MicroRNAs in Alzheimer’s Disease

1Alzheimer’s disease (AD) is an irreversible progressive neurodegenerative disorder, and is the main cause of dementia in the global elderly population. In 2010, it was estimated that there were 35.6 million people living with dementia across the globe; these numbers are expected to double every 20 years until 2050. The characteristics of AD are cognitive impairment and memory loss, causing most AD patients to lose the ability to perform daily activities independently. The cost of caring for these patients is rising, with increasing numbers of AD incidents around the world, posing heavy burdens for individuals, families and society. Although mounting efforts have been made to research the etiology of AD, the causes remain unknown, and effective treatments are not yet available. The pathological hallmarks of AD are senile plaques consisting of accumulated β-amyloid peptides (Aβ) and neurofibrillary tangles (NFTs) primarily containing highly phosphorylated Tau. The two most commonly accepted hypotheses, the Aβ hypothesis and the Tau hypothesis, are also based on these two pathological characteristics. The amyloid cascade hypothesis (see Figure 1) suggests that the imbalance between the production and clearance of Aβ is the key trigger of a cascade of events that leads to AD. The Aβs are produced from processing the amyloid precursor protein through sequential enzymes digested by β-secretase (beta-site amyloid precursor protein cleaving enzyme1, BACE1) and γ-secretase (consisting of presenilin1 and presenilin2). In physiology, Aβ-degrading proteases mediate proteolytic degradation and receptor-modulated endocytosis function to clear the Aβ. The proteolytic system includes neprilysin and an insulin-degrading enzyme; the low density lipoprotein receptor family is involved in the receptor system. The most toxic forms of Aβ are soluble oligomers and intermediate amyloids, which can cause synaptic loss, neurotoxicity, neuron apoptosis, inflammation and mitochondrion dysfunction. Tau are microtubule-associated proteins which stabilize the microtubule and promote vesicle transportation. In neurons, the microtubules are essential for the maintenance of neuronal structure, axonal transportation and neuronal plasticity. Highly phosphorylated Tau can lose its stabilization ability and may start to self-form NFTs. Tau are phosphorylated by types of kinases, and the phosphate residues are removed by phosphatases. The imbalance between the hyper-phosphorylated and de-phosphorylated forms of Tau could lead to the formation of NFTs (see Figure 2). Tauopathies are considered to be an indicator of the severity of AD.

2MicroRNAs, a class of non-coding RNAs, have been acknowledged as important regulators for post-transcriptional gene expression by either repressing translation or degrading target mRNAs. Since their discovery, microRNAs have been identified as the regulators most frequently implicated in many critical biological events, such as development, growth, differentiation and neurodegenerative processes. One microRNA could target numerous genes and one gene could be regulated by multiple microRNAs, making microRNAs a potential tool to investigate multifactorial diseases, for example AD. The study about examining the 13-brain associated microRNAs abundance in human hippocampus samples from fetal and aged adults, age-matched AD patients suggest that misregulation and alterations of specific microRNAs might make contributions to the process of AD. Afterward, tremendous researches have demonstrated the alterations of several microRNAs between AD patients and age-matched control, further prove that microRNAs might play an important role in the pathogenesis of AD. In this review, we will focus on the microRNAs involved in the two most accepted hypotheses of AD pathogenesis: the Aβ hypothesis and the Tau hypothesis.

3Since microRNAs have been used to research AD, accumulating microRNAs have been identified as regulators in the process of Aβ production and clearance. The amyloid cascade hypothesis mentions that the imbalance between the production and clearance of Aβ42 could trigger synaptic loss and neurotoxicity. Section “MicroRNAs Involved in the Aβ Hypothesis” describes the microRNAs implicated in the Aβ hypothesis (as shown in Figure 1 and Tables 1, 2).

4Mounting evidence has shown that specific microRNAs play a key role in regulating the expression of APP and BACE1, which restrict the production of Aβ.

5Accumulating evidence demonstrates that increased APP expression could promote Aβ production, resulting in neurotoxicity, synaptic failure and, eventually, dementia. Many microRNAs have participated in the process of regulating APP expression; for example, found that miR-106a/520c could bind the 3′UTR of APP and repress the expression of APP in human cell lines. Later, the Hebert group found that the miR-20 families (miR-20a, miR-17-5p, and miR-106b) could also regulate APP expression via the binding of APP 3′UTR. In addition, in accordance with miR-20 overexpression in neuron cells, the expression level of APP decreased. Similarly, miR-101, miR-16, and miR-153 have also been identified as negative regulators of APP expression both in vitro and in vivo. Moreover, the low level of miR-153 contributed to the accumulation of Aβ in sporadic AD patients.

6The division of APP by BACE1 is the first and rate-limit step for Aβ formation, and upregulated BACE1 expression levels and enzymatic activities have been detected in sporadic AD brains. Several microRNAs regulating BACE1 expression and activity have been found; for example, the miR-29 family implicated in regulating BACE1 expression has been well studied. performed the microRNAs expression profiles of sporadic AD patients, and found that the expression levels of the miR29a/b1 cluster decreased significantly with high levels of BACE1 expression in sporadic AD brains. Further analysis showed that miR-29a/b-1 could regulate BACE1 expression in vitro and in vivo. Moreover, it was validated that decreased levels of miR-29a/b-1 could promote the production of Aβ and contribute to the pathogenesis of AD. Afterwards, miR-29c, another miR-29 family member, was also found to be downregulated with abnormally high levels of BACE1 in sporadic AD brains; additional validation experiments found that overexpressed miR-29c could induce BACE1 downregulation in human neuroblastoma SH-SY5Y cells via the binding of BACE1 3′UTR.

7Another well studied microRNA is miR-107. The expression of miR-107 decreased significantly with BACE1 increased in AD patients. Then, it was revealed that miR-107 regulated the expression of BACE1 through recognizing and binding the 3′UTR of BACE1 in a cell culture reporter assay. Afterwards, Nelson and Wang also demonstrated that miR-107 levels negatively correlated with BACE1 mRNA levels, leading to Aβ accumulation. Moreover, some studies have shown that miR-107 was implicated in preventing the neurotoxicity and blood–brain-barrier dysfunction induced by Aβ, making miR-107 a potential drug target for researchers.

8Other negative regulators of BACE1 also include miR-298/328, microRNA-195, miR-135a, microRNA-135b, miR-339-5p, and miR-186, all of which negatively correlated with BACE1 expression and exerted regulatory effects by binding the 3’UTR of BACE1.

9Recent cases have reported that the miR-132/212 cluster plays an important role in Aβ production; for example, found that levels of microRNA-132/212 cluster decreased in AD patients, and the deficiency of microRNA-132/212 could promote Aβ production and senile plaque deposition in AD triple transgenic mice. In another study, also confirmed that the loss of miR-132 could promote β amyloid accumulation through regulating inositol 1,4,5-trisphosphate 3-kinase B-ITPKB in AD mouse models. found that miR-98-5p could repress the expression of sorting nexin6 (SNX6), which is involved in increasing the levels of Aβ40, Aβ42, BACE1, soluble amyloid precursor protein β (sAPPβ), and membrane-associated APP β-carboxyl terminal fragments in 293T and SK-N-SH cells. Other microRNAs implicated in Aβ production include miR-34a, miR-146, microRNA-125b, miR-330, miR-24, miR-186, and miR-455. All these provide further evidence that specific microRNAs play a critical role in the pathogenesis of AD.

10Aβ deposits result from an imbalance between the production and clearance of β amyloid peptides. Thus, the dysfunctional clearance of Aβ could also contribute to the accumulation of Aβ peptides. As such, in Section “MicroRNAs Involved in Aβ Clearance” we describe a number of microRNAs involved in Aβ clearance.

11The endosomal–lysosomal system degrades accumulated proteins and functions as a protective factor in the central nervous system. found that upregulated miR-128 could impair the clearance of Aβ through the targeting of lysosomal system enzymes in monocytes in sporadic AD patients. Aβ degradation improved when miR-128 was inhibited in monocytes from AD patients. Another microRNA involved in Aβ clearance was miR-34a which, when upregulated in AD patients, repressed the clearance of Aβ by inhibiting the expression of TREM2. TREM2, a myeloid/microglial cell surface amyloid sensor-receptor implicated in recognizing and digesting Aβ and extracellular amyloidogenic debris, plays a key role in the clearance of over-expressed Aβ. The expression of TREM2 also correlated with blood-derived monocytes (BDMs), which have been implicated in the clearance of β amyloid deposits in AD brains. Immune-related microRNAs (miR-155,-154,-200b,-27b, and 128) which, when expressed differentially in CCL2/CCR2 (chemokine/chemokine receptor) axis impaired BDMs, reportedly participate in the process of Aβ clearance mediated by BDMs. Additionally, several researchers have revealed that differentially expressed apoE isoforms also play a role in Aβ clearance. It has been validated that miR-1908 could inhibit ApoE expression in mRNA and protein levels in human macrophage cell line THP-1 and astrocytoma cell line U87. In AD patients, the level of miR-1908 was negatively related to ApoE expression, which suggested that miR-1908 played a crucial role in inhibiting Aβ clearance through repressing ApoE expression.

12Aβ could induce neurotoxicity, and this section describes several microRNAs involved in this process. In 2016, it was discovered that miR-302 could restore Aβ induced neurotoxicity via the PTEN/Akt/Nrf2/HO-1 pathway in AD neurons. found that miR-137 could reduce Aβ induced neurotoxicity by inhibiting the process of NF-κB via repressing TNFAIP1 expression in N2a cells. It has been reported by many groups that miR-132 decreased significantly in AD brains. recently discovered that the overexpression of miR-132 in cultured cortical neurons could inhibit the neurotoxicity induced by Aβ via the miR-132/PTEN/AKT/FOXO3a pathway. In the same year, found that miR-107 could block Aβ induced neurotoxicity in mice; they injected miR-107 mimics, which could intraventricularly generate native miR-107, into mice and found that miR-107 could restore spatial memory impairment and pyramidal neurons loss caused by Aβ. It was reported that MiR-124 could alleviate neurotoxicity via regulating the expression of BACE-1 in AD cellular models. These microRNAs provide us with a new way to understand the pathogenesis mechanisms of AD.

13Aβ could lead to the synaptic failure and numerous microRNAs involved in this process. It was reported that the level of miR-34c increased significantly in the transgenic mice model, AD patients and Aβ exposed neuron cells. Further investigation demonstrated that inhibiting the expression of miR-34c could rescue the synaptic dysfunction and memory loss induced by Aβ by up-regulating the expression of VAMP2. Additionally, identified that miR-34a, another miR-34 family member, could target multiple genes involved in neuronal synaptic deficiency and energy metabolism. Moreover, further research found that the loss of miR-34a could improve cognitive functions through rescuing synaptic dysfunction by repressing the expression of AMPA and NMDA receptors.

14found that miR-188-5p could restore the synaptic dysfunction and cognitive impairment caused by Aβ in transgenic mice; they also observed that the overexpression of miR-188-5p could alleviate the decrease of dendritic spine density in primary hippocampal neurons exposed to Aβ. Additionally, found that miR-10a was also a negative regulator in synapse remodeling and cell proliferation as a result of a reduction in BDNF-TrkB signals in AD rats. They confirmed that BDNF was a target of miR-10a using a dual luciferase gene reporter assay.

15Recently, a novel pathway, the miR-124/PTPN1 pathway, has been reported to involve in synaptic transmission deficits. The level of miR-124 increased significantly with the expression of its target, PETN1, decreased dramatically in AD brains. The overexpression of miR-124 or the knockdown of PTPN1 in mice could induce AD-like phenotypes, containing synaptic transmissions and plasticity deficits. Moreover, the inhibition of miR-124 expression or the over-expression of PTPN1 could alleviate the synaptic deficits in AD model mice.

16A noticeable thing is that Aβ could induce the synaptic loss, changes in neurotrophin and inflammations, conversely, these deficits could affect the clearance and degradation of Aβ. NF-κB, a pro-inflammatory transcription factor, could be activated by Aβ, and the activation of NF-κB could lead to the upregulated of six inducible miRNAs: miR-7, miR-9, miR-34a, miR-125b, miR-146a, and miR-155. Those NF-κB sensitive upregulated miRNAs in AD brain contributed much to the neuropathological characteristics. For example, upregulated miR-125b may impair the Aβ clearance via repressing the TREM2 expression, induce chronic innate-immune and inflammatory signaling by inhibiting the expression of CFH (complement factor H), repress neurotransmitter package and release, and decrease the neurotrophin within the brain through regulating the expression of SYN-2, 15-LOX, and VDR (synapsin-2, 15-lipoxygenase and vitamin D receptor). Similarly, miR-146a may be responsible for promoting amyloidogenesis through regulating the expression TSPAN12 (tetraspanin12) and/or failing to modulate the NF-κB through regulating IRAK-1 and IRAK-2 (interleukin-1 receptor associated kinase).

17Neurofibrillary tangles were largely composed of hyper-phosphorylated Tau as a result of the imbalance between Tau phosphorylation and de-phosphorylation through regulating the expressions and activities of a number of related kinases and phosphatases. In Section “MicroRNAs Involved in the Tau Phosphorylated Imbalance,” we describe several microRNAs involved in the process of phosphorylation and de-phosphorylation (as shown in Figure 2 and Tables 1, 2).

18It is reported that the expression levels of miR-125b are increased significantly in AD patients. In both neuron cells and in mice, over-expression of miR-125b induced Tau hyper-phosphorylation through the targeting of phosphatases DUSP6 and PPP1CA, while inhibition of miR-125b induced a reduction in Tau phosphorylation and kinase expression/activity. Moreover, found that overexpression of miR-125b could induce cell apoptosis and Tau hyper-phosphorylation by activating the cyclin-dependent kinase 5 (CDK5) and p35/25 in neuron cells. MiR-125b might be implicated in this process through targeting the forkhead box Q1(FOXQ1) directly.

19found that the expression of Tau could be directly regulated by miR-132, and the deletion of miR-132/212 in mice could increase the expression, phosphorylation and accumulation of Tau. They also revealed that the delivery of miR-132 mimics into AD mice could reduce the metabolism of Tau. also reported that the downregulation of miR-132/212 could promote Tau phosphorylation and break the imbalance between Tau phosphorylation and de-phosphorylation via the NOS1-dependent pathway in primary human neurons and neural cells. miR-132 has been recognized in numerous studies as one of the most significant downregulated microRNAs, and it has been implicated in the process of Tau hyper-phosphorylation through the targeting of EP300, GSK3b, Rbfox1, proteases Calpain2 and caspases 3/7.

20The expression level of miR-138 is up-regulated in AD patients. found that miR-138 could directly target the retinoic acid receptor alpha which could repress glycogen synthase kinase-3b (GSK-3β) activity. They further confirmed that miR-138 could promote Tau phosphorylation via the RARA/GSK-3β pathway in N2a/APP and HEK293/tau cells. The level of miR-26b also increased significantly in the pathological areas of postmortem AD brains. It was confirmed that the elevated levels of miR-26b contributed to the pathology of AD through cell cycle entry, Tau hyper-phosphorylation and apoptosis in post-mitotic neurons.

21MicroRNA-922 was also reported to contribute to the pathogenesis of AD via regulating the expressions of UCHL1. The levels of UCHL1 were decreased in sporadic AD patients, and the overexpression of UCHL1 rescued the synaptic and cognitive function in AD model mice. And in the AD model mice, the expression levels of miR-922 were upregulated with UCHL1 protein levels decreased. found that UCHL1 was one target of miR-922, and the phosphorylation levels of Tau were negatively correlated with the expressions of UCHL1. Thus, they suggested that miR-922 involved in the pathogenesis of AD through regulating the phosphorylation levels of Tau by targeting UCHL1.

22The expression levels of microRNA-146a was increased in AD patients’ brains, and the dysregulation of microRNA-146a biogenesis was involved in the tau hyperphosphorylation and AD pathogenesis. In neural cells, it was confirmed that miR-146a targeted the coiled-coil containing protein kinase1 (ROCK1) directly, and the inhibition of ROCK1 could induce abnormal tau phosphorylation, which was related with the low phosphorylation levels of the phosphatase and tensin homolog (PTEN). Additionally, the ROCK1 colocalised with the hyperphosphorylated tau in the early neurofibrillary tangles with decreased ROCK1 protein levels in AD patients. Moreover, in 5xFAD mice, the inhibition of miR-146a through intra-hippocampal delivery induced upregulated ROCK1 protein levels and repressed tau hyperphosphorylation, and restored memory function partly.

23Similarly, the expression level of miR-106b was downregulated with the upregulation of Fyn in the temporal cortex of AD patients. In tau stably expressed SH-SY5Y cells (SH-SY5Y/tau), the overexpression of miR-106b inhibited Aβ42 induced tau phosphorylation at Tyr18 without changes at Ser396/404. It was validated that miR-106b could target Fyn directly and repress the protein levels of Fyn in SH-SY5Y cells. The phosphorylation level of tau at Tyr18 was decreased when Fyn was knockdown, and the inhibitory effects could be rescued when the expression of miR-106b was inhibited. Thus, miR-106b might inhibit Aβ42 induced tau phosphorylation at Tyr18 via regulating the expression of Fyn.

24The expression levels of miR-124b was decreased in the brain of AD patients and in N2a/APP695swe cells. The overexpression of miR-124-3p could rescue cell apoptosis and attenuate the abnormal hyperphosphorylation of tau with the upregulation of Caveolin-1, phosphoinositide 3-kinase (PI3K), phosphor-Akt (Akt-Ser473)/Akt, phosphor-glycogen synthase kinase-3 beta (GSK-3β-Ser9)/GSK-3β in N2a/APP695swe cells. It was suggested that miR-124-3p could inhibit abnormal hyperphosphorylation of tau through targeting Caveolin-1 and regulating the pathway Caveolin-1-PI3K/Akt/GSK-3β.

25Notably, the pathological functions of multiple miRNAs regulated genes in AD-affected brain are overlapped and highly interactive, and the effects of interactive and overlapped functions network are greater than the individual one. Another noticeable problem is that specific brain enriched miRNAs are relatively labile and short-lived, they may be subject to degradation especially in the highly oxidative environment of an AD brain. Thus, down-regulated microRNAs may be an artifact of the degenerative aspects of the disease. While the up-regulated miRNAs and their down-regulated mRNAs targets are a better bet to study especially in post-mortem tissues.

26In 2018, the National Institute of Aging and Alzheimer’s Association proposed that “Alzheimer’s Disease” was a process of accumulated neuropathological changes, then defined as “AD” by biomarkers in vivo or by postmortem examination, not clinical symptoms; thus, the search for reliable, effective and timely biomarkers would be of significant value. The most commonly used biomarkers include CSF (cerebrospinal fluid) Aβ42 or Aβ42/Aβ40 ratio, amyloid PET, CSF phosphorylated Tau, CSF total Tau, Tau PET, anatomic MRI and FDG PET, which all focus on Aβ and Tau in bio fluids and neuroimaging techniques. However, none of them could individually diagnose AD, and these methods still arrived too late for early and effective intervention. In recent decades, many studies have shown that microRNAs are implicated in the pathogenesis of AD (described in sections “MicroRNAs Involved in the Aβ Hypothesis” and “MicroRNAs Involved in the Tau Phosphorylated Imbalance” and summarized in Tables 1, 2), and alterations of microRNAs were found in serum, plasma and CSF, meaning that microRNAs are exceptional candidates for AD biomarkers. Most microRNA biomarker lists were produced by conducting microRNA profiles through qPCR, some of which are even next generation sequences, and comparing differentially expressed microRNAs between AD patients and control groups. Although mounting microRNAs have been identified as potential biomarkers for AD diagnosis, few have been validated in more than two studies, making their widespread use difficult. It was proved that the expression levels of miR-15, miR-181c, miR-125b in AD CSF were up/down-regulated compared with normal controls in more than two studies, and so these specific microRNAs were suggested to be potential biomarkers for AD. Despite this, extensive validation and follow-ups in larger cohorts of patients are still necessary.

27There are still no drugs that are both safe and effective for AD use, even though multiple agents have been researched and tested. Studies about microRNAs in AD have provided prominent insights into our understanding of molecular mechanisms, shedding light on potential drugs by targeting specific microRNAs. The regulatory characteristics of microRNAs are sequence-specific and multiple genes regulation, providing an exciting avenue for regulating this complicated disease networks and pathways. Approximately 500 patents were issued and published in United States and European in 2016, suggesting that the therapeutic potentials of microRNAs have attracted a great deal of attention. A number of preclinical studies also have been performed, for example, performed a preclinical evaluation and suggested that miR-16 was a good drug candidate for AD. They identified that miR-16 could inhibit the expression of APP and BACE1, repress the production of Aβ and the phosphorylation of Tau in cells. Additionally, the delivery of miR-16 into mice induced a decrease in APP, BACE1, and Tau levels in a region dependent method. Moreover, miR-16 has also been found to associate with oxidative stress and inflammation in AD. This successful preclinical study provided support for future microRNA therapy researches. It is important to note that the off target effect of microRNAs should not be ignored at the same time. Although studies about microRNA therapies in AD are still in their infancy, with increasing attention being paid to this field, we believe that researchers will fully understand this at some point, even if there is still a long way to go.

28MW wrote the manuscript. LQ edited it and BT edited the final version of the manuscript.

29The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.