MAFLD progression contributes to altered thalamus metabolism and brain structure

Metabolic associated fatty liver disease (MAFLD), commonly known as non-alcoholic fatty liver disease, represents a continuum of events characterized by excessive hepatic fat accumulation which can progress to nonalcoholic steatohepatitis (NASH), fibrosis, cirrhosis, and in some severe cases hepatocellular carcinoma. MAFLD might be considered as a multisystem disease that affects not only the liver but involves wider implications, relating to several organs and systems, the brain included. The present study aims to investigate changes associated with MAFLD-induced alteration of thalamic metabolism in vivo. DIAMOND (Diet-induced animal model of non-alcoholic fatty liver disease) mice were fed a chow diet and tap water (NC NW) or fat Western Diet (WD SW) for up to 28 weeks. At the baseline and weeks 4, 8, 20, 28 the thalamic neurochemical profile and total cerebral brain volume were evaluated longitudinally in both diet groups using 1H-MRS. To confirm the disease progression, at each time point, a subgroup of animals was sacrificed, the livers excised and placed in formalin. Liver histology was assessed and reviewed by an expert liver pathologist. MAFLD development significantly increases the thalamic levels of total N-acetylaspartate, total creatine, total choline, and taurine. Furthermore, in the WD SW group a reduction in total cerebral brain volume has been observed (p < 0.05 vs NC NW). Our results suggest that thalamic energy metabolism is affected by MAFLD progression. This metabolic imbalance, that is quantifiable by 1H-MRS in vivo, might cause structural damage to brain cells and dysfunctions of neurotransmitter release.


Materials and methods
Animals. Male DIAMOND (diet-induced animal model of non-alcoholic fatty liver disease) mice were purchased from Sanyal Biotechnology (Virginia Beach, VA, USA) kept under standard laboratory conditions in a specific-pathogen-free animal facility and maintained at 22 ± 2 °C with alternating 12 h light-dark cycle. The model is a stable isogenic cross between C57BL/6J (B6) and 129S1/SvImJ (S129) mice. Those mice are prone to recapitulate the key physiological, metabolic, histologic, transcriptomic, and cell-signaling changes seen in humans with progressive NASH only if they are fed with a high fat diet, high carbohydrate diet (Western diet, WD) with 42% kcal from fat and containing 0.1% cholesterol with a high fructose-glucose solution (SW, 23.1 g/L d-fructose + 18.9 g/L d-glucose). Conversely, DIAMOND mice fed with a standard chow diet (NC) with normal water (NW) did not develop the disease 6,28 . All the experimental procedures were performed according to protocols approved by the Animal Care of University Magna Graecia of Catanzaro. The experimental procedures were carried out in compliance with the ARRIVE guidelines. All experiments were performed in accordance with the European Commission guidelines (Directive 2010/63/EU) for the animals used for scientific purposes. Study design. Mice of 8-12 weeks of age and weight of 20.54 ± 0.53 g were divided into two groups and fed ad libitum a normal chow diet (NC, Harlan TD.2019) and tap water (NW) or a high fat/high carbohydrate diet (Western diet, WD, Harlan TD.88137) with a high fructose-glucose water solution (SW, 23.1 g/L d-fructose + 18.9 g/L d-glucose) for up to 28 weeks. The choice of the animals and the diet used to develop steatosis and steatohepatitis have been based on previously published studies 6,28 . One day before starting diet regimen, baseline body weight and MRS were assessed. On weeks 4, 8, 20 and 28, thalamic neurochemical profile was evaluated in the two diet cohorts using 1 H-MRS. Animals body weight was assessed weekly. The day of the sacrifice, animals were exposed to inhaled isoflurane prior to being euthanized. Euthanasia was performed by cervical dislocation. The entire liver was removed from the abdominal cavity and weighed. The liver was sectioned in a sagittal plane and placed in containers of 10% formalin, for later histologic processing and analysis. 1 H-MRS. Before MRS acquisition, mice were anesthetized with 4% of isoflurane (Forane, Abbott) vaporized in O 2 (flow: 2 l/min). During acquisition, anesthesia was kept between 1, 5 and 2% to maintain the breathing rhythm (SA Instruments, Inc., Stony Brook, NY, USA) between 50 and 80 breaths per minute. Body temperature was monitored and maintained at 37 °C (SA Instruments, Inc., Stony Brook, NY, USA). 1 H-MRS spectra were acquired with a Bruker Pharmascan 70/16 US 7 T bore MR scanner (Bruker Biospin MRI GmbH, Ettlingen, Germany), equipped with a Bruker's MRI CryoProbe™ with MRI cryocooler, to increase signal-noise-ratio and reach a higher sensitivity than standard room temperature RF coils. 1 H-MRS acquisition begins with a single-slice, three-axis localizer scan for the purpose of center mouse brain in the imaging field of view. Subsequently, voxel was precisely placed in the thalamus (2 × 2 × 2 mm 3 ) according to anatomical landmarks in Axial and Sagittal T2_turboRARE weighted images with fat suppression (TE: 35 ms, TR: 2500 ms, averages: 2, thickness of 0.7 mm, slices 9, field of view 20 × 20 mm 2 , data matrix 256 × 256). The thalamic voxel (2 × 2 × 2 mm 3 ) was positioned based upon referencing a mouse brain atlas. A localized shimming procedure was performed to improve the B 0 field map homogeneity in the region of interest. Then, a variable power and an optimized relaxation delays (VAPOR pulses) were applied as water suppression scheme. 1 H-MRS suppressed and unsuppressed spectra were acquired with a PRESS_1H sequence with the following parameters: TE 16.6 ms, TE1 8.99 ms, TE2 7.61, TRd 2500 ms, averages 256, dummy scans 2, VOI 2 × 2 × 2 mm 3 .
1 H-MRS data analysis was performed using TARQUIN 4.3.10, a new accurate and robust modeling algorithm 33 which allows to quantify the concentrations of metabolites within the voxel. For the absolute metabolite quantifications were used the two Data File provided by the acquisition, the "fid" data file which are referred to the water suppressed spectra and the "fid.refscan" data file which are referred to the water unsuppressed spectra, both coming from the acquisition in the same voxel location. Volumetric analysis. Volumetric   . By the "Draw tool" the ROIs were traced on the MRI sections. To improve the viewing of cerebral margins, a Default WL/WW, "French Clut" and a linear opacity table were used. Total brain's area was defined every 3 slices, starting from the olfactory bulbs up to the last part of the cerebellum. Subsequently, "Generate Missing ROIs" function was used to outline total brain area for all slices. Then, "Compute ROI Volume" feature was used to merge the ROIs of the entire brain and estimate its volume.
Histological analysis. Liver histology was assessed from paraffin-embedded tissue sections stained with hematoxylin and eosin. Histology was reviewed using the NASH-Clinical Research Network (CRN) criteria and fatty liver inhibition of progression (FLIP) algorithm by a liver pathologist. For each liver slide, the main histological lesions were assessed using the Steatosis-Activity-Fibrosis (SAF) score system 6,34,35 . Steatosis was graded on a scale of 0 to 3 (0: < 5%; 1: 5-33%; 2: 34-66%; 3: > 67%). Grade of activity (0-2) is given by the sum between the presence of ballooning and inflammation. Ballooning hepatocytes was graded as 0 (none), 1 (when few hepatocytes presented a rounded shaped, reticulated, and pale cytoplasm, but with normal dimensions), and 2 (when there is a cluster of prominent ballooning hepatocytes).
The presence of inflammatory foci within the lobule or within the sinusoids, was graded as 0 (none), 1 (< 2 foci per 20 × field), 2 (2-4 foci per 20 × field), and 3 (> 4 foci per 20 × field). The NAFLD activity score (NAS) was calculated by addition of grades of steatosis, inflammation, and ballooning 5 . Steatohepatitis has been diagnosed as previously described 28 . Immunofluorescence. Formalin-fixed, paraffin-embedded sections (5 µm thick) were deparaffinized by exposure to xylene and graded alcohols and then washed in water. Epitopes were retrieved by heating the slides to 98 °C for 30 min in 10 mM Sodium Citrate buffer with pH = 6.0. Sections were washed 3 times with phosphate-buffered saline (PBS) 1X and then permeabilized with a PBS 1X/gelatin (0.2% w/v)/Triton (0.25% v/v) solution, twice for 10 min. Subsequently, slides were blocked in a 5% of bovine serum albumin (BSA) (A7906, Sigma Aldrich, Milan, Italy) blocking solution for 1 h at RT and then incubated overnight at 4 °C in a humidified chamber with the 1% BSA-diluted primary antibodies: anti-Ionized calcium-binding adapter molecule 1 (IBA1) antibody, a microglial and macrophage-specific calcium-binding protein involved in the membrane ruffling and phagocytosis in activated microglia, ab178846, Abcam, Cambridge, UK, 1:300; anti-Transmembrane Protein 119 (TMEM119) antibody, a cell-surface protein and a specific microglial marker which has the advantage of distinguishing microglia from macrophages, ab209064, Abcam, Cambridge, UK, 1:200; anti-Glial fibrillary acidic protein (GFAP) antibody, a marker for the activation of astrocytes in the central nervous system, G3893, Sigma Aldrich, Milan, Italy, 1:200; anti-β III Tubulin (TUBB3) antibody which is widely used as a neuronspecific marker, MAB1637, Sigma Aldrich, Milan, Italy, 1:250. After the incubation period, each slide was twice washed with PBS 1X and, after a permeabilization step, incubated with a fluorescent secondary antibody (Alexa Fluor 488 (green), A11029; Alexa Fluor 594 (red), A21203, Thermo Fisher Scientific, Italy) for 1 h at RT. After the washing step, nuclei were stained with DAPI (0.1 µg/ml, D8417, Sigma Aldrich, Milan, Italy) for 10 min at RT.

Immunofluorescence-derived image acquisition and skeleton analysis.
Fluorescence was detected using a confocal laser scanning microscope TCS SP5 (Leica Microsystems, Wetzlar, Germany). Confocal images were acquired using a 63× objective. 180× magnification was used to obtain better details for all processes. For skeleton analysis, IBA1 positive and TMEM119 positive images were used to visualize all microglia processes, GFAP positive images were used to visualize all astrocyte processes, whereas TUBB3 positive images were used to selectively mark the neurons.
Using ImageJ (version 2.1.0/1.53c), the minimum threshold value (IBA1: 90; TMEM119: 100; GFAP: 110; TUBB3: 110) was picked and kept constant between matching NC NW and WD SW thalamic regions. A noise de-speckling was performed to eliminate single-pixel background fluorescence. Then, the resulting images were converted in binary using "Skeletonize" tool. The analysis technique is based on ImageJ plugin AnalyzeSkeleton (2D/3D) (https:// imagej. net/ plugi ns/ analy ze-skele ton/), that provide an analysis of microglia and astrocytes ramification within entire photomicographs. This plugin was applied on the skeletonized images for collect the data regarding total number of process endpoints and summed process length both for microglia and astrocytes. These data were used as measures of microglia, astrocytes, and neurons morphology. In addition, the changes in neuronal network were analyzed using ImageJ performing a measurement of total area of neurons (Soma + neurite) in TUBB3 positive images. Statistical analysis. Data were analyzed with GraphPad PRISM 9.1.2 (GraphPad Software, Inc., La Jolla, CA, USA). All results were expressed as mean ± S.D. (Standard deviation). Normality was tested using Shapiro-Wilk normality test. Normally distributed data were analyzed by one way ANOVA followed by Tukey's test, while data without normal distribution were analyzed using Kruskal-Wallis analysis of variance and subsequent Dunn's tests. The Unpaired Two-tailed Student's t test was used for comparison of data derived from two groups. Value with p < 0.05 were considered statistically significant. Correlation analysis was assessed using Spearman's correlation coefficient, using NAS as pathology marker.

Results
Mice fed a high fat diet and sugar water develop MAFLD. Animals fed a WD SW developed obesity compared to CD NW-fed mice (Fig. 1A). The weight gain was accompanied by an increase in liver weight at all time points (Fig. 1B). www.nature.com/scientificreports/ www.nature.com/scientificreports/ Mice on a high fat diet with ad libitum sugar water administration developed steatohepatitis (Fig. 1C, D), which was characterized by steatosis, lobular inflammation, and hepatocellular ballooning (Fig. 1D, E). The NAFLD activity score (NAS) increased by week 8 and remained higher than NC NW mice by week 28 (Fig. 1E). Specifically, histology of the liver showed an extensive development of steatosis by week 8 in WD SW-fed mice (Fig. 1D, E). At 8 weeks mice had a mean steatosis grade of 2.5 ± 0.3 ( Fig. 1E) with 60 ± 11% cells with steatosis (not shown). This remained nearly constant after 20 weeks, with a mean steatosis grade of 2.7 ± 0.1 (Fig. 1E), and 68.75 ± 4.7% cells with steatosis (not shown). At week 28 weeks mice had a mean steatosis grade of 2.9 ± 0.1 (Fig. 1E) and 81 ± 3.4% cells with steatosis (not shown). Following initiation of WD SW diet, steatohepatitis developed in 2 out of 6 mice (33.3%) and at week 20, 8 out of 8 mice (100%) had NASH with a prominent inflammation ( Fig. 1C-E), whereas at week 28, 8 out of 10 mice (80%) developed steatohepatitis (Fig. 1C). Stage 1 fibrosis was present by week 20 after initiation of the WD SW diet (Fig. 1E). In contrast, none of the animals on chow diet developed MAFLD (Fig. 1C).
Volumetric analysis of control and high-fat diet-fed animals. Volumetric analysis showed the same total cerebral brain volumes (TCBVs) in both animal cohorts at 0 weeks.
Quantifying microglia and astrocytes morphology in immunofluorescent images of fixed thalamic. At the baseline, immunofluorescence analysis showed the same IBA1% positive pixels, number process of endpoints and summed process length in NC NW group and WD SW animals. Likewise, no differences have been observed for TMEM119 and GFAP between NC NW and WD SW groups (Fig. 4A).
Comparisons at specific times showed that IBA1% positive pixels (p < 0.01), number of process endpoints (p < 0.05) and summed process length (p < 0.01) were significantly higher in WD SW group than in NC NW group at 28 weeks (Fig. 4A,B), whereas TMEM119% positive (p < 0.001), number of process endpoints (p < 0.001) www.nature.com/scientificreports/ and summed process length (p < 0.001) were significantly lower in WD SW group than in NC NW group at 28 weeks (Fig. 4A,B).
In the WD SW group, the number of GFAP positive cells (2.828 ± 1.494, p < 0.01), number of process endpoints (1549 ± 641.3, p < 0.01) and summed process length (1940 ± 705.3 μm, p < 0.01) were significant higher after 28 weeks of high-fat diet, (Fig. 4A). Comparisons at specific times showed that GFAP % positive cells (p < 0.05), number of process endpoints (p < 0.01) and summed process length (p < 0.01) were significantly higher in WD SW group than in NC NW animals at 28 weeks (Fig. 4A,B).  www.nature.com/scientificreports/ In the WD SW group, TUBB3 showed a significant decrease after 28 weeks of high-fat diet for number of process endpoints (2761 ± 562.3, p < 0.05), summed process length (2323 ± 454.9, p < 0.01) as well as area of neurons (165.9 ± 32.76, p < 0.01) (Fig. 5B). Comparisons at specific times showed that TUBB3% positive cells (p < 0.01), number of process endpoints (p < 0.05), summed process length (p < 0.01) and area of neurons (p < 0.001) were significantly lower in WD SW group than in NC NW animals at 28 weeks (Fig. 5A,B). 1 H-MRS spectra of the mouse thalamus (Fig. 6A) for the NC NW and WD SW animals underlined the presence of the main brain metabolites between 0.7 and 4 PPM (Fig. 6B and Supplementary Fig. 1A, 1B). The quantification of 1 H-MRS spectra showed similar concentrations at the baseline in both cohorts (Fig. 7).

Discussion
MAFLD includes a wide spectrum of liver diseases 1 , that can start with the accumulation of lipid molecules in hepatocytes and evolve into NASH 5,6 . Despite the main feature is liver dysfunction, the detrimental impact of lipid accumulation can affect the whole metabolic state also predisposing to cardiovascular and neurological diseases 36 . In this view, the aim of our study was to investigate the effects of the altered lipid metabolism during NASH development in the thalamus. In particular, we monitored and quantified over time the putative indicator of inflammation Taurine, and the levels of Glutamate, tNAA, tCho, tCr to study the impact of brain metabolic fluctuation on its energetic status, structure and function.
In patients suffering from MAFLD, a high-fat diet was associated with a range of systemic dysfunctions that include a gain of weight and fasting glucose, abnormal fasting insulin levels, raised lipid biosynthesis in the liver, elevated levels of circulating fatty acids and glucose intolerance 37 . www.nature.com/scientificreports/ The occurrence of these events was studied in depth in an animal model of MAFLD resembling the human features of disease development 38 , confirming that chronic intake of a diet enriched in fats and carbohydrates contributes to induce inflammation and oxidative damage primarily to the hepatic microcirculation and then, at systemic level 39,40 . This suggests a more generalized endothelial dysfunction, also involving blood-brain barrier that, once damaged, permits the infiltration of circulating inflammatory cells into the brain 41 causing neuroinflammation, oxidative stress, metabolic impairment and neurotransmitter transmission deficit 42,43 .
The occurrence of steatosis, lobular inflammation, and hepatocellular ballooning, characterizing WD SWinduced NASH in our experiments, further support the hypothesis of its possible impact on the brain, as shown by the brain volume reduction and thalamus inflammation during the progression of MAFLD. Indeed, at thalamic level, high fat diet consumption determined a progressive gliosis, revealed evaluating morphology and activity of microglia and astrocytes, respectively.
To better define the role of microglia under NASH, a microglial and macrophage-specific calcium-binding protein, named ionized calcium-binding adaptor molecule 1 (IBA1), and the transmembrane protein 119 (TMEM119), expressed on microglia-derived cells but not on recruited blood-derived macrophages 44 were marked.
At the end of the study, our experiments show that although fat diet induced a loss of the resident microglia at thalamic level, as shown by the reduction of TMEM119 positive cells, the concomitant decrease in the process endpoint number and the summed process length highlighted an amoeboid morphology, indicative of a transformation toward the macrophage phenotype. Conversely, a significant increase of IBA1 positive cells was observed starting from 14 weeks. Before sacrifice, this raise became significant in the thalamus of MAFLD mice as compared to control; in addition, IBA1 positive cells showed an enhancement of the process endpoint number and the summed process length, suggesting the differentiation of infiltered-monocytes in a macrophagic phenotype.
Overall, these data suggest that the time-dependent activation of microglia, in mice fed a high fat diet, caused an inflammatory degeneration at thalamic level that, on the other hand, was accompanied by astrocyte activation, as shown by the enhancement of GFAP positive cells, process endpoint number and summed process length. Astrocyte reactive gliosis is recognized as highly heterogeneous state in which astrocyte activities depend on the specific injury. Indeed, they can both sustain regeneration or promote detrimental effects on surrounding cells and brain parenchyma 45 . In this view, our results need further insights because they might reflect a neuroprotective effect to counterbalance exacerbate inflammation or, differently, a contribution to the inflammatory condition amplification.
NASH-induced inflammation, in turn, probably affects brain metabolism, as shown by spectroscopic analysis. In particular, the changes in the metabolic parameters highlighted a time-dependent increase in the concentration of taurine, considered a hypothetical marker of inflammation, up to twentieth week was detected. This gradual increase was accompanied, until the end of the experiment, by an enhancement in total choline which, normally, represents the sum of the levels of glycerophosphorylcholine and phosphorylcholine, both precursors of phosphatidylcholine and sphingomyelin 31,46,47 .
Taken together, these results indicate that the solubilization of glycerophosphorylcholine and phosphorylcholine, probably due to oxidative/inflammatory insults affecting membranes, could be responsible not only for neuronal demyelination, but also for the alteration of plasma membrane permeability and polarization, and for the dysfunction of neurotransmitter vesicular release 29,[48][49][50] .
At the end of the experiment, in thalamus of WD SW mice, increased NAA and glutamate levels were also highlighted, although they were not associated with any changes in glutamine levels. This suggests that, in the presence of increased glucose tolerance and increased levels of circulating fatty acids, as typically observed at that time point in DIAMOND mice 28 , cerebral tissue activates an alternative mechanism to the use of glucose, capable to equally satisfy its energy needs. Indeed, although glucose has always been recognized as the primary source of brain energy, growing evidence shows that other metabolites, such as glutamate and acetate, are used as energy sources, mainly by astrocytes, both in physiological and pathological conditions 51 . In this perspective, the increase in astrocytic glutamate could represent the substrate needed to an anaplerotic reaction aimed to ensure the right homeostasis of Krebs cycle and to the production of necessary lactate for neuronal survival.
The use of glutamate as a mitochondrial substrate [52][53][54] in turn, could justify its vesicular depletion at synaptic level. Consequently, the lack of glutamate release, compared to the unchanged levels of glutamine measured over time, could explain the cognitive deficits characterizing NASH 55 .
In conditions of impaired metabolism, the brain can also use free fatty acids to produce energy 56 .
The main source of free fatty acids crossing the blood brain barrier may come from long-chain fatty acid/ albumin complexes and, to a lesser extent, from circulating lipoproteins 56 . Once inside the astrocytes, the conversion into acetyl-CoA, operated by the acyl-CoA synthetase, allows its translocation into the mitochondrial matrix for β-oxidation and for the production of ketone bodies, such as acetoacetate, beta-hydroxybutyrate and acetone that results from their spontaneous decomposition 57 . Ketone bodies are synthesized starting from two acetyl-CoA molecules also at the peripheral level, mainly by the liver, especially in conditions of decreased glucose bioavailability. Subsequently, they are transported to the extrahepatic tissues, where they are used, after conversion into acetyl-CoA and introduction into the cycle of tricarboxylic acids, for energy production 58 .
Therefore, the ketone bodies produced by astrocytes or coming from the bloodstream in conditions of more marked metabolic alterations migrate within neurons, where they are converted into acetyl-CoA and used in the Krebs cycle. On the other hand, acetyl-CoA excess is converted into NAA and stored inside neuronal mitochondria for satisfying a possible sudden increase in energy needs 59 . This effect, aimed to preserve neuronal function, might represent a protective mechanism against fat-induced cell damage highlighted by the changes in neuronal morphology. Indeed, TUBB3 tubulin staining showed a time-dependent axon degeneration in WD SW mice, characterized by a decrease in total number of the process endpoint numbers and in the summed process length. www.nature.com/scientificreports/