1. Introduction
The dysregulation of ncRNAs associated with Alzheimer’s Disease (AD) is well acknowledged
[1][2] and lncRNAs in association with AD have been extensively reviewed
[3][4][5][6]. Some studies focus on the role of lncRNAs as potential AD biomarkers
[7], while others focus on lncRNAs showing competing endogenous RNA network (ceRNA) mechanisms
[8]. More recently, a group of lncRNAs has been suggested as potential therapeutic targets for AD
[9].
Since neuroinflammation is a central mechanism of AD, attention has focused on the possible immune-modulatory activities of lncRNAs, revealing that they can positively and/or negatively regulate innate immune gene expression through their general mechanisms of action (miRNA sponge, chromatin remodeling, transcriptional activation/inhibition, post-transcriptional modification) and even regulation of protein activity
[10][11][12]. Transcriptomic studies have highlighted the role of different lncRNA-associated ceRNA networks in the overexpressing APP/PS1 mice model, associated with an early stage of AD, mainly involved in synaptic plasticity, memory, and neuroinflammation
[13][14]. In the brain, inflammatory stimuli, such as the one caused by lipopolysaccharides (LPS), regulate the expression of genes via the upregulation of several lncRNAs
[15]. Furthermore, the lncRNA modulation of neuroinflammation is increasingly acknowledged as a key mechanism underlying nervous system disorders
[16].
A recent review has highlighted the role of non-coding RNA, in regulating inflammation in AD
[17]. In that manuscript, only proinflammatory lncRNA is described. Here, researchers detailed the updated literature that, experimentally, has demonstrated the association of specific lncRNAs with inflammatory processes in the context of AD. According to their reported effect in cellular, animal models, and patients, researchers classify lncRNAs as pro-inflammatory or anti-inflammatory (
Table 1).
Table 1. Long non-coding RNAs related to inflammation and Alzheimer’s disease. LncRNAs are described in alphabetical order. In red are the putative pro-inflammatory lncRNAs and in blue are the putative anti-inflammatory lncRNAs.
2. Pro-Inflammatory lncRNAs, Evidence from Cellular, Animal Models, and Human Studies
To date, the majority of reported neuroinflammation-associated lncRNAs regarding AD progression have a pro-inflammatory role.
LncRNA
Antisense Non-coding RNA in the INK4 locus (ANRIL) is mapped at the INK4 (Inhibitor of Cyclin-dependent Kinase 4) locus and has been identified in several diseases that are related to inflammation and nerve dysfunction
[49]. In pheochromocytoma cells (PC12), a well-known neuronal cellular model, lncANRIL was upregulated by incubating the cells with Aβ(1–42) oligos. Moreover, the lncANRIL knockdown decreased inflammatory cytokine expression, inhibited Aβ-induced apoptosis and autophagy, and led to increased neurite outgrowth, by binding and downregulating miR-125a
[19]. In models of coronary disease, lncANRIL can increase NF
kB expression via miR-181b modulation
[20], but whether miR-181b is targeted in neuronal cells is still unknown. Moreover, whether these changes are reflected in human or animal models of AD remains elusive.
LncRNA
BACE 1 Antisense RNA (BACE1-AS), the antisense of BACE1 (β-secretase), improves BACE1 mRNA stability, preventing the binding of miR-485-5p, thus increasing BACE1 levels
[22]. In a neuronal cellular model of PD, BACE1-AS can regulate apoptosis, inflammatory response, and oxidative stress, through direct regulation of the miR-214-3p/CDIP1 (Cell Death Inducing P53 Target 1) signaling axis
[23].
Interestingly, BACE1-AS/BACE1 dysfunction underlies several human diseases with strong inflammatory components, including multiple tumors and degenerative diseases
[50]. BACE1-AS is upregulated in serum samples of AD patients and brain tissues of AD transgenic (Tg) mice
[21], promoting neuronal damage mediated by autophagy by binding to miR-214-3p and indirectly inhibiting ATG5 expression
[21].
LncRNA
Brain-derived Neurotrophic Factor Antisense (BDNF-AS) has been reported as a target of anti-inflammatory treatments. Specifically, lithium treatment decreased inflammation via decreasing BDNF-AS levels and increasing its target miR-9-5p in a rat model of spinal cord injury (SCI), and it reduced the inflammatory effect caused by H
2O
2 in SH-SY5Y cells
[24]. According to this putative role in facilitating inflammation, increased levels of BDNF-AS can impair cognition in neurodegenerative preclinical models. Moreover, elevated levels of BDNF-AS are found in AD patients’ blood
[25]. Despite this evidence, the direct role of BDNF-AS on inflammation in AD models has not yet been reported.
LncRNA
HOX Antisense Intergenic RNA (HOTAIR) is highly expressed in inflammatory conditions, e.g., tumors, traumatic brain injury mice model, and LPS-treated microglial (BV2) cells. Accordingly, silencing HOTAIR suppresses microglial activation and the release of inflammatory factors
[16]. Supporting a pro-inflammatory role of HOTAIR, sulfasalazine (used for the treatment of autoimmune diseases) reduces HOTAIR expression and prevents the increment of M1-like microglia in a mice model of cuprizone-induced demyelination
[26]. Furthermore, exercise downregulates HOTAIR, and increases its target miR-130a-3p in rat models of AD. In this study, treadmill exercise exerts neuroprotection by reducing inflammatory microglia and oxidative stress, and consequently, improving cognitive function
[27]. In humans, aerobic exercise can attenuate the white matter hyperintensities associated with AD and aging
[51]. Not surprisingly, exercise has been proposed as a useful strategy to prevent AD
[52], due to its potential anti-inflammatory effect
[53]. However, whether the anti-inflammatory effect of exercise is mostly mediated by HOTAIR reduction, or whether this is a downstream event, has not been demonstrated.
LncRNA 17A is a 159-nts antisense transcript, embedded in the human G-protein-coupled receptor 51 gene (GPR51), GABA B2 receptor. The stable expression of 17A in SH-SY5Y cells promotes an alternative GABA B splicing isoform that inhibits GABA B intracellular signaling
[18]. Synthesis of 17A is controlled by inflammatory processes, and it is upregulated in the cerebral cortex of AD patients and appears to enhance the secretion of Aβ in SH-SY5Y cells as a response to inflammatory stimuli
[18]. Impaired GABAergic function plays a significant role in AD, as alterations in this kind of transport account for neurodegenerative diseases, specifically the shift of GABA to depolarizing direction because of the impairment of the KCC2 (
potassium chloride cotransporter 2). In AD11 mice, a model of sporadic AD, the neutralization of NGF (
nerve growth factor) leads to a neurodegenerative pathology such as the one observed in AD patients
[54]; thus, this lncRNA represents an interesting link between inflammation and AD.
LncRNA
Membrane-associated Guanylate Kinase Antisense RNA 3 (MAGI2-AS3) Aβ
25–35 incubation of SH-SY5Y and BV2 leads to increased expression of lncMAGI2-AS3 that results in reduced levels of its target miR-374b-5p levels. This result suggests that MAGI2-AS3/miR-374b-5p axis may regulate the neurotoxicity and neuroinflammation induced by Aβ
25–35 [55]. Furthermore, miR-374b-5p appears to be important in neurogenesis and it is found downregulated in AD patients, coherent with a pro-inflammatory role of MAGI2-AS3.
In that sense, and although no studies on MAGI2-AS3 in AD patients have been reported yet, MAGI2-AS3 appears in a screening of the ceRNA network in human asthma studies
[56], confirming a potential pro-inflammatory role of MAGI2-AS3.
LncRNA N336694 is found up-regulated in APP/PS1 mice brain tissue, suggesting a pro-inflammatory role
[57]. In this study, miR-1066 was also found upregulated, and, although bioinformatic analysis suggested that miR-1066 may be a potential target of lncRNA n3366994, no empirical confirmation has been reported. Interestingly, simvastatin treatment that ameliorated cognition in mice models of AD, was shown to suppress lncRNA n3366994 and miR-106b expression in the brain in APP/PS1 mice
[28].
LncRNA
Neuroblastoma differentiation marker 29 (NDM29) is a lncRNA transcribed by RNA pol III, embedded in the first intron of the ASCL3 (
achaete scute-like homolog 3) gene in humans. NDM29 expression is enhanced in the cerebral cortex of AD patients, its biosynthesis responds to pro-inflammatory molecules, and it is downregulated by anti-inflammatory drugs in different neuroblastoma cell lines
[29].
Nuclear paraspeckles assembly transcript 1 (NEAT1) is upregulated in the temporal cortex and hippocampus of AD patients compared to controls
[32]. NEAT1 can modulate inflammatory processes in several cell types and several human pathologies of inflammatory conditions, via the modulation of several miRNAs, in the AKT, TLR4, TRAF6, and NF-κB signaling pathways
[31]. In SH-SY5Y cells, Aβ protein incubation increased NEAT1 and decreased its target miR27a-3p
[30].
LncRNA
Prostate androgen-regulated transcript 1 (PART1) has a key role in a variety of biological processes. Using Aβ
1–42-incubated endothelial cells as a model of the blood-brain barrier (BBB), lncRNA PART1 increased BBB permeability via binding NOVA2. Thus, in this model, high expression of lncPART1 led to reduced PPP2R3A mRNA levels and subsequently increased NFkB-p65 phosphorylation. NOVA2 (
neuro-oncological ventral antigen 2) expression is reduced in this environment and stabilizes lncPART1, resulting in sustained NFkB-p65 phosphorylation. This signaling contributes to the alteration of BBB proteins (e.g., occludin, claudin-5, and ZO-1), leading to increased permeability. Although this could be a potential mechanism aggravating AD pathogenesis, more studies are needed to unravel the function of this lncRNA in AD, as a diagnostic and therapeutic
[33].
LncRNA
Ribonuclease P RNA component H1 (Rpph1) participates in the maturation of tRNA
[58]. This lncRNA is upregulated in the cortex of an APPswe/PS1deltaE9 mice model of Alzheimer’s, compared to wild-type controls. In this model, Rpph1 upregulates CDC42 (
Cell division cycle 42) regulation, by competing with miR-330-5p and miR326-3p. CDC42 modulates actin dynamics, promoting dendritic spine formation
[47][58]. Furthermore, Rpph1 displays a neuroprotector effect in SH-SY5Y cells incubated with Aβ via miR326-3p/PKM2 (pyruvate kinase isoform M2) axis
[48]. However, Rpph1 overexpression can promote inflammation under low glucose conditions in a model of mesangial cells
[46]; thus, it is still unclear whether the increased levels in the AD mice model are due to early compensation or to a pro-inflammatory role. At this point, the exact function of Rpph1 in AD remains elusive.
LncRNA
SOX21-antisense Transcript 1 (SOX21-AS1) represses the expression of the SOX21 gene, a member of the large SOX (
SRY-related HMG-box genes) family of transcription factors involved in development regulation
[59]. It is mostly studied for its oncogenic properties. In AD mice, knocking down SOX21-AS prevents neuronal oxidative stress and inhibits cell death, through the upregulation of the FZD3–5 (
Frizzled receptor 3)/Wnt signaling pathway
[40] and given the close relationship between oxidative stress and inflammation
[60], SOX21-AS1 can be classified as pro-inflammatory lncRNA.
LncRNA
SRY-box transcription factor 2 overlapping transcript (SOX2-OT) is transcribed from the intron of the Sox2 gene
[61] with a key role in maintaining SOX2 expression
[62]. SOX2-OT is involved in neural embryonic development and adult mouse neurogenesis. Although adult neurogenesis is impaired in AD mice models
[63], it is not known whether SOX2-OT dysfunction may contribute to the progress of the disease. However, SOX2-OT has been shown to mediate inflammation, oxidative stress, and neuronal apoptosis in PD cellular models, acting via miR-942-5p/NAIF1 (
Nuclear apoptosis-inducing factor 1) axis
[39]. Although there is no experimental evidence in AD cellular or animal models, a Logic Mining method used for the analysis of a microarray expression dataset shows SOX2-OT as one of the five genes common to both early and late AD states of the
anti-NGF AD11 transgenic mouse model, a model of sporadic AD
[63]. Further studies are required to validate this gene in human transcriptional studies.
LncRNA
Small nucleolar RNA host gene 1 (SNHG1) is a lncRNA that belongs to the Small Nucleolar RNA host gene (SNHG) family, comprising more than 20 members, many of which have been found associated with cancer progression
[64]. In SH-SY5Y and human primary neuron cells, Aβ incubation increased the expression of SNHG1, while the silencing of this lncRNA attenuated Aβ-induced cellular death and alterations in mitochondrial membrane potential. In this study, SNHG1 was shown to act as a miR-137 sponge targeting KREMEN1 (
Kringles Containing Transmembrane Protein 1)
[36]. KREMEN1 is a transmembrane receptor that blocks the WNT/catenin pathway but can induce apoptosis independently
[65]. Interestingly, silencing of KREMEN1 (by miR-431 overexpression) prevented Aβ-mediated synapse loss in primary cultures from a mice model of AD, suggesting that KREMEN1 may facilitate AD progression
[66]. Another pathway regulated by SNHG1 is the miR-361-3p/ZNF217 axis in neuroblastoma cell lines (SK-N-SH and CHP 92 212); Aβ
25–35 increased SNHG1 and reduced miR-361-3p, increasing its target ZNF217 (
zinc finger gene 217 transcription factor) levels. ZNF217 is also the target of miR-212-3p
[35] and miR-200
[34] in the context of Aβ
25–35-induced inflammation in PC12 cells, where ZNF217 upregulation is associated with increased neurotoxicity.
LncRNA
Small Nucleolar RNA Host Gene 14 (SNHG14) is another member of the SNHG family and has an essential role in promoting pro-inflammatory microglia activation
[37]. In astrocytes from the transgenic APP/PS1 mice model, SNHG14 was reported to sponge miR-223-3p, which directly targets and restrains NLRP3 inflammasome. In this model, angiotensin analogs inhibit inflammation and prevent cognitive impairment by inhibiting SNHG14, thus restoring miR-223-3p function
[37]. Exercise that improves cognition and reduces inflammation markers can also reduce SNG14 levels, in mice models and AD patients
[38].
3. Anti-Inflammatory lncRNAs
Although less studied than the pro-inflammatory, some lncRNAs have an anti-inflammatory action in AD.
LncRNA
Maternally Expressed Gen 3 (MEG3) expression declined in the hippocampus of AD model rats, and over-expressing MEG3 inhibited the activation of the astrocytes, reducing neuronal damage via the PI3K (Phosphoinositide 3 kinase)/AKT pathway
[45]. No data on humans have been reported yet.
LincRNA-p21 was found upregulated by
Bilobalide (the effective component of EGb76, extract of Ginkgo biloba), decreasing neuroinflammation, and promoting autophagy in a mice model of AD
[41].
Another example is represented by MALAT1 (
Metastasis-associated Lung Adenocarcinoma Transcript 1), which is downregulated in the cerebral-spinal fluid (CSF) in AD patients compared to controls
[43][44]. MALAT1 levels correlate positively with alleviated AD severity, as evaluated by the Mini-mental Status examination (MMSE) score, and biomarkers Aβ
42, t-tau, and p-tau. MALAT1 reduction and miR125b increase correlate with AD, but not with PD; suggesting that lncMALAT1/miR-125b are potential biomarkers for AD diagnosis. Consistently, in two cellular models of AD, MALAT1 was found to inhibit inflammation by sponging miR-125b
[43]. Intriguingly, MALAT1 can promote neuroinflammation by NRF2 inhibition in a Parkinson’s Disease (PD) mouse model
[42]. Further research is required to unveil the putative opposite role of MALAT1 in the inflammatory process underlying PD and AD.