Transcriptome analysis of the hippocampus in environmental noise-exposed SAMP8 mice reveals regulatory pathways associated with Alzheimer’s disease neuropathology

Background Chronic noise exposure is one environmental hazard that is associated with genetic susceptibility factors that increase Alzheimer’s disease (AD) pathogenesis. However, the comprehensive understanding of the link between chronic noise stress and AD is limited. Herein, we investigated the effects of chronic noise exposure on AD-like changes in senescence-accelerated mouse prone 8 (SAMP8). Methods A total of 30 male SAMP8 mice were randomly divided into the noise-exposed group, the control group, and aging group (positive controls), and mice in the exposure group were exposed to 98 dB SPL white noise for 30 consecutive days. Transcriptome analysis and AD-like neuropathology of hippocampus were examined by RNA sequencing and immunoblotting. Enzyme-linked immunosorbent assay and real-time PCR were used to further determine the differential gene expression and explore the underlying mechanisms of chronic noise exposure in relation to AD at the genome level. Results Chronic noise exposure led to amyloid beta accumulation and increased the hyperphosphorylation of tau at the Ser202 and Ser404 sites in young SAMP8 mice; similar observations were noted in aging SAMP8 mice. We identified 21 protein-coding transcripts that were differentially expressed: 6 were downregulated and 15 were upregulated after chronic noise exposure; 8 genes were related to AD. qPCR results indicated that the expression of Arc, Egr1, Egr2, Fos, Nauk1, and Per2 were significantly high in the noise exposure group. These outcomes mirrored the results of the RNA sequencing data. Conclusions These findings further revealed that chronic noise exposure exacerbated aging-like impairment in the hippocampus of the SAMP8 mice and that the protein-coding transcripts discovered in the study may be key candidate regulators involved in environment-gene interactions.


Introduction
Alzheimer's disease (AD) is the most common form of dementia, accounting for 50-56% of autopsy and clinical cases [1]. It is marked by a progressive loss of memory and cognitive function, overproduction of amyloid beta (Aβ), and increased hyperphosphorylation of tau [1,2]. However, the pathogenesis of AD is complex, and the disease has no clear cause.
Generally, the association among environmental stress, aging process, and their causal roles plays a role in the development of AD pathogenesis [3][4][5]. Aging is the major risk factor for AD [6]. The incidence of dementia increases exponentially with increasing age [7]. AD-like pathological changes, such as aggregation of Aβ and the phosphorylation of tau protein, are associated with aging [8,9]. Noise-induced health effects were first recognized in industrial settings, such as manufacturers, where intense impulse sound or steady-state long-term exposure with sound pressure levels higher than 85 dBA were associated with auditory and non-auditory effects [10]. Chronic noise exposure in experimental animals can cause not only accelerated Aβ synthesis but also significant and persistent hyperphosphorylation of tau and the formation of prominent pathological neurofibrillary tangles (NFTs) of tau in the hippocampus, key structures in learning and memory, and initial sites of tau pathology in AD [11][12][13][14]. On the other hand, chronic noise exposure could affect the spatial learning and memory [15,16]. Hence, noise exposure may aggravate the development of aging-related AD-like pathological changes, and its related molecular mechanisms need to be explored further.
In the present study, we performed a comparative gene expression analysis of chronic noise-affected and nonnoise-affected brain tissues by using the RNA sequencing technique, an advanced approach that determines the differential expression profiles underlying phenotypic differences [17]. We selected senescence-accelerated mouse prone 8 (SAMP8), which shares behavioral, cognitive, and neuropathological alterations observed in AD patients and is a plausible natural model for exploring the pathogenesis of AD [18]. Our data will serve as a new reference for the prevention and treatment of AD.

Animal use and experimental grouping
The 30 male SAMP8 mice used in our study were provided by the Tianjin University of Traditional Chinese Medicine. The mice were kept under standard housing conditions with controlled ambient temperature (23 ± 2°C) and humidity (50-60%) and a 12-h light/12-h dark cycle (lights on from 06:00 to 18:00). Food and water were provided in their home cages, and they were allowed to adapt to the laboratory environment for 5 days before carrying out the experiment.
Three-month-old male SAMP8 mice were randomly separated into the control group (control group) and noise exposure group exposed to 98 dB SPL white noise (noise group). Eight-month-old male SAMP8 mice were used as positive controls (aging group). The number of animals in each group is equal. The noise group was exposed to 98 dB SPL white noise from 8:00 to 12:00 for 30 days while the control group and the aging group housed in similar cages were exposed to background noise (< 40 dB SPL) from another chamber. After 30 consecutive days of exposure to noise, the mice were sacrificed, and brain samples were immediately collected for biochemical analyses and stored at − 80°C until use. All experiments adhered to the guidelines of the National Institute of Health for the use of experimental animals and were performed in accordance with the approved guidelines specified by the Animal and Human Use in Research Committee of the Tianjin Institute of Environmental Medicine and Operational Medicine.

Noise exposure setup
All noise exposures were performed as previously described [19]. Noise was generated using a noise generator (BK 3560C, B&K Instruments, Denmark), amplified using a power amplifier (Yong-Sheng Audio P-150D; The Third Institute of China Electronics Technology Group, Beijing, China), and transmitted through a loudspeaker (ZM-16S; Tianjin Zenmay Electroacoustic Equipment Co., Tianjin, China). The frequency of the noise emitted from the speaker was in the range of 200-6300 Hz (1/3 octave bands). Animals were exposed to noise in a reverberation chamber, where wire-mesh cages were placed at the center of the sound field. A loudspeaker was suspended directly above the cages. The noise level variation was less than 2 dB in the available space of the animals. The background noise level in the chamber was below 40 dB SPL.

Detection of Aβ by enzyme-linked immunosorbent assay
Frozen mouse hippocampus was homogenized in icecold 1× phosphate-buffered saline (0.02 mol/L, pH 7.0-7.2). Aβ1-40 and Aβ1-42 concentrations were measured in soluble and insoluble brain fractions with mouse ELISA kits (BlueGene Biotech, Shanghai, China) in accordance with the manufacturer's instructions. The mean value of the duplicate samples was considered as the final concentration for each animal.

Western blot analysis
Hippocampus preparation and Western blot were performed as described previously [13,20]. Briefly, each hippocampus was dissected immediately after sacrificing

RNA quantification and qualification
RNA degradation and contamination were monitored using 1% agarose gel. RNA purity was checked using the NanoPhotometer® spectrophotometer (IMPLEN, CA, USA). RNA concentration was measured using the Qubit® RNA assay kit in Qubit® 2.0 Fluorometer (Life Technologies, CA, USA). RNA integrity was assessed using the RNA Nano 6000 assay kit of the Bioanalyzer 2100 system (Agilent Technologies, CA, USA).

Library preparation for transcriptome sequencing
For the RNA sample preparations, 3 μg RNA per sample was used as input material. Sequencing libraries were generated using NEBNext® UltraTM RNA library prep kit for Illumina® (NEB, USA), and index codes were added to attribute sequences to each sample. Briefly, mRNA was purified from total RNA by using poly-T oligo-attached magnetic beads. Fragmentation was carried out using divalent cations under elevated temperature in NEBNext first-strand synthesis reaction buffer (5×). Clustering was performed on a cBot cluster generation system by using TruSeq PE cluster kit v3-cBot-HS (Illumina). The RNA integrity number (RIN) of all samples was greater than 6.8.

Read mapping to the reference genome
Reference genome and gene model annotation files were directly downloaded from the genome website. The index of the reference genome was built using Hisat2 v2.0.5, and paired-end clean reads were aligned to the reference genome by using Hisat2 v2.0.5 (ftp://ftp. ensembl.org/pub/release-94/gtf/mus_musculus/). We selected Hisat2 as the mapping tool to allow Hisat2 to generate a database of splice junctions on the basis of the gene model annotation file and obtain better mapping results than those obtained by other non-splice mapping tools.

Quantification of gene expression level
FeatureCounts v1.5.0-p3 was used to count the read numbers mapped to each gene. Then, the fragments per kilobase of transcript sequence per million (FPKM) of each gene were calculated on the basis of the length of the gene and the number of reads mapped to this gene. The method of using the expected number of FPKM base pairs sequenced considers the effects of sequencing depth and gene length on the read count and is currently the most commonly used technique for estimating gene expression levels. The correlations between samples were evaluated using Pearson's correlation coefficient (Additional file 1: Figure S1, A). And the principal component analysis (PCA) was used to assess the intergroup differences and the repetition of the samples within the group (Additional file 1: Figure S1, B).

Differential expression analysis
Differential expression analysis of two conditions/groups (two biological replicates per condition) was performed using the DESeq2 R package (1.16.1). DESeq2 provides statistical routines for determining the differential expression in the digital gene expression data by using a model based on a negative binomial distribution. The resulting P values were adjusted using the Benjamini and Hochberg's approach for controlling false discovery rates. Genes with an adjusted P value < 0.05 found by DESeq2 were regarded as differentially expressed.

Gene Ontology and Kyoto Encyclopedia of Genes and
Genomes enrichment analysis of differentially expressed genes GO enrichment analysis of differentially expressed genes was implemented using the cluster Profiler R package, in which gene length bias was corrected. GO terms with corrected P value < 0.05 were considered significantly enriched by differentially expressed genes. KEGG is a database resource for understanding the high-level functions and utilities of the biological system. We used the cluster Profiler R package to test the statistical enrichment of the differentially expressed genes in the KEGG pathways. Environmental noise-related genes and pathways were analyzed by Gene Ontology (GO) and Pathway-Relation-Network (Path-net) analysis tools based on Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathway Database using Gene Cloud of Biotechnology Information (GCBI Platform, Shanghai, China) (www.gcbi.com.cn).

Determination of gene expression by real-time PCR
Real-time PCR was performed as previously described [20]. The hippocampus tissues from the exposure and control group mice were homogenized using a rapidly oscillating masher. Total RNA was extracted using an RNeasy Mini kit (TaKaRa Bio, Dalian, China) according to the manufacturer's protocol. Total RNA was converted to cDNA via reverse transcription by using a transcriptor first-strand cDNA synthesis kit (TaKaRa Bio, Dalian, China). A primer pair designed to amplify the GAPDH gene was used as an internal control. Specific primers and probes designed for mouse Arc, Egr1, Egr2, c-Fos, Hmgcr, Hap1, Nauk1, Per2, and GAPDH were used as described in Table 1. Gene expression levels were assessed by quantitative real-time PCR under the following thermal cycling conditions: 2 min at 50°C and 10 min at 95°C followed by 45 cycles of 95°C for 5 s and 57°C for 30 s. Real-time PCR was performed using gene expression assays-on-demand and a Takara PCR thermal cycler dice real-time system (TaKaRa Bio, Dalian, China). The threshold cycle (Ct) of the target genes was normalized to that of GAPDH. mRNA levels in noiseexposed animals were calculated after normalizing the cycle thresholds to the GAPDH expression and were presented as fold-induction values (2 −ΔΔCt ) relative to the control mice.

Statistics
Results were expressed as mean ± standard deviation of the mean. Data were analyzed using the SPSS v. 19.0 software (SPSS Inc., Chicago, IL, USA). Data were subjected to one-way analysis of variance, followed by the least significant difference method test or Dunnett's t test. Values were considered to be significantly different at P < 0.05.

Chronic noise exacerbates AD-like neuropathology in SAMP8 mice
To explore the effects of chronic noise exposure on Aβ in the hippocampus, we determined the concentrations of soluble and insoluble Aβ1-40 and Aβ1-42 in the brain components by using a mouse ELISA kit. The results showed that the amount of Aβ1-40 or Aβ1-42 in the soluble or insoluble fractions was significantly increased in the noise group relative to the control group and was close to that of the aging group ( Fig. 1a-d).
Tau phosphorylation levels were assessed by quantitative immunoblot analyses of cortical extracts from individual mouse in each group after the end of the 30-day experiment period (Fig. 1e, f). In comparison with those in the control group, the levels of tau phosphorylated at Ser202 and Ser404 in the noise group were significantly increased (Fig. 1g, h). These data suggested that chronic noise exposure accelerated AD-like pathological alterations in the SAMP8 mice.

Functional enrichment analysis
A total of 161,813,829 raw reads (80,365,338 for the noise group and 81,448,491 for the control group) were generated. After discarding the reads with adapters, poly-N > 10%, and any other possible contaminants, 156,995,998 clean reads (78,392,394 for the noise group and 78,603,604 for the control group) were obtained. The clean reads were mapped to the mouse reference genome, and the mapping rates were approximately 92.93% and 93.53% for the noise and control group mice, respectively. The cufflink results indicated 30,673 protein-coding transcripts. These mRNAs were used for subsequent analysis.
To identify significant associations of genes with any specific molecular pathway, we performed annotation enrichment analyses. Genes that significantly enriched for P value in GO terms included transmembrane receptor protein serine/threonine kinase signaling pathway, synaptic transmission (GABAergic), sleep and regulation of synaptic transmission (GABAergic), and several terms were closely related to AD, such as learning or memory (GO: 0007611 ), cognition (GO: 0050890), learning (GO: 0007612), and regulation of neuronal synaptic plasticity (GO: 0048168) ( Table 2). By contrast, the KEGG results revealed 20 pathways, of which had significant P values included Toll-like receptor signaling pathway, TNF signaling pathway, and MAPK signaling pathway (Glutamatergic synapse) ( Table 3). We also observed that Egr1, Fos, Egr2, and Arc were the main central genes, and they directly interacted with each other in the gene network analysis.

Differential expression analysis
The mRNA expression levels and transcripts were estimated using FPKMs. A total of 30,673 mRNA transcripts were obtained. Expression analysis showed that 21 mRNA transcripts, including 15 upregulated and 6 downregulated transcripts (P < 0.05), were differentially expressed in the noise group relative to the control group (Table 4). Differentially expressed mRNAs were analyzed using the volcano map and clustering heat map (Fig. 2a, b). Exactly 8 of the 21 differentially expressed genes were related to AD (Table 5).
Validation and quantitative assessment of selected ADrelated differential genes Eight AD-related differentially expressed mRNA transcripts were selected to validate the accuracy of RNA sequencing via qPCR (Fig. 3a-h). The noise-exposed groups exhibited significantly higher mRNA expression of Arc, Egr1, Egr2, Fos, Nauk1, and Per2 than the control mice, thereby confirming the results of the RNA sequencing data. The mRNA expression of Arc, Egr1, and Fos was also markedly higher in the aging mice than in the control mice; the mRNA expression in the latter was similar to that in the noise-exposed mice. No betweengroup differences were observed in the mRNA expression of Hmgcr. The mRNA expression of Hap1 significantly increased in the noise and aging groups, whereas it decreased in the RNA sequencing. To further validate the accuracy of the RNA sequencing, the Egr1 and Fos levels were assessed by Western blot analysis. Immunoblotting confirmed that the expressions of Egr1 and Fos increased more in the noise-exposed and aging mice than in the control mice (Fig. 3i-k). These results indicated that chronic noise exposure induced aging-like impairment in the hippocampi of the SAMP8 mice.

Discussion
Multiple factors contribute to the pathogenesis of AD, and they include aging, genetic variables, and environmental factors [21]. Chronic noise exposure, the main focus of environmental factors, is related to AD [12][13][14]22]. Acute or chronic noise exposure could induce various indicators of AD-like pathological changes that increase the risk of AD development [19,23]. Our recent research further confirmed that chronic noise exposure led to cognitive impairment and Aβ accumulation in young SAMP8 mice, similar to that observed in aging SAMP8 mice [12]. In the present study, we described a reliable transcriptome sequencing to highlight key proteins in the hippocampus that are modulated in response to AD-like neurodegenerations. We also reported genomic signatures that are associated with chronic noise exposure and with age-related changes in the tissue. By overlaying transcriptomic and neuropathological profiles, we identified noise and age-related shifts in underappreciated pathways, such as the Toll-like receptor (TLR) signaling pathway, TNF signaling pathway, MAP kinase pathway, and Ras/ERK signaling pathway, which are involved in the pathogenesis of AD. TLRs, a family of receptor proteins, play a wide role in innate and adaptive immune responses upon the stimulations by exogenous and endogenous TLR ligands. There is an increased endocytosis of TLR4 in the AD model mice brain, which might be a key event for the neurodegeneration signaling in the brain [24]. TNF-α has been assessed in the pathophysiology of AD both in human [25] and animal studies [26], and in our previous study, chronic noise exposure increased levels of TNF-α in the rat hippocampus [20]. The mammalian MAPK family consists of p38-MAPK, ERK, and c-Jun NH2-terminal kinase (JNK), which is  Table 4 Twenty-one significantly dysregulated mRNAs in the noise-exposed and control groups involved in inflammatory reactions and apoptosis processes under conditions of oxidative stress [27]. It was also reported in our previous study that noise exposure could induce neural apoptosis [28], which is influenced by multiple factors and was shown to be closely related to the MAPK pathway [29]. AD is a multifactorial disease, and many theories have been formulated concerning its causes, including neuron loss, Aβ deposition, tau neuropathology, immune system dysfunction, synapse injury, oxidative stress, and mitochondrial dysfunction [30]. In AD, the differential expression of genes can affect different signaling pathways [31]. As expected, our analysis showed that chronic noise exerted a multi-level effect on the acceleration of AD progress, which indicated that chronic noise exposure accelerated AD progress to a certain extent.
Our finding that AD-related genes [32][33][34][35][36], including Egr2, Fos, Arc, Hmgcr, Nuak1, Egr1, Hap1, Per2. Egr2, Fos, Arc, and Hmgcr, show significant differential expression in noise-exposed mice lends support to the idea that environmental hazards including noise exposure may promote the occurrence and development of AD. Egr is a transcription factor and belongs to the family of immediate early genes induced by serum, which is closely related to the differentiation of some tissues and cells and is expressed in the early stage of central nervous system development, which promotes the formation of afterbrain. Upregulation of phagocytic markers Egr2 was observed in Aβ plaque-associated microglia, which reflects an attempt to enhance phagocytosis in plaque-associated microglia [37]. Egr-1 mRNA level is fourfold higher in AD brain than in non-AD brain, and the Egr-1 level is high in brain areas with high NFT density [38]. The overexpression of Egr-1 in rat brain could promote tau phosphorylation at Ser396/404 and Ser262 [39]. In the present study, we found an increase in the phosphorylation levels of tau protein at the Ser202 and Ser404 sites in the noise and aging groups. These alterations were also concomitant with the overexpression of Egr-1 after the end of the stimulus. Furthermore, lasting increase in Fos was found to occur in close correspondence with increase in tau hyperphosphorylation.
Aβ precursor protein (APP) metabolism engages neuronal endolysosomal pathways for Aβ processing and secretion. In AD, the dysregulation of APP leads to excess Aβ. Hap1 may promote the trafficking of APP into the  non-amyloidogenic pathway and reduce the production of Aβ [40]. In the present study, the amount of Aβ increased when Hap1 was decreased; hence, regulating Hap1 expression may help to control Aβ production and affect the development of AD. In our previous study, following noise stress, we detected significantly higher levels not only in Aβ but also in the phosphorylation of tau [16,17,28,41].
Our findings indicate that the long-lasting changes in tau phosphorylation observed after chronic noise exposure are likely to be the result of a complex regulatory network of multiple signaling pathways. Due to the consideration of the influence of too many factors on the results of transcriptome, we did not use wild-type mice to set up the normal control group, which may affect the in-depth analysis of the research results. Additionally, the dose-response relationship of noise exposure was not considered in this study. More perfect experiments will be designed according to the important clues and limitations of the present study.

Conclusion
In summary, our results show the exacerbation of aginglike impairment following chronic noise exposure, as well as candidate genes mediating AD-like pathological changes in the hippocampus of noise-exposed SAMP8 mice. We also identify significant associations of genes with specific molecular pathway, which may be key candidate regulators involved in environment-gene interactions, and suggest that a more detailed look into the underlying mechanism of AD pathogenesis is warranted. Further detailed studies are required to clarify the molecular mechanisms underlying the regulation of the multifaceted signaling system in AD etiology.
Additional file 1: Figure S1. Sample overview of data characteristics. A The correlations between samples. B The inter-group differences and the repetition of the samples within the group.