Galectin-3 in the Gut-liver axis Regulates Autistic-like Behaviors by Mechanism Associated with Cerebral Shank-3+ cell Niches in BALB/c Mice

Galectin-3 stabilizes cell-cell junctions and regulates inammatory pathways in the gut-liver axis. Galectin-3 knockout (Lgals3 −/− ) mice have atypical behaviors by obscure mechanisms. Given that BALB/c mice naturally develop low-sociability, stereotypies and restrict interest, they have been included as autism experimental model. Our major aims were to investigate whether galectin-3 in the gut-liver axis interferes with autistic-like behaviors analyzing BALB/c Lgals3 −/− mice or under partial inhibition of galectin-3 oral intake of cow’s milk for 7 days. Behavioral patterns were assessed using a three-chambers test, open eld, and self-grooming. Histological analysis and immunohistochemistry (Galectin-3, NOS-2, Iba-1, Ki-67, Dll-4, Shank-3, Synaptophysin and Drebrin) were performed in gut, liver, and/or brain. Lgals3 −/− mice amplied stereotypies, social retraction and restrict interest associated with reduction of cerebral Shank-3 + cells. In Lgals3 +/+ mice, cow’s milk intake also amplied atypical behaviors, reduced galectin-3 in enterocytes and Kupffer cells, and disturbed niches of intestinal KI67 + and Dll-4 + cells and hepatic NOS2 + cells. In the brain of milk-treated mice, Iba-1 + microglial cells and NOS2 + Purkinje cells were increased whereas Shank-3 + and Drebrin + Synaptophysin + cells were reduced suggesting, for the rst time, that galectin-3 interferes with autistic behavior. Perhaps, a perspective to new therapies in genetically predisposed individuals to atypical behaviors.


Introduction
Autism spectrum disorder (ASD) is a neurodevelopmental condition characterized by atypical behaviors, including low sociability, restrict interests, stereotypies, and neuroin ammatory conditions 1 . The etiology and pathophysiology of ASD are poorly understood, although is consensus that causative factors are genetic and environmental 2 . The advent of experimental mice models has contributed to identify signaling pathways involved with ASD and associated comorbidity, such as gastrointestinal symptoms 3 .
The absence of reliable biomarker reinforces the necessity to explore experimental models in order to elucidate cellular and molecular mechanisms involved with ASD progression 4 . In this context, BALB/c and Shank-3 de cient (Shank-3 −/− ) mice develop social, behavioral and neuroanatomical patterns compatible with ASD 5, 6 . Shank-3 regulates synaptic functions in the CNS and Shank-3 −/− mice have severe synaptic dysregulation, cortico-striatal dysfunctions and gastrointestinal (GI) disorders correlated with ASD signals 3,6 .
GI comorbidities are common in autistic individuals, frequently associated with intellectual disabilities and behavioral problems 7 . Pathologic intestinal permeability and reduced capacity of hepatic sulfation have been recurrent in these patients 8 , indicating that gut-liver axis can be a therapeutic target to improve GI comorbidities and behavioral-related symptoms. In this context, diets that interfere with gut-liver functions have potential to assuage or exacerbate symptoms in ASD patients with genetic predisposition 9 .
In order to investigate potential gut-liver markers involving extracerebral factors and ASD symptoms, galectin-3 emerges as molecular target. Galectin-3, a β-galactoside binding protein that controls cell-cell and cell-extracellular matrix interactions in distinct tissues 10 , is synthesized by enterocytes, stabilizing desmoglein-2 in desmosomal junctions 11 . Hepatic galectin-3 is produced by Kupffer cells and cholangiocytes but is absent in health hepatocytes 12,13 .
Galectin-3 knockout (Lgals3 −/− ) mice have markedly gut-liver axis disorders, including severe disorganization of intestinal epithelial cells and hepatic progenitor cell niches 14,15 . Recently, it was demonstrated that Lgals3 −/− mice have atypical behavior in home cage movement 16 and impaired emotional response to stress associated with obsessive-compulsive-like disorder 17 . However, galectin-3 is absent in healthy brain of mice 18 . Based on these data, we hypothesize that disturbances on gut-liver axis detected in Lgals3 −/− mice could be associated with atypical behavioral pattern. Here, for the rst time, intestinal and hepatic galectin-3 can be included as experimental target to alleviate comorbidities associated with autistic-like behavioral parameters.

Results
Absence of galectin-3 disturbed behavioral patterns in BALB/c mice To investigate a possible interference of galectin-3 in ASD-associated phenotypes, BALB/c Lgals3 +/+ and Lgals3 −/− mice were submitted to behavioral tests. Repetitive stereotyped behavior was severely increased in Lgals3 −/− mice. These mice expended approximately 20% of time performing stereotyped movements whereas controls used only 5% of experimental time with this behavior (Fig. 1A). In the absence of galectin-3, mice showed limited interest for new object, expending about 25% of time exploring a new object. In contrast, control Lgals3 +/+ mice demonstrated higher interest for new object than Lgals3 −/− mice, using 45% of time exploring a toy (Fig. 1B), indicating that mice lacking galectin-3 have restrict interest.
Parameters of sociability were also evaluated in both mice groups, but they were partially affected by the absence of galectin-3. Social retraction was signi cantly increased in Lgals3 −/− mice and the percentage of time in isolation and/or non-interacting changed from approximately 5% in controls to around 15% of total time in Lgals3 −/− mice (Fig. 1C). On the other hand, both mice groups spent similar percentage of total time seeking social contact with unfamiliar mouse (Fig. 1D). The three chambered test revealed that BALB/c Lgals3 +/+ mice have more interest for toys in detriment to familiar mouse. Instead, Lgals3 −/− mice spent less time on chamber with toy and more time in chamber with familiar mouse (Fig. 1E). Moreover, this test revealed that social contact with familiar mouse was signi cantly increased in the absence of galectin-3 when compared with control mice (Fig. 1E).
The increased repetitive behavior associated with partial changes in sociability in the absence of galectin-3 indicated possible imbalance on Shank-3 pathways, since Shank-3 is a scaffolding protein that stabilizes glutamatergic receptors in CNS 19 . In the brain of control mice, Shank-3 was widely distributed throughout the cortical regions mostly expressed by neurons and glial cells (Fig. 1F). In contrast, Lgals3 −/ − mice showed reduced proportion of Shank-3 + cells (Fig. 1G). Approximately 80% of cortico-cerebral cells were positive to Shank-3 in control mice while only 15% of these cells were marked to Shank-3 in Lgals3 −/ − mice (Fig. 1H). Together, these data suggest that the reduction of Shank-3 + cells in the absence of galectin-3 could be associated with important behavioral disturbances, some of them related to ASD.
Cow's milk supplementation reduces galectin-3 in the gutliver axis of BALB/c mice Given that Lgals3 −/− BALB/c mice showed atypical behaviors, Lgals3 +/+ BALB/c mice were supplemented with cow's milk containing lactose and other glycans that inhibit galectin-3 functions. On day 7 of treatment, the gut of control mice was hallmarked by enterocytes strongly positive to galectin-3 mainly in the tip of villi ( Fig. 2A). In contrast, enterocytes of milk-supplemented mice were rarely positive to galectin-3 ( Fig. 2B). The percentage of galectin-3 + cells was signi cantly reduced in the epithelial tissue in milksupplemented mice when compared with control group (Fig. 2C). However, in the lamina propria, the percentage of galectin-3 + cells was similar between both groups of mice (Fig. 2D). These data indicate that milk daily diet signi cantly reduced the levels of galectin-3 in the intestinal epithelial cells of mice.
Milk intake also interfered with hepatic galectin-3 + cells. In controls, galectin-3 was preferentially found in sinusoidal cells with morphology compatible with Kupffer cells. It is important to note that majority, but not all Kupffer cells were positive to galectin-3 in these mice (Fig. 2E). In milk-treated mice, Kupffer cells were poorly stained to galectin-3 and majority of Kupffer cells was negative to galectin-3 (Fig. 2F). The percentage of hepatic galectin-3 + cells was signi cantly reduced in milk-supplemented mice when compared with control group (Fig. 2G). Indeed, the percentage of Kupffer cells expressing galectin-3 was signi cantly reduced after milk treatment (Fig. 2H). It is important to note that galectin-3 + cells were not detected in the cerebral cortex and cerebellum on day 7 after milk treatment (data not shown). It is plausible to attribute the reduction of galectin-3 expressing cells in the gut-liver axis with the intake of milk.
Cow's milk affects histological niches in the gut and liver of BALB/c mice In order to investigate whether reduction of galectin-3 associated with cow's milk intake affects the histological architecture of gut-liver axis, samples of small intestine and liver were analyzed on day 7 of supplementation. Apparently, general villi organization was not affected by milk consumption (Supplementary Fig. 1A and 1B). However, cryptal KI67 + cell distribution was signi cantly disorganized in mice after milk consumption when compared with control group (Fig. 3A and 3B). Moreover, the reduction of galectin-3 in enterocytes and atypical organization of KI67 + cells were correlated with higher concentration of DLL-4 + cells. In the gut mucosa, these DLL-4 + cells were frequently detected on basal regions of the crypts and lamina propria in controls (Fig. 3C). On the other hand, DLL-4 + cells were widely distributed throughout the crypts and more frequent in the lamina propria of milk supplemented mice ( Fig. 3D).
In the liver, the effects of milk consumption were observed in the cytoplasm of hepatocytes. These cells showed signi cant perinuclear vacuolation when preferentially localized in the acinar zone 1, surrounding the portal triads (Supplementary Fig. 1C and 1D). Vacuoles in hepatocytes are indicative of glycogen or lipid deposits. Using PAS staining, it was found that liver of milk-supplemented mice showed signi cant abnormal glycogen accumulation when compared with control group (Fig. 3E and 3F). To associate this parameter with increased susceptibility to hepatic injury, NOS2 + cells were evaluated, and it was observed a robust number of NOS2 + cells in the liver of milk-supplemented mice, compared to control mice ( Fig. 3G-H). Clearly, the consumption of cow's milk affected the homeostasis of intestinal and hepatic cells.

Cow´s milk increased cerebellar IBA-1 + microglial cells and NOS-2 + Purkinje cells
To investigate whether gut-brain axis was affected by cow's milk consumption, SNC samples were analyzed and associated with reduction of galectin-3 + cells in the small intestine and liver. Cow's milk intake did not affect histological parameters in the cerebellum (Supplementary Fig. 2A and 2B) and cerebellar galectin-3 + cells were not detected in control or milk-supplemented mice (data not shown).
However, the percentage of IBA-1 + microglial cells increased after milk consumption. These cells represented approximately 4% of total cerebellar cells in controls and 10% in milk-treated mice ( Fig. 4A-C).
The possible induction of oxidative stress in the liver after milk consumption was critical to suggest systemic signals. Thus, NOS2 + cells were investigated in the cerebellum of these mice. In fact, milk consumption was associated with signi cant increase of percentage of Purkinje cells expressing NOS2 + .
These cells represented approximately 30% of total cerebellar cells in controls whereas the number increased to 65% in cow's milk supplemented mice (Fig. 4D-F). Together, these data indicate that bovine milk intake affects microglial and Purkinje cell niches in the cerebellum of mice. At least in part, these ndings seem to be associated with atypical distribution of galectin-3 in gut-liver axis.
Cow´s milk disturbed microglial cell niches and synapses in cortico-cerebral regions Consistent with cerebellum, the numbers of IBA-1 + cells were signi cantly increased in cerebral cortex of mice supplemented with cow's milk ( Fig. 5A and 5B). In controls, cortico-cerebral IBA-1 + cells represented approximately 5% of total cells whereas in milk-treated mice this percentage increased to 9% (Fig. 5C).
Again, milk intake also did not interfere with histological architecture (Suppl. Figure 2C and 2D) and no cells expressing galectin-3 were found (data not shown).
The organization of synapses were also investigated in mice supplemented with bovine milk. In the cortico-cerebral regions of control mice, synaptophysin + and drebrin + cells were widely dispersed throughout the brain parenchyma, frequently colocalized in the tissue (Fig. 5D). On the other hand, mice supplemented with cow's milk revealed similar levels of synatoptophysin staining, but signi cant reduction of drebrin immunoreactivity. Co-localization images corroborated with drastic reduction of synaptic puncta (Fig. 5E). Using controls as 100% uorescence intensity, it was possible to identify that synaptophysin reactivity was apparently, but not statistically, reduced in the cortex of milk-supplemented mice in comparison with controls. In contrast, the intensity of drebrin decreased near of 90% in mice after milk consumption. Therefore, the levels of formed synapses, as seen by the colocalization of pre (synaptophysin) and postsynaptic (drebrin) showed similar percentual reduction (Fig. 5F).

Lack of galectin-3 was correlated with behavioral changes in mice supplemented with cow´s milk
Given that Lgals3 −/− BALB/c mice showed atypical behavioral pattern and Lgals3 +/+ BALB/c mice reduced the levels of galectin-3 in the gut-liver axis after consumption of milk, it was plausible to investigate a possible correlation between these ndings. Consistently, shank-3 + cells were widely distributed in the cerebral cortex of control mice (Fig. 6A). As observed in Lgals3 −/− mice (Fig. 1), mice supplemented with cow's milk showed an intermediate pattern of Shank-3 + cell distribution (Fig. 6B).
Furthermore, quantitative analysis revealed that near of 80% of cells in control mice were shank-3 + . In contrast, the percentage signi cantly decreased to approximately 40% of cortical cells in mice supplemented with milk (Fig. 6C).
To correlate low levels of galectin-3 in the gut-liver and behavioral changes in Lgals3 −/− mice, milk-treated mice were submitted to behavioral tests. After milk consumption, stereotypies were signi cantly ampli ed in Lgals3 +/+ mice (Fig. 6D). These mice also demonstrated restrict interest for new object after milk consumption when compared with control mice (Fig. 6E). Three chambered tests revealed that milksupplemented mice spend more time with toys than other (familiar) mouse (Fig. 6F). Together, these data linked behavioral changes detected in the absence of galectin-3 with disturbances in the gut-liver axis hallmarked by reduction of galectin-3 in gut-liver axis.

Discussion
For the rst time, it is suggested that intestinal and hepatic galectin-3 + cells interfere with autistic-like behaviors. Total (Lgals3 −/− mice) or partial (Lgals3 +/+ mice supplemented with cow´s milk) absence of galectin-3 in the gut-liver axis ampli ed stereotypies, restrict interest and social retraction in BALB/c mice. These behavioral phenotypes were related with disturbed niches of intestinal epithelial cells, hepatocytes and Kupffer cells in the liver, and shank-3 + , Iba-1 + , NOS2 + and drebrin + cells in the CNS.
BALB/c mice showed signi cant low sociability, self-aggressive behaviors, and stereotypies when compared with neurotypical C57BL/6 mice 5 . Our data revealed that both total absence (Lgals3 −/− BALB/c mice) or reduction of galectin-3 in the gut-liver axis induced by milk supplementation increased atypical behaviors. The role of galectin-3 as immunomodulatory factor in the gut and liver has been attracted the attention of specialists 15,20 .
Galectin-3 colocalizes with desmoplakin-1 and desmoglein-2 contributing to selective permeability in enterocytes 21 . In Lgals3 −/− mice, enterocytes have ultrastructural defects in cytoskeleton and basolateral membrane domains resulting in aberrant tra c of luminal molecules 14 . In the liver, galectin-3 plays hepatoprotective functions in distinct conditions 12,22,23 . Lgals3 −/− mice develop severe liver steatosis in diet models 24 , NASH in aging 25 , low regenerative capacity 15 , and abnormal organization of collagen bers in experimental brosis 26,27 . Maybe, the reduction or total absence of galectin-3 could modify the intestinal permeability allowing the circulation of luminal components with potential to affect CNS functions.
The strategy to reduce galectin-3 in the gut-liver axis of wild type mice seemed e cient, given that mice supplemented with cow's milk reduced the epithelial levels of galectin-3 in the intestinal villus and hepatic Kupffer cells. Cow's milk carries galectin-3 inhibitors, including lactose 28 . Intriguingly, milk consumption caused a signi cant imbalance in intestinal KI67 + and Dll-4 + cell niches, indicating an abnormal turnover of epithelial cells. In corroboration, KI67 + cell niches are disorganized during intervention with milk oligosaccharides and epithelial barrier dysfunctions that permit the abnormal ux of microbiota compounds [29][30][31][32] . Dll-4 normally synthesized by intestinal epithelial cells is signi cantly increased during mucosa in ammation 33,34 . In addition, Dll-4 −/− mice showed reduced niches of KI67 + cells in the gut 35 .
In the liver, milk supplementation was linked to abnormal glycogen accumulation in hepatocytes.
Although glycogen naturally accumulates after milk consumption within 18 h of last supplementation 36 , long-time glycogen deposit and hydropic degeneration are compatible with subacute liver injury 37,38 .
Regarding to increased number of NOS2 + cells, this phenotype has been associated with oxidative stress induced by chemical hepatotoxic processes [39][40][41] .
The atypical pattern of behavior in the absence or partial reduction of galectin-3 in the gut-liver axis led us to investigate the distribution of shank-3 + and Iba-1 + cells, possible oxidative stress sites and the organization of some synaptic proteins in the CNS. Shank-3 stabilizes glutamatergic receptors in synapses 42 and shank-3 −/− mice exhibited neurodevelopmental disturbances in neurons of prefrontal cortex 43 , GABAergic interneuron dysfunctions 44 , and social de cits followed by repetitive stereotyped behaviors 45,46 . In our experimental conditions, reduction or lack of galectin-3 was concomitant with decreased numbers of shank-3 + cell in the cerebral cortex, suggesting that gut-liver axis disorders and atypical behaviors can be correlated by, at least in part, cellular signaling pathways involving galectin-3 and shank-3.
The cerebral cortex of mice marked by partial inhibition of galectin-3 in the gut-liver axis was also characterized by increased numbers of IBA-1 + microglial cells. The same pattern was observed in the cerebellum of these milk-supplemented mice. It is possible that extracerebral in ammation, including in ammatory bowel diseases, activates microglial cells in the CNS [47][48][49] . Particularly in ASD, gliosis and neuroin ammation events have been described in patients and murine models 50,51 . The cerebellum of BTBR T + tf/J (BTBR) mice, an relevant experimental model to study ASD 52,53 , has increased numbers of Iba-1 + cells and NOS2 expression correlated with stereotypies 54,55 . Our data corroborated with these authors since the cerebellum of milk-supplemented mice was also characterized by increased numbers of Iba-1 + microglial cells and NOS2 + Purkinje cells. These data incite a hypothesis that early damages outside of the CNS can be a trigger to initiate behavioral changes.
Pre and postsynaptic proteins were also investigated in mice supplemented with cow's milk. Drebrin + cells were signi cantly reduced in the cerebral cortex of mice receiving milk. Drebrin is an actin-binding protein present in postsynaptic glutamatergic neurons whose concentration in dendritic spines organizes cytoskeleton and morphology of the cells. Dysfunctions were associated with Alzheimer's disease, schizophrenia and ASD related to post synaptic accumulation of glutamatergic receptors 56,57 . The reduction of shank-3 + and drebrin + cells in the CNS can be directly associated with behavioral changes observed after milk consumption. The distribution of galectin-3 + cells in the gut-liver axis and cortical shank-3 + and drebrin + cells could be monitored in other ASD experimental models. Perhaps, some genetic predisposition is required to establish this phenotype and more detailed studies are necessary.
In our scenario, we suggest that galectin-3 may be related to the onset of ASD symptoms as co-factor together environmental and genetic causes still for the most part unknown. Coherently with our observation in this study, we can hypothesize that the role of galectin-3 could be crucial at different levels. At early stages, during neurodevelopment and subsequently regulating the synaptic plasticity network pruning and maintenance not directly but through several mediators as BNDF, GABA-AR, NGF, Neurtrophin-3, Synaphtophysin A and Drebrin 58,59 . At peripheral level, this study con rms the key role of galectin-3 as cross-link factor between innate immunity, visceral sensitivity and gut tissues development and homeostasis. The delicate balance between these functions and the in uence of dietary habits should be accurately investigated in order to de ne every molecular pathway involved in the onset of behavioral, cognitive and neurological disturbances associated with autism.
In conclusion, Lgals3 −/− mice and Lgals3 +/+ mice supplemented with cow's milk showed atypical behaviors compatible with autistic-like symptoms. Considering that BALB/c Lgals3 +/+ mice have been used as experimental model of ASD, for the rst time, it is suggested that galectin-3 interferes with gastrointestinal comorbidities and autistic behavior. Data revealed that absence or reduction of galectin-3 in the gut-liver axis increased stereotyped movements, affected sociability indexes and ampli ed restrict interest for objects. However, it is important to reinforce that cow's milk did not cause ASD. Although galectin-3 + cells were not detected in CNS of both mice groups, it has been suggested that galectin-3 in the gut-liver axis interferes with brain functions associated with ASD symptoms in genetically predisposed mice. These ndings pointed to mechanisms outside of the CNS can be studied as therapeutic target to alleviate symptoms of gastrointestinal comorbidities associated with autism.

Animals and milk supplementation
Six-week-old male BALB/c Lgals3 +/+ and Lgals3 −/− mice were obtained from the colony bred at the

Histological analysis and immunohistochemistry
Gut, liver and brain of mice were collected and immediately xed in 4% buffered formalin for 72 hours at room temperature. Brain was segmented into cortex and cerebellum fragments. Histological samples were dehydrated, diaphanized, individually embedded in para n, sectioned into 3 µm thick slices, and stained with hematoxylin and eosin dyes 22  Immuno uorescence Para n sections were depara nize with 2 washes with xylene for 15 minutes each. After depara nization, the sections were submitted to a gradient of ethanol solution (100-50%) of 2 minutes each. Then, samples were incubated with PBS 1x/0.05% tween 20 for 30 minutes. After that, the sections were washed with PBS 1x/0.05% tween 20, incubated in citrate buffer at 98 o C for 40 minutes and treated with blocking buffer (3% bovine serum albumin, 5% normal goat serum [Sigma-Aldrich, USA] diluted in PBS 1X/0.05% tween 20). After blocking, the samples were incubated with the primary antibodies overnight, washed 3 times with PBS 1X/0.05% tween 20 and incubated with secondary antibodies. Nuclei were counterstained with DAPI (Sigma-Aldrich, USA). Slices were mounted with DAKO Mounting Media and imaged on a confocal microscope (Leica TCS SPE). Primary antibodies: anti-NOS2 (BD Bioscience, USA), anti-synaptophysin and Drebrin (Milipore, USA). Secondary antibodies: Anti-rat IgG Alexa 488 (BD Bioscience, USA), anti-Mouse 594 for synaptophysin and Anti-Rb 488 for drebrin (Molecular Probes, USA). After capturing 6-8 images per group of cortical external granular layer, the green and red channels were aligned, and the colocalized puncta number was quanti ed using the Puncta Analyzer plug-in in NIH ImageJ as previously described 60 . Experiments were done in duplicate.

Behavioral assays
Prior to behavior tests, mice were acclimated to the testing rooms and platforms over ve consecutive days. After 1 hour in the sound-isolated room test, each mouse was individually adapted inside the testing platforms. To stereotyped motor behavior, each mouse was individually recorded for 15 minutes, 5 frames/second, from a webcam positioned 60 cm in front of a translucent acrylic cage with 10 cm x 10 cm x 15 cm to monitor stereotyped motor attitudes such as paw apping, persistent self-grooming, circling and jumping. The time of each persistent stereotyped motor behavior and summed all time spent on these over the total lmed time were quanti ed in separate data. To sociability essays, each mouse was acclimated for 10 minutes in the three chambered test platform without novel colored cubic toy and same cage mouse, or 15 minutes inside the box with same cage animals. Each group of mice at a time were placed inside an open 90 cm x 45 cm x 7 cm box lled with nesting material and they were recorded, 5 frames/second for 15 minutes, from a webcam situated 60 cm above the box. Attitudes of social interaction and social play such as anal sni ng, nose sni ng, pouncing and pinning were recorded for each animal, and the seconds spent on these behaviors were quanti ed over the total lmed time. Social retraction which means the condition of non-interaction plus isolation from group of a vigil and not moving mouse was also counted and the percentage of time over the total recorded time was likewise quanti ed. To assess cognition in terms of general sociability and interest in social or spatial novelty was performed using one testing mouse which would be presented at the same time to (1) a cage containing a mouse already recognized by the testing mouse which is native from the same cage and from the same group; and (2) a cage containing a never presented before colored cubic toy. To check the sociability and social-spatial interest time, we counted the time spent by the testing mouse near the cage with the already known same cage mice or with the never presented colored cubic toy (near understood as 5.0 cm from the testing mouse nose tip pointing to one of these cages edge direction) or the time spent by the testing mouse touching, sni ng and exploring one of these cages. This experiment was recorded for 10 minutes by a webcam, 5 frames/second, positioned 1.5 m above the cage, and we quanti ed the exploration time over the total recorded time.

Statistical analysis
The statistical tests were accomplished using t-test, and signi cance threshold was xed for α = 0.05. Therefore, P values ≤ 0.05 were considered statistical differences. Each experiment was performed using 3-5 mice per group in three independent assays. Figure 3 Histological disorders in the gut-liver axis caused by cow's milk consumption. Gut samples stained by immunohistochemistry to KI67 (A, controls; B milk-treated mice) and DLL-4 (C, controls; D, milk-treated mice). Liver samples were stained with PAS to observe glycogen distribution in controls (E) and milksupplemented mice (F). Hepatic samples were also marked to NOS-2 and analyzed by uorescence microscopy. Data were obtained from three independent experiments.  Photomicrographs of cortico-cerebral Shank-3 expressing cells and behavioral tests of milksupplemented mice. Cells positive to Shank-3 were identi ed in the cerebral cortex of control mice (A) and