuropathy via controlling MAPK signaling [129].

Hyperphosphorylation of tau protein (KE3)

Abnormal phosphorylation of tau protein by activation of kinases (e.g., ERKs, MAPKs, cyclin-dependent kinase 6 (CDK6), and GSK-3β), and deactivation of phosphatases (e.g., PP2A), is also mediated by miRs.

Brain miRs, including miR-9, miR-124, miR-132, and miR-137, regulated 4R:3 R tau ratios in sporadic progressive supranuclear palsy patients brain [130], contributing to abnormal tau phosphorylation. miR-219 mediated a decreased tau-tubulin kinase 1 protein and GSK-3β by directly binding to their 3’-UTR, inducing tau phosphorylation in human SH-SY5Y cells [131]. Expression of let-7b in cerebrospinal fluid in human CD4+ T lymphocytes positively correlated with total tau and phosphorylated tau [132]. miR-106b inhibited phosphorylation of tau at Tyr18 by targeting Fyn gene expression in human SH-SY5Y cells stably expressing tau [133]. miR-93 activated phosphoinositide 3-kinase (PI3K)/Akt pathway via directly inhibition of phosphatase and tensin homolog (PTEN), besides PH domain leucine rich repeat protein phosphatases (PHLPP) and forkhead box O3 (FOXO3), in human glioma cells and tissues [134], of which tumor-suppressor PTEN affected tau phosphorylation and aggregation through binding to microtubules in AD patient brains [135]. Also, miR-146a induced abnormal neuronal tau phosphorylation by targeting Rho associated coiled-coil containing protein kinase 1 (ROCK1)/ PTEN pathway in human SH-SY5Y neural cells [136]. In blood of AD patients, miR-103a and miR-107 were found to target cyclin dependent kinase 5 regulatory subunit 1 (CDK5R1), among other targets, and to contribute to tau hyperphosphorylation [137].

In human relevant animal studies, among the miR-16 family, miR-15(a/b) emerged as potent regulator of ERK1/2 signaling pathway in mouse or rat neuronal cells. This pathway is involved in tau phosphorylation [138, 139]. miR-15a, miR-195, and miR-497 were demonstrated to directly interact with the 3’-UTR of ERK1 mRNA in mouse N2a cells [138]. miR-124 inhibited abnormal tau hyperphosphorylation by targeting caveolin-1- PI3K/Akt/GSK3β pathway in mouse neuroblastoma cell lines [140]. miR-148a regulated protein kinase B and MAPK/ERK signaling pathways in rat hippocampal neurons [75], known to contribute to tau hyperphosphorylation. miR-125b induced tau hyperphosphorylation and upregulation of p35, cdk5, and p44/42-MAPK signaling in rat primary neurons [141]. miR-101b targeted 5’ adenosine monophosphate (AMP)- activated protein kinase (AMPK) blocking the histone deacetylase 2 (HDAC2), suggested to be involved in tau pathology, besides dendritic abnormalities, and memory deficits in AD mice [142]. miR-98 increased tau phosphorylation in mouse N2a/WT cells [78]. miR-26a directly regulated GSK-3β gene expression in human airway smooth muscle cells [143], and miR-26b activated CDK6 in rat primary cortical neurons [144], both targeting brain-derived neurotrophic factor (BDNF) [145] and implicated in tau hyperphosphorylation in the hippocampus of a mouse model for Down syndrome during aging. miR-34a and miR-34c regulated the tau expression in human neuroblastoma cell lines [146], with the latter having a negative effect on memory formation via suppression of sirtuin 1 (SIRT1) levels in mouse hippocampus [147]. In transfected primary neurons of rat hippocampus or cortex, miR-128a affected tau hyperphosphorylation via targeting co-chaperone BAG2 pathway [148]. In cultured rat cortical neurons, miR-200c repressed PTEN expression [149] and induced tau hyperphosphorylation in human SH-SY5Y cells [150]. miR-326 decreased hyperphosphorylation of tau by inhibiting c-Jun N-terminal kinase pathway and by targeting VAV1 in neurons of AD mice [151].

Overexpression of brain-enriched miR-138 increased tau phosphorylation in human HEK293/tau cells via targeting the retinoic acid receptor alpha/GSK-3β pathway [152]. Lastly, miR-21 has been shown to regulate matrix metalloproteinase-2 via PTEN pathway in murine cardiac fibroblasts [153].

Dysfunctional autophagy (KE4)

Many miRs have been implicated in the regulation of autophagy by targeting autophagy-related proteins (i.e., autophagy genes (ATG), mTOR) or proteins related to lysosome system [117].

miR-17 regulated the ATG7 and autophagic processes in human glioblastoma cells [154]. miR-100 targeted ATG10 in 1-methyl-4-phenyl-pyridinium ion (MPP+)-intoxicated SH-SY5Y cells [155]. miR-137 reduced starvation-induced autophagy via targeting ATG7 in human glioma cell line [156]. Inhibition of miR-132 in human SH-SY5Y cells suppressed mRNA and protein expression levels of autophagy-related protein light chain 3 (LC3) and Beclin 1 [157]. miR-27a and miR-27b regulated PTEN-induced kinase 1 expression and autophagic clearance of damaged mitochondria in human dopaminergic-like M17 cells [158]. miR-124 regulated autophagy via targeting AMPK/mTOR pathway in human dopaminergic neurons [159]. miR-128 targeted key members in mTOR signaling, including mTOR, RICTOR, IGF1, and phosphoinositide-3-kinase regulatory subunit 1 (PIK3R1), in human glioma cells [160]. miR-181b and miR-590 controlled autophagy by targeting the PTEN/Akt/mTOR signaling pathway, in PC12 cells and human glioma cells, respectively [161, 162]. Also, miR-496 has been reported to reduce the expression of mTOR in peripheral blood mononuclear cells of aged subjects [163].

Several animal studies have addressed this key event of human AD pathology. miR-23b was correlated with cognitive impairment after traumatic brain injury through inhibition of ATG12-mediated neuronal autophagic activity in rat model of traumatic brain injury [164]. miR-214 decreased autophagy of hippocampal neurons via targeting Atg12 expression in SAMP8 mouse model of sAD [165]. In murine neurons, miR-299 suppressed Atg5 and antagonized caspase-dependent apoptosis, inhibiting autophagy [166]. miR-9 was implicated in autophagy in rat bone marrow mesenchymal stem cells differentiating into neurons by regulating the ratio of autophagy-related protein LC3 and the expression of neuron specific enolase and microtubule-associated protein 2 (MAP2) [167]. miR-193b directly targeted tuberous sclerosis 1 to regulate mTORC1 activity, promoting protective autophagy in mouse neuroblastoma NSC-34 cells [168]. miR-34a activated SIRT1/mTOR signaling pathway inducing impaired autophagy in brain aging mice after administration of urolithin A supplementation [169]. miR-199a controlled methyl-CpG binding protein 2 (MeCP2) and mTOR signaling in the brain of mice [170]. miR-106b has been involved in autophagy by targeting to sequestosome 1 (SQSTM1)/p62 in mouse hematopoietic myeloid progenitors [171].

In other cell types, miR-20a and miR-106b regulated leucine deprivation-induced autophagy via direct targeting of the Unc51-like kinase 1 (ULK1) protein in myoblasts [172]. There is evidence supporting the involvement of miR-93 in autophagy by regulating core autophagy kinase, ULK1 expression in mouse embryonic fibroblasts or Chinese hamster ovary cells [173], and of miR-30d and miR-101 through inhibition of Beclin1 and Atg4d expression in human breast cancer cells or hepatocellular carcinoma cells [174]. miR-200c upregulated the endoplasmic reticulum stress signaling via LC3-II expression in human prostate adenocarcinoma PC-3 cells [175], while miR-103/miR-107 regulated end-stage autophagy in mouse epidermis via phospholipase D (PLD)1/2–protein kinase C–dynamin pathway [176]. miR-101 has been suggested as key regulator of autophagy via targeting stathmin 1 (STMN1) in human breast cancer cells. miR-183 has been shown to regulate autophagy via targeting ultraviolet radiation resistance-associated gene, an autophagy regulator promoting autophagosome formation and maturation, in colorectal cancer cells [177]. Finally, miR-15a, miR-20a/b, miR-23a, miR-29a, miR-106a/b, miR-195, miR-590 [117], and miR-101 [178] have been reported as autophagy-related miRs, but mainly in cancer.

Synaptic dysfunction (KE7)

The role of miRs in synaptic transmission and plasticity has been described by others [179].

Let-7 family members regulated synaptic function [180] and modulated synaptic plasticity by controlling BDNF translation [181]. In particular, let-7c was implicated in synaptic function in human neurons [182]. miR-26a/b and miR-374 contributed to the regulation of BDNF expression in human peripheral nuclear blood cells [183]. In human neurons, miR-132 and miR-182 decreased the BDNF protein levels [184]. miR-34 regulated synaptogenesis [185] and mediated the regulation of post-synaptic cytoskeletal SH3 and multiple ankyrin repeat domains 3 (SHANK3) in superior-temporal lobe of sAD patients [186]. miR-130b targeted MECP2 in human brain neocortex [187]. In brains of AD patients, miR-30b regulated ephrin type-B receptor 2, SIRT1, and glutamate ionotropic receptor AMPA type subunit 2 (GluA2), which are all important targets for maintaining synaptic integrity [188]. In human plasma, miR-181c and miR-210 targeted neuronal pentraxin 1 (NPTX1) and neuronal pentraxin receptor (NPTXR), both implicated in activation and clustering of AMPA receptors in postsynaptic terminals, and affecting glutamatergic synapses in the hippocampus and synaptic plasticity [189]. miR-572 targeted neural cell adhesion molecule 1 (NCAM1) expression promoting progressive decrease in myelination in human oligodendroglial cell line [190, 191]. miR-485 was found to regulate synaptic genes, PSD95, surface GluR2, and the miniature excitatory post-synaptic current frequency as shown in human hippocampal neurons [192].

Animal studies addressing this key event of human pathology have identified several miRs with potential human relevance. Among brain-specific miRs, miR-9(-3p) has been associated with hippocampal long-term potentiation (LTP), and learning and memory deficits, via targeting LTP-genes, dystrophin (Dmd) and synapse-associated protein 97 (SAP97) in mouse hippocampus [193]. Also, miR-132, miR-134, and miR-138 regulated the neuronal dendritic spines, excitatory synaptic transmission, and synaptic function [193]. miR-26a and miR-384 were downregulated during LTP, and both targeted ribosomal S6 kinase 3 (RSK3), regulating LTP maintenance and long-lasting spine enlargement in hippocampal slices from mice [194]. Inhibition of miR-34b/c in postischemic hippocampal CA1 pyramidal neurons (mice) rescued impaired synaptic plasticity and diminished memory deficits [195]. miR-124 was implicated in long-term plasticity of synapses (rat PC12 cells) [196] and dendritic actin cytoskeleton via regulating the activity of small Rho GTPases (mouse primary cortical neurons) [197]. In a study on adult mouse forebrain, miR-7, miR-29a, miR-137, miR-200c, miR-322, and miR-339 were found highly present in synapses [198]. miR-151 and miR-23a were found to be involved in synaptic reorganization and transcription, respectively, in rat dentate gyrus [199]. The neurite-enriched miR-376b was found to regulate synaptic transmission and plasticity during postnatal period in the mouse hippocampus [200]. miR-16 regulated the neurotrophin receptor-mediated MAPK/ERK pathway in mouse hippocampal neurons and was demonstrated to play an important role in both synapse formation and plasticity [201]. miR-132 targeted MeCP2 in rat neurons [187, 202] and regulated BDNF–ERK–cAMP response element-binding protein (CREB) signaling pathway in rat cortex [203], implicated in dendritic development, synaptogenesis, and synaptic plasticity. In primary rat hippocampal neuronal cells, ectopic expression of miR-34a, miR-193a, or miR-326, modulated endogenous Arc protein expression in response to BDNF treatment [204]. In the nucleus accumbens (in rat basal forebrain), let-7d regulated the expression of BDNF, glutamate ionotropic receptor AMPA Type Subunit 2 (GRIA2), MeCP2, phosphorylation of CREB, supporting its involvement in regulating synapse maturation, growth, and dendritic arborization [205]. miR-19b targets PTEN, regulated dendritic development and synapse formation by modulating the AKT-mTOR signaling pathway in mouse hippocampal neurons [206]. miR-301a directly bound to PTEN in rat cortical neurons, conferring neuroprotection in oxygen-glucose deprivation-induced neuronal injury [207]. Glial-enriched miR-19a has been shown to target Mef2c and neurogenin 1 (Neurog1), known to enhance neuronal differentiation, in primary cultures from postnatal (day 1) rat cortex [208]. miR-134 [209] and miR-138 [210] were found to be involved in synaptic plasticity via targeting SIRT1 in murine neuronal cells. miR-7 targeted neuroligin 2 (NLGN2) and altered synaptic function [188]. miR-124 suppressed GluA2 expression via binding to its 3-UTR, resulting in calcium-permeable AMPARs (CP-AMPARs) formation in rat primary hippocampal neurons [211]. miR-132 and miR-212 affected synaptic transmission and plasticity possibly by regulating the number of postsynaptic AMPA receptors in mice [212]. In mouse hippocampal neurons, miR-146a was demonstrated to bind to microtubule associated protein 1B (MAP1B) mRNA and to inhibit mGluR-dependent AMPA receptor endocytosis [213], while miR-223 bound to the NMDA receptor subunit GluN2B and the AMPA receptor subunit GluA2 [214]. miR-125b (targeting NMDA receptor subunit NR2A) and miR-132 (targeting Rho GTPase-activating protein (p250GAP)) have been reported to exhibit opposing effects on dendritic spine morphology and synaptic function in mouse hippocampal neurons and both to target fragile X mental retardation protein (FMRP) [215, 216]. miR-137 regulated glutamate receptor and synaptic vesicle via binding to GRIA1 enhancing AMPA receptor-mediated synaptic transmission, or via targeting synaptotagmin-1 (Syt1), a synaptic vesicle membrane protein, in rat neurons [217]. miR-125a affected postsynaptic density protein 95 (PSD-95) expression and FMRP phosphorylation as described in mouse cortical neurons [218]. miR-501 mediated the activity-dependent regulation of the AMPA receptor subunit GluA1 expression in rat hippocampal neurons [179]. miR-204 directly targeted EphB2, a component of Eph/Ephrin synaptic signaling, and N-methyl-D-aspartate receptor subunit NR1 in aging mouse hippocampal neurons [219]. miR-188 targeted neuropilin 2 (Nrp-2) to potentiate spine development, miR-135, miR-191, and miR-29a/b targeted tropomodulin 2 (Tmod2), complexin-1/2 and ARP2/3 complex subunit 3 (Arpc3), respectively, and exerted an impact on actin cytoskeleton and spine morphology in rat hippocampal neurons [216]. miR-148a regulated synaptic transmission through targeting known synaptic genes, e.g., Snap25, syntaxin 6 (Stx6), vesicle associated membrane protein 1 and 2 (Vamp1 and Vamp2), calmodulin 1 (Calm1), heat shock protein family A (Hsp70) Member 1A (Hspa1), gap junction protein alpha 1 (Gja1), calcium/calmodulin-dependent protein kinase type II alpha chain (CaMKII), synaptotagmin (Syt), NMDA-R and Vamp in mouse hippocampi [220]. Overexpressed miR-210-5p associated with decreased synaptic number via targeting Snap25 in primary hippocampal neurons of rats [221]. miR-574 expression in hippocampi of mice associated with impaired synaptic and cognitive function in response to exposure to particulate matter [222].

In human or animal studies, miR-10b (human), miR-15a (mouse), miR-206 (mouse) [223–225], miR-30a (human) [226], miR-155 and miR-191 (human) [225] regulated neuronal maturation and promoted BDNF expression in neurons. Functional enrichment studies using human and mouse data demonstrated the involvement of miR-195 in synaptic function by rescuing in vivo ApoE4-associated cognitive deficits [227]. Knockdown of miR-1 in heart resulted in upregulated SNAP-25 expression in synapses of hippocampi from transgenic mice through transportation of exosomes [228].

Neuroinflammation (KE8)

Numerous miRs have been implicated in neuroinflammation leading to AD pathology, with microglia, specialized long-term resident macrophages, to play a key role in the inflammatory responses. Microglial activation is accompanied by expression of recognition, cytokine, and neuronal receptors. Microglia differentiate into to M1 (pro-inflammatory) or M2 (anti-inflammatory) phenotypes. M1 microglia release inflammatory mediators e.g., ROS, matrix metalloproteinase-9 and pro-inflammatory cytokines, e.g., TNFα, interleukin (IL)-6 and IL-1β, while M2 microglia release anti-inflammatory and protective cytokines, e.g., IL-10, transforming growth factor beta 1 (TGF-β), IL-4, and IL-13 [229].

In a review, several miRs are discussed in the context of their involvement in neuroinflammation observed in different models [229]. Let-7 family either inhibited or promoted inflammation. On one hand, let-7 induced polarization of macrophages into the anti-inflammatory M2 phenotype through targeting C/EBP-δ transcription factor and promoted M2 phenotype in microglia. On the other hand, let-7 stimulated inflammatory response via targeting the cytokines IL-6, IL-10, and toll-like receptor 4 (TLR4), and promoted activation of microglia and macrophages via targeting TLR7 [229, 230]. miR-155 and miR-124 were induced in macrophages and microglia in response to nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) dependent TLR signaling and the Th2 cytokines IL-4 and IL-10, and TLR-6 and myeloid differentiation primary response 88 (MyD88), respectively [229]. miR-107 regulated neuroinflammation via targeting granulin [229].

miR-27a inhibited inflammatory response via negatively regulating TLR4, TNFα, and IL-6 as shown in human serum [231]. miR-19b was implicated in inflammatory responses via regulating NF-κB signaling in human brain microvascular endothelial cells upon meningitic E. coli infection [232]. miR-30a was identified as an important inflammatory modulator in human microglia (of chronic active lesions of multiple sclerosis patients) targeting the proinflammatory cytokines, IL-1β, and inducible nitric oxide synthase [233]. miR-132 induced increase in proinflammatory lymphotoxin and TNFα via suppressing SIRT1 in human B cells of multiple sclerosis patients [234]. miR-142 has been suggested as key mediator of IL-1β via downregulation of the glial glutamate-aspartate transporter (GLAST) in human cerebrospinal fluid with active multiple sclerosis [235]. Overexpression of miR-29a/b cluster in immune cells has been linked to chronic inflammation [236] and aging via targeting specific central nervous system microglia modulators, IGF1 and C-X3-C motif chemokine ligand 1 (CX3CL1) in human cortical tissue [69].

miR-9 regulated the NF-κB pathway and induced microglial activation through targeting monocyte chemotactic protein-1-induced protein-1 (MCPIP1) in rat in vitro and in vivo models [237]. miR-125b regulated microglia activation via NF-κB-dependent inflammatory pathway in mouse microglial cells [238]. miR-27b directly targeted Bcl-2-associated athanogene 2 (Bag2) in macrophages regulating TLR-2/MyD88/NF-κB signaling pathway in murine macrophages-like cells [239]. miR-146a regulated TLR/MyD88/NF-κB and Janus kinase-signal transducer and activator of transcription protein (STAT) pathways in mouse microglial cells [240]. miR-125a regulated macrophage activation and inflammation through TLR2 and TLR4 in mouse bone marrow cells [241]. Inhibition of the expression of the miR-15a/16 cluster decreased the expression of inflammatory cytokines, IL-1β and TNFα, and increased the expression of G-protein-coupled receptor kinase 2 (GRK2), in the spinal cords of chronic constriction injury rats [242]. miR-30c targeted inflammatory mediators, including intercellular adhesion molecule 1 (ICAM-1), IL-1β, and TNFα, in spinal cord injury (rats) [243]. miR-101 induced inflammation in lipopolysaccharide-activated microglial cells in rats via targeting IL-1β, IL-6, and TNFα, and inhibiting mitogen-activated protein kinase phosphatase 1 [244]. Downregulation of miR-148a was linked to microglial activation in acute cerebral ischemic stroke via modulating the secretion of inflammatory mediators such as TNFα, IL-1β, and IL-1 in mouse microglial cells [245]. miR-181 was implicated in inflammatory responses of astrocytes, and its aberrant expression in mouse brain cells enhanced the production of pro-inflammatory (TNFα, IL-6, IL-1β, IL-8) and anti-inflammatory (IL-10) cytokines [246]. In mouse primary astrocytes, miR-7 targeted the endoplasmic reticulum (ER) stress protein-HERP2, which mediated ER stress-induced inflammation [247]. miR-137 was found to prevent inflammatory response and cognitive impairment via inhibiting Src-dependent MAPK signaling pathway in brain tissues of mice [248]. Lower levels of miR-143 in dorsal root ganglion neurons due to inflammation were reported in mice [249]. miR-34a along with miR-155 and miR-145 promoted inflammatory response by suppressing c-Maf in microglia from mice [250]. miR-23b modulated neuroinflammation by targeting inositol polyphosphate multikinase (IPMK) in rat BV2 microglial cells and HT22 neuronal cell lines [251]. Increased expression of miR-139 impaired inflammation via targeting eukaryotic translation initiation factor 3 subunit K (EIF3K) in serum of rats with hypoxic-ischemic brain damage [252]. In MPP+-induced Parkinson’s disease mouse model, downregulated miR-205-5p via targeting LncRNA metastasis-associated lung adenocarcinoma transcript 1 (MALAT1) and lncRNA HOX transcript antisense intergenic RNA (HOTAIR) activated neuroinflammation in parkinsonism [253].

In vitro studies suggested that activated microglia induced miR-21 regulating IL6/STAT3 pathway in human models [254]. Lastly, miR-196a stimulated inflammation by regulating NF-κB signaling via annexin A1 (ANXA1) in endothelial cells or c-myc in human breast cancer cells [255].

Mechanistic cross-over between sAD and environmental neurotoxicity

Based on our hypothesis, sAD and environmental neurotoxicity may share common dysregulated processes, which can be controlled by common miRs. Using existing data on the potential function of the selected miRs implicated in AD pathogenesis and/or environmental neurotoxicity, we linked the retrieved miRs to the possible affected starting point or molecular/cellular KEs described in the tau-driven AOP [11]. Taking together, the AD-related miRs with their possible implication in neurotoxicity (human and/or animal), linked to the hypothetical starting point and/or KEs of the proposed AOP toward memory loss, are shown in Fig. 2.

OPEN IN VIEWER

Fig. 1

Overlap between miRs implicated in AD and in environmental neurotoxicity (NT), assigned as human and animal NT, respectively (Venn’s diagrams) [298].

OPEN IN VIEWER

Fig. 2

miRs implicated in AD (and/or neurotoxicity) plugged into the proposed tau-driven AOP toward memory loss. Underlined miRs are implicated in the regulation of more than one hypothetical starting point or key event of the plausible AOP. The relevance of these miRs to human neurotoxicity is indicated in blue color, to animal neurotoxicity in green color, and to both human and animal neurotoxicity in red. While miRs only linked to human AD are indicated in black color.

Functional miR analysis

The complete list of targets of the selected miRs is reported in Supplementary Table 2A and includes human miR–target interactions validated at experimental level retrieved by MiRTarBase. Functional analysis of the identified genes was performed considering the targets of the miRs involved in the proposed starting point and KEs as separate gene sets. PathDip outputs are available in Supplementary Table 2B. Biological relevant pathways were filtered, to finally select those genes that are proven to play a role in each specific starting point or KE. Figure 3 reports the list of curated sources containing the selected pathways in the case of glucose and cholesterol dysmetabolism, described as the hypothetical starting point of the AOP. REACTOME, KEGG, and WikiPathways are the most represented. Figure 4 reports the complexity of human miR–target interactions pathways for the hypothetical starting point. For representative purpose, only the hypothetical starting point is reported. Sankey diagram is a visualization technique that allows representing flows. It depicts how tens of miRs and hundreds of genes are potentially involved in few events. The number of interactions and their intersections enormously increases looking jointly at the starting point and the KEs. Table 2 summarizes only the number of unique biological relevant genes and miRs for each event or component of the proposed AOP.

OPEN IN VIEWER

Fig. 3

Sources reporting biologically relevant pathways for glucose dysmetabolism (A) and cholesterol dysmetabolism (B). The pie chart also describes the contribution given by each source.

OPEN IN VIEWER

Fig. 4

The Sankey diagrams represent the miR–target interactions involving genes that are proven to be involved in the pathways of interest (cholesterol & glucose (dys)metabolism). SP-Cdys refers to starting point-cholesterol dysmetabolism and SP-Gdys refers to starting point-glucose dysmetabolism.

OPEN IN VIEWER

Fig. 5

Networks describe respectively: the link between annotated genes and glucose metabolism related pathway (A), the interaction between miRs and selected gene targets (B), merged interactions (C).

OPEN IN VIEWER

Fig. 6

Networks describe respectively: the link between annotated genes and cholesterol metabolism related pathway (A), the interaction between miRs and selected gene targets (B), merged interactions (C). aNT indicates “human neurotoxicity” and hNT “human neurotoxicity”.

Starting point (SP): glucose and cholesterol dysmetabolism

DipPath revealed 86 pathways related to “Glucose metabolism” regulated by 305 unique genes of which 12 unique miRs were potentially targeted based on their MTIs retrieved by MiRTarBase. Out of these 12 miRs, miR-29b(-3/5p), miR-98(-3/5p), and miR-128(-3/5p) were in overlap with literature search for their implication in AD and in brain glucose metabolism. miR-98-5p and miR-128-3p potentially target IRS2 and MTOR, respectively, while both miR-98-5p and miR-128-3p bind to IGF1 and MAPK6, among other genes. With regards to their link to neurotoxicity, miR-98 and miR-128 are involved in both human and animal neurotoxicity, while miR-29b only in animal neurotoxicity. Further, miR-302(a) has been shown in literature and by functional analysis to be involved in glucose metabolism; however, its relevance to human AD is not clear. While for the AD-related miR-29a(-3/5p) (animal neurotoxicity) and miR-126(-3/5p) (animal neurotoxicity), and for miR-3666 (unknown link to AD and neurotoxicity), which have been identified to be involved in glucose metabolism by this analysis, no data on their link to brain glucose metabolism has been earlier reported.

Similarly, 19 pathways were identified to be involved in “Cholesterol metabolism”, of which 94 unique genes were found to be potentially targeted by 22 unique miRs. Among them, miR-33a(-5p), miR-106a(-3/5p), miR-106b(-3/5p), miR-128(-2-5p), miR-181c(-3/5p), miR-335(-5p), and miR-545(-3p) were also confirmed by literature for their link to both AD and brain cholesterol metabolism. Among their potentially targeted genes, LDLR was controlled by miR-106a-3p, miR-106b-5p, and miR-335-5p, ABCA1 by miR-106a-3p and miR-106b-5p, and SREBF1/2 by miR-335-5p. In addition, miR-106b and miR-335 have been associated with animal neurotoxicity, while miR-128 with both human and animal neurotoxicity. Functional miR analysis revealed miR-1(-3/5p) (human neurotoxicity), miR-27a(-3/5p) (human neurotoxicity), miR-27b(-3p) (human and animal neurotoxicity), miR-30c(-5/3p) (human and animal neurotoxicity), miR-148a(-3/5p), miR-206 (animal neurotoxicity), miR-613 and miR-1229(-3p) to be involved in cholesterol metabolism. However, there is no evidence for the relevance of miR-1, miR-613 and miR-1229 to AD, while for the AD-related miR-27a, miR-27b, miR-30c, miR-148a and miR-206, there is no clear link to brain cholesterol metabolism.

Key events (KEs)

As reported in Fig. 7, some miRs could potentially target several genes with biological relevance to one of the KEs. The functional analysis revealed that most of the identified biological genes (n = 745) were relevant to processes regulating synaptic function.

OPEN IN VIEWER

Fig. 7

Each key event occupies a portion of the diagram proportional to the number of biological relevant genes retrieved by the KE related gene set. The most external section of the diagram reports the miRs regulating these genes, according to miRTarBase.

KE1: Mitochondrial dysfunction

Fifty-six pathways were found related to “mitochondrial function”, and 25 unique miRs associated with these pathways are known to potentially interact with the 346 identified unique genes. Mitochondrial function-related miRs, including miR-7(-3/5p), miR-23b(-3p), miR-27a(-3/5p), miR-34a(-3/5p), miR-34c(-3/5p), miR-103a(-3p), miR-124(-3p), miR-195(-3/5p), miR-210(-3/5p), and miR-338(-5p) were commonly identified by literature and functional analysis. miR-7-3/5p, miR-23b-3p, miR-27a-5p, miR-34a-3p, and miR-210-3p possibly interact with ATP-related genes, such as ATP11 C, ATP7B, ATP7A, ATP1B3, ATP1A1, ATP8B2/3, and ATP8A2, among other miR targets. Besides their link to AD, miR-34a and miR-210 have been shown to be involved in animal neurotoxicity, while miR-23b, miR-27a, miR-34c, miR-103a, miR-124, and miR-195 in human (and/or animal) neurotoxicity. This analysis also revealed the implication of AD-related miR-143 and miR-183 and miR-330 (link to AD is currently unknown) in mitochondrial function, however, no evident data were found in literature for brain cells.

KE2: Oxidative stress

After filtering out the relevant pathways to “oxidative stress”, 51 pathways were obtained, regulated by 218 unique genes which are possibly targeted by 31 unique miRs. miR-9(-3/5p), miR-23b(-3p), miR-29b(-3/5p), miR-34c(-3/5p), miR-125b(-3/5p), miR-128(-3/5p), miR-132(-3/5p), miR-142(-3/5p), miR-200a(-3/5p), miR-200b(-3/5p), miR-200c(-3/5p), miR-210(-3/5p), miR-483(-5p), and miR-486(-5p) were indicated by literature, as well. Among their regulated genes, MAPK-related genes such as MAPK14, MAPK1, and MAPK3, are targeted by miR-9-3p, miR-125b-5p, miR-128-3p, miR-132-3p, miR-142-5p, miR-200a-3p, or miR-483-5p, and E2F1/2/3 genes by miR-9-3p, miR-23b-3p, miR-34c-5p, miR-125b-5p, miR-128-3p, miR-200b-3p, miR-200c-3p, or miR-210-3p. Among these miRs, miR-9, miR-34c, miR-125b, and miR-128 are implicated in human and animal neurotoxicity, whereas miR-29b, miR-132, miR-200a, and miR-210 in animal neurotoxicity, and miR-23b, miR-142, and miR-200c in human neurotoxicity. Notably, the identified miR-153(-3/5p), miR-196a(-5p), and miR-330(-3/5p) by this analysis were not confirmed by literature for their relevance to human AD.

KE3: Hyperphosphorylated tau

There were found 32 pathways linked to “hyperphosphorylated tau”, with 604 unique genes potentially targeted by 43 unique miRs. Commonly found miRs between functional analysis and literature, let-7b(3/5p), miR-9(-3/5p), miR-15a(-3/5p), miR-15b(-3/5p), miR-26a(-3/5p), miR-26b(-3/5p), miR-34a(-3/5p), miR-34c(-3/5p), miR-93(-3/5p), miR-98(-3/5p), miR-101(-3/5p), miR-103a(-3p), miR-106b(-3/5p), miR-124(-3p), miR-125b(-3/5p), miR-128(-3/5p), miR-132(-3/5p), miR-138(-3/5p), miR-146a(-3/5p), miR-200c(-3/5p), miR-219a(-3/5p), and miR-326. PTEN, CDK6, and/or PIK3R1 were targeted by let-7b-5p, miR-26a-3/5p, miR-34a-3/5p, miR-93-5p, miR-101-5p, miR-103a-3p, miR-106b-3/5p, miR-124-3p, and miR-128-3p. While GSK3B by let-7b-3p, miR-9-5p, miR-26a-3p, and miR-98-3p. Only miR-21(-3/5p), which is involved in AD and (human and animal) neurotoxicity, was not directly linked to KE3 in brain cells. Out of the overlapped KE3- and AD-related miRs, let-7b, miR-9, miR-34c, miR-98, miR-124, miR-125b, miR-128, and miR-146a are also involved in neurotoxicity (human and animal). While miR-26a, miR-34a, miR-106b, miR-132, miR-138, and miR-219 are implicated in animal neurotoxicity, and miR-103a and miR-200c in human neurotoxicity.

KE4: Dysfunctional autophagy

DipPath revealed 16 pathways related to “autophagy”, of which 441 unique genes were identified to be possibly targeted by 37 unique miRs. Among them, miR-9(-3/5p), miR-17(-3p), miR-20a(-5p), miR-23b(-3p), miR-27a(-3/5p), miR-27b(-3/5p), miR-34a(-3/5p), miR-93(-3/5p), miR-100(-3/5p), miR-103a(-3p), miR-107, miR-124(-3p), miR-128(-3/5p), miR-132(-3/5p), miR-181b(-3/5p), miR-193b(-3/5p), miR-199a(-5p), miR-214(-3/5p), miR-299(-5p), and miR-590(-3/5p) were shown to be autophagy-related by literature, as well. miR-93-5p, miR-181b-5p, and miR-214-5p potentially target ULK1 or ULK2, while miR-128-3p and miR-181b-5p bind to MTOR. Moreover, miR-183 and miR-496, both involved in animal neurotoxicity, were found to be related to KE4 but they were not confirmed by literature for their relevance either to autophagy in brain cells or to human AD.

KE7: Synaptic dysfunction

There were found 77 unique miRs possibly targeting the 476 unique genes that regulate the identified 42 relevant pathways to “synaptic function”. Most of them, including let-7d(-5p), miR-9(-3/5p), miR-15a(-3/5p), miR-16(-3/5p), miR-19b(-3/5p), miR-23a(-3/5p), miR-26a(-3/5p), miR-26b(-3/5p), miR-29a(-3/5p), miR-29b(-3/5p), miR-30a(-3/5p), miR-30b(-3/5p), miR-34a(-3/5p), miR-34b(-3/5p), miR-124(-3p), miR-125a(-3/5p), miR-125b(-3/5p), miR-130b(-3p), miR-132(-3/5p), miR-134(-3/5p), miR-135a(-3/5p), miR-138(-3/5p), miR-148a(-3/5p), miR-151a(-3/5p), miR-182(-5p), miR-188(-3/5p), miR-191(-5p), miR-193a(-3p), miR-195(-3/5p), miR-204(-3/5p), miR-206, miR-210(-3p), miR-212(-3/5p), miR-223(-3/5p), miR-301a(-3p), miR-326, miR-374a(-3/5p), miR-384, miR-485(-5p), and miR-501(-3p), were also indicated as KE7-related by literature. Among their targets, BDNF is possibly regulated by miR-30a-5p, miR-124-3p, miR-132-3p, miR-206, and miR-210-3p, PTEN by miR-23a-3p, miR-26a-3p, miR-29a-3p, miR-29b-3p, and miR-34a-3p, and CREB-related genes (CREB1/5, CREB3L2) by miR-9-5p, miR-124-3p, miR-125a-5p, miR-206, and miR-374a-5p. In addition, for the identified KE7-related let-7c, miR-1, miR-19a, miR-376b, and miR-574, there are no data on their link to human AD.

KE8: Neuroinflammation

Pathway analysis showed 62 pathways linked to “neuroinflammation”, which are regulated by 745 unique genes, which are potentially targeted by 47 unique miRs. Out of these identified miRs, let-7a(-5p), let-7b(-3/5p), let-7g(-5p), miR-7(-3/5p), miR-9(-3/5p), miR-21(-5p), miR-23b(-3p), miR-27b(-3/5p), miR-29a(-3/5p), miR-29b(-3/5p), miR-30c(-3/5p), miR-34a(-3/5p), miR-101(-3/5p), miR-107, miR-124(-3p), miR-125a(-5p), miR-125b(-3/5p), miR-132(-3/5p), miR-139(-3/5p), miR-142(-3/5p), miR-143(-3/5p), miR-146a(-3/5p), miR-148a(-3/5p), miR-181c(-3/5p), and miR-205(-3p) were also confirmed by literature for their relevance to KE8 in brain cells. Let-7b-3p, miR-9-3/5p, miR-34a-5p, miR-124-3p, and miR-146a-5p potentially bind to NF-κB-related genes. Moreover, for miR-196a no clear evidence for its link to human AD has been described earlier.

Considering all miRs with overlap between functional analysis and literature, we further checked for their potential targeted genes across the KEs of the proposed AOP. Forty miRs were identified, whose common targets were found in at least two different events of the proposed AOP (Fig. 8). In these 40 miRs, for some mature miRs, both 3p- and 5p-arm sequences are included. More details are provided in Supplementary Table 3.

OPEN IN VIEWER

Fig. 8

The heatmap reports 40 miRs, whose common targets (in yellow) affect multiple events (hypothetical starting point and/or KEs) of the proposed AOP (in blue).