Male Sprague–Dawley rats (5–8 weeks, 180–220 g) were purchased from Chengdu Dossy Experimental Animals CO., LTD. (Chengdu, Sichuan, China) and maintained under controlled conditions with no restrictions on food or water on a 12 h light/dark cycle. The rats were given some time to adjust to their new surroundings before the experiment. The Animal Care and Use Committee at Sichuan University approved all of the experimental techniques. The experiments in vivo were performed in 3 parts. In the first part, rats were randomly divided into Sham and CCI groups for the behavior tests, RNA-seq, and molecular and biochemical analysis. In the second part, rats received CCI induction for validation of transcriptome profiles, detection of iron content, and lipid peroxidation levels. In the third part, rats were divided into the Sham + vehicle group, CCI group, CCI + vehicle group, CCI + Ferrostatin-1 (Fer-1) group. Fer-1 was dissolved in 1% dimethyl sulfoxide (DMSO). For CCI + Fer-1, rats were administered with Fer-1 2 h after model establishment (intraperitoneal injections (i.p.), 2 mg/kg), followed three times a week for 21 days. For the Sham + vehicle group and CCI + vehicle group, rats receive an equal volume of 1% DMSO solvent.

The CCI model was established based on the findings of a prior study (Bennett and Xie, 1988). Briefly, the left sciatic nerve was exposed at mid-thigh level after rats were sedated and maintained with 2–3% isoflurane. The nerve trunk was then loosely tied four times using 4-O gut sutures near the branching site. Only anesthesia, incision, and nerve exposure were given to the sham rats. The rats eventually recovered on a heating pad and were returned to their cages.

The rats were sedated with sodium pentobarbital (40 mg/kg, i. p.) on day 21, and the rat’s entire hippocampus was promptly collected under the ice. According to the manufacturer’s protocol, the mirVanaTM miRNA ISOlation Kit (Ambion) was used to extract total RNA from the hippocampus of the CCI and sham groups. The Agilent 2,100 Bioanalyzer assessed RNA integrity (Agilent Technologies, Santa Clara, CA, United States). The samples with an RNA Integrity Number (RIN) of 7 were sent for further testing.

The libraries were built using TruSeq Stranded Total RNA with Ribo-Zero Gold, as directed by the manufacturer. To Ribo-Zero deplete and fragment RNA, about 1 g total RNA was processed using TruSeq Stranded Total RNA with Ribo-Zero Gold Kit. After that, the first-strand cDNA synthesis was treated with Act D Mix, followed by the second-strand cDNA synthesis. The 3′ends of cDNA were then adenylated, and the adapters were ligated to the 3′adenylated ends. The PCR Primer Cocktail and PCR Master Mix were also employed to enrich cDNA fragments. The size and purity of the sample were checked using an Agilent Technologies 2,100 Bioanalyzer. Finally, these libraries were sequenced on an Illumina sequencing device (HiSeqTM 2,500 platform), and paired-end reads of 150 bp/125 bp were generated.

Quality control (QC) was performed on the primary sequencing data generated by RNA-Seq (raw reads). Table 1 summarizes the details of total readings and mapping ratio reads. By deleting the adaptor, low-quality bases, low-quality reads, and reads with unknown N bases, Trimmomatic software (Bolger et al., 2014) filtered the raw reads into high-quality clean reads. HISAT2 (version 2.2.1.0) (Kim et al., 2015) was used to align the clean reads to the reference genome and the RSeQC (version 2.6.4) (Wang et al., 2012) technique was used to count the proportions of various comparison types and evaluate the comparison results. Stringtiew2 software (version 1.3.3b) (Pertea et al., 2015) was used to assemble the reads and splice the new transcript after getting the SAM file with the comparison findings. The candidate lncRNA transcripts were then chosen by comparing the gene annotation information of the reference sequence obtained by Cuffcompare software (version 2.2.1) to the gene annotation information of the candidate lncRNA transcripts (Trapnell et al., 2012). Finally, to acquire lncRNA predicted sequences, transcripts with coding potential were screened out using CPC (Kong et al., 2007), Pfam (version v30) (Finn et al., 2006), and PLEK (version 1.2) (Li et al., 2014). The sequencing reads were aligned with the sequence of mRNA transcript sequences using Bowtie (version 2.2.9) (Langmead and Salzberg, 2012). And then, eXpress (version 1.5.1) (Roberts and Pachter, 2013) was applied to make quantitative gene analysis, the FPKM value, and counts value. Finally, differential expression analysis was done using DESeq2 (version 1.18.0) (Delhomme et al., 2012) with a p-value of 0.05 and |Log2 (fold change)| of 0.58.

The detected DEGs were subjected to gene ontology (GO) (http://geneontology.org/) and Kyoto Encyclopedia of Gene and Genomes (KEGG) (http://www.genome.jp/kegg/) enrichment analysis based on the hypergeometric distribution test. Biological process (BP), cellular component (CC), and molecular function were all included in the GO study (MF). The enrichment scores were used to rank the KEGG analysis pathways.

Rather than using total RNA from the hippocampus for RNA-Seq, HiScript^® III-RT SuperMix (Vazyme Biotech Co., Ltd., China) was used to reverse transcribe total RNA from the hippocampus into cDNA according to the manufacturer’s instructions. Table 2 lists all of the primer sequences used. As an internal reference gene, 18sRNA was employed. The qRT-PCR was performed using a Taq Pro Universal SYBR qPCR Master Mix (Vazyme Biotech Co., Ltd., China) kit with a 20L reaction system and a CFX96 Real-Time System (Bio-Rad Laboratories Inc., Hercules, CA, United States). The expression of the 18sRNA gene was normalized to each reaction of a single sample. The ΔΔCT method was applied for the relative gene expression.

Western blot was used to determine the protein expression levels in the hippocampus corresponding to the identified genes. The tissues were suspended in lysis buffer containing 50 mM Tris (pH 8.0), 150 mM NaCl, and protease inhibitors and further sonicated for 30 s at 10% amplitude (ultrasonic cell crusher, Ningbo) to sufficiently dissolve the total proteins. The total proteins in the supernatant were collected using a 16,000 g centrifuge at 4°C for 20 min. The protein concentration of supernatants was determined using a BCA assay kit (Beyotime, Nanjing, China). Protein samples were separated on polyacrylamide gels with a 15% polyacrylamide content and then transferred to PVDF membranes (0.45 mm, Millipore, Bedford, MA, United States). The membranes were blocked for 1 hour at room temperature with 5 percent nonfat milk (BD Biosciences) in Tris-buffered saline with 0.1 percent Tween (TBST), then incubated overnight at 4°C with the primary antibodies anti-ATF3 (abs136180, Absin), anti-GPX4 (A1933, ABclonal), and anti-SLC7A11 (PA1-16893, Thermo Fisher, United States). The membranes were treated in TBST containing 5% nonfat milk for 1 hour at room temperature with horseradish peroxidase-conjugated anti-Rabbit IgG (Goat mAb, Abcam, ab6721). The Enhanced Chemiluminescence Kit (Thermo Pierce, Waltham, MA, United States) was used to detect the immunoreactivity, which was then viewed using ImageJ (BIO-RAD, United States). The -tubulin expression levels were used to normalize the sample’s expression levels.

The DEGs’ interactive network was built using the STRING database (http://string-db.org/). The interactions score >0.4 was used as the reliability threshold value (Otasek et al., 2019). The Cytoscape program (version 3.8.2) was used to view further and analyze the PPI network (Shannon et al., 2003). Finally, the relevance of this network was determined by calculating the connectivity degree of each protein, which is the number of proteins with which it interacts.

The miRanda database was used to predict interactions between lncRNA and miRNAs. The OmicStudio tools (https://www.omicstudio.cn/tool) were then used to determine the likely target binding of miRNA and mRNA interactions using the TargetScan and miRanda databases.

Following the instructions, ROS was assessed via the ROS ELISA kit (Beyotime, Nanjing, China). The rats were sedated with sodium pentobarbital (40 mg/kg, i. p.) on day 21, and the rat’s entire hippocampus was promptly collected. The rats were sedated with sodium pentobarbital (40 mg/kg, i. p.) on day 21, and the rat’s entire hippocampus was promptly collected. The entire hippocampal tissue was then crushed with a sufficient amount of normal saline to create tissue homogenate. The tissue homogenate was centrifuged at 3,000 rpm for 10 min to obtain the supernatant. ROS, malondialdehyde (MDA), superoxide dismutase (SOD), and reduced glutathione (GSH) were all measured in the supernatant. For ROS detection, all reagents were prepared before starting the assay procedure according to the manufacturer’s instructions. Standards and sample diluent were added into standard and testing sample wells, respectively. After that, 100 μl of HRP-conjugate reagent was added to each test well and incubated for 60 min at 37°C. Following that, chromogen solutions A and B were added to each well, and all of the wells were incubated at 37°C for 15 min. Finally, a stop solution was applied to each well, and the Optical Density (OD) at 450 nm was determined within 15 min using a microplate reader.

The MDA level of the hippocampus was evaluated using an MDA assay kit as a marker of lipid peroxidation (A003-1, Nanjing Jiancheng Bioengineering Institute, Nanjing, China). According to the manufacturer’s instructions, standards, anhydrous alcohol, sample diluent, reagents, and 50% glacial acetic acid were all put to corresponding centrifugal tubes. After mixing, all centrifugal tubes were incubated at 95°C for 40 min before centrifuging for 10 min at 3,000 rpm to extract the supernatant. The ODs were measured at 532 nm using a microplate reader.

The SOD activity was measured using the SOD test kit (A001-3; Nanjing Jiancheng Bioengineering Institute, Nanjing, China). The GSH assay kit was used to determine the amount of GSH in the body (A006-2-1, Nanjing Jiancheng Bioengineering Institute, Nanjing, China). The assay was performed according to the manufacturer’s instructions.

On day 21, the rats were anesthetized with sodium pentobarbital (40 mg/kg, i. p.) and perfused with 0.9 percent saline (4°C) and 4 percent paraformaldehyde in PBS (4°C) via the ascending aorta. The hippocampus was then removed, fixed for 24 h in 4% paraformaldehyde, and dehydrated in a 30% sucrose solution. A frozen microtome was used to cut transverse brain sections (40 μm). The sections were first blocked in 5% normal donkey serum in PBS with 0.1% triton for 1 hour at room temperature. Next, they were treated overnight with rabbit anti-ATF3 (abs136180, Absin), anti-NeuN (66836-1-1g, Proteintech), and anti-GFAP primary antibodies (BSM-33065M, Bioss). After washing with PBS, the sections were incubated for 1 hour at room temperature with the second Cy3-, Cy5-, or fluorescein isothiocyanate (FITC)-conjugated secondary antibodies. The sections were then incubated in DAPI for 10 min at room temperature. Finally, an automatically inverted fluorescent microscope (Olympus, Japan) was used to view the sections blindly. The percentage of the stained area of each selected image was calculated using ImageJ software. Three to five images were randomly selected per rat tissue.

Transverse brain sections (20 μm) were prepared with a frozen microtome. The staining was performed following the manufacturer’s directions using Nissl Staining Solution (C0117, Beyotime, Nanjing, China). The sections were fixed with 4% paraformaldehyde for more than 10 min before staining and then rinsed twice with distilled water. Next, the sections were stained for 10 min with Nissl staining and then rinsed twice with distilled water. All of the sections were dehydrated with 95 percent ethanol and then treated with xylene for transparency. Finally, an automatically inverted fluorescent microscope (Olympus, Japan) was used to view the sections blindly.

DAB-enhanced Perls’ stain was applied as specific ferric staining. Transverse brain sections (10 μm) were prepared with a frozen microtome. The staining procedure was followed exactly as described previously (Moroishi et al., 2011). Briefly, the tissue slices were rinsed with distilled water and treated for 30 min with Perls reagent (5% potassium ferrocyanide, 5% HCl), then washed again in distilled water before incubation for 15 min with DAB (0.05% DAB in distilled water). An automatically inverted fluorescent microscope (Olympus, Japan) was then used to observe the sections blindly.

The iron content of the hippocampus and serum were measured to see if there was any evidence of iron overproduction. The measurements were performed using the tissue iron assay kit (A039-2-1, Nanjing Jiancheng Bioengineering Institute, Nanjing, China) and the serum iron assay kit (A039-1-1, Nanjing Jiancheng Bioengineering Institute, Nanjing, China). The tests were carried out under the manufacturer’s instructions. The ODs were measured at 532 nm using a microplate reader.

GraphPad prism 9.0 was used for statistical analysis. For comparisons between two groups, the Student’s t test was used. When comparing the population mean of various related groups, ANOVA tests are acceptable. In graphs, all data is provided as means ± SD. If p < 0.05, the comparison is regarded as statistically significant.

We started by establishing a CCI rat model with memory impairment. After that, the flowchart was used to look at nociceptive thresholds and cognitive activities after CCI (Figure 1A). PWT and PWL decreased dramatically and consistently in the CCI group (Figures 1B,C). In addition, the CCI group developed prolonged cold hyperalgesia (Figures 1D,E). In the Y maze test, spontaneous alternation was considerably reduced on day 7 after CCI compared to the sham group and continued until the conclusion of the observation period (day 21) (Figure 1F). After CCI, the discrimination index in the NOR test was considerably lower on day 21 (Figure 1G). These findings revealed that a successful model of CCI with a memory problem had been established.

We harvested the hippocampus of CCI with memory impairment rats and sham rats to investigate the causes of CCI-induced memory impairment. RNA-Seq was then used to examine the mRNA and lncRNA expression profiles. The sequencing yielded over 95 million raw reads per sample, with a clean reads Q30 ratio of almost 93.0% (Table 1). RNA-Seq resulted in the mapping and identification of 22,601 mRNAs and 15,683 lncRNAs. We then set the filtering criterion for the DEGs to be p-value 0.05 and |Log2 (fold change)| 0.58, as seen in the volcano figure (Supplementary Figure S1A, B). As a result, 179 DEmRNAs and 191 DElncRNAs have been discovered (Supplementary Figure S1A, B and Supplementary Table S1, B). We next used a heat map to describe the DEGs we found, followed by hierarchical clustering analysis. There was a substantial distinction between the sham and the CCI groups with memory impairment. In addition, the results showed clear segregation between the sham and CCI with memory impairment groups (Supplementary Figure S1C,D).

Some of the DEmRNAs and DElncRNAs we found have been linked to oxidative stress or cognitive disorders, such as SLC27A2 (solute carrier family 27 member 2, fold change = 2.332164385), LOC108348106 (alpha-ketoglutarate-dependent dioxygenase alkB homolog 6, fold change = 3.135245512), SLC18A1 (solute carrier family 18 member A1, fold change = 0.590955033), NDUFA10 (NADH dehydrogenase (ubiquinone) 1 alpha subcomplex 10-like 1, fold change = 0.480971719), BMP4 (bone morphogenetic protein 4, fold change = 0.607791806), GDF7 (growth differentiation factor 7, fold change = 0.194844902), OXTR (oxytocin receptor, fold change = 2.250338978), CBLN1 (cerebellin 1 precursor, fold change = 2.247821962). There were 25 DEmRNAs with expression changes greater than 5-fold, with 13 upregulated mRNAs (such as LOC100911730 with 54.98731663 of fold change) and 12 downregulated mRNAs (such as LOC685716 with 0.025834514 of fold change). TCONS_00012750 (fold change = 288.6572008), TCONS_00012029 (fold change = 226.3665341), TCONS_00030575 (fold change = 0.004728013), TCONS_00026782 (fold change = 0.00562614) were among the 46 upregulated genes and 32 downregulated genes identified in the DElncRNA profile. Tables 3, 4, 5 and 6 describe the Top 20 up- and down-regulated genes in further detail.

We also randomly reselected 9 DEmRNAs (5 downregulated and 4 upregulated) and 8 DElncRNAs (5 downregulated and 3 upregulated) from the DEGs list for qRT-PCR validation to check the correctness and reproducibility of the RNA-Seq results (Figure 2). The expression levels of NFS1, CARTPT, APH1B, NDUFA10, and MRPL53 in the hippocampus of CCI with memory impairment model rats 21 days were all significantly downregulated compared to the sham groups (Figure 2A), while ATF3, CBLN1, MT1, and OXTR were all considerably upregulated (Figure 2B), which is consistent with the RNA-Seq expression profiles. Furthermore, LOC102549072, LOC103693876, LOC102550954, LOC108351425, and LOC103691475 were all strongly downregulated (Figure 2C), whereas LOC108352208, LOC103694103, and LOC103693799 were significantly upregulated (Figure 2D), which was also consistent with the RNA-Seq expression profiles.

We used GO analysis of the discovered DEGs in the hippocampus in the sham group and CCI with the memory impairment group to discover more about the molecular process underlying CCI-induced memory impairment. The results showed that the significantly enriched BP of downregulated DEmRNAs were antigen processing and presentation, smooth muscle cell differentiation, etc. (Figure 3A, Supplementary Table S2). The significantly enriched CC of downregulated DEmRNAs was MHC class I protein complex, cell-cell junction, basement membrane, immunological synapse, mitochondrial large ribosomal subunit, etc. (Figure 3C, Supplementary Table S2). The significantly enriched MF of downregulated DEmRNAs were peptide antigen binding, signaling receptor binding, cytokine activity, etc. (Figure 3E, Supplementary Table S2). In addition, the most enriched BP of upregulated DEmRNAs were skeletal muscle cell differentiation, cholesterol biosynthetic process, regulation of long-term synaptic potentiation, etc. (Figure 3B, Supplementary Table S3). The most enriched CC of upregulated DEmRNAs was gap junction, chloride channel complex, an integral component of mitochondrial membrane, mitochondrial matrix, etc. (Figure 3D, Supplementary Table S3). The most enriched MF of upregulated DEmRNAs were pyridoxal phosphate binding, transaminase activity, lyase activity, ferrous iron-binding, etc. (Figure 3F, Supplementary Table S3).

We then looked at the KEGG pathways of the DEmRNAs we found. The primarily enriched pathways of downregulated DEmRNAs were viral myocarditis, graft-versus-host disease, allograft rejection, etc. (Figure 4A, Supplementary Table S4). On the other hand, the primarily enriched pathways of upregulated DEmRNAs were mineral absorption, glycerophospholipid metabolism, sulfur relay system, etc. (Figure 4C, Supplementary Table S4). In addition, the potential targets of downregulated DElncRNAs mainly were enriched in the hedgehog signaling pathway, wnt signaling pathway, oxidative phosphorylation, etc. (Figure 4B, Supplementary Table S5). In contrast, the significantly enriched pathways of upregulated DElncRNAs were taste transduction, mitophagy, antigen processing, presentation, etc. (Figure 4D, Supplementary Table S5).

Then, to analyze the hub genes of identified DEGs that are involved in CCI induced memory damage, we conducted the PPI network analysis and found the significant hub genes, including ALB, ATF3, NPY, CD3E, PIK3CA, GFAP, CD68, CFTR, and RT1-A2 derived from PPI analysis (Supplementary Figure S2).

Firstly, we compared the CCI with memory impairment model rats’ RNA-Seq dataset with RNA-Seq datasets from SNI (GSE18803) and CCI models (Stephens et al., 2021). Total RNA for RNA-Seq was extracted from the spinal cord and the dorsal root ganglion. To detect DEmRNAs from SNI and CCI datasets, we used the same screening criteria as |Log2 (fold change)| ≥ 0.58, p < 0.05. As a result, 3 DEmRNAs from CCI with memory impairment model rats (ATF3, C1QC, and CD68) and 13 DEmRNAs from SNI and CCI models (ADAMTS2, BCL6B, CLCF1, RBP1, and TPBG, among others) overlapped with SNI and CCI models, respectively (Supplementary Figure S3A, Supplementary Table S6). Besides, no overlapping genes were found in any of the three groups.

Furthermore, we compared the CCI with memory impairment model rats’ RNA-Seq dataset with RNA-Seq datasets from aging cognitive decline (GSE9990), various stages of Alzheimer’s disease (GSE28146), and Huntington’s disease (GSE1767), respectively. We set the same screening criteria as |Log2 (fold change)| ≥ 0.58 and p < 0.05 to identify the DEmRNAs from the cognitive aging and AD datasets. These cognitive models did not have any genes in common (Supplementary Figure S3A, Supplementary Table S6). 11 DEmRNAs (APOC1, TMEM106A, TSC22D4, CYR61, BCL6B, MRVI1, LAMC3, BAIAP3, CTXN3, TPBG, and C4B) and 24 DEmRNAs (ABHD10, ARC, CARTPT, CD3E, CLCF1, KIF14, OXTR, RARRES1, SLC18A1, and SLC2A9, etc.) of CCI with memory impairment model rats overlapped with AD and HD model, respectively. There was no DEmRNA of CCI with memory impairment model rats overlapped with a cognitive aging model. 7 DEmRNAs (HEYL, NFS1, MKI67, MFF, OCLN, MGAT2, and HPRT1) of CCI with memory impairment model rats overlapped with both AD and HD models.

Based on the ceRNA hypothesis, lncRNAs may intervene in the translation of mRNAs by sponging miRNA. The ceRNA regulation network was then built in the CCI with memory impairment rat model to highlight the probable processes of lncRNA-miRNA-mRNA. By counting the number of miRNA response elements, we discovered 106 DEmRNAs, 75 miRNAs, and 90 DElncRNAs in total (Supplementary Figure S4 and Supplementary Table S7). We chose 8 DElncRNAs that were validated by our qRT-PCR to construct the network diagram involving 8 DElncRNAs, 12 miRNAs, and 58 DEmRNAs with 137 edges (Supplementary Figure S5) because the figure could not display the complex interaction (Supplementary Figure S6). The significant miRNAs competitively bound by ceRNA were rno-miR-330-5p, rno-miR-326-3p, rno-miR-326-5p, rno-miR-6334, rno-miR-540-3p, and others, according to the network analysis.

The expression level of ATF3 mRNA was shown to be highly elevated in the CA1 region of the hippocampus of CCI patients with memory impairment (Figure 2B), and it has also been linked to the pathological process of pain (Ding et al., 2020). Furthermore, overexpression of ATF3 has recently been attributed to ferroptosis (Wang et al., 2020). Ferroptosis is also reported to be essential for the development of neuropathic pain (Guo et al., 2021; Wang H et al., 2021). However, it is unknown if ATF3 overproduction in the CA1 region of the hippocampus is related to memory impairment caused by CCI and the mechanism. We, therefore, performed immunostaining of hippocampus (CA1 region) using ATF3, combined with NeuN (a marker for neuron) and GFAP (a marker for astrocyte), respectively. We discovered ATF3 positively stained cells were considerably increased in the hippocampus CA1 area of CCI with memory impairment model rats 21 days compared to the sham group, as shown in Figures 5A–C. Furthermore, the number of ATF3-positive neurons rose noticeably (Figure 5A). In contrast, astrocytes showed little ATF3 positivity (Figure 5B). We also used western blot to confirm ATF3 expression and discovered that the level of ATF3 expression in the hippocampus CA1 area of CCI with memory impairment model rats 21 days is considerably higher (Figure 5D). These findings suggest that ATF3 may be linked to memory loss caused by CCI. qRT-PCR was used to examine the expression of gene markers for ferroptosis. qRT-PCR revealed that GPX4 and SLC7A11 gene expression were dramatically downregulated in the hippocampus of CCI with memory impairment model rats 21 days. However, SLC1A5 and PTGS2 gene expression were significantly elevated (Figure 6A), indicating ferroptosis damage (Tang et al., 2021).

Furthermore, the protein expressions of GPX4 and SLC7A11 in the hippocampus of CCI with memory impairment model rats 21 days were all considerably reduced, according to the western blot (Figures 6B,C). We then used DAB-enhanced Perl’s staining to detect ferric iron (Moroishi et al., 2011). Perl’s positive iron deposits were found in the hippocampus of CCI with memory impairment model mice after 21 days, implying that ferrous iron collected in the hippocampus (Figure 7B). Interestingly, the iron level of the plasma and the hippocampus were both elevated in the CCI with memory impairment group (Figure 7C).

The hallmarks of ferroptosis, including ROS, MDA, GSH, and SOD in the hippocampus, were found to include lipid peroxidation and ROS buildup. As shown in Figures 7E–H, the level of ROS and SOD were dramatically raised, whereas the GSH content and SOD activity were clearly downregulated in the hippocampus of CCI rats compared to sham rats. In addition, the number of Nissl-body neurons was significantly decreased. The shape of Nissl positive bodies was modified in the hippocampus of CCI with memory impairment group compared to the sham group, according to the results of Nissl staining (Figures 7A,D). These findings suggested that the CCI caused hippocampal ferroptosis damage.

To verify whether ferroptosis is related to cognitive impairment induced by CCI, we employed the well-known ferroptosis inhibitor Fer-1 to treat rats. As expected, when compared to CCI or CCI + vehicle groups, Fer-1 treatment could significantly reverse the hyperalgesia generated by CCI (Figures 8A–D). Moreover, Fer-1 treatment alleviated the CCI-induced memory impairment, evidenced by greater spontaneous alternation behavior in the Y maze test and improved capacity to detect novel items in the NOR Test (Figures 8E,F).

After Fer-1 treatment, we further observed the ferroptosis damage changes of CCI rats. As shown in Figures 9A,B, Fer-1 markedly reduced the iron levels of hippocampus and plasma after CCI. Bedsides, the Fer-1 treatment reduced ROS production and lipid peroxidation while increasing GSH levels and SOD activity (Figures 9C–F). qRT-PCR revealed that in the hippocampus of CCI with memory impairment model rats 21 days after Fer-1 therapy, GPX4 and SLC7A11 gene expressions were dramatically increased. In contrast, SLC1A5 and PTGS2 gene expressions were significantly reduced (Figures 9G–J). Those results indicated that ferroptosis is involved in CCI-induced memory impairment.