All experiments were conducted in accordance with internationally accepted principles for the care and use of laboratory mice and approved by the Animal Welfare Committee of the Oswaldo Cruz Foundation (Licence L-012/2018 A2).

A total of 40 male C57BL/6 mice (8 weeks old) were used, all of which were kept in the central animal facility of the Oswaldo Cruz Foundation. The mice were housed in standard cages with controlled temperature (22 °C ± 1 °C) and a 12-h light/dark cycle (dark phase from 18:00). Mice in the HFHC + CHOL2% group were fed a high-fat diet (60% kcal from fat +2% cholesterol; Pragsoluções, São Paulo, RJ, Brazil) for 12 weeks and were given ad libitum access to a high-carbohydrate drinking solution (45 g/L; 55% fructose, 45% sucrose). The control mice received an adapted purified standard diet (10% kcal from fat; Pragsoluções, São Paulo, Brazil) and water for the same duration. A detailed composition of both experimental diets, including macronutrient sources and the functional classification of carbohydrates, lipids, proteins, fiber, vitamins, and minerals, is provided in Table 1. From the sixth to the 12th week, the two diet groups (control and HFHC + CHOL2%) were randomly divided into two treatment groups. The mice assigned to the vehicle groups continued to receive their respective diets and were orally administered the water for the last 6 weeks (n = 20). The mice in the pyridoxamine groups continued to receive the same diet and were treated with pyridoxamine (200 mg/kg/day) by oral gavage for the last 6 weeks (n = 20).

At the end of the experiment, fasting blood glucose levels were measured using the Accu-Chek Performa glucometer (Roche Diagnostics, Basel, Switzerland). Mice were fasted for 6 h, a small incision of less than 1 mm was made with surgical scissors at the tip of the tail to collect a drop of blood, which was applied to a test strip inserted into the glucometer. Glucose values were expressed in mmol/L.

At the end of the experiment, the following parameters were analysed in the liver: microvascular blood flow, leukocyte rolling and adhesion, and vitamin A-positive cells. Mice were anaesthetised by intraperitoneal injection (ip) of ketamine hydrochloride (100 mg/kg) and xylazine (10 mg/kg). The left lobe of the liver was then exteriorised for microcirculation analysis.

Intravital microscopy: interaction between leukocytes and endothelial cells was assessed by counting labelled leukocytes (0.3 mg/kg rhodamine 6G, intravenous (iv), rolling or adherent to the sinusoids and postsinusoidal venules. The number of HSCs was quantified by endogenous fluorescence of vitamin A using a UV filter system. The number of vitamin A-positive HSCs rolling and adherent leukocytes was counted in an area of 170 μm², which included the sinusoids and post-sinusoidal venules (Olympus BX150WI).

At the end of the experimental protocol and the microcirculation analyses had been performed on the anesthetized mice, a cardiac puncture was performed for blood sampling. The blood samples were centrifuged to obtain serum, which was stored at −80 °C for further analysis. The liver and adipose tissue were removed and weighed, and the liver was also stored at −80 °C for further experiments.

To analyse body fat mass, the subcutaneous inguinal and thoracic WAT deposits and the visceral WAT deposits (retroperitoneal, mesenteric, and epididymal) were dissected postmortem. Subcutaneous WAT was defined as the deposit under the skin in the groin area. Retroperitoneal fat refers to the fat in the abdominal cavity extending from the diaphragm to the pelvis, bounded posteriorly by the transverse fascia and anteriorly by the posterior parietal peritoneum. Mesenteric WAT was defined as an accumulation of fat pads adherent to the intestines. Epididymal WAT was identified as the deposit in the lower abdomen, associated with the head of the epididymis. After dissection, the WAT deposits were weighed using an analytical balance (GR-200; A&D Company Limited, Toshima-ku, Japan). The weight of the visceral WAT was calculated by adding the weights of the retroperitoneal, mesenteric and epididymal deposits.

Serum levels of total cholesterol, HDL, LDL, triglycerides, ALT enzyme activity, and hepatic cholesterol and triglyceride levels were measured using commercial kits (Bioclin System II, Belo Horizonte, Brazil). For the determination of total cholesterol and triglycerides in the liver, 50 mg of homogenised liver tissue was extracted in 1 mL of isopropanol and then centrifuged at 2000xG for 10 min. After centrifugation, the supernatant was separated and analysed. Results were normalised to tissue weight and expressed as mg per g of liver tissue (wet weight).

Samples from the left lateral lobe of the liver were fixed in Milloning formalin (pH 7.2) for at least 48 h, processed using standard histological procedures, and then embedded in paraffin for light microscopic analysis (Nikon, Model 80i and DSRi1 digital camera, Nikon Instruments, Inc., Melville, United States). The histological features of sections stained with haematoxylin and eosin (H&E) were classified into the following categories: Steatosis, inflammation and hepatocellular injury. Each sample was analysed using the stereological point-counting method for measuring liver fat content with STEPanizer software (version 1, Berna, Switzerland). Inflammation and ballooning were analysed by examining the slides under a microscope with ×20 magnification. The NAS (Non-Alcoholic Steatohepatitis Activity Score) was used to classify the stage of liver disease as follows: MASH (score ≥5), borderline (score 3-4), or non-MASH (score 0-2). This classification was based on the sum of the scores of the variables assessed, with the disease score categorised according to the parameters described in Kleiner et al. (2005). For each parameter, five histological sections from seven mice per group were analysed. Masson’s trichrome staining was used to assess fibrosis by quantification of blue stained collagen, using ImageJ software (ImageJ version 1.53e, Madison, WI, United States) for image segmentation. Five histological sections from six mice per group were analysed for each parameter.

After deparaffinisation, histological sections were incubated in a citrate buffer at 60 °C and then treated with 3% H₂O₂. Non-specific binding was blocked with normal horse serum (2.5%). Sections were then incubated overnight in a humidified chamber at 4 °C with primary mouse monoclonal antibodies against α-SMA (1:500, sc-53015, Santa Cruz Biotechnology, Santa Cruz, United States), RAGE (1:200, sc-365154, Santa Cruz Biotechnology, Santa Cruz, United States), and CML (1:200, ab125145, Abcam, Cambridge, United Kingdom). Slides were washed in 1x PBS and treated with a biotinylated secondary antibody, followed by incubation with streptavidin-HRP at room temperature. Diaminobenzidine (DAB) was used as the chromogen. The protocol followed the manufacturer’s recommendations (PK7800 and SK-4100, Vectastain Universal Quick kit, Peroxidase, R.T.U., Vector Laboratories, Newark, United States). Slides were counterstained with Mayer’s haematoxylin for light microscopic analysis. Quantification of immunostaining was performed using the ImageJ software (ImageJ version 1.53e, Bethesda, United States).

Quantification of CML in liver was performed using an enzyme -linked immunosorbent assay (ELISA) kit for general CML (Carboxymethyl Lysine) (ELK7896 – ELK Biotechnology CO., Denver, United States) according to the manufacturer’s instructions.

The concentrations of fluorescent AGE in liver tissue and serum were determined according to the method described by Nakayama et al. (1993). In brief, the fluorescence of the samples was measured at an excitation wavelength of 360 nm and an emission wavelength of 460 nm using a SpectraMax M5 ELISA microplate reader (Molecular Devices, San Jose, United States). A native bovine serum albumin (BSA) preparation (1 mg/mL in NaOH 0.1N) was used as a reference, and its fluorescence intensity was defined as one fluorescence unit. The fluorescence values for the samples, adjusted to a protein concentration of 1 mg/mL using the BCA Protein Assay Kit (Thermo Fisher, Waltham, United States) were expressed in arbitrary units (AU) compared to the native BSA fluorescence.

Total RNA was isolated from the middle lobe of the liver using the RNeasy Mini Kit (Qiagen, Venlo, the Netherlands). Complementary DNA (cDNA) synthesis was synthesised using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, United States) with 1 μg RNA in a final volume of 20 μL. The primers used for the amplification of the target genes are listed in Table 2. The qPCR was performed with the Power SYBR Green PCR Master Mix (Applied Biosystems, Foster City, United States) according to the manufacturer’s instructions using the 7,500 Fast platform (Applied Biosystems, Foster City, United States). The relative expression of target genes was calculated using the ΔΔCt method, normalised to β-actin expression.

To measure the specific AGE CML, CEL and MG-H1 and the dicarbonyls MGO, GO and 3-DG, and lysine, 25 μL of serum was used, following the protocol described by Scheijen et al. (2016). In brief, after deproteinisation and hydrolysis of the samples, they were evaporated under nitrogen until dry and then dissolved in 500 μL of water. For the measurement of lysine, CML, CEL, MG-H1, MGO, GO, 3-DG derivatives and non-derivatised lysine, 10 μL of the hydrolysate was diluted in 800 μL of water and analysed by UPLC MS/MS. The concentrations of protein-bound or free CML, CEL, MG-H1, MGO, GO, and 3-DG were determined by comparison with calibration curves. All protein-bound or free AGE fractions were expressed in mmol lysine to match the protein content in each sample.

The Cytometric Bead Array (CBA) Mouse IL-6, INF-γ, TNF-α e IL-2 Cytokine Kit (BD, Becton Dickinson, United States) was used to analyse the cytokines present in the liver. Liver tissue samples were thawed and lysed by sonication in a phosphate buffer containing a cocktail of protease inhibitors and Nonidet P40, both at a final concentration of 1%. The samples were centrifuged at 500 g for 5 min, and the supernatant was collected. Then, 25 µL of the samples, 25 µL of the capture bead mixture, and 25 μL of the fluorochrome phycoerythrin were incubated for 2 h at room temperature. After washing 500 µL of wash buffer, the samples were centrifuged at 200 g for 5 min. The supernatant was discarded, and the samples were resuspended in a 100 µL wash buffer. Data acquisition was performed using the Cytoflex flow cytometer (Beckman Coulter, United States). Total protein determination for liver was performed using the BCA Protein Assay Kit (Thermo Fisher, Waltham, United States), following the manufacturer’s instructions for normalisation.

Oxidative stress in liver tissue was assessed by analysing lipid peroxidation, reactive oxygen species (ROS) formation and endogenous antioxidant defence mechanisms.

Data were expressed as mean ± standard error of the mean (SEM). Normality of data was tested using the Shapiro-Wilk test. Comparisons between groups were performed using two-way ANOVA followed by Tukey’s post hoc test for multiple comparisons to identify significant group differences. Statistical analyses were performed using GraphPad Prism version 8.0.1 (GraphPad Software Inc., La Jolla, United States). Differences were considered statistically significant at p < 0.05.

To establish a preclinical model of MASLD, mice were fed either a standard diet (Control) or a high-fat, high-carbohydrate diet with the addition of 2% cholesterol (HFHC + CHOL2%) for 12 weeks. From the sixth to the 12th week, the mice in the Control and HFHC + CHOL2% groups were given pyridoxamine, while the other mice received the vehicle.

As detailed in Table 1, the HFHC + CHOL2% diet is characterized by a predominance of saturated fat and a reduced content of complex carbohydrates compared with the control diet. Consistent with its higher energy density, HFHC + CHOL2% feeding resulted in a significant increase in total caloric intake relative to the control diet. Pyridoxamine treatment did not reduce total caloric intake or solid food consumption in HFHC + CHOL2%-fed mice, but was associated with a significant reduction in water intake (Supplementary Figure S1). After 12 weeks of dietary exposure, body and liver weights were significantly increased in the HFHC + CHOL2% group compared with control animals, whereas pyridoxamine treatment in HFHC + CHOL2%-fed mice attenuated these increases (Figures 1A,B). Feeding of HFHC + CHOL2% resulted in an impairment of glycaemic homeostasis, as evidenced by a significant increase in fasting glucose levels compared to the control group. The administration of pyridoxamine in the HFHC + CHOL2% group significantly attenuated this increase (Figure 1C). In addition, the HFHC + CHOL2% group showed an increase in subcutaneous and visceral fat tissue compared to the control group, which was significantly reduced by pyridoxamine treatment in the HFHC + CHOL2% fed-mice (Figures 1D,E).

Hepatic histological analyses were performed to characterise the MASLD phenotype. Analysis of H&E staining showed a significant increase in steatosis in the HFHC + CHOL2% group compared to the control group. Conversely, HFHC + CHOL2%-fed mice treated with pyridoxamine showed significantly lower intracellular lipid accumulation than HFHC + CHOL2% group (Figures 1F,G). The MASLD activity score (NAS score) was applied in all groups, with the HFHC + CHOL2% group achieving a score of 7, corresponding to MASH classification, while the other groups scored between 2 and 3, indicating non-MASH status (Figure 1H). Analyses of liver fibrosis using Masson’s trichrome staining revealed greater collagen deposition in the HFHC + CHOL2% group compared to Control group. In contrast, administration of pyridoxamine in the HFHC + CHOL2% group significantly reduced collagen deposition compared to the HFHC + CHOL2% group (Figure 1I).

Biochemical analysis of the serum showed that the HFHC + CHOL2% group had significantly higher levels of total cholesterol, LDL, and HDL compared to the control group (Table 3). Pyridoxamine treatment effectively lowered total cholesterol and LDL concentrations in the HFHC + CHOL2% group while HDL concentrations remained unchanged compared to the HFHC + CHOL2% group (Table 3). No significant differences were observed in serum triglyceride levels among the groups (Table 3). Serum ALT levels were significantly increased in the HFHC + CHOL2% group compared to controls (Table 3). Hepatic cholesterol and triglyceride levels were also significantly higher in the HFHC + CHOL2% group compared to the control group. Importantly, treatment with pyridoxamine in the HFHC + CHOL2% fed-mice attenuated both hepatic lipid accumulation and serum ALT levels in mice fed HFHC + CHOL2% cholesterol (Table 3).

To investigate the effects of pyridoxamine on hepatic microcirculatory function, we used intravital microscopy and laser speckle contrast imaging (LSCI). The mice in the HFHC + CHOL2% group had significantly lower hepatic basal microvascular perfusion compared to the control group. Administration of pyridoxamine significantly improved hepatic perfusion in the HFHC + CHOL2%-fed mice (Figures 2A,B).

Activation of hepatic stellate cells (HSC) was more pronounced in the HFHC + CHOL2% group compared to the control group, as evidenced by a lower number of vitamin A-positive cells. Treatment with pyridoxamine in the HFHC + CHOL2% group restored the number of quiescent HSCs, indicating an attenuation of fibrogenic activation (Figure 2C). Accordingly, immunohistochemical analysis revealed an increase in α-SMA-positive areas in the HFHC + CHOL2% group, a marker of activated HSCs, which was significantly reduced by treatment with pyridoxamine in the MASH mouse model (Figure 2D).

In addition, leukocyte-endothelial interactions were significantly increased in the HFHC + CHOL2% group, as evidenced by a significant increase in rolling and adherent leukocytes in the sinusoids and post sinusoidal venules, compared with the control group. Treatment with pyridoxamine in the HFHC + CHOL2% fed-mice reduced the recruitment of leukocytes to the liver endothelium (Figures 2E,F).

Next, we analysed the hepatic markers of oxidative stress in MASH. The mice on the HFHC + CHOL2% diet showed a significant increase in malondialdehyde (MDA) and reactive oxygen species (ROS) levels compared to the control group. In addition, a reduction in hepatic catalase activity was observed in mice with MASH compared to the control group. Treatment with pyridoxamine did not alter these oxidative stress parameters (Figures 3A–C).

Given the increased leukocyte recruitment observed in MASH mice, we next analysed the cytokine profile in the liver of the mice treated or not with pyridoxamine. Gene expression analysis revealed that the HFHC + CHOL2% group had upregulation of key proinflammatory mediators, including IL-6, TNF-α and NF-κB, compared with the control group. Treatment with pyridoxamine in the HFHC + CHOL2% group significantly attenuated IL-6 and TNF transcriptional activation (Figures 3D–F). At the protein level, this inflammatory signature was confirmed by increased levels of TNF-α, IFN-γ, IL-6 and IL-2 in the HFHC + CHOL2% group compared to the control group and administration of pyridoxamine in the HFHC + CHOL2% group significantly reduced this inflammatory profile (Figures 3G–J).

To investigate the contribution of AGE/ALE metabolism to the development of MASH, immunohistochemical and ELISA quantifications showed a significant accumulation of CML in hepatic tissue of HFHC + CHOL2% mice compared to the control group, which was unaffected by pyridoxamine (Figures 4A,B,E). In contrast, analysis of fluorescent AGE levels determined by the Nakayama method showed a significant increase in both serum and liver of the HFHC + CHOL2% fed-mice, which was reduced by pyridoxamine treatment in the HFHC + CHOL2%-fed mice (Figures 4C,D).

Next, we examined the hepatic expression of AGE receptors and detoxification systems. RAGE was significantly upregulated in the HFHC + CHOL2% group compared with the control group, as determined by both immunohistochemistry and gene expression analyses. In addition, gene expression analysis revealed an increase in DIAPH-1, a downstream adaptor protein that mediates RAGE-dependent intracellular signalling. Treatment with pyridoxamine in the HFHC + CHOL2% group normalised RAGE expression at both the gene and protein levels, as well as DIAPH-1 gene expression, to levels comparable to those of the control group (Figures 4F–I).In parallel, the HFHC + CHOL2% group showed altered expression of other AGE/ALE-related receptors: the expression of AGE-R1 was suppressed, while the expression of galectin-3 (GAL-3) and CD36 was significantly increased compared to the control group. Treatment with pyridoxamine in the HFHC + CHOL2% fed-mice restored AGE-R1 levels and decreased the expression of GAL-3 and CD36 (Figures 4J–L).

The detoxification enzyme glyoxalase-1 (GLO-1), an important regulator of MGO metabolism, was reduced in the HFHC + CHOL2% group compared to the control group. Treatment with pyridoxamine in the HFHC + CHOL2% group restored hepatic GLO-1 expression to control levels (Figure 4M).

We quantified serum levels of dicarbonyls and AGE by UPLC-MS/MS. While plasma levels of MGO remained unchanged in the groups (Figure 5A), mice in the HFHC + CHOL2% group showed a significant increase in glyoxal (GO) and 3-deoxyglucosone (3-DG) compared to the control group. Treatment with pyridoxamine in the HFHC + CHOL2% group effectively attenuated these increases, indicating inhibition of early glycation intermediates (Figures 5B,C).

In contrast, the HFHC + CHOL2% group showed significantly lower levels of free CML and MG-H1, as well as protein-bound CEL and MG-H1, compared to the control group (Figures 5F,H,I). Pyridoxamine treatment did not significantly alter these reductions in HFHC + CHOL2%-fed mice (Figures 5D–I); however, it reduced free MG-H1 levels and protein-bound CEL and MG-H1 levels in control-fed mice (Figures 5F,H,I).