diAcCA, a Pro-Drug for Carnosic Acid That Activates the Nrf2 Transcriptional Pathway, Shows Efficacy in the 5xFAD Transgenic Mouse Model of Alzheimer's Disease

Abstract

The antioxidant/anti-inflammatory compound carnosic acid (CA) is a phenolic diterpene found in the herbs rosemary and sage. Upon activation, CA manifests electrophilic properties to stimulate the Nrf2 transcriptional pathway via reaction with Keap1. However, purified CA is readily oxidized and thus highly unstable. To develop CA as an Alzheimer's disease (AD) therapeutic, we synthesized pro-drug derivatives, among which the di-acetylated form (diAcCA) showed excellent drug-like properties. diAcCA converted to CA in the stomach prior to absorption into the bloodstream, and exhibited improved stability and bioavailability as well as comparable pharmacokinetics (PK) and efficacy to CA. To test the efficacy of diAcCA in AD transgenic mice, 5xFAD mice (or littermate controls) received the drug for 3 months, followed by behavioral and immunohistochemical studies. Notably, in addition to amyloid plaques and tau tangles, a hallmark of human AD is synapse loss, a major correlate to cognitive decline. The 5xFAD animals receiving diAcCA displayed synaptic rescue on immunohistochemical analysis accompanied by improved learning and memory in the water maze test. Treatment with diAcCA reduced astrocytic and microglial inflammation, amyloid plaque formation, and phospho-tau neuritic aggregates. In toxicity studies, diAcCA was as safe or safer than CA, which is listed by the FDA as "generally regarded as safe", indicating diAcCA is suitable for human clinical trials in AD.

Keywords: Alzheimer’s disease; Nrf2 transcriptional pathway; carnosic acid; therapeutic drug for neurodegeneration.

Figure 1

diAcCA treatment rescues deficits in neuronal and synaptic density in 5xFAD mice. Quantitative immunohistochemistry (Q-IHC) of hippocampal sections was performed on 5xFAD and WT littermate mice treated with diAcCA (10 mg/kg, 20 mg/kg, or 50 mg/kg) or olive oil vehicle (Veh). (A) Representative images showing NeuN antibody staining of neurons in WT and 5xFAD mice hippocampal sections. (B) Bar graph quantifying NeuN mean fluorescence intensity (MFI). diAcCA restored NeuN MFI (representing neuronal number) to near normal levels in 5xFAD mice. (C) Representative images showing Synapsin I antibody staining in WT and 5xFAD treatment groups. (D) Bar graph quantifying Synapsin I MFI. diAcCA treatment restored Synapsin I MFI (located in the presynaptic terminal of the synapse) to near normal levels in 5xFAD mice. Each dot on the graph represents the value of brain sections from single mouse and error bars are SEM (n = 4–7 mice/group; 3–6 sections/brain). Scale bar, 500 μm. Significance tested by Student’s t-test of grouped data (* p < 0.05, ** p < 0.005, **** p < 0.0001; ns, not significant).

Figure 2

diAcCA treatment ameliorates AD-related aggregated proteins in 5xFAD mice. Q-IHC was performed on hippocampal sections from WT and 5xFAD littermate mice treated with vehicle (Veh) or diAcCA (10 mg/kg, 20 mg/kg, 50 mg/kg). (A) Representative images showing Aβ(17–24) antibody (clone 4G8) staining of Aβ plaque-like aggregates in WT and 5xFAD treatment groups. (B) Bar graph quantifying percent area (% Ar.) of Aβ aggregates in 5xFAD control and treatment groups. diAcCA treatment significantly reduced the area of Aβ aggregates in 5xFAD mice. (C) Representative images showing pTau antibody (AT8) staining in WT and 5xFAD treatment groups. (D) Bar graph quantifying pTau aggregates in 5xFAD control and treatment groups. diAcCA treatment dramatically reduced the area of pTau aggregates in 5xFAD mice. Each point on the graphs represents the value from brain sections from a single mouse and error bars are SEM (n = 4–7 mice/group; 3 sections/brain). Scale bar, 500 μm. Student’s t-test of grouped data (* p < 0.05, **** p < 0.0001).

Figure 3

diAcCA treatment improves neuroinflammation in 5xFAD mice. Q-IHC was performed on hippocampal sections from WT and 5xFAD littermate mice treated with vehicle (Veh) or diAcCA (10 mg/kg, 20 mg/kg, 50 mg/kg). (A) Representative images showing GFAP antibody (GA5) staining of astrocytes from the WT and 5xFAD treatment groups. (B) Bar graph quantifying GFAP MFI. diAcCA treatment reduced GFAP MFI, representing astrocytic cells in 5xFAD mice. (C) Representative images showing Iba1 antibody staining microglial cells in WT and 5xFAD treatment groups. (D) Bar graph quantifying Iba1 ID. diAcCA treatment improved IBA ID, representing microglial density in 5xFAD mice. Each point on the graphs represents the value from brain sections from a single mouse and error bars are SEM (n = 4–7 mice/group; 3–4 sections/brain). Scale bar, 500 μm. Student’s t-test of grouped data (* p < 0.05, **** p < 0.0001; ns, not significant).

Figure 4

nCounter^® NanoString analysis of 5xFAD mouse brain after diAcCA treatment. (A) Heatmap showing gene expression of selected DAM-related genes in each group. (B–D) Volcano plots showing DEGs for each treatment group compared to vehicle (Veh) control. Grey dots represent genes not significantly affected, while red dots represent upregulated genes, and blue dots, downregulated genes. (E) Relative expression of DAM-related genes C3 and Igf1E, which were significantly lower in 5xFAD-treatment groups compared to 5xFAD-Veh. Values are mean ± SD, significance tested by Student’s t-test (* p < 0.05, ** p < 0.005). (F) GO terms associated with DEGs for 5xFAD-diAcCA treated compared with 5xFAD-Veh. Circle size corresponds to number of genes associated with each GO term (n = 3 for each group).

Figure 5

CA prevents protein aggregate/antibody-induced inflammatory response in hiMG. (A) Representative immunohistochemistry sections of hiMG transplanted into the brains of humanized mice after activation by α-synuclein aggregates plus cognate monoclonal antibody, as in Ref. [36]. CA (2 µM) prevented this effect. Anti-human nuclear antibody (HuNu, blue) was used to identify transplanted hiMG, which were also stained by Iba-1 (green); cleaved caspase-1 staining (red) signified NLRP3 inflammasome activation. (B) Bar graph quantifying immunohistochemical data. Each point represents the value from a section of brain and error bars are SEM (n = 3–8 sections from 17 mice). Log₁₀ transformation normalized the distribution of datapoints, and a Student’s t-test revealed a significant decrease in cleaved/activated caspase-1 integrated density (ID) in response to α-synuclein/Aβ aggregates in the presence of CA (p = 0.0307).

Figure 6

diAcCA treatment improves learning and memory in 5xFAD mice. (A–C) WT and 5xFAD littermate mice were assessed for spatial learning and memory in the Morris water maze probe test after treatment with diAcCA. (A) The 5xFAD vehicle-treated (Veh) and 10 mg/kg diAcCA-treated mice spent significantly less time in the target quadrant and more time in other quadrants compared to WT controls. However, the 5xFAD 20 mg/kg and 50 mg/kg-treated group mice spent a similar percentage of time in the target and other quadrants as the WT controls, signifying dose-dependent improvement. (B) The 5xFAD vehicle-treated, 10 mg/kg, and 20 mg/kg diAcCA-treated mice all spent significantly less time in the platform quadrant zone compared to WT control mice. In contrast, 5xFAD mice treated with 50 mg/kg diAcCA spent similar amount of time in the platform quadrant as WT control mice. (C) The 5xFAD vehicle-treated and 10 mg/kg diAcCA-treated mice took significantly more time to first enter the platform zone compared to WT controls, whereas 5xFAD20 mg/kg and 50 mg/kg-treated groups took a similar amount of time compared to WT controls. (D) On the context test of conditioned fear, 5xFAD-Veh mice, but not mice treated with diAcCA, froze for significantly less time than WT controls. Values are mean + SEM, n = 4–7 mice/group; * p < 0.05 vs. vehicle-treated WT by Fisher’s PLSD test. Note that no statistically-significant difference between WT and 5xFAD mice was noted on cued fear conditioning (Supplementary Figure S3).