Transcriptomic Analysis of the Effects of a Fish Oil Enriched Diet on Murine Brains

The health benefits of fish oil enriched with high omega-3 polyunsaturated fatty acids (n-3 PUFA) are widely documented. Fish oil as dietary supplements, however, show moderate clinical efficacy, highlighting an immediate scope of systematic in vitro feedback. Our transcriptomic study was designed to investigate the genomic shift of murine brains fed on fish oil enriched diets. A customized fish oil enriched diet (FD) and standard lab diet (SD) were separately administered to two randomly chosen populations of C57BL/6J mice from their weaning age until late adolescence. Statistical analysis mined 1,142 genes of interest (GOI) differentially altered in the hemibrains collected from the FD- and SD-fed mice at the age of five months. The majority of identified GOI (∼40%) encodes proteins located in the plasma membrane, suggesting that fish oil primarily facilitated the membrane-oriented biofunctions. FD potentially augmented the nervous system's development and functions by selectively stimulating the Src-mediated calcium-induced growth cascade and the downstream PI3K-AKT-PKC pathways. FD reduced the amyloidal burden, attenuated oxidative stress, and assisted in somatostatin activation—the signatures of attenuation of Alzheimer's disease, Parkinson's disease, and affective disorder. FD induced elevation of FKBP5 and suppression of BDNF, which are often linked with the improvement of anxiety disorder, depression, and post-traumatic stress disorder. Hence we anticipate efficacy of FD in treating illnesses such as depression that are typically triggered by the hypoactivities of dopaminergic, adrenergic, cholinergic, and GABAergic networks. Contrastingly, FD's efficacy could be compromised in treating illnesses such as bipolar disorder and schizophrenia, which are triggered by hyperactivities of the same set of neuromodulators. A more comprehensive investigation is recommended to elucidate the implications of fish oil on disease pathomechanisms, and the result-driven repositioning of fish oil utilization may revitalize its therapeutic efficacy.

Despite of the wide implications of n-3 PUFA on the brain functions, clinical efforts investigating the efficacy of n-3 PUFAs and in that matter, the fish oil as dietary supplements for the treatment of psychiatric diseases have yet met with mixed results. Two separate clinical trials on Alzheimer's patients and Parkinson's patients supplemented by n-3 PUFA enriched diets failed to deliver a robust outcome. The trials found n-3 PUFA effective only if it was administered during the early onset of the AD [10]; in treating PD, its efficacy was restricted to mitigating the depression syndromes only [12]. N-3 PUFAs demonstrated significant positive impact on the major depressive episodes with anxiety co-morbidity [6,13]; however, the supplement's efficacy on the heterogeneous disease profile (with and without comorbid anxiety) was marginal [13]. A pilot trial offering n-3 PUFA supplementation to patients immediately after life-threatening incidents resulted in reduced post traumatic syndromes [19].
A number of clinical, epidemiological, and laboratory studies linked the health benefits of n-3 PUFA to various dietary compositions, including EPA alone [20,21], the EPA:DHA ratio [22], the arachidonic acid (AA):EPA ratio [23,24] the AA: alphalinolenic acid (ALA) ratio [25] and the overall n-3 PUFA; n-6 PUFA ratio [26]. Consequently, the appropriate dietary composition is still a subject of ongoing investigation.
Researchers have been using rodent models for decades to evaluate the efficacy of fish oil as the dietary supplement. Justifying the model, studies reported the increased levels of EPA and DHA accompanied by decreased AA in the brain tissues of rats fed on fish oil enriched diets [27,28]. The n-3 PUFA enriched diet helped balancing behavioral plasticity [29], improved cognitive function [9] and working memory [30], enhanced neuroprotection [31,32] and reduced anxiety and depression-like traits [33,34] in vitro. High throughput genomic studies evaluating the effects of fish oil on the rodent brain identified genes involved in ion channels, neuronal function, signal transduction, synaptic plasticity, cytoskeleton and membrane association, and energy metabolism [35][36][37][38][39][40][41].
The comprehension of the underlying molecular mechanism is potentially critical to understand the effect of the fish oil on neuronal activities. Toward this goal, we studied the effects of high dietary intake of fish oil on the genomic regulations in the mouse brain, and characterized the potentially associated molecular events focused on the nervous system and neurofunctions. The effects of FD on the genes relevant to neurogenesis are of particular interest in the context of the recent study manifesting n-3PUFA as the mediators in generating new neuronal cells [42].

Ethics statement
Research was conducted in compliance with the Animal Welfare Act, and other Federal statutes and regulations relating to animals and experiments involving animals and adheres to principles stated in the Guide for the Care and Use of Laboratory Animals (NRC 2011) in facilities that are fully accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care, International. The protocol was approved by the IACUC committee of Walter Reed Army Institute of Research, Silver Spring, MD.

Diet types and mouse handling
The percentage of the fish oil in the customized FD was restricted to 16% to maintain the food pellet structure consistent with that of SD (Certified Rodent Diet 5002*, Purina LabDiet, MA). Thus, the external appearance and texture of FD and SD appeared similar. Solid food pellets helped maintaining the tidiness of the cages; hence, the cages occupied by FD-or SD-fed mice followed the same schedule of regular lab handling, precluding any concerns associated with psychological bias due to the handling with variable frequency [43]. Possible bias due to the rodents' potential favoring of particular food textures was further averted by introducing the two diets with similar textures.   Increased serum MMP-9 during depression in young patients [125] AD AD patients (N = 38); and controls (N = 34) Higher MMP-9/TIMP-1 ratios and lower TIMP-1 levels compared to cognitively healthy individuals. [126] AD/PD AD patients (N = 30); PD patients (N = 24); and controls (N = 32) Elevated MMP-9 in the plasma of AD patients; but no significant changes of MMP-9 levels in PD cohorts. [127] Schizophrenia Schizophrenic patients (N = 442); and controls (N = 558) Significant preponderance of 21562 C:T polymorphism in patients [128] SOD-1 AD Post-mortem brain samples from AD patient, N = 4; and patients diagnosed with Huntington's Disease, N = 4 Decreased SOD-1 in AD patients' hippocampus [129] Down syndrome (DS) and AD Post-mortem brain samples; DS, N = 9, AD, N = 9; and controls, N = 9 Increased SOD-1 in DS temporal, parietal, and occipital cortex, whereas decreased SOD-1 in the AD temporal cortex [130] BDNF PTSD All women cohort, diagnosed with PTSD, N = 17; with PTSD+ childhood physical neglect, N = 17; controls, N = 15 Lower plasma BDNF in association with PTSD, and more strongly linked with childhood neglect driven PTSD [131] PTSD Patients with history of trauma, N = 34; and controls, N = 34 Higher level of serum BDNF levels right after the traumatic events decreasing over time [84] Bipolar disorder UK Caucasian subjects, BD, N = 962; and control, N = 2100 No overall influence from the Val66Met polymorphism of BDNF, with a moderate association with the susceptibility to the rapid-cycling subset of the disorder. [132] Major depression Korean subjects, major depression, N = 310 and controls, N = 209 Significant association of BDNF with the disease onset among the younger subgroup only [133] PD PD patients, N = 453 and controls, N = 291 Increased PD risk in association with the pesticide exposure [134] AD AD patients with rapid cognitive decline, N = 12; and slow cognitive decline, N = 28 Decreased BDNF serum levels in AD patients with fast cognitive decline [135] FKBP5 PTSD Nonpsychiatric clinic patients with significant levels of childhood abuse and non-child abuse trauma, N = 900 High interaction between four SNPs of the FKBP5 gene (rs9296158, rs3800373, rs1360780, and rs9470080) with severity of child abuse as a predictor of adult PTSD [136] PTSD Subjects linkage and association studies of the genetics of cocaine, opioid, and alcohol dependence, and controls; N = 1146 European Americans (EAs) and 1284 African Americans (AAs) AA subjects: Significant association between one SNP (rs9470080) with childhood abuse eliciting PTSD. EA subjects: Significant association of PTSD risk with alcohol dependence and childhood adverse experiences interacting with FKBP5 polymorphisms [137] Anxiety/depression Patients, newly diagnosed with advanced gastric cancer and supposed to receive first-line chemotherapy, N = 130; and control similar patients after two cycles of chemotherapy, N = 93 Significant association of anxiety with FKBP5 rs9296158, and marginal association with rs9470080 and rs1360780. Marginal association of depression with FKBP5 rs9470080 and rs9296158 [138] Major depression Patients diagnosed with major depression, N = 68; and controls, N = 87 Significant interaction between disease status and FKBP5 risk allele carrier status (minor allele T). [139] Past clinical reports (demography of the cohorts and major findings) associated with five genes (NOS-1, MMP-9, SOD-1, BDNF and FKBP5) are reported. doi:10.1371/journal.pone.0090425.t002 Briefly, these two diets were similarly supplemented by proteins (FD: 18.3%, SD: 20.7%, both by weight), carbohydrates (FD: 53.8%, SD: 51.8%, by weight) and crude fibers (FD: 4%, SD: 4.3%, by weight); vitamins and minerals were also supplemented proportionately including vitamin E (FD: 90 IU/kg; SD: 65 IU/ kg). The different supplements of fat (FD: 16%; SD: 5.5%, both by weight), primarily the higher n-3 PUFA supplements in FD (FD: 4.70%; SD: 0.20%, by weight) outlined the major difference between FD and SD. The n-6 PUFA supplements were marginally lower in FD (FD: 0.66%; SD: 1.22%, by weight). Overall caloric contributions from the two diets were comparable (FD: 4.2 Kcal/ g; SD: 4.07 Kcal/g) ( Table 1). To maintain the food quality across the study duration and to minimize the oxidative damage to n-3 PUFA, freshly prepared food was purchased every two months and stored at 4uC in sealed bags and the food chambers were replenished daily with fresh supplies.
In designing the assay, we refrained from adjusting any potential shortcomings of the fish oil. For instance FD diet contained 0.3% linoleic acid (LA), whose scale of representation in rodent diets has long been a subject of investigation [44,45]; of note the nearly equi-caloric control diet contained 1.22% LA (Table 1), as per the recommendation [44]. Hereby, our result highlighted the diet-induced molecular events potentially explaining the suggested benefits or the adverse effects of fish oil on the brain.
C57BL/6J male mice were purchased from Jackson Laboratory (Bar Harbor, ME) at the age of three weeks, singly housed and immediately introduced to either SD or FD. Mice had free access to their respective diets and standard lab-supplied liquids for next five months (,20 weeks).

Collection of hemibrains and RNA isolation
Following the euthanization of the mice by cervical dislocation the brains were removed from their crania. After the mid-sagittal section, the hemibrains were collected by a highly trained professional experienced with this procedure, who was kept blinded from the animal identities during the dissection period. The dissected brain sections were snap-frozen, saved in individually labeled tubes, and transferred to the freezer at the end of the day. The entire process starting with the brain removal from the skull to the snap-freezing of the last brain region took less than 20 minutes as described in our earlier report [46].
On the day of nucleic acid isolation, organs were weighed before thawing and the appropriate amounts of Trizol TM (Invitrogen, NY) were added. RNA was isolated from the Trizol TM suspensions as per the manufacturer's guidelines. Briefly, the hemibrain was homogenized using the TissueLyser system (Qiagen, MD), phase separated with vigorous agitation, and the aqueous phase was collected. Serial alcohol-based precipitations and washes were performed on the aqueous phase to obtain RNA. The isolated nucleic acids were quantified and qualified using the Agilent BioAnalyzer as reported earlier [47].

High throughput transcriptomic expression analysis
The dual dye microarray was carried out using the Whole Mouse Genome Microarray Kit (Agilent Technologies, Inc.) following the vendor's protocol. 200-2000 ng of purified RNA was labeled with Cy-5 dyes and the reference RNA (Agilent, CA) with Cy-3 dyes (N = 4 from FD-fed mice and N = 3 from SD-fed mice. One SD sample was misplaced). The samples were simultaneously hybridized to Agilent 4644k slides (platform number: 14868) and incubated for 17 h at 55uC. After overnight hybridization, slides were processed in a series of washes. The slides were scanned using an Agilent DNA microarray scanner and the features were extracted using the default setting of the Feature Extraction software (Feature Extraction software v.10.7, Agilent, CA).

Data filtering and process normalization
GeneSpring v.10.1 (Agilent Technologies, Inc., CA) was used to carry out the preliminary data filtration and statistical analysis. Each chip was subjected to intra-chip normalization using the locally weighted scatter-plot smoothing method (LOWESS) [48]. Genes were identified that were differently expressed between the hemibrains of mice fed on FD vs. SD based on 61.5 fold change cut-off with FDR,0.05.
The microarray data was submitted to the Gene Expression Omnibus (GEO). This can be searched using the Platform ID: GPL7202, Series: GSE46933.

Cluster and functional analysis
GeneSpring v.10.1 (Agilent Technologies, Inc., CA) was used to perform two-dimensional hierarchal clustering using Pearson correlation algorithm ( Figure 1).

Real-time PCR validation of microarray results
The mRNA microarray results were validated using two platforms. The high throughput real-time PCR was carried out using the RT 2 Profiler PCR Array System (SABiosciences, MD) on the ABI 7900HT platform (Life Technologies, CA). The PCR was carried out using three biological replicates for each diet. SDS 2.4 software supplied by ABI, CA, was used to compute the changes of threshold cycles (Ct) (i.e., 2 ' (2D Ct), where D Ct = Ct (GOI)2mean (Ct (HKG)); GOI denotes the gene of interest, and HKG denotes the housekeeping gene. In order to eliminate the false positives, the mRNA reads were screened based on two criteria: (i) the controls' threshold cycles were .30, and samples' threshold cycles were ,30 (or vice versa), and (ii) the p-values for the fold-changes were either unavailable or relatively high (p.0.05) from the assay backgrounds.
In addition, qPCR of five genes (NOS1, FKBP5, MMP9, SOD1 and BDNF) was carried out using primers purchased from Fisher Scientific, Inc, PA using their Solaris gene expression assay (Table  S1). The selection of these genes is justified in a brief literature survey ( Table 2). Real-time PCR was performed in the HT700 platform (Applied BioSystems, CA) using five technical replicates. The downstream analysis was conducted using the GAPDH (glyceraldehyde-3-phosphate dehydrogenase) and beta-actin housekeeping genes following the protocol described elsewhere [47].

Results
The diet compositions listed in Table 1 revealed relatively comparable supplements of major nutrients and consumable calories derived from FD and SD. The major differences between the two diet types were attributed to the n-3 PUFA and n-6 PUFA supplements ( Table 1).
The mice weighing 10-12 grams at the age of three weeks had a 61.2% increase in body weight as a result of consuming SD for five months; thus the average weight became 30.96 grams (% CV = 6.12). During the same five month period, FD-fed mice showed 67.2% increase in body weight; the average weight became 36.6 grams (% CV = 13.17). It was a near significant difference in the change of body weights between the FD-vs. SDfed mice, with the FD-fed group gaining 18% more in body mass (p = 0.06).
Transcriptomic analysis identified 1,142 genes altered between the mice fed on FD vs. SD (6.6% of total genes). The GOI demonstrated 61.5 fold changes with FDR,0.05. Figure 1 is a cluster view of these GOI generated using the Pearson correlation algorithm; 659 (58% of GOI) and 483 (42% of GOI) transcripts were elevated and suppressed by FD, respectively. One hundred and fifty two unknown genes with the remaining 990 known genes are documented in Table S2. Figure S1 shows a volcano plot of the same gene set. The entire list is available in the public domain: Gene Expression Omnibus (GEO) and this can be using the Platform ID: GPL7202, Series: GSE46933.
Meeting the objectives of the present study, subsequent convergent functional analysis identified the transcripts linked to the neurogenesis, neuroviability and neurodegenerative disorders.

Biofunction analysis
The 1,142 GOI enriched an ensemble of biofunctions. Table 3 reports the top-ranked biofunctions (p,0.0001) that are enriched by at least 50 genes. The complete list of biofunctions is shown in the Table S3. We identified four biofunctions pertinent to the present objective: ''Neurological disease'', ''Nervous system development and function'', ''Inflammatory response'' and ''Cell death'', and these were further classified into subcategories enriched by at least 30 genes. Of note, we found certain enriched pathways, such as ''LPS/IL-1 Mediated Inhibition of RXR Function'', ''Amyotrophic Lateral Sclerosis Signaling'' and ''Hepatic Fibrosis/Hepatic Stellate Cell Activation''; however, we failed to derive biologically meaningful information from these pathways. Therefore, we focused on exploring the four biofunctional modules identified herein. Together, these four biofunctions enlisted 247 genes (153 up-and 94 down-regulated by FD). The majority of the genes (99 genes, i.e. 40% of the 247 genes) encode the proteins that are located in the plasma membrane. Curating from the Allen Brain Atlas (mouse.brain-map.org) we linked the 20% of these 247 genes to possible brain regions (Table S4).
3.1.i. Neurological diseases. Two hundred and sixty genes (22.7% of GOI) were associated with the ''neurological diseases'' with a range of p values from 1.01610 23 to 5.96610 25 (Table 3). In particular, we focused on three diseases: (i) movement disorder (MvD) (93 genes, p = 8.13610 24 ) ( Tables 3 and 4). The movement disorder was the most enriched candidate on the list. We also selected AD and PD for further discussion in light of the suggested implications of n-3 PUFAs on these particular illnesses [9,10,12,51]. Together, these three diseases explained approximately 34% of the transcripts linked to ''neurological diseases'', with 16 common genes among three disease types of interest. In addition, 3 transcripts were shared by MvD and AD, and 14 transcripts by MvD and PD.
FD elevated the expression of a substantial subset of transcripts associated with MvD (76%), AD (58%) and PD (74%). Proteins associated with 45 transcripts representing 48% of MvD associated genes are located in membranes; likewise, 42% (29 transcripts) and 54% (27 transcripts) of the proteins related to AD and PD, respectively, are located in the membranes.
A comprehensive FD-induced elevation was observed among the transcripts associated with the calcium and potassium channels (CACNA1A, CACNA1D, KCNMB3, KCNQ5, KCN1P1, KCNC4, KCNC3, KCNA5, KCNJ6, KCTD7 and KCNQ2) except one, namely CACNA1S. All transcriptomic members of solute carrier family (SLC18A3, SLC5A7 and SLC6A4), and dopamine, glutamate, adrenergic and cholinergic receptors (DRD5, GRID2, GRIN2A, GRM2, GRM7, ADRAID, ADRB2, CHRM3 and CHRNB3) were also elevated. Alternatively, FD reduced the expression of the transcripts associated with the cytokines (IL1B, IL10, PRL and TNF) and early growth response agents (EGR1 and EGR2). The members of zinc finger family (ZNF385B and ZNF423) displayed over expression, but those of the teashirt zinc finger family (TSHZ1 and TSHZ2) displayed suppressed expression in the FD-fed mice. Transcripts encoding the G-protein coupled receptor (GPR6) were elevated, but that encoding the ionotropic GABA receptors were differentially expressed. FD inhibited the transcripts associated with the domain of alpha 5 (GABRA5) and elevated alpha 4 (GABRA4), respectively. Although GABRG3 and GABRA5 are phenotypically juxtaposed in human chromosome, FD elevated the former and suppressed the later transcript.
DoN and GoN shared five transcripts: BMP6, CDK5R2, CNR1, NTRK3, and ATOH7; all of them were elevated by FD. In addition, there were four transcripts shared by all five subcategories; NGFR was elevated, and FGF2, HGF and MAPK8 were all suppressed by FD.
DoN, GoN and SyT enlisted 80%, 79% and 88% elevated genes, respectively. In DoN, the proteins associated with these transcripts are almost equally distributed among the extracellular space (8 genes), plasma membrane (7 genes), cytoplasm (6 genes) and nucleus (8 genes). On the other hand, proteins encoded by 48% and 71% of the transcripts associated with GoN and SyT, respectively, are localized in the plasma membrane.
FD-elevated transcripts encoding proteins are associated with a wide ensemble of transmembrane receptors and transportation functions (SOX10, CHRNB3, GFRA2, NGFR, APBA1, GJB1, SLC18A3, SLC5A7, SLC6A4 and SLC8A3) and membrane-based G-protein coupled reception (ADBR2, CCR5, CHRM3, CNR1, DRD5, GPR6, GRM2, GRM7, HCRTR1 and NTSR1). FD further increased the expression of most of the genes regulating iii. Apoptosis and immune response. Apoptosis and immune response were the two most enriched sub-categories of ''cell death'' and ''inflammatory response'' respectively, which explained 48% and 95% of the transcripts enlisted by the respective parent terms. Together, they shared 34 genes; 23 of these were suppressed (Table S5).Most of the proteins encoded by these transcripts associated with apoptosis (30%) and immune response (35%) are localized in the plasma membrane.
The pool of cytokine associated transcripts (CSF2, CXCL11, IK, IL10, IL19, IL1B, PRL, TNF and TNFSF14) was suppressed in FDfed mice excluding IL25elevated. Likewise, a large panel of CD markers excluding CD4 was suppressed by FD. The list included CD2, CD3G, CD80, CD28 and CD7, which are associated with Tcells; CD5 and CD48 are associated with B-cells; CD244 is associated with both NK-cells and T-cells; and CD226 is associated with platelets as well as T-cells. Alternatively, all the transcripts associated with transportation (GJA4, LBP and SFTPA1) were elevated. Among the growth factors, BMP6 and PGF transcripts were elevated; whereas FGF2 and HGF were suppressed. All of the kinase-transcripts encoding protein localized in the plasma membrane were elevated (ALK, KIT and NTRK3), while the rest (CHEK1, MAPK8, PDPK1, PRKX and ROCK1) were suppressed by FD.

Validation of gene expression by the real-time PCR analysis
The elevated expressions of DRD3, GABRR1, CCKBR and GABRA3 encoding proteins localized in the membrane were validated by PCR array (Figure 4). The suppressed genes validated by qPCR included MAOA, GAD1, HPRT1, NPY1R, CHRNA1, CHRNB1, GABRD and GABRG2. The assay could not confirm the expressions of CHRNB2, CHRNE, GABRP and GABRB2.
By using the low throughput qPCR assay ( Figure 5), we confirmed that FKBP5 and NOS-1 were elevated by FD while BDNF, MMP9 and SOD1 were suppressed.
Altogether, the qPCR assay validated 76% of the genomic expressions tested herein.

Discussion
A multitude of in vitro results suggests potential benefits of n-3 PUFA enriched diets typically supplemented by fish oil, but translating the knowledge to a viable therapeutic model remains challenging [10,12,13,[52][53][54]. There is a necessity for a more systematic comprehension of the etiologic relationships between the fish oil and their reported effects on various illnesses.
In light of the several studies evaluating the range of the compositions of n-3/n-6 PUFA [21][22][23][24][25]55,56], our study focused on investigating one particular dietary supplement, namely the fish oil enriched with high n-3 PUFAs, EPA and DHA in particular. This high fat composition with a consequence of increased body mass is certainly a concern [57] that should be addressed by a more comprehensive investigation in the future. Furthermore, we investigated the effects on the hemibrains only; dietary effects on various brain sub-regions were not assessed as reported by others [40]. Present study also did not attempt to quantify the shift of fatty acid levels in the brain as a result of consuming a high fat diet that investigated in past [58]. The genomic profiling of the hemibarns could be further biased by the collection of the small midline structures such as thalamic and hypothalamic nuclei.
It is important to note that the present study solicited the molecular impacts of the fish oil, not the consequences of n-3 PUFA. The study was designed to critically comprehend the performances of the fish oil including its benefits and adverse effects. Therefore FD delivered more fish oil derived fatty acids (high n-3 PUFA; but sub-optimal n-6 PUFA) than the controls following many such precedents [42,[59][60][61][62][63]. SD with nearly equivalent caloric contribution had unadjusted fat supplements like many other studies [42,[59][60][61][62][63]. Adjusting the fatty acid presentation in the diet potentially derived from other sources, such as soybeans might alter the molecular makeups of the rodents risking confounding the outcome.
The FD contained 5.2% of n-3 PUFA and 0.65% of n-6 PUFA, presenting a ratio of 8:1 of n-3 PUFA: n-6 PUFA; the SD contained 0.2% and 1.22% of n-3 and n-6 PUFA, respectively, with a ratio of 1:6. By mass, FD consisted of 16% fish oil, and the total fat composition in FD was 1.5 fold higher than SD. FD contained twenty times more n-3 PUFA and 50% less n-6 PUFA as present in SD. Furthermore, FD contained a 1.5:1 ratio of the EPA: DHA and a 7:1 ratio of EPA: AA, which was in agreement with the literature showing health benefits [23][24][25][26]56,64]. By weight, other major diet components including protein, carbohydrate, vitamins and minerals were similar in FD and SD, and the gross energy levels were also equivalent ( Table 1). FD and SD were introduced to randomly chosen mouse groups at three weeks of age. The animals consumed their respective diets for another five months, which is equivalent to their late adolescence. Subsequently, the mice were euthanized to collect the brain samples for transcriptomic analysis.
The two mouse populations gained body weight differently. FD mice gained 18% more weight with a marginal significance (p = 0.06) in accordance with a previous study [65]. This could possibly be a negative effect of the high lipid diet [57]; further optimization of the dietary components is required to control the weight gain and consequent health issues. It is, however, beyond the scope of the present study to discuss the potential markers associated with obesity, such as the genes associated with fatty acids (SLC27a4, FADS2, FA2H, and FADS6; all up-regulated by FD) and insulin treatment (INSIG1 up-regulated and IRS1, IGFBP2; all down-regulated by FD) (Table S2).

Fish oil mediated augmentation of the brain and nervous system
The functional consequences of synaptic plasticity include the power of retention and retrieval of memory; and shifting of learning and behavioral paradigms [66,67]. Environments combined with diets can critically modify the synaptic networks that involve formations, differentiation, repositioning, and elimination of synapse.
The NMDA receptors (NMDARs) are the critical regulators of synaptic plasticity and particularly of the long term consequences such as memory retention [43,66,68,69]. Src-mediated direct phosphorylation of NMDARs initiates the presynaptic influx of Ca +2 [68]. FD apparently activated this presynaptic pathway by elevating the key transcriptomic members, including nSHC (also known as SHC3), NMDAR2A (also known as GRIN2A) and CACNA1A. Promoting the excitatory against the inhibitory synaptic transmissions [70], CACNA1A contributes to a number of neurological disorders [71]. . Unlike CACNA1A, which closely regulates the nervous system, CACNA1S suppressed by FD is associated with photoreception and thermoregulation [72,73].
NMDAR-mediated influx of Ca +2 contributes to the long term potentiation, thereby controlling the behavioral learning, fear response and extinction [66,69,74]. Related observation of elevated mGluR2 (GRM2) in FD-fed mice was rather counterintuitive. mGluR2 in association with moderate Ca +2 influx is linked to long term depression [75]. Addressing the paradox, literatures suggest that the temporal pattern of Ca +2 deposits may shift the consequence from depression to potentiation, and differential buffering of intracellular Ca +2 may also shift the balance [68]. Although we could not rule out the possibility of FD-induced long term depression, our results described hereafter increasingly favored the possibility that FD preferentially promoted long term potentiation.
Supporting the hypothesis, FD enhanced glutamatergic activity ( elevated DRD-2/-3/-5; GRM-2/-7; and GRI-D2/-N2a in the FDfed mice) driving the calcium and dopamine messengers to phosphorylate the protein phosphatase 1 regulatory subunit 1B [69]. Elevated PPP1R1B, also observed by administering DHAenriched diet [40] is critically related to the emotional learning associated with long term potentiation [76]. In addition, FD elevated a group of transcriptional members associated with choline influx, such as CHRM3, CHRNB3 and SLC5A, which typically promote long term potentiation [77]. FD induced elevation of PDPK1 and AKT (alternate name: PKB), two integral downstream members of PI3K pathways, is likely to cause spine    [78]. Together, the accumulated evidence from this study increasingly linked FD to long term potentiation, as opposed to long term depression.
Emerging knowledge elucidating the complexity of BDNF activity regulation begins to dissociate BDNF from the robust synapse as suggested in past [79][80][81]. Researchers have found that the TRKB-independent pro-BDNF signaling pathway is the causative factor of compromised synapse [82], and potentially eliciting depression [83]. In comparison to the remote PTSD patients, recent PTSD patients have been diagnosed with higher levels of BDNF that likely contribute to the biased consolidation of fear memory [84]. Evidently, the memory consolidation process is exon-specific. Prolonged stress applied on the rodents significantly elevated BDNF mRNA levels in exon I (35%), IV (20%) and IX (16%), and substantially elevated mRNA of exon II (20%) [85]. In this context, our microarray data predicted suppressed exon II of BDNF (NM_007540, vendor-delivered information, Table S1) in FD-fed mice, and the qPCR assay ( Figure 5) validated the outcome.
Conversion of pro-BDNF to mature BDNF is mediated by MMP-9, [86,87], which was transcriptionally suppressed by FD. Supporting evidence reported the n-3 PUFA-induced inhibition of MMP-9 in brain cells challenged by an endotoxic assault in vitro [87].
Hence, we postulate that FD promotes axonal growth by steering a BDNF-independent network critically regulated by Src to activate PI3K-AKT-PKC downstream. In fact, the independent actions of Src and BDNF have been reported in the past as a response to a dietary imbalance caused by zinc deficiency [88].

Fish oil mediated neuroprotection
Consistent with a past study [40], here we also reported a host of elevated genes in FD-fed mice associated with the myelin accumulation and preservation, including MOG, MPZ and MAG. Demyelinated lesions in neuronal cells are early indicators of inflammation that lead to the rapid destruction of the myelin sheath, causing increased vulnerability of the axons [89]. In fact, selective loss of MOG was identified as a serologic marker of inflammatory brain disease [90].
Pro-inflammatory cytokines such as TNF-a, IL-1B, IL-10 and TNFSF1 showed decreased transcriptomic expression, while the transcription of interferon gamma antagonist (IK) was elevated by FD in accordance to many past studies [91][92][93][94]. A study conducted on healthy young human adults identified the lower levels of n-3 PUFA (or a higher n-6: n-3 ratio) in serum as the primary cause of the larger influx of IFN-c and TNF-a [6,95]. Of particular note, we observed elevated RELA in FD mice; a rather counterintuitive observation, particularly in light of the suppressed expression of related genes like TNF, and the relevant past studies [96]. One possible reason could be the near obesity condition of the FD-fed mice; a high fat diet allegedly increases NF-kB activities in rodents [97].
The immunological implications of fish oil were further supported by the comprehensive FD induced suppression of CD markers associated with T-cells, B-cells, NK-cells and platelets. Emerging knowledge suggests that there are inhibitory effects of n-3 PUFA on CD cells, the potential markers of depression [98].
Down-regulation of apoptotic pathways and promotion of cell survival are known implications of n-3 PUFA enriched diets. Elevated AKT and BCL21A in FD-fed mice are the established markers of cell survival [99] [100]. The suppressed genes encoding the caspase family (Casp -1/-4/-7/-14) may suggested the diet's potential protective roles against inflammation by regulating the caspase family members [101][102][103]

Potential therapeutic efficacy of FD treating neurodegenarative disorders
The motivation of testing fish oil as the dietary supplements could be derived from the ongoing clinical trial activities evaluating the agonists of GABRG-3, GABRA-4, DRD-5, and THRB (all were elevated by FD), and the antagonists of GABRA-5 (suppressed by FD) for treating depression (www.clinicaltrials.gov). On the other hand, the antagonists targeting GABRG-3, GABRG-4, DRD5, ADR -A1D/-B2, CHRM-3, and SLC6A4 (all elevated by FD) are under investigation for treating bipolar disorder and schizophrenia. In contrast, FD-regulated transcripts could be relevant in eliciting the risk of the schizophrenia, bipolar disorder, and suicidal inclination. FD, for instance activated ADR-A1D/-B2 and SLC6A4; and the activated adrenergic and serotoninergic pathways are linked to augment the suicidal psychopathology [104]. In addition, several of the GABA family members (GABRG-3, GABRG-4) and key dopamine receptors (DRD5) were also elevated by FD. The activities of GABA families and dopaminergic hyperactivity are positively correlated with the prevalence of bipolar disorder and schizophrenia [105] [106]. Contrastingly, dopaminergic hypoactivity has been recognized as a possible precursor of depression [107]. This observation emphasizes the fact that the efficacy of fish oil potentially depends on the disease pathophysiology.
Affective disorder is typically triggered by the decreased activities of antioxidants such as glutathione and the depletion of serological markers such as BDNF and somatostatin [108]. Our report showed FD-induced elevation of a number of transcripts related to transmembrane glutathione transfer (Gstm -4/-6, and Gstt-1) (Table S2) accompanied by CORT encoding cortistatin, and thereby likely activated of the somatostatin receptors. Among other neurodegenerative diseases, the reported impacts of FD on AD and PD were in accordance with a multitude of existing evidence [9,10,12,31]. FD suppressed APBA1, a key gene encoding the amyloid beta (A4) precursor protein-binding family; thereby potentially suppressed the amyloidal toxicity, a vital precursor of AD progression [109]. Possible activation of the dopaminergic cascade, evident from the elevated receptors (DRD-2/-3/-5), indicated a stronger efficacy of FD in treating PD.
FD mice showed a comprehensive elevation of the transcripts associated with voltage-gated potassium channels (KCNMB3, KCNQ5, KCN1P1, KCNC4, KCNC3, KCNA5, KCNJ6, KCTD7 and KCNQ2) and cholinergic modulation (CHRM3 and CHRNB3), which are associated with memory consolidation and stress management [113][114][115]. Among another psychological illnesses, FD showed positive implications in PTSD, regulating some of the potential therapeutic markers like BDNF, FKBP5 and SLC6A4 [116] (Table 2). In addition, Table 2 shows a list of psychological debilities associated with a panel of transcripts investigated by qPCR.

Conclusions
The biomechanisms exhibited by the fish oil that facilitate neurotrophic regulation are still elusive. The knowledge gap limits our predictive aptitude in realistically estimating the efficacy of this dietary supplement. To address this challenge, we customized a diet enriched of 16% fish oil by weight, the optimum fraction of oil added to the rodent food without altering its texture from the standard lab diet. Two randomly chosen groups of C57BL/6J mice consumed one of the two diets from weaning age until their young adulthood. We sacrificed five month old mice and investigated their hemibrains using a genome-wide transcriptomic platform.
Functional analyses of the transcriptomic data focused on the areas of the neurogenesis, neuroviability and neurodegenerative disorders. The GOI were clustered into four most significant functions: ''Neurological disease'', ''Nervous system development and function'', ''Immune response'' and ''Apoptosis''. Together these functions encompassed 247 transcripts (,20%) identified from  1,142 genes displaying the most differential expression between the FD-and SD-fed mice. Approximately 40% of the proteins associated with these genes are located in the plasma membrane, suggesting a biased impact of the fish oil on the membraneassociated activities. Leveraging from the mouse brain atlas (mouse.brain-map.org) we found several of these genes found enriching the brain regions (Table S4) such as olfactory bulb, cerebellar cortex, thalamus and amygdala typically associated with the motor activities, information relay and neuropsychology.
In FD-fed mice, a Src-mediated calcium-controlled network was suggested to reinforce synapse that possibly operated in a BDNF-MMP-9 independent pathway. FD suppressed the transcription of BDNF exons II, which is known to express in response to stress [85]. Activated PI3K-AKT-PKC cascades operated in concert.
We further acknowledged a risk of spurring depression through FD-mediated activation of mGluR2/Ca +2 [75]. However, the FDinduced long term potentiation was deemed as more plausible endpoint in cognition of the additional evidences, including the stimulation of NMDAR and downstream glutaminergic/dopaminergic/cholinergic networks [69,77] and the elevated transcription of WNT family genes [117], PPP1R1B [76] and NEUROD6 [118]. The risk of Ca +2 -induced biased recurrence of excitatory synaptic transmission was possibly mitigated by the FD-induced NLGN1 that typically helps in restoring the balance between the excitatory and inhibitory synapses [119].
FD-induced attenuation of oxidative stress and reduction of amyloidal burden coupled with suppressed inflammatory and apoptosis networks affirmed the potential efficacy of fish oil in  treating a range of neurodegenerative debilities such as affective, Alzheimer's and Parkinson's disorders. In contrast, the FDstimulated neuromodulators like dopamine, choline, GABA and glutamate may diminish the efficacy of the fish oil in treating bipolar disorder, schizophrenia and suicidal inclination, which has been proposed to be triggered by the hyperactivity of the same pool of molecular messengers [105,106]. Taken together, our results suggested a potential susceptibility of the fish oil's efficacy on the disease pathology and highlighted the need for strategic repositioning of the fish oil as a therapeutic adjuvant. Supporting our results a number of past in vitro rodent models [41] [37] [35,36,38,39] reported n-3 PUFA benefits in the restoration of myelin sheath, activation of dopaminergic networks with downstream elevation of PPP1R1B, activation of synaptic plasticity and comprehensive attenuation of inflammasome networks.
In conclusion, fish oil could provide therapeutic benefits for a number of neurodegenerative diseases, while there is a valid concern about its therapeutic efficacy in treating certain other illnesses. The comprehensive investigation to confirm these findings are exigent and the effectiveness of this fish oil diet . Regulatory networks showing clusters of differentially-regulated genes predicted by mRNA microarray and validated by 96-well qPCR assay. The nodes represent the genes and the solid lines depict the interactions between two molecules. Red-and green-colored panels show significantly increased and decreased expression in the FD-fed mice compared to the SD-fed mice (log 1.5 (FD-/SD-fed mice). The right and left panels represent the mRNA microarray and qPCR output, respectively. The molecules without color are not mined by the present analysis; they are used to build the networks. These molecules constitute the major hubs of this network, but are not identified in the data shown here. doi:10.1371/journal.pone.0090425.g004 should be validated by introducing this diet type to specific disease models. Figure S1 The volcano plot of the GOI depicting the expression distribution in FD and SD fed murine hemibrains. Here the x-axis depicts the log 2 (fold change) and y-axis depicts -log 10 (p-value), where p-value is calculated by FDR algorithm. The red and green dots represent the up-and downregulated transcripts, respectively. (TIF) Figure S2 Regulatory networks showing an additional three clusters of differentially-regulated genes involved in ''neurogenesis''. The nodes represent the genes and the solid lines depict the interactions between two molecules. Red-and green-colored genes show significantly increased and decreased expression (log 1.5 (FD-/SD-fed mice) in the FD-fed mice compared to the SD-fed mice (a color scale is at the bottom right corner). The genes are clustered based on their proteins' cellular locations in the nucleus, cytoplasm, plasma membrane, and extracellular space. Three subgroups are ''differentiation of   Table S5 The genomic candidates associated with apoptosis (Apop) and immune response (ImRes). The genes are identified by the Entrez IDs and gene symbols. The log ratio indicates the log 1.5 transformed of the average transcriptomic expressions from FD-/SD-fed mice. (DOCX)