Abstract
Alzheimer’s disease (AD) is a debilitating neurodegenerative disease that affects the cognitive faculties of millions of people worldwide. There is still no known cure for AD, nor a clear understanding of AD etiology. Nevertheless, researchers have made significant strides in understanding various key aspects of AD neuropathology at the cellular and molecular levels. This review is intended to provide a general survey of what is known and unknown, based on the three hallmarks of AD, combined with our knowledge from microRNA research. Our goal is to reevaluate and reassess the current direction of AD research and therapeutic insights, charting a new course and comprehensive plan to combat this imminent global health threat.
Keywords
INTRODUCTION
Alzheimer’s disease (AD) is a neurodegenerative disease and the most common form of dementia that leads to the impairment of cognitive faculties such as learning, memory, language, and behavior [1]. In the US, the official CDC tally of AD ranks the disease as the sixth leading cause of death. However, by including dementia-related conditions into the tally, it is moved up to the third leading cause of death in the US behind heart disease and cancer [2]. With rapid advances in biotechnology for better treatment of cardiovascular and oncologic diseases, increased global life expectancy and population aging, it is anticipated that AD may become a prominent global health threat. The life expectancy for AD patients varies between 3 to 10 years but the disease impact goes far beyond financial burdens of treatment as it places enormous psychological stress and emotional toll on a patient’s family and loved ones [3].
In 1906, Alois Alzheimer first described the pathological hallmarks of AD in the formation of amyloid plaques (Aβ), neurofibrillary tangles (NFT), and neuronal loss [4]. Since this initial report, early AD research focused primarily on understanding and eliminating the formation of Aβ plaques in AD patients, although clinical trial results fell short of anticipated outcomes [5]. More research effort now focuses on the second hallmark of AD, tauopathy, in the form of pathological aggregation of hyperphosphorylated tau proteins, which contributes to the formation of NFT [6]. While the effectiveness of treating AD by tackling NFT remains to be evaluated, it is still equivocal whether Aβ accumulation and NFT formation are directly causative or simply correlative to the etiology of AD. Indeed, transient buildup of Aβ plaques is found in short-term survivors of traumatic head injury and also in patients with brain ischemia [7, 8]. Chronic traumatic encephalopathy in professional boxers and football athletes, resulting from multiple head injuries and concussion, lead to the formation of tau-immunoreactive NFTs with neuropathological symptoms like memory disturbances and behavioral and personality changes very similar to AD [9]. Some studies have proposed that the pathological buildup of soluble Aβ oligomers in the early stage of AD, instead of formation of Aβ plaques in the late stage, causes defects in homeostatic synaptic plasticity, impairment of memory formation, and dysfunction in cognition, hence contributing to the onset of AD pathogenesis [10]. Clearly the verdict regarding the etiology of AD remains outstanding.
The goal of this review is to reexamine what we have learned so far, fine-tune key research directions, and incorporate the latest knowledge from other research fields in a joint effort to make further advances in AD research. In particular, of great interest to this group is to explore the potential regulatory function of non-coding RNA in AD neuropathology, especially in the formation of the three hallmarks of AD.
miRNA AND EXPRESSION REGULATION
microRNAs (miRNAs) are small non-coding RNAs about 18 to 22 nucleotides long. They are known to regulate gene expression post-transcriptionally by silencing mRNA translation into protein [11]. As demonstrated in Fig. 1, miRNA biogenesis can be produced through either regular transcription (intergenic miRNA) or post-transcriptional splicing introns (intronic miRNAs) [12, 13]. For the intergenic miRNAs, the biogenesis begins with RNA polymerase II transcription of a non-protein coding mRNA transcript, formation of a primary pri-miRNA transcript, a secondary process of stem-loop intermediates with DGCR8 and Drosha to make pre-miRNAs, and exportation with Exportin-5 and Ran-GTP into the cytoplasm. The pre-miRNA intermediate then interacts with the Dicer complex made up of Dicer, TRBP, and AGO to produce the mature miRNA. For the intronic miRNAs, the pathway starts with transcription of protein-coding mRNA. miRNA is typically derived from post-transcriptional spliced introns which are picked up by Drosha-DGCR8, treated as intermediate pre-miRNAs, and exported to the cytoplasm like intergenic miRNAs. Once exported into the cytoplasm, mature miRNA remains bound to AGO and forms the RNA-induced silencing complex (RISC). The RISC uses the mature miRNA sequence to bind to the complementary sequence found on mRNA, which, depending on the level of complementarity and relative position, can lead to either silencing or degradation of the target mRNA [14].
microRNA biogenesis.
Clearly, based on the two different biogenesis pathways, the expression of miRNA will be regulated differently. It is conceivable that intergenic miRNA expression will be regulated by a discrete signaling mechanism and transcription factors. Intronic miRNA expression, on the other hand, correlates with the expression of corresponding coding mRNA transcripts, which may be further differentially regulated by alternative exon-intron splicing variants [15]. miRNAs exert their function by binding to the specific binding sites located on the target mRNA transcripts and therefore regulate the downstream translational expression. The binding sites are often located in the 3’ untranslated region (3’ UTR) although there are also reports of binding sites in the 5’ UTR, introns, and exons of mRNA [16]. Furthermore, a perfect complementation of miRNA to its binding site can lead to mRNA degradation while imperfect complementation may lead to silencing only [17]. In this regard, the complexity and multifaceted nature of miRNA regulation suggests a more pleiotropic regulatory role in post-transcriptional regulation of protein expression [18].
Although the majority of miRNA is found within the cells where mRNA is located, some miRNA, known as cell-free circulating miRNA, are found to be highly stable in extracellular environments like blood plasma and in cell culture media, and are suggested to play a role in intercellular signaling and regulation [19]. This remarkable property of miRNA in cell-free bodily fluids led to great interest by many groups, including ours, to study its potential application as markers of diseases [20]. Clearly, dysfunction of miRNA regulation can be pathogenic, leading to the dysregulation of signaling pathways and molecular circuitry which may be critical for all aspects of AD, including clearing Aβ plaque formation, tau protein phosphorylation, and synaptic plasticity in neurogenesis [21, 22].
miRNA FUNCTIONAL ROLES IN AMYLOID-β PROTEIN PRECURSOR PROCESSING PATHWAY
Aβ plaques are formed by the aggregation of Aβ40 and Aβ42 peptides, the cleavage products of the amyloid-β protein precursor (AβPP) processing pathway [23, 24]. AβPP is a conserved type-1 transmembrane glycoprotein that is constitutively expressed in a wide variety of tissues, including tissues from the nervous system, immune system, musculoskeletal system, kidney, lung, pancreas, prostate gland, and thyroid gland [25]. Although the primary function of AβPP has not been fully unraveled, it has been extensively studied for its role as the precursor of the Aβ peptides. Indeed, the highest expression of AβPP is found in neuron and glial cells of the central nervous system (CNS) and is thought to be involved in some aspects of synaptogenesis, repair, neuronal adhesion, neuroprotection, and neural plasticity [25–28]. Normally, AβPP is highly expressed but also rapidly metabolized in neurons [29].
There are two recognized pathways for AβPP processing and cleavage [30–32]. As demonstrated in Fig. 2, in the amyloidogenic pathway, preferentially localized in the endosomal compartment, AβPP is initially cleaved by β-secretase, known as β-site APP cleaving enzyme 1 (BACE1), which yields a soluble N-terminal fragment called sAβPPβ and a membrane bound c-terminal fragment-β, CTFβ [33]. The CTFβ is subsequently cleaved by the γ-secretase complex to generate either 40 or 42 amino acid-long Aβ peptides, known as Aβ40 and Aβ42, which may aggregate and form the Aβ plaques. The c-terminal fragment, called AβPP intracellular domain (AICD), is further processed before internalized. In the non-amyloidogenic pathway, which preferentially occurs at the cell membrane and does not generate Aβ plaques, AβPP is cleaved by α-secretase to produce a longer, soluble N-terminal fragment called sAβPPα and a membrane bound C-terminal fragment-α, CTFα. CTFα is then cleaved by the γ-secretase complex to produce soluble, 24 or 26 amino acid-long peptides known as P3 peptides [34]. The remaining AICD is also internalized as in the amyloidogenic pathway. Therefore, the two processing pathways generate two sets of isomeric secreted peptides with yet unknown functions: sAβPPα and sAβPPβ, P3 peptides and Aβ peptides, and the internalized AICD protein. P3 peptides are soluble peptides that are 16 amino acids shorter than Aβ40/Aβ42 and are not known to aggregate into amyloid peptide plaques.
microRNAs involved in AβPP processing pathways. miRNA involved are listed in boxes,
Based on the latest research, miRNA molecules are heavily involved in all aspects of AβPP expression regulation and relevant cleavage processing molecules. Starting from the AβPP mRNA transcript, at least seven miRNAs—miR-17, -20a, -147, -153, -323-3p, -644, and -655—are known to bind to twelve putative miRNA binding sites located in or near the 3’UTR of AβPP mRNA [35, 36]. (For the complete list of all miRNA quoted in this paper, please see Table 1). Strikingly, this highly regulated binding can be altered by single nucleotide polymorphisms (SNPs) of the miRNA binding targets. For example, the T117C variant of AβPP mRNA 3’UTR inhibits miR-147 binding, which abrogates miR-147-induced suppressive activity of AβPP expression. The A454G variant of AβPP mRNA 3’UTR increases miR-20a binding and therefore exacerbates miR-20a-induced suppressive activity. Elimination of miR-153 miRNA by antisense inhibitor or by altering the miR-153 target binding site on the 3’UTR elevates AβPP mRNA expression. In a mouse model, miR-135a, -193b, -200b, -384, and -429 suppress AβPP expression by binding to the 3’UTR of AβPP mRNA, and the results are correlated with down regulation of serum miR-135a, -193b, -200b, -384, and -429 levels in AD patients [37–39]. Therefore, these studies demonstrate that multiple species of miRNAs regulate AβPP expression and activity, and the potential differential binding site SNP may contribute to susceptibility and risk for AD [40].
miRNA in AD pathogenesis
Bold type indicates additional functional roles in AD pathology (see “Dysregulation of miRNA Regulation and AD Pathogenesis” section).
miRNA species are also involved in regulating critical enzymes involved in both AβPP processing pathways. Some miRNAs tilt the balance to the non-amyloidogenic pathway by suppressing the expression and activity of key molecules in the amyloidogenic pathway. For example, miR-29a/29b-1, -29c, and -107 are shown to bind to the 3’UTR of BACE1 mRNA and suppress BACE1/β-secretase expression and function [41–43]. Since BACE1 is critical for generating Aβ peptides that form Aβ plaques in AD patients, these miRNAs are shown to be correlatively downregulated in AD patients. miR-195 also binds to the 3’UTR of BACE1 mRNA and suppress its expression and function in the amyloidogenic pathway, while inhibition of miR-195 increased BACE1’s protein level and reversed the effect [44]. Furthermore, miR-195 appears to play an additional role by preventing tau phosphorylation, which will be further discussed later in “Dysregulation of miRNA Regulation and AD Pathogenesis” section. There are also a few other miRNAs (miR-135a, -186, -339-5p, and -384) which appear to exhibit similar suppressive activity and correlated down regulation in AD patients [37, 46]. miR-101 suppress BACE1 indirectly through repressing RanBP9, a scaffolding protein that assists in BACE1’s function in promoting BACE1-dependent cleavage of AβPP and Aβ generation, and therefore also suppresses the pathway of Aβ plaque generation [47].
On the other hand, several miRNAs are shown to favor the amyloidogenic pathway and hence the generation of more Aβ peptides. miR-103, -107, and -1306 suppress the expression of “a disintegrin and metalloprotease domain-containing protein 10” (ADAM10), a key enzyme of α-secretase cleavage activity, by binding to its mRNA 3’UTR sites [48]. In a reporter assay, overexpression of miR-103, -107, and -1306 were able to suppress ADAM10 expression by 45%, 52%, and 28% respectively. Interestingly, as mentioned above, miR-107 is also shown to suppress BACE1 expression, thereby suggesting a potential role in suppressing the overall AβPP processing and activity. Similarly, miR-144/451, known regulators of AP-1/c-Jun, bind to the 3’UTR and suppress ADAM10 expression and downstream function [49].
miRNA also regulates other signaling molecules that may affect AβPP processing pathways. For example, miR-24, -186, and -455 targeted Nicastrin, a subunit of the γ-secretase protein complex, at its mRNA 3’UTR sites, which ultimately resulted in reduced levels of Aβ peptide expression [50]. miR-9, -29a/29b-1, and -137/181c regulate serine palmitolytransferase (SPT), a rate limiting enzyme that makes ceramides, the major components of lipid rafts to facilitate AβPP processing and transportation [51]. Overexpression of SPT resulted in an increase of Aβ plaque formation.
Interestingly, miRNAs themselves are targets of feedback regulation by AICD, the internalized final product of AβPP processing pathways [52]. AICD is derived from the c-terminal end of AβPP, following both non-amyloidogenic and amyloidogenic processing pathways. It is the final cleavage product by γ-secretase through a mechanism known as regulated intramembrane proteolysis (RIP) [53]. AICD is relatively unstable in its original state, and therefore needs to be rapidly phosphorylated to form an interactome transcription regulatory complex (by binding to up to 20 cofactors) before transporting into the nucleus and becoming a modular of gene expression [54]. Although both the non-amyloidogenic and amyloidogenic pathways produce AICD after RIP, there are reports showing differential AICD levels from the two pathways and the formation of different AICD by γ-secretase cleavage at either ɛ-site or γ-site [55, 56]. Indeed, in the Presenilins Familial AD mutation study, the mutation of Presenilins, a γ-secretase catalytic component, inhibited γ-secretase cleavage activity at the ɛ-site, which then inhibited the production of signaling AICD while causing the buildup of membrane-bound cytotoxic substrates [57, 58]. In a feedback regulatory control, AICD regulates the expression of miR-663, -3648, and -3687 in human neural stem cells [59]. miR-663 has been shown to suppress the expression of multiple genes implicated in neurogenesis, including FBXL18 and CDK6. In murine fetal developing brain, the increased AβPP expression suppresses miR-574-5p expression, presumably through AICD transcriptional activity in the nucleus, and therefore blocks miR-574-5p induced neurogenesis [60]. In this system, reduced miR-574-5p promotes neurogenesis, which further reduces the neural progenitor pool. In addition to AICD, other factors in the AβPP pathway appear to play regulatory roles on miRNA as well. For example, the increase of presenilin-1 (PS1) activity, a component of the γ-secretase complex, would lead to an increase in miR-342 levels in the hippocampus in AD mouse models, inhibition of Ankyrin G (AnkG), and promotion of neurodegeneration by disrupting the axon initial segment, action potential initiation, and neuronal polarity [61]. It would be of great interest to further investigate how the differential expression levels of AICD, as a result of γ-secretase cleavage activity using either ɛ-site or γ-site, may impact or alter the regulation and expression of miRNAs.
miRNAs play critical roles in regulating AβPP expression and various AβPP processing molecules, such as ADAM10, BACE1, and components of the γ-secretase complex, or indirectly through other regulator genes, such as SPT, RANBP9, and AnkG. Remarkably, miRNA expression is also regulated by AICD and others in a feedback loop, suggesting a very dynamic, complex, yet strict regulatory mechanism in AβPP pathways [62]. Furthermore, SNPs of miRNA binding sites, and the resulting dysregulation of miRNA-induced effects, might contribute to an individual’s susceptibility and risk of AD pathogenesis. Clearly, further study is needed to sort through the complex network of miRNA regulation in AβPP processing pathways and functional mechanisms. This may lead to a better understanding of AD pathogenesis, predictive markers for early stage AD, and perhaps a better drug design or treatment to counter the formation of Aβ plaques.
miRNA FUNCTIONAL ROLES IN THE FORMATION OF NEUROFIBRILLARY TANGLES
The second pathological hallmark of AD is the formation of NFT, which are characterized by the aggregates of hyperphosphorylated tau proteins. Tau proteins are microtubule-associated proteins that are commonly found in the CNS and are abundant along the axons of neurons where they help stabilize the microtubule structure [63]. As shown in Fig. 3, normal phosphorylation of tau proteins by protein kinases regulates how tau binds to the microtubules while hyperphosphorylation of tau disrupts this binding and leads to cytoskeletal instability in the neuron [64]. Furthermore, hyperphosphorylated tau detaches from the microtubules and aggregates into insoluble filaments, which leads to the sequestration of normal tau and other microtubule-associated proteins and causes structural disruption of the neuronal axons and neuronal organelles like the endoplasmic reticulum [65]. The detached filamentous tau proteins aggregate in neuronal cell bodies and form NFT [66, 67].
microRNAs involved in tau phosphorylation and hyperphosphorylation. miRNA are listed in boxes,
Similar to the regulation of AβPP expression, several miRNAs are known to suppress tau protein expression by directly binding to its mRNA 3’UTR sites. miR-34a is known to be highly expressed and regulated in brain tissues, and likely to be involved in neuronal development, neural stem differentiation, and AD pathology [68, 69]. In human neuroblastoma cell lines, miR-34a represses tau expression by binding to its mRNA 3’UTR sites, and the effect can be reversed by the inhibition of endogenously expressed miR-34 family members [70]. miR-219 is known to be downregulated in brain tissues from AD patients and those with severe age-related tauopathy, binds directly to the 3’UTR of Tau mRNA and represses tau synthesis at the post-transcriptional level [71]. Likewise, miR-132/212, on chromosome 17 is known to be involved in neuronal development, the expression level correlated with tauopathy and cognitive impairment [72]. miR-132/212 target Tau mRNA by binding to the mRNA 3’UTR sites, and the effect can be restored by miR-132 inhibitor.
miRNAs may also alter tauopathy by regulating the expression and function of kinases and phosphatases which are involved in tau phosphorylation. Cyclin-dependent kinase 5 (CDK5) is a serine, threonine kinase that plays a critical role in neurotransmission, neuronal trafficking, neuronal development, and synaptic plasticity [73]. CDK5 belongs to the cyclin-dependent kinase family but is not activated by cyclins, nor does it play a significant role in the cell cycle. Instead, its activity is restricted to mitotic neurons and is required to bind to either p35 or p39, neuron-specific regulatory subunits, for its regulatory function. Excessive neuronal stress and toxic factors can induce hyperactivation of CDK5, which can lead to aberrant hyperphosphorylation of cytoskeletal proteins such as tau [74]. Apparently, during oxidative stress, which results in an influx of intracellular calcium, calpain, a calcium dependent protease, becomes active and cleaves both p35 and p39 into more stable truncated derivatives of p25 and p29. The truncated p25 and p29 then bind to CDK5 as CDK5/p25 and CDK5/p29 complex, which have a 6-fold longer half-life than CDK5/p35 and CDK5/p39 complex, and hence increase downstream phosphorylation activity, including the phosphorylation of tau proteins [75]. Both miR-103 and -107 inhibited CDK5R1 in SK-N-BE neuroblastoma cell lines by binding to the CDK5R1 mRNA 3’UTR site, and the overexpression affect CDK5/p35 interaction and cause decreased neuronal migration [76]. miR-107 in particular are also known to affect both BACE1 and ADAM10 activity in the AβPP pathways as mentioned earlier [43, 77]. Similarly, miR-195, which suppresses BACE1 protein expression, also regulates tau phosphorylation by binding to the 3’UTR of p35 on the Cdk5r1 gene and interfering with CDK5/p35 (murine CDK5/p25) function in rats, thereby reducing phosphorylation of tau protein [78]. Therefore, it would be of great interest to study further how miR-103, -107, and -195 may play dual roles in regulating both tau phosphorylation and Aβ plaque formations in AD pathogenesis.
There are also miRNAs that indirectly affect CDK5 activity by regulating other relevant targets. miR-26b is upregulated in human AD brains, and the ectopic overexpression of miR-26b in rat primary postmitotic neurons led to the DNA replication, aberrant cell cycle entry, and increased tau phosphorylation [79]. miR-26b appears to target retinoblastoma (Rb), which is a tumor suppressor, and therefore causes upregulation of E2F and cyclin E1 (CNE1) and downregulation of p27/kip1 cell cycle inhibitor, which ultimately leading to nuclear export and activation of CDK5. Overexpression of miR-125b in primary hippocampal neurons led to tau phosphorylation, and it directly targeted the mRNA 3’UTR of two key phosphatases, DUSP6 and PPP1CA, and the anti-apoptotic factor Bcl-W [80]. Inhibition of DUSP6 and PPP1CA prevents dephosphorylation of ERK1/2 and hence sustains ERK1/2 activity to increase CDK5/p35 activation [81]. Also, miR-125b suppression of Bcl-W, a pro-apoptotic gene, leads to an influx of intracellular calcium and, as mentioned before, promotes calpain activity, induces p35 truncation into p25, and hence activates and prolongs CDK5/p25 activity in phosphorylation of tau [75].
Glycogen synthase kinase 3β (GSK3β) is a downstream target of CDK5/p25 which has been shown to be involved in neurogenesis and synaptic plasticity, while its overexpression leads to tau phosphorylation and NFTs in AD patients [82]. In N2a and HEK293 cell lines, miR-138 is upregulated and directly binds to the 3’UTR of retinoic acid receptor alpha (RARA) mRNA, and the overexpression of miR-138 led to the inhibition of RARA, increased GSK3β activity and tau phosphorylation [83]. miR-218 suppresses the protein tyrosine phosphatase α (PTPα) by binding to its mRNA 3’UTR site, which then causes phosphorylation and activation of GSK3β, inactive protein phosphatase 2A, and hyperphosphorylation of tau protein [84]. The effect can be reversed by limiting miR-218 levels or using miR-218 inhibitor. On the other hand, miR-124 is known to be low in AD patients, and its overexpression can reduce tau phosphorylation and apoptosis by downregulating scaffolding protein Caveolin-1, which is known to be involved in the Caveolin-1-PI3K/Akt/GSK3β pathway in AD [85]. Several other miRNAs, including miR-26a, -769, -940, and -1303, are shown to directly repress GSK3β although the studies were done in different disease models [86–90]. miR-106b suppresses tau phosphorylation by suppressing the tyrosine protein kinase, Fyn [91]. It would be of great interest to see how these microRNAs interact with each other, and how their combinatorial effect contributes to NFT and AD pathogenesis.
miRNA can affect AD pathology through yet another mechanism. The insulin-like growth factor-1 (IGF-1) is highly regulated in the brain tissues of mice and humans, and shown to be involved in neurogenesis, CNS development, and cognitive faculties [92]. miR-98, in the mouse models and HEK293 cell lines, downregulates IGF-1 protein level and promoted Aβ production and phosphorylation of tau. Inhibition of miR-98 reversed the trend by upregulating IGF-1, suppressing Aβ production and reducing tau phosphorylation [93].
MEMORY LOSS AND COGNITIVE DECLINE IN EARLY STAGE ALZHEIMER’S DISEASE
Since the first publication by Alois Alzheimer, research into AD pathogenesis has focused primarily on studying and understanding the process and generation of Aβ and NFTs. This is best represented by the Amyloid Cascade Hypothesis, which proposed that the deposition of the Aβ peptide in the brain parenchyma leads to AD [94, 95]. However, the failure of clinical trials using the immunotherapeutic drugs Bapineuzumab and Solanezumab that purported to reduce Aβ42 peptide production and hence its neurotoxicity and degenerative effect, led to a call for reassessment of the focus and general strategy of AD research [5, 96]. Indeed, the impact of AβPP on AD has been disputed in another recent study using targeted gene deletion of the APP gene in trisomic Down syndrome-induced pluripotent stem cells, which challenged the notion that increased AβPP levels are solely responsible for Down syndrome-associated AD pathogenesis [97]. The parochial emphasis on studying the structural effect of accumulating Aβ plaques and NFTs in late stage AD, while overlooking the symptoms of functional memory loss and cognitive impairment found in early stage AD, seems to be rather counterintuitive [98–100]. Therefore, we sought to survey the known basic cellular and molecular mechanisms in learning and memory formation as well as the early signs of impairment in neurogenesis and synaptogenesis, which may shed a light on the etiology of AD.
Loss of memory and impairment of cognitive function are some of the first symptoms reported by patients suffering from early stage AD before the formation of Aβ plaques and NFTs [10]. The current consensus states that AD starts with memory complaints, especially in episodic memory, speech, naming, and semantic problems [101]. Semantic memory is first impaired in the language of AD patients, affecting verbal fluency and object-naming ability, which may occur several years prior to diagnosis [102, 103]. Impaired memory processing and function appear to be associated with damage in the hippocampus since the dysfunction of hippocampal formation makes the storage of new memories impossible [104]. Evidently, lesions of the hippocampus in humans are also known to prevent the acquisition of new episodic memories, as found in the famous HM patient, suggesting the critical role of hippocampus-dependent memory formation [105]. For AD patients, this learning and memory formation process is likely degenerated with a correlative impairment in neurogenesis, synaptogenesis, and loss of synaptic plasticity [99]. The prevalent theory about learning and memory formation, which includes working memory, implicit and explicit memories, from a cellular perspective all require changes in the strength of neuronal connections and synaptic activities. As proposed originally by Santiago Ramon y Cajal in 1894 and refined later by Donald Hebb in 1949, during the learning and memory formation process there is a constant modification of cellular structure and formation of new neuron synapses, which may be transient or permanent depending on the type of stimulation, synapses, and neuronal structure [10, 106]. The increased frequency of postsynaptic activation results in a rapid strengthening of the synapses because there is an enhanced release of neurotransmitters and increased trafficking of neurotransmitter receptors and protein kinase activity in the postsynaptic neuron [107, 108].
The cellular model of learning and memory formation can be best described by the theory of long-term potentiation (LTP) [109]. LTP occurs at excitatory synapses throughout the different regions of the brain but has been most widely studied in the hippocampus where learning and memory occur. LTP can be divided into two phases, E-LTP and L-LTP. The E-LTP is an early, short term (lasting a few hours) phase, which is induced with a low frequency stimulation (LFS) and requires no new protein synthesis [110]. The E-LTP phase is initiated when the presynaptic neuron releases neurotransmitters that cause postsynaptic stimulation and rapid depolarization. The L-LTP is a late, long-term (L-LTP) phase, which can last several hours or longer. L-LTP requires high frequency stimulation (HFS), which likely increases frequency of postsynaptic activation, rapid strengthening of the synapses, increased trafficking of neurotransmitter receptors and protein kinase activity in the postsynaptic neuron [107, 108]. Furthermore, the L-LTP phase requires protein synthesis since memory formation can be effectively blocked by translational inhibitors [111, 112]. Memory formation can also occur by an alternative mechanism through brain derived neurotrophic factor (BDNF), a commonly expressed neural growth factor, and neurotrophin-3 (NT-3), which are known to induce gradual but long lasting memory [113, 114]. Similar to the effect of HFS in L-LTP, the neurotrophin-induced plasticity requires protein synthesis especially in the local axonal or dendritic regions. Considering the differential regulatory mechanism between E-LTP and L-LTP, and neurotrophin-induced memory all base on protein translation instead of gene transcription, it is conceivable that miRNA, which functions by regulating protein translation, may play a major role in regulating LTP and hence memory formation [115].
miRNA FUNCTIONAL ROLES IN LONG TERM POTENTIATION (LTP) AND MEMORY FORMATION
Most excitatory presynaptic terminals form synapses on small protuberances, referred to as spines, which stud the dendrites of cortical pyramidal cells or strengthen the existing synapses. The postsynaptic side of the synapse forms electron-dense thickening known as the postsynaptic density (PSD), which contains the receptors, scaffolding proteins, adhesion proteins, cytoskeletal proteins, and numerous signaling proteins. Glutamatergic synapse is one of the best studied molecular models of LTP (see Fig. 4), which can be found in the CA1 region of the hippocampus [116, 117]. In this model, the neurotransmitter glutamate is released from synaptic vesicles in presynaptic terminals by the voltage-dependent calcium channels. Of the two well characterized glutamate receptors on the postsynaptic membrane, N-methyl-D-aspartate receptor (NMDAR) and α-amino-3-hydroxy-t-methyl-4-isoxazole proprionate receptor (AMPAR), both contain binding sites for the agonists (glutamate) but with different affinity, and NMDAR also contains internal Mg2+ binding sites [118]. During the LFS, glutamates bind to AMPARs and render them permeable to the monovalent cations Na+ and K+ but not to the divalent cation Ca2+, and NMDARs remain closed throughout the process because of Mg2+ ion blockade [119]. AMPARs rapidly desensitize and close the channel after channel opening. In the event of HFS, strong postsynaptic depolarization triggers intense activation of postsynaptic AMPARs, and lifted the Mg2+ blockade on NMDARs, which now allows both Na+ and Ca2+ to enter through NMDA receptor channels [117, 120]. The elevated Ca2+ level activates calmodulin (CaM), which then activates calcium/calmodulin-dependent kinase II (CaMKII), a serine/threonine protein kinase, other members of the CaMK kinases (CaMKK, CaMKI, CaMKIV), and kinases such as protein kinase C [121, 122]. The activated kinases translocate and bind to PSD to activate synaptic plasticity activity, thereby inducing spine structural modification and maintenance [123]. Kinases also augment potentiation by phosphorylating NMDARs and AMPARs [121, 124]. Similarly, BDNF also induces long term memory formation through AMPARS and NMDARS [125, 126], and as discussed before, this effect is likely through a rapid activation of local protein translation using mRNAs that are positioned in situ.
LTP mechanism.
The activated calcium/calmodulin kinase pathway also activates the adenylyl cyclases, which induce another critical secondary messenger, cyclic AMP (cAMP) [127]. In a series of breakthrough studies by Eric Kandel and colleagues using the marine mollusk Aplysia californica, several factors play into the regulation of gene and protein expression in L-LTP [128]. The cAMP-dependent protein kinase (also known as protein kinase A) regulates L-LTP through the ERK signaling pathway, which altered downstream expression levels of growth factors and nuclear proteins that support synaptic plasticity and maintenance [112]. Nuclear proteins such as cAMP response element-binding protein (CREB) alters the expression levels of a large pool of conserved downstream targets that are involved in synaptogenesis, neurotransmission, Ca2+ signaling, cytoskeletal remodeling, and kinase activity [129]. Methyl CpG binding protein 2 (MeCP2) acts as a transcriptional repressor or activator at methylated sites of DNA, and the loss or gain of function of MeCP2 has been linked to impaired learning and memory formation [130, 131].
Many miRNAs play critical roles in learning and memory processes. In an inducible neuron-specific knockout of the Dicer1 gene, which is critical for the last stage of miRNA synthesis as shown in Fig. 1, the deletion of the Dicer1 gene causes a reduction of brain-specific miRNA expression, including miR-29a, -29c, -124, -132, -137, and -138, in neurons in 8- to 10-week-old adult mice, and as a result, the mutant mice displayed an enhanced memory and learning with regard to spatial tasks as well as stronger fear conditioning [132]. Furthermore, the altered miRNA expression led to changes in morphology of dendritic spines and increased translation of dendritic proteins and others, such as BDNF, MMP-9, GluR1 and GluR2 (GluA1 and GluA2), and the PSD95 protein. For example, in the Aplysia model, the binding of serotonin (5-HT) to its receptor leads to downregulation of miR-124, which inversely causes increased CREB expression [133]. miR-124 has a binding site on the CREB mRNA 3’UTR and thereby inhibits the protein translation of CREB [134]. miR-137 and miR-124 target the 3’UTR of GluA1 and GluA2 mRNA, respectively [135, 136]. Overexpression of miR-137 in the hippocampus leads to post-synaptic silencing and reduced LTP while miR-124 regulation of GluA2 may be redundant. miR-138 targets the 3’ UTR of acyl protein thioesterase 1 (APT1) mRNA, which promotes depalmitoylation of membrane bound proteins such as AMPARs, and is considered to negatively regulate protein localization at the synapse [137]. Since miR-138 regulation correlates with short-term memory recognition, its regulation of APT1 appears to play a role of maintaining the functional stability of new dendritic spines and allowing for the proper localization of membrane bound synaptic proteins [138].
The expression of miR-132/212 and together with miR-219 and miR-34a-5p is closely regulated by various transcription factors, including CREB, mGluR, and NMDAR in the postsynaptic neurons, and is known to play critical functional roles in neuronal plasticity, gene expression, and cognitive capacity [139–142]. miR-132/212 was first identified through a genome-wide screen as a target of the transcription factor CREB, which has direct binding sites on the miR-132/212 transcripts. Interestingly, HFS induces mGluR-dependent transcription upregulation of miR-132/212, while NMDAR activation selectively downregulates mature miR-34a-5p, -132, -212, and miR-219 levels, indicating an accelerated decay of mature miRNAs [141, 142]. miR-132/212 double knockout mouse has a reduction in postsynaptic depolarization in the hippocampus while conditional induction of transgenic miR-132/212 triggers a marked increase in dendritic spine density [143–145]. miR-132 regulates GTPase-activating protein, p250GAP, which is responsible for actin turnover [140]. Another miR-132 targets is methyl CpG–binding protein 2 (MeCP2), which is implicated in Rett Syndrome and other mental retardation disorders, and the increased expression of MeCP2 promotes synaptic maturation as well as BDNF expression [146]. The expression of BDNF also promotes miR-132 expression and results in a feedback loop mechanism where a rise in miR-132 expression can reduce the expression of various genes that promote neurogenesis. miR-219 targets CaMKIIγ expression in mouse dorsal root ganglia, regulates BDNF expression, and suppresses the expression of NMDAR subunit 1 (NR1) [147, 148]. The miR-34a deficiency in APP/PS1 mice also significantly attenuates cognitive deficits in early stage AD [149].
Sirt1 (silent mating type information regulation 2 homolog 1) is highly regulated by various microRNAs, including miR-9, -34a-5p, miR-134, and miR-138. Sirt1 activity is involved in LTP by enhancing synaptic plasticity through the targeted suppression of miR-134 via the YY1 transcription factor [150]. Sirt1 loss of function led to increased expression of miR-134 and impaired dendritic spine formation because miR-134 inhibited the translation of key L-LTP genes such as CREB and BDNF. miR-9 inhibits Sirt1 activity in AD during neurodevelopmental stages in order to promote neuronal differentiation, and miR-9 also regulates the translation of synapse-associated protein 97 (SAP97) which can negatively regulate AMPAR [151, 152]. Sirt1 was identified as a target of miR-34a, which controls neurite elongation [153]. miR-138 suppresses axon regeneration by suppressing Sirt1 activity, and Sirt1 also acts as a transcriptional respressor to suppress miR-138 in a feedback regulatory mechanism [154].
There are several other microRNAs crucial to L-LTP protein synthesis and maintenance. miR-26a and miR-384-5p are downregulated during LTP, thereby alleviating their suppressive activity on Ribosomal S6 kinase 3 (RSK3), which is critical for rpS6 phosphorylation and subsequently increased protein translation activity [155].
DYSREGULATION OF miRNA REGULATION AND AD PATHOGENESIS
Thus far, we have discussed a long list of miRNA species which play a pleiotropic role throughout the various stages of neurogenesis and AD pathogenesis. We also described how the miRNAs are feedback-regulated by some downstream factors that they themselves regulate. This begs the question: could the dysregulation of miRNA be causative of AD etiology? Although there is no direct evidence to support this supposition so far, various reports, as discussed here and additionally by others, appear to point in this direction. If miRNA dysregulation has a causal relationship with AD pathogenesis, the dysregulation of miRNAs should be directly or indirectly associated with the three hallmarks of AD.
Some of the most direct evidence of RNA’s role in memory formation comes from a recent study by the Glanzman team [156]. Bedecarrats et al. showed, at least in Aplysia, that a mere transfer of RNA from a sensitized neuron of a trained sea slug to a naïve sea slug is sufficient to transfer the sensitized memory/engram to the naïve untrained sea slug. Although it is uncertain whether the “transferred memory” was based on mRNA, long noncoding RNA (lncRNA), miRNAs, or the combination thereof, this surprise finding nevertheless suggests that RNA is the central element or engram for memory formation and cognitive capability. Similarly, as discussed in this paper, much circumstantial evidence supports the notion that the dysregulation of miRNA may be one cause of memory loss and cognitive decline. We have noted that the deletion of the Dicer1 gene in a neuron-specific manner causes downregulation of miRNA maturation, which actually causes memory and learning enhancement, changes in morphology of dendritic spines, and increased translation of dendritic proteins such as BDNF in mutant mice [132]. miRNAs such as miR-9, -26a, -32a-5p, -124, -132, -134, -137, -138, -212, -384-5p, and let-7a suppress proteins involved in key memory formation and learning, like CREB, and various other factors, including BDNF and Sirt1, that are involved in AMPAR and NMDAR signaling and activation, the dendritic spine formation pathway and synaptogenesis [133, 155].
Likewise, as we also discussed earlier and summarized in (Table 1, miRNAs are involved in regulating both AβPP pathways and NFT formation. miR-17, -20a, -93b, -147, -153, -200b, -323-3p, -384, -429, -644, and -655 are known to suppress AβPP protein expression, while miR-34a, -132/212, and -219 have been shown to suppress the expression of tau protein [35–39, 70–72]. For the AβPP processing pathways, miRNAs also regulate various key downstream factors in both the amyloidogenic and non-amyloidogenic pathways, such as BACE1/β-secretase, RanBP9, ADAM10, AP-1/c-Jun, SPT, Nicastrin, and AnkG [41–43, 61]. In the same way, miRNAs regulate the expression and function of kinases and phosphatases relevant to tau phosphorylation, such as CDK5, Rb1, DUSP6, PPP1CA, Bcl-2, PTPα, caveolin-1, and GSK3β [76, 86–90].
As discussed earlier and shown in Figs. 2 and 3 and Table 1, some miRNA species play multifaceted roles in different aspects of AD pathology. These miRNAs are of particular interest since their regulatory function, or the dysregulation of their regulators, may hold the key to understanding AD pathology. For example, miR-98 regulates IGF-1 protein translation, and the expression of IGF-1 protein would affect downstream Aβ production and phosphorylation of tau [92, 93]. Similarly, miR-103 and miR-195 suppress BACE1 expression while miR-107 is known to suppress both BACE1 and ADAM10 activity, suggesting an overall suppressor of AβPP processing and activity [44, 78]. Together, miR-103, -107, and -195 suppress the expression of CDK5R1 and downstream kinase activity in tau phosphorylation, and they also appear to play a role in associated learning [157]. Another interesting miRNA is miR-34a, which appears to play multiple roles in all aspects of AD. The tissue knockout of miR-34a causes significant cognitive deficits in early stage of AD and reduction of the buildup of Aβ plaque deposition and suppression of tau expression in late stage AD [70, 158]. miR-132/212 regulates both memory formation and neuronal development as well as suppress tau expression [72, 143–145]. There are few others as listed in Table 1, including miR-9, 26a, 29a/29b-1, 29c, -137, -138, -219, and-384; all play multiple roles in various aspect of AD.
Therefore, it would be of great interest to study how these miRNA species may play multifarious roles in various aspects of AD pathogenesis and how their dysregulation may contribute to the root cause of AD pathogenesis.
In the same manner, it would be of great interest to also study the regulators of miRNA maturation. Of particular interest are the miRNA regulators that also feedback-regulated miRNA expression and maturation. As mentioned throughout this paper, there are quite a few miRNA-regulated factors that feedback-regulated miRNA expression, including MAPK1, AICD, and CREB. For example, miR-132-3p is known to regulate MAPK1, which is involved in synaptic plasticity and L-LTP. MAPK1 (Mitogen-activated protein kinase 1) is known to belong to the MAP kinase family, whose members are critical to signaling transduction pathways [142]. MAPK1 protein in turn can influence the stability of the Dicer complex and, hence, affect the maturation of new miRNAs. Sirt1 expression is regulated by various miRNAs, including miR-9, -32a-5p, -34a, -134, and -138; and Sirt1 also acts as a transcriptional repressor of the expression of miR-138 [150, 154]. AICD, the downstream final product of the AβPP processing pathway, is shown to regulate the expression of several miRNAs, such as miR-663, -3648, and -3687 in human neural stem cells and miR-574-5p, which promotes neurogenesis [59, 60]. NMDAR activation leads to both rapid downregulation of mature miR-132/212 and inhibition of processing the miR-132/212 precursor, suggesting how the feedback regulation works in regulating miRNA function [141].
POTENTIAL MECHANISMS OF miRNA-DYSREGULATION INDUCED AD PATHOGENESIS
In general, the key mechanism of miRNA-exerted regulatory function remains the same—it suppresses the target protein expression by binding to its mRNA 3’ UTR (or other regions) site [17]. Depending on a collection of factors, including the abundance and combination of different miRNA species, complementation sequence on the binding site, presence of other signaling factors, and stimulation signal strength and time interval, the protein translation may be initiated, suppressed, or shut down completely after induced mRNA degradation [16, 18]. The complexity and feedback-loop regulation of miRNA function by various downstream factors demonstrates this as a highly regulated mechanism. Clearly, any breach of the regulatory cycle can be pathogenic, leading to dysregulation of miRNA expression and maturation; dysfunctions in regulating synaptic plasticity in neurogenesis and memory formation; interference with the functions of kinases and phosphatases, which are critical for tau hyperphosphorylation; and interruption of molecular circuitry for clearing intermediate signaling molecules such as Aβ peptides, which ultimately precipitates into Aβ plaques [21].
As discussed previously, long-term memory formation requires either HFS-induced L-LTP or BDNF/NT-3 induced long-lasting synaptic plasticity and synaptogenesis [113]. Both mechanisms require new protein synthesis and, in particular, an immediate protein translation since L-LTP can be effectively blocked by treatment with translational inhibitors around the time of L-LTP induction (typically a 30–45 minute window) [112]. How can synapse specificity be achieved within a neuronal body without a transient and elaborate intracellular protein trafficking network? Due to temporal and spatial constraints, it is likely that the activity-dependent synaptic plasticity is dependent on local protein translation [159, 160]. In other words, all necessary mRNAs and translational machinery are trafficked to the specific subcellular compartments in advance, but remain translationally silenced until the initiation of L-LTP, which would then activate protein translation within the subcellular compartment. Frey and Morris in 1997 proposed the model of “synaptic tagging,” which put forth the idea of short-lived protein synthesis-independent “synaptic tags” existing at the local synapse [161]. These synaptic tags sequestered the expression of relevant proteins in situ (from localized mRNAs within the postsynaptic spine) generated from a prior activity within the neuron, such as E-LTP with LFS. However, upon repeated stimulation of HFS, the specific synaptic tag suppression is alleviated, which results in protein expression within the specific synapse in situ and achieving input-specific synaptogenesis and memory formation [162]. Since the chief molecular mechanism of miRNA function is to suppress protein translation and expression, and miRNA are also relatively small in size with high stability, it is fitting that miRNA may play the role of “synaptic tags” [163].
By combining the idea that RNA is the engram of memory formation with the theory that miRNAs function as “synaptic tags”, a mechanistic regulatory picture of RNA-induced memory formation begins to emerge. To visualize this interaction, as shown in Fig. 5, one possible scenario is that memory formation is a process involving a set of “suppressive miRNA tags/switches.” During the initial stage of a memory forming event, a set of relevant “memory forming” mRNAs are transcribed but remain translationally silenced by a corresponding set of miRNA tags while being trafficked to a specific subcellular compartment, i.e., nascent postsynaptic spines. Following a secondary memory event, such as HFS or repetitive stimulations, the “miRNA tag” suppression is removed, protein translation is then turned on, and a cascade of “memory forming activity” is initiated, including phosphorylation of NMDAR and AMPAR. The cascade induces a strong depolarization and potentiation of the PSD compartment, activation of scaffolding and structural proteins, as well as signal transduction activity that leads to synaptogenesis, morphologically transforming growth cones into a new synapse or strengthening the existing synapse with the adjacent neurons [164]. This process will be repeated as the axon terminal of the current neuron forms a new synaptic junction with the adjacent recipient neuron. In the event of a weak or absent secondary memory event, such as stimulation cessation or stimulation below an acceptable threshold, the “memory forming” mRNA may begin to degrade, resulting in the gradual loss of the ability to form new memory.
microRNA’s functional role in synaptogenesis and memory formation.
It is conceivable that the initial “suppressor miRNA tags” may be derived from the intronic miRNAs, which are the product of RNA splicing of pre-mRNA. In this way, the miRNA-induced mRNA suppression occurs soon after the initial transcription process. The suppressive activity may be adjusted by an alternative exon-intron splicing variation or by factors that regulate intronic miRNA maturation. The secondary memory event occurs during subsequent HFS or repeat stimulation, which possibly initiates the de novo transcription of a new set of secondary activators, either in the form of intergenic or other intronic miRNAs, or a new set of regulator proteins. The secondary memory activators remove the suppressive effect of miRNA tags, hence inducing protein translation and the subsequent cascade of memory forming activity. Another potential mechanism is through the regulatory effect of factors which interfere with the half life and maturation of miRNA tags, thereby reducing miRNA tag activity. Likewise, the input-specific synaptogenesis can also be explained by the “suppressor miRNA tags.” During the initial stage of a memory event, we may have different sets of “memory forming” mRNAs transcribed following differential stimulations, translationally silenced by their own respective sets of suppressor miRNA tags, and trafficked into their corresponding subcellular compartments. In the event of a secondary memory event, the translation of a unique set of activators for the specific signal, will selectively remove miRNA tags corresponding with the secondary memory event. This would leave the other “memory forming” mRNA translationally silenced and unresponsive, hence achieving the intracellular signaling specificity.
Clearly, any interference of the “suppressor miRNA tags” formation or suppressive activity and mechanism can be detrimental and pathogenic. For example, the inability to remove miRNA suppressor function effectively can be inhibitory to synaptogenesis, thus causing the inability to form new memory. Likewise, the inability to regulate miRNA function properly due to, for instance, an interruption of the feedback control loop or dysregulated activator miRNA function, may switch the signaling transduction pathway into overdrive. This can cause hyperphosphorylation of kinases and phosphatase, and hence hyperphosphorylated tau protein, leading to the formation of NFT. As mentioned before, the high expression of AβPP in neuron and glial cells of the CNS is thought to be involved in some aspects of synaptogenesis, neuronal adhesion, neuroprotection, and neural plasticity. Therefore, the inability to regulate miRNA suppression properly may trigger massive AβPP synthesis with dysregulated AβPP processing and metabolizing process, which may then lead to the aggregation of Aβ plaques. Also, it is conceivable that the SNP on the miRNA binding sites on the mRNA 3’UTR of key factors may play a role in increasing susceptibility and risk of developing AD. The dysregulation of miRNA function could then theoretically lead to AD pathogenesis.
DISCUSSION
In this review, we focused on the potential cause of AD in the form of dysregulation of miRNA function, and especially on its effect on the three hallmarks of AD. As discussed in detail, we have observed that many miRNA species play multifaceted roles in all aspects of AD pathology, suggesting its potential central role in AD pathogenesis. We propose the “suppressive miRNA tags” as the potential key regulators of an RNA-induced input specific memory formation process. Most evidence is circumstantial, but more and more studies are supporting this hypothesis. This is especially true as demonstrated by Bedecarrats et al. that the mere transfer of RNA is sufficient to transfer sensitized memory/engram to a new host [156]. It is beyond the scope of this review, but miRNA appears to play a critical role in other neurogenesis as well [165]. Here, we focused on the assessment of miRNA’s functional role in memory formation in the L-LTP and the excitatory glutamatergic synaptic system. However, it has also been studied and published that miRNA likewise plays a major role in other types of synaptic transmission, such as GABA [166], acetylcholine [167], dopamine [168], serotonin signaling [169], and μ opioid system [170]. Therefore, it would be of great interest to continue investigating the various aspects of miRNA regulation.
Since RNA appears to be the central element of memory formation, it is conceivable to begin careful study of the specific RNA species, such as mRNA, lncRNA, miRNA, or combination thereof, regarding their role in memory formation. miRNAs can potentially be used as markers for detecting, diagnosing, and determining the prognosis of AD. The combination of exogenous miRNAs may also be used as a drug treatment to counter the effects of AD. As shown by Lee et al., the replenishment of miRNA-188-5p is sufficient to restore the synaptic and cognitive deficits in 5XFAD mice [171]. Also, it is possible to design antisense oligonucleotides (ASO) based on siRNA or a Crispr-CAS9 system to target degradation of crucial miRNA targets [172]. DeVos et al. demonstrated through non-human primate models that the use of short sequenced ASOs can reduce phosphorylated tau levels, spare hippocampal neurons, and delay AD progression [173]. Farr et al. used an AβPP targeted ASO strategy in mouse models to reduce Aβ levels; as a result, the study was able to demonstrate that memory and learning improved along with a reduction in neuroinflammatory cytokines [174]. As far as targeting miRNAs are concerned, Wang et al. used miR-34a ASOs in mouse models to show that delivery of mir-34a ASOs increased Bcl-2 protein levels and reduced capase-3 activity [175]. The miR-34a ASO may inhibit cell death pathways and reduce Ca2+ levels. As a result, mir-34a ASOs might one day work in combination with memantine for late stage AD in order to prevent neurons from becoming overexposed to Ca2+ [176].
We have discussed throughout this review how the regulatory function of miRNA is dynamic and complex. For example, at least seven miRNAs are known to have binding sites on the 3’UTR of AβPP mRNA. Some miRNAs, such as miR-195, have multiple binding sites on different mRNAs and hence regulate different signaling pathways at the same time. Also, the complementary nature of a miRNA sequence to its binding site may exhibit differential effects since a perfectly complementary target site may lead to mRNA degradation while a less perfect complement may result in only a translational silencing effect [16]. This redundancy potential could complicate any analysis of an individual miRNA’s effect on a specific aspect of AD pathology, where care should be taken to sort through individual versus common effects. Nevertheless, a good start would be to initiate a general screening and categorization of various miRNA species throughout the different aspects of neurogenesis and AD pathogenesis. This will help to identify some key players in the processes, which may require further study in the future.
Footnotes
ACKNOWLEDGMENTS
The funding of this research is supported by the Valley Hospital Foundation and General Operation Fund. We would like to acknowledge the generous support from the research community of the Valley Hospital, which is located in Ridgewood and Paramus Townships, New Jersey. We thank all the sponsors and former patients who donated and funded our research. We thank Ms. Audrey Meyers, President/CEO and Ms. Julia Karcher, Vice President, of the Valley Hospital for their generous and continuing support. We thank all the physicians and technicians from the Valley Hospital who contributed to our Biobank for their expert advice. We thank all the coordinators and staff for their help in our research endeavors.
