MicroRNAs (miRNAs) are small (≈18–22 nt) noncoding single-stranded RNA molecules. They predominantly act by binding to the 3′ untranslated regions (UTRs) of complementary mRNA targets, leading to reduced target expression. Most are ubiquitously expressed in mammals, while others display tissue-specific enrichment, suggesting distinct functions in these tissues (Landgraf et al., 2007; Ludwig et al., 2016). After their transcription, miRNAs are subjected to two cleavage steps. The first takes place in the nucleus, where the complex consisting of the DGCR8 (Di George syndrome critical region gene 8) and the Drosha ribonuclease processes the primary miRNA (pri-miRNA) to the precursor miRNA (pre-miRNA). Following transport to the cytoplasm by Exportin 5′ and the Ran-GTP factor, RNA III enzyme Dicer cuts off the terminal loop to generate the mature miRNA duplex. Ultimately, one of the strands (3p and 5p miRNAs isoforms) is loaded onto the Argonaute protein of the RNA-induced silencing complex to guide it to its target mRNAs (Cammaerts et al., 2015b; Sadlon et al., 2019; Zeng et al., 2005).

The involvement of miRNAs in neurodegenerative processes has become more evident following increased research focus on the noncoding part of the genome. Expression profiling in the human brain has shown dysregulated miRNAs in neurodegenerative phenotypes, like Alzheimer’s disease (AD) and Parkinson’s disease (PD) (Schulz et al., 2019; Takousis et al., 2019). Similar studies in serum and plasma also showcase the use of miRNAs as diagnostic biomarkers for neurodegenerative brain diseases (NBDs) (Sheinerman et al., 2017), including frontotemporal dementia (FTD) (Grasso et al., 2019) and amyotrophic lateral sclerosis (ALS) (Freischmidt et al., 2015).

Genetic variation can impact miRNA function at different levels. For instance, mutations in 3′ UTRs can create or distort existing miRNA binding sites, leading to differential mRNA expression of the target gene. Such cases have been described for NBD-associated genes, such as α-synuclein in PD (Junn et al., 2009; Su et al., 2018) and progranulin (GRN) in FTD (Rademakers et al., 2008). Alternatively, genetic variants within miRNA genes can modulate their functions in different ways, for example, by affecting the processing during maturation or by altering the “seed” sequence with which the miRNA binds to its complementary mRNA target (Cammaerts et al., 2015a). Accordingly, meta-analyses of GWAS performed on AD and PD patients identified miRNA variants associated with disease pathogenesis (Ghanbari et al., 2016a; Ghanbari et al., 2016b).

In the present study, we are investigating the implication of miRNA variants in FTD. Based on a list of brain-expressed miRNAs (Cammaerts et al., 2015b), we are looking for miRNA variants in FTD patients with available whole exome sequencing (WES) data. We believe that our approach, focusing exclusively on variants in noncoding molecules like miRNAs, could improve our understanding of the genetic etiology of FTD, as such variation is regularly overlooked in most GWAS.

By now, miRNAs are well known to play a critical role in brain development and functions of diverse neuronal populations (Budde et al., 2010; Tan et al., 2013). miRNA dysregulation in the brain results in impaired molecular pathways that are shared by NBDs, such as AD and PD (reviewed in (Sadlon et al., 2019)). Genetic variants can lead to impairment of miRNA functions via different mechanisms, but there is still limited evidence associating miRNA variants with brain neurodegeneration.

For this study, we investigated the possible involvement of miRNA variants in the pathogenicity of FTD and AD. We identified 4 rare seed variants in 4 miRNAs (MIR122, MIR656, MIR423, MIR885) in our FTD cohort. Although rare variant association analysis was not feasible due to the small sample size, the presence of such rare variants in our patient cohort could indicate an involvement of these miRNAs in neurodegenerative processes. Indeed, bioinformatic analyses suggest functional implications of these miRNAs/variants related to FTD. miR-885-3p is predicted to bind the 3′ UTR of GRN by both miRmap and TargetScan and this is also validated by HITS-CLIP performed in the human brain cortex (Boudreau et al., 2014). This interaction warrants functional investigation in a FTD context to elucidate whether progranulin (PGRN) protein levels could be downregulated by miR-885-3p, which would align with the described haploinsufficiency mechanism linking GRN loss-of-function mutations with FTD. We also reported seed variants creating new miRNA binding sites at the 3′ UTRs of C9orf72 (MIR423, MIR656, MIR856) and FUS (MIR423), with a significant probability score (miRmap, P.over exact <0.05). FUS plays a critical role in RNA processing and DNA repair, while C9ORF72 haploinsufficiency has been linked with ALS/FTD via impaired membrane trafficking and autophagic function (Pang and Hu, 2021; Shi et al., 2018). These observations need to be interpreted cautiously as predictions need to be experimentally validated. Further, expression patterns for each miRNA-mRNA interaction must be investigated in disease-related tissue to confirm biological relevance. Nonetheless, our genetic and bioinformatic approach suggests miRNA-mediated molecular processes that could contribute to FTD pathogenicity (Figure 2).

MIR885 is the most intriguing case. Both the 3p and the 5p isoform are enriched in brain (Human miRNA tissue atlas (https://ccb-web.cs.uni-saarland.de/tissueatlas2, (Keller et al., 2021; Ludwig et al., 2016)), suggesting a distinct function for MIR885 in brain. The 5p isoform is the leading strand, which is quite common for mature miRNAs derived from the same precursor (McCall et al., 2017). miR-885-5p is predicted to bind the SORT1 gene, encoding sortilin 1, with very strong affinity metrics (TargetScan v.8, miRDB and miRmap v.1.1). Sortilin 1 is a known regulator of PGRN levels, as it mediates PGRN lysosomal trafficking, leading to reduced extracellular PGRN levels (Hu et al., 2010). Common variants adjacent to the SORT1 gene have been associated with reduced PGRN levels in plasma in control subjects and FTD patients (Carrasquillo et al., 2010), while rare SORT1 variants have been associated with increased risk for FTD in different populations (Philtjens et al., 2018). Also, a recent GWAS identified a SORT1 missense mutation as the sentinel SNP in a novel risk locus for AD (Bellenguez et al., 2022). Involvement of miR-885-5p in disease pathogenesis could be investigated via direct interaction with SORT1, where altered sortilin 1 expression might dysregulate unique or shared pathways between FTD and AD leading to neurodegeneration. Another study described the miR-885-5p interaction with matrix metalloprotease 9 (MMP9) in glioma cells (Yan et al., 2011). MMP9 is a potential AD biomarker for disease progression, with higher serum levels of MMP9 correlated with faster cognitive decline in patients with mild cognitive impairment attributed to AD (Abe et al., 2020). This is aligned with another study, which identified higher MMP9 levels in the brain of AD patients with a later Braak stage (Hernandes-Alejandro et al., 2020). Taken together, a potential involvement of miR-885-5p in AD could be attributed to its interaction with MMP9.

Both SORT1 and MMP9 are reported to assist in the degradation and clearance of toxic Αβ aggregates (Radosinska and Radosinska, 2025; Ruan et al., 2018). However, their neuroprotective function can be attenuated by the presence of APOE Ε4 (Asaro et al., 2020; Shackleton et al., 2019). Therefore, any mechanistic impact derived from their interaction with miR-885-5p could be masked by the inarguably higher effect in the presence of an APOE E4 isoform. A recent study showcased that circ_0003611 (circLPAR1), a circular RNA upregulated in AD (Li et al., 2020), acts as a sponge for miR-885-5p (Pan et al., 2022). Downregulation of circPARL1 significantly reduced Αβ-induced apoptosis, inflammation and oxidative stress in a human neuroblastoma cell line (SK-N-SH cells) (Pan et al., 2022). A similar effect was observed in Aβ-treated SK-N-SH cells after exogenous miR-885-5p administration. This mitigation of Αβ-induced pathology was attributed to miR-885-5p binding and reducing transcript levels of KREMEN1, a gene previously implicated in AD via noncoding RNA interactions (Wang et al., 2019). These results further support a protective role for miR-885-5p in AD, independent of APOE status, via alleviating Aβ-induced inflammation, which appears to be exacerbated when miR-885-5p levels are depleted (Figure 2). Investigation of miR-885-5p function in a knockdown model using neuronal-like cell types would provide valuable insights of its implication in neurodegenerative processes.

We identified 2 ultra-rare MIR885 variants (rs941703617, rs897551430) in 3 FTD patients (Table 3), both absent from control subjects and the HEX database (https://www.alzforum.org/exomes/hex). Extending our screening in our AD cohort, we identified rs897551430 in 4 LOAD patients, as well as two other rare variants in 2 LOAD patients (Table 4). Variants rs897551430 and rs993255773 are predicted to strongly decrease pre-miRNA-885 levels (miRNASNP v3, http://bioinfo.life.hust.edu.cn/miRNASNP/). This is also supported by the prominent changes caused to the secondary miRNA structure in the presence of each mutant allele (Figure 1). Both variants are located at the 5′ end of the 5p isoform, which corresponds to the lower stem and could thus affect processing of the pri-miRNA by the DGCR8-Drosha complex (Bofill-De Ros et al., 2019; Ma et al., 2013). Rare variant association analysis initially showed a significant enrichment for MIR885 variants in our AD cohort (SKAT-O p-value = 0.026). However, upon the integration of sex and APOE genotypes as covariates, the genetic association was lost. Genetic studies on rare variants are inherently challenging. A recent study performed region-based rare variant association analysis using whole-genome sequencing data of >6,000 amyotrophic lateral sclerosis (ALS) patients and >70,000 healthy controls to explore the noncoding genome (Eitan et al., 2022). Despite the large cohort size, no disease association was found for any of the identified miRNA variants. This was, partially, attributed to the small genomic size of miRNAs (∼120 nucleotides), which significantly hinders gene-burden analysis for rare variants, as it requires a dramatic increase in sample size to prevent missing risk or potentially causal genetic variation. The importance of using large population datasets was also underscored in the most recent phenome-wide association study which used genetic and clinical data from over 400,000 participants of the UK BioBank and showcased the pleiotropic effect of common miRNA variants (Mustafa et al., 2023). That said, we believe that the observed genetic enrichment of ultra-rare variants before APOE status and sex correction in our relatively small cohort size endorses genetic screening in extended cohorts or integration of publicly available AD and FTD genetic data to substantially increase the power to test for a possible association with disease risk.

In conclusion, we identified rare genetic variants of brain-expressed miRNAs in patients with NBDs. Functional investigation for all variants is warranted, as they are predicted to target disease-associated genes. Elucidation of miRNA function in neurodegeneration will pave the way for novel therapeutic approaches for brain disorders. Furthermore, given that miRNAs constitute robust biomarkers, establishing miRNA expression profiles in a disease-related context could offer opportunities for earlier and improved differential diagnosis. We believe that the genetic findings presented for MIR885 suggest an implication for this miRNA gene in AD pathology and underscore the disease relevance of genetic variation in noncoding genomic regions.