2.1 Animals and Experiment Design

Sixty male C57BL/6J mice were obtained from SPF Biotechnology Co. Ltd. (Beijing, China). (licensed ID: SCXK 2019-0010). Then animals were divided into four mice per cage and housed for 12 h:12 h light–dark cycles with a normal chow diet. C57BL/6J mice were randomly allocated into six groups: FO (D-gal + fish oil), PSO (D-gal + peony seed oil), CO (D-gal + corn oil), OO (D-gal + olive oil), LO (D-gal + lard oil), and the model group, respectively. Then killed and collected tissue after 10 weeks of feeding. All experimental procedures were approved by the Institutional Animal Care at Qingdao University (QDU-AEC-2023432).

2.2 Fatty Acid and Fat-Soluble Vitamins Detection Analysis

The detailed method of the fatty acid composition refers to the previous method by gas chromatography (GC) (Agilent 7890A, USA). Fatty Acid Methyl Ester standards mix (FAME) (CRM47885, Supelco, USA) as a reference (Cai et al. 2020). Fat-soluble vitamin standards were mixed and determined by high-performance liquid chromatography (HPLC) (Agilent 1200, USA).

2.3 RT-qPCR Analysis

Total RNA of brains and livers was extracted by TRIzol and reversed transcribed into a cDNA with a cDNA synthesis kit. Quantitative PCR was performed using SYBR green with the QuantStudio 1 PCR System (A40425, ThermoFIsher, USA) following the previous method (Zhang et al. 2019). Then the relative expression was calculated as $$$ {2}^{\Delta \Delta {C}_{\mathrm{t}}} $$$ = $$$ {2}^{\Delta \left(\Delta {C}_{\mathrm{t}}\ \mathrm{Target}\ \mathrm{gene}-\Delta {C}_{\mathrm{t}}\ \mathrm{Actb}\right)} $$$, The Actb gene was used as a reference. The target gene's primer information was shown in Table S1.

2.4 Western Blot

Brain tissue was rinsed in pre-cooling PBS, and about 50 mg was added to RIPA lysis buffer to extract total protein brain samples, and western blot (WB) analysis was performed. Firstly, the homogenate was cracked and centrifuged. Combined with PVDF membrane by SDS-PAGE electrophoresis After which the membrane was incubated with the primary antibody PPARG (A11183 Abclonal, WuHan, China) and RXRA (A15242) in a 4°C refrigerator overnight. On the second day, TBST was washed and then incubated with the second antibody at room temperature for 2 h. ActB (AC038) was used as the control reference protein and normalized. According to the previous analysis methods, ImageJ software was used to analyze the relative expression levels of the detected proteins (Zhang et al. 2022).

2.5 RNA-Seq and Differential Gene Expression Analysis

Brain's RNA samples were conducted on the Illumina HiSeq 2000 platform, and then quantity control of the sequence was performed using FastQC v0.10 (de Sena Brandine and Smith 2019). Clean reads were sequenced to mouse reference and obtained the location information by STAR 2.5.3a software (Dobin et al. 2013). RNAseq data were uploaded to the National Genomics Data Center (NGDC) (https://ngdc.cncb.ac.cn/) (CNCB-NGDC Members and Partners 2021) and the BioProject number is CRA004459. Differential gene expression was determined using DESeq2. The p-value was corrected by Benjamin & Hochberg (BH) multiple tests to get the Q-value (adjusted p-value) < 0.05, and |FoldChange|> 1.5 (Love, Huber, and Anders 2014).

The R Bioconductor/cluster_profile and Metascape (http://metascape.org) were applied for Gene Ontology (GO), and the Kyoto Encyclopedia of Genes and Genomes (KEGG) has been previously described (Wang, Wang, and Li 2012; Wang et al. 2020). The protein–protein interaction (PPI) network was visualized with Cytoscape (v3.1.2).

2.6 Transcription Factor Prediction and Multivariate Analysis of Transcript Splicing

We used RNA-seq to screen differential transcription factors (TFs), combined with UCSC Genome Browser (http://genome.ucsc.edu/) and JASPAR 2020 (http://jaspar.genereg.net/) to analyze the confidence and correlation (Fornes et al. 2020). Then, alignment files (Bam) and reference annotations (GTF) were passed to rMATS software, which is a computational tool to detect differential AS events from RNA-Seq data. Differential expression of alternative splice was determined using Junction Counts, significant DAGs|IncLevelDifference|> 0.01 and FDR < 0.05 (Shen et al. 2014).

2.7 Enzyme-Linked Immunosorbent Assay

Brain was homogenized in 50 mg tissue/mL cold PBS and was centrifuged at 12,000 g for 15 min, and protein was detected using the BCA method. According to the ELISA kit manufacturer's protocol to measure the level of TNF-α, AGERs, BDNF, IL-1β, IL-10, and PGF2α (Mlbio, Shanghai, China) in serum and brain.

2.8 Statistical Analysis

All experiments were repeated at least three times, and data were performed as means with standard deviation (mean ± SD). All correlations were calculated using Pearson's correlation coefficient, respectively. One-way analysis of variance (ANOVA) followed by t-test using Graph-Pad Prim software is considered significantly different if p-value < 0.05 (*) or p-value < 0.01 (**).

3.1 Peony Seed Oil Inhibits Neuroinflammation in the Brain

We first recorded that there were no significant differences in feed intake and body weight among the groups (Figure S1A). Then we found that PSO, FO, and LO inhibited the values of AGERs, which is harmful to inflammation induced by D-gal (Figure 1A; p < 0.05). To investigate the effect of peony seed oil and fish oil on neuroinflammation, we detected the pro-inflammatory and anti-inflammatory cytokines in brain homogenate and serum; the concentration of TNF-α and IL-1β was significantly decreased in the PSO and FO groups compared to the model group, but the concentration of IL-10 was significantly increased in the PSO and FO groups compared to the model group (Figure 1B–D; p < 0.05).

We found that the concentration of retinol-binding protein 4 (RBP4), which is used to bind blood-derived retinyl esters and retinol (Klöting et al. 2007). Retinol acid (RA)’ increases were significant differences and correlations with RBP4's concentration in the FO group (R² = 0.711; p = 0.017) and LO group (R² = 0.787; p < 0.008) (Figure 1E; p < 0.05). Meanwhile, PGF-2α, as an important metabolite of AA, its increase was a significant difference and correlation with Lp-PLA2's concentration in the FO group (R² = 0.743; p = 0.013) and PSO groups (R² = 0.805; p = 0.006) compared to the model group (R² = 0.783; p = 0.008) (Figure 1F; p < 0.05). In this part, we found that n-3 PUFA could inhibit neuroinflammation by inhibiting AA metabolism and promoting RA metabolism.

3.2 Peony Seed Oil Promotes the n-3/n-6 PUFAs Biosynthesis

First of all, no changes in liver coefficient among the groups. Then the liver's fatty acid composition was detected because both of them are the most important organs in the process of fatty acid synthesis, metabolism, and transportation (Pawlosky, Barnes, and Salem 1994). Compared with the model group, the proportion of DHA was significantly higher in the PSO and FO groups, and the proportion of AA was significantly lower in the PSO and FO groups. However, there was no significant change in the proportion of palmitic acid (PA, C16:0), palmitoleic acid (POA, C16:1n-7), stearic acid (SA, C18:0), and oleic acid (OA, C18:1n-9) in the liver (Table 1; p < 0.05).

So we investigated the critical enzymes in the process of long-chain fatty acid (LCFA) biosynthesis, elongation, and β-oxidation in the peroxisome, such as fatty acid desaturase (FADS1/2) and elongation of very long fatty acid protein (ELOVL2/3/5/6), and β-oxidation in peroxisome requires acetyl-Coenzyme A acyltransferase 1A (ACAA1A) and acyl-Coenzyme A oxidase (ACOX1/3) (Figure 2A). We found that the expression of Fads1, Fads2, Elovl2, Elovl5, and Acaa1a was significantly higher, and the expression of Acox1 and Acox3 was significantly lower in both the PSO and FO groups compared to the model group (Figure 2B; p < 0.05). While in the brain, the expression of Fads1, Fads2, Elovl2, and Acaa1a was significantly higher but the expression of Acox3 was significantly lower in both PSO and FO groups compared to the Model group (Figure 2C; p < 0.05). In addition, there was a significant correlation between the liver and brain increase of DHA in PSO group (R² = 0.897; p < 0.01) and FO group (R² = 0.972; p < 0.01). Meanwhile, AA's decrease between the liver and brain has a significant correlation in Model group (R² = 0.870; p < 0.01), CO group (R² = 0.801; p < 0.05), FO group (R² = 0.848; p < 0.01) and PSO group (R² = 0.926; p < 0.01) (Figure 2D; p < 0.05). In the section, peony seed oil and fish oil supplements improved the n-3/n-6 PUFAs biosynthesis and elongation in the liver and brain.

3.3 Peony Seed Oil and Fish Oil Affect Fatty Acid Metabolic/Transport-Related Gene Function

To find out the relationship of n-3 PUFA biosynthesis, metabolism, and detailed mechanism, RNA-seq analysis of the brain helped us find that PSO, and FO groups were well differentiated from the model group (Figures 3A and S2A). Compared with the model group, a total of 34, 49, 98, 188, and 193 DEGs were up-regulated, and a total of 51,44, 66, 139, and 147 DEGs were down-regulated in CO, OO, LO, PSO, and FO groups, respectively (Figures 3B and S2B–F; p < 0.05), A total eight common DEGs in the CO, OO, LO, PSO, and FO groups were compared to the model group (Figure 3C,D).

GO functional analysis in intervention groups compared to the model group could be divided into two levels; we found that in PSO and FO-vs-model, it mainly involves cell–cell signaling, cell adhesion, fatty acid metabolism process, regulation of apoptotic process, and long-term memory of the biological process. In cellular components, it mainly affects the membrane, an integral component of the plasma membrane. In molecular function, cellular response to retinoic acid and retinoid X receptor (RXR) binding. (Figures 3E and S3A,B; p < 0.05).

3.4 Peony Seed Oil Activates in PPAR/RXR Signaling Pathway

Then, in the DEGs enrichment analysis by KEGG and GSEA, we found that in PSO, FO, and up-regulated PPAR signaling pathways and down-regulated arachidonic acid metabolism. In the PSO-vs-model group, up-regulated in alpha-linolenic acid metabolism and Fatty acid biosynthesis, down-regulated Linoleic acid and TNF signal pathway. While in the FO-vs-model group, up-regulated retinol metabolism and fatty acid elongation, down-regulated AGE-RAGE signaling pathway, and apoptosis. (Figures 4A,B and S4A and S3B; p < 0.05).

Then we investigated the main TFs of PPARs and RXRs; both of them have three isoforms in the brain, respectively. We find that Pparg ‘s gene expression was significantly higher in the PSO and FO groups compared to the model group. The Rxra's gene expression was higher in the PSO, FO, OO, and LO groups compared to the model group (Figure 4B; p < 0.05). Then, the relative protein level of PPARG was also significantly higher in the PSO and FO groups than in the model group. the relative protein level of RXRA was significantly higher in the PSO, FO, OO, and LO groups compared to the model group (Figure 4C,D; p < 0.05). So we found that the n-3 PUFA supplement activates the PPAR/RXRs signaling pathway.

3.5 Peony Seed Oil Affects Splicing of the Mapt Gene and Inhibits the Tau Protein

Aim to further explore the regulatory mechanism and correlation of n-3 PUFA on key risk factors of differential AS, which is recognized as a regulator of protein structure and function. According to the types of AS, it can be divided into alternative 5′ splice site (A5SS), alternative 3′ splice site (A3SS), mutually exclusive exons (MXE), and skipped exons (SE), and intron retention (IR) by rMATS (Hsieh et al. 2019). We found that the numbers of AS events were lower in the PSO and FO groups compared to the model group (Figure 5A; p < 0.05). Next, aim to find whether the types and numbers of AS are related to the key risk genes in AD, including Psen1/2, Bace1, App, Apoe, Mecp2, and Mapt. Most notably, the AS numbers of the Mapt gene, which is located on chromosome 11 and mainly functional, were encoded by tau protein and were decreased in the PSO and FO groups, and the main AS type is MXE (Figure 5B,C). Then, we want to investigate the changes of tau protein and phosphorylation when the AS state of the exon near the Mapt gene promoter region changes. Then, the TAU's protein expression in the PSO, FO, and LO groups was significantly higher than in the model group (Figure 5D; p < 0.05). The series of evidence indicated that n-3 PUFA disturbed splicing Mapt and inhibited Tau protein expression.