Gut Microbiota Alterations and Cognitive Impairment Are Sexually Dissociated in a Transgenic Mice Model of Alzheimer’s Disease

Experimental design

Hemizygous APP/PS1 male (Tg-M) and female (Tg-F) mice expressing a chimeric mouse/human amyloid precursor protein (Mo/HuAPP695swe) and a mutant human presenilin 1 (PS1-dE9) transgene on a C57BL/6 genetic background (Jackson Laboratory, Bar Harbor, ME, USA, RRID: MMRRC-34829-JAX) and their wild-type (WT) littermates (males, WT-M and females, WT-F) were used for the study. Animal management was supervised by licensed veterinarian care and approved by the Bioethics Committee of the Center for Research and Advanced Studies of the National Polytechnic Institute (Cinvestav) in accordance with the Mexican official standard NOM-062-ZOO-1999 and the principles set forth in the 8^th edition of the National Institute of Health (NIH) guide for the care and use of laboratory animals and endorsed by the Institute for Laboratory Animal Research, the Division on Earth and Life Studies, and the National Research Council of the National Academies. Once animals reached 4 months of age, they were individually housed with ad libitum access to food (AIN-93 diet) [48] and water under optimal vivarium conditions (12 h/12 h light–dark cycle, 20°C, and 40–50% relative humidity) for 2 months.

Estrous cycle evaluation

Cognitive evaluations in female mice were done only during the metaestrous/diestrous, the reproductive stages with the lowest hormonal levels [49], as estradiol have a direct impact in cognition and memory [50]. Cytology and typification of the reproductive stage was based on previous descriptions [51, 52].

Cognitive assessment

One week before sacrifice, behavioral testing was performed. To habituate the animals for the behavioral tests, animals were placed in the behavioral room 2 days before start testing trials. Anxiety-related behaviors were assessed by the Elevated Plus Maze (EPM) [50]. Briefly, the 50 cm elevated plus-shaped maze was made of medium-density fiberboard with a matte black acrylic surface and four arms (50 cm long and 10 cm wide) where two arms were open, and two arms were enclosed by 30 cm height walls. The task consisted of one trial per animal, where the animals were placed in the center of the EPM facing an open arm. Usually, rodents avoid open spaces and spend most of the time in enclosed places, such as the closed arms of the EPM [50]. Entries into the open/close arms were recorded during 5 min, and the animal’s trajectory was used to calculate percentage of time spent in each arm [50] (Fig. 1A, B).

OPEN IN VIEWER

Fig. 1

Effect of gender and genotype in anxiety and memory performance. A) Anxiety was inversely proportional to the time spent in the open arm [F (1, 23) = 2.35, 3.03 and 12.42 for interaction, row and column factor respectively] and B) directly proportional to the time spent in the closed arms [F (1, 25) = 2.15, 4.07 and 6.30 for interaction, row and column factor respectively] in the elevated plus maze. C) Working memory was directly proportional to the spontaneous alternations [F (1, 25) = 2.25, 24.31 and 30.05 for interaction, row and column factor respectively] in the T-maze. D) Recognition memory was directly proportional to the discrimination index [F (1, 24) = 3.50, 0.18 and 26.80 for interaction, row and column factor respectively] in the novel object recognition test. E) Spatial learning curves show the latency to find the hidden platform during learning phase [F (33, 396) = 1.03, F (11, 396) = 1.69 and F (3, 396) = 14.68 for interaction, row and column factor respectively] in the WM. X axis corresponds to the 12 consecutive learning assays. F) Spatial learning was inversely proportional to the latency to find the platform during learning phase [F (1, 30) = 10.95, 0.09 and 10.07 for interaction, row and column factor respectively] in the water maze. G) Learning performance was compared between trial 1 and 12 during learning phase in the WM. H) Spatial memory was inversely proportional to the latency to find the place where the platform was originally located [F (1, 31) = 4.02, 2.80 and 6.34 for interaction, row and column factor respectively] during WM memory phase. I) Representative schemes of WT-M and Tg-M mice trajectory for spatial learning trials from the start point to the platform area. J) Representative schemes of WT-F and Tg-F mice trajectory for the spatial memory trial from the start point to the platform area. Cardinal points are marked in maze walls as a star-shaped polygon (north), a square (south), a triangle (west), and a circle (east). Black arrow indicates start point, white arrow indicates finish point, central square indicates platform area and dotted line indicates mouse trajectory in the WM test. For A, WT-M n = 9, WT-F n = 7, Tg-M n = 9, and Tg-F n = 6. For B and C, WT-M n = 10, WT-F n = 7, Tg-M n = 10, and Tg-F n = 6. For D, WT-M n = 10, WT-F n = 6, Tg-M n = 10, and Tg-F n = 6. For E, F, and G, WT-M n = 10, WT-F n = 11, Tg-M n = 10, and Tg-F n = 6. For H, WT-M n = 10, WT-F n = 9, Tg-M n = 10, and Tg-F n = 6. For A-C and F-H, white color circles within bars indicate experimental subjects. For D and E, white, grey and black circles indicate experimental groups. Data are shown as mean±SD bars. For A-F and H, two-way ANOVA with Tukey post-hoc correction was performed comparing experimental groups. For G, unpaired t tests were performed comparing learning assay no.1 with no. 12 for each experimental group. For A-F and H, data between square brackets indicate F (DFn, DFd) values of two-way ANOVA with Tukey post-hoc corrections. Statistical significances are shown as ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, and ^****p < 0.0001. For E, statistical significance between WT-M versus WT-F is shown by □, Tg-M versus Tg-F by ∘, WT-M versus Tg-F by epsfbox G:/Tex/IOSPRESS/JAD/201367/IF-01.eps, WT-F versus Tg-M by #, WT-M versus Tg-M by ♂and WT-F versus Tg-F by ♀one symbol represents p < 0.05, two symbols p < 0.01, three symbols p < 0.001, and four symbols p < 0.0001. T1 = learning trial no. 1, T12 = learning trial no.12, F, female; M, male; Tg, transgenic; WT, wild type.

Working memory was evaluated by the T-maze (TM) [53]. The TM is an appropriate test to evaluate hippocampal and PFC functions in rodents [54] and to examine cognitive deficits in early stages of AD pathology [55]. Briefly, a T-shaped maze apparatus of polymethylmethacrylate with starting arm (8.5×10.5×33.0 cm), and two choice arms (8.5×10.5×30.0 cm each, left and right) was used as previously described [56]. The TM task consisted of two phases: 1) Sample phase, animals were placed at the start arm and allowed to choose between left or right choice arms. A central divider was then inserted into the start arm to create a start box. Once animals have chosen an arm, they were confined there for 30 s, thereafter the door was reopened and allow mice to return to the start arm, where animals were removed and returned to their cage. 2) Choice phase, 2 min later, animals were placed again in the start arm without the central divider and allowed them to choose an arm. When animals faced the TM apparatus, they display a tendency to alternate and avoid the entry into the previously entered arm. If the animal chooses two different arms in both phases, this is counted as spontaneous alteration. Two hours later the procedure was repeated. Three trials were given on each day for 2 days, with a total of 6 trials per animal. Percentage of spontaneous alternations were calculated from total of correct arm entries (Fig. 1C).

Recognition memory was evaluated by the Novel Object Recognition task (NOR) [57]. NOR is an appropriate test to evaluate recognition memory related to hippocampus [58] and prefrontal cortex (PFC) [59] functions, both regions heavily affected in AD [56]. This task is based on the natural tendency of rodents to explore a novel object more than a familiar one [57]. NOR test was carried out in a 40 cm² open field arena with 40 cm height walls of plywood surrounding the floor. NOR task consisted of three phases: 1) Habituation phase: animals explored the open field for 5 min without objects in the box; 2) Training phase: Twenty-four hours after habituation phase, animals were placed for 5 min in the box from the same starting position, and with two objects with identical features (weight, size, texture, shape and color) positioned in adjacent corners at 10 cm from the walls; 3) Trial phase: The recognition memory test was carried out 1.5 h after the training trial with the mice exploring the open field for 5 min in the presence of one familiar and one novel object, this last one different features than the former. A discrimination index (DI) was calculated for each animal after the test session (Fig. 1D) as DI = (time with novel object — time with familiar object) / (time with novel object+time with familiar object). Between all testing trials, all surfaces and objects were washed with 40 % ethanol solution. Exploration was defined as smelling and/or touching the object with the nose and/or forepaws [57].

Spatial memory was assessed by Water Maze (WM) [60]. The WM is an appropriate test to evaluate spatial memory related to hippocampus [61] and PFC [62] functions. Briefly, the WM task was carried out in a 100 cm diameter and 30 cm height round pool half filled with 26°C white water and an escape platform located in the center, as previously described [60]. Spatial clues were posted on each wall of the room with paper figures of different shapes and colors. WM task consisted of three phases, followed by an adaptation phase that consisted of placing the mice for 20 s on the scape platform visible for the animal: 1) Training phase: mice were released and allowed to swim in order to reach the scape platform in a maximum time of 60 s. This phase was repeated three times, each starting from a different cardinal point. 2) Learning phase: the animal was allowed to swim in order to find the hidden scape platform in a maximum time of 60 s for 12 consecutive trials, starting 3 non-consecutive times from different cardinal points. 3) Memory phase: immediately after all animals have completed the 12 consecutive learning trials, they performed 1 probe trial, in which the scape platform was removed from the pool. The animal was released from the north and allowed to swim to find the missing platform in a maximum time of 30 s. The time spent (latency) to find the hidden platform (learning phase) and the platform area (memory phase) in the pool was recorded and assigned as an inversely proportional measure to the spatial learning and memory respectively (Fig. 1E-H). Experiments were recorded with a Logitech DH Pro Webcam C920 camera connected to a computer equipped with Logitech Capture 1.10.110 with the Eniima software (inhouse made software) in order to track animal’s behavior and trajectory (Fig. 1I, J).

Determination of short chain fatty acids in feces

Fecal samples were collected for each animal before the cognitive tests. Animals were placed for 1 h in a Super Mouse 750™ Micro-Isolator® system cage for housing mice (15 cm width x 30 cm long x 17 cm high) previously cleaned and disinfected. At least 100 mg of feces pellets were collected per animal, immediately frozen and store at –70°C until use, for short-chain fatty acids (SCFAs) determination or microbiota analysis.

For SCFAs determination, the collected fecal samples were dried to constant weight and subsequently processed using the solid-phase extraction (SPE) method to analyze via HPLC (PerkinElmer Series N3896). At least 50 mg of dried feces were suspended completely in 1 ml of deionized water by vigorous mixing in a vortex at maximum speed for 10 min. The suspension was centrifuged at 13000 rpm for 10 min, the supernatant was transferred to a fresh tube and passed through activated C-18 max 100 mg/mL GracePure™ Reversed-Phase SPE Columns according to the manufacturer’s instructions as previously described [63]. The SCFAs were eluted with 1 mL of filtered water (Nylon Syr Filter, 13 mm, 0.22μm) and analyzed via HPLC using 20:80 acetonitrile:NaH₂PO₄ (J.T. Baker No. 9012-03 and Sigma-Aldrich No. S8282 respectively) at a pH of 2.2 using phosphoric acid, (J.T. Baker No. 0260-05) as mobile phase [64]. Standard curves were prepared for acetate (Sigma-Aldrich No. 45754-100ML-F), propionate (Sigma-Aldrich No. P1386-500ML), and butyrate (Sigma-Aldrich No. B103500-500ML). SCFAs proportion was calculated taking the total amount of acetate + propionate + butyrate for each experimental subject as 100% .

Fecal microbiota analysis

Bacterial DNA was extracted using the Favor-Prep™ Stool DNA Isolation Mini Kit (Favorgen Biotech Corp, Cat. No: FASTI001-1) according to the manufacturer's instructions. The purified DNA quantity was measured using 260/280 nm absorbance using a Nano Drop 2000 spectrophotometer (Thermo Scientific), and its quality was evaluated by 0.5% agarose gels electrophoresis. Genomic libraries of ∼281 base pairs (bp) 16S rDNA 3rd hypervariable region (V3) amplicons were generated for each experimental subject using specific V3-341F primers against V3 that also contained a different 12-bp Golay barcode (set of barcodes 1–100) complementary to position 340–356 of Escherichia coli 16S rRNA molecule rrnB GeneBank J01859.1 and an overhang adapter specified by Ion Torrent, and V3-518R reverse primers complementary to position 517–533 of the same molecule [65]. The amplicons of the V3 regions were generated by 1X 50μl polymerase chain reactions (PCR) containing 1μl of 10μM forward primer, 1μl of 10μM reverse primer (settings specified above), 0.25μl of 5 U/μl high thermostable DNA recombinant polymerase (Thermo Scientific No. EP0402), 1μl of 10 mM dNTP mix (Thermo Scientific No. R0191), 4μl of 25 mM MgCl₂ (Thermo Scientific No. R0971), 5μl of 10X taq buffer with KCl (100 mM Tris-HCl [pH = 8.8 at 25°C], 500 mM KCl and 0.8% volume/volume [% v/v] nonidet P40, Thermo Scientific No. B38) and 20 ng of genomic DNA in 37.75μl of nuclease-free water (Thermo Scientific No. R0581) per reaction. Amplification was carried out using Applied Biosystem Veriti Thermal Cycler (Supplementary Table 1). For sequencing, equivalent amounts of 1 to 100 barcoded amplicons (∼10μg each) were combined. Each mix was purified using E-Gel iBase Power System (Invitrogen). The libraries size and concentration were determined using an Agilent 2100 Bioanalyzer, and libraries for each run were adjusted to 26 pM prior to clonal amplification. Emulsion PCR was carried out using the Ion OneTouch™ 200 Template Kit v2 DL (Life Technologies), according to the manufacturer's instructions. Amplicon enrichment with ion spheres was done using Ion OneTouch ES. The sequencing was done using the Ion 316 Chip Kit v2 and the Ion Torrent Personal Genome Machine (PGM) System. After sequencing, reads were filtered by the PGM software to remove low quality and polyclonal sequences. During this process, sequences matching the 3' adapters were automatically trimmed and filtered. Ion Torrent PGM software, Torrent Suite v4.0.2, was used to demultiplex the sequenced data based on their barcodes. Poor quality reads were eliminated from the datasets, i.e., quality score ≤20, containing homopolymers >6, length <200 nucleotides (nt), and containing errors in primers or barcodes. Filtered data were exported as FASTQ files. Demultiplexed sequencing data were analyzed using quantitative insights into microbial ecology (QIIME) software v1.9.0 pipeline [66]. FASTQ files were converted into FASTA files, and all demultiplexed files were concatenated into a single file. Closed reference operational taxonomic units (OTUs) were determined at the 97% similarity level using the UCLUST algorithm. Chimeras were detected and removed from the datasets using the ChimeraSlayer. Sequence alignments were done against the Greengenes 13.9 core set (sequencing summary is shown in Supplementary Table 2). Microbial sequence data were pooled for OTUs analysis and comparison of taxonomic relative abundance was obtained (Figs. 23, Supplementary Figure 1, Table 1, and Supplementary Table 3). Prior to calculate microbial diversity, the OTU table was rarefied at 8,700 sequences per sample (samples with <8,700 were omitted) using a “single_rarefaction.py” QIIME script for alpha diversity metrics. Taxa richness measurements (Observed and Chao1) and alpha diversity indexes (Shannon and Simpson) were calculated (Fig. 4A, Table 2, and Supplementary Figure 2). Weighted and unweighted UniFrac distances were used to perform the principal coordinate analysis (PCoA; Fig. 4B). Community diversity was determined by the number of OTUs and beta diversity, measured by UniFrac unweighted and weighted distance matrices in QIIME software v1.9.0. Analysis of similarities (ANOSIM) and permutational multivariate analysis of variance (ADONIS), were used to determine statistically significant clustering of groups based upon microbiota structure distances. Furthermore, linear discriminant analysis (LDA) effect size (LefSe) program v1.0 was used to perform LDA (scores ≥2; Fig. 5 and Supplementary Table 4) in order to estimate significant effect size of each bacterial taxa between experimental groups [67].

OPEN IN VIEWER

Fig. 2

Relative abundance of bacterial taxa in fecal samples. A) Pie charts of phyla’s relative abundance. The numbers on/beside charts indicate the percentage of abundance. For A, the color of each tag indicates bacterial phyla. Others refers to phyla abundances < 0.1% of relative abundance. B) Relative abundance of bacterial taxa. The color of each tag indicates order, family and genus bacterial taxa. Others refers to taxa <0.5% of relative abundance. WT-M n = 9, WT-F n = 8, Tg-M n = 6, and Tg-F n = 6. Data are shown as mean. Complete abundances and statistical significances are shown in Table 1. F, female; f, family; g, genus; M, male; o, order; Tg, transgenic; WT, wild type.

OPEN IN VIEWER

Fig. 3

Relative abundance of prominent bacterial taxa in fecal samples. Relative abundance values of Bacteroidetes [F (1, 21) = 5.20, 6.93 and 0.92 for interaction, row and column factor respectively], Bacilli [F (1, 21) = 6.38, 9.92 and 0.42 for interaction, row and column factor respectively], Lactobacillales [F (1, 21) = 5.24, 8.15 and 0.29 for interaction, row and column factor respectively], Turicibacterales [F (1, 21) = 5.27, 6.69 and 2.29 for interaction, row and column factor respectively], Clostridiaceae [F (1, 21) = 8.99, 0.44 and 0.64 for interaction, row and column factor respectively], Lactobacillaceae [F (1, 21) = 11.16, 10.98 and 4.26 for interaction, row and column factor respectively], S24-7 [F (1, 21) = 4.96, 6.54 and 0.003 for interaction, row and column factor respectively], Turicibacteraceae [F (1, 21) = 5.20, 6.60 and 2.34 for interaction, row and column factor respectively], Klebsiella [F (1, 25) = 0.05 for interaction, row and column factor], Lactobacillus [F (1, 22) = 7.59, 8.67 and 0.30 for interaction, row and column factor respectively], Lactococcus [F (1, 21) = 7.50, 11.60 and 8.23 for interaction, row and column factor respectively] and SMB53 [F (1, 21) = 20.96, 1.55 and 0.17 for interaction, row and column factor respectively] present in fecal samples are shown. WT-M n = 9, WT-F n = 8, Tg-M n = 6, and Tg-F n = 6. White color circles within bars indicate experimental subjects. Data are shown as mean. Data between square brackets indicate F (DFn, DFd) values of two-way ANOVA with Tukey post-hoc correction. Statistical significances are shown as ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, and ^****p < 0.0001. Complete abundances and statistical significances are shown in Table 1 and Supplementary Table 3. F, female; M, male; Tg, transgenic; WT, wild type.

OPEN IN VIEWER

Fig. 4

Fecal microbiota bacterial diversity. Alfa diversity analyses for experimental groups are shown in (A) by observed number of taxa, Chao1 richness [F (1, 21) = 2.74, 0.0001 and 0.01 for interaction, row and column factor respectively], Shannon index [F (1, 21) = 0.38, 1.19, and 8.02 for interaction, row and column factor respectively] and Simpson index [F (1, 21) = 0.09, 1.15 and 8.65 for interaction, row and column factor respectively]. Beta diversity analyses for experimental groups plotted as principal coordinates analysis (PCoA) in (B) of weighted and unweighted variables. WT-M n = 9, WT-F n = 8, Tg-M n = 6, and Tg-F n = 6. For A, white color circles within bars indicate experimental subjects. For B, each subject is represented by a color sphere. Data are shown as mean±SD bars. For A, data between square brackets indicate F (DFn, DFd) values of two-way ANOVA with Tukey post-hoc correction and statistical significances are shown as ^*p < 0.05, ^**p < 0.01, ^***p < 0.001 and ^****p < 0.0001. For B, p values for PCoA schemes with significative clustering are showed according to the following results: ADONIS: R = 0.07, p = 0.33 and ANOSIM: R = 0.02, p = 0.34 for weighted and ADONIS: R = 0.10, p = 0.05 and ANOSIM: R = 0.16, p = 0.06 for unweighted variables comparing WT mice; ADONIS: R = 0.18, p = 0.08 and ANOSIM: R = 0.11, p = 0.14 for weighted and ADONIS: R = 0.09, p = 0.47 and ANOSIM: R = –0.01, p = 0.46 for unweighted variables comparing Tg mice; ADONIS: R = 0.10, p = 0.18 and ANOSIM: R = 0.06, p = 0.24 for weighted and ADONIS: R = 0.09, p = 0.20 and ANOSIM: R = 0.11, p = 0.14 for unweighted variables comparing M mice; and ADONIS: R = 0.10, p = 0.23 and ANOSIM: R = 0.15, p = 0.07 for weighted and ADONIS: R = 0.09, p = 0.08 and ANOSIM: R = 0.26, p = 0.01 for unweighted variables comparing F mice. F, female; M, male; Tg, transgenic; WT, wild type.

OPEN IN VIEWER

Fig. 5

Fecal enriched bacterial taxa associated to gender and genotype. The figure shows the results of the linear discriminant analysis (LDA) effect size (LEfSe) by gender (A) and genotype (B) comparison of differentially abundant bacterial taxa (orders, families and genera) significantly associated with WT-M (n = 9), WT-F (n = 8), Tg-M (n = 6), and Tg-F (n = 6). For A and B, the color tags on the top right side indicate experimental groups. LDA score (shown at logarithmic scale underneath the bars and indicate with vertical dotted lines) cutoff of 2 was used to discriminate bacterial taxa. Specific LDA scores and statistical significances are shown in Supplementary Table 4. F, female; f, family; g, genus; M, male; o, order; Tg, transgenic; WT, wild type.

Metagenome prediction

We used Phylogenetics Investigation of Communities by Reconstruction of Unobserved States (PICRUSt v.1.1.1) in order to predict related meta-bolic profiles of the metagenomes from 16S rDNA gene data set by Kyoto Encyclopedia of Genes and Genomes (KEGG) orthologs classification database at hierarchy level 3 pathways. Statistical Analysis of Taxonomic and Function (STAMP v2.1.3) software was used to determinate significant differences in abundance of OTUs related to metabolic pathways (Fig. 6 and Supplementary Table 5) [68].

OPEN IN VIEWER

Fig. 6

Prediction of the functional metagenome of the fecal bacterial microbiota. The figure shows graphic representations of significant predicted metabolic pathways determined by PICRUSt v1.1.1 regarding the corresponding bacterial OTU table. Y-axis shows relative abundance to corresponding pathway according to bacterial taxa composition for each experimental group (WT-M n = 9, WT-F n = 8, Tg-M n = 6, and Tg-F n = 6). Data are shown as mean±SD bars. Statistical significances calculated by t-test are shown as ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, and ^****p < 0.0001. F, female; M, male; Tg, transgenic; WT, wild type.

Correlations between determined parameters

Associations between cognition, SCFAs concentration and proportion, bacterial taxa relative abundance, microbial diversity, and proportion of predicted metabolic pathways-related sequences were assessed by Pearson’s correlation coefficient (Graph-Pad Prism version 6) (Fig. 7).

OPEN IN VIEWER

Fig. 7

Correlations between determined parameters. Pearson’s correlation between butyrate concentration and spontaneous alternation proportion, butyrate concentration and discrimination index, acetate concentration and discrimination index, acetate concentration and open arm time proportion, propionate concentration and spontaneous alternation proportion, propionate concentration and discrimination index, Ruminococcaceae relative abundance and spontaneous alternation proportion, butyrate concentration and Simpson index, butyrate concentration and Shannon index, propionate concentration and propanoate metabolism predicted pathway, propionate concentration and type I diabetes mellitus predicted pathway, and Phe, Tyr and Trp biosynthesis predicted pathway and Lachnospiraceae relative abundance are shown. WT-M n = 9, WT-F n = 8, Tg-M n = 6, and Tg-F n = 6. Data are shown as mean. One-tailed p values (p), Pearson’s correlation coefficients (r), trend (solid line) and 95% confidence interval (dotted line) are shown for each correlation. F, female; M, male; Tg, transgenic; WT, wild type.

Statistical analysis

Data are expressed as the mean±standard deviation (SD). Open arm time, closed arms entries, spontaneous alternations, DI, WM latencies, SCFAs concentration and proportion, GM relative abundances, alpha diversity indexes, food intake, and body weight changes were analyzed by a two-way ANOVA followed by a Tukey post-hoc test. Beta diversity was analyzed by ANOSIM and ADONIS, whereas differences for taxa enrichment, abundance of sequences related to predictive metabolic pathways and correlations between determined parameters were evidenced by LefSe, t- and one-tailed tests, respectively. All results were considered statistically significant at p < 0.05.

Behavioral and cognitive performance

Behavioral tests in female mice were done only at metestrus-/diestrus phase of the estrous cycle. Anxiety was determined as the proportion of time spent in the open/closed arms in the EPM test. Tg-F mice spent significantly less time in the open arms compared to WT-F, whereas time spent in closed arms was higher in Tg-F compared to WT-M. This data indicates that Tg-F mice showed higher anxiety than their WT-counterpart, which was not observed in males (Fig. 1A, B). Working memory was assessed by the T-maze test. Percentage of spontaneous alternations (%) was higher in WT-F compared to WT-M, Tg-M, and Tg-F mice. WT-M also showed better working memory than Tg-M mice (Fig. 1C). Object recognition memory was determined by the NOR test. Impaired recognition memory was observed in both Tg-M and Tg-F mice, as they presented lower DIs compared to their WT-counterparts (Fig. 1D). Spatial-learning and spatial-memory were assessed by the WM test. Learning curves are shown in Fig. 1E. Average time to find the scape platform during all learning trials indicate that Tg-M mice had higher latencies compared to their WT-counterpart (Fig. 1E, F). To determine effective learning, we compared the latencies to find the scape platform in trial 1 (E1) compared to trial 12 (E12). We observed that only WT mice (both, males and females) were able to show an effective learning as E12 <E1, but Tg mice (both, males and females) did not showed significant differences between E12 and E1 (Fig. 1G). Spatial memory test was carried out after learning trials. WT-F showed better spatial memory than WT-M, Tg-M and Tg-F (Fig. 1H). Representative trajectories during learning and memory phases for M and F subjects are shown in Fig. 1I and 1J, respectively.

Short chain fatty acids concentration

SCFAs were quantified in fecal samples. The concentration of acetate was decreased in Tg mice (male: 14939.03 ± 5496.59μg/g, female: 11978.96±5480.01μg/g) compared to their WT counterparts (male: 24776.47±6380.95μg/g, female: 23768.91±3387.26μg/g). In contrast, propionate concentration increased in Tg mice (male: 9765.39±3128.70μg/g, female: 7696.35±5746.66μg/g) compared to their WT counterparts (male: 3056.42±2163.77μg/g, female: 2071.83±742.01μg/g). Butyrate concentration was higher in WT-F mice (5263.79±1455.44μg/g) compared to WT-M (2639.47±1039.08μg/g), Tg-F (1284.66±432.52μg/g), and Tg-M mice (1623.90±558.06μg/g).

Fecal microbiota analysis

To determine whether 6-month-old mice presented sex- and/or genotype-dependent GM dysbiosis, we analyzed the fecal microbiota taxa relative abundance in all experimental groups. At phylum level, we found a higher abundance of Bacteroidetes in Tg-F compared Tg-M (Figs. 2A3, and Table 1). At class level, a higher relative abundance of Bacilli was observed in Tg-M compared to Tg-F (Fig. 3, Supplementary Figure 1, and Supplementary Table 3). At order level, higher abundances of Lactobacillales and Turicibacterales were observed in Tg-M compared to Tg-F mice and compared to WT-F only for Turicibacterales (Fig. 3, Table 1, Supplementary Figure 1, and Supplementary Table 3). At family level, Clostridiaceae abundance was higher in WT-F compared to WT-M mice. Lactobacillaceae relative abundance was higher in Tg-M compared to the rest of the groups. S24-7 showed an increase in Tg-F compared to Tg-M, and Turicibacteraceae was higher in Tg-M compared to Tg-F and WT-F (Fig. 3, Table 1, Supplementary Figure 1, and Supplementary Table 3). At genus level, abundance of Klebsiella was higher in Tg-F compared to WT-F and WT-M, while Lactobacillus was higher in Tg-M compared to Tg-F. Lactococcus showed higher proportions in Tg-M compared to the other three groups, and SMB53 showed a higher abundance in WT-F compared to WT-M and Tg-F, increasing its abundance in Tg-M in comparison to WT-M mice (Figs. 2B3 and Table 1).

Alpha diversity was estimated by the Observed number of taxa, Chao1 richness and Shannon and Simpson indexes (rarefaction curves are shown in Supplementary Figure 2). A decreased bacterial diversity was observed in Tg-M compared to WT-F according to Simpson index (Fig. 4A). No significant differences were found in observed number of taxa, Chao1 richness and Shannon index (Table 2). The clustering the bacterial communities using PCoA of weighted and unweighted variables revealed a sex-dependent segregation between WT mice comparing weighted variables, without statistical clustering for the rest of comparisons (Fig. 4B).

Next, we used a linear discriminant analysis effect size (LEfSe) to determine enriched taxa per condition. Seven taxa were enriched in WT-M versus WT-F (Stenotrophomonas, Parabacteroides, Prevotella, Slackia, Acetobacter, Paraprevotella and Coprobacillus), whereas nine taxa were enriched in WT-F versus WT-M (Clostridiaceae, Ruminococcaceae , SMB53, EW055, Fusibacter, Flexispira, Peptostreptococcaceae, Kaistobacter and Mogibacteriaceae). Additionally, four taxa were enriched in Tg-M versus to Tg-F (Lactobacillus, Dietziaceae, Lactococcus and Variovorax) and four in Tg-F versus Tg-M (Koribacteraceae, Erythrobacteraceae , Koribacter and Stramenopiles) (Fig. 5A and Supplementary Table 4). Also, ten taxa were enriched for WT-M versus Tg-M (Allobaculum, Lachnospiraceae, Ruminococcus, Nitrospira, Ruminococcaceae, Coprococcus, Oscillospira, Lachnobacterium, Blautia and Christiansenellaceae), whereas nine taxa were enriched in Tg-M versus WT-M (Lactobacillus, Clostridiaceae, Adlercreutzia, SMB53, Dorea, Variovorax, Turicibacter, Clostridium and Fusibacter). Regarding females, seven taxa were enriched in WT-F versus Tg-F (Clostridiaceae, Ruminococcaceae, Ruminococcus, Fusibacter, SMB53, Peptostreptococcaceae, and Clostridium), while six enriched taxa were observed in Tg-F versus WT-F (Allobaculum, Adlercreutzia, Koribacter, Dorea, Coprobacillus and Christiansenellaceae; Fig. 5B and Supplementary Table 4).

Metagenome prediction analysis

Functional metabolic pathways related to fecal microbiota abundances was determined by PICRUSt analysis using the OTU table. The main 20 pathways showing statistically significant difference between experimental groups were mostly linked with metabolism, genetic information processing, human diseases, organismal systems, and environmental information processing (Fig. 6 and Supplementary Table 5). Among these, arginine (Arg) and proline (Pro) metabolism, colorectal cancer, Parkinson’s disease, and toxoplasmosis were increased in WT-F with respect to WT-M, while primary bile acid biosynthesis was decreased in WT-F regarding WT-M. On the other hand, carbohydrate digestion and absorption, G protein-coupled receptors, tyrosine (Tyr), and glutathione metabolism were decreased in Tg-F in comparison to Tg-M mice. Protein digestion and absorption was decreased in Tg-M compared to Tg-F. Tg-M presented increases in propanoate, arachidonic acid and pyruvate metabolic related-pathways, and decreases in insulin signaling pathway and phenylalanine (Phe), Tyr and tryptophan (Trp) biosynthesis compared to WT-M. On the other hand, Tg-F presented increases in type I diabetes mellitus and histidine (His) and Trp metabolism, and decreases in naphthalene degradation, DNA repair, and recombination proteins compared to WT-F (Fig. 6).

Correlations analysis

The determine the possible association between behavioral/cognitive performance in the experimental groups versus SCFAs concentration, taxa relative abundance, alpha diversity metrics, and predicted metabolic pathways we used the Pearson’s correlation coefficient test (Fig. 7). Butyrate concentrations correlated positively with working- and object recognition memory with the highest values in WT-F mice, and the lowest in Tg-F mice. Acetate concentration correlated positively with a better object recognition memory and with lower anxiety index, as WT mice showed the higher values and Tg-F mice the lower values. Propionate levels in fecal samples showed a negative correlation with working memory and object-recognition memory, as Tg-M mice showed the highest values. Among all detected taxa, only Ruminococcaceae and Lachnospiraceae families were positively correlated with working memory and Phe, Tyr, and Trp biosynthesis predicted pathway, respectively. Simpson and Shannon indexes were positively correlated with butyrate concentrations, with highest values found in WT-F mice. Finally, propionate concentrations in feces were positively correlated with propanoate metabolism, and Type I diabetes mellitus predicted pathways, with highest values in Tg-M mice (Fig. 7).