tRF^Ala-AGC-3-M8 attenuates neuroinflammation and neuronal damage in Alzheimer’s disease via the EphA7-ERK_1/2-p70S6K signaling pathway

Abstract

Background

Alzheimer’s disease (AD) is a chronic, progressive neurodegenerative disorder clinically characterized by memory decline, cognitive dysfunction, language impairment, deterioration of visuospatial skills, and personality changes. Pathologically, AD is marked by the deposition of β-amyloid (Aβ) plaques in the brain, the formation of neurofibrillary tangles, and progressive neuronal loss. Recent research has highlighted transfer RNA (tRNA)-derived small RNAs (tsRNAs) as crucial regulators in various biological processes; however, their roles in the pathophysiology of AD remain largely unexplored. The erythropoietin-producing hepatocellular (Eph) receptor family has recently drawn attention in the study of neurodegenerative diseases due to their role in regulating critical processes, including cell migration, neural development, angiogenesis, and tumor formation. This study aimed to investigate specific tsRNAs associated with AD by performing RNA sequencing on the cortex of APP/PS1 transgenic mice and to explore the relationship between tsRNAs and their target genes within the Eph receptor family, thereby elucidating insights into the specific regulatory functions of these molecules.

Methods

Eight-month-old male C57BL/6 and APP/PS1 transgenic mice were used in the study. BV-2 and HT22 cells were cultured and treated with Aβ_25-35 at concentrations ranging from 0 µM to 40 µM. RNA was extracted from cortical tissues, and tRNA-derived fragments were analyzed after pre-treatment to remove RNA modifications. Differential expression of tRFs and tiRNAs was identified through sequencing, followed by bioinformatics analysis of target genes using TargetScan and miRanda. Transfection of BV-2 and HT22 cells with EphA7-siRNA and tRF^Ala-AGC-3-M8-mimic was conducted, and their interaction was validated using dual-luciferase reporter assays. Protein expression levels were assessed by western blotting and immunofluorescence. Statistical analyses were performed using R and GraphPad Prism, with significance set at p < 0.05.

Results

We identified for the first time that EphA7 expression is upregulated in aggregated microglia and neuronal cells in the dentate gyrus region of the hippocampus, with increased phosphorylation of ERK_1/2 and p70S6K in AD. This upregulation occurred following the downregulation of tRF^Ala-AGC-3-M8 due to Aβ stimulation and was confirmed via in vitro experiments. By inhibiting EphA7 expression and increasing tRF^Ala-AGC-3-M8 expression, we suppressed the ERK_1/2-p70S6K signaling pathway in BV-2 and HT22 cells. This intervention alleviated neuronal damage and tau hyperphosphorylation in HT22 cells and reduced the M1-type polarization state of BV-2 cells induced by Aβ_25-35 (see Graphical Abstract).

Conclusions

This study clarifies the specific role of tRF^Ala-AGC-3-M8 in AD pathology and offers a promising target for therapeutic interventions.

Graphical Abstract

Background

Alzheimer’s disease (AD) is a progressive, neurodegenerative disorder and the most common type of dementia among older adults. Its pathological features include the formation of β-amyloid (Aβ) plaques and neurofibrillary tangles (NFTs). Aβ aggregation activates microglia, triggering an inflammatory response. Early in the disease, microglia can recognize and clear abnormal Aβ aggregates. However, as AD progresses, microglia may polarize to the M1-type, releasing excessive pro-inflammatory factors (such as tumor necrosis factor-α and interleukin [IL]-1β) and neurotoxic substances [1,2,3]. These factors negatively affect nearby neurons, exacerbating neuroinflammation. Tau, a microtubule-associated protein, normally supports microtubule structure in neurons and can become hyperphosphorylated. This hyperphosphorylation impairs its ability to bind to microtubules, leading to microtubule disassembly. Consequently, hyperphosphorylated tau protein aggregates into insoluble NFTs, disrupting neuronal function and inducing apoptosis, resulting in neuronal death [4].

The amyloid precursor protein/ human presenilin 1 double-transgenic mouse model (APP/PS1 transgenic mice) is a valuable tool for studying AD. This model concurrently expresses the human APP and PS1 gene mutations [5]. The APP mutation enhances the β-secretase-mediated cleavage of APP, increasing Aβ precursor production, while the PS1 mutation alters γ-secretase activity, promoting Aβ production. This dual effect significantly increases Aβ accumulation in the brain, forming senile plaques and leading to neuroinflammation and neuronal damage, thereby recapitulating key pathological features of AD [6].

Transfer RNA (tRNA)-derived small RNAs (tsRNAs) are a newly identified class of small RNA molecules generated under various physiological and pathological conditions and are mediated by specific nucleases, such as Dicer and ANG [7]. Depending on their origin and mechanism of generation, tsRNAs can be categorized into tRNA-derived fragments (tRFs) and tRNA-derived stress-induced RNA (tiRNAs). Specifically, tRFs are further classified into tRF-1, tRF-2, tRF-3, and tRF-5 based on their cleavage patterns, while tiRNAs are classified into 3’-tiRNA and 5’-tiRNA. Each tRF or tiRNA subtype has different biological effects. tsRNAs play important roles in regulating gene expression and cellular functions through interactions with specific target genes, impacting post-transcriptional regulation and signal transduction pathways [8]. Recently, tsRNAs have been associated with various diseases, including cancer, neurological disorders, viral infections, and metabolic diseases [9]. However, the exact mechanisms by which tsRNAs influence AD remain largely unexplored.

The erythropoietin-producing hepatocellular (Eph) receptor family is the largest subgroup of receptor tyrosine kinases and plays essential role in various biological processes, including cell migration, neurodevelopment, angiogenesis, and tumor formation. Eph receptors are classified into two major types: EphA and EphB. Their ligands, known as ephrins, are membrane-bound proteins classified into ephrin-A and ephrin-B. The interaction between Eph receptors and ephrin ligands mediates their functions through bidirectional signaling [10]. Recently, the Eph receptor family has gained attention in the study of neurodegenerative diseases [11, 12]. Research has demonstrated that EphA4 and EphB2 are significant in AD. Activation of the EphA4 signaling pathway is linked to dysfunction of neuronal synapses, while downregulation of EphB2 is related to loss of synaptic plasticity [13,14,15]. However, few studies have investigated how tsRNAs regulate Eph signaling and influence AD pathology, prompting further exploration of tsRNAs as a regulatory factor in AD.

In our experiments, we identified a specific tsRNA, tRF^Ala-AGC-3-M8, derived from 5’-tRFs in the APP/PS1 transgenic mice expression profile. Following Aβ stimulation, the expression of tRF^Ala-AGC-3-M8 was significantly downregulated, leading to the upregulation of EphA7 within the Eph receptor family. EphA7 enhances hyperphosphorylation of the ERK_1/2 -p70S6K signaling pathway, exacerbating neuroinflammation and neuronal damage.

Methods

Animals and design

All animal precedures were approved by the Animal Experiment Ethics Committee of Southern Medical University. Eight-month-old male C57BL/6 mice and APP/PS1 transgenic mice(25 ± 3 g) were purchased from the Guangdong Experimental Animal Center. Animals were housed in groups of six per cage under a 12-hour light/dark cycle (lights on at 8:00 AM) with ad libitum access to food and water.

Cell culture and design

BV-2 (CL-0493) and HT22 (CL-0595) cells were provided by Wuhan Procell Life Science & Technology Co., Ltd. The cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM), containing 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin. When the cell growth density reached approximately 50%, BV-2 and HT22 cells were induced with Aβ_25-35 (HY-P0128, MedChemExpress, Shanghai, China) at concentrations of 0, 5, 10, 20, 30, and 40 µM. BV-2 cells were exposed for 24, 36, and 48 h, while HT22 cells were treated for 24 h.

RNA extraction, collection, and sample quality control

Total RNA was extracted from samples using RNAiso Plus (Takara Bio Cat. #9109) following the manufacturer’s protocol. The purity and concentration of total RNA samples were determined using a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific), with samples demonstrating an A₂₆₀/A₂₈₀ ratio between 1.8-2.0 being retained for downstream analysis.

tsRNAs pretreatment and library construction

Before preparing libraries from total RNA samples, the following treatments were performed: 3’-deacylation (charged) to 3’-OH for 3’adaptor ligation, 3’-cP (2’,3’-cyclic phosphate) removal to 3’-OH for 3’ adaptor ligation, 5’-OH phosphorylation to 5’-P for 5’ adaptor ligation, and demethylation of m1A and m3C for efficient reverse transcription. RNA biotype selection for sequencing libraries was performed using an automated gel cutter. Library quality and absolute quantification were assessed using an Agilent BioAnalyzer 2100.

Sequencing data analysis

Raw sequencing data generated by Illumina NextSeq 500 underwent Illumina purity filtering and was used for subsequent analysis. Read sequences were trimmed (5’ and 3’ adaptor bases trimmed), allowing a single mismatch alignment with mature tRNA sequences, followed by alignment of unmatched read sequences with precursor tRNA sequences using Knot software. Alignment statistics (alignment rate, read length, and fragment sequence bias) guided the decision on the suitability of subsequent analyses. If applicable, expression profiling and differential expression analysis of tRF and tiRNA were performed. All visualizations were generated using R or Perl environments.

Prediction and bioinformatics analysis of tsRNAs-Target genes

Differentially expressed tsRNAs that met the criteria of fold change > 1.5 and p-value < 0.05 were selected from the sequencing dataset. Subsequently, quantitative reverse transcription polymerase chain reaction (qRT-PCR) was used to further screen and validate differentially expressed tsRNAs consistent with sequencing results, designating them as candidate tsRNAs( Ca-tRFs). The interaction seed sequences between tsRNAs and their target genes were predicted using TargetScan and miRanda. Significant relationships between Ca-tRFs and their respective target genes were visualized using Cytoscape (version 3.9.1). Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway annotations were used for functional enrichment analysis of Ca-tRFs target genes to correlate tsRNAs with different biological functions. These analyses were performed using R (version 4.3.1), with a significance threshold set at p-value < 0.05.

Quantitative reverse transcription polymerase chain reaction (qRT-PCR)

Total RNA was extracted using RNAiso Plus reagent (Takara Bio Cat. #9109) following the manufacturer’s protocal. RNA concentration and purity were evaluated using a NanoDrop ND-1000 spectrophotometer. Subsequently, total RNA was reverse transcribed into cDNA using the PerfectStart^® Uni RT&qPCR Kit (Transgen AUQ-01). PCR amplification was performed using PerfectStart^® Green qPCR SuperMix (Transgen AQ602-01). qRT-PCR analysis was performed using an Applied Biosystems 7500 Real-Time PCR System (Thermo Fisher Scientific, USA) following the recommended PCR premix conditions. To ensure the reliability and reproducibility of the experiment, each reaction was independently repeated three times. RNA expression was normalized to those of β-actin and tsRNAs expression levels were normalized to those of U6. The 2^ΔΔCt method was used to analyze the relative RNA expression. Primer sequences are showed in Table 1.

Table 1 Primer sequence

EphA7-siRNA and tRF^{Ala−AGC−3−M8} -mimic transfection

EphA7-targeting siRNA (EphA7-siRNA) and non-specific control siRNA (NC-siRNA), as well as chemically synthesized double-stranded tRF^Ala-AGC-3-M8 mimic (tRF^Ala-AGC-3-M8 -mimic) and non-specific control mimics (NC-mimic), were purchased from RIBOBIO Biotechnology Co., Ltd., China. EphA7-siRNA or NC-siRNA, tRF^Ala-AGC-3-M8-mimic or NC-mimic were transfected using Lipofectamine 3000 (Invitrogen, USA) into BV-2 or HT22 cells according to the manufacturer’s protocol, and the transfection process lasted for 24–36 h.

Dual-luciferase assay

After designing wild-type (WT) and mutant(MUT) EphA7 sequences, the sequences were cloned downstream of the Renilla luciferase gene (hRluc) in the pmiR-RB-Report™ plasmid vector. The WT plasmid vector with tRF^Ala-AGC-3-M8-mimic, WT plasmid vector with NC-mimic, MUT plasmid vector with tRF^Ala-AGC-3-M8-mimic, or MUT plasmid vector with NC-mimic was co-transfected into BV-2 cells. After 36 h, the cells were treated with cell lysis buffer. A suitable amount of lysis buffer (typically 100 − 200 µL) was added to each well, and the cells were gently pipetted to ensure complete lysis. Firefly luciferase substrate and Renilla luciferase substrate were prepared according to the manufacturer’s instructions. 100 µL of cell lysate was added to the wells of the luciferase detection plate. Firefly luciferase substrate was added and mixed. The firefly luciferase signal was immediately measured using a luminometer. Renilla luciferase substrate was then added to the same lysate. Reading parameters were set according to the luminometer protocol, with firefly and Renilla luciferase signals read separately.

Western blotting (WB)

Lysates from tissues and cells were prepared using RIPA buffer with protease/phosphatase inhibitors. Proteins (10–30 µg) were separated on 7.5 − 10% SDS-PAGE gels (Thermo Fisher) and transferred to PVDF membranes (Merck Millipore). Membranes were blocked with 5% bovine serum albumin (BSA; Sigma-Aldrich, Cat. #A7906) (1 h, RT) and incubated overnight at 4 °C with primary antibodies (p-tau, EphA7, ERK_1/2, p-ERK_1/2, p70S6K, p-p70S6K, iNOS, β-actin, β-Tubulin; 1:1,000 − 1:10,000, Bioworld). After washing, membranes were incubated with HRP-conjugated secondary antibodies (1:1,000 − 1:10,000, Bioworld; 2 h, RT). Proteins were detected using ECL substrate (Pierce, Cat. #32106) and imaged with a ChemiDoc™ System (BioRad). Each experiment was repeated three times independently. Band intensities were quantified using Fiji ImageJ and normalized to β-actin/β-tubulin.

Immunofluorescence staining (IF)

Tissues were embedded, sectioned, and deparaffinized. Cells were washed, centrifuged, and resuspended in PBS. Samples were fixed with 4% paraformaldehyde (PFA) (20mins, RT), permeabilized with 0.1% Triton X-100 (10–15 min, RT), and washed 3x with PBS. Non-specific binding was blocked with 5% goat serum (NGS; Sigma-Aldrich, Cat.#G9023) (1 h, RT). Samples were incubated overnight at 4°C with primary antibodies (iNOS, NeuN, Iba-1, p-tau, EphA7; 1:100 − 1:500, abcam), washed 3x with PBS, and incubated with fluorescent secondary antibodies (AlexaFluor 555/488/647; 1:1000, abcam; 1 h, RT, protected from light). Nuclei were stained with DAPI (5–10 mins, RT), washed 2x with PBS, and imaged using Leica fluorescence microscope or Leica TCS SP8 confocal microscope.

Statistical analysis

The data were analyzed using R (Version 4.3.1) and GraphPad Prism (Version 9.5.1). For comparisons between two groups, two-tailed Student’s t-test was used for parametric data. For comparisons between multiple groups, one-way analysis of variance was performed, followed by post-hoc tests (e.g., Tukey’s HSD or Bonferroni correction) for parametric data. Statistical significance was set at p < 0.05.

Result

Downregulation of tRF^Ala-AGC-3-M8 expression in APP/PS1 transgenic mice cortex

We comprehensively analyzed the expression profile of tsRNAs in APP/PS1 transgenic mice cortex using tsRNA-Seq technology. Our results showed an increase in the proportion of 5’-tiRNA and tRF-5c with a decrease in the proportion of tRF-5a in APP/PS1 mice (Fig. 1A). We identified tsRNAs with significant differential expression based on fold change (cut-off value of 1.5) and p-value (only when there were replicates, cut-off value of 0.05). This analysis revealed 25 tsRNAs with significant differential expression in the APP/PS1 transgenic mice expression profile, including 13 upregulated and 12 downregulated tsRNAs, among which there were 12 types of 5’tsRNAs (Fig. 1B and C). The verification results for three 5’tsRNAs, including tiRNA^Gly-CCC-2 (upregulated, tiRNA-5, p-value = 0.0202), tRF^Ala-AGC-3-M8 (downregulated, tRF-5a, p-value = 0.0491), and tRF^Gly-TCC-1 (downregulated, tRF-5c, p-value = 0.0346), were consistent with the sequencing results (Fig. 1D).

Fig. 1

Comprehensive tsRNA expression profile in APP/PS1 transgenic mice analyzed using tsRNA-Seq. (A) Distribution of tsRNA subtypes, showing an increased proportion of 5’-tiRNA and tRF-5c and a decreased proportion of tRF-5a in APP/PS1 transgenic mice cortex compared to controls (Control and APP/PS1: n = 3, biologically independent samples). (B) Volcano plot of tsRNA expression, highlighting 25 tsRNAs with significant differential expression (13 upregulated and 12 downregulated). (C) Specific information of differentially expressed tsRNAs, with 5’ tsRNAs marked in red. (D) qRT-PCR analysis of 5’ tsRNAs. tiRNA^Gly-CCC-2, tRF^Ala-AGC-3-M8, and tRF^Gly-TCC-1 showed consistent results with the sequencing data ( n = 6, biologically independent samples). Statistical significance was assessed using Student’s t-test (D) (p-value < 0.05; p values are indicated). All data are presented as means ± SEM. Each data point represents the average of 3 technical replicates

tRF^Ala-AGC-3-M8 directly binds to the 3’ untranslated region (3’ UTR) region of EphA7

To further identify significant tsRNAs and their associated biological processes, we used bioinformatics tools. Leveraging the miRNA-like mechanism of tsRNAs, where their seed sequences (two to seven nucleotides at the 5’ end) recognize target mRNAs, we predicted all mRNAs targeted by tiRNA^Gly-CCC-2, tRF^Ala-AGC-3-M8, and tRF^Gly-TCC-1. We then identified the top 150 target genes for each tsRNA based on higher structure score, lower free energy, and context plus score and presented these findings through a tsRNA-Target genes interaction network (Fig. 2A). Notably, EphA7 was the only member of the Eph family interacting with tiRNA^Gly-CCC-2 (structure score: 152, energy score: 12.98, context plus score: -1.74), tRF^Ala-AGC-3-M8 (structure score: 146, energy score: 16.75, context plus score: -4.27), and tRF^Gly-TCC-1 (structure score: 144, energy score: -2.86, context plus score: -5.29). We further illustrated the binding patterns of these three tsRNAs with the 3’ UTR region of EphA7 based on their binding sequences. The results indicated that tRF^Ala-AGC-3-M8 binds to the 3’ UTR region of EphA7 with an 8mer-1a binding pattern, a highly accurate matching mode (Fig. 2B).

To further verify the binding with the 3’ UTR region of EphA7, we designed WT (5’ccTGAAAAACATCCa3’) and mutant (5’ccACTAAATGTAGGGa3’) 3’ UTR sequences of EphA7 and cloned these sequences into plasmid vectors to create WT and MUT pmiR-RB-Report™ vectors. We transfected these WT or MUT plasmids (Group: EphA7-WT, EphA7-MUT), alongside tRF^Ala-AGC-3-M8-mimic or NC-mimic (Group: AGC-3-M8, NC) into BV-2 cells. After 36 hours, we measured luciferase activity (normalized to firefly luciferase). The results showed that tRF^{Ala-AGC- 3-M8}-mimic significantly decreased firefly luciferase activity in the WT group compared to the NC-mimic group (p-value<0.0001), while no significant difference was observed in the MUT group. This result convincingly confirmed that tRF^Ala-AGC-3-M8 binds to the EphA7 3’ UTR region and inhibits EphA7 translation (Fig. 2C).

Fig. 2

Identification and validation of tsRNA-Target genes interactions and binding patterns. (A) The tsRNA-Target genes interaction network identifies EphA7 as a common target. (B) tRF^Ala-AGC-3-M8 binds to the EphA7 3’ UTR through an 8mer-1a pattern. 2D structure: The specific location of the binding site in the 3’ UTR (mRNA) or throughout a particular sequence. Local AU: AU weight distribution diagram around the seed sites, which influences site accessibility. The evaluation function slightly varies depending on the type of seed. The red bar indicates the position is A, with a higher number being more favorable. Position: The relative location of the site on the UTR, with proximity to the edges being more optimal. (C) Dual-Luciferase Reporter Assay results show that tRF^Ala-AGC-3-M8-mimic significantly decreases firefly luciferase activity (adjusted by renilla luciferase, normalized to NC mimic) in the EphA7-WT group (n = 9, biologically independent samples). Statistical significance was assessed using Student's t-test (p-value < 0.05; P values are indicated). All data are expressed as means ± SEM. Each data point represents the average of 3 technical replicates

Enrichment analysis of tRF^Ala-AGC-3-M8 target genes

To understand the role and relationship of the predicted tRF^Ala-AGC-3-M8 target genes in signaling pathways, we performed KEGG and GO enrichment analyses. KEGG enrichment analysis reveals that the axon guidance pathway had the highest score among the signaling pathways associated with tRF^Ala-AGC-3-M8 target genes (Fig. 3A). We further illustrated the gene codes involved in the axon guidance pathway using a Pathway–Gene network diagram (Fig. 3B), which indicates the involvement of EphA7. GO enrichment analysis demonstrates that the synapse organization pathway had the highest score among the biological processes associated with tRF^Ala-AGC-3-M8 target genes (Fig. 3C). We also constructed Biological Process-Gene waterfall plot (Fig. 3D) to elaborate the biological processes involving EphA7, which highlights synapse organization, regulation of synapse organization and activity, post-synapse organization, and cell junction assembly. These results suggest that the target genes of significantly differentially expressed tsRNAs in the APP/PS1 transgenic mice expression profiles are closely related to axon guidance and synapse organization, with EphA7 being significantly involved in both processes.

Fig. 3

KEGG and GO enrichment analyses of tRFAla−AGC−3−M8 target genes. (A) KEGG analysis identified the axon guidance pathway as the most significant for the predicted target genes. (B) Pathway-Gene network diagram showing that EphA7 (Universal Protein Resource ID: 13041, labeled with a black box) is involved in the axon guidance pathway. (C) GO analysis revealed synapse organization as the key biological process. (D) Biological Process-Gene Waterfall plot illustrating EphA7's role in synapse-related activities (labeled with a black box)

Upregulation of EphA7 was detected in microglial aggregates and neurons localized to the dentate gyrus (DG)

To investigate the characteristic expression changes of EphA7 in APP/PS1 transgenic mice, we analyzed the brain slices. The results showed scattered Iba-1⁺/EphA7⁺ cell clusters in the cortex (Fig. 4A), with significantly increased fluorescence area and intensity compared to the control group (p-value<0.0001). These cells often appeared as irregular circular shapes surrounding cell with weak NeuN⁺ fluorescence signals (Fig. 4C). Additionally, in both the head (Fig. 5A) and tail (Fig. 5E) hippocampus of APP/PS1 transgenic mice, NeuN⁺/EphA7⁺ cells with higher EphA7 fluorescence signals (p-value<0.0001) were distributed throughout the DG of the hippocampus, whereas these cells were not observed in Cornu Ammonis (CA) region.

Fig. 4

EphA7 expression in the cortex of APP/PS1 transgenic mice. (A) Brain slices from the APP/PS1 transgenic mice show Iba-1⁺ cell clusters (marked with white arrows) in the cortex, with higher expression of EphA7. Scale bar: 1000 μm. (B) Mean fluorescence intensity analysis of EphA7 expression in the cortex (n = 6, biologically independent samples). (C) Zoomed-in immunofluorescence image of the cortex. Scale bar: 20 μm. The fluorescence signal was manually traced to enhance visualization in schematic diagram (black: microglia; red: neuron). (D) Immunofluorescence analysis of Iba-1 (orange) and EphA7 (green) localization. Statistical significance was assessed using Student’s t-test (P values are indicated). All data are presented as means ± SEM. Each data point represents the experimental measure

Fig. 5

EphA7 expression in the hippocampus of APP/PS1 transgenic mice. (A, E) Immunofluorescence image show significant EphA7 overexpression in the hippocampus of DG; Scale bar: 500μm. (B, F) Mean fluorescence intensity of EphA7 in the dentate gyrus (n = 6, biologically independent samples). (C, G) Zoomed-in immunofluorescence image of the dentate gyrus; Scale bar: 50 μm. (D, H) Immunofluorescence analysis of NeuN (red) and EphA7 (green) localization. Statistical significance was assessed using Student’s t-test (B and F) (p values are indicated). All data are presented as means ± SEM. Each data point represents the experimental measure

Downregulation of tRF^Ala-AGC-3-M8 and upregulation of EphA7 after Aβ_25-35 induction

The downregulation of tRF^Ala-AGC-3-M8 (Fig. 1D) and upregulation of Epha7 (Fig. 4 and Fig. 5) have been observed in the APP/PS1 transgenic mice. To further explore these changes, we conducted in vitro experiments using BV-2 and HT22 cells.

In the HT22 cells, compared to the control group (Aβ_25-35, 0 µM), all treatment groups (Aβ_25-35, 5 µM, 10 µM, 20 µM, 30 µM, and 40 µM) exhibited irregular shrinkage of neuronal cell bodies and shortened processes, with damage increasing at higher Aβ_25-35 stimulation concentrations. After 24 h, the 30 µM and 40 µM concentration groups showed nearly complete loss of intact cell structures. For tRF^Ala-AGC-3-M8, the 5 µM (p-value<0.0001), 10 µM (p-value<0.0001), 20 µM (p-value<0.0001), 30µM (p-value = 0.0012) groups demonstrated downregulation of tRF^Ala-AGC-3-M8 (Fig. 6A).

In the BV-2 cells, cell morphology changes were assessed at induction of Aβ_25-35 for 24, 36 and 48 h to modify the difference response of BV-2 cell simulated by Aβ_25-35. After 24 h of Aβ_25-35 induction, all treatment groups showed slight aggregation; however, the morphological changes were not obvious. At this point, qRT-PCR analysis indicated that tRF^Ala-AGC-3-M8 expression levels were significantly higher in the treatment groups compared to the control group (fold changes > 5, peaking at> 40-fold).tRF^Ala-AGC-3-M8 expression levels exhibited a concentration-dependent increase though the 40 µM group was lower compared to the 30 µM group (Fig. 6B). After 36 h of induction, cells in the treatment group showed increased aggregation and cluster areas compared to 24 h of Aβ_25-35 stimulation. At this stage, the cell morphology became irregular and star-shaped. qRT-PCR detection showed that tRF^Ala-AGC-3-M8 expression levels were significantly higher in the treatment groups compared to the control group, but the fold changes decreased, with only the 30 µM and 40 µM groups showing fold changes > 10, while other concentration groups exhibited fold changes < 10 (Fig. 6C). After 48 h of induction, cells of the treatment group showed significant morphological changes, tending toward M1-type polarization (enlarged cell bodies and radiating pseudopodia). Morphological changes were more predominantly toward the center of the clusters (Fig. 6D). Immunofluorescence analysis of iNOS expression showed that iNOS⁺ cells were clustered, with higher iNOS⁺ fluorescence intensity closer to the center (Fig. 6E and F). Cells in the 20 µM treatment group were mostly star-shaped, while cells in the < 20 µM treatment groups were merely round (Fig. 6E and G). Western blot analysis of iNOS protein expression levels after 48 h of Aβ_25-35 induction showed that all treatment groups exhibited higher levels than the control group (Fig. 6H). Both HT22 (Aβ_25-35, 5 µM, 10 µM, 20 µM) and BV-2 cells (Aβ_25-35, 5 µM, 10 µM, 20 µM, 30µM, 40µM) expressed high EphA7 after Aβ_25-35 induction for 24 h (HT22) or 48 h (BV-2) (Fig. 6I and J).

Fig. 6

Characterization of tRFAla−AGC−3−M8 and EphA7 expression in in vitro models of AD. (A–D) qRT-PCR analysis of tRFAla−AGC−3−M8 expression in HT22 cells (after 24 h induction with Aβ25–35) and BV-2 cells (after 24, 36, or 48 h induction with Aβ25–35; n = 6, biologically independent samples); Scale bar: 75μm (overview), 20μm (insert). (E, F) Immunofluorescence analysis of iNOS localization in BV-2 cells after 48 h induction with Aβ25–35. The distance–intensity average curve demonstrates the characteristic expression profile of iNOS in BV-2 cells treated with Aβ25–35 (n = 6, biologically independent samples). (G) Measurement of cell length in iNOS + BV-2 cells (n = 6, biologically independent samples); Scale bar: 75μm (overview), 20 μm (insert). (H–J) Western blot analysis of iNOS and EphA7 expression in BV-2 and HT-22 cells following treatment with Aβ25–35 for 24 or 48 h (n = 6, biologically independent samples); Statistical significance was assessed using one-way ANOVA (P values are indicated). All Data are presented as means ± SEM. Each data point represents the average of 3 technical replicates

Activation of ERK_1/2-p70S6K signaling pathway after Aβ stimulation

Since the specific signaling pathways of EphA7 in AD are not well defined, we investigated potential signaling pathways involving EphA7 in APP/PS1 transgenic mice. WB analysis showed that in APP/PS1 transgenic mice, the phosphorylation of ERK_1/2 (p-value < 0.0001)and p70S6K (T421/S424) (p-value < 0.0001) were significantly increased (Fig. 7A). Parallel in vitro experimentsindicated that in BV-2 cells, the phosphorylation levels of ERK_1/2 (p-value < 0.0001) and p70S6K (T421/S424) (p-value < 0.0001) were significantly increased in the 5 µM and 20 µM Aβ_25-35 treatment groups (p-value < 0.0001) (Fig. 7B). Similarly, in HT22 cells, phosphorylation levels of ERK_1/2 (p-value < 0.0001) and p70S6K (T421/S424) (p-valuep < 0.0001) were significantly elevated in the 10 µM and 20 µM Aβ_25-35 treatment groups (Fig. 7C).

Fig. 7

Western blot analysis of ERK_1/2 and p70S6K phosphorylation in vivo and vitro model of AD. (A) Western blot analysis revealed a significant increase in the phosphorylation of ERK_1/2 and p70S6K (T421/S424) in the APP/PS1 transgenic mice group compared to the control group (n = 6, biologically independent samples). (B) Phosphorylation of ERK_1/2 and p70S6K (T421/S424) was increased in BV-2 cells treated with varying concentrations of Aβ_25-35 (5 μM and 20 μM), n=6 samples/group. (C) HT22 cells treated with Aβ_25-35 at concentrations of 10 μM and 20 μM exhibited elevated phosphorylation of ERK_1/2 and p70S6K (T421/S424) (n = 6, biologically independent samples). Statistical significance was assessed using Student’s test (A) and one-way ANOVA (B and C) (P values are indicated). All Data are presented as means ± SEM. Each data point represents the average of 3 technical replicates

Inhibition of EphA7 or upregulation of tRF^Ala-AGC-3-M8 suppresses ERK_1/2 and p70S6K phosphorylation, reducing M1-type polarization and Tau phosphorylation

The ERK_1/2-p70S6K pathway is activated after Aβ_25-35 stimulation (Fig. 7). To explore whether tRF^Ala-AGC-3-M8 affects the ERK_1/2-p70S6K signaling pathway, we transfected the tRF^Ala-AGC-3-M8-mimic (Group: ON-AGC-3-M8) and NC-mimic (Group: Control-mimic-NC; AD-mimic-NC) into BV-2 cells. After 24 h, qRT-PCR analysis was performed to measure the expression levels of tRF^{Ala-AGC−3−M8} and EphA7. The results showed that tRF^Ala-AGC-3-M8 was significantly increased in BV-2 cells after transfection (fold change > 5, p-value < 0.0001), whereas EphA7 did not show any significant changes (p-value = 0.5695). (Fig. 8A). Subsequently, 48 h of Aβ_25-35 stimulation was conducted in transfected cells, total protein expression levels were measured by WB. The results demonstrated that, in BV-2 cells transfected with NC-mimic, EphA7 expression levels were increased with hyperphosphorylation levels of ERK_1/2 (p-value<0.0001)and p70S6K (p-value<0.0001), while cells transfected with tRF^Ala-AGC-3-M8-mimic showed decreased EphA7 (p-value<0.0001) (Fig. 8B and D) expression and significantly reduced phosphorylation levels of ERK_1/2 (p-value<0.0001) and p70S6K (p-value<0.0001) (Fig. 8C and D). Accordingly, cells transfected with the tRF^Ala-AGC-3-M8-mimic exhibited reduced aggregation, increased cell density, and a more normal morphology (Fig. 8E), along with a significant decrease in average iNOS fluorescence expression (p-value<0.0001) (Fig. 8F and G) and the distribution of iNOS ⁺ BV-2 cells is relatively uniform (Fig. 8F and H). To verify that whether the inhibition of the ERK_1/2-p70S6K pathway by tRF^Ala-AGC-3-M8 was achieved through suppressing EphA7 expression, we performed EphA7-siRNA (Group: Si-EphA7) or NC-siRNA (Group: Control-NC; AD-NC) transfections in BV-2 cells. After 24 h of transfection and following 48 h of Aβ_25-35 stimulation, total protein expression levels were measured by WB. The result shows that the EphA7-siRNA significantly downregulates the expression of EphA7 mRNA (p-value<0.0001) while exerting no discernible impact on the levels of tRF^Ala-AGC-3-M8 (p-value = 0.6415) (Fig. 8I). Additionally, EphA7-siRNA transfected BV-2 cell were consistent with those of the tRF^Ala-AGC-3-M8-mimic treatment, with normalization of cell morphology, reduced iNOS⁺-BV2 cell aggregation (Fig. 8M), decreased EphA7 (p-value<0.0001)expression (Fig. 8J and L), significantly reduced phosphorylation of ERK_1/2 (p-value<0.0028)and p70S6K (p-value<0.0001) (Fig. 8K and L), and decreased iNOS expression (Fig. 8N, O and P).

Fig. 8

Effects of tRF^{Ala−AGC−3−M8} on ERK_1/2-p70S6K pathway in BV-2 cells. (A, I) qRT-PCR analysis of tRF^{Ala−AGC−3−M8} and EphA7 in BV-2 cells after transfection with tRF^{Ala−AGC−3−M8}-mimic or EphA7-siRNA (n = 6, biologically independent samples). (B-D, J-L) Western blot analysis of EphA7 and phosphorylation of ERK_1/2 and p70S6K (T421/S424) in BV-2 cells (n = 6, biologically independent samples). (E, M) Representative light microscopy images of BV-2 cells morphological features; Scale bar: 75μm (overview), 20μm (insert). (F-H, N-P) Immunofluorescence localization an mean fluorescence intensity of iNOS in BV-2 cells (n = 6, biologically independent samples); Scale bar: 75μm (overview), 20μm (insert); Statistical significance was assessed using Student’s test (A and I) and one-way ANOVA (D, L, G and O) (P values are indicated). All data are presented as the means ± SEM. Each data point represents the average of 3 technical replicates

Based on the BV-2 cell validation results, we then transfected HT22 cells with tRF^Ala-AGC-3-M8-mimic (Group: ON-AGC-3-M8), NC-mimic (Group: Control-mimic-NC; AD-mimic-NC), EphA7-siRNA (Group: Si-EphA7), or NC-siRNA (Group: Control-NC; AD-NC). After 24 h of transfection, qRT-PCR analysis was performed to measure the transcript levels of tRF^Ala-AGC-3-M8 and EphA7. The results showed that tRF^Ala-AGC-3-M8-mimic only increased tRF^{Ala−AGC−3−M8} levels (p-value <0.0001)without affecting Epha7 transcription (p-value = 0.6555) and the EphA7-siRNA downregulates the transcription level of Epha7 (p-value<0.0001) without impact on the levels of tRF^Ala-AGC-3-M8 (p-value = 0.3536) (Fig. 9A), which is consistent with the results in BV-2 cells. Subsequently, treatment groups were stimulated with 20µM Aβ_25-35 for 24 h. Total protein expression levels were measured by WB. The results showed that EphA7 expression (AD-mimic-NC: p-value<0.0001; AD-NC: p-value<0.0001) was significantly higher in the AD-mimic-NC and AD-NC groups than in the control group, with significantly increased phosphorylation of ERK_1/2 (AD-mimic-NC: p-value<0.0001; AD-NC: p-value<0.0001) and p70S6K (AD-mimic-NC: p-value<0.0001; AD-NC: p-value<0.0001) (Fig. 9B and C). HT22 cell morphology were changed to varying degrees, mostly linearly, with clustering distances between neuronal cell nuclei (Fig. 9D) accompanied by significantly increased tau protein phosphorylation (Fig. 9E and F). Conversely, the ON-AGC-3-M8 and Si-EphA7 groups showed decreased EphA7 expression levels, reduced phosphorylation of ERK_1/2 (ON-AGC-3-M8: p-value<0.0001;Si-EphA7: p-value<0.0001) and p70S6K (ON-AGC-3-M8: p-value<0.0001; Si-EphA7: p-value<0.0001) (Fig. 9B and C), more directional neuron extension (Fig. 9D), and significantly lower tau protein phosphorylation (ON-AGC-3-M8: p-value<0.0001; Si-EphA7: p-value<0.0001) (Fig. 9E and F).

Fig. 9

Effects of tRF^{Ala−AGC−3−M8} on ERK_1/2-p70S6K pathway in HT22 cells. (A) qRT-PCR analysis of tRF^{Ala−AGC−3−M8} and EphA7 in HT22 cells after transfected with tRF^{Ala−AGC−3−M8}-mimic or EphA7-siRNA (n = 6, biologically independent samples). (B-C) Western blot analysis of EphA7 expression and phosphorylation levels of ERK_1/2 and p70S6K (T421/S424) (n = 6, biologically independent samples). (D) Representative image of HT22 morphological features; Scale bar: 75μm (overview), 20μm (insert). (E, F, G) Immunofluorescence localization and mean fluorescence intensity of p-tau (Red) in HT22 cells (n = 6, biologically independent samples), p-tau was labelled with solid white line in zoomed-in immunofluorescence image; Scale bar: 75μm (overview), 20μm (insert); Statistical significance was assessed using Student’s test (A) and one-way ANOVA (C and F) (P values are indicated). All data are presented as the means ± SEM. Each data point represents the average of 3 technical replicates

Discussion

AD is the most prevalent form of dementia among older adults. Although its etiology is not completely understood, it is generally believed to involve a combination of genetic, environmental, and lifestyle factors [16]. Current diagnostic methods rely on cognitive assessments, neuropsychological tests, neuroimaging techniques (such as magnetic resonance imaging and positron emission tomography), and cerebrospinal fluid analysis [17]. To date, there is no effective cure for AD; existing treatments only alleviate symptoms and improve quality of life. Therefore, developing and validating novel biomarkers, exploring novel drug targets, and formulating effective therapeutic strategies are critical for early and accurate diagnosis and treatment. Over the years, tsRNAs have garnered significant attention as emerging non-coding RNAs because of their high stability and specificity. tsRNAs play crucial roles in various biological processes, including gene regulation, cellular stress responses, and disease development [9]. Our study is the first to identify that tRF^Ala-AGC-3-M8 is significantly reduced in case of Aβ deposition in patients with AD. This reduction was negatively correlated with the expression of the target gene EphA7. EphA7 upregulation activates the ERK_1/2-p70S6K phosphorylation pathway, which promotes microglial M1-type polarization and enhances tau protein phosphorylation in neurons, leading to neuronal damage.

tRF and tiRNA from 5’end of tRNAs may be crucial in the pathology of AD. Current evidence suggests that tsRNAs from different sources have distinct biological functions [18], with 5’tRNAs playing vital roles in immunity, hematopoiesis, and intercellular communication [19]. In the context of AD, where neuroinflammation and dysregulated gene expression are hallmark features, the exploration of specific 5’tRNAs may uncover novel regulatory mechanisms underlying disease progression. In our study, we focused on identifying and characterizing 5’tRNAs with potential roles in AD pathogenesis. Through a combination of sequencing analysis and in vivo validation, we identified three key 5’tRNAs tiRNA^Gly-CCC-2, tRF^Ala-AGC-3-M8, and tRF^Gly-TCC-1 as major regulatory molecules. Notably, our statistical results suggest that tRF^Ala-AGC-3-M8 interacts with EphA7 through an 8mer-1a binding pattern, indicating a perfect match between tRF^Ala-AGC-3-M8 and nucleotides 2 − 8 in the 3’ UTR of EphA7, with A at position one, suggesting a novel mechanism by which tsRNAs may modulate gene expression in AD.

We confirmed the differential expression of EphA7 in AD and, for the first time, reported that microglia in an aggregated state exhibit higher EphA7 expression. EphA7, a member of the Eph receptor family, is well-known for its expression in neurons and its critical role in neuronal growth, differentiation, and synaptic connections [20]. It is involved in neuron-related biological behavior, such as axon guidance [21, 22], synaptic organization, and cortical dendritic development. EphA7 has not been reported to play important physiological functions in microglia. However, there is evidence that EphA2, another Eph receptor, is expressed in microglia and activates the EphA2/p38 MAPK pathway, promoting M1-type polarization, accompanied by reprogramming from oxidative phosphorylation to glycolysis in AD [23]. EphA7 has structural domains similar to those of EphA2, suggesting it may have comparable functions in microglia. Since our sequencing was initially conducted in the cortex and we have established the relationship between tRF^Ala-AGC-3-M8 and EphA7, we hypothesized that the expression of EphA7 in the cerebral cortex would be abnormally distributed due to the reduction of tRF^Ala-AGC-3-M8. labeling of Iba1⁺ /EphA7⁺ cell in the APP/PS1 transgenic mice brain confirmed this hypothesis. The cerebral cortex of APP/PS1 mice exhibited aggregated microglia, and these aggregated microglia significantly expressed EphA7.

The DG-specific upregulation of EphA7 may be closely associated with the DG region’s heightened sensitivity to Aβ toxicity. We subsequently examined EphA7 expression in hippocampal sections. Interestingly, EphA7 was highly expressed in DG neurons. Emerging evidence suggests that DG neurons exhibit increased vulnerability to Aβ-induced synaptic dysfunction and cell death compared to other hippocampal subregions [24]. This selective vulnerability may drive the specific upregulation of EphA7 as a potential protective response or pathological marker. The observed expression pattern could reflect either a compensatory mechanism to counteract Aβ toxicity or an indicator of early pathological changes, given that EphA7 is known to regulate synaptic plasticity and neuronal survival. This region-specific response highlights the DG’s unique position in AD progression and suggests that EphA7 may serve as a key mediator in the differential response to Aβ toxicity across hippocampal subregions.

The downregulation of tRF^Ala-AGC-3-M8 provides another clear pathway to elucidate the toxic effects caused by Aβ deposition. tRF^Ala-AGC-3-M8 belongs to the tRF-5 category, which is primarily derived from the 5’ end of mature tRNA. Studies have shown that cellular stress, such as oxidative stress, hypoxia, and viral infections, can induce the generation of tRF-5 [25,26,27,28,29]. Similarly, The APP/PS1 transgenic mice carry mutant human APP and PS1 genes that promote excessive Aβ aggregation in the brain, leading to amyloid plaque deposition in the cerebral cortex and hippocampus, which subsequently triggers neurotoxic responses. These stresses may cause cells to modulate the expression of tRF-5, especially tRF^Ala-AGC-3-M8. Our in vitro experiments provide compelling evidence that the reduction of tRF^Ala-AGC-3-M8 is directly associated with Aβ-induced neurotoxicity. The temporal dynamics of tRF^Ala-AGC-3-M8 downregulation in HT22 and BV-2 cells reveal a cell-type specific response to Aβ exposure. The earlier downregulation observed in HT22 cells suggests that neurons may be more vulnerable to Aβ toxicity compared to microglia, consistent with previous reports of neuronal susceptibility to Aβ-induced stress [30]. The differential response between BV-2 and HT22 cells can be explained by the inherent capacity of microglia to mitigate Aβ toxicity through phagocytic activity. Our findings support the hypothesis that microglia maintain tRF^Ala-AGC-3-M8 levels through active Aβ clearance during the initial phase of Aβ exposure (24–36 h). However, prolonged Aβ stimulation ultimately overwhelms the microglial defense mechanisms, leading to cellular dysfunction and significant tRF^Ala-AGC-3-M8 reduction. This biphasic response pattern suggests that tRF^Ala-AGC-3-M8 may serve as a sensitive indicator of cellular stress response capacity.

We propose that the interaction between tRF^Ala-AGC-3-M8 /EphA7 axis holds significant implications in the clinical diagnosis and treatment of AD. Dysregulated tsRNA expression has been linked to various diseases. For instance, the level of tRF^His-GTG-1 positively correlates with neutrophil extracellular traps (NETs) formation in patients with systemic lupus erythematosus (SLE), and tRF^His-GTG-1 inhibitors effectively suppress IC-induced NETs formation and hyperactivation, alleviating symptoms in patients with SLE [31]. Additionally, tiRNA^Val-CAC-2 has been shown to interact with far upstream element-binding protein 1 (FUBP1), stabilizing the FUBP1 protein and increasing its enrichment at the c-MYC promoter region, thereby promoting c-MYC transcription and facilitating pancreatic cancer metastasis [32]. These findings highlight the potential of tsRNAs in disease modulation and underscore the importance of exploring their roles in neurodegenerative diseases like AD. In our study, we systematically analyzed the target genes of tRF^Ala-AGC-3-M8 using bioinformatics approaches, revealing significant enrichment in critical biological processes such as axon guidance and synaptic organization, with EphA7 emerging as a central player. Subsequent in vitro validation demonstrated that EphA7 exerts profound effects on key cells of the nervous system, including microglia and neurons, through the ERK_1/2-p70S6K signaling pathway. Current research indicates that the Eph receptor family, including EphA7, influences neurodegenerative disease progression through diverse biological mechanisms. For instance, EphA7 regulates dendritic arborization and spine complexity via Src kinase and the mTOR regulator Tsc1 [33]. Overexpression of EphA7 reduces dendritic branching, while its suppression enhances dendritic complexity and synaptic function. Although direct evidence linking the EphA7-ERK_1/2-p70S6K signaling pathway to AD pathogenesis is still limited, EphA7 is recognized as a critical mediator of ERK_1/2 activation [34]. ERK_1/2 can phosphorylate p70S6K at the T421 and S424 sites, significantly regulating tau protein phosphorylation [35]. Furthermore, p70S6K regulates the expression of cytokines (e.g., IL-6 and TNF-α) and chemokines (e.g., CCL2), as well as transcription factors, such as NF-κB, playing roles in inflammation [36]. Our findings provide compelling evidence that Aβ stimulation downregulates tRF^Ala-AGC-3-M8 expression, leading to aberrant EphA7 expression in hippocampal neurons and activated microglia. This dysregulation enhances phosphorylation of ERK_1/2 and p70S6K at T421 and S424 sites, driving pathological processes such as excessive tau phosphorylation and heightened inflammatory responses, ultimately exacerbating neuronal damage in AD. Importantly, our study identifies the EphA7-ERK_1/2-p70S6K signaling pathway as a novel therapeutic axis that can be targeted for clinical translation. By modulating this pathway, we propose that the detrimental effects of tRF^Ala-AGC-3-M8 dysregulation in AD could be mitigated, offering a promising strategy for therapeutic intervention.

Conclusions

This study is the first to confirm that tRF^Ala-AGC-3-M8 plays a role of Aβ toxicity in AD pathology by regulating EphA7, elucidating its specific mechanisms. This finding is significant for the development of novel therapeutic drugs and the prediction of novel diagnostic biomarkers for AD. Nonetheless, our study has certain limitations. We propose that changes in tRF^Ala-AGC-3-M8 expression in AD are dynamic, especially in microglia undergoing polarization and morphological changes. A sustained reduction in tRF^Ala-AGC-3-M8 may indicate a state where microglia can no longer resist Aβ stimulation. Moreover, since tRF^Ala-AGC-3-M8 is commonly dysregulated in both neurons and microglia, it might play a significant role as a communication molecule in the crosstalk between neurons and microglia in the hippocampal DG in AD. Further experimental data are required to demonstrate the changes and specific causes of tRF^Ala-AGC-3-M8 under Aβ stimulation. Addressing these issues will enhance our understanding of the mechanism of action of tRF^Ala-AGC-3-M8.