We analyzed post-mortem medulla oblongata and cervical spinal cord tissues of two female C9ORF72-related ALS patients and two sex-matched donors deceased for cerebral hemorrhage, stored at the Fondazione IRCCS Istituto Neurologico “Carlo Besta” neuropathology biobank. The time recorded between decease and post-mortem tissue sampling was within 72 h. Tissues were stored in formalin-fixed paraffin-embedded (FFPE) material.

Genomic DNA was extracted from post-mortem tissues according to a standard phenol-chloroform procedure. The GGGGCC hexanucleotide repeat in C9ORF72 gene was analyzed by a two-step protocol, including a first PCR amplification step using the genotyping primers previously reported (DeJesus-Hernandez et al., 2011). The normal range fragment length analysis was performed on 2% agarose gel. Samples presenting two distinct amplification products in the normal size range are definitively considered negative. Samples resulting in a single amplification product were further analyzed in a second step by the repeat-primed polymerase chain reaction method (AmplideX PCR/CE C9ORF72 Kit (RUO)—Asuragen, Biotechne—Minneapolis—MN, United States) on a 3100XL ABI Prism Genetic Analyzer (Thermo Fisher Scientific—Waltham—MA, United States). The presence of C9ORF72 repeat expansion was assigned when the sample displayed a typical electropherogram profile with decaying stutter amplification peaks. This method is highly sensitive and robust, single-tube, 3-primer C9ORF72 PCR reagents that can flag all expanded samples irrespective of length and provide accurate sizing up to ~145 repeat units using capillary electrophoresis.

We obtained 5 μm-thick consecutive sections of medulla oblongata and cervical spinal cord tissues from both ALS and healthy donor post-mortem samples. Sections were then stained using hematoxylin & eosin (H&E) staining protocol for FFPE tissues, applying the following incubation steps: Bioclear (1 min) twice, 100% EtOH (1 min) twice, 95% EtOH (1 min) twice, wash in H2O; hematoxylin (1 min); two washes in H2O; eosin (40 s); two washes in H2O; 70% EtOH (1 min); 90% EtOH (1 min); 100% EtOH (1 min). To perform H&E histological analysis, tissue images were acquired using NanoZoomer-XR (Hamamatsu Photonics K. K.—Hamamatsu City, Japan) digital slide scanner 3.0.

Consecutive sections were counterstained with thionin to further visualize spinal cord cytoarchitecture. The following incubation steps was applyed: Bioclear (1 min) twice, EtOH 100% (1 min) twice, 95% EtOH (1 min) twice, wash in H2O; thionin (5 min); wash in tap H2O; 70% EtOH (1 min); 95% EtOH (to differentiate between white and gray matter); 100% EtOH (1 min); Bioclear (1 min); mounting medium Eukitt® (O. Kindler GmbH). To perform Nissl staining histological analysis, tissue images were acquired using Aperio CS2 (Leica Microsystems GmbH—Wetzlar, Germany).

The presence of C9ORF72-related RNA foci was assessed using the BaseScope v2 Assay for the detection of sense G4C2-repeat expansions in FFPE tissues (ACD, Bio-Techne, Minneapolis, MN, United States). Sections of medulla oblongata and cervical spinal cord were baked in a HybEZ™ II dry oven (ACD, Bio-Techne) for 1 h at 60 °C. Tissue deparaffinization was performed with two incubations in xylene (5 min each, RT), followed by two incubations in 100% EtOH (2 min each, RT), and drying for 10 min at 60 °C. Staining was conducted according to the BaseScope v2 protocol using a target-specific GGGGCC probe and a negative control probe (bacterial dihydrodipicolinate reductase, DapB). Slides were first treated with RNAscope Hydrogen Peroxide (10 min, RT) to block non-specific binding sites, then immersed in RNAscope Target Retrieval Reagent (15 min, 99 °C) to unmask the target sites. Afterwards, they were transferred to 100% EtOH (3 min) and dried completely overnight. The following staining steps were performed using the HybEZ™ Humidity Control Tray (ACD, Bio-Techne) to prevent desiccation between incubations: Protease IV solution (30 min, 40 °C), G4C2 probe hybridization (120 min, 40 °C), Amp1 (30 min, 40 °C), Amp2 (30 min, 40 °C), Amp3 (15 min, 40 °C), Amp4 (30 min, 40 °C), Amp5 (30 min, 40 °C), Amp6 (15 min, 40 °C), Amp7 (30 min, RT), and Amp8 (15 min, RT). Fast RED solution (1:60 ratio of Fast RED-B to Fast RED-A) was then applied for 10 min at RT for signal detection. Nuclei were counterstained with DAPI (1:1000, 10 min, RT). Slides were dried at 60 °C for 15 min and mounted with FluorSave Reagent. Confocal fluorescence images were acquired with a laser-scanning microscope Eclipse TE 2000-E (Nikon Inc.) and analyzed using EZ-C1 3.70 (Nikon Inc.) and Fiji v1.53 (ImageJ, Fiji). C9ORF72-RNA foci quantification was assessed using the Fiji-ImageJ software, according to manufacturer’s protocol. Two images for each tissue type were quantified.

Leica LMD7 Laser Microdissection Microscope (Leica Microsystems GmbH—Wetzlar, Germany) was used. For each sample, five 5-μm thick consecutive sections were mounted on the LCM membrane slide, stained with H&E and fixed in RNase-free 75–100% EtOH. Slides were left to completely dry in a completely RNase-free environment under a fume hood for 2 h. The hypoglossal nuclei and spinal cord ventral horns were selected using the Leica Laser Microdissection V8.5 software (Leica Microsystems GmbH), cut and pooled in a RNase-free 0.5 mL Eppendorf and resuspended in lysis buffer (Thermo Fisher Scientific—Waltham—MA, United States) to extract RNA for molecular analysis.

Each diffuser isolation cap containing LCM sections was treated with the PureLink FFPE Total RNA Isolation Kit (Thermo Fisher Scientific). To lysate the tissue, melting buffer and proteinase K were added to the tube and incubated at 60 °C for 60–80 min in the Eppendorf ThermoMixer® C (Eppendorf S.r.l.—Hamburg, Germany), with occasional agitation. To isolate tissue total RNA, the following procedures were applied: to bind RNA, binding Buffer (L3) and 100% EtOH were added to the sample; the resulting solution was filtered using the kit Spin Cartidge and centrifuged at 800 x g for 1 min; to wash the Spin Cartridge from unwanted molecule, 3 wash steps with Wash Buffer (W5) followed by 15.000 x g centrifugation for 1 min was performed; to elute the sample, the Spin Cartridge was placed in a clean RNA Recovery Tube and RNase-free water, previously heated at 65 °C, was added to the center of the cartridge. After a 2-min incubation, the solution was eluted through 15.000 x g centrifugation for 1 min; a second elution step was performed using the same water on the same Spin Cartridge to collect the maximum amount of RNA possible. RNA concentration and quality was checked using NanoDrop 2000c Spectophotometer (Thermo Fisher Scientific).

Total RNA extracted from the obtained LCM sections was reverse transcribed using Megaplex RT primers Human Pool A and B and MultiScribe Reverse Transcriptase Kit (Thermo Fisher Scientific) following the recommended thermal protocol: 40 cycles consisting of 16 °C for 2 min, 42 °C for 1 min and 50 °C for 1 s; hold stage at 85 °C for 5 min; final hold stage at 4 °C to preserve the sample until further use. cDNA, corresponding to 20 ng of total RNA, was pre-amplified with Preamplification Reaction Mix that was prepared according to manufacturer’s instructions (Thermo Fischer Scientific) and following the recommended thermal protocol: hold at 95 °C for 10 min; hold at 55 °C for 2 min; hold at 72 °C for 2 min; 12 cycles consisting of 95 °C for 15 s and 60 °C for 4 min; hold at 99.9 °C for 10 min; final hold stage at 4 °C. Next, the resulting cDNA was combined with TaqManTM Fast Advanced Master Mix and dispensed into each port of the TaqMan Human MicroRNA array card A and B v2.0, following the manufacturer’s instructions. The arrays were run on the Viia 7 Real-Time PCR System (Thermo Fisher Scientific) following the recommended thermal protocol: hold at 92 °C for 10 min, 40 cycles consisting of 95 °C for 1 s and 60 °C for 20 s, final hold stage at 4 °C. Human array A and B cards contained primers for 754 miRNAs, including 3 positive controls and 1 negative control, with the controls run in technical duplicates. For both samples, a pool of replicates A and B was used. All raw real-time PCR data were imported into the DataConnect cloud platform, and the automatic Crt threshold was applied using the Design and Analysis software (DA2; Thermo Fisher Scientific, Design & Analysis 2 software). Only miRNAs with high-quality amplification (AmpStatus = AMP, Amplification Score > 1, and Cq Confidence > 0.8) were included in the downstream analysis. miRNAs not meeting these quality thresholds were classified as null and consequently considered as not expressed. To find the most stable miRNAs, in both cards separately we used the user-friendly web-based tool RefFinder¹ developed for evaluating and screening reference genes/miRNAs from extensive experimental datasets. Expression stability ranking identified hsa-miR-34a-000426 as the most suitable miRNA reference for Pool A and hsa-miR-30e-3p-000422 for Pool B, respectively.

Due to the low number of biological replicates, our analysis relied solely on a qualitative presence/absence assessment of miRNAs, without performing group-wise differential expression testing. miRNAs were considered exclusively expressed if detected in only one condition considering the expression in both replicates of either healthy donors or C9ORF72-ALS patients.

A comprehensive in silico pipeline was implemented to identify putative miRNA target genes and characterize their functional relevance. Target prediction was performed by integrating three miRNA–mRNA interaction databases: miRTarBase, TarBase v9 and TargetScan 8.0. For each miRNA, all predicted or experimentally validated interactions were retrieved; only genes predicted by at least two out of the three databases were retained to increase confidence in the interactions.

These genes were further filtered to include only those previously implicated in ALS pathogenesis, based on manual curation of the literature. To annotate candidate miRNAs and their putative target mRNAs, we performed a systematic PubMed search. We included studies reporting experimentally validated miRNA–target interactions, functional assays, or high-confidence bioinformatic predictions supported by experimental evidence, and excluded studies lacking primary data, purely computational predictions, or not relevant to human neurodegenerative disease. For each miRNA–mRNA pair, we recorded the type of supporting evidence, including assay type, experimental model, and outcomes. In cases of conflicting results, both were noted, giving preference to studies with direct experimental validation in human samples or well-established models. In addition, genes involved in molecular pathways associated with C9ORF72-related mechanisms (autophagy, nucleocytoplasmic transport, excitotoxicity, RNA metabolism) were prioritized. Functional enrichment analysis was conducted using ClueGO (Cytoscape v3.10.3). Gene Ontology (GO) terms (Biological Process, Molecular Function and Cellular Component) and pathway databases (KEGG 2024 and Reactome 2024) were queried in order to elucidate terms associated with the selected miRNAs.

Total RNA extracted from the LCM captured ventral horn of cervical spinal cord tissue sections derived from the two C9ORF72-ALS patients and two healthy donors, previously examined for miRNA expression, was reverse transcribed using SuperScript Vilo cDNA Synthesis kit (Invitrogen, Thermo Fisher Scientific) and the following thermal protocol: hold stage at 25 °C for 10 min, hold stage at 42 °C for 60 min, hold stage at 85 °C for 5 min, final hold stage at 4 °C. cDNA (corresponding to 10 ng total RNA) was amplified by quantitative real-time PCR, in duplicate, using TaqMan Fast Advanced Master Mix and TaqMan gene expression assays specific for CASP3, KMT2C, HOXA11, SOCS1, STAT3, TGFBR2, SMAD4, TSG101, DPP9, JAK2, ROCK1, ROCK2, TXNIP, FOXO1, SOX2, CREB1, GRIN2A, SMAD2, BAP1, TET3, BCL6, GSK3B, TERT genes on ViiA7 Real-time PCR system (Thermo Fisher Scientific). 18S has been chosen as housekeeping (Thellin et al., 1999). Specific TaqMan assay IDs are reported in Supplementary Table 1. The following thermal protocol was used: hold stage at 95 °C for 20 s, 40 cycles consisting of 95 °C for 1 s and 60 °C for 20 s, final hold stage at 4 °C. mRNA expression levels were normalized against 18S, and relative expression was calculated using the 2^−∆Ct method. The Heatmapper tool (https://www.heatmapper2.ca/) was used to perform hierarchical clustering of the columns following standard procedures for expression heatmap visualization (Babicki et al., 2016; Kernick et al., 2025). The Expression module was employed, with the Euclidean distance metric to measure pairwise similarity and the Average Linkage method to define cluster relationships. This approach follows standard procedures for visualizing relative expression patterns while maintaining reproducibility.

The G4C2 hexanucleotide repeat analysis showed the presence of G4C2-rich repeat expansion in ALS tissues (Supplementary Figure 1).

We focused on the hypoglossal and ambiguous nuclei of post-mortem medulla oblongata tissue sections, which can be detected close to the fourth ventricle (Figure 1). At both lower and higher magnification, no pathological changes were identified in C9ORF72-ALS samples, consistent with the spinal onset and progression of the disease at the time of the death. To confirm this, we quantified the amount of stained motor neurons in both C9ORF72-ALS patients’ and healthy donors’ medulla oblongata motor nuclei. Healthy donor’s hypoglossal and ambiguous nuclei showed, respectively, an average of 76.19 and 104.04 motor neurons, while the same regions in C9ORF72-ALS patient’s medulla oblongata displayed an average of 83.09 and 99.25 stained motor neurons, respectively.

In the same two patients, the ventral horn regions of the cervical spinal cord showed an altered architecture with disassembled motor neurons, numerous vacuoles both in the gray and white matter at low magnification (Figure 2). At higher magnification, motor neuron bodies showed reduced size and altered morphology consequently leaving empty spaces, pyknotic nuclei indicating chromatin and nucleus shrinkage during apoptosis and necrosis processes, while the whole sections tissues showed sponge-like degenerating features. Coalescing vacuole areas and big cavitations were found in both white and gray matter, indicating the complete degeneration of motor neurons in the ventral spinal cord. In addition, abundant activated glial cells were observed both in gray and white matter of ventral spinal cord sections (Figure 3).

The Nissl staining displayed small motor neurons with different circumferential cytoplasmic size and “skein-like” inclusions in the ventral spinal cord tissues of C9ORF72-related ALS patient (Figure 4). In addition, the ALS peripheral nerve sections showed an altered band irregular morphology, associated to a great amount of glial cells nuclei (Figure 4). Neuronal loss in C9ORF72-ALS patients compared to healthy donors was additionally assessed by stained motor neuron quantification, which resulted in an average of 59.33 motor neurons detected in the ventral horns of cervical spinal cord of ALS-related patients, whereas healthy donors showed the presence on average of 81.17 motor neurons in the same regions.

To further investigate the pathological features linked to C9ORF72-ALS pathology, we performed BaseScope Assay RNA scope assay in human post-mortem sections of medulla oblongata and ventral horn of cervical spinal cord of C9ORF72-ALS patients. Motor neurons were stained with a G4C2 red fluorescent probe to detect the presence of RNA foci. As shown in Figure 5, C9ORF72-ALS medulla oblongata and spinal motor neurons displayed aggregates of G4C2 repeats in the nuclei confirming the toxic gain of function of the expiation in the C9ORF72 gene. C9ORF72-RNA foci quantification in motor neurons of C9ORF72-ALS patients revealed an average of 1.9 foci per cell in the ventral horns of cervical spinal cord samples, while motor nuclei in the medulla oblongata displayed 1.5 foci per cell. Although both regions displayed similar RNA foci accumulation, overt motor neuron degeneration was mainly observed in the spinal cord, suggesting a region-specific susceptibility likely due to higher vulnerability of spinal motor neurons to repeat RNA toxicity and impaired RNA metabolism, whereas medullary neurons may activate compensatory mechanisms that delay cell death.

To identify miRNAs potentially implicated in motor neurodegeneration, we performed a miRNA expression profiling of 754 human miRNAs in LCM captured ventral horn of cervical spinal cord tissue sections (Supplementary Figure 2) derived from the two C9ORF72-ALS patients and two healthy donors. Our molecular and bioinformatics analyses revealed a specific group of miRNAs, including miR-10b, miR-30d-5p, miR-93-3p, miR-106b-5p, miR-127-3p, miR-135a-5p and miR-590-5p, exclusively expressed in the ventral horn of cervical spinal cord regions of healthy donors but not in C9ORF72 ALS patients (Table 1). Through an integrated in silico analysis of predicted target genes and a thorough review of the literature (Brett et al., 2011; Gupta et al., 2012; Li et al., 2021; Liu et al., 2016; Martínez-Fábregas et al., 2018; Pang and Hu, 2023; Petrocca et al., 2008; Shu et al., 2022; van Battum et al., 2018; Wang et al., 2020; Xia et al., 2019; Yang et al., 2013). We selected miR-30d-5p, miR-106b-5p and miR-135-5p for downstream analyses, due to their known cellular functions related to autophagy, neuroprotection and axonal regrowth (Table 2; Supplementary Tables 2, 3).

Conversely, we found that miR-200b-3p, miR-346, and miR-1225 were exclusively expressed in the ventral spinal cord horns of C9ORF72-ALS patients (Table 3). Subsequently, only two miRNAs were selected according to predicted target genes and evidence reported in the literature (Fletcher et al., 2022; Fu et al., 2019; Guo et al., 2020; Jiang et al., 2023; Kmetzsch et al., 2021; Trümbach and Prakash, 2015; Wei et al., 2015), since the increased expression of miR-200b-3p and miR-346 could be associated with the modulation of autophagy machinery, ER stress, cell survival, excitotoxicity, DNA modifications and vesicle trafficking (Table 4; Supplementary Tables 2, 3).

Based on miRNA-target prediction databases, Gene Ontology enrichment analysis and previously published studies (Figure 6A; Table 5; Supplementary Table 2), the following target mRNAs and their biological pathways were identified for miRNAs expressed in controls but not C9ORF72-ALS spinal cord: (i) Caspase 3 (CASP3) implicated in apoptosis and cell death pathways (Dafinca et al., 2016); Home box A11 (HOXA11), and Suppressor of cytokine signaling 1 (SOCS1) implicated in inflammation and immunity (Li et al., 2022; Spiller et al., 2018); Lysine Methyltransferase 2C (KMT2C) implicated in DNA modifications (Park et al., 2022), for miR-30d-5p; (ii) Transforming growth factor beta receptor 2 (TGFB-R2), SMAD family member 4 (SMAD4) implicated in apoptosis and cell death pathways (Petrocca et al., 2008; Meroni et al., 2019; Jiang et al., 2005); Tumor susceptibility 101 (TSG101) implicated in autophagy and vesicle trafficking (Kaul et al., 2020; Walker et al., 2016), Signal transducer and activator of transcription 3 (STAT3) implicated in inflammation and immunity (Pang and Hu, 2023; Toral-Rios et al., 2020); Dipeptidyl peptidase 9 (DPP9) implicated in protein metabolism (Finger et al., 2020), for miR-106b-5p; (iii) Forkhead Box O1 (FOXO1) and Thioredoxin interacting protein (TXNIP) implicated in apoptosis and cell death pathways (Tang et al., 2014; Andrés-Benito et al., 2017), Janus kinase 2 (JAK2), Rho associated coiled-coil containing protein kinase (ROCK1 and ROCK2) implicated in inflammation and immunity (Satriotomo et al., 2006; Wang et al., 2020; Tönges et al., 2014), for miR-135a-5p.

Similarly, integrating target-prediction resources, functional enrichment results and evidence from the literature (Figure 6B; Table 5; Supplementary Table 2), the following target mRNAs and their biological pathways were identified for miRNAs exclusively expressed in C9ORF72-ALS spinal cord tissue: (i) CAMP responsive element binding protein 1 (CREB1), SRY-Box Transcription Factor 2 (SOX2) implicated in cell growth and survival (Peng et al., 2013; Pandey et al., 2015); Glutamate ionotropic receptor NMDA type subunit 2A (GRIN2A) implicated in excitotoxicity (Gunasekaran and Omkumar, 2022); BRCA1 associated protein 1 (BAP1), SMAD2 implicated in autophagy and vesicle trafficking (Kim et al., 2022; Zhou et al., 2020); Tet methylcytosine dioxygenase 3 (TET3) implicated in DNA modifications (Yang et al., 2020), for miR-200b-3p; (ii) BCL6 Transcription Repressor (BCL6), Telomerase reverse transcriptase (TERT) implicated in cell growth and survival (Jiang et al., 2023; Song et al., 2015); Glycogen synthase kinase 3 beta (GSK3β) implicated in autophagy and vesicle trafficking (Guo et al., 2020) for miR-346.

Gene expression levels were quantified in LCM ventral horn spinal cord sections of the same healthy donors and C9ORF72-related ALS patients. Molecular analysis by RT-PCR revealed a trend toward lower expression of SOCS1 and DPP9 genes in healthy tissues. Although preliminary, these results suggest that the inflammatory signaling pathway leading to SOCS1 activation, as well as the protein metabolism regulated by DPP9, is not engaged in control tissues, unlike in C9ORF72-related ALS, where both pathways appear to be upregulated (Figure 6C). Moreover, SOX2 expression was elevated in C9ORF72-related ALS tissues compared to controls, suggesting a pro-survival response aimed at counteracting motor neuron degeneration (Figure 6C). Hierarchical clustering was performed on the samples, as visualized by the column dendrograms of the heatmap and identified two clusters of miRNAs and their predicted targets: one characterizing healthy donor tissue; the other characterizing C9ORF72-related tissues. This separation reflects the expected divergence between C9ORF72-ALS and healthy donor tissues, given that only those genes that were actually expressed were included and that the selected miRNAs and their predicted targets display condition-specific expression patterns. Therefore, the heatmap highlights the differential expression signature distinguishing C9ORF72-related tissues from controls, rather than revealing unexpected clustering among individual miRNAs or target genes. These findings suggest that mutations in ALS-associated genes may disrupt miRNA biogenesis and function, potentially contributing to disease pathogenesis.