Identification and Characterization of a Translational Mouse Model for Blood-Brain Barrier Leakage in Cerebral Small Vessel Disease

Abstract

Blood-brain barrier (BBB) dysfunction is a hallmark of cerebral small vessel disease (cSVD). This study aimed to identify a mouse model that replicates BBB impairment and shares key cSVD risk factors. Transgenic db/db and LDLr^-/-.Leiden mice, both prone to obesity and hypertension, were compared to C57BL/6J controls. BBB leakage was assessed using DCE-MRI and sodium fluorescein (NaFl); cerebral blood flow (CBF) by MRI. Dyslipidemia and vascular inflammation were measured by plasma tests. Tight junction integrity, endothelial dysfunction (glucose transporter 1, GLUT-1) and neuroinflammation were evaluated with immunohistochemistry and PCR. Both transgenic models developed an obese phenotype with hyperinsulinemia, but only LDLr^-/-.Leiden mice showed human-like dyslipidemia. When fed a high-fat diet (HFD) or HFD plus cholesterol, LDLr^-/-.Leiden mice showed reduced CBF, endothelial dysfunction (lowered GLUT-1), elevated vascular inflammation (ICAM-1, VCAM-1, S-selectin), and BBB leakage, as evidenced by DCE-MRI and NaFl, together with reduced ZO-1 and claudin-5 expression. Contrastingly, db/db mice showed endothelial dysfunction without BBB leakage. Neuroinflammation (IBA-1, GFAP) was observed only in LDLr^-/-.Leiden groups, consistent with BBB disruption. These findings indicate that LDLr^-/-.Leiden mice, but not db/db mice, are a promising translational model for studying BBB dysfunction in cSVD, offering insights into disease mechanisms and a platform for therapeutic development.

Keywords: blood–brain barrier dysfunction; cerebral small vessel disease (cSVD); dynamic contrast-enhanced MRI (DCE-MRI); endothelial dysfunction; neuroinflammation.

Figure 1

Study design. Mice were housed in groups in digital ventilated cages (DVC) at arrival at 12 weeks and 32 weeks. Mice on the same diet and of the same strain were randomly assigned to groups and housed in separate cages. All mice were housed under identical conditions with ad libitum access to food and water. The mice of group 1 (WT) and group 2 (db/db) were fed a standard diet (chow) throughout the study. Group 3 (younger LDLr^−/−.Leiden mice) were fed chow from birth to 12 weeks old and switched to a high-fat diet enriched with 1% cholesterol (HFD+C) until 16 weeks old. Group 4 (older LDLr^−/−.Leiden mice) was fed a high-fat diet (HFD) for 12 weeks starting from 24 weeks of age. Thereafter the mice underwent MRI scanning and were then sacrificed for postmortem studies. Abbreviations: high-fat diet (HFD), high-fat diet with 1% cholesterol (HFD+C), magnetic resonance imaging (MRI), sodium fluorescein (NaFl), immunohistochemistry (IHC), quantitative PCR (qPCRs). The figure was created in BioRender. Jia, R. (2025).

Figure 2

Body weight, plasma lipid changes and home-cage activity in different groups. (A) Body weight (group size: WT: n = 10; db/db: n = 11; LDLr^−/−. Leiden+HFD+C: n = 12; LDLr^−/−. Leiden+HFD: n = 12). (B) Relative home-cage activity during the daytime and nighttime was measured in all groups across three days before the MRI scan (group size: WT: n = 8; db/db: n = 4; LDLr^−/−. Leiden+HFD+C: n = 5; LDLr^−/−. Leiden+HFD: n = 5). (C) Heatmaps show the average home-cage activity per day in all groups recorded during the three days before the MRI scan. (D) Plasma insulin was determined by ELISA (group size: WT: n = 11; db/db: n = 11; LDLr^−/−. Leiden+HFD+C: n = 12; LDLr^−/−. Leiden+HFD: n = 12). (E) Plasma total cholesterol and lipoprotein profile, and (G) plasma ApoB concentration and (H) plasma triglyceride (TG) concentration of all groups were determined by enzymatic assays (group size: WT: n = 11; db/db: n = 11; LDLr^−/−. Leiden+HFD+C: n = 12; LDLr^−/−. Leiden+HFD: n = 12). Lipoprotein profiles were analyzed in plasma pools; in the respective fractions (F), cholesterol and (I) TG concentrations were determined and plotted as profiles. Data were presented as mean ± SEM. Box plot with 95% confidence intervals for the mean. * p < 0.05, ** p < 0.01, *** p < 0.001.

Figure 3

Vascular conditions and brain structure changes. (A) Representative high-resolution voxel-wise images of the bregma (−1.94) and (B) quantification of cerebral blood flow (CBF) for each mice group under vasodilation condition and vasoconstriction condition (group size: WT: n = 10; db/db: n = 9; LDLr^−/−.Leiden+HFD+C: n = 12; LDLr^−/−.Leiden+HFD: n = 12). (C) Cortical thickness was measured in different groups (group size: WT: n = 11; db/db: n = 10; LDLr^−/−.Leiden+HFD+C: n = 12; LDLr^−/−.Leiden+HFD: n = 11). (D) Brain volumes and ventricular volumes were calculated with FMRIB Software Library (version 5.0.10) based on threshold. Hippocampal volumes were measured on 6 consecutive sections of hippocampus (bregma −1.34 to −3.40) (group size: WT: n = 11; db/db: n = 11; LDLr^−/−.Leiden+HFD+C: n = 12; LDLr^−/−.Leiden+HFD: n = 12). Immunohistochemical staining for Glucose Transporter 1 (GLUT-1) was performed on the left hemisphere of mice. (E) Representative images of GLUT-1 staining in different ROIs (yellow scale bar = 400 µm; red scale bar = 50 µm). (F) GLUT-1 gene expression measured by qPCR (group size: WT: n = 11; db/db: n = 10; LDLr^−/−.Leiden+HFD+C: n = 12; LDLr^−/−.Leiden+HFD: n = 11). (G) The number of GLUT-1 positive blood vessels per mm² and (H) the percentage of GLUT-1 positive staining per area were representative of capillary density (Group size: WT: n = 11; db/db: n = 11; LDLr^−/−.Leiden+HFD+C: n = 10; LDLr^−/−.Leiden+HFD: n = 11). (I) GLUT-1 amount per vessel represented GLUT-1 transporter density per vessel (group size: WT: n = 11; db/db: n = 11; LDLr^−/−.Leiden+HFD+C: n = 10; LDLr^−/−.Leiden+HFD: n = 11). Plasma concentration of (J) E-selectin, (K) P-selectin, (L) vascular cell adhesion molecule-1 (VCAM-1), (M) intercellular adhesion molecule-1 (ICAM-1), (N) plasminogen activator inhibitor-1 (PAI-1) and (O) leptin measured with ELISA (group size: WT: n = 11; db/db: n = 11; LDLr^−/−.Leiden+HFD+C: n = 12; LDLr^−/−.Leiden+HFD: n = 12). Data were presented as mean ± SEM. * p < 0.05, ** p < 0.01, *** p < 0.001.

Figure 3

Vascular conditions and brain structure changes. (A) Representative high-resolution voxel-wise images of the bregma (−1.94) and (B) quantification of cerebral blood flow (CBF) for each mice group under vasodilation condition and vasoconstriction condition (group size: WT: n = 10; db/db: n = 9; LDLr^−/−.Leiden+HFD+C: n = 12; LDLr^−/−.Leiden+HFD: n = 12). (C) Cortical thickness was measured in different groups (group size: WT: n = 11; db/db: n = 10; LDLr^−/−.Leiden+HFD+C: n = 12; LDLr^−/−.Leiden+HFD: n = 11). (D) Brain volumes and ventricular volumes were calculated with FMRIB Software Library (version 5.0.10) based on threshold. Hippocampal volumes were measured on 6 consecutive sections of hippocampus (bregma −1.34 to −3.40) (group size: WT: n = 11; db/db: n = 11; LDLr^−/−.Leiden+HFD+C: n = 12; LDLr^−/−.Leiden+HFD: n = 12). Immunohistochemical staining for Glucose Transporter 1 (GLUT-1) was performed on the left hemisphere of mice. (E) Representative images of GLUT-1 staining in different ROIs (yellow scale bar = 400 µm; red scale bar = 50 µm). (F) GLUT-1 gene expression measured by qPCR (group size: WT: n = 11; db/db: n = 10; LDLr^−/−.Leiden+HFD+C: n = 12; LDLr^−/−.Leiden+HFD: n = 11). (G) The number of GLUT-1 positive blood vessels per mm² and (H) the percentage of GLUT-1 positive staining per area were representative of capillary density (Group size: WT: n = 11; db/db: n = 11; LDLr^−/−.Leiden+HFD+C: n = 10; LDLr^−/−.Leiden+HFD: n = 11). (I) GLUT-1 amount per vessel represented GLUT-1 transporter density per vessel (group size: WT: n = 11; db/db: n = 11; LDLr^−/−.Leiden+HFD+C: n = 10; LDLr^−/−.Leiden+HFD: n = 11). Plasma concentration of (J) E-selectin, (K) P-selectin, (L) vascular cell adhesion molecule-1 (VCAM-1), (M) intercellular adhesion molecule-1 (ICAM-1), (N) plasminogen activator inhibitor-1 (PAI-1) and (O) leptin measured with ELISA (group size: WT: n = 11; db/db: n = 11; LDLr^−/−.Leiden+HFD+C: n = 12; LDLr^−/−.Leiden+HFD: n = 12). Data were presented as mean ± SEM. * p < 0.05, ** p < 0.01, *** p < 0.001.

Figure 4

BBB integrity in different mouse models. (A) K^trans and (B) V_e of the TOFTS model are two important DCE-MRI parameters to detect BBB permeability (group size: WT: n = 7; db/db: n = 7; LDLr^−/−.Leiden+HFD+C: n = 8; LDLr^−/−.Leiden+HFD: n = 9). (C) The NaFl content normalized to plasma in total brain and different subregions (group size: WT: n = 10; db/db: n = 9; LDLr^−/−.Leiden+HFD+C: n = 10; LDLr^−/−.Leiden+HFD: n = 8). Immunohistochemical staining for zonula occludens-1 (ZO-1) was performed on the left hemisphere of the brain. (D) Representative images of ZO-1 staining in different ROIs (yellow scale bar = 300 µm, white scale bar = 100 µm). (E) The number of ZO-1 positive blood vessels per mm² and (F) the percentage of ZO-1 positive staining per area were representative of ZO-1 positive vascular density (group size: WT: n = 11; db/db: n = 11; LDLr^−/−.Leiden+HFD+C: n = 12; LDLr^−/−.Leiden+HFD: n = 12). (G) ZO-1, (H) occludin, and (I) claudin-5 gene expression measured by qPCR (group size: WT: n = 11; db/db: n = 10; LDLr^−/−.Leiden+HFD+C: n = 12; LDLr^−/−.Leiden+HFD: n = 11). Data were presented as mean ± SEM. * p < 0.05, ** p < 0.01, *** p < 0.001.

Figure 5

Neuroinflammation changes in different groups. Immunohistochemical staining for IBA-1 and glial fibrillary acidic protein (GFAP) was performed on the left hemisphere of the brain. (A) Representative images of IBA-1 immunohistochemistry staining in different ROIs (yellow scale bar = 400 µm; red scale bar = 100 µm). (B) The intensity of IBA-1 represented the amount of IBA-1 (group size: WT: n = 11; db/db: n = 11; LDLr^−/−.Leiden+HFD+C: n = 12; LDLr^−/−.Leiden+HFD: n = 12). (C) IBA-1 gene expression measured by qPCR (group size: WT: n = 11; db/db: n = 10; LDLr^−/−.Leiden+HFD+C: n = 12; LDLr^−/−.Leiden+HFD: n = 11). (D) Representative images of GFAP immunohistochemistry staining in different ROIs (yellow scale bar = 400 µm; red scale bar = 100 µm). (E) The intensity of GFAP represented the amount of GFAP (group size: WT: n = 10; db/db: n = 11; LDLr^−/−.Leiden+HFD+C: n = 11; LDLr^−/−.Leiden+HFD: n = 12). (F) The percentage of GFAP-positive area in different ROIs represented the distribution of activated astrocytes (group size: WT: n = 10; db/db: n = 11; LDLr^−/−.Leiden+HFD+C: n = 11; LDLr^−/−.Leiden+HFD: n = 12). (G) GFAP gene expression measured by qPCR (group size: WT: n = 11; db/db: n = 10; LDLr^−/−.Leiden+HFD+C: n = 12; LDLr^−/−.Leiden+HFD: n = 11). Data were presented as mean ± SEM. * p < 0.05, ** p < 0.01, *** p < 0.001.