For this study, we used the transgenic mouse line 5xFAD (MMRRC catalog #034840-JAX). Nine-month-old 5xFAD mice overexpressing the 695-amino acid isoform of the human amyloid precursor protein (APP695) carrying the Swedish, London, and Florida mutations under the control of the murine Thy-1 promoter were used. In addition, they express human presenilin-1 (PSEN-1) carrying the M146L/L286V mutation, also under the control of the murine Thy-1 promoter (Oakley et al., 2006). Also, for experiments of inhibition of hippocampal activity, we used 9-month-old C57BL6/J male mice (catalogue #000664-JAX). All mice were housed together in numerical birth order in groups of mixed genotypes (three-to-five mice per cage). All the animals were housed with access to food and water ad libitum in a colony room kept at 19–22°C and 40%–60% humidity, under an inverted 12:12 h light/dark cycle (from 08:00 to 20:00). All animal procedures were approved by local committees [Universitat de Barcelona, CEEA (10,141) and Generalitat de Catalunya (DAAM 315/18)], in accordance with the European Communities Council Directive (86/609/EU).

Marine compounds were obtained from the available sample collections at the University of Barcelona. Briefly, the marine organisms were extracted with organic solvents and the extracts were further purified through chromatographic methods (HPLC) as previously reported (Núñez-Pons et al., 2010; Núñez-Pons et al., 2012). meridianins were tested as mixtures in the experiments.

Primary hippocampal cultures were performed as previously described (Anglada-Huguet et al., 2016). Hippocampus from E17.5 WT mouse embryos (C57BL6/J mice, MMRRC catalog #000664-JAX) were dissected and gently dissociated with a fire-polished Pasteur pipette. Cells were seeded (50,000 cells/cm² for immunochemical staining) onto 24 mm culture plates pre-coated with 0.1 mg/ml poly-d-lysine (Sigma Chemical Co., St. Louis, MO, United States) and cultured in neurobasal medium supplemented with B27 (Gibco, Paisley, United Kingdom, ×50) and GlutaMAX (Gibco, ×100) at 37°C in a humidified atmosphere containing 5% CO₂. At day in vitro 3, cultures were transduced with, AAV9·CamKIIα-(0.4). eGFPWPRE.rBG diluted 5,000 times (from Perelman School of Medicine, University of Pennsylvania, AV-9-PV1917). At day in vitro 20, the cultures were treated for 24 h with meridianins 500 nM or vehicle as previously described in vitro (Llorach-Pares et al., 2020). Fixed primary cultures were subjected to immunofluorescence as previously described (Giralt et al., 2017), enhancing GFP labeling using GFP FITC-conjugated antibody (1:500, #Ab6662, Abcam).

Animals were euthanized by cervical dislocation. Left hemispheres were removed and fixed for 5 days in 4% paraformaldehyde in phosphate-buffered saline (PBS). Free-floating coronal brain sections (40 μm) were obtained using a Leica vibratome (Leica VT1000S). Immunofluorescence procedure was carried out as described elsewhere (Giralt et al., 2017). The following primary antibodies were used: anti-GFAP rabbit polyclonal (1:500, Dako, REF#Z0334, Lot#20049468), anti-Iba1 rabbit polyclonal (1:500, Wako, #09-19741), and anti-APP (1:800, Novus Biologicals, #NBP2-62566). The following secondary antibody was used: Anti-rabbit AlexaFluor555 (1:300, Cat#SC-2048, Lot#L1815) from Thermo Fisher Scientific.

Dorsal hippocampus in fixed tissue and fixed primary cultures were imaged using a Leica Microsystems Confocal SP5-II at the Medical School Imaging facility, with a ×20 or ×40 numerical aperture lens with 5× digital zoom and standard (one airy disc) pinhole (1 AU) when required and frame averaging (three frames per z step) were held constant throughout the study. Confocal z stacks were taken every 0.2 µm for in vitro experiments and every 2 µm for in vivo experiments, and at 1,024 × 1,024 pixel resolution. The imaging analysis was performed with the Fiji freeware (Schindelin et al., 2012). Briefly, for in vivo imaging analysis, for each mouse, at least two slices of 40 µm containing dorsal hippocampal tissue were analyzed. Up to two representative images, were obtained from each mouse. To estimate the density of dendritic spines in primary cultures, 31–39 dendrites from pyramidal-shaped neurons (one or two dendrites per neuron) from three different cultures were counted.

Hippocampal samples were collected in cold lysis buffer containing 50 mM Tris base (pH 7.5), 10 mM EDTA, 1% Triton X-100, and supplemented with 1 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride, 1 mg/ml leupeptin, and 1 mg/ml aprotinin. Samples were centrifuged at 16,000 g for 15 min and the supernatants collected. After incubation (1 h) in blocking buffer containing 2.5% BSA and non-fat powdered milk in Tris buffered saline Tween (TBS-T) (50 mM Tris–HCl, 150 mM NaCl, pH 7.4, 0.05% Tween 20), membranes were blotted overnight at 4°C with primary antibodies. Antibodies used for immunoblot analysis were: GSK3β (1:1,000; Cell Signaling, #9315), phosphor-GSK3β at Ser9 (1:1,000; Cell Signaling, #9336xz), BDNF (1:1,000; Icosagene, #327-100, clone 3C11), catalytic PKAα (1:1,000; Santa Cruz #sc-903), phospho-PKA at Thr197 (1:1,000 Cell Signalling #5661S), TNFα (1:500, Abcam #ab1793). The membranes were then rinsed three times with TBS-T and incubated with horseradish peroxidase-conjugated secondary antibody for 1 h at room temperature. After washing for 30 min with TBS-T, the membranes were developed using the enhanced chemiluminescence (ECL) kit (Santa Cruz Biotechnology). The ImageLab densitometry program (ImageLab from ChemiDoc system from Bio-Rad) was used to quantify the different immunoreactive bands relative to the intensity of the α-tubulin or, in the case of phospho-GSK3β and phospho-PKA, they were relativized with respect to total GSK3β and total catalytic PKAα, respectively.

C57BL6/J or 5xFAD mice were deeply anesthetized with ketamine-xylacine (100 and 10 mg/kg, respectively) and placed in a stereotaxic apparatus for unilateral osmotic minipump (model 1,002; Alzet, Palo Alto, CA, United States) implantation. The brain infusion kit (#0008663) was also used to deliver into the third ventricle 0.11 μl per hour of vehicle or meridianins 500 nM (0.1 mm posterior to bregma, ± 0.8 mm lateral to midline and −2.5 mm ventral to the parenchyma surface). Cannulas were fixed on the skull with the Loctite 454 (from Alzet). Minipumps, previously equilibrated overnight at 37°C in PBS, were implanted subcutaneously in the back of the animal. Mice were allowed to recover for 7 days before starting the behavioral assessment.

Fresh brain hemispheres were processed, following the Golgi–Cox method as described previously (Giralt et al., 2017). Essentially, mouse brain hemispheres were incubated in the dark for 21 days in filtered dye solution (10 g L⁻¹ K₂Cr₂O₇, 10 g L⁻¹ HgCl2, and 8 g L⁻¹ K₂CrO₄). The tissue was then washed 3 × 2 min in water and 30 min in 90% ethanol (EtOH) (v/v); 200 µm sections were cut in 70% EtOH on a vibratome (Leica Microsystems) and washed in water for 5 min. Next, they were reduced in 16% (v/v) ammonia solution for 1 h before washing in water for 2 min and fixation in 10 g L⁻¹ Na₂S₂O₃ for 7 min. After a 2 min final wash in water, sections were mounted on superfrost coverslips, dehydrated for 3 min in 50%, then 70%, 80%, and 100% EtOH, incubated for 2 × 5 min in a 2:1 isopropanol:EtOH mixture, followed by 1 × 5 min in pure isopropanol and 2 × 5 min in xylol. Brightfield images of Golgi-impregnated stratum radiatum secondary apical dendrites from hippocampal CA1 pyramidal neurons and secondary apical dendrites from cortical pyramidal neurons of the layer V in frontal cortex and entorhinal cortex were obtained. Images were captured with a Nikon DXM 1200F digital camera attached to a Nikon Eclipse E600 light microscope (×100 oil objective). Only fully impregnated pyramidal neurons with their Soma found entirely within the thickness of the section were used. Image z stacks were taken every 0.2 mm and at 1,024 × 1,024 pixel resolution, yielding an image with pixel dimensions of 49.25 × 49.25 mm. The total number of spines counting were performed by using the Fiji freeware (Schindelin et al., 2012). At least 60 dendrites per group from at least five mice per genotype were counted. Picture acquisition and subsequent analysis were performed independently by two investigators blind to genotypes, and results were then pooled. Overall differences between the results were minor.

Sample sizes were determined by using the power analysis method: 0.05 alpha value, 1 estimated sigma value, and 75% of power detection. All data are expressed as mean ± SEM. Normal distribution was tested with d’Agostino and Pearson omnibus normality test. If the test was passed, statistical analysis was performed using parametric statistical analysis. Before pairs of comparisons, we performed the F test to compare variances. In experiments with normal distribution, statistical analyses were performed using the unpaired two-sided Student’s t-test (95% confidence) and the two-way ANOVA with the Bonferroni’s or Tukey’s post-hoc tests as appropriate and indicated in the figure legends. The t-test with Welch’s correction was applied when variances were unequal. Values of p < 0.05 were considered statistically significant. Grubbs and ROUT tests were performed to determine the significant outlier values. All experiments in this study were blinded and randomized by blocks of animals. All mice bred for the experiments were used for pre-planned experiments and randomized to experimental groups. Data were collected, processed, and analyzed randomly. The experimental design and handling of mice were identical across experiments. Littermates were used as controls with multiple litters (3–5) examined per experiment.

We previously reported that meridianins are capable to induce an increase on cell arborization (Llorach-Pares et al., 2020). Thereby, we hypothesized that these compounds have the potential to induce structural synaptic plasticity. We then transduced primary hippocampal neurons with an AAV9-CaMKII-GFP at day in vitro 3. At day in vitro 21, spine density was quantified just after a 24 h treatment with meridianins 500 nM. We observed that the number of spines in meridianin-treated neurons was increased compared with neurons treated only with vehicle (Figures 1A,B). These results reveal a significant potential to induce structural synaptic plasticity by meridianins and provide a rationale for the use of these compounds to counteract cognitive decline and progressive neuropathology in AD mouse models.

Although we already reported that meridianins can inhibit GSK3β in vitro (Llorach-Pares et al., 2020), this potential function has never been tested in vivo. From the same previous report, in silico simulations suggested that meridianins could have difficulties to cross the blood–brain barrier. Thus, we first verified whether meridianins could exert GSK3β inhibition in vivo by injecting the compound in the third ventricle of the brain in living adult mice (Figure 2A). We evaluated GSK3β inhibition in the hippocampal tissue by estimating the phosphorylation levels of the protein in the serine 9 residue (Ser9), which is widely established to be a substrate directly related with GSK3β inhibition (Fang et al., 2000). We observed that GSK3β inhibition in the hippocampus was significant 20 and 60 min after intraventricular injection of meridianins 500 nM (Figure 2B), as observed by an increase in the phosphorylation levels at the Ser9 residue of the protein. These results encouraged us to evaluate whether meridianins could improve or ameliorate cognitive deficits in the 5xFAD mouse model of AD.

The 5xFAD transgenic mice display a strong phenotype with a relatively fast time course compared with other AD mouse models. We have previously observed that 8-month-old 5xFAD mice already display deficits in novel object recognition memory and in cognitive flexibility measured by the NORT and the spontaneous alternation in a T-maze or T-SAT (Giralt et al., 2018; de Pins et al., 2019; Pérez-Sisqués et al., 2021). Thus, we implanted in 9-month-old WT and 5xFAD mice a mini-osmotic pump to continuously deliver vehicle or meridianins 500 nM into the third ventricle of the brain for 14 days (Figure 3A). Seven days after surgical intervention, WT and 5xFAD mice treated with vehicle or meridianins were subjected first to the NORT and then to the T-SAT. First, in the NORT, we observed that both WT and 5xFAD mice explored more the new object with respect to the old one (Figure 3B). However, the preference of 5xFAD mice treated with vehicle for the new object was significantly reduced compared with that registered in 5xFAD mice treated with meridianins. These results indicate the presence of cognitive deficits in 5xFAD treated with vehicle compared with WT treated with vehicle but that such deficits were rescued in 5xFAD mice treated with meridianins (Figure 3B). In the T-SAT task, the spontaneous alternation was evaluated 1 h after a training session. WT vehicle and 5xFAD meridianins mice significantly alternated, whereas 5xFAD vehicle mice did not (Figure 3C). Intriguingly, WT meridianins also showed impaired spontaneous alternation, suggesting that permanent GSK3β inhibition in healthy animals can be deleterious for some cognitive skills. In summary, meridianins rescued the observed impairments on object recognition memory and spontaneous alternation in the 5xFAD mice.

Since we observed a significant enhancement of synaptic density in cultured hippocampal neurons upon meridianin treatment, we then hypothesized that the potential cognitive improvements displayed by 5xFAD mice treated with meridianins could be associated to an induction of structural synaptic plasticity. We then performed a Golgi staining in all four groups of mice, and spine density analysis was carried out in the frontal cortex, entorhinal cortex, and CA1 of the hippocampus. We analyzed three regions because meridianins were delivered into the third ventricle and, consequently, many brain regions could be reached by the compound. First, we did not observe any change from any group with regard to the spine density in layer V pyramidal neurons of the frontal cortex (Figure 4A). In contrast, spine density was decreased in layer V pyramidal neurons of the entorhinal cortex in 5xFAD mice treated with vehicle whereas spine density was completely rescued in 5xFAD mice treated with meridianins (Figure 4B). Similarly, spine density was also decreased in CA1 pyramidal neurons of the hippocampus in 5xFAD mice treated with vehicle whereas spine density was completely rescued in 5xFAD mice treated with meridianins (Figure 4C). Taken altogether, meridianins rescued synapse loss in 9-month-old 5xFAD mice in, at least, two brain regions, namely, entorhinal cortex and hippocampal CA1.

Since we observed a prominent improvement on the synapse loss and recovery of some cognitive functions, we then sought for potential improvements in hippocampal neuropathological hallmarks such as the plaque number, astrogliosis and microgliosis. First, the plaque number did not differ between 5xFAD mice treated with vehicle and 5xFAD mice treated with meridianins in any hippocampal subregion, namely, CA1, CA3, or DG (Figure 5A). Next, we evaluated astrogliosis in all four groups by evaluating GFAP staining in all three hippocampal subregions: CA1, CA3, and DG (Figure 5B). Our analysis showed a trend of increased GFAP in all three subareas in 5xFAD mice treated with vehicle. However, this increase was significant only in the CA1 when compared with WT treated with vehicle. Interestingly, 5xFAD mice treated with meridianins did not show changes in GFAP staining in any hippocampal subarea compared to WT mice treated with vehicle. We next measured Iba1 staining intensity as a marker of microgliosis (Figure 5C) in all four groups. Changes in microgliosis using Iba1 as a marker were more prominent and clearer. Concretely, the intensity of Iba1 staining was upregulated in the 5xFAD mice treated with vehicle in all three hippocampal subareas (CA1-DG). In contrast, the intensity of Iba1 staining was upregulated only in the CA1 but not in CA3 and DG in 5xFAD mice treated with meridianins. Altogether, these results suggest that meridianins do not affect the plaque number, but it ameliorates neuroinflammation in the hippocampus of 5xFAD mice.

Since we observed a restoration of synapse loss, we then tested for a potential concomitant effect in the PKA-CREB-BDNF pathway in the hippocampus of 5xFAD mice treated with meridianins. For this purpose, we performed Western blot experiments in hippocampal lysates from all four groups of mice that underwent behavioral characterization (Figure 3) to evaluate total protein levels for BDNF (Figure 6A), phosphorylation levels of PKA in the threonine 197 (Figure 6B), and total catalytic PKA levels (Figure 6C), but no changes were observed in any group compared with WT mice treated with vehicle. To further investigate biochemical changes related with neuroinflammation induced by meridianins in the hippocampus, we interrogated TNFα levels (Figure 6D). We did not detect differences in any group compared with WT mice treated with vehicle. We concluded that meridianins may exert their effects through alternative pathways to ultimately rescue the loss of excitatory synapses and to ameliorate microgliosis and astrogliosis in 5xFAD mice.