Transcriptional Profiles Reveal Deregulation of Lipid Metabolism and Inflammatory Pathways in Neurons Exposed to Palmitic Acid

The effects of the consumption of high-fat diets (HFD) have been studied to unravel the molecular pathways they are altering in order to understand the link between increased caloric intake, metabolic diseases, and the risk of cognitive dysfunction. The saturated fatty acid, palmitic acid (PA), is the main component of HFD and it has been found increased in the circulation of obese and diabetic people. In the central nervous system, PA has been associated with inflammatory responses in astrocytes, but the effects on neurons exposed to it have not been largely investigated. Given that PA affects a variety of metabolic pathways, we aimed to analyze the transcriptomic profile activated by this fatty acid to shed light on the mechanisms of neuronal dysfunction. In the current study, we profiled the transcriptome response after PA exposition at non-toxic doses in primary hippocampal neurons. Gene ontology and Reactome pathway analysis revealed a pattern of gene expression which is associated with inflammatory pathways, and importantly, with the activation of lipid metabolism that is considered not very active in neurons. Validation by quantitative RT-PCR (qRT-PCR) of Hmgcs2, Angptl4, Ugt8, and Rnf145 support the results obtained by RNAseq. Overall, these findings suggest that neurons are able to respond to saturated fatty acids changing the expression pattern of genes associated with inflammatory response and lipid utilization that may be involved in the neuronal damage associated with metabolic diseases.


Introduction
The chronic consumption of high-fat diets (HFD) has been involved in the development of several pathological conditions such as obesity, type-2 diabetes, and even cognitive impairments that may lead to dementia [1][2][3][4][5]. Different components of the HFD have been studied to unravel the molecular pathways they are altering in order to understand the link between increased caloric uptake and cellular dysfunction. The most abundant saturated fatty acid present in the HFD is palmitic acid (PA) which was found to be increased in the circulation of obese and diabetic people [6][7][8]. The deleterious effects of the exposure to PA have been extensively studied in peripheral tissues; mainly in the liver, muscle, and pancreas [9]. The reported effects have been associated with insulin resistance, endoplasmic reticulum stress, mitochondrial dysfunction, and systemic inflammation [10][11][12][13]. Few studies have addressed the mechanisms for such deleterious effects, using transcriptome studies. For instance, in cultured human hepatocytes, genes linked to lipid transport, lipogenesis, lipid droplet growth, glucose, and fatty acid metabolism have been shown to be upregulated after PA exposure [14]. Similar results were obtained in primary cultures of human pancreatic islets, where genes related to glucose and lipid metabolism were upregulated after PA exposure at nontoxic dose [15]. Another group reported the transcriptomic effect of PA in a myoblast cell line model. These authors found alterations in several pathways such as interleukin, apoptosis, and insulin/PI3K signaling, among others, that are important for the ability to respond to hormones, cytokines, and metabolism [16].
However, little is known regarding PA effects in neurons. This is an important question to be resolved in view of the dramatic and chronic increase in the intake of HFD and PA in modern life that has become a risk to the healthy brain function. In the brain, the astrocytes are the main cells responsible of fatty acid oxidation, but information on the metabolic responses of neurons when exposed to a high concentration of PA is still scarce [17,18]. Although it is generally accepted that neuronal energy demands rely exclusively in glucose oxidation and that fatty acids are not largely metabolized, recent evidence points that neurons may use both glucose and fatty acids for ATP production under specific conditions [19]. In this regard, a study with hypothalamic neurons has demonstrated that they are able to sense and metabolize long-chain free fatty acids to produce ATP by mitochondrial ß-oxidation. The increased cellular levels of ATP close the K ATP channels causing neuronal depolarization, involved in the hypothalamic control of energy balance of the body [20]. Furthermore, in hippocampal neurons and in differentiated human neuroblastoma cells, PA induces a reduction of the NAD + /NADH ratio, and the activity and expression of the energy-sensing molecule, Sirtuin-1, are compromised and insulin resistance is generated [21,22]. These reported effects strongly suggest the metabolization of fatty acids by neurons under specific stress conditions, such as high concentrations of PA. Additionally, one of the brain regions with high plasticity and metabolic rate is the hippocampus. The cells in this structure are particularly vulnerable to deleterious conditions such as exposure to the components of HFD, neurodegeneration, and aging [23][24][25]. Given this, it is important to shed light on the consequences of the harmful effects of saturated fatty acids in these particular neurons.
To our knowledge, there is no reported evidence about the transcriptomic response and characterization of how the cellular program of a neuron integrates the pleiotropic effects of the exposure to PA. In the current study, we looked for the response in gene transcription after neuronal exposure to PA in non-toxic doses. By generating transcription profiles through an RNA sequencing approach, we observe the signaling pathways and neuronal responses that are consequently altered by the exposure to this saturated fatty acid.

Cell Culture and PA Treatment
Primary hippocampal neuronal cultures were prepared from Wistar rat brains obtained from 17-day-old embryos as previously reported [22]. Animals were handled with all precautions necessary to diminish their suffering consistent with the Regulations for Research in Health Matters (México) and with the approval of the local Animal Care Committee. Briefly, hippocampi were dissected, minced with a scalpel in Krebs solution (121 mM NaCl, 4.8 mM KCl, 1.2 mM KH 2 PO 4 , 25.4 mM NaHCO 3 , 14.2 mM Glucose, 0.004 mM Phenol Red), and incubated with 0.25% trypsin at 37 °C for 10 min. The hippocampi were mechanically dissociated using a cell strainer (Corning ®), and the cellular suspension was homogenized in neurobasal medium (Gibco 21103049) supplemented with 2% B27 (Gibco 17504044), 0.5 mM L-Glutamine (Gibco 25030-081) and 20 µg/mL penicillin/streptomycin (Gibco 15140-122). For Oil Red O staining, hippocampal neurons were plated at 1.97 × 10 5 cells/cm 2 density on 12 well plates; for RNA extraction, cells were plated at 1.6 × 10 5 cells/cm 2 density on 60 mm in plastic dishes; and for immunodetection, neurons were plated at 1.97 × 10 5 cells/cm 2 density on 12 well plates with glass coverslips. Every plate and coverslip were previously coated with 10 µg/mL poly-l-lysine for 24 h. Cytosine arabinoside (10 µM) was added to cultures 3 days after plating to inhibit the growth of non-neuronal cells. The astrocyte population in these cultures is near 5% as measured by immunofluorescence against the glial fibrillary acidic protein. Hippocampal neurons were used for experiments after 12 days in vitro (DIV) and were maintained at 37 °C in a humidified 5% CO 2 /95% air atmosphere. After the 12 DIV, PA (Sigma-Aldrich) or vehicle as a control condition (BSA/PBS) was added for 24 h. PA was prepared as a stock solution in ethanol and the working solution was prepared the same day of use in 10% bovine serum albumin (BSA) (Sigma-Aldrich A9647)/phosphate-buffered saline (PBS) and was incubated at 37 °C for at least 2 h before adding it to the cell cultures.

Oil Red O Staining Quantification
Lipid droplet detection was performed by Oil Red O (Sigma-Aldrich) staining. Oil Red O was prepared as a stock solution 5% in isopropanol in constant agitation overnight at 4 °C. Afterwards, the solution was filtrated through a Whatmann® Filter and a 6:4 (Oil Red O/Mili-Q water) was prepared and incubated at room temperature (RT) at least 20 min before its use. The solution was filtrated through a 0.2 µm Millex-GP ® Filter. For the Oil Red O staining, cell culture medium was removed; the cells were washed twice with PBS and were fixed immediately with PFA 4%/PBS overnight at 4 °C. Afterwards, PFA was removed, and they were washed twice with PBS and once with isopropanol 60%. Then, cells were left to dry and when they were completely dry, Oil Red O stain was added and incubated for 2 h. Finally, the Oil Red O stain was removed, and the cells were washed with bidistilled water until background staining was removed. For its quantification, pure isopropanol was added, and the Oil Red O stain was solubilized. The final solution was read in a multiplate reader at 520 nm.

RNA Extraction
Total RNA was isolated using TRIzol™ reagent (Thermo Fisher Scientific) as specified by the manufacturer and cDNA was synthesized from 200 ng of RNA using the High Capacity, cDNA Reverse Transcription Kit (Applied Biosystems/Thermo Fisher, 4374966) with random primers. The quantification of total RNA was achieved using a NanoDrop 2000 (Themo Scientific). The integrity of RNA was assessed by Agarose-Gel Electrophoresis.

RNA Sequencing
RNA was quantified using Qubit 2.0 (Invitrogen, USA) and quality was assessed using the Agilent 2100 Bioanalyzer (Agilent Technologies, USA). Only samples with an RNA Integrity Number (RIN) > 8.0 were used. Libraries were constructed using 500 ng of RNA, using the Truseq Stranded mRNA library prep kit from Illumina according to the manufacturer's instruction. The libraries were sequenced using an Illumina HiSeq2500 equipment (Illumina, Inc.) in Pair-end (2 × 125 bases). Depth of sequencing was > 25 million reads.

RNA-Seq Analysis
Adapter trimming and low-quality reads were filtered out using trimmomatic v.0.39 [26]. Reads were mapped to the rat genome with STAR v.2.7.1a [27], and also to the rat transcriptome with salmon v.0.14.1 [28], using assembly version Rnor6 with ensembl annotations version 6.0.95. Gene level counts from reads mapped to the genome were quantified using featureCounts in the rsubread [29] package. Differential expression analysis was performed with DEseq2 [30] and edgeR [31], with a paired design ~ specimen + treatment. Differentially expressed genes were defined as having FDR < 0.1 and |Log2FC|> = 0.5. Low expressed genes were filtered out before edgeR using the filterByExpr function and the design matrix. GO term and pathway overrepresentation were performed with the clusterProfiler [32] Bioconductor package on all genes found DE in at least one of the 4 workflows using all expressed genes as background. Gene set enrichment analysis [33] was performed on all expressed genes ordered by their DE − log10 p value.

Quantitative RT-PCR
It was performed using the Maxima SYBR Green/ROX qPCR Master Mix (Thermo Scientific, K0221) in a QuantStudio 3 (Applied Biosystems). All reactions were performed five or seven times, and the expression was normalized using the glyceraldehyde-3-phosphate dehydrogenase (Gapdh) mRNA. The sequences of the primers used are listed in Table 1.

Immunofluorescence and Image Analysis
After different PA treatments, the cell culture media was removed, and cells were washed three times with ice-cold PBS. Then, the cells were fixed with ice-cold PFA 1%/PBS for 5 min and washed three times with PBS. Afterwards, cells were permeabilized in 0.3% Triton X-100/PBS for 30 min at RT. Then, cells were incubated in blocking solution (BSA 4%/PBS) with gentle agitation for 1 h at RT. Next, cells were incubated with anti-MAP2 antibody (1:1000, Millipore #MAB378) in blocking solution for 48 h at 4 °C. After washing three times in 0.3% Triton X-100/PBS, cells were incubated with secondary antibody (1:1000, Alexa Fluor 488 donkey anti-mouse Invitrogen # A21202) in blocking solution for 2 h, at RT. Immediately after incubation with the secondary antibody, nuclei were stained with Hoechst

PA Exposure Changes the Transcriptional Profile Without Modifying the Neuronal Morphology
To characterize the morphological consequences and neurite integrity after 24 h of exposure to PA, we first performed a qualitative analysis of the distribution of the cytoskeletal protein MAP2 in the hippocampal neurons (Fig. 1a). As shown, the MAP2 immunodetection was located mainly in the neurites, scarcely found in the neuronal soma and we did not observe evident effects on the neuronal morphology with 100 and 200 µM PA compared to the control condition. However, neurons exposed to 300 µM of PA showed a slight increase in the concentration of MAP2 into the soma and a fragmented pattern in the localization of this protein, suggesting dendritic blebbing. This distribution of MAP2 is consistent with the induction of localized swelling and may indicate toxic effects of PA at concentrations of 300 µM and above. To estimate the neuronal lipid content after PA exposure, we stained neurons with Oil Red. Spectrophotometric quantification showed that PA caused an increase in the neuronal content of lipid droplets (Fig. 1b), demonstrating the uptake and metabolization of this saturated fatty acid by hippocampal neurons. Although the presence of glial cells in the cell cultures is scarce and only represent approximately 5%, we cannot rule out their contribution in the observed effects in the Oil Red quantification. Hence, considering this effect and previous viability assays [22], we continued the experiments with the dose of 200 µM PA. Given that the fatty acids per se can alter gene regulation in periphery cells and knowing that PA has pleiotropic effects, we then asked to what extent hippocampal neurons modify their transcriptomic profile after PA exposure. By RNAseq after 24 h of 200 μM PA, since different mapping methods can have different transcript detection sensitivities [34], we mapped sequences to both the rat's genome and transcriptome. In addition, for each mapping and gene quantification method, we used two well-established methods (DEseq2 and edgeR) for differential expression analyses, making a total of 4 different pipelines. This multiple analysis showed that the gene expression was not drastically altered (Fig. 2) and it revealed a total of 45 upregulated and 30 downregulated genes (Fig. 2a, b) in at least one pipeline. Genes such as Angptl4, Ugt8, Hmgcs2, Ccl2, and Insig1 were shared between the different pipelines (Supplementary Table 1), which suggest that these genes are strongly affected in the neurons by PA.
Clustering of differentially expressed genes (DEGs) showed that each experimental condition in a group behaves with a similar expression pattern (Fig. 2c) and suggests that the exposure to PA can cluster experimental conditions into two different groups but that between individuals, there are specific changes and alterations. Furthermore, as shown in the differential expression analysis (Fig. 2c, d), not only protein-coding genes are deregulated by this exposure but also long non-coding intergenic RNAs (LincRNAs) and small nuclear RNAs (snRNAs). All together these findings demonstrate that PA exposure is sufficient to induce changes in the transcriptional profile of neurons, suggesting that cellular mechanisms, even those involved in nuclear function, are being altered.

Changes in Neuronal Lipid Metabolism and Inflammatory Pathways Are Induced by PA
To determine the biological processes and the DEGs that are modified after the PA exposure, we performed pathway enrichment analyses. Gene Ontology (GO) enrichment analysis (Fig. 3a) showed that among the top 20 functionally enriched biological processes the chemokine-mediated signaling pathway, fatty acid and chemokine responses, cholesterol biosynthetic pathway, and insulin function were altered during this exposure, indicating that these processes are importantly involved in the neuronal response to PA.
In order to determine the relationships between enriched GO terms and DEGs, we constructed an interaction network using the clusterProfiler R package (Fig. 3c). We observe that the network forms a single-connected component, pointing that most genes are GO terms closely related and work together in similar processes. In this regard, Hmgcs2 and Insig1 are part of an important node for energy metabolism since it is known they are involved in the response to fatty acid, insulin, starvation, and cholesterol biosynthesis. The interaction network also showed that genes, such as Ccl2, involved in the cellular response to chemokines and the chemokine-mediated signaling pathway are related to lipid metabolism. These results suggest that PA might be Fig. 2 Transcriptome analysis of hippocampal neurons exposed to PA. a Number of DEG found with the RNAseq employing different pipelines. b Pie chart depicting the total number and percentage of differentially down-and upregulated genes found by RNAseq analy-sis. c Heat map showing the clustering of control group (V) vs. the PA exposed hippocampal neurons. d Volcano plot of DEG. Genes not significantly different are depicted by dark-colored dots altering the neuronal energy metabolism, particularly the lipid metabolism, and the inflammatory response through the induction of specific genes that are linked to both processes.
To further confirm and unravel some other biological processes affected by the exposure of neurons to PA, we performed the same enrichment analysis with Reactome [35] pathways (Fig. 4). This analysis confirmed the GO categories previously found not only by showing metabolism of lipids and the fatty acid cycle as important responsive pathways in neurons after PA exposure, but also revealed additional deregulated processes such as ion-channel transport and, strikingly, mitochondrial fatty acid beta-oxidation (Fig. 4a, b). The analysis of the interaction network showed that multiple genes involved in lipid metabolism are deregulated, such as Ugt8, Cyp51, Acot1, and Echs1, suggesting that the metabolism of lipids is one of the most relevant biological processes activated in response to PA (Fig. 4c). Furthermore, Echs1 and Acot1 also link the lipid metabolism with the mitochondrial beta-oxidation (Fig. 4c), which portraits them as important effectors in both processes. Overall, together GO and Reactome enrichment analyses and the interaction networks strongly indicate that PA is affecting mainly lipid utilization pathways, which is interesting and unexpected because they are believed to be not completely active in neurons.

Genes Dependent on PPAR Signaling and Lipid Metabolism Are Deregulated in Neurons
GO term and pathway enrichment have the disadvantage that only DEGs are considered; however, other relevant processes may be driven by subtle changes in multiple genes of the same biological pathway. In order to elucidate these, we performed gene set enrichment analysis (GSEA) which takes into account the whole list genes, both on KEGG [36] pathways (Fig. 5) and GO term gene sets (Supplementary Fig. 1). First, we found that after PA exposure, the IL-17, TNF, and MAPK signaling pathways are deregulated (Fig. 5), suggesting that PA triggers inflammatory components in neurons. It is also shown that the fatty acid metabolism and catabolism Fig. 3 GO term enrichment analysis reveals metabolic and inflammatory processes altered in neurons after PA exposure. a Top 20 biological processes that were found to be significantly enriched (hypergeometric test FDR < 0.1) in response to PA exposure. b Heatmap of genes contained in the GO terms, each square represents the fold change of individual genes (x-axis) and to which biological processes (y-axis) they are associated with. c Interaction network of genes and GO terms, where gray nodes represent pathways and the green to red color-scaled nodes are DE genes linking or associating them Fig. 4 Reactome pathway enrichment analysis reveals metabolic and lipid processes altered in neurons when exposed to PA. a Top Reactome pathways that were found to be significantly enriched (hypergeometric test FDR < 0.1) in response to PA exposure. b Heatmap of genes contained in the GO terms. Each square represents the fold change of individual genes (x-axis) and to which pathways (y-axis) it is associated with. c Interaction network of genes and pathways, where gray nodes represent pathways and the green to red colorscaled nodes are DE genes linking or associating them The top pathways and biological processes analyzed by KEGG (y-axis) and the number of genes involved in each process that are altered (x-axis) are depicted in the graph. The affected metabolic pathways (red rectangles), lipid processes (yellow rectangles), immunological pathways (purple rectangles), and other important biological and cellular processes (green rectangles) when neurons are exposed to PA are highlighted ( Fig. 5 and Supplementary Fig. 1), PPAR signaling pathway, the synthesis of ketone bodies (Fig. 5), cholesterol biosynthetic and metabolic processes (Supplementary Fig. 1) are likewise affected. This result indicates that neurons have lipid metabolism elements that sense and respond to a high dose of saturated fatty acids. Additionally, this analysis showed that PA can affect general cellular and biological processes like apoptosis and cellular senescence (Fig. 5), as well as specific neuronal processes like ensheathment of neurons and axons, axonogenesis and regulation of neurotransmitter levels (Supplementary Fig. 1).
In order to validate the results obtained by the RNAseq and to show that the PPAR signaling pathway is responding to the PA stimulus, we analyzed 4 genes by quantitative RT-PCR (qRT-PCR). Hmgsc2 and Angptl4 are two genes implicated in lipid metabolism and regulated by the PPAR signaling pathway. Their mRNA quantification by qRT-PCR showed that both genes (Fig. 6a, b) are significantly increased in the hippocampal neurons that were exposed to 200 μM PA compared to the control condition. On the other hand, Ugt8 and Rnf145, two other genes implicated in lipid metabolism but not regulated by the PPAR signaling pathway, were downregulated (Fig. 6c, d). Overall, these results and the presence of lipid bodies (Fig. 1b) demonstrate that when exposed to PA, neurons modify their transcriptional profile affecting the lipid metabolism through signaling pathways that are known to respond to fatty acid stimulus in periphery cells.

Discussion
The intake of HFD has been associated with the development of metabolic diseases and to the onset of neurodegenerative diseases [37][38][39][40][41]. These types of diets harbor PA as one of the main saturated fatty acids [42][43][44]. Although several mechanisms regarding its effects have been characterized in peripheral tissues [10][11][12][13], it remains poorly understood the neuronal responses activated by PA. Despite evidence that neurons are barely able to metabolize and respond directly to saturated fatty acids, new reports have shown that these cells are affected by pathological concentrations of PA [21,22]. Additionally, it is known that the PA has pleiotropic effects since it can bind to transcription factors, activate signaling pathways through membrane-bound or nuclear receptors, be used in energy metabolism or as a precursor in the synthesis of other molecules, among other described effects [45]. Here we showed several functional pathways and cellular processes that are altered in neurons exposed to a non-toxic but a high concentration of PA, particularly those involved in lipid metabolism, insulin signaling and inflammatory responses.
As shown in the GSEA of KEGG pathways analysis, signaling through IL-17 is one of the most sensitive pathways activated in neurons after PA exposure. This signaling pathway belongs to the pro-inflammatory response well characterized mainly in T-cells and macrophages [46][47][48][49], as well as in hypothalamic neurons [50]. The family of Biological replicates were performed at least five times with *p ≤ 0.005 transcription factors NF-κB has been implicated in the IL-17 signaling and in the regulation of other pathways and transcription factors such as p53, MAPK, and the Peroxisome Proliferator-Activated Receptors (PPARs) [51][52][53][54]. Interestingly, we have previously found that PA increases the acetylated form of p65 in hippocampal neurons, which might be involved in the regulation of IL-17 effects [22]. Furthermore, it has been reported that PA can also trigger inflammatory response through the Toll-Like Receptor 4 (TLR4) inducing the transcriptional activity of p65 [55,56]. In addition, p65 is also a central component of the inflammatory response of the inflammasome, NLRP3 [57]. It is known that the NLRP3 can be induced not only by p65 activation but also by external stimuli that can bind to the TLR4 and/or generate ER stress, mitochondrial dysfunction, and ROS elevation [56][57][58]. In this regard, the GSEA shows that several of these mechanisms are being impacted in hippocampal neurons, suggesting that neurons could be responding to the presence of high PA concentrations through this inflammatory component.
On the other hand, it is known that one of the characteristics of the inflammasome response is cytoplasmic swelling due to the ion influx/efflux. Additionally, MAPK is an important signaling pathway that can be regulated by inflammatory stimuli, and usually, it is activated along with these inflammatory pathways; either by a direct phosphorylation of p65 by p38 or through the activation of JNK [59,60]. Our results show that all of these pathways are altered when neurons are exposed to PA and are significantly involved in the response to this saturated fatty acid. The inflammatory response might be correlated to the dendritic bebbling we observed when neurons were exposed to cytotoxic concentrations of PA. Thus, present results provide useful evidence about the inflammatory reaction that PA may exert on neurons giving support to the hypothesis regarding the role of PA in the induction of chronic neuronal damage.
Besides the inflammatory process, we have also found metabolic pathways that respond to the PA exposure. Previously, we have reported a diminished consumption of glucose and a reduction of the NAD + /NADH ratio after neuronal exposure to PA, suggesting that this saturated fatty acid can be used as energetic fuel by the neurons [22]. Although the utilization of saturated fatty acid as energy substrates for the brain is still controversial, the results from present analysis showed that neurons can respond to PA activating diverse lipid metabolic pathways to produce energy or to synthesize several other compounds. In this regard, it is well known that the PPAR nuclear transcription factor proteins can bind and respond to different types of lipids [61]. Particularly, it has been described in HepG2 cells that PA and the monosaturated fatty acid, oleic acid, can specifically bind and activate PPARa and/or PPARg, in a time and concentration-dependent manner [62]. Furthermore, our results showed that many other metabolic pathways that respond to nutrients are also altered by PA, such as FoxO, PI3K-Akt, synthesis and degradation of ketone bodies, and cholesterol metabolism. Several research groups have reported similar effects of PA in different cell models when using RNAseq or microarrays [15,16,[63][64][65]. These groups show an alteration of genes involved in the PPAR signaling pathway, fatty acid degradation, chemokine signaling pathway, inflammation pathways, beta-oxidation, insulin signaling, among others. In fact, we have previously reported that PA induces insulin resistance in neurons, similar to the one observed in periphery cells [21]. Thus, current results strongly suggest that these signaling networks and cellular processes are impacted by PA in different cell types.
Several genes encoding enzymes involved in lipid and energy metabolism were evaluated with qRT-PCR. One of the metabolic pathways that is used for energy production under fasting conditions is the ketone body synthesis. In this pathway, the mitochondrial enzyme 3-hydroxy-3-methylglutaryl-CoA Synthase 2 (Hmgcs2) catalyzes the first reaction of ketogenesis and is activated in the presence of acetyl-CoA, mostly derived from mitochondrial beta-oxidation. The transcriptional control of this gene is mediated by fatty acids through PPAR binding [66]. Herein, we demonstrated that the transcript that encodes Hmgcs2 was upregulated, suggesting that this lipid-dependent pathway is being activated in the neurons. Another gene shown to be upregulated is Angptl4. Interestingly, this gene is also a target of PPAR and is induced under hypoxic conditions in various cell types. The encoded protein is a serum hormone directly involved in regulating lipid metabolism [67].
Interestingly, the other two genes that we measured by qRT-PCR, which are involved in lipid metabolism but not regulated by the PPAR transcription factor family, were found to be downregulated, Ugt8 [68] and Rnf145 [69]. Ugt8 is a brain-specific-key enzyme in the last step for the synthesis of galactocerebrosides [70,71]. The recently characterized Rnf145 enzyme is a E3-ubiquitin ligase involved in the homeostasis of cholesterol, negatively regulating this process when high levels of this protein are found in a cell [72,73]. The downregulation of Rnf145 transcript suggests alterations in the cholesterol pathway leading to an increase of the cholesterol biosynthetic genes expression and the novo synthesis of cholesterol [72].
Since lipid metabolism remains controversial in neurons, the genes we have validated are mainly involved in this process. Nevertheless, previously, we have reported the activation of p65, a well-characterized transcription factor involved in inflammatory responses in hippocampal neurons exposed to PA [22] as well as the effects of this saturated fatty acid on the insulin signaling [21]. Other research groups have also reported the induction of inflammatory mediators after neuronal exposure to PA [74]. Furthermore, transcriptional expression levels of different genes involved in PA effects on neurons showed a significant impact on different neuronal processes associated with apoptosis, senescence, and autophagy that are known to participate in the pathophysiology of neurodegenerative diseases, such as Alzheimer's Disease.
Deregulation of synaptic function by PA exposure is a very interesting and current research topic in the field, since it can help unravel some mechanisms by which neurodegenerative diseases can be triggered by a HFD consumption. Interestingly, there are some research groups that have reported associated effects of HFD and PA to synaptic dysfunction through different mechanisms, mainly involving protein palmitoylation and pro-inflammatory mechanisms [74][75][76][77]. Few studies, such as the one published by Roy et al. (2016), identify lipidbinding transcription factors that are capable of modulating synaptic function [78]. In our study, GSEA of KEGG (Fig. 5) and GO terms (Supp. Figure 1) show that processes like longterm potentiation, a proposed mechanism underlying learning or memory, can be affected when hippocampal neurons are exposed to PA. Therefore, this first approach using a global analysis of gene expression can help to understand the pleiotropic effects of PA on neurons and the possible pathways activated under pathological conditions which may be involved in the deleterious effects of the consumption of saturated fatty acids.
Author Contribution MFL designed and performed experiments, analyzed data, prepared the figures, and drafted the manuscript; NA performed the bioinformatical data analyses, prepared figures, and reviewed the manuscript; MPD prepared the neuronal cultures and assisted with the confocal analysis; KTA assisted with the qRT-PCR experiments and prepared the figures; RRV prepared the sequencing libraries and performed the RNAseq; IARV performed the RNAseq and reviewed the manuscript; CAC contributed valuably to the discussion, writing and approval of manuscript; LAH contributed to the revision of the manuscript; CA and RGB funded, designed, and oversaw the whole project including experimental design, data analysis, drafting and reviewing the manuscript.