Transcriptomic analysis of glutamate-induced HT22 neurotoxicity as a model for screening anti-Alzheimer’s drugs – Scientific Reports

The quality of RNA sequencing data

After filtering the low-quality reads, high-quality clean reads of 5mMGlu, vehicle control, AE-Et50, and SA-Et50 sample were 12,164,574 (99.77% of the raw reads), 12,161,689 (99.75% of the raw reads), 12,160,701 (99.74% of the raw reads), and 12,159,420 (99.73% of the raw reads), respectively. Clean reads were mapped to reference sequences. Greater than 86.74% of the total clean reads from the RNA-Seq data were aligned and mapped uniquely to the reference genome for all samples.

Expression profiling of DEGs among treatment conditions in cultured HT22 cells

To investigate whether glutamate treatment ± AE or SA extracts alter the transcriptome profiles of cultured HT22 cells, we conducted separate RNA-seq analyses of HT22 cells treated with glutamate, co-treated with glutamate and AE extract, and co-treated with glutamate and SA extract. We found that 285 genes were significantly differentially expressed with glutamate treatment compared with the vehicle control, while 322 and 459 genes were significantly differentially expressed in cells that were co-treated with glutamate + AE extract and glutamate + SA extract, respectively (P < 0.05 and FDR < 0.05; Table 1) relative to the glutamate-treated cells. The lists of DEGs are shown in Supplementary Tables S1–S3. The results of Venn diagram analyses representing overlapping DEGs are shown in Fig. 1 and Supplementary Table S4. The heatmaps of DEGs are shown in Fig. 2.

Hierarchical clustering heatmap of DEGs among treatment conditions in HT22 cells. (a) Heatmap analyses of DEGs obtained using the hierarchical clustering method based on the expression pattern of the overlapping gene set (41 DEGs) from a three-way comparison of DEGs from the different experimental groups, and (b) the complete gene sets (285/322/459 DEGs) resulting from the three treatment-comparison groups. Each row represents a gene, and each column represents each comparison group as indicated at the bottom. The color bar represents the log2 fold change in expression level and ranges from blue (downregulation) to red (upregulation).

Biological pathways and network analyses of DEGs among treatment conditions in cultured HT22 cells

The lists of DEGs among treatment conditions in HT22 cells were analyzed using IPA software to predict canonical pathways, diseases/disorders, and biological functions significantly associated with DEGs. The bubble chart of canonical pathway analysis (Fig. 3a) showed that DEGs induced by glutamate treatment were significantly associated with the changes of canonical pathways in “Neurotransmitters and Other Nervous System Signaling” category, including the activation state of Neurovascular Coupling Signaling Pathway (DEGs: CACNA1C, ENTPD5,GRM1, PLA2G4A, PLA2G4B, and PTGS2) and the inhibition state of Neuroinflammation Signaling Pathway (DEGs: HMOX1, IKBKG, IL1R1, PLA2G4A, PLA2G4B, PTGS2, and TLR1) and Synaptogenesis Signaling Pathway (DEGs: APOE, GRM1, LRP1, NLGN2, THBS1, and THBS2).

The canonical pathways associated with DEGs identified from cells treated with glutamate (a), glutamate with AE extract (b), and glutamate with SA extract (c). The list of DEGs from RNA-seq analysis were analyzed using IPA software to predict canonical pathways associated with DEGs. Bubble plot indicates the canonical pathways, where bubble size corresponds to number of genes enriched for corresponding pathway and color indicates z-score. Orange bubbles indicate predicted activation and a positive z-score, blue bubbles indicate predicted inhibition and a negative z-score.

We found that DEGs from cells co-treated with AE extracts were significantly associated with the changes of canonical pathways in “Cellular Immune Response” category (Fig. 3b), including the activation state of Natural Killer Cell Signaling Pathway (DEGs: CD247, HLA-F, MAP3K1, NFAT5, and ULBP1) and “Ingenuity Toxicity List Pathways” category, including the activation state of p53 Signaling Pathway (DEGs: APAF1, CCNG1, HIPK2, RB1, and THBS1), Xenobiotic Metabolism AHR Signaling Pathway (DEGs: ALDH3A1, HDAC5, NRIP1, and UGT1A6), and Xenobiotic Metabolism CAR Signaling Pathway (DEGs: ALDH3A1, NRIP1, PPM1L, SOD3, and UGT1A6).

In addition, our results revealed that DEGs from cells co-treated with SA extracts were significantly associated with the changes of canonical pathways in “Neurotransmitters and Other Nervous System Signaling” category (Fig. 3c), including the activation state of CREB Signaling in Neurons Pathway (DEGs: ACKR3, CAMK2B, CMKLR1, GPR162, GPRC5A, GPRC5B, GRIK2, GRM1, P2RY14, PDGFB, PIK3R1, PIK3R5, PLCD3, PRKAG2, S1PR1, and TGFB1), Neuropathic Pain Signaling In Dorsal Horn Neurons Pathway (DEGs: CAMK2B, GRM1, PIK3R1, PIK3R5, PLCD3, and PRKAG2), Amyotrophic Lateral Sclerosis Signaling (DEGs: APAF1, CAPN5, GLUL, GRIK2, PIK3R1, and PIK3R5), Reelin Signaling in Neurons (DEGs: APOE, AMK2B, ITGA5, PDK2, PIK3R1, and PIK3R5), Synaptogenesis Signaling Pathway (DEGs: APOE, CAMK2B, GRM1, ITSN2, NECTIN1, NRXN2, PIK3R1, PIK3R5, PRKAG2, SYT7, THBS1, and THBS2), Neuroinflammation Signaling Pathway (DEGs: GLUL, HLA-F, HMOX1, NFAT5, PIK3R1, PIK3R5, PLA2G4B, PTGS2, SLC1A3, TGFB1, and TLR1) and the inhibition state of Neurovascular Coupling Signaling Pathway (DEGs: ENTPD6, GRM1, KCNJ15, NPR1, PLA2G4B, PRKAG2, PTGS2, and SLC1A3) and Semaphorin Neuronal Repulsive Signaling Pathway (DEGs: CD44, CFL2, CSPG4, ITGA5, ITGB2, PIK3R1, PIK3R5, PRKAG2, and VCAN).

We found that “neurological disease” was present among the top diseases/disorders significantly associated with DEGs from cells treated with glutamate relative to vehicle control and DEGs from cells co-treated with glutamate and AE extracts relative to the glutamate treated cells (P < 0.05; Supplementary Table S5). DEGs from cells co-treated with glutamate and SA extracts relative to the glutamate treated cells also were significantly associated with “neurological disease” (P = 1.49E−06 to 4.47E−15). Our results revealed that DEGs induced by glutamate treatment were significantly associated with neurological disorders and diseases, including “cerebral disorder”, central nervous system cancer, “brain glioma”, “glioma”, and “cerebrovascular dysfunction” (P < 0.05; Supplementary Table S6). In addition, several nervous system functions, including “sensory system development”, “activation of microglia”, “differentiation of synapse”, “morphogenesis of nervous tissue”, and “neuritogenesis” were also significantly associated with DEGs (P < 0.05; Supplementary Table S6). Furthermore, we found that DEGs from cells co-treated with glutamate and AE extracts were associated with several neurological diseases, including “early-onset Alzheimer’s disease” (P < 0.05; Supplementary Table S6). Our results revealed that both sets of DEGs from cells co-treated with AE extracts and cells co-treated with SA extracts were significantly associated with “development of neurons” (P < 0.05; Supplementary Table S6).

Biological networks of DEGs were also predicted using IPA software. A representative interactome network of DEGs between control and glutamate treatments showed associations with several molecules including “pro-inflammatory cytokine”, “cytokine receptor”, “growth hormone”, and “apolipoprotein” (Fig. 4a). We found that “Alzheimer’s disease” was present among the diseases/disorders significantly associated with DEGs induced by glutamate treatment. In addition, canonical pathways, including “neuroinflammation signaling pathway”, “NF-κB signaling”, “apoptosis signaling”, “autophagy”, and “synaptogenesis signaling pathway” were also significantly associated with DEGs (Fig. 4a).

Network analyses of DEGs among treatment conditions in HT22 cells. Graphical representation of the interactions between DEGs identified from cells treated with glutamate (a), glutamate with AE extract (b), and glutamate with SA extract (c). Red, up-regulation; green, down-regulation.

The interactome network of DEGs from cells co-treated with AE extracts showed associations with several diseases/disorders and canonical pathways, including “central nervous system cancer”, “congenital anomaly of central nervous system”, “apoptosis of neural precursor cells”, “neuroinflammation signaling pathway”, “NF-κB signaling”, and “amyloid processing” (Fig. 4b). The interactome network of DEGs from cells co-treated with SA extracts showed associations with several diseases/disorders and canonical pathways, including “neurovascular coupling signaling pathway”, “neuroinflammation signaling pathway”, “NF-κB signaling”, “amyloid processing”, “synaptic long-term depression”, and “Parkinson’s signaling” (Fig. 4c). Interestingly, the hub gene in the interactome network of DEGs from cells co-treated with AE extracts and cells co-treated with SA extracts was “NF-κB”, which is the key gene responsible for inflammation and several neurodegenerative diseases (Fig. 4b, c).

RT-qPCR validation of selected DEGs from all treatment conditions

To validate the reliability of the transcriptome sequencing data, the relative expression levels of 9 DEGs involved in Alzheimer’s disease (Apoe, Ptgs2, Rest, Zbed6, Loxl2, Ccl2, Synpo, Ablim1, and Glis3; see also Table 2) were analyzed by quantitative RT-qPCR. RNA-seq analysis and RT-qPCR produced similar gene expression profiles (Fig. 5a). A strong correlation between the log2 fold change of the two methods was observed (Spearman’s correlation coefficient = 0.874, P < 0.001) (Fig. 5b).

Correlation between gene expression ratios obtained from the transcriptome data and RT-qPCR data. (a) Expression levels obtained from transcriptome data (blue) and RT-qPCR data (red). (b) Correlation analysis between the transcriptome and RT-qPCR data. Each point represents a value of the log2 of the relative fold change between pairs of treatment-comparison groups (glutamate treated cells vs. vehicle control, glutamate with AE extracts vs. glutamate treated cells, glutamate with SA extracts vs. glutamate treated cells, respectively). Regression line and Spearman’s correlation coefficient are shown.

In RT-qPCR experiment, we found that Apoe, Ptgs2 and Ccl2 expression was significantly increased while Loxl2 tended to increase in glutamate treatment compared with controls (P < 0.05; Fig. 6a, b, e, f). Co-treatment with AE extract and SA extract significantly attenuated expression of these genes. In addition, Rest, Zbed6, Synpo, Ablim1, and Glis3 expression was significantly decreased in glutamate-treated cells compared with controls (P < 0.05; Fig. 6c, d, g, h, i). Moreover, we found that co-treatment with AE extract and SA extract both significantly induced expression of these genes.

The expression levels of 9 selected DEGs associated with brain function and/or AD pathogenesis determined by RT-PCR for validation of RNA-seq data. Relative expression levels were calculated using the delta-delta threshold cycle (Ct) method and ß-actin as the reference gene. The results are expressed as the mean ± SE, n = 3. P < 0.05 is considered significant. *P < 0.05 glutamate treated cells vs. vehicle control, ^#P < 0.05 glutamate with extracts vs. glutamate treated cells.