Empagliflozin reduces vascular damage and cognitive impairment in a mixed murine model of Alzheimer's disease and type 2 diabetes

Abstract

Background: Both Alzheimer's disease (AD) and type 2 diabetes (T2D) share common pathological features including inflammation, insulin signaling alterations, or vascular damage. AD has no successful treatment, and the close relationship between both diseases supports the study of antidiabetic drugs to limit or slow down brain pathology in AD. Empagliflozin (EMP) is a sodium-glucose co-transporter 2 inhibitor, the newest class of antidiabetic agents. EMP controls hyperglycemia and reduces cardiovascular comorbidities and deaths associated to T2D. Therefore, we have analyzed the role of EMP at the central level in a complex mouse model of AD-T2D.

Methods: We have treated AD-T2D mice (APP/PS1xdb/db mice) with EMP 10 mg/kg for 22 weeks. Glucose, insulin, and body weight were monthly assessed. We analyzed learning and memory in the Morris water maze and the new object discrimination test. Postmortem brain assessment was conducted to measure brain atrophy, senile plaques, and amyloid-β levels. Tau phosphorylation, hemorrhage burden, and microglia were also measured in the brain after EMP treatment.

Results: EMP treatment helped to maintain insulin levels in diabetic mice. At the central level, EMP limited cortical thinning and reduced neuronal loss in treated mice. Hemorrhage and microglia burdens were also reduced in EMP-treated mice. Senile plaque burden was lower, and these effects were accompanied by an amelioration of cognitive deficits in APP/PS1xdb/db mice.

Conclusions: Altogether, our data support a feasible role for EMP to reduce brain complications associated to AD and T2D, including classical pathological features and vascular disease, and supporting further assessment of EMP at the central level.

Keywords: Alzheimer’s disease; Amyloid-β; Cognition; Empagliflozin; Hemorrhage; Microglia; Tau; Type 2 diabetes.

Fig. 1

EMP treatment limits metabolic alterations in db/db and APP/PS1xdb/db mice. a Long-term EMP treatment significantly reduced postprandial glucose levels in diabetic mice. No differences were detected at week 6 ([F_{(7, 36)=}0.864, p = 0.543], statistical power 0.317), although differences were detected from week 10 to week 22 (week 10 [F_{(7, 33)}=18.91], week 14 [F_{(7, 67)} = 32.92], week 18 [F_{(7, 69)} = 31.68], week 22 [F_{(7, 66)} = 25.12]; ^‡‡p < 0.01 vs. control, control-EMP, APP/PS1, APP/PS1-EMP, and APP/PS1xdb/db-EMP; ^††p < 0.01 vs. control, control-EMP, APP/PS1, and APP/PS1-EMP]). b When insulin levels were analyzed, individual weekly assessment revealed that EMP treatment helped to maintain high insulin levels in db/db and APP/PS1xdb/db mice as the disease progresses (week 6 [F_{(7, 75)} = 2.29, p = 0.036], week 10 [F_{(7, 74)} = 4.47], week 14 [F_{(7, 69)} = 5.23], week 18 [F_{(7, 68)} = 5.42], week 22 [F_{(7, 69)} = 4.49], week 26 [F_{(7, 69)} = 6.89]; ^╫╫p < 0.01 vs. control, control-EMP, and APP/PS1]; ^‡‡p < 0.01 vs. control, control-EMP, APP/PS1, APP/PS1-EMP, and db/db; ^††p < 0.01 vs. control, control-EMP, APP/PS1, and APP/PS1-EMP; ^##p < 0.01 vs. control, control-EMP, APP/PS1, APP/PS1-EMP, db/db, and APP/PS1xdb/db). c Body weight was maintained by EMP, as revealed by weekly assessment (week 6 [F_{(7, 75)} = 6.21], week 10 [F_{(7, 73)} = 34.13], week 14 [F_{(7, 76)} = 42.39], week 18 [F_{(7, 77)} = 44.34], week 22 [F_{(7, 77)} = 55.71], week 26 [F_{(7, 77)} = 52.55]; ^††p < 0.01 vs. control, control-EMP, APP/PS1, and APP/PS1-EMP]; ^##p < 0.01 vs. control, control-EMP, APP/PS1, APP/PS1-EMP, and db/db; ^††p < 0.01 vs. control, control-EMP, APP/PS1, and APP/PS1xdb/db; **p < 0.01 vs. rest of the groups) (control n = 13, control-EMP n = 10, APP/PS1 n = 9, APP/PS1-EMP n = 11, db/db n = 11, db/db-EMP n = 10–12, APP/PS1xdb/db n = 9, APP/PS1xdb/db-EMP n = 10)

Fig. 2

Cognitive impairment in APP/PS1xdb/db mice was ameliorated by EMP treatment. a EMP treatment significantly improved the performance in the NOD test for all paradigms under study (“what” [F_{(7, 207)} = 2.88], “where” [F_{(7, 208)} = 3.17], “when” [F_{(7, 214)} = 3.79]; ^##p < 0.01 vs. control, control-EMP, and APP/PS1-EMP; ^††p < 0.01 vs. control, control-EMP, APP/PS1, APP/PS1-EMP, db/db, and db/db-EMP; ^‡‡p < 0.01 vs. control, control-EMP, APP/PS1, APP/PS1-EMP, db/db-EMP, and APP/PS1xdb/db-EMP). b In the MWM, EMP treatment reduced acquisition times in APP/PS1, db/db, and APP/PS1xdb/db mice (day 1 [F_{(7, 298)=}3.85], day 2 [F_{(7, 288)} = 14.33], day 3 [F_{(7, 280)} = 13.27], day 4 [F_{(7, 284)} = 18.24]; **p < 0.001 vs. the rest of the groups; ^††p < 0.001 vs. control, control-EMP, APP/PS1, APP/PS1-EMP, and db/db-EMP; ^╫╫p < 0.01 vs. control, control-EMP, and APP/PS1-EMP; ^₸₸p < 0.01 vs. control and control-EMP). c EMP ameliorated the impairment observed along retention 1 (24 h);, however no statistical differences were detected among the groups [F_{(7, 69)} = 1.65, p = 0.134]. On retention 2 (72 h), reduced cognitive abilities in db/db and APP/PS1xdb/db mice were reversed by EMP treatment [F_{(7, 68)} = 4.40, ^‡‡p < 0.001 vs. control and control-EMP] (control n = 11, control-EMP n = 13, APP/PS1 n = 10, APP/PS1-EMP n = 10, db/db n = 9, db/db-EMP n = 10, APP/PS1xdb/db n = 5, APP/PS1xdb/db-EMP n = 9)

Fig. 3

Brain atrophy was ameliorated by EMP treatment. a Brain atrophy in db/db and APP/PS1xdb/db mice was improved after EMP treatment, when the hemisection size and cortical areas were analyzed (hemisection [F_{(7, 205)} = 6.23]; hemicortex [F_{(7, 191)} = 3.59]; ^††p < 0.01 vs. control, control-EMP, APP/PS1, and APP/PS1-EMP]; ^┬p < 0.01 vs. control). While a similar profile was observed in the hippocampus, differences did not reach statistical significance [F_{(7, 107)} = 0.718; p = 0.657] (control n = 7, control-EMP n = 9, APP/PS1 n = 6, APP/PS1-EMP n = 5, db/db n = 6, db/db-EMP n = 7, APP/PS1xdb/db n = 6, APP/PS1xdb/db-EMP n = 5). b Cortical thickness was also improved after EMP treatment [F_{(7, 419)=}2.53; ^♯♯p < 0.01 vs. APP/PS, db/db-EMP, and APP/PS1xdb/db-EMP]. c Illustrative example of cresyl violet staining where cortical thinning observed in db/db and APP/PS1xdb/db mice is restored by EMP treatment. Scale bar = 125 μm. d NeuN/DAPI ratios were reduced in the proximity and far from SP, and EMP treatment reversed this situation (close to plaques [F_{(3, 933)} = 6.017, ^††p < 0.001 vs. APP/PS1-EMP and APP/PS1xdb/db-EMP; ^‡‡p < 0.01 vs. APP/PS1xdb/db-EMP]; far from plaques [F_{(7, 3029)} = 15.86, ^₸₸p < 0.001 vs. control, control-EMP, APP/PS1, APP/PS1-EMP, db/db-EMP, and APP/PS1xdb/db-EMP; ^##p < 0.01 vs. control-EMP and APP/PS1-EMP]) (control n = 5, control-EMP n = 4, APP/PS1 n = 5, APP/PS1-EMP n = 4, db/db n = 5, db/db-EMP n = 4, APP/PS1xdb/db n = 5, APP/PS1xdb/db-EMP n = 4). e Illustrative example of NeuN/DAPI staining in the proximity and far from SP in the cortex. Scale bars = 25 μm

Fig. 4

Spontaneous bleeding and inflammation were reduced after EMP treatment. a EMP reduced hemorrhage burden in the cortex, and while a similar profile was observed in the hippocampus, differences were not statistically significant (cortex [F_{(7, 262)} = 2.65, ^†p = 0.011 vs. control, control-EMP, APP/PS1, and APP/PS1-EMP]; hippocampus [F_{(7, 118)} = 01.85, p = 0.081]) (control n = 8, control-EMP n = 6, APP/PS1 n = 5, APP/PS1-EMP n = 4, db/db n = 7, db/db-EMP n = 6, APP/PS1xdb/db n = 4, APP/PS1xdb/db-EMP n = 6). b Illustrative example of Prussian blue staining in all groups under study. Green arrows point at individual hemorrhages. Scale bar = 100 μm. c Cortical microglia burden was lower in the proximity of SP in APP/PS1xdb/db mice ([F_{(3, 534)} = 7.036], ^‡‡p < 0.01 vs. APP/PS1 and APP/PS1-EMP, ^┬┬p < 0.01 vs. APP/PS1]), while in SP-free areas, increased burden was reduced after EMP treatment ([F_{(7, 4465)=}137.36, **p < 0.01 vs. the rest of the groups, ^₸₸p < 0.01 vs. control, control-EMP, APP/PS1, APP/PS1-EMP, db/db-EMP, and APP/PS1xdb/db-EMP; ^††p < 0.01 vs. control, control-EMP, APP/PS1-EMP, and APP/PS1xdb/db-EMP, ^##p < 0.01 vs. APP/PS1-EMP and APP/PS1xdb/db-EMP]). A similar profile was observed in the hippocampus, and microglia burden was reduced after EMP treatment in SP-free areas (close to plaques ([F_{(3, 39)}=2.21, p = 0.91]); far from plaques [F_{(7, 768)} = 23.79, ^╫╫p < 0.01 vs. control, control-EMP, APP/PS1, APP/PS1-EMP, and APP/PS1xdb/db-EMP, ^††p < 0.01 vs. control, control-EMP, APP/PS1-EMP, and APP/PS1xdb/db-EMP, ^♯♯p < 0.01 vs. control and control-EMP]) (control n = 5, control-EMP n = 5, APP/PS1 n = 5, APP/PS1-EMP n = 4, db/db n = 6, db/db-EMP n = 4, APP/PS1xdb/db n = 4, APP/PS1xdb/db-EMP n = 5). d Illustrative example of the cortical areas immunostained for Iba-1 (green) and SP (4G8, red). Circles point out representative regions close and far from SP. Scale bar = 125 μm

Fig. 5

Amyloid pathology is reduced by EMP treatment. a SP burden was lower in APP/PS1xdb/db mice when compared with APP/PS1 animals. EMP treatment slightly reduced SP burden in the cortex after TS and 4G8 staining (TS [F_{(3, 107)} = 20.55], 4G8 [F_{(3, 104)} = 11.89]; ^††p < 0.01 vs. APP/PS1 and APP/PS1-EMP). While individual SP size was increased in APP/PS-EMP mice (TS [F_{(3, 5191)} = 8.98], 4G8 [F_{(3, 5039)} = 21.11]; ^‡‡p < 0.01 vs. APP/PS1-EMP; ^††p < 0.01 vs. APP/PS1 and APP/PS1-EMP), EMP treatment reduced SP density (TS [F_{(3, 107)} = 19.74], 4G8 [F_{(3, 107)} = 17.81]; ^‡‡p < 0.01 vs. APP/PS1; ^††p < 0.01 vs. APP/PS1 and APP/PS1-EMP). To a lesser extent, we observed a similar trend when we analyzed SP burden in the hippocampus (TS [F_{(3, 49)} = 8.01], 4G8 [F_{(3, 49)} = 6.08]; ^‡‡p < 0.01 vs. APP/PS1]), SP density (TS [F_{(3, 50)} = 11.57], 4G8 [F_{(3, 50)} = 9.42]; ^‡‡p < 0.01 vs. APP/PS1), and individual plaque size (TS [F_{(3, 643)} = 0.975, p = 0.404]; 4G8 [F_{(3, 638)} = 2.39, p = 0.067]) (APP/PS1 n = 5, APP/PS1-EMP n = 5, APP/PS1xdb/db n = 5, APP/PS1xdb/db-EMP n = 5). b Illustrative example of cortical regions stained with TS and 4G8 where reduced SP density can be observed after EMP treatment. Scale bar = 50 μm. c A similar profile was observed in the hippocampus when we measured SP burden, density, and individual SP size (^‡‡p < 0.01 vs. APP/PS1) (APP/PS1 n = 5, APP/PS1-EMP n = 5, APP/PS1xdb/db n = 5, APP/PS1xdb/db-EMP n = 5). d EMP treatment reduced soluble Aβ40 levels in the cortex ([F_{(3, 27)} = 4.20, ^‡p = 0.015 vs. APP/PS1xdb/db]). While a similar trend was observed in the hippocampus, differences were not statistically significant ([F_{(3, 26)}=0.479, p = 0.7]). No differences were observed for soluble Aβ42 levels in the cortex ([F_{(3, 27)} = 0.832, p = 0.488]) or the hippocampus ([F_{(3, 26)} = 1.02, p = 0.397]). EMP also reduced insoluble Aβ40 levels in the cortex ([F_{(3, 25)} = 5.98, **p = 0.003 vs. the rest of the groups]), while differences did not reach statistical significance for insoluble Aβ40 in the hippocampus ([F_{(3, 22)} = 2.68, p = 0.072]) as well as for insoluble Aβ42 levels (cortex [F_{(3, 27)} = 1.43, p = 0.254]; hippocampus [F_{(3, 26)} = 1.54, p = 0.228]) (APP/PS1 n = 7, APP/PS1-EMP n = 9, APP/PS1xdb/db n = 7, APP/PS1xdb/db-EMP n = 7)