Saturated and unsaturated fat diets impair hippocampal glutamatergic transmission in adolescent mice

Consumption of high-fat diets (HFD) has been associated with neuronal plasticity deficits and cognitive disorders linked to the alteration of glutamatergic disorders in the hippocampus. As young individuals are especially vulnerable to the effects of nutrients and xenobiotics on cognition, we studied the effect of chronic consumption of saturated (SOLF) and unsaturated oil-enriched foods (UOLF) on: i) spatial memory; ii) hippocampal synaptic transmission and plasticity; and iii) gene expression of glutamatergic receptors and hormone receptors in the hippocampus of adolescent and adult mice. Our results show that both SOLF and UOLF impair spatial short-term memory. Accordingly, hippocampal synaptic plasticity mechanisms underlying memory, and gene expression of NMDA receptor subunits are modulated by both diets. On the other hand, PPARγ gene expression is specifically down-regulated in adolescent SOLF individuals and up-regulated in adult UOLF mice.


Introduction
Consumption of high-fat diets (HFD) has been associated with both neuronal plasticity deterioration and cognitive deficits linked to the impairment of glutamatergic neurotransmission within the hippocampus (HIP) (Del Olmo and Ruiz-Gayo, 2018). These deficiencies are particularly intense in individuals that start to consume this type of diets during pre-adulthood, as demonstrated by studies comparing the effect of HFD in young and adult mice and showing that HFD worsen both relational and spatial memory capacities, specifically in animals that consume these diets during the juvenile period (Boitard et al., 2012;Del Rio et al., 2016;Kaczmarczyk et al., 2013;Valladolid-Acebes et al., 2013. Memory decline induced by HFD overlaps with the impairment of synaptic efficacy and blunting of NMDA-induced long-term depression (LTD) within the HIP (Valladolid-Acebes et al., 2012). The deterioration of synaptic plasticity triggered by HFD occurs concomitantly with changes in glutamatergic neurotransmission, as assessed in ex vivo neurochemical studies showing that HFD consumption decreases the efficiency of glutamate (GLU) up-take and evokes an adaptive up-regulation of glial GLU transporters (GLT-1 and GLAST) as well as a down-regulation of glutamine synthase within the HIP (Valladolid-Acebes et al., 2012).
The above-mentioned changes are apparently not due to the resulting obesity as they are detected before obesity is established and have been linked to brain insulin resistance (Beilharz et al., 2015;Grillo et al., 2015;Vinuesa et al., 2016). Other studies have reported that deficient leptin receptor (LepR) signalling triggered by HFD intake might also account for HIP function impairment (Cordner and Tamashiro, 2015;Dodds et al., 2011;Hwang et al., 2010;Mainardi et al., 2017;McGregor and Harvey, 2018;Valladolid-Acebes et al., 2012). It is suggested that either specific fatty acids (FAs) or a certain ratio of saturated vs. unsaturated FAs yielded by HFD consumption could account for HIP Abbreviations: AMPA, α-amino-3-hydroxy-5-methyl-4-isoxazolpropionic acid; AMPAR, AMPA receptor; BST, Basal synaptic transmission; BW, Body weight; FA, Fatty acid; fEPSP, Field excitatory postsynaptic potentials; GLU, Glutamate; HFD, High-fat diet; HIP, Hippocampus; LepR, Leptin receptor; NMDA, N-methyl-D aspartate; NMDAR, NMDA receptor; SOLF, Saturated oil-enriched food; UOLF, Unsaturated oil-enriched food. dysfunction independently of other factors (Del Olmo and Ruiz-Gayo, 2018). Moreover, although no previous study has demonstrated that saturated FA intake is an independent risk factor for memory and mood decline, population studies carried out in children/adolescents point to a negative correlation between HIP-dependent relational memory and saturated fat intake, independently of metabolic factors (Baym et al., 2014). Furthermore, HIP-dependent memory deficits have been observed in adolescent mice that display HFD-induced obesity, but not systemic insulin resistance .
Related to the work from other groups , the present study is based on the hypothesis that regular intake of diets containing elevated amounts of saturated, but not unsaturated FAs is detrimental for HIP mechanisms that underlie learning/memory, particularly in individuals that consume these diets during the adolescence/early juvenile period. To address this issue, diets enriched with either saturated (SOLF, Saturated OiL-enriched Food) or unsaturated fat (UOLF, Unsaturated OiL-enriched Food) were provided for 8 weeks to 5-(adolescent) and 8-weekold (young adult) mice, and their effects on: i) spatial short-term memory; ii) synaptic transmission and plasticity, and iii) gene and protein expression of glutamate, hormone and peroxisome proliferator-activated (PPARs) receptors, were characterized in the HIP.

Diets, animals, and experimental design
The two HFDs used in this study were manufactured in our laboratory as previously described (Plaza et al., 2019)  All experiments were performed in C57BL/6J male mice (Charles River, France). Briefly, mice were housed under a 12-h light/12-h dark cycle, in a temperature-controlled room (22ºC) with water available ad libitum, in accordance with the European Communities Council Directive (86/609/EEC) for the care and use of laboratory animals (PCD-CEU08-112-16). Animals were weaned at P21 and were kept on standard chow until P35, when they were divided into three groups with similar mean body weight (BW), single housed to properly monitor food intake, and assigned (free access) either to SD (Teklad global 2018, Harlan Laboratories, IN, USA), UOLF or SOLF. After 8 weeks, animals were used for the procedures detailed bellow. Mice were killed by decapitation, blood was collected and preserved, and HIP were dissected and used for electrophysiological studies or frozen for further analyses.
Two studies were conducted to identify the differential effect of SOLF and UOLF on 1) spatial memory performance (Y-maze test) and HIP gene expression of hormone receptors (insulin, leptin and adiponectin), PPARs, and genes involved in glutamatergic transmission, and 2) basal synaptic transmission (BST) and synaptic plasticity, as well as on protein levels of GLU receptor subunits within the HIP.
Study 1 was carried out simultaneously in mice that started to consume the assigned diet either on P35 (these animals received the diet during the adolescent period and early adulthood and will be referred as the adolescent cohort from now, n = 7 per group) or on P56 (referred as adult cohort, as they consumed the diet only during adulthood, n = 7 per group). These animals were the same used in a previous report (Plaza et al., 2019). Study 2 (n = 17 per group) was carried out only in adolescent mice (Fig. 1).

Plasma biochemistry
Plasma leptin, insulin, gastric inflammatory polypeptide (GIP), resistin, interleukin-6 (IL6), tumour necrosis factor α (TNFα), and The research was organized in two studies. Study 1 was carried out both in adolescent and adult animals that started to consume the experimental diets on postnatal days 35 (P35) and 55 (P55), respectively. Study 2 was carried out only in adolescent animals. In both studies food intake and BW were monitored weekly. Animals from Study 1 were tested in the Y-maze at the end of the dietary treatment, then killed and HIP samples were used for WB and RT-PCR. Animals from Study 2 were used in electrophysiological experiments. monocyte chemoattractant protein-1 (MCP1) were measured in a multiplex immunoassay (Millipore, Billerica, MA) in a Bio-Plex suspension array system 200 (Bio-Rad Laboratories, Hercules, CA, USA). Adiponectin was analysed by EIA (Abcam, UK). Triglycerides (Spinreact, Spain) and non-esterified FA (NEFA) (Wako Bioproducts, USA) were measured by spectrophotometric methods.

Evaluation of spatial memory in the Y maze behavioral testing
Spatial short-term working memory was analyzed by recording spontaneous alternation in a black plexiglass Y maze conformed by three identical arms (50 cm long x 19 cm wide) with 35 cm high walls (Contreras et al., 2019). Animals were placed in the testing room, which was rich in visual cues, 60 min before testing for habituation. For testing, mice were placed into one arm, and allowed to explore the maze for 10 min, during which arm entries (all four paws within an arm) and decision time (time in the center of the maze) were recorded. A correct alternation was defined as an entry into three different arms (A, B and C) in overlapping successive sequences of 3 arm entries (e.g., a successive entry sequence A, C, B, C, B, A, C, A, B). The percent alternation score was calculated as [actual alternations]/[possible alternations] X 100.

Electrophysiology assays
Briefly, transverse HIP slices (400 µm thick) were prepared by using a manual tissue chopper (Stoelting Tissue Slicer, Illinois) and placed in gassed (95% O 2 , 5% CO 2 ) ice-cold Krebs-Ringer bicarbonate (KRB) solution containing (mM); 119 NaCl, 2.5 KCl, 1 KH 2 PO 4 , 1.3 MgSO 4 , 2.5 CaCl 2 , 26.2 NaHCO 3 and 11 glucose, in a humidified interface chamber at 20-25ºC, as described previously (Del Olmo et al., 2000, 2003. After 2 h incubation, the slices were transferred to the submersion recording chamber, where they were continuously perfused (2 ml/min) with standard KRB solution. Field excitatory postsynaptic potentials (fEPSPs) were evoked by stimulating Schaffer collateral commissural fibers with biphasic electrical pulses (30-70 μA and 100 μs at 0.033 or 0.066 Hz) delivered through bipolar tungsten insulated microelectrodes (0.5 MΩ) and recorded in the CA1 stratum radiatum using tungsten electrodes (1 MΩ). Electrical pulses were generated by a pulse generator Master 8 (AMPI, Israel) and the recording electrode was connected to an AI-402 amplifier (Axon Instruments, USA) connected in turn to a CyberAmp 320 signal conditioner (Axon Instruments, USA). Evoked responses were digitized at 25-50 Hz using a Digidata 1320 A (Axon Instruments, USA) and stored on a Pentium IV IBM-compatible computer using pCLAMP 9.0 software (Axon Instruments, USA). The Schaffer collateral axons contacting the cell population of interest were stimulated every 30 s and after obtaining stable synaptic responses for at least 20 min (baseline period) an experimental protocol was performed. Synaptic plasticity was studied by inducing long-term potentiation (LTP) by four high-frequency stimulation (HFS) trains (100 Hz, 1 s, at test intensity) separated by 20 s. In another set of experiments, slices were perfused with 15 μM NMDA for 6 min to induce NMDA-LTD, following the protocol previously used in our laboratory (Lee et al., 1998;Naranjo et al., 2019;Valladolid-Acebes et al., 2012). Synaptic strength was assessed by measuring the initial slope of the fEPSP and the data were normalized with respect to the mean values of the responses of each animal during the 20 min baseline period. A single slice from each individual animal was considered as n = 1. All electrophysiological experiments were carried out at 31-32 • C. In some of the experiments and during the baseline period, input/output (I/O) curves were applied to evaluate basal synaptic transmission (BST).

RT-PCR
Total HIP RNA was extracted by using the Tri-Reagent protocol (Sigma, USA). cDNA was then synthesized from 1 µg total mRNA by using a high-capacity cDNA RT kit (BioRad, CA). Quantitative RT-PCR was performed by using designed primer pairs (Integrated DNA Technologies, USA. Table 2, Supplementary Material). SsoAdvanced Universal SYBR Green Supermix (BioRad, CA) was used for amplification according to the manufacturer's protocols, in a CFX96 Real Time System (BioRad). Values were normalized to the housekeeping genes Actb and 18 s. The ΔΔC(T) method was used to determine relative expression levels. Statistics were performed using ΔΔC(T) values (Livak and Schmittgen, 2001).

Statistical analysis
All data are presented as the means ± S.E.M. Statistical significance was determined by one-way ANOVA (ANOVA-1) followed by Bonferroni's post hoc test or two-way ANOVA (ANOVA-2) when appropriate. Outliers were identified by using the ROUT method (Q = 1%; GraphPad Prism software).

Spontaneous alternation is modulated by diets enriched either in saturated or in unsaturated fatty acids
Spontaneous alternation performance was tested in the Y maze both in adolescent and adult mice that consumed either SOLF or UOLF. The percentage of spontaneous alternation was analyzed in adolescent and adult cohorts after dietary treatment, as illustrated in Fig. 2. Statistical analysis by ANOVA-2 revealed an effect of dietary treatment (P < 0.05) while no effect of age nor significant interaction dietary treatment x age were observed. Fig. 3A and B show that dietary treatment had no effect on Gria1 and Gria2 gene expression, which suggests that AMPAR subunit levels are not modulated by SOLF/UOLF diets. With respect to the NMDAR subunits ( Fig. 3D and E), a down-regulation of both Grin2A and Grin2B gene expression was triggered by the dietary treatment (ANOVA-2; P < 0.05 and, P < 0.05, respectively). In the case of Grin2A, a significant effect of age (P < 0.01) was found, whereas no effect in the interaction was detected. For Grin2B, both age and the interaction age x diet were not significant (Fig. 3B). No effects on Grin1 subunit expression were detected (Fig. 3C).

SOLF and UOLF regulate the expression of both NMDA2A (Grin2A gene) and NMDA2B (Grin2B gene) subunits of the NMDAR in the HIP
To assess the effect of SOLF on Grin1, Grin2A and Grin2B expression, their correspondent encoded proteins were quantified by WB in adolescent mice. As illustrated in Fig. 3G-H, SOLF tended to reduce both NMDA2A and NMDA2B immunoreactivity within the HIP although this effect did not reach statistical significance.

Peroxisome proliferator activated receptor gamma (Pparg) was down-regulated by SOLF in the hippocampus of adolescent mice
Gene expression of hormone receptors potentially sensitive to HFDs, such as Ppara and Pparg, insulin (Insr), adiponectin (Adipor1 and Adi-por2) and leptin (Lepr) receptors, was measured in the HIP. In the case of Ppara, no effect of the dietary treatment was observed either in adolescent or in adult mice (Fig. 3I). In contrast, Pparg expression was dependent on dietary treatment (ANOVA-2; P < 0.001), animal age (P < 0.001) and interaction diet x age P < 0.001). Lepr, Adipor and Insr were not affected by the dietary treatment in either adolescent or adult mice ( Fig. 3K-N).

SOLF and UOLF differently affected basal synaptic transmission and LTP/LTD in the hippocampus of adolescent mice
Since previous results, obtained by other authors and ourselves (Boitard et al., 2012;, indicate a more potent effect of HFD in adolescent mice, electrophysiological studies were carried out only in this group of animals. As illustrated in Fig. 4A, both diets affected BST. SOLF blunted input/ output (I/O) curves (Fig. 4A) indicating a negative impact of this diet on synaptic efficacy in CA1 pyramidal neurons. A similar effect was triggered by UOLF, although in this case the difference did not reach statistical significance.
The effect of both SOLF and UOLF on HIP synaptic plasticity was further characterized by analyzing their effect on LTP and LTD. Regarding LTP, statistical analysis of the recording by ANOVA-2 indicates significant differences in the factors diet (P < 0.05), time (P < 0.001) and interaction diet x time (P < 0.001; post hoc analysis are included in Fig. 4B). As illustrated in Fig. 4B and C, the magnitude of LTP observed 10 min after HFS was significantly attenuated by SOLF (mean ± S.E.M. of fEPSP 10 min after HFS; SD, 148.9 ± 5.0 vs. SOLF, 113.4 ± 2.5). Moreover, SOLF mitigated LTP maintenance 2 h after induction (mean ± S.E.M. of the fEPSP during the last ten minutes of the recording; SD, 144.6 ± 3.6 vs. SOLF, 117.6 ± 3.2; Fig. 4B and D). In contrast, UOLF failed to inhibit LTP induction (SD, 148.9 ± 5.0 vs. UOLF, 138.6 ± 3.3; ns. 10 min after HFS, Fig. 4B and C), but fully abolished LTP maintenance (last 10 min of the recording; SD, 144.6 ± 3.6 vs. UOLF, 107.9 ± 2.8; Fig. 4B and D). In summary, our results show that LTP induction is specifically impaired by SOLF while UOLF interferes with LTP maintenance.
In regard to NMDA-induced LTD, two-way repeated measures ANOVA (from the beginning until before HFS application) indicated statistical significance in diet, time and interaction diet x time (P < 0.01, P < 0.01 and P < 0.001 respectively; post hoc analysis are shown in the graph). Fig. 5 shows that this mechanism was impaired by UOLF since long-lasting depression of the fEPSP triggered by NMDA was not observed in UOLF-treated mice (SD: 53.2 ± 3.5 vs. UOLF: 85.7 ± 3.3). In contrast, SOLF mice displayed a stronger NMDA-LTD than controls (SD, 53.2 ± 3.5 vs. SOLF, 37.3 ± 2.9).
To further identify the extent to which NMDA responses were due either to synaptic plasticity or to an eventual toxic effect, HFS was applied 50 min after NMDA, once LTD was established (Fig. 5A). In control slices, HFS induced a re-potentiation of fEPSP until baseline values (109.0 ± 5.4, Fig. 5C), which was also observed in UOLF slices (99.8 ± 3.7, Fig. 5C), thus suggesting that HFS was able to produce synaptic plasticity. In contrast, HFS applied to SOLF slices did not evoke any change (65.9 ± 1.5) compared to the Control and UOLF groups; Fig. 5C.) This result would indicate that SOLF-treated HIP present less plasticity, which could be due, at least in part, to the apparent toxic effect of the combination of lauric and palmitic acids contained in SOLF (see LDH cytotoxic assay; Fig. 2, Supplementary Material).

Plasma biochemistry
Biochemical data from Study 1 have been previously reported (Plaza et al., 2019) and Table 1 shows data corresponding to Study 2. An increase in NEFA values was observed only in SOLF-fed mice (P < 0.01), whereas the increase in leptin and insulin was statistically significant only in UOLF mice (P < 0.05 and P < 0.001, respectively). No changes in GIP, resistin, IL6, TNFα and MCP1 (data not shown), triglycerides and adiponectin were found. BW gain and food intake are shown in Supplementary Material (Fig. 1).

Discussion
Our study shows that SOLF and UOLF intake impaired short-term spatial HIP-dependent memory. Moreover, SOLF was more detrimental than UOLF for synaptic plasticity mechanisms underlying spatial memory, such as LTP and LTD. These differences were not dependent on the increase in BW as SOLF caused smaller (this study) or, at most, similar weight gain than UOLF (Plaza et al., 2019). This finding supports the hypothesis that the HIP is a target for saturated FAs. In addition, the vulnerability toward saturated fat was only perceptible in animals that consumed the diet during the juvenile period. To our knowledge, this is the first report studying differential effects of saturated vs unsaturated fat diets on HIP synaptic plasticity mechanisms. It must be highlighted that alternation in the Y-maze is used to interrogate short-term working spatial memory, but additional studies using the radial arm maze or the Morris water maze would be necessary to properly characterize the impairment of learning and memory in response to SOLF/UOLF (Ernyey et al., 2019;Tarantini et al., 2019;Vorhees and Williams, 2014).
We have previously reported that obesity induced by conventional HFDs, which contain elevated amounts of both saturated and unsaturated fat and simple carbohydrates, occurs concomitantly with the impairment of spatial learning and memory in the eight-arm radial maze (Valladolid-Acebes et al., 2011) as well as in the novel recognition test (Valladolid-Acebes, 2013), specifically in mice consuming the diet during the adolescent period. These results, corroborated by other authors also reporting the inhibition of HIP neurogenesis in adolescent mice exposed to HFD (Boitard et al., 2012), are coherent with population studies showing that poor dietary quality is associated with a Fig. 2. SOLF impaired memory performance in the Y-maze test specifically in adolescent mice. The graph illustrates the percentage of spontaneous alternations in the Y-maze both in adolescent and adult mice (n = 7 per group) after 8-weeks of SOLF and UOLF. Data are expressed as means ± SEM. Comparison was made by ANOVA-2 test (Diet, F (2,38) = 3.245, P < 0.05; age, F (1,38) = 1.617, P = 0.211; interaction F (2,38) = 2.601, P = 0.087).

Fig. 3. Effect of 8-week SOLF and UOLF on mRNA levels of glutamate receptors, PPAR, and hormone receptors in adolescent and adult mice hippocampus.
Panels A-E illustrate the effect of UOLF and SOLF on gene expression of GLU receptor subunits (Grin2a: Diet, F (2,34) = 2.988, P < 0.05; age, F (1,34) = 3.113, P < 0.01; interaction, F (2,34) = 2.163, P = 0.063; Grin2b: Diet, F (2,30) = 2.998, P < 0.05; age F (1,30) = 1.591, P = 0.219; interaction, F (2,30) = 1.521, P = 0.221). mRNA levels corresponding to PPARα, PPARγ, leptin, insulin and adiponectin receptors appear illustrated in panels I, J, K, L, M and N, respectively (Pparg: Diet, F (2,28) = 9.552, P < 0.001; age, F (1,28) = 10.522, P < 0.001; interaction, F (2,28) = 5.712, P < 0.001, followed by Bonferroni post hoc test; * P < 0.05 adolescent SD vs. adolescent SOLF and adult SD vs. adult UOLF; # P < 0.05 adolescent UOLF vs. adult UOLF). Values are means ± SEM (adolescent mice, n = 6 for SD, n = 7 for UOLF and SOLF groups; adult mice, n = 6 for SD, n = 7 for UOLF and SOLF groups). Effect of SOLF/UOLF on GluA1, GluN2A and GluN2B receptor subunit protein levels are shown in panels 3 F, 3 G and 3 H, respectively. decline of psychological functioning in adolescents (Jacka et al., 2011), and poor learning and memory capacities (Wright et al., 2017). However, none of these studies has identified the contribution to memory impairment of obesity itself vs that of particular nutrients (fat vs sugar) contained in HFD. The current research focusses on the effect of two sucrose/glucose-free diets containing an elevated proportion of either saturated (SOLF) or unsaturated FAs (UOLF); it has to be highlighted that SOLF contains a high proportion of palmitic and, especially, of lauric acid, which adds a supplementary interest to our work, since coconut oil, whose main FA is lauric acid, has been shown to impair memory function and alter HIP morphology in rats (Granholm et al., 2008). Taken together with other studies showing that palmitic acid induces morphological changes in primary HIP cultures, able to compromise cell function and excitability (Loehfelm et al., 2020), and to trigger microglia activation (Tu et al., 2019;Lin et al., 2020), our current data suggest that the combination of lauric and palmitic acids contained in SOLF has a critical role in cognitive deficits associated to SOLF consumption. Moreover, our data corroborate that consumption of HFDs is detrimental and further support the negative impact that regular intake of fat-enriched meals has on cognitive processes. In this sense, other authors have recently shown that acute exposure (7-9 days) to HFD during juvenility is sufficient to impair HIP functions depending on glucocorticoid receptors (Khazen et al., 2019).
The possibility that diet composition might be an obesityindependent triggering factor for cognitive impairment is stressed by other studies concluding that western-style diets cause neurocognitive damage in absence of obesity/overweight (Beilharz et al., 2015). It has to be highlighted that, under our experimental conditions, SOLF mice displayed elevated HOMA indexes (Plaza et al., 2019), a circumstance that might have an impact on memory (Rom et al., 2019). In fact, memory/learning deficits have been shown to be linked to brain insulin resistance (Grillo et al., 2015), a condition that can prematurely develops during HFD interventions before obesity is established (Vinuesa et al., 2016). It has to be noted that, in our study, plasma insulin levels were higher in UOLF than in SOLF mice, a data that needs to be interpreted cautiously considering that mice were not fasted before blood sampling. Therefore, a certain degree of insulin resistance, associated to the intake of saturated fat, might contribute to the effects observed here. In any case, the influence of HIP insulin sensitivity on spatial memory is a controversial matter since HFD has a different impact on HIP insulin sensitivity in males and females, which is not accompanied by intersex differences in terms of spatial memory and HIP excitability (Underwood and Thompson, 2016). We want to emphasize, that, in the current study, we have decided to use only males to minimize the number of animals, but new research aimed at characterizing sex differences concerning SOLF/UOLF effects in cognition and synaptic plasticity will be necessary to properly characterize this issue.
Another circumstance that could account for cognitive deficits is the development of HIP leptin resistance triggered by SOLF. This is a relevant issue considering that leptin has been shown to promote HIPdependent learning and memory (Van Doorn et al., 2017) and to regulate HIP synaptic transmission (Moult et al., 2010; Moult and Harvey,
2012; Oomura et al., 2006;Shanley et al., 2001). In fact, leptin-insensitive mice have been shown to display a significant impairment of both LTP and LTD (Li et al., 2002;Winocur et al., 2005), and a desensitization of the protein kinase B (Akt) pathway coupled to HIP leptin receptors (LepR) has been detected in adolescent mice consuming HFD and displaying deficits in spatial memory and moderate hyperleptinemia (Valladolid-Acebes et al., 2013). In the same vein, engineered mice lacking LepR specifically in astrocytes display impaired BST and LTD in the HIP (Naranjo et al., 2019). The possibility that SOLF would promote leptin resistance seems unlikely as plasma leptin levels in adolescent SOLF mice were lower than in their UOLF counterparts, in accordance with previous findings showing that SOLF represses leptin gene expression in visceral WAT (Plaza et al., 2019). However, this possibility cannot be fully discarded inasmuch as diets enriched in saturated fat have been shown to disrupt the blood-brain barrier (Hsu and Kanoski, 2014). Thus, it can be speculated that leptin access to the brain would be increased by SOLF with the consequent development of leptin resistance.
The electrophysiological data suggest that the impairment of spatial memory in SOLF-treated animals is related, at least in part, to the effect of saturated FAs on HIP plasticity and particularly on BST, as previously observed in a study carried out with a classical HFD (Valladolid-Acebes et al., 2012). A surprising result was the shortening of LTP triggered by both SOLF and UOLF, even though SOLF was more potent than UOLF in inhibiting LTP induction. This finding is coherent with studies carried out in animals treated with HFD (Hao et al., 2016), as well as with in . Forty-five min after NMDA withdrawal, four trains of HFS were applied to slices from SD (white circles, n = 6), SOLF (black circles, n = 8) and UOLF (grey circles, n = 8) animals. (Repeated measures ANOVA-2 was performed until before HFS application (Diet, F (2,16) = 7.969, P < 0.01; time, F (59,944) = 1.771, P < 0.001; interaction, F (118,944) = 2.423, P < 0.001, followed by Bonferroni post hoc test * * P < 0.01 and * P < 0.05 SD vs. SOLF and ## P < 0.01 and # P < 0.05 SOLF vs. UOLF). Bars represent means ± S.E.M. of fEPSP slopes 10 min prior (B) and 10 min after HFS (C). ANOVA-1 F (2128) = 42.51, P < 0.001 and F (2140) = 34.62, P < 0.001, respectively, followed by Bonferroni post hoc test; * ** P < 0.001 and * P < 0.05 SD vs. SOLF and SD vs. UOLF; ### P < 0.001 SOLF vs. UOLF. Upper traces show representative recordings from one of each type of experiment (a and b show the time points corresponding to each curve). No more than 1 slice/mouse was used in each experiment. vitro assays showing that HIP slices treated with palmitic acid display a partial inhibition of LTP (Contreras et al., 2017), and confronts with the lack of effect of HFD identified in other studies (Valladolid-Acebes et al., 2012). In any case, the negative impact of UOLF on LTP duration and BST strongly suggests that unsaturated FAs may also have a negative impact on HIP plasticity. Nevertheless, the molecular mechanisms that account for SOLF and UOLF effects in LTP seem to be different, since SOLF affected the induction of the phenomenon whereas UOLF impaired its maintenance. Wong et al. (1989) observed differences between unsaturated and saturated fat-enriched diets in synaptic plasticity mechanisms and observed that a HFD containing both lauric and myristic acids reduced the phosphorylation of the substrate protein F1 (aka GAP43), a factor that accounts for HIP synaptic plasticity (Linden and Routtenberg, 1989). On this basis, one could speculate that a deficient activity of PKC signaling pathways might account for LTP impairment triggered by SOLF.
Regarding LTD, our results are striking considering that it was inhibited by UOLF but apparently potentiated by SOLF. The effect of UOLF, which is similar to that provoked by classical HFDs (Hwang et al., 2010;Valladolid-Acebes et al., 2012), would suggest that unsaturated FAs impair this plasticity mechanism. As changes in NR2A/NR2B ratios have been related to LTD thresholds (Xu et al., 2009), the modulation of NMDAR subunits detected in SOLF-treated mice could underlie changes in synaptic transmission and plasticity. In any case, the effect of SOLF in LTD is difficult to interpret since HFS was unable to re-potentiate the fEPSP slope, as it did in control and UOLF-treated animals. This condition suggests a toxic effect of SOLF, able to exacerbate NMDA-evoked LTD. Such a possibility would be compatible with cytotoxicity detected in HIP slices incubated with a combination of lauric and palmitic acids, used in a proportion similar to that contained in SOLF (Fig. 2,  Supplementary Material).
The electrophysiological and behavioural findings could be related to the down-regulation of Grin2A and Grin2B gene expression triggered by SOLF and UOLF. Such an adaptive regulation is coherent with the specific role of GluN2A and GluN2B in both LTP and LTD (Foster et al., 2010;Kellermayer et al., 2018;Liu et al., 2004;Shipton and Paulsen, 2014). In fact, GluN2A deletion mitigates HIP LTP and impairs spatial learning (Kannangara et al., 2015;Kiyama et al., 1998;Sakimura et al., 1995). In addition, changes in NMDAR density have been detected after synaptic plasticity induction as well as by spatial memory formation (Baez et al., 2018). Some authors point to the increase in the synaptic GluN2A/GluN2B ratio as a stabilizer of synaptic changes, contributing to memory consolidation, particularly concerning spatial representation (Baez et al., 2018). Our data are coherent with this idea and would suggest that modulation of NMDAR subunits by SOLF might underlie the deleterious effect of unsaturated FAs on spatial memory. It must be noted that the inhibition of Grin2A and Grin2B expression was not accompanied by a significant decrease of the corresponding GluN2A and GluN2B proteins. This circumstance could be due to the fact that protein levels change more slowly than the corresponding mRNA (Vogel and Marcotte, 2013) or even to a poor efficiency of translation mechanism, previously reported by other authors (Boutej et al., 2017;Schindler et al., 1990). In addition, the stability observed in AMPAR subunit mRNA levels (Gria1 and Gria2) also points to the lack of effect of the diets on AMPAR subunit content. The lack of effect of SOLF/UOLF on Gria1 levels contrasts with the results reported by Spinelli et al. (2017). Although we cannot provide any precise interpretation to explain this difference, one could speculate that SOLF/UOLF and the diet used by Spinelli et al., probably affect GLU turnover (Valladolid-Acebes et al., 2012) differently and trigger a distinct regulation of GLU receptor subunits, as a consequence of specific adaptive compensatory mechanisms. Unfortunately, we lack data regarding the influence of SOL-F/UOLF on HIP GLU uptake kinetics that would have allowed proper discussion of this issue.
Finally, the inhibition of PPARγ gene expression suggests that the impairment of synaptic plasticity induced by SOLF could be related to PPARγ modulation, in addition to the decrease in NMDA subunits 2A and 2B. In this regard, our previous studies have shown that commercial HFD containing lard decreases NMDA2B gene expression concomitantly with deficits in spatial learning . Moreover, the activation of PPARγ has been demonstrated to ameliorate spatial memory deficits (Chen et al., 2016). For these reasons, further studies using pharmacological tools to investigate whether NMDA and PPARγ agonists prevent/reverse SOLF-evoked spatial memory deficits would be necessary.
Taken together, our data suggest that changes of synaptic plasticity could be mainly related to post-translational modifications of GLU receptors that modulate their functionality; otherwise, one can speculate that changes in plasma levels of insulin and leptin, eventually leading to insulin and/or leptin resistance, might also be pivotal in modulating both HIP synaptic plasticity and short-term memory, particularly in SOLF-treated mice (Harvey, 2007;Harvey et al., 2006). Moreover, PPARγ might also play a role in SOLF effects. In conclusion, our study shows that diets enriched with either saturated or unsaturated FAs modulate spatial HIP-dependent spatial memory through modulation of synaptic transmission and plasticity mechanisms in this area.