The SH-SY5Y human neuroblastoma cell line, which is widely used as a model for various ND such as Parkinson’s, Alzheimer’s, and HD, was obtained from the American Type Culture Collection (#ATCC/CRL-2266 ™, United States). Cells were maintained in growth medium Dulbecco’s modified eagle medium (DMEM) supplemented with 10% fetal bovine serum and 1% penicillin–streptomycin at 37°C in a humidified incubator (Thermo Scientific™, Heracell 150i Cell incubator, United States) with 5% CO₂. The SH-SY5Y was transfected with pEGFP-Q23 and pEGFP-Q74 (Cat.No. #40261; #40262; Addgene, Watertown, MA, United States), which have normal and expanded Poly-Q tracts in HTT. Stable cell lines were created by using selection media DMEM containing 100 μg/mL of G418 (Sigma, UK).

For protein extraction, SH-SY5Y cells were rinsed with phosphate-buffered saline (PBS), and 0.25% trypsin/EDTA was used to detach cells from the surface. Protein lysates were prepared from cultured cells, and protein levels were quantified by DS-11 Series Spectrophotometer/Fluorometer (DENOVIX, United States). Approximately 50 μg of each sample was diluted with 2 x Laemmli’s loading buffer (60 mM Tris–HCl pH 6.8, 10% glycerol, 2% SDS, 50 mM dithiothreitol, and 0.01% bromophenol blue). The protein samples were run at 150 V for 80 min. The proteins were blotted onto an immobilon PVDF (Immobilon-P, transfer membrane Millipore, Sigma, Ireland, UK) membrane using a Trans-Blot Turbo Transfer System (Bio-Rad, United States) at 25 V for 60 min. The membranes were blocked with 5% non-fat dry milk (Marvel, Tesco, UK), diluted with TBST (10x Tris-buffered saline containing 0.1% Tween 20) on a tube roller for 1 h and then washed three times for 5 min in TBST. The membrane was incubated overnight at 4°C in a blocking solution with primary antibodies (1:500 mouse monoclonal anti-GFP (BD Bioscience, UK) or 1:2000 anti-mouse β-actin antibody (Sigma, UK) diluted in milk with TBST). Anti-β-actin was used as a loading control. The membrane was washed with TBST four times for 5 min and incubated with a secondary antibody for 1 h [goat anti-mouse HRP or goat anti-rabbit antibody HRP diluted in 5% milk with TBST (Sigma, UK)]. After washing, TBST Chemiluminescent Substrate (Thermo Fisher Scientific Inc., United States) was added to the membrane, and bands were visualized on ChemiDoc Imaging System (Bio-Rad Laboratories, Hercules, CA, United States).

For immunofluorescence analysis, cells were seeded onto 16 mm coverslips at a density of 2 × 10⁵ cells well in 12-well culture plates for 24 h. The following day, the coverslips were washed three times with PBS and then fixed for 10 min with 100% methanol at −20°C. Then, the coverslips were incubated with PBS containing 0.5% bovine serum albumin (BSA) for 1 h at RT to block the non-specific binding of antibodies. Next, the coverslips were incubated in the diluted primary antibody (mouse monoclonal anti-α-tubulin; R&D Systems, MAB1195, dilution 1:500) in 1% BSA in PBS for 1 h at room temperature. The cells were then washed three times in PBS for 5 min each. Then, the cells were incubated with the secondary antibody (anti-mouse Alexa Fluor™ 647; 1:1000; Thermo Fisher, UK) in 1% BSA for 1 h at room temperature in the dark. The samples were washed three times again with PBS for 5 min each in the dark, followed by incubation with 30 nM DAPI (Sigma, UK), and washed a further three times with PBS for 5 min. Next, the coverslips were washed once with dH₂0 prior to mounting in Mowiol-4-88. Images were obtained using the scanning disk immunofluorescent microscope (cellVivo, Olympus inverted microscope IX83. Tokyo, Japan).

The HTG EdgeSeq miRNA Whole Transcriptome Assay (miRNA WTA) (HTG Molecular Diagnostics, Tucson, United States) allows the measurement of the expression of 2083 human miRNA transcripts described in the miRBase database using next-generation sequencing (NGS). The test uses the HTG’s quantitative nuclease protection assay and utilizes the strong sensitivity and wide range of NGS. The HTG EdgeSeq device automates the nuclease protection phase, streamlining the library setup and making NGS platforms convenient for miRNA analysis. The chemistry, which only involves lysis and no extraction, notably cuts down the sample needed, unlike other techniques. The HTG EdgeSeq chemistry can be used with preserved tissue samples, biological fluids, and cultured cell lines. First, HTG lysis buffer (LB) was prewarmed for 20 min at 50°C. The cells were counted, and the pellets were washed with PBS prior to adding HTG LB. After 10 μL with a cell density of 4000 cells/μL, HTG LB was added, and the sample was stored at −80°C. Four aliquots of cells for each of the cell lines, Q23 and Q74, were submitted to HTG Molecular Diagnostics for analysis. Quantification of miRNA expression was performed using counts per million (CPM). To standardize and account for variations in total reads among samples, the CPM values were subjected to a log2 transformation.

The nucleotide sequences of DE miRNAs were downloaded from miRBase.¹ The nucleotide sequences of the human mRNA genes were obtained from GenBank.² The miRNA binding sites (BSs) were predicted by the MirTarget program. The MirTarget program defines the following features of binding: the start of the initiation of miRNA binding to mRNAs; the localization of miRNA BS in 5’UTRs, CDSs, and 3’UTRs; the free energy of binding; and the schemes of nucleotide interactions between miRNAs and mRNAs. The ΔGm equals the free energy of the miRNAs binding with their fully complementary nucleotide sequence. The MirTarget program finds hydrogen bonds between adenine (A) and uracil (U), guanine (G) and cytosine (C), G and U, and A and C. The distance between A and C was 1.04 nm, the distance between G–C and A–U was 1.03 nm, and the distance between G–U was 1.02 nm. For comparison, MirTarget differs from other (predicting miRNA target sites in mammalian mRNAs) programs in terms of finding the BSs of miRNA on the mRNAs in the following: (1) it accounts for the interaction of the miRNAs with mRNA over the entire miRNAs sequence; (2) it considers non-canonical G–U and A–C pairs; and (3) it calculates the free energy of the interaction of the miRNAs with mRNA (Ivashchenko et al., 2016; Belkozhayev A. et al., 2022).

Total RNA was extracted from growing cell samples (>2 × 10⁶ to 5 × 10⁶) using the ReliaPrep RNA Cell Miniprep System (Promega, United States) following the manufacturer’s protocol. The concentration of RNA samples was quantified using a DS-11 series spectrophotometer. The total RNA samples were stored at −80°C until use.

cDNA was prepared using the First Strand cDNA Synthesis Kit (Thermo Fisher, UK) following the manufacturer’s protocol. Approximately, 600 ng of RNA was prepared in a 10 μl sample. The samples were incubated using a PCR machine (3Prime, TECHNE. United States) at 25°C for 5 min, 37°C for 60 min, and 70°C for 5 min, and then the tubes were placed on ice, spun down in microfuge, and stored at −20°C for subsequent analysis. Single-strand DNA concentration was measured using a DS-11 series spectrophotometer and normalized to 10 ng/mL for the following experiments. The expression of different genes involved in NDs was quantified by Real-time PCR (qPCR) reactions performed with SYBR Green SuperMix (SsoAdvanced Universal SYBR^® Green Supermix; Bio-Rad Laboratories Ltd. UK), using 1 μL of cDNA template. A total reaction mixture was amplified in a 96-well PCR plate using the following standard thermocycling program: 95°C denaturation for 10 min, followed by 40 cycles of 95°C for 15 s, and 60°C for 30 s. All primer pairs were designed with Primer-BLAST³ and were obtained from Eurofins Genomics (Ebersberg, Germany) (Supplementary Table S1). The expression levels were measured in triplicate using the 2^−∆∆CT Ct (cycle threshold) method.

Normalization of miRNA expression data was performed using the EdgeSeq REVEAL⁴ from the HTG EdgeSeq System. The statistical results of RT-qPCR were analyzed by Student’s t-test, and a p-value of <0.05 was considered statistically significant. All gene fold change graph analyses were performed using the statistical software GraphPad Prism 9.

The cellular model for HD was established by stably transfecting SH-SY5Y cells with vectors encoding wild-type Htt (Q23) and mutant Htt (Q74). As expected, band intensity of β-Actin expression was equal in all of the cell lines, demonstrating that there was equal loading of the protein samples. The wild-type Q23 migrated at approximately 34 kDa, while mutant Q74 migrated at 50 kDa, as it carries a larger CAG trinucleotide repeat (Figure 1A). The identification of components of Q74 aggregates containing Poly-Q proteins has proven difficult because of the insolubility of such complexes. Therefore, we dissolved it by utilizing acid and alkaline reagents previously demonstrated (Hazeki et al., 2000).

Immunofluorescence staining was performed on the parental SH-SY5Y and the stably transfected Q23-Htt and Q74-Htt cell lines. The results showed that the Q74-Htt protein was prone to aggregation compared to Q23. Furthermore, these results corroborated that Q74-Htt was localized in aggregates close to the nucleus, while Q23-Htt was evenly distributed in the cytoplasm (Figure 1B).

In order to evaluate altered miRNA expression in SH-SY5Y cells stably expressing wild-type and mutant Htt, we performed differential expression of 2083 miRNA probes that passed expression filtering, quantified from small-RNA sequencing using the HTG EdgeSeq system. Profiles of wild-type Htt (Q23) and mutant Htt (Q74) were generated alongside multi-tissue control (MTC) control technical replicates using the HTG EdgeSeq system and all of them passed (Supplementary Table S2). Of the 2083 miRNAs, 354 miRNAs (top 18 miRNAs) were significantly differentially expressed (DE) in Q23 and Q74 cell lines, with 126 upregulated and 228 downregulated miRNAs (Supplementary Table S3).

From the DE 354 miRNAs, the top eight miRNAs (miR-3687, miR-4417, miR-1273 g-5p, miR-3648, miR-612, miR-1273 h-5p, miR-4484, and miR-5088-5p) were confirmed to have significantly increased in expression in HD. Moreover, 10 miRNAs (miR-411-5p, miR-32-5p, miR-301a-3p, miR-135a-5p, miR-876-3p, miR-889-3p, miR-148a-3p, miR-126-5p, miR-495-3p, and let-7f-2-3p) were downregulated in the Q23 and Q74 samples (Table 1).

Expression profiles of the 18 most significant miRNA heatmaps are depicted in Figure 2A. These 18 miRNAs were significantly dysregulated among the normal and mutant Htt groups (Figure 2B). As depicted in Figure 2C, the volcano plot shows sample clustering based on the expression of the 354 significant miRNAs. The wild-type Htt (Q23) and mutant Htt (Q74) groups clustered in distinct sets based on this group of miRNAs.

The characteristics of the interaction of DE 126 upregulated and 228 downregulated miRNAs with the mRNA of 17,494 human genes were studied using the MirTarget program. It was found that the differentially upregulated 126 miRNAs have 52,911 BSs on 8,482 target mRNAs with ΔG/ΔGm values ranging from 80 to 100% in 3’UTRs, 5’UTRs, and CDSs. Moreover, thirteen 412 BSs were also identified for 228 downregulated miRNAs in the 3’UTRs, 5’UTRs, and CDSs mRNA of 6,618 genes with ΔG/ΔGm values equal to 80 to 100%.

The obtained characteristics of the BSs of differentially upregulated and downregulated miRNAs (miR-3687, miR-612, miR-4417, miR-4261, miR-504-3p, miR-126-5p, miR-411-5p, miR-889-3p, and miR-22-5p) were identified in the 3’UTRs, 5’UTRs, and CDSs mRNAs of 18 genes, which play an important role in NDs, including HD. The list of target gene associations with neurodevelopmental disorders is mined from a weekly updated web resource of DISEASES⁵ and PubMed,⁶ (Table 2).

miR-3687 binds to the mRNA of the B3GNT2 gene with −113 kJ/mole free binding energy in the 5′UTR. Furthermore, the BSs of miR-612 in CDS mRNAs of the CSMD2, GNAO1, and ATXN2 genes show the highest free binding energy, equal to −114 and − 119 kJ/mole. miR-4417 and miR-4261 are bound in mRNAs of the DUSP15, EPN3, GLIS1, CREBBP, ATN1, and HAP1 genes, with free binding energy changing from −70 kJ/mole to −98 kJ/mole in the 3’UTRs, 5’UTRs, and CDSs. The mRNAs of some genes have BSs for miR-4261 within their 3′UTRs and CDSs. For example, the 3′UTRs and CDSs of the HAP1 gene have miR-4261 BSs. Furthermore, miR-4261 had two BSs in the mRNA of the ATN and HAP1 genes, whose BSs started at 1758÷2,319 nt and 180÷3,147 nt with a length of 16 nt for 3′UTRs and CDS. The ATXN1 gene contains BSs for miR-504-3p in the CDS with −96 kJ/mole free energy.

Downregulated miR-126-5p, miR-411-5p, miR-889-3p, and miR-22-5p bind to the mRNA of the DNM1L, GEMIN4, EFNA5, NOS3, CNPY1, KIAA1324L, and ATXN10 genes in the UTRs and CDSs. The free energy values of these miRNA interactions with the target genes were equal to −88 ± 5 kJ/mole. According to Table 2, among the genes associated with NDs, the HAP1 and CREBBP genes play an important role in HD (Jiang et al., 2006; Li et al., 2019). In addition, the ATXN2, ATN1, ATXN1, and ATXN10 genes belong to a group of genes that are associated with microsatellite-expansion diseases, a class of neurological and neuromuscular disorders caused by the expansion of short stretches of repetitive DNA, such as spinocerebellar and cerebellar ataxia (Suzuki and Yazawa, 2011; Ashizawa, 2012; Laffita-Mesa et al., 2021; Chen J. M. et al., 2022).

The schemes of interaction of miR-3687, miR-612, miR-4417, miR-4261, miR-504-3p, miR-126-5p, miR-411-5p, miR-889-3p, and miR-22-5p with mRNAs of B3GNT2, CSMD2, GNAO1, ATXN2, DUSP15, EPN3, GLIS1, CREBBP, ATN1, HAP1, ATXN1, DNM1L, GEMIN4, EFNA5, NOS3, CNPY1, KIAA1324L, and ATXN10 genes are shown in Figure 3. The schemes of the interaction of miRNAs with the mRNA of genes indicate the complementarity of nucleotides, including canonical (A–U, G–C) and non-canonical (A–C, G–U) bonds.

In silico analysis of DE miRNAs and their interactions with human mRNA genes, which play an important role in NDs, offers a valuable approach to deciphering intricate gene regulatory networks. This computational method provides insights into potential miRNA-mediated regulatory mechanisms and their implications for various biological processes and diseases. The integration of computational predictions with experimental validation and functional enrichment analysis contributes to our understanding of the complex interplay between miRNAs and their target genes, and these results show that miRNAs can regulate the expression of genes associated with the development of ND, including HD. Therefore, we also determined the expression of target genes by RT-qPCR to obtain a more accurate result for the prediction.

Next, the genes identified by a combination of miRNA profiling of Q23/Q74 cell line models with bioinformatic analysis were selected for validation by RT-qPCR analysis. To study the effect of Q74 on the gene expression of the selected genes, 18S was used as a housekeeping gene.

There was no statistical difference between the GNAO1, ATXN2, DUSP15, EPN3, GLIS1, HAP1, DNM1L, NOS3, CNPY, KIAA1324, and ATXN10 genes. The study of 18 genes identified 3 genes to be upregulated and 4 genes to be downregulated.

The upregulated genes were ATN1, GEMIN4, and EFNA5 (Figure 4A). From those genes upregulated, expression was the highest for GEMIN4 1.5-fold (p = 0.0022), while marginal upregulation was observed for ATN1 (p = 0.0400) and EFNA5 (p = 0.0228).

The RT-qPCR results indicated that the downregulated genes were CSMD2, CREBBP, ATXN1, and B3GNT2 (Figure 4B). From those genes downregulated, expression was the highest for CSMD2 with a 2-fold (p = 0.0022), ATNX1 with a 1-fold (p = 0.0043), and CREBBP with a 0.5-fold (p = 0.0152). These results demonstrate that the miRNAs identified in this study could impact gene expression and thus validate their associated link to HD.