Abstract

   To investigate the underlying mechanisms of decreased plasmalogens
   (Pls) levels in neurodegenerative diseases, here the effects of
   seafood-derived Pls on undifferentiated and differentiated human
   SH-SY5Y neuroblastoma cells exposed to amyloid-β[1–42] was analyzed.
   Transcriptional profiles indicated that a total of 6,581 differentially
   expressed genes (DEGs) were significantly identified among different
   experimental groups, and KEGG analysis indicated that these DEGs were
   related to AD, endocytosis, synaptic vesicle cycle, autophagy and
   cellular apoptosis. After Pls treatment, the striking expression
   changes of ADORA2A, ATP6V1C2, CELF6, and SLC18A2 mRNA strongly suggest
   that Pls exerts a beneficial role in alleviating AD pathology partly by
   modulating the neurotransmitter release and synaptic transmission at
   the transcriptional level. Besides these, GPCRs are also broadly
   involved in Pls-signaling in neuronal cells. These results provide
   evidence for supporting the potential use of Pls as an effective
   therapeutic approach for AD.

   Keywords: Alzheimer’s disease, plasmalogens, SH-SY5Y cells, gene
   expression, transcriptional profiles

Introduction

   Alzheimer’s disease (AD), a progressive neurodegenerative disorder,
   often occurs among aging people ([35]Chung et al., 2019). This disease
   is characterized by memory loss and cognitive impairment. The
   pathological features of AD are the deposition of amyloid-β (Aβ)
   plaques and formation of neurofibrillary tangles, which is primarily
   constituted by hyperphosphorylated tau proteins in extra- and
   intra-nerve cells, leading to synaptic dysfunction and neuronal death.
   Although different therapeutic approaches have been proposed, only few
   are tested effectively to block AD progression ([36]Sorrentino et al.,
   2017). In addition to aging, studies have shown that several risk
   factors play a crucial role in predisposing for AD. For example,
   smoking, midlife obesity, hypertension, type 2 diabetes,
   hypercholesterolemia and history of traumatic brain injury are strongly
   associated with the onset of AD ([37]Meco et al., 2020).

   The molecular pathogenesis of this age-related cognitive decline
   remains incompletely defined ([38]Wan et al., 2020). Intraneuronal Aβ
   accumulation is believed to trigger neurodegeneration by disrupting
   synaptic transmission and endosomal, lysosomal/proteasomal, and
   mitochondrial functions as well as by facilitating the
   hyperphosphorylation of microtubular tau protein ([39]Omtri et al.,
   2018). Genome-wide association studies on AD patient brains have
   demonstrated significant expression changes in genes that regulate
   vesicular trafficking, cytoskeleton, energy metabolism, inflammation,
   ubiquitin-proteosome system, and autophagy ([40]Liang et al., 2007).
   Besides these, it has been reported that the phosphoinositide 3-kinase
   (PI3K)/AKT-mTOR signaling pathway, nuclear factor kappa B (NF-κB), and
   mitogen-activated protein kinase (MAPK) pathways are activated in
   neurons in several neuropathological conditions ([41]Zheng et al.,
   2017; [42]Coelho et al., 2019; [43]Saha et al., 2020).

   Plasmalogens (Pls) are a special class of dietary glycerophospholipids
   which can be easily found in animal foodstuffs such as poultry,
   livestock and seafood. Pls are abundant in the nervous system,
   especially in the white matter, where Pls are localized in the
   cytoplasmic side of myelin ([44]Che et al., 2018). Besides their
   contribution to membrane integrity as structural components, Pls are
   also involved in multiple cellular processes such as membrane fusion,
   ion transport, cholesterol efflux, and generation of secondary
   messengers ([45]Su et al., 2019). Several literatures have reported a
   direct link between Pls deficiency and AD pathology, whereas the
   administration of Pls (1 mg/day) was effective to improve cognitive
   function of mild AD patients ([46]Ginsberg et al., 1995; [47]Wood et
   al., 2011; [48]Fujino et al., 2017). [49]Che et al. (2018) have
   reported Pls (20 μg/mL in the medium) could significantly decrease
   intracellular and extracellular Aβ[42] levels of CHO-APP/PS1 cells,
   whereas another study suggested that Pls (5 and 20 μg/mL in the medium)
   prevented neuronal cell death by activating G-protein coupled receptors
   (GPCRs) to induce ERK and Akt cellular signaling ([50]Hossain et al.,
   2013, [51]2016). However, limited genes have been analyzed in these
   studies. Therefore, it is necessary to explore the role of Pls in the
   pathogenesis of AD as a whole.

   Aβ peptide is a product of its precursor amyloid polypeptide protein by
   sequential proteolytic cleavages, and it has various conformations.
   Among them, Aβ[1–42] easily forms insoluble aggregates, which are the
   predominant fibers found in senile neuritic plaques of AD brains
   ([52]Selkoe, 2001). Previous studies proved that Aβ[1–42] has
   neurotoxic potential, interfering with synaptic plasticity and
   affecting several cellular signaling pathways ([53]Selkoe, 2001). Thus,
   neurons exposed to appropriate Aβ[1–42] (less than 10 μM) are routinely
   used to obtain the in vitro AD models ([54]Arbo et al., 2017;
   [55]Coelho et al., 2019).

   In this study, SH-SY5Y neuroblastoma cells were used to obtain human
   neuron-like cells using the sequential all-trans retinoic acid (ATRA)
   differentiation and brain-derived neurotrophic factor (BDNF) maturation
   program ([56]Encinas et al., 2000; [57]Goldie et al., 2014). Then, in
   vitro AD model cells were constructed through Aβ[1–42] induction, and
   they were treated with seafood-derived Pls (10 μg/mL) for 24 h.
   Finally, the transcriptional profiles of AD model cells before and
   after Pls treatment were characterized, and the potential effects of
   seafood-derived Pls on the AD-related pathological process were
   uncovered. Results of this study may help us to fully understand the
   beneficial effects of seafood-derived Pls on pathology of patients with
   AD.

Materials and Methods

Cell Culture and Treatment

   The SH-SY5Y human neuroblastoma cell line obtained from BeNa Culture
   Collection (Beijing, China). The SH-SY5Y cells were grown for three
   generations before experiments, and they were used in a low passage
   number (<13). Cells were cultured in flasks in a humidified incubator
   at 37°C and 5% CO[2] (MCO-15AC, Sanyo Electric Co., Osaka, Japan), and
   the culture medium was replaced thrice per week. The SH-SY5Y cells were
   cultured in Dulbecco′s Modified Essential Medium (DMEM, Gibco, San
   Francisco, CA, United States) supplemented with 10% heat-inactivated
   fetal bovine serum (FBS, Gibco) and 100 U/mL penicillin/streptomycin
   (Gibco), and were split into three groups at 80% confluence.

   For the control group, the cells were still cultured in DMEM containing
   10% FBS and 100 U/mL penicillin/streptomycin to reach 80% confluence,
   and were harvested at this stage (at day 3). For the other experiment
   groups, ATRA (Sigma-Aldrich, St. Louis, MO, United States) was added in
   the medium (DMEM + 10% FBS + 10 μM ATRA + 100 U/mL
   penicillin/streptomycin), and the same culture medium was changed on
   day 3. On day 6, the culture medium was changed to serum-free BDNF
   (Life Technologies, Carlsbad, CA, United States) medium (DMEM + 50
   ng/mL BDNF + 100 U/mL penicillin/streptomycin). After 7 days of BDNF
   exposure, neuron-like phenotype were obtained on day 13. To obtain
   cells of Pls group (Pls), these neuron-like cells were further treated
   with seafood-derived Pls (10 μg/mL in medium) for 24 h. For the AD
   model group (AD), the neuron-like cells were treated with Aβ[1–42]
   (1.0–4.0 μM, Sigma-Aldrich) for 24 h before cells collection, whereas
   cells in the AD model + Pls group (AD_Pls) were further treated with 10
   μg/mL seafood-derived Pls in medium for another 24 h and harvested at
   day 15.

   The seafood-derived Pls was extracted and purified from Mussel (Mytilus
   edulis) in the laboratory, according to the method described in a
   previous study ([58]Wang et al., 2021). The purity of the obtained Pls
   was 91.56%, with the major components of phosphatidylethanolamine
   plasmalogens (50.13%) and phosphatidylcholine plasmalogens (41.43%)
   when analyzed by HPLC-ELSD ([59]Wang et al., 2021). The remaining 8.44%
   contained 3.12% PC and 5.08% PE. The proportions of the unsaturated
   fatty acid in Pls and phospholipids were 63.07 and 58.43%.
   eicosapentaenoic acid was the major constituent of unsaturated fatty
   acids, which accounted for 45.82 and 42.35% in Pls and phospholipids,
   respectively ([60]Song et al., 2020; [61]Wang et al., 2021; [62]Zhang
   et al., 2021). During the culture and differentiation stages, SH-SY5Y
   cells were visualized by a Zeiss Axiovert 200 inverted fluorescence
   microscope (Carl Zeiss, Oberkochen, Germany), and documented with
   AxioVisionLE (Carl Zeiss).

The 4′,6-Diamidino-2-Phenylindole and Propidium Iodide (PI) Staining

   The death of SH-SY5Y cells treated with different concentrations of
   Aβ[1–42] was measured by 4′,6-Diamidino-2-Phenylindole (DAPI)
   (Sigma-Aldrich) and Propidium Iodide (PI) (Sigma-Aldrich) staining.
   Briefly, differentiated SH-SY5Y cells were cultured on coverslips in a
   24-well plate. After the cells adhered, 5 mg/L DAPI was added to the
   culture supernatant and cultured overnight in dark. Then, Aβ[1–42] was
   added to each well to the final concentrations of 1, 2, and 4 μM. After
   treatment with Aβ[1–42] for 24 h, 50 mg/L PI was added to each well and
   stained for 3–5 min. Then, cells were observed under a fluorescence
   microscope (Carl Zeiss) and analyzed by AxioVisionLE software. Five
   images were obtained for each cell sample. The total number of nuclei
   was assessed by DAPI staining, while the nuclei of dead cells were
   determined by PI staining. Therefore, the cell death rate was
   determined by comparing the number of PI positive cells to the number
   of total cells in the images.

RNA Extraction and RNA-Seq Library Preparation

   Total RNA was extracted from SH-SY5Y cells of different groups using
   TRIzol Reagent (Invitrogen, CA, United States). The RNA quantity and
   quality were analyzed using a NanoDrop ND-1000 instrument (Agilent, CA,
   United States). The mRNA was purified from 5 μg of total RNA using
   poly-T oligo-attached magnetic beads after two rounds of purification.
   The obtained mRNA was fragmented using divalent cations under elevated
   temperature, and then reverse-transcribed to construct the cDNA library
   using the mRNASeq sample preparation kit (Illumina, San Diego, United
   States). The sequencing was performed on an Illumina Novaseq^TM 6000
   platform (LC Sciences, Houston, Texas, United States). Three biological
   repeats were assessed for each sample, and all of the data were
   expressed as the mean ± standard deviation (SD). The raw datasets of
   sequencing have been submitted to the NCBI Short Read Archive under the
   accession code [63]PRJNA728528.

RNA-Seq Reads Analysis

   The raw transcriptome data were first processed by removing
   adaptor-containing sequences, poly-N, reads shorter than 150 bp, and
   low-quality reads. Then the valid reads were aligned to the homo-genome
   ([64]http://www.ensembl.org/Homo_sapiens/Info/Index) using the HISAT
   package, which allowed multiple alignments (up to 20 by default) and 2
   mismatches while mapping the reads to the reference genes. Mapped reads
   of each sample were assembled using StringTie, and all of the
   transcriptome data were merged to generate a comprehensive
   transcriptome using Perl scripts software. StringTie and edgeR were
   used to calculate the expression levels of all transcripts by
   calculating FPKM. These mRNAs were further analyzed using GO (gene
   ontology) enrichment analysis and KEGG (Kyoto Encyclopedia of Genes and
   Genomes) signaling pathway enrichment analysis.

Quantitative RT-PCR

   To validate the RNA-seq results, the abundance of ADORA2A, APP,
   ATP6V1C2, Bcl-2, DGKK, GSAP, GSK3, IL33, PSEN1, and SLC18A2 mRNAs in
   different experimental groups were analyzed in parallel using an ABI
   Prism 7500 Sequence Detection System (PE Applied Biosystems, CA, United
   States). Briefly, total RNA was extracted from samples with three
   biological repeats, and converted to cDNA using the cDNA reverse
   transcription kit (Applied Biosystems). Quantitative RT-PCR (qRT-PCR)
   primers used for determining the transcript abundance of selected mRNAs
   are listed in [65]Supplementary Table 1. GAPDH was used as the
   housekeeping gene. The qPCR amplification was initiated by denaturation
   at 95°C for 10 s, followed by 40 cycles of 95°C for 10 s and 60°C for
   30 s. Each reaction was repeated three times. The specificity of the
   amplification was examined by melting curve analysis. The relative fold
   changes of the tested genes were calculated using the 2^–ΔΔCt method.

Statistics Analysis

   All results in this study are represented as the average ± standard
   deviation (SD) from three independent experiments. Statistical analysis
   was performed using Microsoft Excel, and one-way analysis of variance
   (ANOVA) followed by Duncan’s new multiple range tests. Differences were
   assumed to be statistically significant, and indicated with different
   letters, for P < 0.05.

Results

Morphological Characteristics of Cultured SH-SY5Y

   In the present study, SH-SY5Y cells were imaged at each stage of
   treatment using the Zeiss Axiovert 200 inverted microscope. As shown in
   [66]Figure 1A, SH-SY5Y cells treated with ATRA for 5 days had reduced
   proliferation and presented a more polar morphology, with cell bodies
   being extended and networks beginning to develop. After further
   maturation with BDNF, cells migrated to clusters, and the cellular
   networks became increasingly complex, which were similar to the mature
   neurons. These mature neuron-like cells were further treated with Pls
   and Aβ[1–42] alone or in combination, to obtain cells of Pls group, AD
   group, and AD_Pls, respectively. However, no obvious morphological
   differentiation was observed during Pls or Aβ[1–42] treatments.

FIGURE 1.

   [67]FIGURE 1
   [68]Open in a new tab

   Characteristics of cultured SH-SY5Y cells and effects of Aβ[1–42] on
   death of SH-SY5Y cells. (A) Morphology of SH-SY5Y cells at different
   differentiation stages observed using an Axiovert inverted microscope.
   A1, untreated SH-SY5Y cells (×200 magnification). A2, SH-SY5Y cells
   treated with ATRA for 5 days (×200 magnification). These cells
   demonstrated an increased death rate and a more polar morphology, as
   cell bodies extended longer and networks began to form. A3, after
   further maturation with BDNF, the cells migrated to clusters and
   networks became increasingly complex (×200 magnification). (B) The
   SH-SY5Y cells were treated with different concentrations of Aβ[1–42]
   for 24 h, the cell death rates were determined by comparing the number
   of dead cells to the number of total cells in the images (×200
   magnification). (C) Quantification of cell death ratios after 24 h
   treatment with the indicated concentrations of Aβ[1–42]. The different
   letters indicated a significant difference between experiment groups at
   p < 0.05 (Duncan’s new multiple range test).

Effects of Aβ[1–42] on Death of SH-SY5Y Cells

   To assess the effects of Aβ[1–42] on cell death, equal numbers of
   differentiated SH-SY5Y cells were seeded in each well of a 24-well
   plate. Nevertheless, the clusters and network structures of the
   neuron-like SH-SY5Y cells disappeared after trypsin hydrolysis during
   transferring from the culture flasks to coverslips. [69]Figure 1B
   presented that the cell death ratio notably increased with an
   increasing in Aβ[1–42] concentrations, indicating that the damage of
   Aβ[1–42] on SH-SY5Y cells was dose-dependent. [70]Figure 1C represented
   that treatment with 2 μM of Aβ[1–42] leads to 26% cell death, while 55%
   of the SH-SY5Y cells died when treatment with 4 μM of Aβ[1–42] for 24
   h. To ensure the optimal induction result and an appropriate cell
   concentration, 2 μM of Aβ[1–42] was finally used in subsequent
   experiments.

Transcriptome Sequencing of SH-SY5Y

   To investigate the effect of Pls on AD development, the transcriptomes
   of samples from normal SH-SY5Y cells, AD model and AD_Pls SH-SY5Y cells
   were sequenced. The RNA sequence results are summarized in [71]Table 1.
   A total of 466,789,804 raw reads were obtained, of which the valid
   reads was 428,666,272. Further analysis of these high-quality cleaned
   reads generated 208,460 assembled transcripts, which corresponded to
   58,826 expression genes. In the alignment analysis, the ratio of valid
   reads which were mapped to homo-genome came up to 96.40–96.72%. The Q30
   values were greater than 97.43%, and the GC contents were ranged from
   51 to 53%. Besides these, the Pearson correlation coefficients of these
   transcriptome profiles among libraries and biological repeats further
   proved the RNA sequence data were reliable ([72]Supplementary Figure
   1).

TABLE 1.

   Summary of the transcriptome sequencing database.
   Group Total reads Valid reads Mapped reads Unique mapped reads Multiple
   mapped reads GC% ≥ Q30%
   ADmodel 160,677,104 147,325,300 142,448,010 (96.69%) 114,367,872
   (77.62%) 28,080,138 (19.06%) 51.50 97.51
   AD model + Pls 147,231,378 137,964,148 133,011,556 (96.40%) 105,793,995
   (76.68%) 27,217,561 (19.72%) 51.00 97.43
   Control 158,881,322 143,376,824 138,685,824 (96.72%) 107,595,429
   (75.02%) 31,090,395 (21.70%) 53.00 97.51
   Total 466,789,804 428,666,272 414,145,390 – – – –
   [73]Open in a new tab

Transcriptional Profiles of Different Experimental Groups

   As shown in [74]Figure 2, 6,581 differentially expressed genes (DEGs)
   were identified as significant when the experimental groups were
   compared with each other. Using a fold change cutoff ratio of ≥ 2 or ≤
   0.5, 4,139 DEGs (3721 up-regulated, 418 down-regulated) were further
   identified when the AD model was compared with control cells. The
   comparison of AD_Pls and control groups yielded 4,002 DEGs (3578
   up-regulated, 424 down-regulated), whereas the comparison of AD and
   AD_Pls groups yielded 192 DEGs (105 up-regulated, 87 down-regulated).
   And overall, 59 DEGs were common in all of the three comparisons.

FIGURE 2.

   FIGURE 2
   [75]Open in a new tab

   Numbers of DEGs between experimental groups, and their corresponding GO
   and KEGG enrichment data. (A) Venn diagram showing the DEGs
   significantly identified by comparison of AD model and control groups,
   AD model + Pls and control groups, and AD model and model + Pls groups
   (fold change > 2 or fold change < 0.5). (B) GO function enrichment
   analysis of DEGs. (C) KEGG pathway enrichment analysis of DEGs.

   Then, the potential biological functions of these DEGs were analyzed.
   In the GO analysis performed among control, AD and AD_Pls groups, the
   most significantly enriched GO terms were “protein binding,” “nucleus,”
   cytoplasm,” “cytosol,” and “nucleoplasm”([76]Figure 2B). Similarly,
   KEGG enrichment analysis indicated that “PI3K-Akt signaling pathway,”
   “Endocytosis,” “MAPK signaling pathway,” “Alzheimer disease,” “mTOR
   signaling pathway,” “protein processing in endoplasmic reticulum,” and
   “Autophagy” were the most represented pathways ([77]Figure 2C).
   Therefore, these pathways were specifically analyzed.

Differentially Expressed Genes Directly Involved in Alzheimer’s Disease

   The 86 DEGs directly involved in “Alzheimer disease” pathway were first
   analyzed. As results presented in [78]Table 2, the expression of
   several hallmarks of AD, such as APAF1, APP, and PSEN1 were obviously
   up-regulated in AD model and AD_Pls cells. Compared with AD model
   cells, the increased levels of these DEGs were less significant in the
   AD_Pls group. Moreover, the results indicated that more than half of
   the AD-related DEGs (48 out of 86) were involved in oxidative
   phosphorylation (OXPHOS). And expression levels of most OXPHOS
   transcripts were decreased in AD model and AD_Pls groups. Besides
   these, several DEGs involved in AD-related calcium signaling pathway
   (such as CACNA1D and ITPR2), were also significantly altered
   ([79]Supplementary Table 2).

TABLE 2.

   List of the significant (p < 0.05) AD-associated DEGs and respective
   fold changes identified among different groups.
   Gene name Fold change (AD model/control) Fold change (AD model +
   Pls/control) Description
   Amyloid β formation
   APAF1 3.85 2.72 Apoptotic peptidase activating factor 1
   APP 3.92 3.73 Amyloid beta precursor protein
   IDE 3.77 2.80 Insulin degrading enzyme
   MME 5.22 3.23 Membrane metalloendopeptidase
   PSEN1 4.73 4.34 Presenilin 1
   AD-related oxidative phosphorylation
   ATP5F1E 0.61 0.70 ATP synthase F1 subunit epsilon, OXPHOS complex V
   COX4I2 0.24 0.37 Cytochrome c oxidase subunit 4I2, OXPHOS complex IV
   COX7A2L 0.39 0.43 Cytochrome c oxidase subunit 7A2 like, OXPHOS complex
   IV
   CYC1 0.52 0.57 Cytochrome c1, OXPHOS complex III
   NDUFA3 0.54 0.72 NADH:ubiquinone oxidoreductase subunit A3, OXPHOS
   complex I
   NDUFS8 0.57 0.74 NADH:ubiquinone oxidoreductase core subunit S8, OXPHOS
   complex I
   SDHD 1.58 1.76 Succinate dehydrogenase complex subunit D, OXPHOS
   complex II
   UQCRB 0.48 0.52 Ubiquinol-cytochrome c reductase binding protein,
   OXPHOS complex III
   [80]Open in a new tab

Differentially Expressed Genes Involved in PI3K-Akt/mTOR and
Mitogen-Activated Protein Kinase Signaling Pathways

   Then, 178 DEGs involved in PI3K-AKT/mTOR signaling pathway were
   identified. As the results shown in [81]Table 3, most of these DEGs
   were up-regulated in either AD or AD_Pls groups compared with the
   control cells. However, the up-regulation extents were less significant
   after Pls treatment, showing by the 102 of 178 down-regulated DEGs
   involved in PI3K-AKT/mTOR signaling pathway when the AD_Pls group was
   compared with AD models. Among them, the down-regulation of SOS1 was
   most significantly, whereas ATP6V1C2 was notably up-regulated after Pls
   treatment.

TABLE 3.

   List of the significant (p < 0.05) PI3K-AKT/mTOR and MAPK signaling
   pathways related DEGs and respective fold changes identified among
   different groups.
   Gene name Fold change (AD model/control) Fold change (AD model +
   Pls/control) Description
   PI3K-AKT/mTOR signaling pathway
   ATP6V1C2 0.15 2.10 ATPase H^+ transporting V1 subunit C2
   FN1 2.02 1.64 Fibronectin 1
   FOXO3 4.91 3.94 Forkhead box O3
   ITGA9 3.19 2.27 Integrin subunit alpha 9
   KIT 4.55 3.37 KIT proto-oncogene receptor tyrosine kinase
   LAMA1 3.49 2.47 Laminin subunit alpha 1
   SOS1[82]^* 4.56 1.64 SOS Ras/Rac guanine nucleotide exchange factor 1
   MAPK signaling pathway
   MAP3K13 2.57 1.48 Mitogen-activated protein kinase kinase kinase 13
   NFATC3 3.47 1.53 Nuclear factor of activated T cells 3
   [83]Open in a new tab

   *Indicate genes involved in both PI3K-AKT/mTOR and MAPK signaling
   pathways.

   The activation of MAPK pathway plays a pivotal role in Aβ-induced
   neuroinflammation ([84]Zaretsky and Zaretskaia, 2021). Consistent with
   previous study, 110 DEGs that participated in the MAPK pathway were
   identified as significant, and 84 of 110 DEGs were up-regulated in the
   AD group compared with controls. After further treatment with Pls, most
   of these DEGs were down-regulated, especially SOS1, MAP3K13, and
   NFATC3.

Differentially Expressed Genes Involved in Endocytosis and Synaptic Vesicle
Cycle

   Increasing evidences have reported that the onset and progression of AD
   are associated with endocytosis, as Aβ enters a cell by endocytosis,
   and then the endocytic vesicle is merged with a lysosome for
   degradation the peptide ([85]Muraleva et al., 2019). In this study, 115
   DEGs related to endocytosis were identified, and most of them were
   up-regulated in both AD and AD_Pls groups. Further comparison between
   the AD and AD_Pls groups indicated that the up-regulation of CBL and
   down-regulation of ARF1 were the most significant after treatment with
   Pls.

   Communication between neurons is mediated by the release of
   neurotransmitters from the synaptic vesicle, thus impairment of
   synaptic vesicle dynamics is believed to be one cause of cognitive
   defects in AD. In this study, 29 DEGs involved in synaptic vesicle
   cycle were identified. Among them, the calcium sensor SYT1 was
   up-regulated by 4.68- and 4.62-folds in AD and AD_Pls groups,
   respectively ([86]Table 3). The expression levels of ATP6V1C2
   (responsible for the vesicle acidification) and SLC18A2 (involved in
   the vesicle retrieved), were notably decreased to 0.03- and 0.0005-fold
   of control cells in the AD group, whereas their levels recovered to
   2.10- and 0.47-fold of controls in AD_Pls group.

Differentially Expressed Genes Related to Autophagy and Apoptosis

   The dramatic increase in autophagic vacuoles is another feature of AD.
   To explore the changes in autophagy in AD and AD_Pls cells, 81
   autophagy-related DEGs were assessed. [87]Table 4 showed that most of
   these DEGs were up-regulated in AD and AD_Pls groups, but the
   up-regulation levels of MTMR3 and STX17 were markedly reduced after Pls
   treatment. Nevertheless, the transcript levels of genes encoding
   lysosomal proteolysis (CTSB and CTSD) did not significantly change
   among different experimental groups.

TABLE 4.

   List of significant (p < 0.05) endocytosis, synaptic vesicle cycle,
   autophagy, and apoptosis-associated DEGs and respective fold changes
   identified among different groups.
   Gene name Fold change (AD model/control) Fold change (AD model +
   Pls/control) Description
   Endocytosis
   ARF1 1.22 0.44 ADP ribosylation factor 1
   CBL 0.35 1.94 Cbl proto-oncogene
   CHMP4A 1.60 2.45 Charged multivesicular body protein 4A
   SH3KBP1 1.64 2.28 SH3 domain containing kinase binding protein 1
   Synaptic vesicle cycle
   ATP6V1C2 0.15 2.10 ATPase H + transporting V1 subunit C2
   SLC18A2 0.00 0.47 Solute carrier family 18 member A2
   SYT1 4.68 4.62 Synaptotagmin 1
   Autophagy
   ATG2B 4.15 3.76 Autophagy related 2B
   ATG4A 2.74 2.37 Autophagy related 4A cysteine peptidase
   ITPR1 4.87 4.25 Inositol 1,4,5-trisphosphate receptor type 1
   LAMP1 1.74 1.42 Lysosomal associated membrane protein 1
   MTMR3 5.47 3.79 Myotubularin related protein 3
   PIK3R1[88]^* 2.10 3.24 Phosphoinositide-3-kinase regulatory subunit 1
   STX17 3.64 1.35 Syntaxin 17
   Apoptosis
   CASP2 2.33 1.13 Caspase 2
   DDIT3 1.54 2.75 DNA damage inducible transcript 3
   ITPR2 5.06 2.92 Inositol 1,4,5-trisphosphate receptor type 2
   PTPN13 5.34 5.35 Protein tyrosine phosphatase, non-receptor type 13
   [89]Open in a new tab

   *indicate genes involved in both autophagy and apoptosis.

   Autophagy is closely related to cell apoptosis, thus the apoptosis
   pathway is also assessed. KEGG analysis indicated that 65 DEGs related
   to apoptosis were identified as significant ([90]Table 4). When the
   three experimental groups were compared with each other, PTPN13 was
   notably up-regulated in both AD and AD_Pls groups. ITPR2 was
   up-regulated by 5.06-folds in the AD group, whereas only 2.92-folds in
   the AD_Pls group. When the AD and AD_Pls groups were compared, the
   expression levels of DDIT3 were increased whereas CASP2 level was
   decreased after treatment with Pls.

Other Differentially Expressed Genes

   Besides the aforementioned DEGs, the transcription level of BHLHB9 was
   up-regulated thousands of times in both AD and AD_Pls groups, whereas
   KLHL11, AKAP2, DGKK, ZNF445, PDE1A, and MYH15 mRNAs were up-regulated
   dozens of times ([91]Table 5). On the contrary, the transcription
   levels of GPCR22 and TRAPPC were significantly down-regulated in both
   AD and AD_Pls groups. Finally, when the AD and AD_Pls groups were
   compared, the expression levels of MATR3, CELF6, and ADORA2 in AD_Pls
   cells were significantly down-regulated after treatment with Pls
   ([92]Table 5).

TABLE 5.

   Other DEGs identified among different groups.
   Gene name Fold change (AD model/control) Fold change (AD model +
   Pls/control) Description
   BHLHB9 1347.47 7905.61 Basic helix-loop-helix family member b9
   PDE1A 27.51 16.27 Phosphodiesterase 1A
   ARMCX4 17.29 7.53 Armadillo repeat containing X-linked 4
   TIAF1 16.03 2.03 TGFB1-induced anti-apoptotic factor 1
   CELF6 0.77 0.03 CUGBP Elav-like family member 6
   ADORA2A 0.92 0.05 Adenosine A2a receptor
   MATR3 5.09 0.03 Matrin 3
   [93]Open in a new tab

Quantitative RT-PCR

   To validate the transcriptome results, 10 transcripts were selected and
   further analyzed by qRT-PCR. As expected, the expression patterns of
   ADORA2A, APP, ATP6V1C2, Bcl-2, DGKK, GSAP, GSK3, IL33, PSEN1, and
   SLC18A2 mRNAs were consistent with those obtained by RNA-seq
   ([94]Figure 3). Compared with control cells, the expression levels of
   DGKK were markedly increased in AD and AD_Pls groups. SLC18A2 and
   ADORA2A were the most significantly down-regulated transcripts in the
   AD group and AD_Pls groups, respectively. However, discrepancies were
   also observed between data obtained from the qRT-PCR and RNA-seq. For
   example, the up-regulation levels of DGKK mRNA in AD and AD_Pls groups
   obtained from qRT-PCR were 25.53 and 16.82, whereas these values were
   30.68 and 27.51 when detected by RNA-seq. These discrepancies might
   result from different sensitivities among different technologies.

FIGURE 3.

   [95]FIGURE 3
   [96]Open in a new tab

   A comparison of the expression levels of several genes in AD model, AD
   model + Pls, and control groups. These genes were selected as KEGG
   analysis indicated that they were AD-related (APP, GSAP, GSK3, and
   PSEN1), autophagy-related (Bcl-2), fatty acid metabolism-related
   (DGKK), or their mRNAs levels were significantly changed among
   different groups (ATP6V1C2, SLC18A2, and IL33). The expression level of
   each gene in the control group was set as 1, and that in AD and AD_Pls
   groups was quantified relative to it. GAPDH was selected as an
   endogenous control. The results were represented as average ± SD (n =
   3), and different letters denoted a significant difference between each
   group at p < 0.05 (Duncan’s new multiple range test).

Discussion

   A key limitation in the field of neuroscience is the lack of suitable
   in vitro models resembling mature neurons. Although animal models are
   informative, they cannot provide a full explanation of the molecular
   mechanisms underlying neuronal function as human transcriptional
   networks are complex, with species-specific gene expression modulation
   patterns ([97]Goldie et al., 2014). SH-SY5Y neuroblastoma cells have
   been used extensively to model networks and pathways related to human
   cognitive disorders. Although there is debate about the need to
   differentiate SH-SY5Y, an increasing number of studies have suggested
   the consideration of experimental methodology and applicability of the
   cell model in answering complex functional questions related to human
   neural architecture ([98]Goldie et al., 2014). Previous studies have
   reported that the ATRA differentiation-BDNF maturation program not only
   induced significant increases in the expression of neuron-specific
   marker genes (such as Sv2, NeuN, and NPY), but also yielded SH-SY5Y
   cells with phenotype approaching mature neurons ([99]Agholme et al.,
   2010; [100]Goldie et al., 2014). Consistent with these studies, the
   expression levels of Sv2C in AD model and AD_Pls cells were 3.88- and
   3.36-folds higher than that of the controls, and the NPY2R levels were
   up-regulated by 8.78- and 6.98-folds in AD and AD_Pls groups when
   compared with the controls ([101]Supplementary Table 2). These data
   indicated that differentiated SH-SY5Y cells displayed mature
   neuron-like appearance and function. These neuron-like cells were
   further treated to obtain AD models, and subsequently used to
   investigate the potential role of seafood-derived Pls in AD
   pathogenesis.

   Down-regulation of OXPHOS genes in animal models of AD has been
   reported in previous studies, and it is suggested that these reductions
   could trigger an overall bioenergetic crisis in the neurons, resulting
   in cell death ([102]Area-Gomez and Schon, 2017; [103]Mastroeni et al.,
   2017). Altered Ca^2+ homeostasis in AD and AD_Pls cells was also
   observed at the transcriptional level in this study, which has been
   considered as an upstream event of AD pathogenesis and often occurs
   before the development of overt symptoms ([104]Tong et al., 2018).
   Collectively, these data indicate that exposure to Aβ[1–42] cause
   energy defects and Ca^2+ dysregulation in SH-SY5Y cells, whereas PLs
   might effectively attenuate this crisis due to less significant
   down-regulation of these genes.

   The PI3K-AKT/mTOR signaling pathway has been considered as the root
   cause of neuropsychiatric disorders such as AD ([105]Sharma and Mehan,
   2021). Up-regulation of PI3K-AKT/mTOR signaling pathway is associated
   with axonal dysregulation, leading to harmful consequences including
   over-production of reactive oxygen species, mitochondrial instability,
   lower oxidative phosphorylation and ATP levels, and even neuronal
   apoptosis ([106]Seitz et al., 2012). Conversely, PI3K-AKT/mTOR
   signaling pathway inhibitors play a neuroprotective role, because of
   their effects in inhibiting cell apoptosis ([107]Seitz et al., 2012).
   Consistent with those previous studies, [108]Table 3 shows that the
   up-regulation degrees of DEGs related to PI3K-AKT/mTOR and MAPK
   signaling pathways were less significant after Pls treatment. These
   data indicate the PI3K-AKT/mTOR and MAPK signaling pathway is activated
   after Aβ[1–42] exposure, whereas Pls treatment inhibits the activation
   of these signaling pathways.

   ATP6V1C2 encodes vacuolar-H^+ ATPase (V-ATPase), which is a proton pump
   that is required for acidification of lysosome/vacuole ([109]Zhao et
   al., 2018). Thus, V-ATPase would influence cellular processes such as
   endocytosis, vacuole fusion and protein degradation in eukaryotes. As
   lysosomes are the final-degradation organelles of Aβ peptide and acidic
   environment is the determinant of hydrolytic enzyme activation, it is
   speculated that down-regulation of ATP6V1C2 mRNA levels in AD model
   cells would lead to increasing of pH value in lysosomes, resulting in
   inactivation of peptidases and inefficient Aβ clearance ([110]Zhao et
   al., 2018; [111]Zhou et al., 2021). However, the acidic lysosomal
   environment recovered after Pls induction by elevated ATP6V1C2
   transcript levels in AD_Pls cells ([112]Table 3). In addition, ATP6V1C2
   also modulates the concentration of neurotransmitters into synaptic
   vesicles and is crucial in synaptic transmission ([113]Zhao et al.,
   2018). A very recent study has reported that the pleiotropic roles of
   low ATP6V1A levels in AD pathogenesis are mediated via the synaptic
   vesicle cycle, phagosome, and oxidative phosphorylation ([114]Zhou et
   al., 2021), which is consistent with results of this study.

   SLC18A2 is another notable DEG involved in synaptic vesicle cycle,
   which encodes a neurotransmitter transporter responsible for the
   packaging of small molecule neurotransmitters (acetylcholine,
   histamine, dopamine, norepinephrine, epinephrine, and serotonin) into
   synaptic vesicles ([115]Hu et al., 2020). SLC18A2 is expressed in
   monoaminergic neurons of the central nervous system, and abnormal
   expression of SLC18A2 has been proposed to contribute to vulnerability
   toward epilepsy-related psychiatric disorders and cognitive impairment
   in adults ([116]Markos et al., 2016; [117]Treble-Barna et al., 2017).
   The SLC18A2

   inhibitor (Deutetrabenazine) is proved to be effective in the cure of
   involuntary movements in patients with tardive dyskinesia, and the
   SLC18A2 blocker (Tetrabenazine) is the only US Food and Drug
   Administration-approved drug for Huntington’s disease ([118]Hu et al.,
   2020). In this study, the transcript levels of ATP6V1C2 and SLC18A2
   were significantly down-regulated in AD model, yet their levels were
   quickly recovered after further treatment with Pls ([119]Table 4).
   Thus, it is suspected that Pls may also alleviate Aβ[1–42]-induced
   neurotoxicity through modulating neurotransmitter system.

   Autophagy plays a modulatory role in the internalization and uptake of
   Aβ, and likely impacts the degradation or formation of Aβ plaques
   ([120]Bordi et al., 2016). Nevertheless, sometimes conflicting results
   were obtained according to earlier investigations of autophagy
   induction in AD models ([121]Bordi et al., 2016). Through a
   custom-designed microarray analysis, a previous study reported that
   genes related to autophagosome formation and lysosomal biogenesis were
   up-regulated, whereas the autolysosomal proteolytic function was not
   evidently altered at early stages of AD ([122]Lipinski et al., 2010).
   Results of this study were consistent with these reports, as shown by
   the overall up-regulation of autophagy Initiation, autophagophore
   formation and elongation, and autophagosome-lysosome fusion-related
   genes, whereas the expression levels of lysosomal proteolysis were not
   evidently changed. Therefore, it is hypothesized that the activation of
   autophagy in AD model cells may represent an acute attempt by the
   affected neuronal cells to rid themselves of the harmful effects of
   Aβ[1–42] exposure. However, abnormally accelerated endocytosis and
   accumulation of Aβ[1–42] eventually become counterproductive as
   lysosomal proteolysis function is insufficient, leading to the
   acceleration of AD onset.

   Among other DEGs, the up-regulation of BHLHB9 was most notable
   ([123]Table 5). BHLHB9 is also known as GPCR-associated sorting protein
   3 (GASP-3), it promotes neurosynaptogenesis by influencing the
   phosphorylation and proteolytic processing of APP and PSENs in
   transgenic mice ([124]Mishra and Heese, 2011). The marked up-regulation
   of BHLHB9 mRNA levels in both the AD and AD_Pls groups may be related
   to the complex neural network and synaptic structures formed in AD and
   AD_Pls cells compared with control cells. However, it is noticed the
   BHLHB9 mRNA level is 5.87-fold higher after further treatment with Pls.
   This result is consistent with the identified roles of BHLHB9 in
   improving memory and learning abilities in animal models, as well as
   the expected effect of Pls on alleviating AD pathology ([125]Mishra and
   Heese, 2011). Other DEGs in [126]Table 5 are also relevant to
   neurodegenerative diseases. PDEs are targets for therapy of AD, as many
   PDE inhibitors have shown encouraging cognitive improvement effects
   ([127]Nabavi et al., 2019). ARMCX4, also known as GASP-4, is one of the
   highly dysregulated priority genes in brains of Parkinson’s disease
   ([128]Abu-Helo and Simonin, 2010). TIAF1 encodes a small TGF-β1-induced
   factor, and a significant up-regulation of Aβ levels occurred rapidly
   following TIAF1 self-association in degenerating and dead neurons
   ([129]Chang, 2009).

   [130]Table 5 also showed that expression levels of ADORA2, CELF6, and
   MATR3 were significantly altered after Pls treatment. MATR3 is one of
   the newly identified dementia-causing genes ([131]Park et al., 2020).
   ADORA2A encodes the adenosine A2A receptor, which is a GPCR that
   mediates synaptic transmission and neuronal excitability in the central
   nervous system ([132]Domenici et al., 2019). Moreover, ADORA2A has been
   reportedly essential for Aβ[1–42] toxicity as Aβ[1–42] did not induce
   learning deficits or synaptotoxicity in ADORA2A knockout mice
   ([133]Chen et al., 2007). In addition, preclinical data have also
   supported the use of adenosine A2A receptor as therapeutic target in
   neuropsychiatric disorders ([134]Domenici et al., 2019). CELF6 encodes
   a RNA-binding protein, which is highly expressed in several
   monoaminergic cell populations and in cells of the hypothalamus
   commonly targeted in psychiatry. A recent study has revealed that many
   targets mRNA of CELF6 encode proteins involved in synaptic transmission
   ([135]Rieger et al., 2020). When the transcriptional alterations of
   ATP6V1C2, SLC18A2, ADORA2, and CELF6 are considered together, it is
   speculated that seafood-derived Pls alleviate the pathology of AD
   mainly by modulating synaptic vesicle trafficking, promoting
   neurotransmitter transport, and synaptic transmission.

   The expression of several mRNAs encoding GPCRs (ADORA2A) and GASPs
   (BHLHB9 and ARMCX4) were significantly altered in this study. These
   findings were consistent with previous research, as it has been
   reported that Pls activate orphan GPCRs to enhance the phosphorylation
   of ERK and Akt kinases, resulting in the inhibition of caspase-3
   activity and thus inhibiting neuronal cell death ([136]Hossain et al.,
   2016). Besides this, GPCRs also regulate tau phosphorylation and Ca^2+
   dysregulation through various cellular kinases in AD neurons
   ([137]Chidambaram and Chinnathambi, 2020). Based on these reports and
   findings of this study, it is suspected that GPCRs are extensively
   involved in AD pathogenesis, and they may contribute to the protective
   role of seafood-derived Pls in SH-SY5Y neuronal-like cells.

   It is worth noting that although the transcription levels of many genes
   related to AD pathogenesis, autophagy, endocytosis, synaptic vesicle
   trafficking, and apoptosis were significantly altered among different
   groups, obvious morphology differences were not observed among these
   neuron-like cells after further treatment with Pls or Aβ[1–42]. These
   results may be related to the short treatment time (24 h) applied in
   this study. Therefore, longer treatment time will be used in the next
   study, to further validate these transcriptional alterations by
   morphological changes. In addition, the protect effective of sea-food
   derived Pls needs to be further confirmed at the protein level.

Conclusion

   In summary, the alteration of transcriptional profiles in AD model and
   AD_Pls cells were investigated, the relevance of PI3K-Akt/mTOR and MAPK
   signaling pathway, autophagy, endocytosis, synaptic vesicle
   trafficking, autophagy and apoptosis to the pathogenesis of AD, as well
   as the potential role of seafood-derived Pls in relieving AD
   progression were analyzed. The obtained data confirmed that the
   activation of autophagy in AD model cells is the first response to
   Aβ[1–42] impairment, while ATP depletion, deficient lysosomal
   proteolytic function are the second step. DEGs significantly identified
   among different groups suggested that therapeutic roles of
   seafood-derived Pls are mediated through accelerating the toxic
   Aβ[1–42]clearance, promoting neurotransmitter transport and synaptic
   transmission, and facilitating the formation of complex neural network
   and synaptic structures in AD model SH-SY5Y neuronal-like cells.
   Results of this work also provided evidence that the GPCRs implicate in
   Pls-signaling in neuronal cells. However, further in vivo studies are
   needed to validate the effect of Pls on AD proposed in this study at
   protein level, and to assess the potential use of seafood-derived Pls
   as an effective therapeutic agent for AD.

Data Availability Statement

   The datasets presented in this study can be found in online
   repositories. The names of the repository/repositories and accession
   number(s) can be found below: NCBI under accession [138]PRJNA728528.

Author Contributions

   JF and GS: study concept and design and participated in the drafting of
   the article. JF and XC: acquisition of data. QW and SG: analysis of
   samples and data interpretation. QS and MZ: critical revision of the
   manuscript for important intellectual content, and study supervision.
   All authors contributed to the article and approved the submitted
   version.

Conflict of Interest

   The authors declare that the research was conducted in the absence of
   any commercial or financial relationships that could be construed as a
   potential conflict of interest.

Publisher’s Note

   All claims expressed in this article are solely those of the authors
   and do not necessarily represent those of their affiliated
   organizations, or those of the publisher, the editors and the
   reviewers. Any product that may be evaluated in this article, or claim
   that may be made by its manufacturer, is not guaranteed or endorsed by
   the publisher.

Funding

   This work was supported by the National Key R&D Program of China
   (2018YFD0901103), the Key Project of Medical Health Science and
   Technology of Zhejiang Province (WKJ-ZJ-1715), the Medical Health
   Science and Technology Project of Zhejiang Province (2020KY443), and
   the Hangzhou Medical and Health Science and Technology Project
   (B20200116).

Supplementary Material

   The Supplementary Material for this article can be found online at:
   [139]https://www.frontiersin.org/articles/10.3389/fnagi.2021.773713/ful
   l#supplementary-material
   [140]Click here for additional data file.^ (245KB, pdf)
   [141]Click here for additional data file.^ (35KB, DOC)
   [142]Click here for additional data file.^ (1.4MB, xls)

References