Non-coding RNAs and Mitochondrial Dysfunction in Alzheimer’s Disease: A Systematic Review Authors: Ștefania-Maria Mocrei-Rebrean, Vlad Răzniceanu, Sebastian Romeo Pintilie, Ioana Berindan-Neagoe Published: 2026 ABSTRACT AbstractAlzheimer’s disease (AD) is responsible for 70% of dementia cases worldwide, with tau hyperphosphorylation and amyloid-β plaque accumulation representing its core pathological hallmarks. Genetic predisposition, oxidative stress, and neuroinflammation contribute to disease onset and progression. Non-coding ribonucleic acids (ncRNAs) are a class of RNAs which control gene expression and whose dysregulation in AD patients has been linked to amyloid production, neuroinflammation, and mitochondrial dysfunction, which ranges from impaired energy metabolism to disrupted mitochondrial biogenesis and dynamics. Our descriptive systematic review surveyed the involvement of ncRNAs in mitochondrial dysfunction in AD across experimental and clinical literature. We identified multiple microRNAs (miRNAs), long non-coding RNAs (lncRNAs), and circular RNAs (circRNAs) that directly regulate mitophagy, mitochondrial biogenesis, mitochondrial autophagic, and apoptotic pathways, mitochondrial dynamics, and protein import mechanisms in AD models. Among the most important candidates demonstrating clinical dysregulation, miR-140 and lncRNA NEAT1 regulate mitophagy, while miR-9, miR-34a, miR-146a, miR-155, and miR-485 are implicated in mitochondrial biogenesis and miR-204 in mitochondrial autophagy. LncRNA BDNF-AS, miR-148a-3p, miR-21-5p, and miR-103a-3p emerged as regulators of the mitochondrial apoptosis pathway with confirmed clinical dysregulation. Multiple ncRNAs control mitochondrial dynamics, of which miR-195, miR-124, and miR-455-3p have also been studied in AD patients. Additionally, several ncRNAs were found to indirectly regulate mitochondrial fission, autophagy, and apoptosis, although the underlying mechanisms require further characterization. Thus, while ncRNA-centered AD research is in its early stages, current mechanistic and translational evidence supports mitochondrially relevant ncRNAs as promising candidates for biomarker and therapeutic development.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1007/s12035-026-06198-9.Keywords: Alzheimer’s disease, Mitochondrial dysfunction, Non-coding RNA, MicroRNA, Long non-coding RNA, Circular RNA Introduction Alzheimer’s disease (AD) accounts for 70% of dementia cases worldwide [1], with 58 million cases documented as of 2020 and an estimated threefold increase in prevalence by 2050 [2]. AD is characterized by tau hyperphosphorylation [3], amyloid-beta (Aβ) plaque buildup [4, 5], and synaptic failure [6]. Additional pathophysiological contributors include oxidative stress [7], neuroinflammation [8], and mutations in the genes β-amyloid precursor protein (APP), presenilin 1 (PSEN1), presenilin 2 (PSEN2), and apolipoprotein E (APOE) [9]. Early diagnosis and disease-modifying treatment in AD are of great research interest in order to reduce disability and to alleviate healthcare system burden [10, 11]. Non-coding RNA (ncRNA) is a type of RNA which is not translated into proteins, but instead helps regulate gene expression [12]. As shown in Fig. 1, ncRNAs can be classified into long non-coding RNAs (lncRNA) and small non-coding RNAs by using a threshold of 200 nucleotides [13–16]. MicroRNAs (miRNAs) belong to the latter class [17] while circular RNAs (circRNAs) are a distinct category of ncRNAs with a closed-loop structure [18, 19]. LncRNAs induce various epigenetic modifications [20–22]. By contrast, miRNAs bind to complementary mRNA sequences to inhibit their translation [23, 24] and circRNAs act as a “sponge” for miR, thereby preventing it from repressing target messenger RNA (mRNA) [25]. LncRNAs influence amyloid production, tau hyperphosphorylation, mitochondrial dysfunction, oxidative stress, and synaptic impairment [26]. Accumulating evidence indicates that ncRNAs play an important role in AD-related oxidative stress [27, 28] and that AD may benefit from the usage of ncRNAs as biomarkers and therapeutic targets [29, 30]. Main non-coding RNA classes relevant to mitochondrial dysfunction in Alzheimer’s disease Alterations in mitochondrial biogenesis, dynamics, mitophagy, proteostasis, and mitochondrial DNA (mtDNA) have been observed in AD [31]. However, it is unclear whether mitochondrial defects precede or are a consequence of amyloid pathology [32]. On the one hand, Aβ and hyperphosphorylated tau alter neuronal energy production and antioxidant defense mechanisms [31], which subsequently disturbs mitochondrial physiology, cytochrome oxidase activity, and intracellular calcium homeostasis [32]. On the other hand, mitochondrial reactive oxygen species (ROS) favor Aβ buildup [33]. Although numerous reviews analyze the role of ncRNAs in AD pathogenesis in the general sense [28, 34–37], there is a scarcity of literature summarizing the data available on ncRNA involvement in mitochondrial dysfunction in AD. To our knowledge, only one narrative review published in 2024 has investigated the mechanistic links between the various ncRNA classes, amyloid pathology and mitochondrial dysfunction in AD [38]. Three other studies reviewed the role of mitochondrially-localized microRNAs in AD pathology [39–41]. The purpose of this paper is to provide a systematic and updated overview on the state of ncRNA research in the field of AD with a focus on mitochondrial dysfunction and potential therapeutic and diagnostic applications. Methods Search Strategy A systematic literature search of the PubMed, Embase, and Web of Science Core Collection databases was conducted using search strings combining ‘Alzheimer’ and ‘mitochondrial’ with Boolean operators and the following ncRNA terms: ‘microRNA’, ‘miRNA’, ‘miR’, ‘long non-coding RNA’, ‘lncRNA’, ‘non-coding RNA’, ‘circular RNA’, ‘circRNA’, ‘small nucleolar RNA’, ‘snoRNA’, ‘short hairpin RNA’, ‘small hairpin RNA’ and ‘shRNA’. Studies published before 18 November 2025 were assessed for eligibility. Study Eligibility Criteria Articles investigating the pathological mechanisms, therapeutic or biomarker applications of the selected ncRNA classes, or their regulatory enzymes in patients or experimental AD models were included. The inclusion criteria were (1) studies written in English; (2) research conducted on human patients or experimental AD models; (3) studies investigating endogenous ncRNAs or synthetic ncRNA-based reagents that mimic in vivo processes in relation to mitochondrial parameters in AD; (4) investigations addressing the pathological mechanisms of ncRNAs or their regulatory enzymes in relation to mitochondrial dysfunction in AD; (5) investigations evaluating therapeutic applications of the selected ncRNA classes or their regulatory enzymes in relation to mitochondrial function in AD; (6) investigations evaluating the biomarker applications of the selected ncRNA classes or their regulatory enzymes in AD; and (7) original research articles. The exclusion criteria were (1) conference/meeting abstracts; (2) non-English publications; (3) reviews and meta-analyses; (4) bibliometric analyses and editorials; (5) articles investigating diseases other than AD; (6) articles whose primary outcomes were unrelated to ncRNA expression or ncRNA biogenesis in AD; (7) articles whose primary outcomes included mitochondrial parameters but not ncRNAs or ncRNA biogenesis in AD; (8) articles whose models induced cognitive decline via secondary AD pathogenetic mechanisms without reproducing essential histopathological characteristics of AD (e.g., chemically induced AD models via scopolamine, AlCl3 and streptozotocin [42]). Data Extraction Screening was performed independently by three reviewers (S.M.M.R., V.R. and S.R.P.). Conflicts were resolved by a fourth experienced reviewer (I.B.N.). The extracted data were organized into tables featuring publication details, experimental model, patient cohort or bioinformatics dataset, sample type, RNA quantification method, and mitochondria-relevant results. Given the heterogeneity of models, outcomes, and sample sizes, a meta-analysis and risk of bias assessment were not performed. Results and Discussion Database Screening and Characteristics of the Included Studies Article identification and selection following PRISMA guidelines [43] are summarized in Fig. 2. PubMed, Embase, and Web of Science Core Collection searches retrieved 2578 records. After automatic duplicate identification using Rayyan [44] and manual assessment, 1787 duplicates were removed. Screening yielded 42 studies, an overview of which is provided in Table 1. PRISMA flowchart illustrating the identification and screening process of the studies included in this review Table 1Overview of studies included in this systematic review: design, models, and main ncRNA findingsNo.ReferenceYearStudy typeModel and sample typeRNA quantification methodsNcRNA class studiedPrimary objective 1.[45]2019FundamentalCell-based modelRT-qPCRmiRNAlncRNAAβ25–35 ↑ SNHG1;SNHG1 knockdown restores MMP and reduces apoptosis via the miR-137/KREMEN1 axis 2.[46]2020MixedaHuman (brain)Rodent (brain)Cell-based modelRT-qPCRmiRNAlncRNA↓ WT1-AS and ↑ miR-375 in AD;WT1-AS overexpression increases MMP/ATP and reduces oxidative stress via the WT1/miR-375/SIX4 axis 3.[47]2021FundamentalRodent (brain)Cell-based modelRT-qPCRlncRNAmiRNAAβ25–35 ↑ H19 and ↓ miR-129;H19 silencing restores MMP and reduces oxidative stress via the miR-129/HMGB1 axis 4.[48]2025FundamentalCell-based modelRT-qPCRmiRNAcircRNAAβ ↑ circ_0049472 and ↓ miR-22-3p;circ_0049472 knockdown restores mitochondrial function via the miR-22-3p/PDE4A axis 5.[49]2023FundamentalCell-based modelPoly(A) and total RNA-seqlncRNAcircRNAPSEN1 A79V/L150P neurons: ↓ MEG3, LOXL1-AS1, A2M-AS1;CircRNAs predominantly ↑;↑ CDR1as/circ_0001946 in L150P 6.[50]2020FundamentalRodent (brain)Cell-based modelRT-qPCRmiRNAAPP/PS1 mice/Aβ42: ↓ Dicer1 via Nrf2 repression;Dicer1 loss increases oxidative stress and decreases MMP;Dicer1 overexpression is neuroprotective 7.[51]2022TranslationalaHuman (brain)N/AmiRNA↓miR-132-3p, miR-212-3p in AD;miR-9-5p variably dysregulated in AD 8.[52]2025TranslationalaHuman (brain)N/AmiRNAlncRNA↓ miR-129-5p and miR-132 in AD brains;Higher miR-129-5p/miR-132 associated with greater brain resilience;miR-124-3p, NEAT1 and PTEN emerged as key nodes in a mitochondrial-related ceRNA network 9[53]2025MixedaHuman (brain, blood)Cell-based modelN/AmiRNA↓ HIBCH and MGME1 in AD;miR-922 directly targets MGME1; miR-98-5p does not 10.[54]2025TranslationalaHuman (brain)N/AmiRNAmiR-103 and miR-484 predicted to co-regulate AD-relevant OXPHOS and TCA-cycle genes 11.[55]2021TranslationalHuman (brain)RT-qPCRmiRNA↑ miR-485 in AD frontal cortex;Associated with ↓ PGC-1α and impaired antioxidant response 12.[56]2024TranslationalaHuman (brain)N/AmiRNAmiR-106b-5p, miR-17-5p and miR-16-5p linked to altered protein networks in AD hippocampus;Protein alterations associated with impaired mitochondrial function 13.[57]2020TranslationalHuman (brain)miRNA array-based analysismiRNA↑ miR-9, miR-34a, miR-146a and miR-155 in AD temporal cortex;All target the SIRT1 3′UTR, consistent with ↓ SIRT1 14.[58]2021MixedHuman (plasma, CSF, brain)Cell-based modelmiRNA microarray;RT-qPCRmiRNA↑ miR-1273g-3p in AD plasma and CSF;miR-1273g-3p suppresses TIMM13, GLRX5 and MTCH1 and induces mitochondrial dysfunction15.[59]2022TranslationalHuman (brain)RT-qPCRmiRNA↑ miR-146a in AD temporal cortexmiR-146a inversely associated with VDAC and impairs oxidative phosphorylation 16.[60]2024MixedHuman (brain)Rodent (brain)RT-qPCRmiRNA↓ miR-218-5p and miR-124-3p in AD hippocampus;Loss of miR-218-5p/miR-124-3p increases RTN3 17.[61]2024MixedHuman (brain)Rodent (brain)Cell-based modelmiRNA microarray;RT-qPCRmiRNA↑ miR-323-5p and miR-370-3p in AD-model mice;miR-150-3p, miR-323-5p and miR-370-3p ↓ OPA1;miR-150-3p directly targets MFN2 18.[62]2024TranslationalaHuman (brain)N/AmiRNAAD-associated miRNAs predicted to target autophagy/mitochondrial biomarkers TFEB, TOMM20 and GABARAPL1 19.[63]2021FundamentalRodent (brain)Cell-based modelRT-qPCRmiRNA↑ miR-351-5p and ↓ Miro2 in AD-model hippocampus;miR-351-5p suppresses Miro2 20.[64]2024MixedHuman (brain)Rodent (brain)Cell-based modelRT-qPCRmiRNA↑ miRNA PC-5P-12969 in AD brains and models;PC-5P-12969 targets GSK3α and improves mitochondrial respiration/ATP while reducing fragmentation 21.[65]2025FundamentalRodent (brain)Cell-based modelNanostring miRNA expression assay; RT-qPCRmiRNAAPP/PSEN1 mice: ↑ miR-7a, miR-148a. ↓ miR-34a, miR-128a;miR-7a targets Klf4, reducing ferroptosis-associated mitochondrial fragmentation and MMP loss 22.[66]2025FundamentalRodent (brain)Cell-based modelRT-qPCRmiRNA↑ miR-9-5p and ↓ SIRT1 in APP/PS1 hippocampus;Rg1 ↓ miR-9-5p, restores SIRT1, improves mitochondrial function/mitophagy 23.[67]2025FundamentalCell-based modelRT-qPCRmiRNAmiR-148a-3p, miR-21-5p, miR-103a-3p derived from neural stem cells exert mitochondrial protective effects in AD 24.[68]2021FundamentalRodent (brain)Cell-based modelRT-qPCRmiRNA↑ miR-140 and ↓ PINK1 in AD models;miR-140 directly targets PINK1;miR-140 silencing improves mitochondrial function and autophagy 25.[69]2020FundamentalCell-based modelRT-qPCRmiRNAAβ25–35: ↓ miR-9-5p and ↑ GSK-3β;miR-9-5p targets GSK-3β, reducing mitochondrial dysfunction and oxidative stress 26.[70]2016FundamentalRodent (brain)Cell-based modelmiRNA microarray;RT-qPCRmiRNASAMP8 hippocampus: ↑ miR-195, ↓ miR-15b;miR-195 directly targets MFN2, causing loss of MMP 27.[71]2021FundamentalRodent (brain)Cell-based modelRT-qPCRmiRNA↑ miR-204 and ↓ TRPML1 in AD models;miR-204 targets TRPML1, promoting ROS production and impaired mitochondrial autophagy 28.[72]2021FundamentalCell-based modelRT-qPCRmiRNAAD neuronal models: ↑ miR-124, miR-125b and miR-21;↑ exosomal miR-124 and miR-155;miR-124 mimic improves mitochondrial membrane potential 29.[73]2022FundamentalCell-based modelRT-qPCRmiRNA↑ miR-124 in APP-Swedish neurons and their secretome;miR-124 mimic preserves mitochondrial membrane potential and reduces inflammatory activation 30.[74]2019FundamentalCell-based modelRT-qPCRmiRNAmiR-455-3p promotes mitochondrial biogenesis and fusion (↓ DRP1/FIS1; ↑ OPA1/MFN1/MFN2), reducing mutant APP-induced mitochondrial fragmentation 31.[75]2024MixedHuman (brain)Rodent (brain)Cell-based modelRNAscope ISH;RT-qPCR; Northern blotlncRNA↓ lncMtDloop in human AD brains and 3xTg mice;lncMtDloop regulates TFAM-dependent mitochondrial transcription, improving mitochondrial homeostasis 32.[76]2018MixedHuman (CSF, serum)Cell-based modelRT-qPCRlncRNA↑ lncRNA-ATB in AD CSF/serum and Aβ-treated cells;lncRNA-ATB suppression restores MMP and reduces oxidative injury via the miR-200/ZNF217 axis 33.[77]2022FundamentalRodent (brain)Cell-based modelRT-qPCRlncRNA↑ MIR600HG in AD mice;MIR600HG promotes NEDD4L-mediated PINK1 degradation, impairing autophagy and mitochondrial function 34.[78]2020FundamentalRodent (brain)Cell-based modelRT-qPCRlncRNA↑ NEAT1 in AD miceNEAT1 promotes NEDD4L-mediated PINK1 ubiquitination, impairing PINK1-dependent autophagy 35.[79]2018FundamentalCell-based modelRT-qPCRlncRNAAβ25–35: ↑ BDNF-AS and ↓ BDNFBDNF-AS silencing restores MMP and reduces mitochondrial apoptosis/oxidative stress 36.[80]2023FundamentalRodent (brain)Cell-based modelRT-qPCRcircRNAAβ1–42: ↑ circ_0004381 and ↓ miR-647;circ_0004381 sponges miR-647 to ↑ PSEN1; its knockdown reduces oxidative stress and mitochondrial dysfunction 37.[81]2023FundamentalCell-based modelmiRNA microarray;RT-qPCRmiRNAAuNP treatment: ↑ miR-21-5p, miR-132, miR-107, miR-369, miR-181c and miR-212↓ miR-34a and miR-138;miR-21-5p/SOCS6 signaling contributes to protection against Aβ-induced mitochondrial injury 38.[82]2024FundamentalRodent (brain)Cell-based modelRT-qPCRmiRNAmiR-375-3p overexpression enhances mitophagy;Reduces Aβ-induced mitochondrial dysfunction and apoptosis via PS1/mTOR signaling 39.[83]2023TranslationalWhole bloodSmall RNA-SequencingmiRNA↓ miR-6501-5p and ↓ miR-143-3p in ADmiR-6501-5p predicted to target ATP5E; miR-143-3p predicted to target NDUFB9, NDUFA9, NDUFS1 and COX4I1 40.[84]2008MixedHuman (brain)Cell-based modelRT-qPCRmiRNABraak stages 5–6 AD vs Braak stages 0–1 controls:Hippocampus: ↑ miR-423Medial frontal gyrus: ↑ miR-423Cerebellum: ↓ miR-98miR-423 and miR-98 target IDH2 41.[85]2025FundamentalRodent (brain)Cell-based modelRT-qPCRmiRNACeO₂-DRP/MA treatment resulted in ↓ relative miRNA-204 levels, increased MMP, and ↓ PINK1, Beclin-1 and LC3B 42.[86]2020FundamentalRodent (brain)Cell-based modelRT-qPCRmiRNAAD osteoblasts: miR-96-5p inhibitor increases Foxo1, decreases Lcn-2Hippocampus of AD mice: miR-96-5p inhibitor reduced oxidative stress markersᵃStudy used an in silico approach instead of/in addition to performing experimental RNA quantificationThis table summarizes the publication year, study type, experimental model, RNA processing method, ncRNA class studied as well as the main findings of each study selected for data extraction after screening. Studies using human tissue like CSF, serum, plasma, blood or human omics datasets were termed “translational,” while in vitro and in vivo studies were referred to as “fundamental.” Studies employing both approaches were termed “mixed studies.” The downregulation and upregulation of ncRNAs, other genes and proteins are represented using downward arrows (↓) and upward arrows (↑).The RNA quantification column indicates only experimental RNA expression assessment methods. N/A indicates that no experimental RNA quantification methods relevant to the study design were reportedAD, Alzheimer’s disease; circRNA, circular RNA; CSF, cerebrospinal fluid; lncRNA, long non-coding RNA; miR, microRNA; N/A, not applicable; ncRNA, non-coding RNA; Poly(A) RNA-seq, polyadenylated RNA sequencing; RNA-seq, RNA sequencing; ROS, reactive oxygen species; RT-qPCR, reverse transcription quantitative polymerase chain reaction; total RNA-seq, total RNA sequencingGene and protein symbols: A2M-AS1, alpha-2-macroglobulin antisense RNA 1; BDNF-AS, brain-derived neurotrophic factor antisense RNA; Dicer1, double-stranded RNA-specific endoribonuclease 1; GABARAPL1, GABA type A receptor-associated protein-like 1; LOXL1-AS1, lysyl oxidase-like 1 antisense RNA 1; MDH2, malate dehydrogenase 2; MEG3, maternally expressed gene 3; MFN2, mitofusin 2; MGME1, mitochondrial genome maintenance exonuclease 1; MIR600HG, miR-600 host gene; Miro2, mitochondrial Rho GTPase 2; NDUFB10, NADH:ubiquinone oxidoreductase subunit B10; NDUFS1, NADH:ubiquinone oxidoreductase core subunit S1; NDUFS2, NADH:ubiquinone oxidoreductase core subunit S2; NDUFS3, NADH:ubiquinone oxidoreductase core subunit S3; NDUFS8, NADH:ubiquinone oxidoreductase core subunit S8; NDUFV1, NADH:ubiquinone oxidoreductase core subunit V1; NEAT1, nuclear paraspeckle assembly transcript 1; OPA1, optic atrophy 1; PDHA1, pyruvate dehydrogenase E1 subunit alpha 1; PINK1, PTEN-induced putative kinase 1; SDHB, succinate dehydrogenase complex iron sulfur subunit B; SNHG1, small nucleolar RNA host gene 1; TFEB, transcription factor EB; TOMM20, translocase of the outer mitochondrial membrane 20; UQCRFS1, ubiquinol-cytochrome c reductase Rieske iron-sulfur polypeptide 1; WT1-AS, Wilms tumor 1 antisense RNA Overview of studies included in this systematic review: design, models, and main ncRNA findings miRNA lncRNA Aβ25–35 ↑ SNHG1; SNHG1 knockdown restores MMP and reduces apoptosis via the miR-137/KREMEN1 axis Human (brain) Rodent (brain) Cell-based model miRNA lncRNA ↓ WT1-AS and ↑ miR-375 in AD; WT1-AS overexpression increases MMP/ATP and reduces oxidative stress via the WT1/miR-375/SIX4 axis Rodent (brain) Cell-based model lncRNA miRNA Aβ25–35 ↑ H19 and ↓ miR-129; H19 silencing restores MMP and reduces oxidative stress via the miR-129/HMGB1 axis miRNA circRNA Aβ ↑ circ_0049472 and ↓ miR-22-3p; circ_0049472 knockdown restores mitochondrial function via the miR-22-3p/PDE4A axis lncRNA circRNA PSEN1 A79V/L150P neurons: ↓ MEG3, LOXL1-AS1, A2M-AS1; CircRNAs predominantly ↑; ↑ CDR1as/circ_0001946 in L150P Rodent (brain) Cell-based model APP/PS1 mice/Aβ42: ↓ Dicer1 via Nrf2 repression; Dicer1 loss increases oxidative stress and decreases MMP; Dicer1 overexpression is neuroprotective ↓miR-132-3p, miR-212-3p in AD; miR-9-5p variably dysregulated in AD miRNA lncRNA ↓ miR-129-5p and miR-132 in AD brains; Higher miR-129-5p/miR-132 associated with greater brain resilience; miR-124-3p, NEAT1 and PTEN emerged as key nodes in a mitochondrial-related ceRNA network Human (brain, blood) Cell-based model ↓ HIBCH and MGME1 in AD; miR-922 directly targets MGME1; miR-98-5p does not ↑ miR-485 in AD frontal cortex; Associated with ↓ PGC-1α and impaired antioxidant response miR-106b-5p, miR-17-5p and miR-16-5p linked to altered protein networks in AD hippocampus; Protein alterations associated with impaired mitochondrial function ↑ miR-9, miR-34a, miR-146a and miR-155 in AD temporal cortex; All target the SIRT1 3′UTR, consistent with ↓ SIRT1 Human (plasma, CSF, brain) Cell-based model miRNA microarray; RT-qPCR ↑ miR-1273g-3p in AD plasma and CSF; miR-1273g-3p suppresses TIMM13, GLRX5 and MTCH1 and induces mitochondrial dysfunction ↑ miR-146a in AD temporal cortex miR-146a inversely associated with VDAC and impairs oxidative phosphorylation Human (brain) Rodent (brain) ↓ miR-218-5p and miR-124-3p in AD hippocampus; Loss of miR-218-5p/miR-124-3p increases RTN3 Human (brain) Rodent (brain) Cell-based model miRNA microarray; RT-qPCR ↑ miR-323-5p and miR-370-3p in AD-model mice; miR-150-3p, miR-323-5p and miR-370-3p ↓ OPA1; miR-150-3p directly targets MFN2 Rodent (brain) Cell-based model ↑ miR-351-5p and ↓ Miro2 in AD-model hippocampus; miR-351-5p suppresses Miro2 Human (brain) Rodent (brain) Cell-based model ↑ miRNA PC-5P-12969 in AD brains and models; PC-5P-12969 targets GSK3α and improves mitochondrial respiration/ATP while reducing fragmentation Rodent (brain) Cell-based model APP/PSEN1 mice: ↑ miR-7a, miR-148a. ↓ miR-34a, miR-128a; miR-7a targets Klf4, reducing ferroptosis-associated mitochondrial fragmentation and MMP loss Rodent (brain) Cell-based model ↑ miR-9-5p and ↓ SIRT1 in APP/PS1 hippocampus; Rg1 ↓ miR-9-5p, restores SIRT1, improves mitochondrial function/mitophagy Rodent (brain) Cell-based model ↑ miR-140 and ↓ PINK1 in AD models; miR-140 directly targets PINK1; miR-140 silencing improves mitochondrial function and autophagy Aβ25–35: ↓ miR-9-5p and ↑ GSK-3β; miR-9-5p targets GSK-3β, reducing mitochondrial dysfunction and oxidative stress Rodent (brain) Cell-based model miRNA microarray; RT-qPCR SAMP8 hippocampus: ↑ miR-195, ↓ miR-15b; miR-195 directly targets MFN2, causing loss of MMP Rodent (brain) Cell-based model ↑ miR-204 and ↓ TRPML1 in AD models; miR-204 targets TRPML1, promoting ROS production and impaired mitochondrial autophagy AD neuronal models: ↑ miR-124, miR-125b and miR-21; ↑ exosomal miR-124 and miR-155; miR-124 mimic improves mitochondrial membrane potential ↑ miR-124 in APP-Swedish neurons and their secretome; miR-124 mimic preserves mitochondrial membrane potential and reduces inflammatory activation Human (brain) Rodent (brain) Cell-based model RNAscope ISH; RT-qPCR; Northern blot ↓ lncMtDloop in human AD brains and 3xTg mice; lncMtDloop regulates TFAM-dependent mitochondrial transcription, improving mitochondrial homeostasis Human (CSF, serum) Cell-based model ↑ lncRNA-ATB in AD CSF/serum and Aβ-treated cells; lncRNA-ATB suppression restores MMP and reduces oxidative injury via the miR-200/ZNF217 axis Rodent (brain) Cell-based model ↑ MIR600HG in AD mice; MIR600HG promotes NEDD4L-mediated PINK1 degradation, impairing autophagy and mitochondrial function Rodent (brain) Cell-based model ↑ NEAT1 in AD mice NEAT1 promotes NEDD4L-mediated PINK1 ubiquitination, impairing PINK1-dependent autophagy Aβ25–35: ↑ BDNF-AS and ↓ BDNF BDNF-AS silencing restores MMP and reduces mitochondrial apoptosis/oxidative stress Rodent (brain) Cell-based model Aβ1–42: ↑ circ_0004381 and ↓ miR-647; circ_0004381 sponges miR-647 to ↑ PSEN1; its knockdown reduces oxidative stress and mitochondrial dysfunction miRNA microarray; RT-qPCR AuNP treatment: ↑ miR-21-5p, miR-132, miR-107, miR-369, miR-181c and miR-212 ↓ miR-34a and miR-138; miR-21-5p/SOCS6 signaling contributes to protection against Aβ-induced mitochondrial injury Rodent (brain) Cell-based model miR-375-3p overexpression enhances mitophagy; Reduces Aβ-induced mitochondrial dysfunction and apoptosis via PS1/mTOR signaling ↓ miR-6501-5p and ↓ miR-143-3p in AD miR-6501-5p predicted to target ATP5E; miR-143-3p predicted to target NDUFB9, NDUFA9, NDUFS1 and COX4I1 Human (brain) Cell-based model Braak stages 5–6 AD vs Braak stages 0–1 controls: Hippocampus: ↑ miR-423 Medial frontal gyrus: ↑ miR-423 Cerebellum: ↓ miR-98 miR-423 and miR-98 target IDH2 Rodent (brain) Cell-based model Rodent (brain) Cell-based model AD osteoblasts: miR-96-5p inhibitor increases Foxo1, decreases Lcn-2 Hippocampus of AD mice: miR-96-5p inhibitor reduced oxidative stress markers ᵃStudy used an in silico approach instead of/in addition to performing experimental RNA quantification This table summarizes the publication year, study type, experimental model, RNA processing method, ncRNA class studied as well as the main findings of each study selected for data extraction after screening. Studies using human tissue like CSF, serum, plasma, blood or human omics datasets were termed “translational,” while in vitro and in vivo studies were referred to as “fundamental.” Studies employing both approaches were termed “mixed studies.” The downregulation and upregulation of ncRNAs, other genes and proteins are represented using downward arrows (↓) and upward arrows (↑).The RNA quantification column indicates only experimental RNA expression assessment methods. N/A indicates that no experimental RNA quantification methods relevant to the study design were reported AD, Alzheimer’s disease; circRNA, circular RNA; CSF, cerebrospinal fluid; lncRNA, long non-coding RNA; miR, microRNA; N/A, not applicable; ncRNA, non-coding RNA; Poly(A) RNA-seq, polyadenylated RNA sequencing; RNA-seq, RNA sequencing; ROS, reactive oxygen species; RT-qPCR, reverse transcription quantitative polymerase chain reaction; total RNA-seq, total RNA sequencing Gene and protein symbols: A2M-AS1, alpha-2-macroglobulin antisense RNA 1; BDNF-AS, brain-derived neurotrophic factor antisense RNA; Dicer1, double-stranded RNA-specific endoribonuclease 1; GABARAPL1, GABA type A receptor-associated protein-like 1; LOXL1-AS1, lysyl oxidase-like 1 antisense RNA 1; MDH2, malate dehydrogenase 2; MEG3, maternally expressed gene 3; MFN2, mitofusin 2; MGME1, mitochondrial genome maintenance exonuclease 1; MIR600HG, miR-600 host gene; Miro2, mitochondrial Rho GTPase 2; NDUFB10, NADH:ubiquinone oxidoreductase subunit B10; NDUFS1, NADH:ubiquinone oxidoreductase core subunit S1; NDUFS2, NADH:ubiquinone oxidoreductase core subunit S2; NDUFS3, NADH:ubiquinone oxidoreductase core subunit S3; NDUFS8, NADH:ubiquinone oxidoreductase core subunit S8; NDUFV1, NADH:ubiquinone oxidoreductase core subunit V1; NEAT1, nuclear paraspeckle assembly transcript 1; OPA1, optic atrophy 1; PDHA1, pyruvate dehydrogenase E1 subunit alpha 1; PINK1, PTEN-induced putative kinase 1; SDHB, succinate dehydrogenase complex iron sulfur subunit B; SNHG1, small nucleolar RNA host gene 1; TFEB, transcription factor EB; TOMM20, translocase of the outer mitochondrial membrane 20; UQCRFS1, ubiquinol-cytochrome c reductase Rieske iron-sulfur polypeptide 1; WT1-AS, Wilms tumor 1 antisense RNA The mitochondria-relevant results for each ncRNA class were organized into Tables 2, 3, and 4. Studies describing cross-class mechanisms were listed in the ncRNA table according to the primary mechanistic focus [45–47]. In all other multi-class ncRNA articles, findings were organized separately into the relevant tables [48, 49]. Tables 2 and 3 were divided into patient-based (“translational”) and model-based (“fundamental”) sub-sections in order to highlight clinically relevant evidence. The complete data extraction in Tables 2, 3, and 4 including sample type and RNA-quantification-methods are present in the Supplementary Material. Statistical data related to ncRNA expression in the form of p-values were reported in the Results column of each table, where available, and only the fold-change values explicitly reported in numerical format in the original studies were documented. Table 2MicroRNA dysregulation and mitochondrial dysfunction in AD: overview of translational and fundamental studiesMiRNA—translational studiesReferenceSubjects or experimental modelResults[51]AD patients and controlsMeta-analysis:- 36 transcriptomic datasets (7 for RNA-seq and 29 for microarray)- 45 differential miR expression studiesAD vs. controls:- ↓ miR-132-3p (logOR = − 5.08; 95% CI − 6.41 to − 3.74; FDR = 1.19 × 10⁻12)- ↓ miR-212-3p (logOR = − 6.37; 95% CI − 7.53 to − 5.21; FDR = 2.38 × 10⁻25)- miR-9-5p frequently reported as differentially expressed, but not among the statistically significant meta-analytic DEmiRNAs[52]Discovery datasets:- Religious orders study and the rush memory and aging project- miR cohorts: control (n = 180), MCI (n = 111), AD (n = 147), other (n = 12)- mRNA cohorts: control (n = 200), MCI (n = 165), AD (n = 251), other (n = 12)Validation dataset:- Alzheimer’s disease neuroimaging initiative- mRNA cohorts: control (n = 246), MCI (n = 383), AD (n = 114)AD vs controls:- ↓ miR-129-5p (FDR < 0.01)- ↓ miR-132 (FDR < 0.01)ceRNA regulatory axes:- miR-132-3p–NEAT1–circ_0132052- miR-129-5p–MALAT1–circ_0066962[53]eTraining set: GSE5281n = 87 controlsn = 74 AD samplesAD vs. controls:- ↓ HIBCH and MGME1 (> twofold; p < 0.001)- miR-922 predicted and experimentally validated as a direct regulator of MGME1Validation set: GSE63061n = 139 AD samplesn = 249 controls[54]GSE118553GSE48350GSE33000n = 377 AD patientsn = 223 controlsAD vs. controls:- miR-103 and miR-484 identified as predicted regulators of mitochondrial hub genes (UQCRFS1, NDUFS2, NDUFS3, NDUFS1, SDHB, MDH2, PDHA1, NDUFS8, NDUFB10, and NDUFV1)- miR-103 showed a reported interaction with NDUFV1; the role of miR-103/miR-484 in AD remains insufficiently characterized[55]n = 10 ADn = 12 controlsn = 12 nondemented with Alzheimer’s neuropathology (NDAN)AD vs controls and NDAN:- ↑ miR-485-5p vs controls (p = 0.007)- ↑ miR-485-5p vs NDAN (p = 0.003)- no significant difference between controls and NDAN (p = 0.0909)- ↓ PGC1α in AD; elevated miR-485-5p was proposed to contribute to PGC1α inhibition[56]NeuroPro database: 11 proteomic studies of human AD hippocampal tissueAD hippocampus:- miR-106b-5p, miR-17-5p and miR-16-5p identified as regulators associated with altered proteins- associated target proteins enriched in ribosomal function, autophagy and immune-system pathways[57]n = 3 AD patients (79.3 ± 9.2 years)n = 3 controls (78.4 ± 5.7 years)AD vs age-, sex- and PMI-matched controls:- ↑ miR-9, miR-34a, miR-146a and miR-155 (1.7- to 3.3-fold)- miR-9, miR-34a, miR-146a and miR-155 have recognition sites in the SIRT1 mRNA 3′UTR- their upregulation was proposed to contribute to ↓ SIRT1 expression in AD[58]en = 36 controlsn = 24 aMCIn = 36 ADAD vs. controls (|FC|> 1.5; p < 0.05):- ↑ miR-1273g-3p- ↑ miR-8063- ↓ miR-6798-3p- ↓ miR-106b-5p- ↓ miR-363-3paMCI vs controls (|FC|> 1.5; p < 0.05):- ↑ miR-1273g-3p- ↑ miR-375n = 31 controln = 12 PSAD (presymptomatic AD)n = 20 PDAD (prodromal AD)n = 20 ADAD vs. controls:- ↑ miR-1273g-3p (p < 0.05)AD vs. PSAD:- ↑ miR-1273g-3p (p < 0.01)PDAD vs. controls:- no significant difference in miR-1273g-3pn = 13 controlsn = 10 PSADn = 13 PDADn = 14 ADPSAD vs. controls:- ↑ miR-1273g-3p (p < 0.01)PDAD vs. controls:- ↑ miR-1273g-3p (p < 0.01)AD vs. controls:- ↑ miR-1273g-3p (p < 0.001)CSF miR-1273g-3p negatively correlated with CSF Aβ42 (r = −0.501; p < 0.001)Prefrontal cortex RNA-seq dataset:n = 383 ADn = 165 controlsPost-mortem hippocampus:n = 8 ADn = 5 controlsAD vs. controls:- ↓ GLRX5, MTCH1 and TIMM13 mRNA in prefrontal cortex- ↓ TIMM13 protein in hippocampal CA1, CA2 and CA3[59]en = 13 AD patientsn = 10 controlsAD vs. age-matched controls:- ↑ miR-146a (3.6-fold; p ≤ 0.05)- miR-146a negatively correlated with VDAC1 protein levels (ρ = −0.80; p ≤ 0.01)[60]en = 6 ADn = 6 controlsAD vs. controls:- ↓ miR-218-5p and miR-124-3p (p < 0.05)- ↑ RTN3 protein (p < 0.001)[61]eGSE173955n = 8 AD (hippocampus)n = 10 controlsAD vs. controls:- ↓ OPA1 (p = 0.031)- ↓ MFN2, not significant (p = 0.058)AD vs. controls:- ↓ OPA1 (p = 0.029)- ↓ MFN2 (p = 0.00042)GSE183260n = 19 AD (whole brain)n = 21 controls[62]GSE5281:n = 87 ADn = 74 controlsValidation dataset GSE132903:n = 97 ADn = 98 controlsmiRDB/HMDD databases:hub gene-targeting AD-associated miRNAsGSE5281, AD vs. controls:- ↑ TFEB (logFC = 1.80; Padj = 1.75 × 10−14)- ↓ TOMM20 (logFC = −1.01; Padj = 1.77 × 10−6)- ↓ GABARAPL1 (logFC = −1.01; Padj = 1.57 × 10−9)Expression patterns of TFEB, TOMM20 and GABARAPL1 confirmed in GSE132903 (p < 0.001)miRDB/HMDD:- TFEB predicted targets: miR-29a-3p, miR-124-3p, miR-29b-3p, miR-29c-3p- TOMM20 predicted targets: let-7a-2-3p, miR-204-3p- GABARAPL1 predicted targets: miR-195-5p, miR-16-5p, miR-155-5p, miR-497-5p, miR-15b-5p, miR-143-3p, miR-133b[64]en = 27 AD (65–95 years old)n = 15 controls (65–103 years old)AD vs. controls:- ↑ miRNA PC-5P-12969 (p = 0.0008)[83]Statistical and machine learning analysisN = 7 AD (> 65 years old)N = 4 controls (> 65 years old)AD compared to controls (peripheral blood):↓miR-6501-5p (log2FC = − 1.34159, p = 0.003957)↓miR-143-3p (log2FC = − 0.69404, p = 0.048802)Predicted mitochondrial target genes:- miR-6501-5p: predicted to target ATP5E- miR-143-3p: predicted to target NDUFB9, NDUFA9, NDUFS1 and COX4I1Associated mitochondria-relevant terms as per Gene Ontology (GO) analysis:- miR-6501-5p: respiratory electron transport chain; generation of precursor metabolites and energy- miR-143-3p: respiratory electron transport chain; generation of precursor metabolites and energy; mitochondrial inner membrane; mitochondrial electron transport; NADH dehydrogenase activity; mitochondrial respiratory chain complex I[84]eHippocampus,medial frontal gyrus, and cerebellum samplesfrom age-matched n = 27 subjects:n = 2 Braak stage 0 (non-demented)n = 5 Braak stage 1 (non-demented)n = 5 Braak stage 3 (non-demented)n = 5 each Braak stages 4, 5, and 6 (AD)Braak stages 5–6 AD vs. Braak stages 0–1 controls:Hippocampus: ↑ miR-423 (p < 0.05)Medial frontal gyrus: ↑ miR-423 (p < 0.05)Cerebellum: ↓ miR-98 (p < 0.05)MiRNA—fundamental studiesReferenceSubjects or experimental modelResults[48]SK-N-SH and CHP212 cells treated with AβAβ-treated vs. control cells:- ↓ miR-22-3p (p < 0.001)Mechanistic validation:- miR-22-3p mimic vs. miR-NC: ↓ PDE4A- miR-22-3p directly targets the PDE4A 3′UTR- miR-22-3p overexpression protected against Aβ-induced neuronal injury[50]APP/PS1 miceAPP/PS1 vs. WT littermates:- ↓ Dicer1 mRNA (4 months: p = 0.0335; 11 months: p = 0.0167)- ↓ Dicer1 protein (4 months: p = 0.0092; 11 months: p = 0.0002)- ↓ Nrf2 mRNA (4 months: p = 0.0304; 11 months: p = 0.0195)- ↓ Nrf2 protein (4 months: p = 0.0001; 11 months: p = 0.0038)- ↑ activated caspase-3 protein (4 months: p = 0.0098; 11 months: p = 0.0008)Primary murine hippocampal neuronsDicer1 siRNA vs NC siRNA:- increased ROS (p < 0.0001)- decreased MMP (p = 0.0399)- ↑ activated caspase-3 protein (p = 0.020)Neuro-2a cellsAβ42 vs. vehicle:- ↓ Dicer1 protein (p = 0.0004)- ↓ Nrf2 proteinNrf2 overexpression vs. vehicle plasmid:- ↑ Dicer1 mRNA (p = 0.0482)- ↑ Dicer1 protein (p = 0.024)Mechanistic validation:- Nrf2 directly regulates Dicer1 transcription through ARE sites in the Dicer1 promoter[53]293 T cells co-transfected with MGME1 3′UTR luciferase reporter constructs and miR-922 or miR-98-5pMGME1-3′UTR-WT + miR-922 vs. MGME1-3′UTR-WT + miR-NC:- decreased luciferase activity by 42% (p = 0.003)- confirmed direct binding of miR-922 to MGME1 3′UTRMGME1-3′UTR + miR-98-5p:- no specific binding to MGME1[58]H4-APPswe and SH-SY5Y cellsH4-APPswe and SH-SY5Y + miR-1273g-3p mimic vs. mimic NC:- ↑ miR-1273g-3p (p < 0.01)- decreased MMP and mitochondrial respiration- increased mitochondrial fragmentation and oxidative stress- ↓ GLRX5, MTCH1 and TIMM13Rescue experiments:- restored mitochondrial respiration and reversed BACE1/Aβ42-related changes[59]Primary cortical neurons from Sprague Dawley rats treated with AβAβ-treated vs. untreated primary cortical neurons:- ↑ intracellular miR-146a (p ≤ 0.05)- ↑ EV miR-146a (p ≤ 0.0001)Primary mixed glial cells from Sprague Dawley ratsAβ-treated vs untreated primary mixed glial cells:- ↑ intracellular miR-146a (p ≤ 0.0001)- ↑ EV miR-146a (p ≤ 0.01)miR-146a overexpression vs. vector control in primary mixed glia:- ↑ intracellular miR-146a (> twofold)- ↑ EV miR-146a (53–116-fold)- decreased ATP-linked respiration, maximal respiration and spare respiratory capacity, glycolysis, glycolytic capacity and glycolytic reserve[60]Male 3 × Tg-AD mice subjected to social isolationAdult male C57BL/6 mice (wild-type) subjected to social isolationSocially isolated vs. group-housed AD and WT mice:- ↓ miR-218-5p and miR-124-3pSocially isolated AD mice + miR-218-5p/miR-124-3p agomirs vs. scramble:- restored DG-CA3 synaptic plasticity and pattern separationMechanistic validation:- miR-218-5p and miR-124-3p directly bind the Rtn3 3′UTR and posttranscriptionally suppress RTN3 protein- RTN3 upregulation/aggregation disrupted mitochondrial distribution in presynaptic boutons[61]3xTg-AD mice3 × Tg-AD vs. WT mice:- ↑ miR-323-5p and miR-370-3p (> twofold, p < 0.001)- ↓ OPA1 and MFN2 proteinAdult rat hippocampal neural stem/progenitor (HCN) cells- insulin withdrawal- miR-150-3p, miR-323-5p or miR-370-3p transfectionInsulin withdrawal vs. control:- ↑ miR-150-3p, miR-323-5p and miR-370-3pmiRNA transfection vs control:- ↓ OPA1- miR-150-3p additionally ↓ MFN2- decreased mitochondrial respiration and ATP production- increased mitochondrial fissionMechanistic validation:- miR-150-3p directly binds the MFN2 3′UTR (30% reduction in reporter activity, p < 0.001)- OPA1 is a potential target of all three miRNAs[63]HCN cellsmiR-351-5p mimic vs. miR-Con:- ↓ Miro2 mRNA by 71% (p < 0.0001)- ↓ Miro2 protein by 41% (p < 0.01)- increased mitochondrial fission, mitophagy, autophagy-dependent cell death- decreased MMP and ATP productionMechanistic validation:- miR-351-5p directly targets the Miro2 3′UTR (37% reduction in reporter activity; p < 0.001)AD model: 5xFAD mice, 3xTg-AD miceWT mice3xTg-AD and 5xFAD vs. WT mice:- ↑ miR-351-5p- ↓ Miro2[64]APP transgenic mice (Tg2576), 6 months old (n = 5)Tau transgenic mice (P301L), 6 months old (n = 5)Aβ knock-in mice (hA-KI), 12 months old (n = 5)Wild-type C57BL/6 mice (n = 5)vs respective WT controls:- ↑ miRNA PC-5P-12969 in APP mice (p = 0.0117)- ↑ miRNA PC-5P-12969 in Tau mice (p = 0.0110)- no significant change in 6-month hA-KI mice (p = 0.9606)- ↑ miRNA PC-5P-12969 in 12-month hA-KI mice (p = 0.0362)HT22 cells (untreated)HT22 cells + miRNA PC-5P-12969 vectorHT22 cells + mAPP plasmid vectorHT22 cells + miRNA PC-5P-12969 + mAPP vectors (pre-condition)HT22 cells + mAPP + miRNA PC-5P-12969 vectors (post-condition)vs untreated HT22 cells:- ↑ miRNA PC-5P-12969 with mAPP (p < 0.0001)- ↑ miRNA PC-5P-12969 with miRNA PC-5P-12969 vector (p < 0.0001)- ↑ miRNA PC-5P-12969 in post-condition (p = 0.0001)- no significant change in pre-condition (p = 0.9693)Post-condition vs. mAPP:- ↓ GSK3α and APP- increased ATP production and maximal mitochondrial respiration- increased mitochondrial length- decreased mitochondrial fragmentationMechanistic validation:- miRNA PC-5P-12969 directly binds the GSK3α 3′UTR (p < 0.0001)[65]n = 3 APP/PSEN1 double transgenic micen = 3 wild-type (WT) miceAPP/PSEN1 vs. WT mice:- cortex: ↑ miR-7a (2.05-fold); ↓ miR-34a- hippocampus: ↑ miR-7a and miR-148a; ↓ miR-128a- ↑ Klf4 protein (p = 0.015 by immunofluorescence; p < 0.05 by Western blot)N9 microglial cells primed with lipopolysaccharide, followed by treatment with Aβ oligomers for 24 hLPS-primed N9 + AβO + miR-7a mimic vs. LPS-primed N9 + AβO:- decreased ROS- ↓ iNOS and NLRP3SH-SY5Y cells treated with RAS-selective lethal 3 and vehicle (ferroptosis model)RSL3 + miR-7a mimic vs. RSL3:- ↓ Klf4- decreased ROS- restored normal long tubular mitochondrial morphology- increased MMP (complete recovery at 250 nM miR-7a)Mechanistic validation:-miR-7a directly binds the Klf4 3′UTR and suppresses Klf4SH-SY5Y cellsmiR-7a mimic vs negative control: ↓ Klf4miR-7a inhibitor treatment: ↑ Klf4[66]n = 42 APP/PS1 micen = 6 wild type C57BL/6J miceAD vs. WT:- ↑ miR-9-5p (p < 0.001)- ↓ SIRT1- decreased ATP, MMP, and mitochondrial respiratory functionAD + Rg1 vs. AD + Vehicle 1:- ↓ miR-9-5p (p < 0.001)- ↑ SIRT1- improved mitochondrial functionAD + Rg1 + miR-9-5p agomir vs. AD + Rg1 + agomir NC:- ↑ miR-9-5p (p < 0.05)- ↓ SIRT1- increased mitochondrial dysfunctionMouse hippocampal neuron HT-22 treated with Aβ1–42 (20 µM)Aβ1–42 vs. blank:- ↑ miR-9-5p (p < 0.001)- ↓ SIRT1- ↓ PINK1, Parkin, LC3B II/I, and Beclin-1Aβ1–42 + Rg1 vs. Aβ1–42 + Vehicle 1:- ↓ miR-9-5p (p < 0.001)- ↑ SIRT1- ↑ PINK1, Parkin, LC3B II/I and Beclin-1Aβ1–42 + Rg1 + miR-9-5p mimic vs Aβ1–42 + Rg1 + mimic NC:- ↑ miR-9-5p (p < 0.05)- ↓ SIRT1- ↓ PINK1, Parkin, LC3B II/I and Beclin-1- decreased ATP, MMP, and mitochondrial respiratory function[67]Primary human neuronal cultures from human induced pluripotent stem cells (hiPSCs) exposed to 1 µM Aβ−42ohiPSC-NSC-EVs (containing miR-148a-3p, miR-21-5p and miR-103a-3p) + Aβ−42o vs Aβ−42o alone:- decreased mitochondrial superoxide- ↓ BAD and BAX; ↑ BCL-2- increased MMP and preserved TOM20 + mitochondria- ↑ Beclin-1 and MAP1LC3B- normalized mitochondrial complex gene expression[68]Male Sprague–Dawley rats, induced via aggregated injection Aβ1–42 into the hippocampusAD vs. normal rat hippocampus:- ↑ miR-140 (p < 0.05)- ↓ PINK1miR-140 antagomir vs. NC-antagomir:- ↑ PINK1- increased autophagyPrimary hippocampal neurons from 1-day-old Sprague–Dawley rats treated with Aβ-derived diffusible ligandsAβDDLs-treated vs control neurons:- ↑ miR-140 (p < 0.05)- ↓ PINK1- increased ROS and mitochondrial dysfunction; decreased autophagymiR-140 inhibitor vs. NC-inhibitor:- ↓ miR-140 (p < 0.05)- ↑ PINK1- increased MMP and autophagy; decreased ROS[69]Mouse hippocampal neuron cell line HT22 treated with Aβ25–35Aβ25–35-treated vs untreated HT22 cells:- ↓ miR-9-5p (p < 0.05)- ↑ GSK-3βAβ25–35 + miR-9-5p mimic vs Aβ25–35 + NC mimic:- ↑ miR-9-5p (p < 0.01)- ↓ GSK-3β- decreased ROS/oxidative stress and apoptosis- ↓ cytochrome C, AIF, cleaved caspase-3 and BAX; ↑ BCL-2[70]SAMP8 mice aged 3-, 6- and 9- months (hippocampus)SAMP8 vs age-matched SAMR1 hippocampus:- ↑ miR-195 at 3 and 6 months (p < 0.05); no significant change at 9 months- ↓ miR-15b at 3, 6 and 9 months (p < 0.05)- ↓ MFN2 mRNA and proteinHT-22 hippocampal neuronal cells transfected with miR-195 mimicsmiR-195 mimic vs scramble in HT-22 cells:- ↑ miR-195 (p < 0.05)- ↓ MFN2 protein- decreased MMP[71]Rat hippocampal neurons treated with Aβ1–42Normal hippocampal neurons (control)Aβ1–42-treated vs. control neurons:- ↑ miR-204 (p < 0.05)- ↓ TRPML1- increased mitochondrial dysfunction/ROS and impaired mitochondrial autophagyAβ1–42 + miR-204 inhibitor vs. Aβ1–42 + inhibitor negative-control:- ↑ TRPML1- ↓ PINK1, Parkin, LC3II, and Beclin1- increased MMP; decreased ROS and mitochondrial autophagyAβ1–42 + CLN (STAT3 activation) vs. Aβ1–42 + vehicle:- ↓ miR-204 (p < 0.05)- ↑ TRPML1- increased MMP and ATP; decreased ROS and mitochondrial autophagyAPP/PS1 transgenic AD mouse modelAPP/PS1 vs. normal hippocampus:- ↑ miR-204 (p < 0.05)- ↓ TRPML1- decreased mitochondrial autophagy and increased mitochondrial ROS[72]SH-SWE cells (transfected with human APP695)SH-SWE vs. SH-WT:- ↑ miR-124, miR-125b, and miR-146a (p < 0.05); ↑ miR-21 and miR-155 (p < 0.01)- ↑ exosomal miR-124 (p < 0.05) and miR-155 (p < 0.01)miR-124 mimic vs. mock in SH-SWE:- ↑ miR-124 (70-fold; p < 0.01)- ↓ exosomal miR-125b (p < 0.05), miR-146a and miR-155 (p < 0.01); miR-21 decreased, NS- increased mitochondrial activity/MitoTracker intensitymiR-124 inhibitor vs. mock in SH-SWE:- ↓ miR-124 (≥ 80% decrease; p < 0.01)- ↑ miR-21 (p < 0.05) and miR-146a (p < 0.01); ↓ miR-125b (p < 0.05)- increased MFN2 protein; no significant change in DRP1iNEU-PSEN (iNeurons from iPSCs generated from a patient carrying the PSEN1ΔE9 mutation) and iNEU-wild type from a healthy female controliNEU-PSEN vs. iNEU-WT:- ↑ miR-124, miR-125b, and miR-21 (p < 0.05)- ↓ miR-146a and miR-155 (p < 0.05)- ↑ exosomal miR-124 and miR-155 (p < 0.01)miR-124 mimic vs. mock in iNEU-PSEN:- ↑ miR-124 (~ 12-fold; p < 0.01)- ↓ miR-125b (p < 0.01) and miR-21 (p < 0.05)- ↓ exosomal miR-21 and miR-146a (p < 0.05) and miR-155 (p < 0.01); miR-125b decreased, NSmiR-124 inhibitor vs. mock in iNEU-PSEN:- ↓ miR-124 (≥ 80% decrease; p < 0.01)- ↓ miR-125b (p < 0.05) and miR-21 (p < 0.01); ↑ miR-146a (p < 0.01)[73]APP Swedish SH-SY5Y cells (SWE)SWE vs. wild-type SH-SY5Y (SH):- ↑ miR-124 in cells and secretome (cellular, exosomal, and EXO-free; p < 0.05)Human CHME3 microglial cells treated with IFNγ and co-cultured with SWE cellsIFNγ-microglia co-cultured with SWE vs IFNγ-microglia monoculture:- ↑ miR-124 in microglia (nearly 90-fold; p < 0.001)IFNγ-microglia co-cultured with miR-124-mimic SWE vs IFNγ-microglia co-cultured with mock SWE:- ↑ miR-124 in SWE cells (70-fold; p < 0.01) and secretome (exosomal and EXO-free; p < 0.05)- ↑ miR-124 and ↓ miR-146a in co-cultured microglia (p < 0.01)- ↓ iNOS and TNF-α expression in microglia- increased mitochondrial activity/MitoTracker Red intensity[74]N2a mouse neuroblastoma cells transfected with mutant APP cDNAMutant APP + miR-455-3p vs. mutant APP- increased mitochondrial biogenesis- decreased mitochondrial fragmentation- ↓ mitochondrial fission markers (DRP1 and FIS1)- ↑ mitochondrial fusion markers (OPA1, MFN1, and MFN2)[81]Human neural stem cells (hNSCs) (H9-derived) treated with Aβ1–42 ± gold nanoparticles (AuNPs)Aβ + AuNPs vs. Aβ- ↑ miR-21-5p (FC = 4.78), miR-132, miR-107, miR-369, miR-181c, and miR-212- ↓ miR-34a and miR-138- ↑ DRP1, NRF1, and TFAM mRNA- increased MMP and decreased mitochondrial ROS- decreased apoptosisAβ + AuNPs + miR-21-5p inhibitor vs Aβ + AuNPs + NC- ↓ miR-21-5p (p < 0.01)- ↓ DRP1, NRF1, and TFAM mRNA- decreased cell viability- decreased MMP, increased mitochondrial ROS, and increased apoptosismiR-21-5p mimic vs. mimic-NC- ↑ miR-21-5p (p < 0.01)- ↓ SOCS6 protein- increased cell viability and decreased apoptosis[82]SH-SY5Y cells treated with Aβ25–35SH + Aβ25–35 + miR-375-3p vs. SH + NC + Aβ25–35- decreased Aβ25–35-induced apoptosis- decreased intracellular ROS- increased MMP/JC-1 polymer:monomer fluorescence ratioAPP/PS1 transgenic miceAD + miR-375-3p vs. AD control- ↑ LC3B, Beclin-1, MFN2, and PINK1 mRNAs[85]n = 10 3 × Tg-AD mice receiving PBS injections in the tail veiln = 10 3 × Tg-AD mice receiving CeO2-DRP/MA 1 mmol/L (CeO2 nanoparticles covalently linked with mitochondrial aptamers (MA) and DNA-RNA hybrid chains (DRP) to detect and silence mitochondrial miRNA-204) injections in the tail veinn = 10 C57BL/6 mice3 × Tg-AD treated with CeO2-DRP/MA vs. PBS-treated 3 × Tg-AD mice:- Decreased escape latency (p < 0.05)- Increased swimming speed (Morris water maze)- Decreased depression-related and anxiety-related behaviorsHippocampal tissue of CeO2-DRP/MA treated vs. PBS-treated 3 × Tg-AD mice:- ↓ PINK1, Beclin1, and LC3B mRNA/protein- ↑TRPML1 and p-STAT3 mRNA/proteinSY5Y + Aβ cellsSY5Y + Aβ + CeO2-DRP/MA (1 mmol/L) vs. untreated SY5Y + Aβ cells:- increased cell viability from ~ 50% at 0 mmol/L to 98% at 1 mmol/L after 24 h- increased MMP- ↓ mitochondrial miRNA-204 (p < 0.01)- ↓PINK1, Beclin-1, and LC3B mRNA- ↑TRPML1 and STAT3 mRNAN2aSW cellsCeO2-DRP/MA-treated vs. untreated N2aSW cells:- ↓PINK1, Beclin-1, LC3B mRNA/protein- ↑TRPML1, STAT3 mRNA/protein[84]SK-N-AS (human neuroblastoma cells)Pre-miR-423 vs. negative control-transfected cells:- ↓ IDH2 mRNA (59.33% and 60.19% of control in two independent experiments; p < 0.05)HepG2 (human hepatocellular carcinoma)Pre-miR-98 vs. negative control-transfected cells:- ↓ IDH2 mRNA (70.03% and 43.98% of control in two independent experiments; p < 0.05)[86]Kunming mice:n = 6 normaln = 6 shamn = 66 successfully modeled AD mice after bilateral hippocampal Aβ1–40 injection, divided into 11 groups (n = 6/group)AD-mice derived osteoblasts:- AD + naringin vs. AD + PBS: ↓ miR-96-5p (p < 0.05)- AD + miR-96-5p inhibitor vs. AD + inhibitor-NC: ↓ miR-96-5p (p < 0.05), ↑ Foxo1, ↓Lcn-2- AD + miR-96-5p mimic vs. AD + miR-96-5p mimic-NC: ↓ Foxo1, ↑Lcn-2Hippocampus of AD mice:- AD + miR-96-5p inhibitor vs. AD + inhibitor-NC:- Reduced OH- and MDA- Increased SOD, CAT and GSH-Px- AD + sh-Lcn2 vs. AD + sh-NC: Reduced mitochondrial structural damage (qualitative assessment)- AD + oe-Foxo1 vs. oe-NC:- Reduced mitochondrial structural damage (qualitative assessment)- Reduced OH- and MDA- Increased SOD, CAT and GSH-Px- ↓ Lcn2eStudies marked with a superscript “e” included additional experimental validation directly assessing the reported ncRNA or its proposed regulatory interaction which was accordingly included in the “Fundamental” section of the tableThis table synthesizes the experimental model and mitochondrially relevant results of microRNA-focused studies. Studies using human tissue like CSF, serum, plasma, blood or human omics datasets were termed “translational,” while in vitro and in vivo studies were referred to as “fundamental.” The first part of the table summarizes translational data while the second part of the table summarizes fundamental experimental findings. For those studies which featured both translational and fundamental results, the findings were placed in the respective table segment. The downregulation and upregulation of ncRNAs, other genes and proteins are represented using downward arrows (↓) and upward arrows (↑)3xTg-AD mice, triple-transgenic Alzheimer’s disease model mice; AD, Alzheimer’s disease; aMCI, Aβ-positive amnestic mild cognitive impairment; APP, β-amyloid precursor protein; ATP, adenosine triphosphate; AuNPs, gold nanoparticles; Aβ, amyloid-beta; ceRNA, competing endogenous RNA; CSF, cerebrospinal fluid; DEmiRNA/DEmiRNAs, differentially expressed microRNA(s); EV, extracellular vesicles; HCN, adult hippocampal neural progenitor cells; hiPSC, human induced pluripotent stem cell; hNSC, human neural stem cell; iPSC, induced pluripotent stem cell; lncRNA, long non-coding RNA; MAP, Rush Memory and Aging Project; mAPP, mutant APP; MCI, mild cognitive impairment; miR/miRNA, microRNA; MMP, mitochondrial membrane potential; mRNA, messenger RNA; N/A, not applicable; ncRNA, non-coding RNA; NDAN, nondemented patients with Alzheimer’s neuropathology; PDAD, prodromal Alzheimer’s disease; PSAD, presymptomatic Alzheimer’s disease; RNA-seq, RNA sequencing; ROS, reactive oxygen species; RT-qPCR, reverse transcription quantitative polymerase chain reaction; SAMP8, senescence-accelerated mouse prone 8; WT, wild-type; PBS, phosphate-buffered saline; CeO₂, cerium dioxide; DRP, DNA–RNA hybrid chains; MA, mitochondrial aptamers; DNA, deoxyribonucleic acid; RNA, ribonucleic acid; FAM, fluorescein amidite; IVIS, in vivo imaging system; NC, negative control; MDA, malondialdehyde; SOD, superoxide dismutase; CAT, catalase; GSH-Px, glutathione peroxidase; shRNA, short hairpin RNA; oe, overexpression, RSL3, RAS-selective lethal 3; log2FC, log2 fold change; GO, Gene OntologyGene and protein symbols: AIF, apoptosis-inducing factor; BAD, Bcl-2-associated death promoter; BAX, BCL2-associated X protein; BCL-2, B-cell lymphoma 2; Dicer1, double-stranded RNA-specific endoribonuclease 1; DRP1, dynamin-related protein 1; FIS1, mitochondrial fission protein 1; GABARAPL1, GABA type A receptor-associated protein-like 1; GSK-3β, glycogen synthase kinase-3 beta; GSK3, glycogen synthase kinase 3; HIBCH, 3-hydroxyisobutyryl-CoA hydrolase; iNOS, inducible nitric oxide synthase; Klf4, Krüppel-like factor 4; LC3B, microtubule-associated protein 1 light chain 3B; MAP1-LC3B, microtubule-associated proteins 1A/1B light chain 3B; MDH2, malate dehydrogenase 2; MFN1, mitofusin 1; MFN2, mitofusin 2; MGME1, mitochondrial genome maintenance exonuclease 1; Miro2, mitochondrial Rho GTPase 2; NDUFB10, NADH:ubiquinone oxidoreductase subunit B10; NDUFS1, NADH:ubiquinone oxidoreductase core subunit S1; NDUFS2, NADH:ubiquinone oxidoreductase core subunit S2; NDUFS3, NADH:ubiquinone oxidoreductase core subunit S3; NDUFS8, NADH:ubiquinone oxidoreductase core subunit S8; NDUFV1, NADH:ubiquinone oxidoreductase core subunit V1; NLRP3, NLR family pyrin domain-containing protein 3; NRF1, nuclear respiratory factor 1; NRF2, nuclear factor erythroid 2-related factor 2; OPA1, optic atrophy 1 gene; Parkin, E3 ubiquitin-protein ligase parkin; PDE4A, phosphodiesterase 4 A; PDHA1, pyruvate dehydrogenase E1 subunit alpha 1; PGC-1α, peroxisome proliferator-activated receptor-gamma coactivator 1-alpha; PINK1, PTEN-induced putative protein kinase 1; PSEN1, presenilin 1; RTN3, reticulon 3 gene; SDHB, succinate dehydrogenase complex iron sulfur subunit B; SIRT1, sirtuin 1; SOCS6, suppressor of cytokine signalling 6; STAT3, signal transducer and activator of transcription 3; TFEB, transcription factor EB; TIMM13, translocase of the inner mitochondrial membrane 13; TNF-α, tumor necrosis factor alpha; TOMM20, translocase of the outer mitochondrial membrane 20; TRPML1, transient receptor potential mucolipin-1; UQCRFS1, ubiquinol-cytochrome c reductase Rieske iron-sulfur polypeptide 1; UTR, untranslated region; VDAC1, voltage-dependent anion channel 1, IDH2, isocitrate dehydrogenase 2, Beclin1, beclin 1; FOXO1, forkhead box O1; LCN2, lipocalin 2, ATP5E, ATP synthase F1 subunit epsilon; NDUFB9, NADH:ubiquinone oxidoreductase subunit B9; NDUFA9, NADH:ubiquinone oxidoreductase subunit A9; COX4I1, cytochrome c oxidase subunit 4I1 MicroRNA dysregulation and mitochondrial dysfunction in AD: overview of translational and fundamental studies AD patients and controls Meta-analysis: - 36 transcriptomic datasets (7 for RNA-seq and 29 for microarray) - 45 differential miR expression studies AD vs. controls: - ↓ miR-132-3p (logOR = − 5.08; 95% CI − 6.41 to − 3.74; FDR = 1.19 × 10⁻12) - ↓ miR-212-3p (logOR = − 6.37; 95% CI − 7.53 to − 5.21; FDR = 2.38 × 10⁻25) - miR-9-5p frequently reported as differentially expressed, but not among the statistically significant meta-analytic DEmiRNAs Discovery datasets: - Religious orders study and the rush memory and aging project - miR cohorts: control (n = 180), MCI (n = 111), AD (n = 147), other (n = 12) - mRNA cohorts: control (n = 200), MCI (n = 165), AD (n = 251), other (n = 12) Validation dataset: - Alzheimer’s disease neuroimaging initiative - mRNA cohorts: control (n = 246), MCI (n = 383), AD (n = 114) AD vs controls: - ↓ miR-129-5p (FDR < 0.01) - ↓ miR-132 (FDR < 0.01) ceRNA regulatory axes: - miR-132-3p–NEAT1–circ_0132052 - miR-129-5p–MALAT1–circ_0066962 Training set: GSE5281 n = 87 controls n = 74 AD samples AD vs. controls: - ↓ HIBCH and MGME1 (> twofold; p < 0.001) - miR-922 predicted and experimentally validated as a direct regulator of MGME1 Validation set: GSE63061 n = 139 AD samples n = 249 controls GSE118553 GSE48350 GSE33000 n = 377 AD patients n = 223 controls AD vs. controls: - miR-103 and miR-484 identified as predicted regulators of mitochondrial hub genes (UQCRFS1, NDUFS2, NDUFS3, NDUFS1, SDHB, MDH2, PDHA1, NDUFS8, NDUFB10, and NDUFV1) - miR-103 showed a reported interaction with NDUFV1; the role of miR-103/miR-484 in AD remains insufficiently characterized n = 10 AD n = 12 controls n = 12 nondemented with Alzheimer’s neuropathology (NDAN) AD vs controls and NDAN: - ↑ miR-485-5p vs controls (p = 0.007) - ↑ miR-485-5p vs NDAN (p = 0.003) - no significant difference between controls and NDAN (p = 0.0909) - ↓ PGC1α in AD; elevated miR-485-5p was proposed to contribute to PGC1α inhibition AD hippocampus: - miR-106b-5p, miR-17-5p and miR-16-5p identified as regulators associated with altered proteins - associated target proteins enriched in ribosomal function, autophagy and immune-system pathways n = 3 AD patients (79.3 ± 9.2 years) n = 3 controls (78.4 ± 5.7 years) AD vs age-, sex- and PMI-matched controls: - ↑ miR-9, miR-34a, miR-146a and miR-155 (1.7- to 3.3-fold) - miR-9, miR-34a, miR-146a and miR-155 have recognition sites in the SIRT1 mRNA 3′UTR - their upregulation was proposed to contribute to ↓ SIRT1 expression in AD n = 36 controls n = 24 aMCI n = 36 AD AD vs. controls (|FC|> 1.5; p < 0.05): - ↑ miR-1273g-3p - ↑ miR-8063 - ↓ miR-6798-3p - ↓ miR-106b-5p - ↓ miR-363-3p aMCI vs controls (|FC|> 1.5; p < 0.05): - ↑ miR-1273g-3p - ↑ miR-375 n = 31 control n = 12 PSAD (presymptomatic AD) n = 20 PDAD (prodromal AD) n = 20 AD AD vs. controls: - ↑ miR-1273g-3p (p < 0.05) AD vs. PSAD: - ↑ miR-1273g-3p (p < 0.01) PDAD vs. controls: - no significant difference in miR-1273g-3p n = 13 controls n = 10 PSAD n = 13 PDAD n = 14 AD PSAD vs. controls: - ↑ miR-1273g-3p (p < 0.01) PDAD vs. controls: - ↑ miR-1273g-3p (p < 0.01) AD vs. controls: - ↑ miR-1273g-3p (p < 0.001) CSF miR-1273g-3p negatively correlated with CSF Aβ42 (r = −0.501; p < 0.001) Prefrontal cortex RNA-seq dataset: n = 383 AD n = 165 controls Post-mortem hippocampus: n = 8 AD n = 5 controls AD vs. controls: - ↓ GLRX5, MTCH1 and TIMM13 mRNA in prefrontal cortex - ↓ TIMM13 protein in hippocampal CA1, CA2 and CA3 n = 13 AD patients n = 10 controls AD vs. age-matched controls: - ↑ miR-146a (3.6-fold; p ≤ 0.05) - miR-146a negatively correlated with VDAC1 protein levels (ρ = −0.80; p ≤ 0.01) n = 6 AD n = 6 controls AD vs. controls: - ↓ miR-218-5p and miR-124-3p (p < 0.05) - ↑ RTN3 protein (p < 0.001) GSE173955 n = 8 AD (hippocampus) n = 10 controls AD vs. controls: - ↓ OPA1 (p = 0.031) - ↓ MFN2, not significant (p = 0.058) AD vs. controls: - ↓ OPA1 (p = 0.029) - ↓ MFN2 (p = 0.00042) GSE183260 n = 19 AD (whole brain) n = 21 controls GSE5281: n = 87 AD n = 74 controls Validation dataset GSE132903: n = 97 AD n = 98 controls miRDB/HMDD databases: hub gene-targeting AD-associated miRNAs GSE5281, AD vs. controls: - ↑ TFEB (logFC = 1.80; Padj = 1.75 × 10−14) - ↓ TOMM20 (logFC = −1.01; Padj = 1.77 × 10−6) - ↓ GABARAPL1 (logFC = −1.01; Padj = 1.57 × 10−9) Expression patterns of TFEB, TOMM20 and GABARAPL1 confirmed in GSE132903 (p < 0.001) miRDB/HMDD: - TFEB predicted targets: miR-29a-3p, miR-124-3p, miR-29b-3p, miR-29c-3p - TOMM20 predicted targets: let-7a-2-3p, miR-204-3p - GABARAPL1 predicted targets: miR-195-5p, miR-16-5p, miR-155-5p, miR-497-5p, miR-15b-5p, miR-143-3p, miR-133b n = 27 AD (65–95 years old) n = 15 controls (65–103 years old) AD vs. controls: - ↑ miRNA PC-5P-12969 (p = 0.0008) Statistical and machine learning analysis N = 7 AD (> 65 years old) N = 4 controls (> 65 years old) AD compared to controls (peripheral blood): ↓miR-6501-5p (log2FC = − 1.34159, p = 0.003957) ↓miR-143-3p (log2FC = − 0.69404, p = 0.048802) Predicted mitochondrial target genes: - miR-6501-5p: predicted to target ATP5E - miR-143-3p: predicted to target NDUFB9, NDUFA9, NDUFS1 and COX4I1 Associated mitochondria-relevant terms as per Gene Ontology (GO) analysis: - miR-6501-5p: respiratory electron transport chain; generation of precursor metabolites and energy - miR-143-3p: respiratory electron transport chain; generation of precursor metabolites and energy; mitochondrial inner membrane; mitochondrial electron transport; NADH dehydrogenase activity; mitochondrial respiratory chain complex I Hippocampus, medial frontal gyrus, and cerebellum samples from age-matched n = 27 subjects: n = 2 Braak stage 0 (non-demented) n = 5 Braak stage 1 (non-demented) n = 5 Braak stage 3 (non-demented) n = 5 each Braak stages 4, 5, and 6 (AD) Braak stages 5–6 AD vs. Braak stages 0–1 controls: Hippocampus: ↑ miR-423 (p < 0.05) Medial frontal gyrus: ↑ miR-423 (p < 0.05) Cerebellum: ↓ miR-98 (p < 0.05) Aβ-treated vs. control cells: - ↓ miR-22-3p (p < 0.001) Mechanistic validation: - miR-22-3p mimic vs. miR-NC: ↓ PDE4A - miR-22-3p directly targets the PDE4A 3′UTR - miR-22-3p overexpression protected against Aβ-induced neuronal injury APP/PS1 vs. WT littermates: - ↓ Dicer1 mRNA (4 months: p = 0.0335; 11 months: p = 0.0167) - ↓ Dicer1 protein (4 months: p = 0.0092; 11 months: p = 0.0002) - ↓ Nrf2 mRNA (4 months: p = 0.0304; 11 months: p = 0.0195) - ↓ Nrf2 protein (4 months: p = 0.0001; 11 months: p = 0.0038) - ↑ activated caspase-3 protein (4 months: p = 0.0098; 11 months: p = 0.0008) Dicer1 siRNA vs NC siRNA: - increased ROS (p < 0.0001) - decreased MMP (p = 0.0399) - ↑ activated caspase-3 protein (p = 0.020) Aβ42 vs. vehicle: - ↓ Dicer1 protein (p = 0.0004) - ↓ Nrf2 protein Nrf2 overexpression vs. vehicle plasmid: - ↑ Dicer1 mRNA (p = 0.0482) - ↑ Dicer1 protein (p = 0.024) Mechanistic validation: - Nrf2 directly regulates Dicer1 transcription through ARE sites in the Dicer1 promoter MGME1-3′UTR-WT + miR-922 vs. MGME1-3′UTR-WT + miR-NC: - decreased luciferase activity by 42% (p = 0.003) - confirmed direct binding of miR-922 to MGME1 3′UTR MGME1-3′UTR + miR-98-5p: - no specific binding to MGME1 H4-APPswe and SH-SY5Y + miR-1273g-3p mimic vs. mimic NC: - ↑ miR-1273g-3p (p < 0.01) - decreased MMP and mitochondrial respiration - increased mitochondrial fragmentation and oxidative stress - ↓ GLRX5, MTCH1 and TIMM13 Rescue experiments: - restored mitochondrial respiration and reversed BACE1/Aβ42-related changes Aβ-treated vs. untreated primary cortical neurons: - ↑ intracellular miR-146a (p ≤ 0.05) - ↑ EV miR-146a (p ≤ 0.0001) Aβ-treated vs untreated primary mixed glial cells: - ↑ intracellular miR-146a (p ≤ 0.0001) - ↑ EV miR-146a (p ≤ 0.01) miR-146a overexpression vs. vector control in primary mixed glia: - ↑ intracellular miR-146a (> twofold) - ↑ EV miR-146a (53–116-fold) - decreased ATP-linked respiration, maximal respiration and spare respiratory capacity, glycolysis, glycolytic capacity and glycolytic reserve Male 3 × Tg-AD mice subjected to social isolation Adult male C57BL/6 mice (wild-type) subjected to social isolation Socially isolated vs. group-housed AD and WT mice: - ↓ miR-218-5p and miR-124-3p Socially isolated AD mice + miR-218-5p/miR-124-3p agomirs vs. scramble: - restored DG-CA3 synaptic plasticity and pattern separation Mechanistic validation: - miR-218-5p and miR-124-3p directly bind the Rtn3 3′UTR and posttranscriptionally suppress RTN3 protein - RTN3 upregulation/aggregation disrupted mitochondrial distribution in presynaptic boutons 3 × Tg-AD vs. WT mice: - ↑ miR-323-5p and miR-370-3p (> twofold, p < 0.001) - ↓ OPA1 and MFN2 protein Adult rat hippocampal neural stem/progenitor (HCN) cells - insulin withdrawal - miR-150-3p, miR-323-5p or miR-370-3p transfection Insulin withdrawal vs. control: - ↑ miR-150-3p, miR-323-5p and miR-370-3p miRNA transfection vs control: - ↓ OPA1 - miR-150-3p additionally ↓ MFN2 - decreased mitochondrial respiration and ATP production - increased mitochondrial fission Mechanistic validation: - miR-150-3p directly binds the MFN2 3′UTR (30% reduction in reporter activity, p < 0.001) - OPA1 is a potential target of all three miRNAs miR-351-5p mimic vs. miR-Con: - ↓ Miro2 mRNA by 71% (p < 0.0001) - ↓ Miro2 protein by 41% (p < 0.01) - increased mitochondrial fission, mitophagy, autophagy-dependent cell death - decreased MMP and ATP production Mechanistic validation: - miR-351-5p directly targets the Miro2 3′UTR (37% reduction in reporter activity; p < 0.001) AD model: 5xFAD mice, 3xTg-AD mice WT mice 3xTg-AD and 5xFAD vs. WT mice: - ↑ miR-351-5p - ↓ Miro2 APP transgenic mice (Tg2576), 6 months old (n = 5) Tau transgenic mice (P301L), 6 months old (n = 5) Aβ knock-in mice (hA-KI), 12 months old (n = 5) Wild-type C57BL/6 mice (n = 5) vs respective WT controls: - ↑ miRNA PC-5P-12969 in APP mice (p = 0.0117) - ↑ miRNA PC-5P-12969 in Tau mice (p = 0.0110) - no significant change in 6-month hA-KI mice (p = 0.9606) - ↑ miRNA PC-5P-12969 in 12-month hA-KI mice (p = 0.0362) HT22 cells (untreated) HT22 cells + miRNA PC-5P-12969 vector HT22 cells + mAPP plasmid vector HT22 cells + miRNA PC-5P-12969 + mAPP vectors (pre-condition) HT22 cells + mAPP + miRNA PC-5P-12969 vectors (post-condition) vs untreated HT22 cells: - ↑ miRNA PC-5P-12969 with mAPP (p < 0.0001) - ↑ miRNA PC-5P-12969 with miRNA PC-5P-12969 vector (p < 0.0001) - ↑ miRNA PC-5P-12969 in post-condition (p = 0.0001) - no significant change in pre-condition (p = 0.9693) Post-condition vs. mAPP: - ↓ GSK3α and APP - increased ATP production and maximal mitochondrial respiration - increased mitochondrial length - decreased mitochondrial fragmentation Mechanistic validation: - miRNA PC-5P-12969 directly binds the GSK3α 3′UTR (p < 0.0001) n = 3 APP/PSEN1 double transgenic mice n = 3 wild-type (WT) mice APP/PSEN1 vs. WT mice: - cortex: ↑ miR-7a (2.05-fold); ↓ miR-34a - hippocampus: ↑ miR-7a and miR-148a; ↓ miR-128a - ↑ Klf4 protein (p = 0.015 by immunofluorescence; p < 0.05 by Western blot) LPS-primed N9 + AβO + miR-7a mimic vs. LPS-primed N9 + AβO: - decreased ROS - ↓ iNOS and NLRP3 RSL3 + miR-7a mimic vs. RSL3: - ↓ Klf4 - decreased ROS - restored normal long tubular mitochondrial morphology - increased MMP (complete recovery at 250 nM miR-7a) Mechanistic validation: -miR-7a directly binds the Klf4 3′UTR and suppresses Klf4 miR-7a mimic vs negative control: ↓ Klf4 miR-7a inhibitor treatment: ↑ Klf4 n = 42 APP/PS1 mice n = 6 wild type C57BL/6J mice AD vs. WT: - ↑ miR-9-5p (p < 0.001) - ↓ SIRT1 - decreased ATP, MMP, and mitochondrial respiratory function AD + Rg1 vs. AD + Vehicle 1: - ↓ miR-9-5p (p < 0.001) - ↑ SIRT1 - improved mitochondrial function AD + Rg1 + miR-9-5p agomir vs. AD + Rg1 + agomir NC: - ↑ miR-9-5p (p < 0.05) - ↓ SIRT1 - increased mitochondrial dysfunction Aβ1–42 vs. blank: - ↑ miR-9-5p (p < 0.001) - ↓ SIRT1 - ↓ PINK1, Parkin, LC3B II/I, and Beclin-1 Aβ1–42 + Rg1 vs. Aβ1–42 + Vehicle 1: - ↓ miR-9-5p (p < 0.001) - ↑ SIRT1 - ↑ PINK1, Parkin, LC3B II/I and Beclin-1 Aβ1–42 + Rg1 + miR-9-5p mimic vs Aβ1–42 + Rg1 + mimic NC: - ↑ miR-9-5p (p < 0.05) - ↓ SIRT1 - ↓ PINK1, Parkin, LC3B II/I and Beclin-1 - decreased ATP, MMP, and mitochondrial respiratory function hiPSC-NSC-EVs (containing miR-148a-3p, miR-21-5p and miR-103a-3p) + Aβ−42o vs Aβ−42o alone: - decreased mitochondrial superoxide - ↓ BAD and BAX; ↑ BCL-2 - increased MMP and preserved TOM20 + mitochondria - ↑ Beclin-1 and MAP1LC3B - normalized mitochondrial complex gene expression AD vs. normal rat hippocampus: - ↑ miR-140 (p < 0.05) - ↓ PINK1 miR-140 antagomir vs. NC-antagomir: - ↑ PINK1 - increased autophagy AβDDLs-treated vs control neurons: - ↑ miR-140 (p < 0.05) - ↓ PINK1 - increased ROS and mitochondrial dysfunction; decreased autophagy miR-140 inhibitor vs. NC-inhibitor: - ↓ miR-140 (p < 0.05) - ↑ PINK1 - increased MMP and autophagy; decreased ROS Aβ25–35-treated vs untreated HT22 cells: - ↓ miR-9-5p (p < 0.05) - ↑ GSK-3β Aβ25–35 + miR-9-5p mimic vs Aβ25–35 + NC mimic: - ↑ miR-9-5p (p < 0.01) - ↓ GSK-3β - decreased ROS/oxidative stress and apoptosis - ↓ cytochrome C, AIF, cleaved caspase-3 and BAX; ↑ BCL-2 SAMP8 vs age-matched SAMR1 hippocampus: - ↑ miR-195 at 3 and 6 months (p < 0.05); no significant change at 9 months - ↓ miR-15b at 3, 6 and 9 months (p < 0.05) - ↓ MFN2 mRNA and protein miR-195 mimic vs scramble in HT-22 cells: - ↑ miR-195 (p < 0.05) - ↓ MFN2 protein - decreased MMP Rat hippocampal neurons treated with Aβ1–42 Normal hippocampal neurons (control) Aβ1–42-treated vs. control neurons: - ↑ miR-204 (p < 0.05) - ↓ TRPML1 - increased mitochondrial dysfunction/ROS and impaired mitochondrial autophagy Aβ1–42 + miR-204 inhibitor vs. Aβ1–42 + inhibitor negative-control: - ↑ TRPML1 - ↓ PINK1, Parkin, LC3II, and Beclin1 - increased MMP; decreased ROS and mitochondrial autophagy Aβ1–42 + CLN (STAT3 activation) vs. Aβ1–42 + vehicle: - ↓ miR-204 (p < 0.05) - ↑ TRPML1 - increased MMP and ATP; decreased ROS and mitochondrial autophagy APP/PS1 vs. normal hippocampus: - ↑ miR-204 (p < 0.05) - ↓ TRPML1 - decreased mitochondrial autophagy and increased mitochondrial ROS SH-SWE vs. SH-WT: - ↑ miR-124, miR-125b, and miR-146a (p < 0.05); ↑ miR-21 and miR-155 (p < 0.01) - ↑ exosomal miR-124 (p < 0.05) and miR-155 (p < 0.01) miR-124 mimic vs. mock in SH-SWE: - ↑ miR-124 (70-fold; p < 0.01) - ↓ exosomal miR-125b (p < 0.05), miR-146a and miR-155 (p < 0.01); miR-21 decreased, NS - increased mitochondrial activity/MitoTracker intensity miR-124 inhibitor vs. mock in SH-SWE: - ↓ miR-124 (≥ 80% decrease; p < 0.01) - ↑ miR-21 (p < 0.05) and miR-146a (p < 0.01); ↓ miR-125b (p < 0.05) - increased MFN2 protein; no significant change in DRP1 iNEU-PSEN vs. iNEU-WT: - ↑ miR-124, miR-125b, and miR-21 (p < 0.05) - ↓ miR-146a and miR-155 (p < 0.05) - ↑ exosomal miR-124 and miR-155 (p < 0.01) miR-124 mimic vs. mock in iNEU-PSEN: - ↑ miR-124 (~ 12-fold; p < 0.01) - ↓ miR-125b (p < 0.01) and miR-21 (p < 0.05) - ↓ exosomal miR-21 and miR-146a (p < 0.05) and miR-155 (p < 0.01); miR-125b decreased, NS miR-124 inhibitor vs. mock in iNEU-PSEN: - ↓ miR-124 (≥ 80% decrease; p < 0.01) - ↓ miR-125b (p < 0.05) and miR-21 (p < 0.01); ↑ miR-146a (p < 0.01) SWE vs. wild-type SH-SY5Y (SH): - ↑ miR-124 in cells and secretome (cellular, exosomal, and EXO-free; p < 0.05) IFNγ-microglia co-cultured with SWE vs IFNγ-microglia monoculture: - ↑ miR-124 in microglia (nearly 90-fold; p < 0.001) IFNγ-microglia co-cultured with miR-124-mimic SWE vs IFNγ-microglia co-cultured with mock SWE: - ↑ miR-124 in SWE cells (70-fold; p < 0.01) and secretome (exosomal and EXO-free; p < 0.05) - ↑ miR-124 and ↓ miR-146a in co-cultured microglia (p < 0.01) - ↓ iNOS and TNF-α expression in microglia - increased mitochondrial activity/MitoTracker Red intensity Mutant APP + miR-455-3p vs. mutant APP - increased mitochondrial biogenesis - decreased mitochondrial fragmentation - ↓ mitochondrial fission markers (DRP1 and FIS1) - ↑ mitochondrial fusion markers (OPA1, MFN1, and MFN2) Aβ + AuNPs vs. Aβ - ↑ miR-21-5p (FC = 4.78), miR-132, miR-107, miR-369, miR-181c, and miR-212 - ↓ miR-34a and miR-138 - ↑ DRP1, NRF1, and TFAM mRNA - increased MMP and decreased mitochondrial ROS - decreased apoptosis Aβ + AuNPs + miR-21-5p inhibitor vs Aβ + AuNPs + NC - ↓ miR-21-5p (p < 0.01) - ↓ DRP1, NRF1, and TFAM mRNA - decreased cell viability - decreased MMP, increased mitochondrial ROS, and increased apoptosis miR-21-5p mimic vs. mimic-NC - ↑ miR-21-5p (p < 0.01) - ↓ SOCS6 protein - increased cell viability and decreased apoptosis SH + Aβ25–35 + miR-375-3p vs. SH + NC + Aβ25–35 - decreased Aβ25–35-induced apoptosis - decreased intracellular ROS - increased MMP/JC-1 polymer:monomer fluorescence ratio AD + miR-375-3p vs. AD control - ↑ LC3B, Beclin-1, MFN2, and PINK1 mRNAs n = 10 3 × Tg-AD mice receiving PBS injections in the tail veil n = 10 3 × Tg-AD mice receiving CeO2-DRP/MA 1 mmol/L (CeO2 nanoparticles covalently linked with mitochondrial aptamers (MA) and DNA-RNA hybrid chains (DRP) to detect and silence mitochondrial miRNA-204) injections in the tail vein n = 10 C57BL/6 mice 3 × Tg-AD treated with CeO2-DRP/MA vs. PBS-treated 3 × Tg-AD mice: - Decreased escape latency (p < 0.05) - Increased swimming speed (Morris water maze) - Decreased depression-related and anxiety-related behaviors Hippocampal tissue of CeO2-DRP/MA treated vs. PBS-treated 3 × Tg-AD mice: - ↓ PINK1, Beclin1, and LC3B mRNA/protein - ↑TRPML1 and p-STAT3 mRNA/protein SY5Y + Aβ + CeO2-DRP/MA (1 mmol/L) vs. untreated SY5Y + Aβ cells: - increased cell viability from ~ 50% at 0 mmol/L to 98% at 1 mmol/L after 24 h - increased MMP - ↓ mitochondrial miRNA-204 (p < 0.01) - ↓PINK1, Beclin-1, and LC3B mRNA - ↑TRPML1 and STAT3 mRNA CeO2-DRP/MA-treated vs. untreated N2aSW cells: - ↓PINK1, Beclin-1, LC3B mRNA/protein - ↑TRPML1, STAT3 mRNA/protein Pre-miR-423 vs. negative control-transfected cells: - ↓ IDH2 mRNA (59.33% and 60.19% of control in two independent experiments; p < 0.05) Pre-miR-98 vs. negative control-transfected cells: - ↓ IDH2 mRNA (70.03% and 43.98% of control in two independent experiments; p < 0.05) Kunming mice: n = 6 normal n = 6 sham n = 66 successfully modeled AD mice after bilateral hippocampal Aβ1–40 injection, divided into 11 groups (n = 6/group) AD-mice derived osteoblasts: - AD + naringin vs. AD + PBS: ↓ miR-96-5p (p < 0.05) - AD + miR-96-5p inhibitor vs. AD + inhibitor-NC: ↓ miR-96-5p (p < 0.05), ↑ Foxo1, ↓Lcn-2 - AD + miR-96-5p mimic vs. AD + miR-96-5p mimic-NC: ↓ Foxo1, ↑Lcn-2 Hippocampus of AD mice: - AD + miR-96-5p inhibitor vs. AD + inhibitor-NC: - Reduced OH- and MDA - Increased SOD, CAT and GSH-Px - AD + sh-Lcn2 vs. AD + sh-NC: Reduced mitochondrial structural damage (qualitative assessment) - AD + oe-Foxo1 vs. oe-NC: - Reduced mitochondrial structural damage (qualitative assessment) - Reduced OH- and MDA - Increased SOD, CAT and GSH-Px - ↓ Lcn2 eStudies marked with a superscript “e” included additional experimental validation directly assessing the reported ncRNA or its proposed regulatory interaction which was accordingly included in the “Fundamental” section of the table This table synthesizes the experimental model and mitochondrially relevant results of microRNA-focused studies. Studies using human tissue like CSF, serum, plasma, blood or human omics datasets were termed “translational,” while in vitro and in vivo studies were referred to as “fundamental.” The first part of the table summarizes translational data while the second part of the table summarizes fundamental experimental findings. For those studies which featured both translational and fundamental results, the findings were placed in the respective table segment. The downregulation and upregulation of ncRNAs, other genes and proteins are represented using downward arrows (↓) and upward arrows (↑) 3xTg-AD mice, triple-transgenic Alzheimer’s disease model mice; AD, Alzheimer’s disease; aMCI, Aβ-positive amnestic mild cognitive impairment; APP, β-amyloid precursor protein; ATP, adenosine triphosphate; AuNPs, gold nanoparticles; Aβ, amyloid-beta; ceRNA, competing endogenous RNA; CSF, cerebrospinal fluid; DEmiRNA/DEmiRNAs, differentially expressed microRNA(s); EV, extracellular vesicles; HCN, adult hippocampal neural progenitor cells; hiPSC, human induced pluripotent stem cell; hNSC, human neural stem cell; iPSC, induced pluripotent stem cell; lncRNA, long non-coding RNA; MAP, Rush Memory and Aging Project; mAPP, mutant APP; MCI, mild cognitive impairment; miR/miRNA, microRNA; MMP, mitochondrial membrane potential; mRNA, messenger RNA; N/A, not applicable; ncRNA, non-coding RNA; NDAN, nondemented patients with Alzheimer’s neuropathology; PDAD, prodromal Alzheimer’s disease; PSAD, presymptomatic Alzheimer’s disease; RNA-seq, RNA sequencing; ROS, reactive oxygen species; RT-qPCR, reverse transcription quantitative polymerase chain reaction; SAMP8, senescence-accelerated mouse prone 8; WT, wild-type; PBS, phosphate-buffered saline; CeO₂, cerium dioxide; DRP, DNA–RNA hybrid chains; MA, mitochondrial aptamers; DNA, deoxyribonucleic acid; RNA, ribonucleic acid; FAM, fluorescein amidite; IVIS, in vivo imaging system; NC, negative control; MDA, malondialdehyde; SOD, superoxide dismutase; CAT, catalase; GSH-Px, glutathione peroxidase; shRNA, short hairpin RNA; oe, overexpression, RSL3, RAS-selective lethal 3; log2FC, log2 fold change; GO, Gene Ontology Gene and protein symbols: AIF, apoptosis-inducing factor; BAD, Bcl-2-associated death promoter; BAX, BCL2-associated X protein; BCL-2, B-cell lymphoma 2; Dicer1, double-stranded RNA-specific endoribonuclease 1; DRP1, dynamin-related protein 1; FIS1, mitochondrial fission protein 1; GABARAPL1, GABA type A receptor-associated protein-like 1; GSK-3β, glycogen synthase kinase-3 beta; GSK3, glycogen synthase kinase 3; HIBCH, 3-hydroxyisobutyryl-CoA hydrolase; iNOS, inducible nitric oxide synthase; Klf4, Krüppel-like factor 4; LC3B, microtubule-associated protein 1 light chain 3B; MAP1-LC3B, microtubule-associated proteins 1A/1B light chain 3B; MDH2, malate dehydrogenase 2; MFN1, mitofusin 1; MFN2, mitofusin 2; MGME1, mitochondrial genome maintenance exonuclease 1; Miro2, mitochondrial Rho GTPase 2; NDUFB10, NADH:ubiquinone oxidoreductase subunit B10; NDUFS1, NADH:ubiquinone oxidoreductase core subunit S1; NDUFS2, NADH:ubiquinone oxidoreductase core subunit S2; NDUFS3, NADH:ubiquinone oxidoreductase core subunit S3; NDUFS8, NADH:ubiquinone oxidoreductase core subunit S8; NDUFV1, NADH:ubiquinone oxidoreductase core subunit V1; NLRP3, NLR family pyrin domain-containing protein 3; NRF1, nuclear respiratory factor 1; NRF2, nuclear factor erythroid 2-related factor 2; OPA1, optic atrophy 1 gene; Parkin, E3 ubiquitin-protein ligase parkin; PDE4A, phosphodiesterase 4 A; PDHA1, pyruvate dehydrogenase E1 subunit alpha 1; PGC-1α, peroxisome proliferator-activated receptor-gamma coactivator 1-alpha; PINK1, PTEN-induced putative protein kinase 1; PSEN1, presenilin 1; RTN3, reticulon 3 gene; SDHB, succinate dehydrogenase complex iron sulfur subunit B; SIRT1, sirtuin 1; SOCS6, suppressor of cytokine signalling 6; STAT3, signal transducer and activator of transcription 3; TFEB, transcription factor EB; TIMM13, translocase of the inner mitochondrial membrane 13; TNF-α, tumor necrosis factor alpha; TOMM20, translocase of the outer mitochondrial membrane 20; TRPML1, transient receptor potential mucolipin-1; UQCRFS1, ubiquinol-cytochrome c reductase Rieske iron-sulfur polypeptide 1; UTR, untranslated region; VDAC1, voltage-dependent anion channel 1, IDH2, isocitrate dehydrogenase 2, Beclin1, beclin 1; FOXO1, forkhead box O1; LCN2, lipocalin 2, ATP5E, ATP synthase F1 subunit epsilon; NDUFB9, NADH:ubiquinone oxidoreductase subunit B9; NDUFA9, NADH:ubiquinone oxidoreductase subunit A9; COX4I1, cytochrome c oxidase subunit 4I1 Table 3Long non-coding RNA dysregulation and mitochondrial dysfunction in AD: overview of translational and fundamental studiesLncRNA—translational studiesReferenceSubjects or experimental modelResults[46]eGEO GSE4757 (n = 10 controls, n = 10 AD human cortices)GEO GSE16759 (n = 4 controls, n = 4 AD human samples—parietal lobe cortex)GSE4757, AD vs. controls:- ↓ WT1-AS (p = 0.0082)- ↓ SIX4 (p = 0.0055)GSE16759, AD vs. controls:- ↑ miR-375 (p = 0.0438)[75]en = 8 controls (5 female, 3 male)n = 8 AD patients (4 female, 4 male, Braak stages IV–VI)AD vs age- and sex-matched controls:- ↓ lncMtDloop in PFC (p = 0.0057)- ↓ lncMtDloop in hippocampal CA1 (p = 0.0076)- ↓ lncMtDloop in dentate gyrus, trend only (p = 0.063)- ↓ lncMtDloop by RT-qPCR in PFC (p = 0.0031) and hippocampus (p = 0.0027)[76]en = 16 probable AD patients (7 males and 9 females, 70.3 ± 8.8 years)n = 18 healthy individuals (8 males and 10 females, 68.3 ± 9.6 years)AD patients vs. healthy controls:- ↑ lncRNA-ATB in CSF (p < 0.001)- ↑ lncRNA-ATB in serum (p < 0.05)LncRNA—fundamental studiesReferenceSubjects or experimental modelResults[45]SH-SY5Y cells treated with Aβ25–35human primary neurons treated with Aβ25–35Aβ25–35-treated vs. untreated SH-SY5Y and human primary neurons:- ↑ SNHG1 (20 μM: p < 0.001)- decreased MMPAβ25–35 + si-SNHG1 vs. Aβ25–35 + si-NC:- increased MMP and decreased apoptosisMechanistic validation:- SNHG1 knockdown ↑ miR-137 (p < 0.01)- SNHG1 sponges miR-137; miR-137 directly targets KREMEN1- miR-137 inhibition or KREMEN1 overexpression partially reversed the protective effects of SNHG1 knockdown[46]SH-SY5Y cells treated with Aβ25–35Aβ25–35-treated vs. untreated SH-SY5Y cells:- ↑ miR-375 (p < 0.05)- ↓ WT1-AS and SIX4 (p < 0.05)Aβ25–35 + miR-375 mimic vs. Aβ25–35 + mimic-NC:- ↓ SIX4 mRNA and protein (p < 0.05)- decreased MMP and ATP production- increased oxidative stress and apoptosisMechanistic validation:- miR-375 directly binds SIX4 mRNA- WT1 transcriptionally promotes miR-375 expressionMale BALB/c mice, Aβ25–35-induced AD modeln = 10/groupAβ25–35 AD-model vs. sham mice:- ↑ miR-375 (p < 0.05)- ↓ WT1-AS and SIX4; ↑ WT1AD + WT1-AS overexpression vs. AD + oe-NC:- ↓ miR-375 (p < 0.05)- ↑ SIX4 (p < 0.05)- decreased oxidative stress and neuronal apoptosis[47]PC12 cells treated with Aβ25–35Aβ25–35-treated vs. untreated PC12 cells:- ↑ H19 and ↓ miR-129 (p < 0.05)- ↑ HMGB1 (p < 0.05)Aβ25–35 + si-H19 vs. Aβ25–35 + si-NC:- ↑ miR-129 and ↓ HMGB1 (p < 0.05)- decreased ROS and increased MMP (p < 0.05)Mechanistic validation:- H19 directly binds/sponges miR-129- miR-129 directly targets the HMGB1 3′UTRBALB/c male mice injected with Aβ25–35 (intraventricular)n = 10/groupAβ25–35 AD-model vs. sham mice:- ↑ H19 and ↓ miR-129- ↑ HMGB1[49]hiPSC-derived glutamatergic forebrain neurons from PSEN1-mutant patients (L150P, A79V) and isogenic controls (fibroblast-derived)A79V and L150P PSEN1-mutant neurons vs. respective gene-corrected isogenic controls:- ↓ MEG3 (A79V: L2FC = −5.88, Padj = 1.47 × 10⁻2⁹; L150P: L2FC = −6.31, Padj = 2.75 × 10⁻⁹4)- ↓ LOXL1-AS1- ↓ A2M-AS1[75]WT and 3xTg-AD mouse-derived primary hippocampal neuron cultures3xTg-AD vs. WT primary hippocampal neurons:- ↓ lncMtDloop (RT-qPCR: p = 0.0075)3xTg-AD + LV-lncMtDloop vs. 3xTg-AD + LV-control:- ↑ mtRNA (p = 0.0363)- ↑ mitochondrial genome-encoded OXPHOS genes/proteins- ↓ Drp1 protein (p = 0.0352)- ↑ Opa1 protein (p = 0.0296)- increased basal and reserve mitochondrial respiration3xTg-AD miceWild-type mice12-month-old 3xTg-AD vs. age-matched WT mice:- ↓ lncMtDloop in PFC and hippocampus (RT-qPCR: p = 0.0059 and p = 0.0082, respectively)3xTg-AD + AAV-lncMtDloop vs. 3xTg-AD + AAV-control:- restored mitochondrial ultrastructure and morphology in PFC and hippocampus- ↑ PINK1, Parkin and LC3BMechanistic validation:- lncMtDloop directly interacts with TFAM and promotes its recruitment to mtDNA promoters, increasing mitochondrial transcription[76]Neuronal PC12 cells treated with Aβ 25–35Aβ25–35-treated vs. untreated PC12 cells:- ↑ lncRNA-ATB (p < 0.001)- increased ROS- decreased MMPAβ25–35 + sh-ATB vs. Aβ25–35 + sh-NC:- ↓ lncRNA-ATB (p < 0.01)- decreased ROS- increased MMPsh-ATB vs. sh-NC:- ↑ miR-200 (p < 0.01)Aβ25–35 + sh-ATB + miR-200 inhibitor vs Aβ25–35 + sh-ATB + inhibitor NC:- increased ROS- decreased MMPmiR-200 mimic vs. mimic NC:- ↓ ZNF217[77]APP/PS1 mice5xFAD miceAPP/PS1 and 5xFAD mice > 6 months vs. age-matched WT:- ↑ MIR600HG (p < 0.05; age-dependent)APP/PS1 + AAV-shMIR600HG vs. untreated APP/PS1:- increased cytochrome c oxidase activitySH-SY5Y cellsSH-SY5Y + MIR600HG overexpression vs control:- ↑ PINK1–NEDD4L interaction- ↑ PINK1 ubiquitination- ↓ PINK1- ↓ autophagy markers (LC3, OPTN and p62)SH-SY5Y + shMIR600HG vs. control:- ↑ PINK1- ↑ autophagy markers (LC3, OPTN and p62)[78]APP/PS1 transgenic miceAPP/PS1 vs. WT:- ↑ lncRNA NEAT1 in older APP/PS1 mice (p < 0.01)APP/PS1 + AAV-NEAT1 vs. untreated APP/PS1:- decreased cytochrome c oxidase activityAPP/PS1 + AAV-shNEAT1 vs. untreated APP/PS1:- increased/restored cytochrome c oxidase activity and ATP- decreased AβSH-SY5Y cellsNEAT1 overexpression vs. vector:- ↑ PINK1 ubiquitination and degradation- decreased PINK1-dependent autophagyNEAT1 knockdown vs. vector:- ↓ PINK1 ubiquitination- ↑ PINK1 and PINK1-dependent autophagyN2A-APPsw cellsshNEAT1 vs. control:- decreased Aβ42 (p < 0.01)shNEAT1 + siPINK1/siOPTN/siLC3 vs shNEAT1:- reversed the reduction in Aβ42[79]PC12 cells treated with Aβ25–35Aβ25–35 vs. untreated PC12 cells:- ↑ lncRNA BDNF-AS (p < 0.05)- ↓ BDNFAβ25–35 + si-BDNF-AS vs. Aβ25–35 + si-NC:- ↓ lncRNA BDNF-AS (p < 0.05)- ↑ BDNF and BCL-2; ↓ BAX and cleaved caspase-3- increased MMP- decreased cytochrome c release and apoptosis- decreased ROS and MDA; increased SOD and CAT activityeStudies marked with a superscript “e” included additional experimental validation directly assessing the reported ncRNA or its proposed regulatory interaction which was accordingly included in the “Fundamental” section of the tableThis table synthesizes the experimental model and mitochondrially relevant results of lncRNA-focused studies. Studies using human tissue like CSF, serum, plasma, blood, or human omics datasets were termed “translational,” while in vitro and in vivo studies were referred to as “fundamental.” The downregulation and upregulation of ncRNAs, other genes and proteins are represented using downward arrows (↓) and upward arrows (↑). The first part of the table summarizes translational data while the second part of the table summarizes fundamental experimental findings. For those studies which featured both translational and fundamental results, the findings were placed in the respective table segment3xTg-AD mice, Triple-transgenic Alzheimer’s disease model mice; AD, Alzheimer’s disease; APP, β-amyloid precursor protein; ATP, adenosine triphosphate; AuNPs, gold nanoparticles; Aβ, amyloid-beta; CAT, catalase; CSF, cerebrospinal fluid; DG, dentate gyrus; GEO, Gene Expression Omnibus; hiPSC, human induced pluripotent stem cell; LC3, microtubule-associated proteins 1A/1B light chain 3; lncMtDloop, long non-coding mitochondrial displacement loop RNA; lncRNA, long non-coding RNA; lncRNA BDNF-AS, long non-coding RNA brain-derived neurotrophic factor antisense; lncRNA NEAT1, lncRNA nuclear paraspeckle assembly transcript 1; lncRNA WT1-AS, Wilms tumor 1 antisense RNA; MDA, malondialdehyde; miRNA, microRNA; MMP, mitochondrial membrane potential; mtRNA, mitochondrial RNA; N/A, not applicable; OSI, oxidative stress index; OXPHOS, oxidative phosphorylation; PFC, prefrontal cortex; Poly(A) RNA-seq, polyadenylated RNA sequencing; PSEN1, presenilin 1; RT-qPCR, reverse transcription quantitative polymerase chain reaction; SOD, superoxide dismutase; total RNA-seq, total RNA sequencingGene and protein symbols: A2M-AS1, alpha-2-macroglobulin antisense RNA 1; BAX, BCL2-associated X protein; BCL-2, B-cell lymphoma 2; BDNF-AS, BDNF antisense RNA; DRP1, dynamin-related protein 1; HMGB1, high mobility group box 1; LC3B, microtubule-associated protein 1 light chain 3B; LOXL1-AS1, lysyl oxidase-like 1 antisense RNA 1; MIR600HG, miR-600 host gene; NEDD4L, neural precursor cell expressed, developmentally down-regulated 4-like, E3 ubiquitin protein ligase; OPA1, optic atrophy 1 gene; OPTN, optineurin; p62, sequestosome-1 (SQSTM1); Parkin, E3 ubiquitin-protein ligase parkin; PINK1, PTEN-induced putative protein kinase 1; SIX4, SIX homeobox 4; WT1, Wilms Tumor 1 gene; WT1-AS, Wilms tumor 1 antisense RNA; ZNF217, zinc finger protein 217 Long non-coding RNA dysregulation and mitochondrial dysfunction in AD: overview of translational and fundamental studies GEO GSE4757 (n = 10 controls, n = 10 AD human cortices) GEO GSE16759 (n = 4 controls, n = 4 AD human samples—parietal lobe cortex) GSE4757, AD vs. controls: - ↓ WT1-AS (p = 0.0082) - ↓ SIX4 (p = 0.0055) GSE16759, AD vs. controls: - ↑ miR-375 (p = 0.0438) n = 8 controls (5 female, 3 male) n = 8 AD patients (4 female, 4 male, Braak stages IV–VI) AD vs age- and sex-matched controls: - ↓ lncMtDloop in PFC (p = 0.0057) - ↓ lncMtDloop in hippocampal CA1 (p = 0.0076) - ↓ lncMtDloop in dentate gyrus, trend only (p = 0.063) - ↓ lncMtDloop by RT-qPCR in PFC (p = 0.0031) and hippocampus (p = 0.0027) n = 16 probable AD patients (7 males and 9 females, 70.3 ± 8.8 years) n = 18 healthy individuals (8 males and 10 females, 68.3 ± 9.6 years) AD patients vs. healthy controls: - ↑ lncRNA-ATB in CSF (p < 0.001) - ↑ lncRNA-ATB in serum (p < 0.05) SH-SY5Y cells treated with Aβ25–35 human primary neurons treated with Aβ25–35 Aβ25–35-treated vs. untreated SH-SY5Y and human primary neurons: - ↑ SNHG1 (20 μM: p < 0.001) - decreased MMP Aβ25–35 + si-SNHG1 vs. Aβ25–35 + si-NC: - increased MMP and decreased apoptosis Mechanistic validation: - SNHG1 knockdown ↑ miR-137 (p < 0.01) - SNHG1 sponges miR-137; miR-137 directly targets KREMEN1 - miR-137 inhibition or KREMEN1 overexpression partially reversed the protective effects of SNHG1 knockdown Aβ25–35-treated vs. untreated SH-SY5Y cells: - ↑ miR-375 (p < 0.05) - ↓ WT1-AS and SIX4 (p < 0.05) Aβ25–35 + miR-375 mimic vs. Aβ25–35 + mimic-NC: - ↓ SIX4 mRNA and protein (p < 0.05) - decreased MMP and ATP production - increased oxidative stress and apoptosis Mechanistic validation: - miR-375 directly binds SIX4 mRNA - WT1 transcriptionally promotes miR-375 expression Male BALB/c mice, Aβ25–35-induced AD model n = 10/group Aβ25–35 AD-model vs. sham mice: - ↑ miR-375 (p < 0.05) - ↓ WT1-AS and SIX4; ↑ WT1 AD + WT1-AS overexpression vs. AD + oe-NC: - ↓ miR-375 (p < 0.05) - ↑ SIX4 (p < 0.05) - decreased oxidative stress and neuronal apoptosis Aβ25–35-treated vs. untreated PC12 cells: - ↑ H19 and ↓ miR-129 (p < 0.05) - ↑ HMGB1 (p < 0.05) Aβ25–35 + si-H19 vs. Aβ25–35 + si-NC: - ↑ miR-129 and ↓ HMGB1 (p < 0.05) - decreased ROS and increased MMP (p < 0.05) Mechanistic validation: - H19 directly binds/sponges miR-129 - miR-129 directly targets the HMGB1 3′UTR BALB/c male mice injected with Aβ25–35 (intraventricular) n = 10/group Aβ25–35 AD-model vs. sham mice: - ↑ H19 and ↓ miR-129 - ↑ HMGB1 A79V and L150P PSEN1-mutant neurons vs. respective gene-corrected isogenic controls: - ↓ MEG3 (A79V: L2FC = −5.88, Padj = 1.47 × 10⁻2⁹; L150P: L2FC = −6.31, Padj = 2.75 × 10⁻⁹4) - ↓ LOXL1-AS1 - ↓ A2M-AS1 3xTg-AD vs. WT primary hippocampal neurons: - ↓ lncMtDloop (RT-qPCR: p = 0.0075) 3xTg-AD + LV-lncMtDloop vs. 3xTg-AD + LV-control: - ↑ mtRNA (p = 0.0363) - ↑ mitochondrial genome-encoded OXPHOS genes/proteins - ↓ Drp1 protein (p = 0.0352) - ↑ Opa1 protein (p = 0.0296) - increased basal and reserve mitochondrial respiration 3xTg-AD mice Wild-type mice 12-month-old 3xTg-AD vs. age-matched WT mice: - ↓ lncMtDloop in PFC and hippocampus (RT-qPCR: p = 0.0059 and p = 0.0082, respectively) 3xTg-AD + AAV-lncMtDloop vs. 3xTg-AD + AAV-control: - restored mitochondrial ultrastructure and morphology in PFC and hippocampus - ↑ PINK1, Parkin and LC3B Mechanistic validation: - lncMtDloop directly interacts with TFAM and promotes its recruitment to mtDNA promoters, increasing mitochondrial transcription Aβ25–35-treated vs. untreated PC12 cells: - ↑ lncRNA-ATB (p < 0.001) - increased ROS - decreased MMP Aβ25–35 + sh-ATB vs. Aβ25–35 + sh-NC: - ↓ lncRNA-ATB (p < 0.01) - decreased ROS - increased MMP sh-ATB vs. sh-NC: - ↑ miR-200 (p < 0.01) Aβ25–35 + sh-ATB + miR-200 inhibitor vs Aβ25–35 + sh-ATB + inhibitor NC: - increased ROS - decreased MMP miR-200 mimic vs. mimic NC: - ↓ ZNF217 APP/PS1 mice 5xFAD mice APP/PS1 and 5xFAD mice > 6 months vs. age-matched WT: - ↑ MIR600HG (p < 0.05; age-dependent) APP/PS1 + AAV-shMIR600HG vs. untreated APP/PS1: - increased cytochrome c oxidase activity SH-SY5Y + MIR600HG overexpression vs control: - ↑ PINK1–NEDD4L interaction - ↑ PINK1 ubiquitination - ↓ PINK1 - ↓ autophagy markers (LC3, OPTN and p62) SH-SY5Y + shMIR600HG vs. control: - ↑ PINK1 - ↑ autophagy markers (LC3, OPTN and p62) APP/PS1 vs. WT: - ↑ lncRNA NEAT1 in older APP/PS1 mice (p < 0.01) APP/PS1 + AAV-NEAT1 vs. untreated APP/PS1: - decreased cytochrome c oxidase activity APP/PS1 + AAV-shNEAT1 vs. untreated APP/PS1: - increased/restored cytochrome c oxidase activity and ATP - decreased Aβ NEAT1 overexpression vs. vector: - ↑ PINK1 ubiquitination and degradation - decreased PINK1-dependent autophagy NEAT1 knockdown vs. vector: - ↓ PINK1 ubiquitination - ↑ PINK1 and PINK1-dependent autophagy shNEAT1 vs. control: - decreased Aβ42 (p < 0.01) shNEAT1 + siPINK1/siOPTN/siLC3 vs shNEAT1: - reversed the reduction in Aβ42 Aβ25–35 vs. untreated PC12 cells: - ↑ lncRNA BDNF-AS (p < 0.05) - ↓ BDNF Aβ25–35 + si-BDNF-AS vs. Aβ25–35 + si-NC: - ↓ lncRNA BDNF-AS (p < 0.05) - ↑ BDNF and BCL-2; ↓ BAX and cleaved caspase-3 - increased MMP - decreased cytochrome c release and apoptosis - decreased ROS and MDA; increased SOD and CAT activity eStudies marked with a superscript “e” included additional experimental validation directly assessing the reported ncRNA or its proposed regulatory interaction which was accordingly included in the “Fundamental” section of the table This table synthesizes the experimental model and mitochondrially relevant results of lncRNA-focused studies. Studies using human tissue like CSF, serum, plasma, blood, or human omics datasets were termed “translational,” while in vitro and in vivo studies were referred to as “fundamental.” The downregulation and upregulation of ncRNAs, other genes and proteins are represented using downward arrows (↓) and upward arrows (↑). The first part of the table summarizes translational data while the second part of the table summarizes fundamental experimental findings. For those studies which featured both translational and fundamental results, the findings were placed in the respective table segment 3xTg-AD mice, Triple-transgenic Alzheimer’s disease model mice; AD, Alzheimer’s disease; APP, β-amyloid precursor protein; ATP, adenosine triphosphate; AuNPs, gold nanoparticles; Aβ, amyloid-beta; CAT, catalase; CSF, cerebrospinal fluid; DG, dentate gyrus; GEO, Gene Expression Omnibus; hiPSC, human induced pluripotent stem cell; LC3, microtubule-associated proteins 1A/1B light chain 3; lncMtDloop, long non-coding mitochondrial displacement loop RNA; lncRNA, long non-coding RNA; lncRNA BDNF-AS, long non-coding RNA brain-derived neurotrophic factor antisense; lncRNA NEAT1, lncRNA nuclear paraspeckle assembly transcript 1; lncRNA WT1-AS, Wilms tumor 1 antisense RNA; MDA, malondialdehyde; miRNA, microRNA; MMP, mitochondrial membrane potential; mtRNA, mitochondrial RNA; N/A, not applicable; OSI, oxidative stress index; OXPHOS, oxidative phosphorylation; PFC, prefrontal cortex; Poly(A) RNA-seq, polyadenylated RNA sequencing; PSEN1, presenilin 1; RT-qPCR, reverse transcription quantitative polymerase chain reaction; SOD, superoxide dismutase; total RNA-seq, total RNA sequencing Gene and protein symbols: A2M-AS1, alpha-2-macroglobulin antisense RNA 1; BAX, BCL2-associated X protein; BCL-2, B-cell lymphoma 2; BDNF-AS, BDNF antisense RNA; DRP1, dynamin-related protein 1; HMGB1, high mobility group box 1; LC3B, microtubule-associated protein 1 light chain 3B; LOXL1-AS1, lysyl oxidase-like 1 antisense RNA 1; MIR600HG, miR-600 host gene; NEDD4L, neural precursor cell expressed, developmentally down-regulated 4-like, E3 ubiquitin protein ligase; OPA1, optic atrophy 1 gene; OPTN, optineurin; p62, sequestosome-1 (SQSTM1); Parkin, E3 ubiquitin-protein ligase parkin; PINK1, PTEN-induced putative protein kinase 1; SIX4, SIX homeobox 4; WT1, Wilms Tumor 1 gene; WT1-AS, Wilms tumor 1 antisense RNA; ZNF217, zinc finger protein 217 Table 4Circular RNA dysregulation and mitochondrial dysfunction in AD: overview of translational and experimental studiesReferenceSubjects or experimental modelResults[48]SK-N-SH and CHP212 cells treated with 20 μM Aβ exposure for 24 hAβ-treated vs. control cells:- ↑ circ_0049472 (p < 0.001)- ↓ miR-22-3p (p < 0.001)Aβ + si-circ_0049472 vs. Aβ + si-NC:- ↓ circ_0049472 (p < 0.001)- ↑ miR-22-3p (p < 0.001)- ↑ BCL-2; ↓ BAX and cleaved-caspase-3- increased/restored MMP[49]hiPSC-derived glutamatergic forebrain neurons from PSEN1-mutant patients (L150P, A79V) and isogenic controls (fibroblast-derived)L150P vs. isogenic control:- ↑ CDR1as/circ_0001946 (L2FC = 2.38; Padj = 2.04 × 10−10)- ↓ UBE2AA79V vs. isogenic control:- circ_0001946: no significant change (L2FC = 0.67; Padj = 0.37)- ↓ UBE2A[80]Primary mouse hippocampal neurons treated with Aβ1–42miR-647 inhibitor + sh-circ_0004381 vs. sh-circ_0004381:- ↓ miR-647 (p < 0.05)- increased ROS and cytoplasmic cytochrome c; decreased SOD- attenuated the increases in ATP and MMP- ↑ BAX/BCL-2 ratioMicroglial BV2 cells treated with lipopolysaccharideLPS + sh-circ_0004381 vs. LPS + sh-NC:- ↓ iNOS and COX-2- decreased M1 and increased M2 polarization- decreased IL-1β, IL-6 and TNF-αMale APPswe/PS1dE9 transgenic miceAPPswe/PS1dE9 vs. non-transgenic controls:- ↑ circ_0004381 and ↓ miR-647 (p < 0.05)APPswe/PS1dE9 + sh-circ_0004381 vs. untreated APPswe/PS1dE9:- ↓ circ_0004381 and ↑ miR-647 (p < 0.05)- decreased ROSThis table synthesizes the tissue sample, experimental model, RNA quantification method and mitochondrially-relevant results of 3 circRNA-focused studies. No eligible circRNA studies using patient-derived data were identified after the screening process. All circRNA evidence was derived from experimental in vitro and in vivo studies, which we refer to as “fundamental studies.” The downregulation and upregulation of ncRNAs, other genes and proteins are represented using downward arrows (↓) and upward arrows (↑)AD, Alzheimer’s disease; APP, β-amyloid precursor protein; ATP, adenosine triphosphate; Aβ, amyloid-beta; circRNA, circular RNA; hiPSC, human induced pluripotent stem cell; MMP, mitochondrial membrane potential; N/A, not applicable; ncRNA, non-coding RNA; Poly(A) RNA-seq, polyadenylated RNA sequencing; RNA-seq, RNA sequencing; ROS, reactive oxygen species; RT-qPCR, reverse transcription quantitative polymerase chain reaction; SOD, superoxide dismutase; total RNA-seq, total RNA sequencingGene and protein symbols: BAX, BCL2-associated X protein; BCL-2, B-cell lymphoma 2; COX-2, cyclooxygenase-2; iNOS, inducible nitric oxide synthase; PSEN1, presenilin 1; UBE2A, ubiquitin-conjugating enzyme E2A Circular RNA dysregulation and mitochondrial dysfunction in AD: overview of translational and experimental studies Aβ-treated vs. control cells: - ↑ circ_0049472 (p < 0.001) - ↓ miR-22-3p (p < 0.001) Aβ + si-circ_0049472 vs. Aβ + si-NC: - ↓ circ_0049472 (p < 0.001) - ↑ miR-22-3p (p < 0.001) - ↑ BCL-2; ↓ BAX and cleaved-caspase-3 - increased/restored MMP L150P vs. isogenic control: - ↑ CDR1as/circ_0001946 (L2FC = 2.38; Padj = 2.04 × 10−10) - ↓ UBE2A A79V vs. isogenic control: - circ_0001946: no significant change (L2FC = 0.67; Padj = 0.37) - ↓ UBE2A miR-647 inhibitor + sh-circ_0004381 vs. sh-circ_0004381: - ↓ miR-647 (p < 0.05) - increased ROS and cytoplasmic cytochrome c; decreased SOD - attenuated the increases in ATP and MMP - ↑ BAX/BCL-2 ratio LPS + sh-circ_0004381 vs. LPS + sh-NC: - ↓ iNOS and COX-2 - decreased M1 and increased M2 polarization - decreased IL-1β, IL-6 and TNF-α APPswe/PS1dE9 vs. non-transgenic controls: - ↑ circ_0004381 and ↓ miR-647 (p < 0.05) APPswe/PS1dE9 + sh-circ_0004381 vs. untreated APPswe/PS1dE9: - ↓ circ_0004381 and ↑ miR-647 (p < 0.05) - decreased ROS This table synthesizes the tissue sample, experimental model, RNA quantification method and mitochondrially-relevant results of 3 circRNA-focused studies. No eligible circRNA studies using patient-derived data were identified after the screening process. All circRNA evidence was derived from experimental in vitro and in vivo studies, which we refer to as “fundamental studies.” The downregulation and upregulation of ncRNAs, other genes and proteins are represented using downward arrows (↓) and upward arrows (↑) AD, Alzheimer’s disease; APP, β-amyloid precursor protein; ATP, adenosine triphosphate; Aβ, amyloid-beta; circRNA, circular RNA; hiPSC, human induced pluripotent stem cell; MMP, mitochondrial membrane potential; N/A, not applicable; ncRNA, non-coding RNA; Poly(A) RNA-seq, polyadenylated RNA sequencing; RNA-seq, RNA sequencing; ROS, reactive oxygen species; RT-qPCR, reverse transcription quantitative polymerase chain reaction; SOD, superoxide dismutase; total RNA-seq, total RNA sequencing Gene and protein symbols: BAX, BCL2-associated X protein; BCL-2, B-cell lymphoma 2; COX-2, cyclooxygenase-2; iNOS, inducible nitric oxide synthase; PSEN1, presenilin 1; UBE2A, ubiquitin-conjugating enzyme E2A The findings presented mechanistic evidence of variable strength and are discussed accordingly. Some studies performed direct mitochondrial assessments such as measures of ATP production, mitochondrial morphology, mitophagy, and mitochondrial dynamics while other studies reported indirect evidence such as that derived from bioinformatics analyses or oxidative stress and apoptosis assessments. Mitophagy and ncRNAs in AD Mitophagy is the selective autophagy of dysfunctional mitochondria. This process is dependent on the PTEN-induced putative protein kinase 1 (PINK1) and the E3 ubiquitin ligase Parkin. Normally, PINK1 is imported to the inner mitochondrial membrane, but mitochondrial membrane depolarization leads to PINK1 accumulation on the outer membrane. This causes the recruitment of cytosolic Parkin, which ubiquitinates outer membrane proteins and leads to autophagic receptor accumulation [87]. Mitophagy involves various selective autophagy adaptors such as p62, optineurin (OPTN) and nuclear dot protein 52 (NDP52), which are recruited to dysfunctional mitochondria under the regulation Parkin/PINK1. The adaptors contain a ubiquitin-binding domain and a LC3-interacting region [88]. The LC3 proteins are part of the Atg8 family of proteins which conjugate to the autophagosomal membranes, which are essential for autophagosome-lysosome fusion [89]. Two experimental studies have demonstrated the involvement of miRNAs in regulating PINK1 in AD. Firstly, miR-351-5p suppressed mitochondrial Rho GTPase 2 (Miro2) in adult hippocampal neural progenitor cells, thus disrupting mitochondrial transport, inducing mitochondrial fission and stimulating mitophagy through the PINK1/Parkin pathway. In AD mice hippocampi, miR-351-5p was upregulated while Miro2 was downregulated. While miR-351-5p itself was not quantified in the human hippocampus, Miro2 was downregulated in AD patients [63]. Hence, therapeutic avenues to restore spatial mitochondrial distribution in neurons and maintain synaptic integrity may involve modulating miR-351-5p. Secondly, miR-140 was shown to downregulate PINK1. MiR-140 overexpression was detected in both AD mice hippocampal tissues and Aβ oligomer-treated primary hippocampal neurons, which was associated with impaired autophagy. Functional inhibition of miR-140 in both models preserved mitochondrial function via the upregulation of PINK [68]. Other AD studies linked miR-140-5p upregulation in the human brain to ADAM10 downregulation and increased amyloidogenic processing of APP [90, 91]. miR-140 is also involved in mitochondrial dynamics and mitophagy in other cell types [92–94], encouraging further investigation. PINK1 is also modulated by a series of long non-coding RNAs. Firstly, lncRNA NEAT1 upregulation was observed in the hippocampus of AD mice, alongside increased PINK1 ubiquitination and reduced autophagy markers (p62, OPTN, LC3). Its knockdown reversed these effects, confirming its inhibitory role on PINK1-dependent mitophagy [78]. The disruptive effect of NEAT1 on mitochondrial homeostasis observed by Huang et al. (2020) is in agreement with other studies. Elsewhere, NEAT1 overexpression was shown to disrupt mitochondrial biogenesis by interacting with peroxisome proliferator-activated receptor-gamma coactivator 1-alpha (PGC-1α) and inhibiting B-cell lymphoma 2 protein (BCL-2) and sirtuin 1 (SIRT1) [95]. NEAT1 knockdown reduces neuronal apoptosis rates in AD [96]. Interestingly, one study using PD mice found that NEAT1 prevented PINK1 degradation [97]. Nonetheless, the evidence suggests that NEAT1 regulates mitophagy and other mitochondrial processes in AD. Secondly, Liu et al. (2022) found that lncRNA MIR600HG was upregulated in AD mice. MIR600HG upregulates E3 ligase NEDD4L (E3 ubiquitin protein ligase, neural precursor cell expressed, developmentally down-regulated 4-like), which decreases PINK1 activity via ubiquitination, consequently impairing mitophagic activation [77]. NEDD4L has been shown to suppress autophagy and negatively regulate mitochondrial energy metabolism [98, 99]. The findings of Liu et al. [77] regarding the NEDD4L/PINK1 axis are corroborated by a recent endothelial injury study [100]. Finally, lncRNA BACE1-AS knockdown via lentiviral modulation resulted in a significant increase in PINK1 and LC3-I/II expression in SH-SY5Y cells [101]. LncRNA BACE1-AS suppression increases cognitive performance in AD mice [102] and promotes autophagy [103], though further in vitro studies better replicating the hallmarks of AD pathophysiology are warranted. In summary, PINK1 levels are decreased by miR-140 [68], lncRNA NEAT1 [78] and lncRNA MIR600HG [77] via various mechanisms in the preclinical AD studies identified in this review. Dysregulated levels of these ncRNAs have been reported in patients. MiR-140-5p is upregulated in post-mortem AD brains [90] while plasma miR-140-3p levels are predictive for mild cognitive impairment [104] and are elevated in stroke [105]. LncRNA BACE1-AS is a biomarker candidate and is upregulated in the plasma [106, 107] and serum of AD patients [108]. NEAT1 is upregulated in the plasma of AD [109] and PD patients [110]. MIR600HG is dysregulated in autism spectrum disorder [111], but additional studies are necessary to identify MIR600HG dysregulation patterns in AD. Finally, miR-351-5p, which stimulates PINK1/Parkin-dependent mitophagy, is upregulated in AD mouse hippocampi [63] but further studies are needed to confirm its dysregulation in patients. NcRNAs and Mitochondrially Mediated Cell Death in AD Multiple cell death pathways are dysregulated in AD, from the autophagy-lysosome pathway [112] to high neuronal apoptosis rates [113] and ferroptosis [114]. In this section, we discuss the ncRNAs which help regulate autophagy, apoptosis and ferroptosis in AD, highlighting how they affect mitochondrial activity in each case. Caspases are key agents in apoptosis. They can be activated through death receptors such as those in the tumor necrosis factor receptor superfamily or via the intrinsic, mitochondria-dependent pathway [115]. The mitochondrial pathway involves mitochondrial outer membrane permeabilization (MOMP), which enables cytochrome c to leave the intermembrane space and enter the cytosol, where it activates executioner caspases-3 and −7. MOMP is regulated by pro-apoptotic effector proteins BCL-2-associated X protein (BAX), BCL-2 homologous antagonist/killer (BAK) and BH3-only proteins and anti-apoptotic BCL-2 proteins [116]. A number of BH3-only proteins activate BAX or BAK and cause them to form pores in the inner membrane, ensuring MOMP. Anti-apoptotic BCL-2 proteins prevent the activation of BAX and BAK or directly sequester BH3-only proteins, thus preventing apoptosis despite death signaling [117]. Among the various regulatory mechanisms of the mitochondrial apoptosis pathway, the tumor suppressor protein p53 acts as a simultaneous BAX activator and BCL-2 suppressor [115]. By contrast, the PI3K/AKT pathway prevents apoptosis. Phosphatidylinositol-3-kinases (PI3K) can activate numerous downstream targets, including the protein kinase B (AKT), which is an important regulator of cellular proliferation, as it inhibits the pro-apoptotic activity of BAD, BAX, and glycogen synthase kinase-3 beta (GSK-3β) [118]. We identified several studies on the involvement of ncRNAs in mitochondrially mediated apoptosis in AD. Firstly, Guo et al. (2017) found that knockdown of the lncRNA BDNF-AS in Aβ-treated PC12 cells led to a significant upregulation of the mitochondrial anti-apoptotic marker BCL-2, alongside a reduction in pro-apoptotic proteins cleaved caspase-3 and BAX [79]. These findings complement preexisting literature. BDNF-AS has been shown to promote amyloid deposition [119] and mitochondrial dysfunction [120] and is upregulated in murine AD models [121] as well as in AD human blood [119]. BDNF-AS knockdown has been shown to suppress apoptosis and inflammatory pathway-related proteins in vivo [121]. Secondly, circ_0049472, a mainly cytoplasmic ncRNA, was upregulated in a cell model of AD, while miR-22-3p, which was shown to have neuroprotective and anti-apoptotic effects in other diseases [122], was downregulated [48]. Mechanistically, it was shown that circ_0049472 binds to miR-22-3p and that circ_0049472 knockdown resulted in the normalization of mitochondrial membrane potential (MMP), increased BCL-2 expression and decreased BAX and cleaved-caspase-3 [48]. Thirdly, in another study, neurons treated with Aβ−42o were exposed to extracellular vesicles containing miR-148a-3p, miR-21-5p, miR-103a-3p. This reduced BAX and BAD expression, increased anti-apoptotic BCL-2 levels, improved MMP, restored mitochondrial populations and normalized mitochondrial electron transport-related gene expression [67]. These results suggest that miR-148a-3p, miR-21-5p, miR-103a-3p exert protective effects on mitochondrial function and prevent mitochondrially mediated apoptosis, but further mechanistic confirmation is required to confirm which miR has the most powerful protective effect. miR-148a-3p is downregulated in AD patients’ serum [123, 124] and blood [125] and can be upregulated via exercise interventions [126]. Meanwhile, miR-21-5p levels in cerebrospinal fluid (CSF) were upregulated in a small group of AD patients [127] while whole blood and plasma studies report conflicting regulation patterns [128–130]. MiR-21-5p upregulation also indirectly protected against neuroinflammation and mitochondrial dysfunction in AD neurons treated with gold nanoparticles [81]. Similarly abundant studies are available on miR-103a-3p. It has been linked to several AD hub genes [131] as well as childhood trauma-related disorders [132], multiple sclerosis[133], and PD [134]. Abyadeh et al. (2025) identified links between miR-103 and mitophagy genes in AD, as discussed in another section of this article [54]. One other study found that miR-103a-3p is downregulated in AD fundamental models [135] but miR-103a-3p levels in patients’ serum were also lowered by exercise and cognitive rehabilitation [136]. Thus, the dysregulation pattern of miR-103a-3p has yet to be thoroughly investigated. Fourth, Zhang demonstrated that H19 is upregulated while miR-129 is downregulated in the brains of AD mice. Silencing lncRNA H19 increased miR-129 levels, which in turn reduced p53 and Ki-67 expression, promoting Aβ25–35-induced PC12 cell viability [47]. Overall, the available clinical evidence on H19 dysregulation in AD [137] and mild cognitive impairment [138] is inconclusive. Lnc-RNA H19 demonstrates, however, broader mitochondrial relevance across various organ systems [139–142]. Fifth, miR-106b-5p is downregulated AD plasma [58] and whole-blood samples [143] and it has been designated as a hub miR in AD [144]. In their bioinformatics analysis, Nguyen et al. (2024) linked miR-106b-5p, miR-17–5p, and miR-16–5p to autophagy and immune signaling in the AD hippocampus [56]. There is evidence that miR-106b-5p promotes mitochondrial oxidative stress [145, 146] and modulates the BAX/BCL-2 and caspase-dependent apoptotic signaling pathways in SH‐SY5Y cells [146]. Therefore, future research could investigate the link between miR-106b-5p dysregulation and mitochondrial dysfunction in AD. miR-16-5p is upregulated in AD mouse brains [147], but the data on its pro-apoptotic effects via BCL-2 downregulation [147] is in conflict with its anti-apoptotic function mediated by BACE1 inhibition [148]. Comparatively less evidence exists for miR-17–5p; it demonstrates protective effects on mitochondrial autophagy in the myocardium [149] while disrupting the amyloid pathway in the microglia [150]. Sixth, experimentally induced overexpression of miR-9-5p in murine hippocampal neurons exposed to Aβ resulted in reduced oxidative stress via downregulated GSK-3β, which in turn decreased BAX and cytochrome C and increased BCL-2 [69]. Interestingly, another in vitro AD study concluded that miR-9-5p inhibits autophagy [151]. LncRNA BDNF-AS acts as a competing endogenous RNA for BACE1 by competitively binding miR-9-5p, which promotes amyloid pathology [119]. Further studies are necessary to clearly establish which molecular mechanism of interest should be employed therapeutically as far as miR-9-5p is concerned. It must be noted that the direction of observed dysregulation trends of miR-9-5p across experimental models and patient tissue samples vary. For example, the above-mentioned study reports miR-9-5p downregulation in Aβ-treated neurons [69], whereas another study noted the opposite trend in the hippocampus of APP/PS1 transgenic mice [66]. In patient studies, Yuen et al. assessed miR-9-5p across eight independent human AD-versus-control brain comparisons spanning six brain regions through an omic dataset meta-analysis, but no general direction of dysregulation for miR-9-5p was established because the pooled analysis did not yield statistical significance [51]. Meanwhile, Cieślik et al. 2020 [57] used temporal-lobe tissue from three AD brains and three controls and observed miR-9-5p upregulation in AD brains [57]. Further diverging reports come from exosome-enriched compared to raw AD CSF [152] and AD brain tissue datasets [153]. These discrepancies may be associated with differences in sample sizes and types, choice of sampled brain region and the various factors influencing in vivo transgenic AD mouse microRNA signatures compared to cell monocultures. So far, we have discussed the ncRNAs relevant to the mitochondrial apoptosis pathway in AD. In the following, we focus on autophagy. An additional important regulator of autophagic flux and mitochondrial bioenergetics is GABA type A receptor-associated protein-like 1 (GABARAPL1), which is an Atg8-family protein contributing to the later stages of autophagosome maturation [154, 155]. GABARAPL1 is downregulated in the human AD frontal lobe [156]. Loss of GABARAPL1 impairs autophagic flux and mitochondrial clearance, leading to the accumulation of dysfunctional mitochondria [154]. Our review identified that several miRNAs are relevant to GABARAPL1 function in AD. Xu et al. (2024) identified GABARAPL1 as an AD-associated biomarker, with its downregulation confirmed across human AD transcriptomic datasets. Bioinformatic analyses performed within the same study revealed that miR-195-5p, miR-16-5p, miR-155-5p, miR-497-5p, miR-15b-5p, miR-143-3p, and miR-133b are predicted to target GABARAPL1 [62]. MiR-195-5p was additionally linked to neuroinflammation and amyloid toxicity in another study investigating differentially expressed miRNAs in AD [51]. Among these miRNAs predicted to target GABARAPL1, several have also been reported as dysregulated in human AD samples, as will be discussed below. Plasma, serum or whole blood dysregulation has been documented for miR-195-5p and miR-16-5p [157], miR-497-5p [158], miR-15b-5p [159, 160], miR-143-3p [161, 162], and miR-133b [163, 164]. MiR-195-5p [165] and miR-155-5p are upregulated in the AD human brain [166], while miR-16-5p was additionally dysregulated in CSF [167]. The second autophagy regulator of interest is TFEB (transcription factor EB), which acts as a major transcriptional regulator of autophagy-lysosomal genes [168]. TFEB activity is inhibited by GSK3β and mTOR signaling, which promotes Aβ and phosphorylated tau accumulation and cognitive decline in AD [112]. TFEB also regulates electron transport chain complex I, supports mitochondrial biogenesis via transcriptional regulation and also ensures the expression of mitophagy receptors nuclear dot protein 52 (NDP52) and optineurin (OPTN) [169]. TFEB has been validated as an AD biomarker in SH-SY5Y neuronal cells and is predicted to be targeted by miR-29a-3p, miR-124-3p, miR-29b-3p, and miR-29c-3p [62]. One in vitro study used miR-124-3p mimics to increase GSK-3β phosphorylation and consequently reduce tau protein-induced apoptosis [170]. miR-29b-3p exhibited significant downregulation in the whole blood of PD but not AD patients [171]. By contrast, miR-29c-3p was not only downregulated in the serum of AD patients, but was also shown to upregulate the mitochondrial epistasis gene signal transducer and activator of transcription 3 (STAT3) activity in vitro and improve cognitive function in rats [172]. Mesenchymal stem cell-derived extracellular vesicles containing miR-29c-3p ameliorated AD pathology in vivo and in vitro. The therapeutic effect of miR-29c-3p was mediated by BACE1 inhibition [173]. As with apoptosis and autophagy, there are ferroptosis-relevant ncRNAs implicated in AD pathophysiology. Ferroptosis is a cell death mechanism which is iron-dependent and reactive oxygen species-reliant [174]. It promotes Aβ, hyperphosphorylated tau aggregation [175] and neurodegeneration [176]. The AD brain displays increased iron deposition [177]. Iron is necessary for mitochondrial electron transport chain function, heme biosynthesis, such that disruptions in iron homeostasis also induce mitochondrial dysfunction [178]. We identified one study which investigated the involvement of miR-7a in ferroptotic phenomena and the effect thereof on the mitochondria. Ramesh et al. (2025) showed that miR-7a was upregulated in AD mice cortices and hippocampi while the pro-inflammatory Krüppel-like factor 4 (Klf4) was upregulated in AD mouse brain tissue. MiR-7a mimics decreased Klf4 protein expression in untreated SH-SY5Y cells while, in a separate experiment, ferroptosis-induced SH-SY5Y cells displayed increased Klf4 protein. MiR-7a mimic treatment of SH-SY5Y cells followed by ferroptosis induction prevented mitochondrial fragmentation and restored MMP levels. Importantly, the cells used in the above experiments were untreated with Aβ oligomers, nor was miR-7a expression evaluated following Aβ oligomer treatment [65]. The observed upregulation of both miR-7a and Klf4 in AD mice coupled with the protective effects of miR-7a mimics in ferroptosis models could be interpreted as follows: the observed miR-7a upregulation in vivo plays a partial compensatory role in the context of AD-related Klf4 overexpression, as also noted by the authors [65]. Klf4 belongs to the family of zinc finger transcription factors with various regulatory functions [179], such as iron level modulation and lipid peroxidation [65]. The mitochondria has been implicated in the initiation and mediation of ferroptotic signaling due to its role in mtROS production, iron accumulation [180], the TCA cycle and electron transfer chain activity [181]. Conversely, ferroptotic stimuli lead to mitochondrial morphology and metabolism abnormalities [182]. To summarize this section, preclinical AD model data indicates that lncRNA BDNF-AS [79] and circ_0049472 [48] promote mitochondrial-dependent apoptosis by downregulating BCL-2 and upregulating BAX and caspase-3. Conversely, miR-148a-3p, miR-21-5p, miR-103a-3p [67] exert the opposite effect, preventing apoptosis. LncRNA H19 silencing results in microRNA-129 upregulation and consequent anti-apoptotic effects in vitro via the regulation of p53 and Ki-67 expression [47] and other experimental data from non-AD models point to its involvement in the regulation of PTEN and PINK1 [123, 139, 142]. Additionally, miR-7a seems to attenuate ferroptosis-associated mitochondrial fragmentation [65], but no clinical data is available on its dysregulation in AD patients, although decreased miR-7 expression has been observed in PD post-mortem brains [183]. Bioinformatic evidence converging with preexisting mechanistic and clinical studies further implicate miR-106b-5p in autophagic regulation in AD [56], possibly via BAX/BCL-2 signaling and mitochondrial oxidative stress [146]. Similarly, miR-195-5p, miR-16-5p, miR-155-5p, miR-497-5p, miR-15b-5p, miR-143-3p, and miR-133b are predicted regulators of GABARAPL1 [62], which importantly regulates autophagosome formation and mitochondrial clearance [154, 155]. TFEB, another major autophagy regulator which also regulates mitophagy receptor expression [112, 169] is predicted to be targeted by miR-29a-3p, miR-124-3p, miR-29b-3p, miR-29c-3p in AD [62]. These predicted miRNA-target relationships warrant further experimental validation to clarify their involvement in autophagic and mitochondrial dysregulation in AD. NcRNAs Influence Mitochondrial Biogenesis and Mitochondrial Gene Expression in AD Altered mitochondrial biogenesis is one of the causes for reduced mitochondrial populations in AD hippocampi [184]. Multiple mitochondrial transcription factors and coactivators are involved in mitochondrial biogenesis regulation. PGC-1α plays a central role due to its role in activating mitochondrial gene transcription and function. Two key regulators of PGC-1α in low energy conditions are the energy sensor AMP-activated protein kinase (AMPK), which activates PGC-1α via phosphorylation when the AMP/ATP ratio increases, and SIRT1, which deacetylates PGC-1α when the NAD+/NADH ratio increases [185]. Our study identified that several ncRNAs intervene in this mitochondrial regulatory axis in AD. MiR-34a, miR-146a, and miR-155 [57] and miR-9-5p [57, 66] are negative regulators of SIRT1 while miR-485 downregulates PGC-1α [55] in AD. MiR-9-5p, miR-132-3p, and miR-212-3p all have been shown to inhibit SIRT1 in other disease models [186–189]. MiR-9-SIRT1 interactions have previously been reported in neurodegeneration studies [189]. MiR-132-3p inhibition conserves SIRT1-PGC-1α-NRF2 signaling in oxidative stress [187]. Below, we will discuss reports on the dysregulation of these miRNAs in AD, having already addressed miR-9-5p dysregulation trends in the previous section. MiR-485 is a highly conserved brain-enriched miR whose dysregulation in the AD brain [55], blood and CSF [190], salivary exosomes [191] as well as in PD [190] has been previously reported. Notably, miR-34a, miR-146a, miR-155 [57] and miR-485 [55] were significantly upregulated in AD patient brains. Finally, Yuen et al. observed statistically significant dysregulation for miR-132-3p, miR-212-3p in AD brains their meta-analysis [51]. Alongside PGC-1α, mitochondrial genome maintenance exonuclease 1 (MGME1) is also involved in mitochondrial gene expression. It encodes an exonuclease responsible for mtDNA processing and mitochondrial genome maintenance [192, 193]. There is evidence that links mtDNA damage in AD to energy depletion and increased oxidative stress [194]. Our review identified a bioinformatics study which showed that MGME1 is significantly downregulated in the human AD brain and that miR-922 can bind to the 3′ UTR of MGME1 and inhibit its expression [53]. Finally, Xu et al. (2025) found miR-132 and miR-129-5p to be downregulated in AD brains and identified potential interactions between miR-124-3p and the mitochondrial epistasis genes STAT3 and cyclin-dependent kinase 6 (CDK6) [52]. STAT3 acts as a transcription factor which regulates mitochondrial calcium homeostasis, respiration [195], electron transport chain activity [196], and participates in mtDNA transcription [197, 198]. Additionally, it has an anti-apoptotic function, stimulating BCL-2 transcription [199]. CDK6 regulates exit from the G1 cell cycle phase [200]. It has been shown to interact with the promoters of oxidative phosphorylation (OXPHOS) genes and one mitochondrial transcriptional regulator [201]. One murine study on neurotoxicity confirmed that miR-124-3p downregulates STAT3 and lowers neuroinflammation [202]. Interestingly, serum miR-124-3p is upregulated by rosuvastatin in AD patients [203]. Its upregulation was also reported in post-mortem AD human brains [204]. Therapeutically, exosomally delivered miR124-3p attenuated neuroinflammation in a triculture AD model [205] and exerted neuroprotective effects in vitro and in vivo in other AD studies [170, 206]. MiR124-3p dysregulation has also been observed in PD [207] and traumatic brain injury [208, 209]. The STAT3/miR-204/transient receptor potential mucolipin-1 (TRPML1) signaling axis was also implicated in mitochondrial autophagy and oxidative stress regulation in a study by Zhang (2021), wherein miR-204 downregulation by STAT3 resulted in increased TRPML1 expression, reduced mitochondrial dysfunction and ROS production and promoted mitophagy in AD [71]. Li (2025) developed CeO2-DNA-RNA hybrid chain (DRP)/mitochondrial aptamers (MA) nanoclusters designed to silence mitochondrial miR-204. In vitro, treatment improved mitochondrial membrane potential and ROS levels and ameliorated AD-related pathological TRPML1/STAT3 autophagic signaling. In AD mice, the nanoclusters accumulated in brain tissue and exerted similar effects on hippocampal autophagic signaling. However, miRNA-204 levels were only qualitatively assessed via fluorescence studies in mice and no fold changes were reported for the RT-qPCR data on the Aβ-treated SY5Y cells [85]. One clinical study found that miR-204 was downregulated in neuron-derived serum exosomes in AD patients [210] while blood lymphocyte miR-204 was upregulated in another study [211]. MiR-204 downregulation in CSF has been reported in frontotemporal dementia [212] as well as AD [213]. MiR-423 was upregulated in the hippocampus and medial frontal gyrus, whereas miR-98 was downregulated in the cerebellum of AD patients [84]. In vitro studies in non-AD cell lines confirmed that miR-423 and miR-98 downregulate isocitrate dehydrogenase 2 (IDH2), which helps maintain mitochondrial redox balance. IDH2 levels in the selected brain regions were, however, not measured directly [84]. Lower plasma levels of miR-423-5p correlated with cognitive deterioration and increased amyloid deposition in AD patients [214] and experimental evidence from non-AD models supports miR-423-5p involvement in the regulation of mitochondrial function [215, 216]. MiR-98 reportedly displays dysregulated expression in AD patients [217] and has also demonstrated a connection with mitochondrial dysfunction in a scopolamine model of AD [218]. In summary, given that SIRT1 exerts a protective effect on mitochondrial biogenesis, hippocampal-dependent memory and helps ameliorate neurotoxic Aβ peptide accumulation [219], the modulation of miR-9, miR-34a, miR-146a, miR-155 [57], miR-9-5p [51, 66], miR-132-3p, and miR-212-3p may be a viable therapeutic approach in optimizing mitochondrial biogenesis in AD. MiR-485 [55] antagomir treatments to stimulate mitochondrial biogenesis by increasing PGC-1α could also be of great research interest given the broadly observed dysregulation of miR-485 in the blood, CSF and brain of AD patients. MiR-204 inhibition via STAT3 [71] and nanoclusters [85] has positive effects on mitochondrial autophagy and mitochondrial function, respectively. As human full-length tau accumulation [220] and Aβ pathology [221] influence STAT3 levels, miR-204 antagomirs could be employed to compensate for the detrimental effects of low STAT3 on mitochondrial autophagy. MiR-423 and miR-98 may influence mitochondrial redox balance [84]. Additionally, the preliminary bioinformatic data suggest that miR-922 may contribute to MGME1 downregulation [53] and to the accumulation of mitochondrial genome defects in AD. Finally, given the documented involvement of miR-124-3p in neuroinflammatory signaling [222], mitochondrially-mediated apoptosis [223] and mitochondrial localization processes [224] in the context of neurodegeneration, the bioinformatic findings of Xu et al. regarding the regulation of STAT3 and CDK6 by miR-124-3p [52] also require additional experimental validation. NcRNAs Regulating Mitochondrial Dynamics in AD Disturbances in the expression of mitochondrial fusion and fission genes have been observed across neurodegenerative diseases [225]. The AD human hippocampus displays altered mitochondrial dynamics with intensified mitochondrial fission [226] while therapeutic measures stimulating fusion may ameliorate disease course [227]. Mitochondrial fusion involves the fusion of two mitochondrial outer membranes via GTP hydrolysis and a subsequent fusion of the inner membranes. In mammals, there are three key dynamin-related GTPases controlling mitochondrial fusion: mitofusin 1 (MFN1), mitofusin 2 (MFN2), and optic atrophy 1 (OPA1). MFN1/MFN2 fuse the outer membranes, while OPA1 ensures the fusion of the inner membranes [228]. Mitochondrial fission is coordinated by dynamin-related protein 1 (DRP1), a cytosolic protein recruited by various mitochondrial adaptors to form an oligomeric ring around the constricted mitochondria. DRP1 undergoes GTP hydrolysis and dynamin 2 is recruited to the constricted site to complete the division. Inner mitochondrial membrane constriction takes place at mitochondria-ER contact regions and depends on Ca2+ [229]. Mitochondrial fission protein 1 (FIS1) is a mitochondrial outer membrane protein which is involved in the fission complex [230] and facilitates DRP1 recruitment and assembly [231]. It has also been shown to inhibit MFN1, MFN2 and OPA1 activity [232]. Miro1 and Miro2 are outer mitochondrial membrane GTPases involved in inter-organelle interactions, mitochondrial transport on microtubule and actin cytoskeleton and mitophagy [233]. Miro proteins inhibit mitochondrial fission in the presence of normal Ca2+ concentrations and promote fission in the presence of high Ca2+ via DRP1 [234]. Our systematic review has identified several miRNAs which affect mitochondrial dynamics in AD. Woo et al. (2020) demonstrated that Miro2 is a critical target of miR-351-5p. Murine AD hippocampal tissue showed upregulated miR-351-5p and downregulated Miro2, a finding that was supported by transcriptomic data. In HCN cells, miR-351-5p was shown to directly downregulate Miro2 mRNA, leading to increased mitochondrial fragmentation and dysfunction and autophagy-dependent cell death mediated via PINK1/Parkin activity [63]. In another study, across human and 3xTg-AD model mice hippocampi and HCN cells, Han et al. (2024) observed a downregulation of OPA1 and MFN2. This downregulation was linked to the upregulation of miR-323-5p and miR-370-3p, observed in both 3xTg-AD mice and in HCN cells following insulin withdrawal. These microRNAs decreased OPA1 expression, while miR-150-3p was additionally shown to directly target MFN2, promoting mitochondrial fragmentation and neuronal cell death [61]. Zhang et al. bring complementary results by finding that miR-195 was upregulated in the SAMP8 mice and that miR-195 inhibits MFN2 translation and therefore mitochondrial fusion [70]. There is evidence to support the therapeutic potential of miR-195 inhibition in AD. Intraventricular delivery of miR-195 inhibitors improves cognitive performance, enhances postsynaptic density, and increases the activity of mitochondrial respiratory chain complexes I and II in AD mouse models, supporting its role as a regulator of mitochondrial and synaptic function [235]. Conversely, the studies of Kumar et al. showed that miR-455-3p upregulates OPA1, MFN1, and MFN2 and downregulates the fission-associated genes DRP1 and FIS1 in mouse neuroblastoma cells transfected with mutant APP [74] and ameliorates mitochondrial dynamics and biogenesis in mice [236]. These findings are in agreement with a subsequent study which confirms the beneficial effect of miR-455-3p on mitochondrial dynamics and mitophagy [237]. Furthermore, miR-455-3p dysregulation has been observed across post-mortem AD brains, B-lymphocytes, and fibroblasts [238, 239], making it a potent biomarker and therapeutic development focus. On the other hand, lncMtDloop, which was found to be downregulated in human and murine AD mouse brains, was also proved to promote fusion by increasing OPA1 and to decrease fission by downregulating DRP1 [75], but additional clinical and preclinical data is necessary. Vijayan et al. (2024) found miRNA PC-5P-12969 to be elevated in the human AD frontal cortex, in APP and tau transgenic mice, Aβ knock-in mice as well as in HT22 cells treated with mutant APP (mAPP). MiRNA PC-5P-12969 was shown to downregulate GSK3 and ameliorates mitochondrial respiration, length and ATP production while lowering mitochondrial fragmentation in mAPP-treated cultures [64]. GSK3 is a multifunctional Ser/Thr protein kinase which is dysregulated neurodegeneration [240]. GSK-3β overexpression negatively regulates mitochondrial biogenesis via PGC-1α and PI3K/Akt signaling, promotes mitochondrial fragmentation by increasing DRP1 and mitochondrial fission factor and disturbs mitochondrial bioenergetics by suppressing complex I activity [241]. Wang et al. (2019) found that lncRNA SNHG1 knockdown upregulates miR-137, which in turn contributed to neuroprotection by repressing kringle-containing transmembrane protein 1 (KREMEN1) expression in Aβ-treated neurons [45]. KREMEN1 is involved in the Wnt/β-catenin pathway regulating metabolism and homeostasis [242] and one of its cytoplasmic regulatory components, GSK-3β, has been shown to exhibit regulatory effects on mitochondrial activity [241, 243]. MiR-137 dysregulation has been linked to AD progression [244, 245], mitochondrial biogenesis [246] and mitophagy [247], but the neuroprotective role of miR-137 is not supported across all reports. Garcia et al. (2021) found that miR-146a was upregulated in SH-SWE cells and downregulated in PSEN1ΔE9 mutated neurons [72]. MiR-124 mimic resulted in a downregulation of exosomal miR-125b/miR-21/miR-146a/miR-155 and an increase in mitochondrial transmembrane potential in both of the above-mentioned cell lines. Interestingly, miR-124 inhibition promoted mitochondrial fusion by increasing MFN2 levels [72]. Another study using an in vitro model of chronic neuroinflammation found that TNF-α exposure in hippocampal neurons results in elevated miR-34a and miR-146a, which correlated with mitochondrial dysfunction [248]. MiR-34a is also elevated in d-galactose (d-gal)-induced oxidative stress. Interventions like swimming and miR-34a inhibitors resulted in decreased hippocampal DRP1 and MFN2, as well as increased mitochondrial biogenesis (PGC-1α) in rats [249]. Elsewhere, it was reported that miR-146a downregulates Parkin [250]. Both miR-34a and miR-146a downregulate BCL-2 [251]. Therefore, miR-146a antagomirs [252] could restore mitophagic balance and mitochondrial function in the AD brain. miR-146a therapies have been preliminarily investigated in relation to cortical impact injury [253] and ALS [254]. Briefly, in the preclinical AD studies, we have identified that mitochondrial fission is promoted by miR-351-5p via Miro2 downregulation [63], by miR-150-3p, miR-323-5p, and miR-370-3p via OPA1 and MFN2 downregulation [61], and by miR-195 [70] and miR-124 [72] via MFN2 inhibition. By contrast, miR-455-3p [74] and lncMtDloop [75] upregulate fusion-related genes, while miRNA PC-5P-12969 [64, 241] and miR-137 [45] seem to reduce mitochondrial fission rates indirectly by downregulating GSK3. NcRNAs Regulating Mitochondrial Protein Import in AD Amyloid precursor protein causes mitochondrial protein import dysfunction by blocking the translocase of the outer membrane (TOM) and translocase of the inner membrane 23 channels (TIM23), which causes Complex IV dysfunction and favors ROS production. Aβ also prevents the entrance of precursor proteins through the mitochondrial outer membrane [255]. APP accumulation in the import channels is proportional to AD severity and to mitochondrial dysfunction [256]. In this section, we summarize the available data on ncRNA regulation of the mitochondrial import machinery in AD. There are five mitochondrial protein import pathways, and the classical import pathway relies on TOM and TIM23 [257]. TOMM20 is part of the translocase of the outer mitochondrial membrane complex (TOMM). TOMM20 functions as a receptor for mitochondrial precursor proteins and nuclear-encoded oxidative phosphorylation subunits. Dysregulated TOMM20 expression has been linked to ischemic injury damage, Meniere’s disease and cancer [258]. Based on bioinformatics analyses, TOMM20 was predicted to be targeted by Let-7a-2-3p and miR-204-3p in AD [62]. Little data is available regarding these miRNAs in AD: miR-204-3p is downregulated in the murine AD hippocampus and relieves oxidative stress [259] and neuroinflammation [260]. Let-7a-2-3p is downregulated in the epileptic murine hippocampus [261] but has not been quantified in AD models. TIMM13 is part of the translocase of the inner membrane (TIM) complex and plays a role in mitochondrial import and biogenesis [262] as well as mitophagy [263]. Kim et al. (2021) observed TIMM13 downregulation in the hippocampi of AD patients. In H4-APPswe and SH-SY5Y cells, miR-1273g-3p directly downregulated TIMM13 [58]. Only one other preclinical study investigated miR-1273g-3p, namely, as an autophagic promoter in SH-SY5Y cells [264]. Additional clinical validation of the biomarker potential of this miR is necessary and more detailed descriptions of its role in mitochondrial protein import are warranted. Voltage-dependent anion channel 1 (VDAC1) is an outer mitochondrial membrane protein responsible for metabolite, ROS, and Ca2+ transport as well as apoptotic signaling via BAX and BCL-2 [265]. VDAC1 upregulation has been reported in the AD post-mortem brain and in APP mice [266]. In AD, VDAC1 interacts with Aβ and phosphorylated tau [267], causing mitochondrial pore dysfunction, which affects protein import into the mitochondria and downstream oxidative phosphorylation and structural abnormalities [266]. In the temporal cortex of AD patients, miR-146a levels were upregulated and inversely correlated with VDAC1 expression. MiR-146a compromised oxidative phosphorylation in mixed glial cells exposed to Aβ [59]. Since VDAC1 downregulation seems to ameliorate mitochondria function in AD mice, [268, 269], VDAC1 modulation via mir-146a may be an attractive research direction. Finally, in a familial AD cell line carrying the PSEN1 L150P mutation, circ_0001946 was found to be upregulated. Circ_0001946 downregulates ubiquitin-conjugating enzyme E2A (UBE2A) [49]. UBE2A encodes a neuronal-specific ubiquitin conjugating enzyme [270] which promotes the mitochondrial translocation of Parkin [271] and also ubiquitinates mitofusins, TOM20, TOM70 and VDAC1 [272]. In AD, outside of its mitochondrial function, UBE2A has been shown to contribute to the proteolytic clearance of amyloid peptides [273]. At the time of writing this article, no additional studies regarding circ_0001946 in relation to AD or the mitochondria could be identified, although circ_0001946 has been studied in glioblastoma and other cancers [274, 275]. Altogether, the strongest experimental and clinical evidence supporting ncRNA involvement in mitochondrial protein import dysfunction concerns miR-146a and miR-1273g-3p as downregulators of VDAC1 and TIMM13, respectively. A singular study suggests that circ_0001946 downregulates UBE2A and thus indirectly helps regulate TOM and VDAC1 expression while only bioinformatic data nominate Let-7a-2-3p and miR-204-3p as regulators of TOMM20 in AD. NcRNAs with Emerging Mitochondrial Roles in AD The major ncRNAs directly implicated in AD mitochondrial regulation and the mechanisms at play have been synthesized in Fig. 3. In this section, we discuss several ncRNAs of emerging mitochondrial relevance in AD mediated by oxidative stress, neuroinflammation, and endoplasmic reticulum-mitochondria interactions. The most abundant evidence involves miR-375. A bioinformatics analysis, whose findings were replicated in Aβ-treated SH-SY5Y cells and AD mice, identified miR-375 upregulation in the cortex of AD patients while lncRNA WT1-AS and the target gene of miR-375—transcription factor SIX homeobox 4 (SIX4)—were downregulated [46]. Interestingly, another study found that miR-375 was upregulated in the plasma and CSF of aMCI but not in AD patients [58], which points to the need of establishing the expression profiles of miR-375 across distinct tissue types in AD. Wang et al. further showed that lncRNA WT1-AS overexpression suppressed miR-375 and consequently upregulated SIX4 while improving mitochondrial oxidative status and ATP production in vitro. [46] Thus, the miR-375/SIX4 axis may be indirectly involved in mitochondrial homeostasis. Conversely, miR-375-3p mimic treatment seems to improve neuronal mitochondrial function in Aβ-treated neuroblastoma cells and promote mitochondrial autophagy in AD mice [82] without promoting apoptosis in AD [276]. The diverging findings in the effects of miR-375 on mitochondrial function in AD reported in the two studies could be explained by differences in protocols. Wang et al. (2020) differentiated SH-SY5Y cells with retinoic acid and BDNF [46] while the second study does not mention pre-treatment of SH-SY5Y cells before Aβ exposure [82]. It is additionally noted that under physiological conditions, miR-375 can inhibit autophagy, while in early AD pathogenesis, it exerts the opposite effect [82]. The differences in signaling pathway studied and cell culture likely contribute to the differences in mitochondrial outcomes. Thus, the mechanistic details of the effects of miR-375 on mitochondrial function in AD require further investigation. Schematic representation of the major ncRNAs which directly regulate mitochondrial functions in Alzheimer’s disease (AD). The figure summarizes the key miRNAs, lncRNAs, and circRNAs reported to modulate mitochondrial dynamics, biogenesis, mitophagy, ferroptosis, apoptosis, and protein import in AD. Arrows indicate activation or positive regulation, whereas T-bar-ended lines indicate inhibition or negative regulation To continue, there is evidence of miR-218-5p, miR-124-3p downregulation in the human and murine AD brain. These miRNAs downregulate the endoplasmic reticulum protein reticulon 3 (RTN3), which regulates mitochondrial fission and influences neuronal mitochondrial localization [60]. Neuronal RTN3 is upregulated in AD and Parkinson’s disease as a nonspecific endoplasmic reticulum stress response [277]. RNT3 promotes BCL-2/Beclin-1 interactions [278] and mediates BCL-2 translocation to the mitochondria under endoplasmic reticulum stress [277]. In APP-transfected SH-SWE cells, miR-124-3p also inhibits receptor for advanced glycation end-products (RAGE)/high mobility group box 1 (HMGB1) axis signaling, which alleviates neuroinflammation and indirectly optimizes mitochondrial MMP [73, 279]. Hence, miR-124-3p indirectly influences mitochondrial function by regulating endoplasmic reticulum-mitochondria interactions and neuroinflammation signaling. Meanwhile, several downregulated lncRNAs demonstrating dysregulation in AD have been directly linked to mitochondrial processes in other disease models. LncRNAs MEG3, LOXL1-AS1, ADAMTS9-AS2, and A2M-AS1 are downregulated in a preclinical AD study by Corsi et al. [49]. MEG3 and ADAMTS9-AS2 dysregulation in AD has also been reported in other studies [280–282]. MEG3 [283] and A2M-AS1 [284] have been linked to mitophagy while LOXL1-AS1 helps regulate mitochondrially mediated apoptosis [285, 286]. ADAMTS9-AS2 appears to indirectly regulate pyroptosis [287], an inflammatory cellular death pathway which causes mitochondrial damage [288]. Overall, MEG3 and ADAMTS9-AS2 show promise as potential AD biomarkers and mitochondrial regulators. A mouse study suggests that the miR-96-5p/Foxo1/Lcn2 axis may be involved in mitochondrial dysfunction and oxidative stress in AD. MiR-96-5p inhibition upregulated Foxo1 and downregulated Lcn2 in AD mouse osteoblasts while reducing AD mouse hippocampal oxidative stress. Foxo1 overexpression and Lcn2 silencing separately improved hippocampal mitochondrial ultrastructure [86]. Serum miR-96-5p and neuronal inflammatory mediator Cathepsin B were shown to have good diagnostic accuracy for AD-related cognitive impairment [289] and miR-96-5p inhibition was seems to promote Aβ clearance [290]. The literature thus suggests miR-96-5p involvement in AD, making the future confirmation the direct impact of miR-96-5p inhibition on mitochondrial parameters in AD a reasonable research objective. Two preclinical studies investigate circ_0004381 and lncRNA-ATB, which could potentially be used to target AD-related mitochondrial dysfunction induced by oxidative stress. Firstly, Li et al. observed that the knockdown of circ_0004381 in AD neurons reduced COX-2 (cyclooxygenase-2) expression and microglial iNOS (inducible nitric oxide synthase) [80], the latter being involved in inflammatory neurodegeneration [291] and mitochondrial dysfunction [292]. Secondly, lncRNA-ATB was upregulated in Aβ25–35 induced PC12 cell injury and in turn repressed miR-200, which resulted in oncogene zinc finger protein 217 (ZNF217) overexpression, oxidative stress and MMP reduction [76]. ZNF217 suppresses PI3K-AKT activation in Aβ-damaged neurons [293], which increases oxidative stress and mitochondrial vulnerability in neurons [294]. While no clinical data are available for circ_0004381, lncRNA-ATB upregulation has been reported in the CSF and serum of AD patients. A small clinical study observed the downregulation of miR-6501-5p and miR-143-3p in the whole blood of AD patients compared to controls. These miRNAs were predicted to target mitochondrial genes ATP5E and NDUFB9, NDUFA9, NDUFS1 and COX4I1, respectively, contributing to the regulation of the various functions such as the electron transport chain and respiration [83]. So far, miR-143-3p has been linked to mitochondrial apoptosis in a myocardial ischemia model [295]. Two multiomics studies provide additional research avenues concerning ncRNA-mitochondrial gene interactions. One multiomics study predicted that miR-103 and miR-484 are coregulatory miRNAs for several AD-relevant oxidative phosphorylation and tricarboxylic acid cycle genes [54]. In AD patients, miR-103 demonstrates downregulation in plasma [296] while miR-484 is downregulated in brain tissue [297]. In other disease models, miR-484 was linked to oxidative phosphorylation [298] and mitochondrial fission [297, 299] while miR-103 inhibited mitophagy [134, 300]. Another bioinformatic study proposes miR-7113-5P and miR-3130-3P as regulators of ABAT gene AD-risk alleles [301]. ABAT (4-aminobutyrate aminotransferase) interacts with APOE, potentially contributing to AD pathogenesis [302], and is involved in several mitochondrial processes, from the catabolism of inhibitory neurotransmitter γ-aminobutyric acid to OXPHOS enhancement [303, 304]. However, the dysregulation of miR-7113-5P and miR-3130-3P in AD has yet to be clinically and experimentally confirmed. Lastly, the downregulation of the cytoplasmic miR processing enzyme Dicer1 due to amyloid pathology appears to impact mitochondrial homeostasis indirectly by inducing oxidative stress [50]. Dicer1 exerts protective effects on mitochondrial metabolism [305, 306] and OXPHOS [307] in several preclinical AD studies. This is an incentive to further study Dicer1 expression in the post-mortem AD brain and to identify the main mechanism by which Dicer1 ameliorates mitochondrial function in AD. A number of studies not included in our systematic data extraction because their models simulated limited aspects of AD neuropathology indicate that additional ncRNAs may be considered in the exploration of mitochondrial dysfunction in future research. MiR-98 indirectly prevented mitochondrial apoptosis pathway activation in a scopolamine AD-like model [218]. Several studies using AlCl3 to induce AD-like neurodegeneration indicate a possible link between miR-134 and altered mitochondrial function as a result of oxidative stress [308–310]. Lnc-NKILA similarly exacerbated oxidative stress and lowered MMP in streptozotocin-induced neuronal injury [311]. While the experimental models employed above do not ideally reflect the characteristics of AD, these mechanistic insights could inform future in vitro and in vivo studies verifying the above-mentioned mechanism in AD models of higher histopathological fidelity. To summarize this section, we have identified several ncRNAs of potential mitochondrial relevance in AD. There is evidence of miR-218-5p, miR-124-3p, miR-375, miR-103, miR-484, MEG3, ADAMTS9-AS2, miR-96-5p, miR-6501-5p, and miR-143-3p dysregulation in AD models or patients. MiR-218-5p and miR-124-3p regulate endoplasmic reticulum-mitochondria interactions while miR-124-3p additionally inhibits neuroinflammation. Preliminary in vitro data supports the indirect involvement of miR-375 in mitochondrial autophagy in AD, although the exact mechanisms and pathways require further clarification. Despite the documented mitochondrial relevance of miR-103, miR-484, and MEG3 in other pathologies, there is no confirmatory experimental data using AD models. Less evidence is available for circ_0004381, lncRNA-ATB and miR-96-5p which seem to promote mitochondrial dysfunction in vitro by exacerbating oxidative stress, and for miR processing enzyme Dicer1, which has so far not been studied in patient populations, and for miR-7113-5P, miR-3130-3P, which are predicted to target the ABAT gene. Limitations and Future Directions The current study has various limitations. Firstly, the heterogeneity of experimental models and relatively small sample sizes of the translational studies made it impossible to perform a meta-analysis on the collected data. A risk of bias or quality assessment was not performed due to the diverse nature of study designs and documented outcomes, of which some were qualitative and some quantitative. The systematic review protocol has not been pre-registered in a platform such as PROSPERO. Secondly, various discrepancies in the RNA quantification methods and in the observed dysregulation patterns of certain ncRNAs between studies indicate that further research using uniform methodology is needed to establish their pathogenetic relevance. Thirdly, given the intricate relationships between amyloid pathology, oxidative stress and mitochondrial dysfunction [312], whether ncRNA dysregulation is a cause or a consequence of AD and related disorders remains to be determined [313]. Fourth, only a minority of translational studies employed plasma or CSF samples while many used post-mortem transcriptomic data, making it a priority for future AD ncRNA research to focus on larger patient cohorts and to explore body fluid diagnostic methods. In line with this issue, the sample sizes employed in the translational studies were relatively small and larger cohorts will be necessary to validate preexisting findings in human subjects. Notably, there are limited large-scale [314, 315] and longitudinal [316, 317] studies investigating the interplay between ncRNA levels and the evolution of AD. Further studies will be necessary to reach standardized biomarker expression level thresholds. Additionally, although certain ncRNA knockdown and overexpression methods have yielded promising results in the preclinical studies we identified, cell and rodent AD or neuroinflammation models do not reflect the entirety of the biological states involved in human disease and, as such, preclinical data requires validation in humans. In a similar fashion, ncRNAs which show therapeutic promise pose the challenge of firstly configuring a viable delivery mechanism for human subjects and subsequently evaluating their in vivo dose–effect dynamics [318]. Overall, the employment of ncRNAs as diagnostic and biomarker molecules in clinical practice requires further wider scale validation and inter-molecule sensitivity and specificity comparisons. Conclusion This study has outlined the major ncRNAs which are dysregulated in AD and which are involved in the regulation of mitophagy, mitochondrial-mediated apoptosis, mitochondrial dynamics and metabolism. While the state of biomarker and therapeutic research in the field of mitochondrially relevant ncRNAs is in its primary stages, the growing body of evidence surrounding the role of ncRNAs in neurodegenerative pathologies makes this a dynamic and open-ended research avenue. We have discussed the various ncRNAs which have been shown to directly regulate mitochondrial processes in AD. Among mitophagy-relevant ncRNAs which interact with PINK1, miR-140 and lncRNA NEAT1 have the most clinical dysregulation evidence. The miRNAs regulating mitochondrial biogenesis with the best translational research potential are miR-9, miR-34a, miR-146a, miR-155, miR-485, and miR-204. Meanwhile, the best further research candidate in mitochondrial autophagic regulation is miR-106b-5p, as the predicted regulator of AD-relevant autophagic genes has yet to be explored in vitro. Mitochondrially mediated apoptosis is promoted by lncRNA BDNF-AS while miR-148a-3p, miR-21-5p, and miR-103a-3p have antiapoptotic effects, all being dysregulated across various patient tissue types. Mitochondrial dynamics is regulated by several ncRNAs, of which miR-195, miR-124, and miR-455-3p are the most broadly studied in clinical settings as biomarkers, followed by miRNA PC-5P-12969 and miR-137. Finally, experimental and translational evidence points to miR-146a and miR-1273g-3p as regulators of mitochondrial protein import mechanisms in AD. To conclude, miRNAs are by far the most widely studied class of ncRNAs in relation to mitochondrial dysfunction and biomarker development in AD, followed by lncRNAs. The body of research available mainly focuses on elucidating the multi-faceted mechanistic roles of ncRNAs in mitochondrial homeostasis and on identifying viable diagnostic and monitoring candidates. Although promising, the development of ncRNA therapeutics for AD will require extensive further translationally oriented experimental research. Supplementary Information Below is the link to the electronic supplementary material. (DOCX 76.8 KB) Abbreviations Triple-transgenic Alzheimer’s disease model mice Alpha-2-macroglobulin antisense RNA 1 4-Aminobutyrate aminotransferase Alzheimer’s disease A Disintegrin and Metalloproteinase domain 10 Apoptosis-inducing factor Protein kinase B Amyotrophic lateral sclerosis Aβ-positive amnestic mild cognitive impairment Adenosine monophosphate AMP-activated protein kinase Apolipoprotein E β-Amyloid precursor protein Adenosine triphosphate Gold nanoparticles Amyloid-beta Beta-site amyloid precursor protein cleaving enzyme 1 BACE1 antisense RNA Bcl-2-associated death promoter Bcl-2 homologous antagonist/killer BCL2-associated X protein B-cell lymphoma 2 B-cell lymphoma 2 protein Brain-derived neurotrophic factor BDNF antisense RNA Catalase Cyclin-dependent kinase 6 Competing endogenous RNA Circular RNA Cyclooxygenase-2 Cerebrospinal fluid Differentially expressed microRNA Dentate gyrus Double-stranded RNA-specific endoribonuclease 1 Dynamin-related protein 1 Endoplasmic reticulum Extracellular vesicles Mitochondrial fission protein 1 GABA type A receptor-associated protein-like 1 Gene Expression Omnibus Glycogen synthase kinase 3 Glycogen synthase kinase-3 beta Guanosine triphosphate Adult hippocampal neural progenitor (cells) 3-Hydroxyisobutyryl-CoA hydrolase Human induced pluripotent stem cell High mobility group box 1 Human neural stem cell Interleukin-1 beta Inner mitochondrial membrane Inducible nitric oxide synthase Induced pluripotent stem cell Krüppel-like factor 4 Kringle-containing transmembrane protein 1 Microtubule-associated proteins 1A/1B light chain 3 Microtubule-associated protein 1 light chain 3B Long non-coding mitochondrial displacement loop RNA Long non-coding RNA brain-derived neurotrophic factor antisense LncRNA nuclear paraspeckle assembly transcript 1 NF-κB-interacting long noncoding RNA Wilms tumor 1 antisense RNA Long non-coding RNA Lysyl oxidase-like 1 antisense RNA 1 Rush Memory and Aging Project Microtubule-associated proteins 1A/1B light chain 3B Mutant APP Mild cognitive impairment Malondialdehyde Maternally expressed gene 3 Mitofusin 1 Mitofusin 2 Mitochondrial genome maintenance exonuclease 1 MiR-600 host gene MicroRNA Mitochondrial Rho GTPase 1 Mitochondrial Rho GTPase 2 Mitochondrial membrane potential Mitochondrial outer membrane permeabilisation Messenger RNA Mitochondrial DNA Mechanistic target of rapamycin Mitochondrial RNA Nicotinamide adenine dinucleotide (oxidized form) Nicotinamide adenine dinucleotide (reduced form) Nicotinamide adenine dinucleotide phosphate Non-coding RNA Nondemented patients with Alzheimer’s neuropathology Nuclear dot protein 52 Neural precursor cell expressed, developmentally down-regulated 4-like, E3 ubiquitin protein ligase NLR family pyrin domain-containing protein 3 Nuclear respiratory factor 1 Nuclear factor erythroid 2-related factor 2 Optic atrophy 1 gene Optineurin Oxidative stress index Oxidative phosphorylation Sequestosome-1 (SQSTM1) E3 ubiquitin-protein ligase parkin Parkinson’s disease Prodromal Alzheimer’s disease Phosphodiesterase 4A Prefrontal cortex Peroxisome proliferator-activated receptor-gamma coactivator 1-alpha Phosphoinositide 3-kinase PTEN-induced putative protein kinase 1 Preferred Reporting Items for Systematic Reviews and Meta-Analyses Presenilin-1 Presymptomatic Alzheimer’s disease Presenilin 1 Presenilin 2 Phosphatase and tensin homolog Receptor for advanced glycation end-products Ribonucleic acid RNA sequencing Reactive oxygen species Reticulon 3 gene Reverse transcription quantitative polymerase chain reaction Senescence-accelerated mouse prone 8 Short interfering RNA Sirtuin 1 SIX homeobox 4 Small nucleolar RNA host gene 1 Suppressor of cytokine signalling 6 Superoxide dismutase Signal transducer and activator of transcription 3 Traumatic brain injury Tricarboxylic acid cycle Transcription factor EB Translocase of the inner membrane Translocase of the inner membrane 23 Translocase of the inner mitochondrial membrane 13 Tumor necrosis factor alpha Translocase of the outer membrane Translocase of the outer mitochondrial membrane (complex) Translocase of the outer mitochondrial membrane 20 Transient receptor potential mucolipin-1 Ubiquitin-conjugating enzyme E2A Untranslated region Voltage-dependent anion channel 1 Wild-type Wilms tumor 1 gene Wilms tumor 1 antisense RNA Zinc finger protein 217 Malate dehydrogenase 2 NADH:ubiquinone oxidoreductase subunit B10 NADH:ubiquinone oxidoreductase core subunit S1 NADH:ubiquinone oxidoreductase core subunit S2 NADH:ubiquinone oxidoreductase core subunit S3 NADH:ubiquinone oxidoreductase core subunit S8 NADH:ubiquinone oxidoreductase core subunit V1 Pyruvate dehydrogenase E1 subunit alpha 1 Succinate dehydrogenase complex iron sulfur subunit B Ubiquinol-cytochrome c reductase Rieske iron-sulfur polypeptide 1 Polyadenylated RNA sequencing Author Contributions All authors contributed to the study conception and design. Project supervision was performed by Ș.M.M.R. and I.B.N. Screening, data collection and analysis were performed by Ș.M.M.R., V.R., and S.R.P. Disagreements during screening were resolved by I.B.N. Literature synthesis for the tables and discussion sections was carried out by Ș.M.M.R. and V.R. The first draft of the manuscript was written by Ș.M.M.R., V.R., S.R.P., and I.B.N. Figures were designed by Ș.M.M.R. All authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. Funding The authors declare that no funds, grants, or other support were received during the preparation of this manuscript. Data Availability No original data were generated in the present study. All data analyzed in this systematic review were extracted from previously published articles cited in the reference list. Declarations Ethics Approval Not applicable. Consent to Participate Not applicable. Consent for Publication Not applicable. Competing Interests The authors declare no competing interests. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials (DOCX 76.8 KB) Data Availability Statement No original data were generated in the present study. All data analyzed in this systematic review were extracted from previously published articles cited in the reference list.