Specific pathogen free C57BL/6 wild-type mice were fed under a controlled temperature (23°C) and kept on a 12 h light/dark cycle with food and water provided ad libitum. 3–4 mice were kept in individually ventilated cages. Littermate male mice (4 and 20 months old) were used, and three different tissue samples for RNA extraction were obtained per strain. All animal handling protocols were approved by the Animal Ethics Committee of Shanghai University.

Hippocampus was rapidly dissected in 0°C PBS and placed in RNA extraction solution (Promega, United States). RNA extraction was carried out in accordance with the manufacturer’s protocol. 2 μg total RNA and 4 μl 5X RT Master Mix (TaKaRa, Japan) were added to RNA-free water to a final reaction volume of 20 μl for cDNA synthesis under different temperature of 25°C for 5 min, 37°C for 30 min, 85°C for 10 s, and 12°C for 10 min. The target genes were quantified by real-time PCR using qPCR SYBR Green Master Mix (Yeasen, China). Each reaction contained 1 μl cDNA sample (100–200 ng/μl), 10 μl qPCR SYBR Green Master Mix, 0.8 μl (10 μM) designated primers, and RNA-free water to a final volume of 20 μl. The PCR conditions were as follows: 95°C for 30 s, 55°C for 20 s, 72°C for 30 s, 40 thermal cycles. The mRNA expression was normalized to GAPDH. The primers of each gene were shown in Table 1.

For comparison of gene expression in the hippocampus of 3-, 24-, and 29 month-old mice, the following methods were used. Raw RNA-Seq reads of the gene expression series GSE61915 were downloaded from the NCBI Gene Expression Omnibus database (Stilling et al., 2014). Low quality bases and 5′ and 3′ (Q20 ≤ 20) ends of each read were trimmed using a locally developed Perl script, and adapters were removed by trimmomatic (version 0.36) (Bolger et al., 2014). The clear reads were mapped against mouse genes (GRCm38.p5) using HISAT2 (version 2.0.5) (Kim et al., 2015). The read counts were generated by BEDTools (Quinlan, 2014) and expression values were calculated using the RPKM (Reads per Kilobase per Million mapped reads) method. DEGs were obtained using the fold change of ≥1.50 or ≤0.67, and a t-test p-value of ≤0.05 was also required.

For comparison of gene expression in the hippocampus of mice treated with PBS or human cord plasma protein, human young plasma, and human advanced-age plasma, the processed data from GSE75416 were used directly (Castellano et al., 2017). DEGs were obtained using the fold change of ≥1.2 or ≤0.83, as well as a t-test p-value of ≤0.05.

Enriched functions of significantly changed genes were generated using DAVID 6.8 (Huang da et al., 2009).

Human and mouse molecular interactions were collected from the BioGRID (version 3.4.153) (Chatr-Aryamontri et al., 2017) and IntAct (release 2017/9/2) (Orchard et al., 2014) databases, respectively. The BioGRID and IntAct databases contain various experimentally validated interactions, such as: two-hybrid, affinity capture, co-localization, far-western blotting, proximity label-MS, phage display, coimmunoprecipitation, pull-down, x-ray crystallography, and cross-linking. We were especially interested in the genes that directly interacted with the DEGs of special functions. All the interactions were collected from published literature, and could be retrieved from the original paper via the PubMed IDs. Thus, these data are more reliable than other computationally predicted interactions. All interactions were represented by gene symbol pairs, the BioGRID and IntAct dataset were merged based on these gene symbol pairs. Only human–human or mouse–mouse interactions were retained. For molecular interaction network construction, the first neighbors (most of which represents more reliable and usually direct interactions) of the genes related to the immune response (Figure 2A) were selected, and the interactions were displayed using Cytoscape 3.5.1 (Shannon et al., 2003).

The gene functions in hippocampal development were collected from Uniprot (Boutet et al., 2016) according to their GO annotation. The hippocampal development (Supplementary Table S1)-associated network was constructed using the same method. A slight difference was that all genes that interacted with hippocampal development-associated genes were DEGs, while the immune response-related genes were DEGs in the above network.

The human and mouse networks were constructed.

PsiRNA-YWHAE was constructed for the intracellular silence of YWHAE expression. The hairpin sequence containing YWHAE (YWHAE-for 5′- TCC CAG CTA ACA CTG GCG AGT CCA AGG TTT CAA GAG AAA ACC TTG GAC TCG CCA GTG TTA GCT T -3′ and YWHAE-rev 5′- CAA AAA GCT AAC ACT GGC GAG TCC AAG GTT TTC TCT TGA AAC CTT GGA CTC GCC AGT GTT AGC T -3′) was inserted into the psiRNA-hH1neo vector at two restriction enzyme sites (Bpi I). The negative control (psiRNA-control) with a random sequence (non-targeting siRNA) was also constructed in the same way (Supplementary Figure S1).

HEK293T, an inbreeding of HEK293 (Human embryonic kidney cells 293) cells, were cultured in DMEM (Invitrogen, United States) containing 10% fetal bovine serum (Invitrogen, United States), at 37°C with 5% CO₂. Cells were plated on 6-well plates at a density of 5 × 10⁵/ml, and the following day transfected with 4 μg psiRNA-YWHAE or psiRNA-control as a negative control and cultured for 36 h. Transfection was carried out using Lipofectamine2000 (Invitrogen, United States), according to the manufacturer’s protocol.

The human brain sections (provided by Huashan Hospital, Fudan University, Shanghai, China) were washed three times with PBS, treated with 0.2% Triton X-100 (diluted with PBS) for 30 min, and then blocked in 5% albumin from bovine serum for 60 min. Next, the sections were incubated with an YWHAE antibody at room temperature overnight. The following day, the sections were washed three times with PBS (5 min per wash) and then incubated with a secondary antibody for 1.5 h. Finally, the sections were stained with 4′,6-diamidino-2-phenylindole for 20 min. The expression of YWHAE was detected by confocal microscopy. The experiment protocols were approved by the Human Ethic Committee of Huashan Hospital (Fudan University, Shanghai, China), and Shanghai University Ethics Committee.

To determine the difference at the molecular level during the hippocampal aging process, we compared the gene expression in the hippocampus between young (3 months) and advanced-age (24 and 29 months) mice (Stilling et al., 2014). In comparison with 3 month-age mice, 29 month-age mice showed a larger difference than that of 24 month-age mice, as can be seen from the greater discrete degree in the scatter plot (Figure 1A). These DEGs were subsequently classified into two categories, which included 510 up-regulated and 132 down-regulated genes (Figure 1B). Specifically, 378 and 271 genes were up-regulated by at least 1.5-fold (t-test p-value ≤ 0.05) in the 24- and 29 month-age mice respectively and 139 genes were consistently up-regulated (Figure 1B). 44 and 99 genes, respectively, were down-regulated by at least 1.5-fold, and 11 genes were collectively down-regulated (Figure 1B). The small overlap between the up and down regulated genes of the 24 and 29 months old mice is due to the variable gene expressions between these stages, indicating the aging is a dynamically changing process. It can be inferred that there are more genes up-regulated than those down-regulated. Moreover, the number of overlapping genes in the up-regulated group (139) is substantially larger than in the down-regulated group (11), illustrating that these common genes may play a general role during the process of hippocampal aging. Furthermore, DAVID (Huang da et al., 2009) functional annotation shows that the majority of the up-regulated genes are associated with the innate and inflammatory immune responses (Figure 1C). Interestingly, the immune response-related functions were also found to have significant changes during aging (High, 2004; Weinberger et al., 2008; Balivada et al., 2017), indicating that immune response-related genes may play important roles during the hippocampal aging process.

The above analysis demonstrates the importance of the immune response (Figure 1C) in hippocampal aging, however, the underlying mechanism of this genes network is still unknown. Thus, we investigated the molecular interactions (see section “Materials and Methods”) of these genes in the BioGRID (Chatr-Aryamontri et al., 2017) and IntAct (Orchard et al., 2014) databases. The molecular interactions formed a network (Figures 2A,B) that included 97 nodes and 78 edges, with 24 immune response-related genes being up-regulated (indicated in red nodes) and interacted with its partners. Meanwhile, there are 12 single genes without interactions to others. Through interaction with these 24 genes, their partner genes (green nodes) may indirectly participate in the aging-related immune response. Notably, Ywhae is shown to be linked to many immune response-related genes (Table 2), forming a hub in the network (Figure 2B). Considering its connection to many immune response genes (Figure 2B), it can be speculated that Ywhae may not only have an influence in the nervous system (Baxter et al., 2002; Toyo-oka et al., 2003) but also in the immune response during the aging process.

We also investigated the interactions in human. The hub gene Ywhae of the above immune response associated molecular interaction network in mice was not observed in human (Supplementary Figure S2). However, this does not mean the immune response associated molecular interaction network is not conserved between human and mouse, since the interaction dataset of both human and mouse were incomplete, the overlapped interactions were limited.

In addition to the 139 up-regulated genes (Figure 1B), 11 down-regulated genes were also observed, as shown in Figure 3. The majority of these genes, including Ret, Sftpc, Gpr17, Tpsb2, and Glp1r are associated with disulfide bonds that are crucial to protein structure (Butera et al., 2014), it has been reported that disulfide bonds may influence the function of proteins in the blood (Butera et al., 2014). In addition to disulfide bonds, cell membrane and receptor-associated genes were also observed, including Ret, Lims2, Gpr17 and Glp1r. Membrane receptors embedded in the plasma membrane bind extracellular molecules and allow communication between the extracellular and intracellular space, meaning that environmental exposure may influence aging through transmembrane receptors (Bettio et al., 2017). As shown in Figure 3, Sftpc, for example, encodes the membrane protein surfactant-associated protein C (SP-C) (Glasser et al., 2013). It has been reported that Sftpc plays an important role in innate host defense, enhancing macrophage-mediated phagocytosis and clearing, and limiting inflammatory responses (Glasser et al., 2008), which are all important during aging (Figure 1C). Further, GPR17 is a marker for progenitor progression within the oligodendroglia lineage (Boda et al., 2011) and is expressed in all regions of the brain, indicating that downregulation of GPR17 in our network may interfere with normal brain function and lead to aging. In conclusion, these down-regulated genes take part in many different regulatory pathways together with the up-regulated genes. The current results demonstrate that the identified genes have important biological significance in hippocampal aging research.

It has been reported that a change in hippocampal volume and structure is observed with age-related decline in learning and memory (Verbitsky et al., 2004; Arias-Cavieres et al., 2017; Bettio et al., 2017; Moret-Tatay et al., 2017). Additionally, in the aging hippocampus, neurobiological alterations have been clearly observed including those that alter intracellular signaling and gene expression, and cause neuroinflammation response (Bettio et al., 2017). In recent years, studies regarding of the age-related hippocampus have focused on cognitive decline and impaired memory (Arias-Cavieres et al., 2017; Bettio et al., 2017; Moret-Tatay et al., 2017). Several studies revealed the DEGs in hippocampus between young and the aging mice linked to the cognitive deficient (Xiong et al., 2000; Cheng et al., 2007; Thomas et al., 2014), synaptic plasticity (Dempsey and Ali, 2014) and various dysregulated pathways related to immune and inflammatory response (Landel et al., 2016). In order to understand how significantly up- and down-regulated genes are involved in hippocampal function during aging, their interactions were investigated using public databases and established a network using the privous methods (von Mering et al., 2003; Liu and Zhao, 2004). Interestingly, most of these genes interact with Ywhae (Figure 4), which is the hub gene in the immune response network during hippocampal aging (Figure 2B). Nucleotide polymorphisms in Ywhae would lead to abnormal hippocampal development (Kido et al., 2014), implying that Ywhae maybe influence the function of hippocampus during aging through certain signaling pathway. As shown in Figure 4, YWHAE interacts with VIM, which is an integral regulator of cell adhesion (Dave and Bayless, 2014), neurite extension (Dave and Bayless, 2014), and myelination (Triolo et al., 2012), suggesting that Ywhae may be involved in the regulation of neural function through Vim during aging. Furthermore, GFAP, which is expressed in astrocytes of the central nervous system (Eng and Ghirnikar, 1994), is the partner of YWHAE (Figure 4) and is important for the brain to accommodate neural activities or changes during development (Kim et al., 2011). Moreover, GFAP also plays a crucial role in neuro-inflammation of central nervous system injury (Otani et al., 2003), which is consistent with the speculation that Ywahe is involved in the immune response (Figure 2B) together with GFAP. In conclusion, the interaction between YWHAE and its partners is obviously important for hippocampal development in the nervous system, implying an important function of these genes in the aging process. Interestingly, YWHAE interacts with C4B, which is up-regulated in aging mice and may influence the autoimmune process in diseases (Paul et al., 2002) or modulate interstitial inflammation (Welch et al., 2001), participating in the immune response. Moreover, C4B is also associated with ITGB2, which regulates the lineage distribution of hematopoietic cells in the blood and bone marrow (Gomez and Doerschuk, 2010), implying that Ywhae may be involved in different signaling pathways contributing to the aging process. Essentially, aging is a slow and complicated process that is accompanied by numerous changes in gene expression, especially those related to hippocampal development. This also partially explains why these genes influence hippocampal development and impair memory and learning ability, accelerating aging.

In order to further verify the expression of the genes in our current research (Figure 4), a number of genes in the network were randomly chosen and their expression was evaluated by qPCR in advanced-age and young mice. Hippocampal tissues were extracted from 4- and 20 month-old male mice and the mRNA amplified and compared with that of 4 month-old mice as a control. As shown in Figure 5, the majority of hippocampal development genes (Dclk2, Rara, Hdac1, and Ezh2) were up-regulated, with the exception of Ywhae. Moreover, Tgm1, Cd74, Rbm3, and C4b were also up-regulated compared with the young mice group, which is consistent with the network shown in Figure 4.

In order to further explore the expression pattern of YWHAE (Figure 4) in humans, the expression of YWHAE was evaluated in human brain (Figure 6A). It was found that compared with the young human brain (15- and 25-years-old), YWHAE is down-regulated in the older human brain (69-year-old). These results are consistent with those seen in mice (Figures 5A,B). Furthermore, to evaluate whether the interaction network of YWHAE in mice is conserved in humans, qPCR was carried out to detect the mRNA expression in HEK293T cells after silencing YWHAE. The psiRNA-YWHAE vector was constructed to silence it in HEK293T cells. As shown in Figures 6B,C, with the obvious down-regulated expression of YWHAE, the mRNA levels of DCLK2, RARA, HDAC1, CD74, and RBM3 were significantly increased in HEK293T cells, which is consistent with the results observed in mice (Figure 5A). However, due to the diversity of gene expression in different species, not all genes exhibit the same expression pattern (Figure 5A). For instance, there is no significant difference in the expression of EZH2, TGM1, or C4B at the mRNA level. In conclusion, the gene expression pattern in humans is consistent with that in mice, indicating that the interaction network of YWHAE is conserved between humans and mice.

It has been reported that human umbilical cord plasma proteins can revitalize hippocampal function in advanced-age mice (Castellano et al., 2017), and that tissue inhibitor of metalloproteinases 2, which is an umbilical cord plasma protein, provides a reservoir for the neuroplasticity-promoting process. In order to gain more information regarding this process, microarray data from mice treated with human cord plasma (Castellano et al., 2017) were re-analyzed (Figure 7). In comparison with the PBS control, a total of 148 up-regulated and 190 down-regulated genes (up- or down-regulated by least 1.2-fold, p-value ≤ 0.05) were observed in the samples treated with human cord plasma (Figure 7A). Function enrichment analysis indicates that these genes exhibit functions including sensory perception, skeletal muscle cell differentiation, G protein-coupled receptor signaling, and detection of chemical, amino acid, hormone stimuli (Figures 7B,C). The hypothesis exists that changes in textural perception (Conroy et al., 2017) and loss of skeletal muscle stem cells with age may drive the degeneration of age-related neuromuscular junctions (Liu et al., 2017), indicating that genes with these functions (Figures 7B,C) may participate in aging progression or the revitalization progress. Interestingly, certain inflammatory- and immune response-related genes were also observed among those genes that were down-regulated (Figure 7D). However, no overlapping genes (Figure 7E) were found between those up-regulated during hippocampal aging (Figure 2A) and those down-regulated following treatment with human cord plasma (Figure 7D), indicating that mice adopt a different process to become younger after human umbilical cord plasma treatment. Our current data suggest that advanced-age mice do indeed benefit from these DEGs; however, it is noteworthy that this appears not to be a reverse process from advanced-age to young mice following treatment with umbilical cord plasma. In other words, the genes which contribute to the revitalization of advanced-age mice are different from the genes which are induced by aging during the normal hippocampal aging process.