Aβ triggers VDAC1 and p53 overexpression in primary neuronal cultures, leading to apoptosis, which was prevented by VBIT-4
Previously, we have shown that VDAC1 overexpression leads to its oligomerization followed by apoptosis [34, 41, 51], and both are inhibited by the VDAC1-interacting molecule, VBIT-4 (Additional file 1: Fig. S1a). Consistently, in this study, cisplatin applied to the cell line SH-SY5Y of neuronal origin induced VDAC1 overexpression, VDAC1 oligomerization, and cell death, and reduced cell viability (Additional file 1: Fig. S1b–f), which were all inhibited by VBIT-4.
To determine whether Aβ-induced cell death [28, 29, 31] is also associated with VDAC1 overexpression, we used rat hippocampal primary neuronal culture and expressed the human APP gene carrying the Swedish and London mutations (APP swe/lnd ) [52] together with GFP using a viral expression system (Additional file 1: Fig. S1g). In this system, the overproduced Aβ was secreted to the medium (conditioned medium) [53]. An ELISA-based assay indicated that this conditioned medium contained Aβ at 24 ± 1.2 pg/ml (n = 3 cultures).
Neuronal cultures incubated with the conditioned medium (diluted 1:1 with the culture medium) [53] showed a time-dependent increase in VDAC1 level (Fig. 1a,b), which was reduced by VBIT-4 (Fig. 1c,d). Neurons with increased VDAC1 levels also showed increased expression of p53 (Fig. 1e,f). Similar results were obtained by immunoblotting analysis (Fig. 1g). Interestingly, both neurons that were infected to express APP and those that were exposed to the conditioned medium showed high levels of trimeric VDAC1, suggesting that Aβ leads to VDAC1 oligomerization (Fig. 1g) [28].
Fig. 1 Aβ induces VDAC1 overexpression, oligomerization, and apoptotic cell death in primary neural cultures. Primary neural cultures were infected with App swe/lnd-EGFP Sindbis virus for 16 h (Additional file 1: Fig. S1g) to overexpress and secrete Aβ into the medium (conditioned medium, Cond. Med.). a Cells were incubated with and without 50% conditioned medium for 6, 24, and 48 h and IF stained for VDAC1. b Quantification of VDAC1 IF staining. c, d IF staining for VDAC1 of cells incubated for 48 h with 50% conditioned medium in the presence and absence of VBIT-4 (10 μM), and its quantification. e–f Cultured neurons incubated with and without 50% conditioned medium for 24 h were co-immunostained for VDAC1 and p53 (e) and their expression levels were quantified (f). g Primary neural cultures were infected with App swe/lnd-EGFP Sindbis virus for 20 h (Inf.), then conditioned medium (Cond. Med.) was collected, and control neuronal culture was incubated with and without 50% conditioned medium for 48 h and subjected to immunoblotting for VDAC1 and p53. The p53 and monomeric VDAC1 levels are shown in the bottom in relative units (RU). The low exposure (Low Exp.) is presented to show the increase in monomeric VDAC1 levels. The positions of VDAC1 monomers and oligomers and of the molecular weight standards are indicated. h VDAC1 promoter sites that match the sequence profiles generated from Aβ ID decamers [54]. The distance from the VDAC1 transcription start site and the q-value of a motif occurrence are presented. i Immunostaining for activated caspase-3 and VDAC1, in control and conditioned medium treated culture. Quantification of activated caspase-3 levels are shown in (f). j Proposed coupling of VDAC1 overexpression induced by apoptosis stimuli or Aβ and VDAC1 oligomerization forming a large channel, with and without Aβ participation, mediating the release of apoptogenic protein cytochrome c (Cyto c) and apoptosis-inducing factor (AIF) from the intermembrane space (IMS). The VDAC1-interacting molecule, VBIT-4, prevents VDAC1 oligomerization and apoptosis. The functions of VDAC1 in cell life include (blue arrows): control of metabolic cross-talk between the mitochondria and the rest of the cell; transport of Ca2+ to and from the IMS; mediation of cellular energy production by transporting ATP/ADP and NAD +/NADH and fatty acid transport as acyl-CoA (FA-CoA) form, and regulation of glycolysis via binding of hexokinase (HK). The TCA cycle, the electron transport chain (ETC), and the ATP synthase (F0F1) are also presented Full size image
Aβ is proposed to act as a putative transcription factor by binding to the Aβ–interacting domain (AβID) in the DNA sequence (G/T)GG(A/G)(G/T)TGGGG, which is found in APP, BACE1, and APOE promoters [54, 55]. Additionally, soluble Aβ translocates to the nucleus where it regulates gene transcription [54, 55], including activation of the p53 promoter [56]. Here, we identified the AβID consensus sequence at seven sites in the VDAC1 promoter region, with the most significant site being GGGGATGGGG (Fig. 1h, Additional file 1: Table S3). These results suggest that Aβ can enhance VDAC1 expression directly by binding to the VDAC1 promoter or indirectly by activating the p53 promoter.
To determine whether Aβ-induced cell death [28, 29, 31] is associated with VDAC1 overexpression, we analyzed the expression of activated caspase-3 and found that it was increased in cells with increased VDAC1 expression (Fig. 1i,f).
These results together with previous findings [22, 41] suggest that Aβ [18] and certain pathological conditions [37] lead to VDAC1 overexpression and oligomerization, forming a channel large enough for pro-apoptotic proteins and mtDNA to cross the outer mitochondrial membrane (OMM) and to subsequently induce apoptosis and inflammation (Fig. 1j, Additional file 1: Fig. S1h). VBIT-4, by preventing VDAC1 oligomerization, protects against mitochondria dysfunction, apoptosis and inflammation (Fig. 1j, Additional file 1: Fig. S1h).
VBIT-4, by protecting against mitochondria dysfunction, allows VDAC1 to function to control the metabolic cross-talk between the mitochondria and the rest of the cell, and transport of Ca2+ and fatty acid as acyl-CoA mediates cellular energy production by transporting ATP/ADP and NAD +/NADH, and regulates glycolysis via binding of hexokinase (HK) (Fig. 1j). All are important for cell life.
VBIT-4 has a stable metabolic profile, and it crosses the blood–brain barrier
To determine if VBIT-4 can cross the blood–brain barrier and mitigate brain pathology when administered in drinking water (either encapsulated in poly lactic-co-glycolic acid (PLGA)-nano-particles, naked), or by gavage, its concentration in the brain was analyzed using liquid chromatography mass spectroscopy (LC/MS/MS) (Fig. 2a). In a preliminary study, VBIT-4 showed an elimination half-life (PK) of 7.6 h (Fig. 2b) indicating a stable metabolic profile, and high plasma protein binding with the bound compound fraction possibly serving as a reservoir from which a slow release can occur (Additional file 1: Table S4). A single-dose toxicity study for VBIT-4 in rats showed no treatment-related mortality or clinical signs, and no significant changes in hematology or in serum chemistry parameters (Additional file 1: Table S4). Overall, the pharmacokinetics data and in-vivo efficacy of VBIT-4 appear useful in predicting an effective therapeutic dose.
Fig. 2 VBIT-4 improves the cognitive performance of 5 × FAD mice. a Representative LC–MS/MS analysis of VBIT-4 concentration in VBIT-4-treated mouse brain extracts [1]. Control (PBS no peak detected); (2) PLGA nano-particle-VBIT-4 administered in drinking water; (3) PLGA-VBIT-4 administered through gavage; (4) VBIT-4 in drinking water. The retention time (RT) and VBIT-4 concentration in the brain is indicated. b PK profile studied in rats following administration of VBIT-4 by IV (5 mg/kg) and PO (20 mg/kg). The observed PK parameters showed moderate-high oral bioavailability. (F 65%), T 1/2 = 7.6 h, Cmax = 3310 ng/ml, Tmax = 1.33 h, AUCinf = 38,369 h*ng/ml. c Disease progression timeline in 5 × FAD mice and the experimental protocol for VBIT-4-treatment. Mice behavioral tests were performed at the age of 7–7.5 months, about 5 months after initiating VBIT-4 or control (0.36% DMSO) treatment (number of mice in each group is indicated). Effect of VBIT-4 on WT was also tested (n = 8). d, e Performance at RAWM was analyzed at trial 6 at the end of day 1, and was expressed as the number of errors (d) or time it took to reach the platform (e). For (d), a one-way ANOVA yielded a significant difference among the groups [f(3,28) = 5.4, P = 0.005], and Tukey post-hoc analysis revealed that 5 × FAD mice performed more poorly than WT (P = 0.008) and WT-VBIT-4 mice (P = 0.013). 5 × FAD-VBIT-4-treated mice performed much better than the 5 × FAD mice (P = 0.008). For (e), a one-way ANOVA yielded a significant difference among the groups [f(3,30) = 6.9, P = 0.001]. A Tukey post-hoc analysis revealed that 5 × FAD mice performed more poorly than the WT mice (P = 0.001), and the WT + VBIT-4 (P = 0.021); 5 × FAD-VBIT-4 mice performed better than 5 × FAD mice (P = 0.003). f An open field habituation test yielded a significant difference among the groups in the time they spent in a mobile state [f(3,30) = 4.5, P = 0.009]. 5 × FAD mice spent less time than WT mice, and 5 × FAD-VBIT-4 treated mice spent longer than 5 × FAD mice (P = 0.053), and similar to that of WT mice. WT-VBIT-4-treated mice (P = 0.001) spent a longer time than the WT. A one-tailed t-test revealed significantly better performance of WT than 5 × FAD mice (P = 0.013). g Number of entries in a T-maze. A one-way ANOVA yielded a significant difference among the groups [f(3,28) = 4.22, P = 0.014], Tukey post-hoc analysis revealed that 5 × FAD mice performed more poorly than WT mice (P = 0.026), and that the 5 × FAD-VBIT-4-treated mice performed better than the 5 × FAD mice (P = 0.0242). 5 × FAD-VBIT-12-treated mice performed better than 5 × FAD but less than the VBIT-4-treated mice Full size image
VBIT-4 prevents cognitive deterioration in 5 × FAD mice
In previous studies, we demonstrated that VBIT-4 prevents apoptosis-associated processes such as cytosolic Ca2+ elevation and ROS production [41], leading to amelioration of the disease-associated processes in several disease models [26, 42, 43]. Therefore, we tested the effects of VBIT-4 on the AD-like pathology in a 5 × FAD transgenic mouse model carrying five familial AD (FAD) mutations [45]. This model exhibits Aβ plaques at 2 months, synaptic degeneration and neuronal loss beginning at about 4 months, and massive neuronal loss at 8–9 months of age [45]. The experimental timeline of VBIT-4 treatment and its molecular structure, are presented in Fig. 2c. and Additional file 1: Fig. S1a.
To evaluate the effects of VBIT-4 on several aspects of cognitive performance, 5 × FAD mice were given VBIT-4 in drinking water twice a week (20 mg/kg), and at the age of 7–7.5 months, they were subjected to four behavioral tests (radial arm water maze, Y-maze, T-maze, and open field tests) (Fig. 2d–g). Results of the radial arm water maze showed that the 5 × FAD mice made about twice the errors of wild-type (WT) mice and took twice the time to reach the platform (Fig. 2d, e). In contrast, the performance of VBIT-4-treated 5 × FAD mice was similar to that of the WT in both error numbers and time to reach the platform (Fig. 2d, e).
Next, we used the open field habituation test to evaluate long-term non-associative, non-aversive spatial learning [48]. The 5 × FAD mice spent about 25% of the time that the WT spent in a mobile state. When treated with VBIT-4, they performed similarly to the WT mice (Fig. 2f). Interestingly, in this test, VBIT-4 also improved the performance of the WT group, perhaps affecting the exploratory activity and reactivity of the mice to a novel environment.
The T-maze test assesses spatial long-term memory and alternation behavior, including the mouse's ability to recognize and differentiate between a new and a familiar compartment. Here, 5 × FAD mice made significantly fewer entries to the correct arm than the WT mice. The VBIT-4-treated mice had similar entries as the WT group (Fig. 2g). In this test, we also examined another VDAC1-interacting molecule, VBIT-12 [41], and found it to be less effective than VBIT-4 (Fig. 2g, blue bar).
Finally, the Y-maze test revealed that the 5 × FAD mice had fewer correct triads than the WT group, but when treated with VBIT-4, their performance was similar to that of the WT group (Additional file 1: Fig. S2a). Thus, oral administration of VBIT-4 rescued several aspects of cognitive function in the AD-like 5 × FAD mice.
It should be noted that treating WT with VBIT-4 had no effect on the expression of specific markers for astrocytes, microglia, or neurons (Additional file 1: Fig. S2b–e).
The finding that VBIT-4 had no effect in WT healthy mice could be due to the fact that the VDAC1 levels are too low to lead a shift of VDAC1 from monomeric to oligomeric form. Therefore, in WT mice no apoptosis occurs, and VDAC1 mediates the normal metabolic processes. This is supported by a cell-based study showing that VBIT-4 does not interfere with normal mitochondria function [41].
VDAC1 is overexpressed in the neuropil surrounding Aβ plaques in the 5 × FAD mouse brain, and VBIT-4 protects against neuronal loss
Next, mice that underwent behavioral testing were sacrificed, and their brains were used to examine the pathological features of the disease. As pathologies in the hippocampus and cortex are closely associated with AD development [57], we focused on these brain regions. IHC staining of brain sections of 5x FAD mice showed that Aβ was distributed throughout the cortex and hippocampus, with formation of numerous plaques (Fig. 3a, b). Immunostaining showed that VDAC1 in the WT mice was evenly distributed, while the 5 × FAD mice showed punctate staining throughout the sections and strong staining in ring-like structures (Fig. 3c, d). Similar staining patterns for Aβ and VDAC1 were obtained in VBIT-4-treated 5 x FAD mice.
Fig. 3 VDAC1 is highly expressed in the neuropil surrounding the Aβ plaques of the 5 × FAD mouse model. a–d Representative cortical and hippocampal sections from WT and 5 × FAD mice treated and untreated with VBIT-4, IHC stained for Aβ (a, b) or VDAC1 (c, d). Higher magnifications of selected areas are shown within the dashed-line squares. e Confocal IF images of cortical and hippocampal sections from 5 × FAD mice co-IF-stained for Aβ and VDAC1. The over-expressed VDAC1 rings are formed around the Aβ plaques. f–h Quantitative analysis of VDAC1 expression levels in cortical sections outside the plaques (g), (area a in f) and in the neuropil surrounding the Aβ plaques (h), (area b in f); in h, numbers are relative to levels outside of the plaque (a). Results show means ± SEM (n = 3). **P < 0.01, ***P < 0.001, ****P < 0.0001. i Representative Co-IF staining of cortical sections from 5 × FAD mice for VDAC1 and neuronal markers (TUBB3, NeuN, synaptophysin); microglia (IBA1) or astrocytes (GFAP) Full size image
To determine the relationship between Aβ plaques and VDAC1 rings, we carried out co-immunofluorescence (co-IF) staining for Aβ and VDAC1. Results demonstrated that the strongly stained ring-like structures represented overexpressed VDAC1 in the neuropils (composed of mostly unmyelinated axons, dendrites, and glial cell processes that form a synaptically dense region) surrounding the Aβ plaques (Fig. 3e, Additional file 1: Fig. S3a). The intensity of VDAC1 in regions outside the rings (area a in Fig. 3f) was significantly higher in the VBIT-4-treated 5 × FAD mice (1.4-fold) than in the untreated 5 × FAD mice (Fig. 3g). In VDAC1-expressing rings (area b in Fig. 3f), the VDAC1 level was dramatically elevated (by 15-fold) relative to area a, and in the VBIT-4 treated mice, this elevation was significantly smaller (12-fold) (Fig. 3h).
To identify the cell compartments surrounding the Aβ plaques that overexpressed VDAC1, we used four neuron-specific markers: synaptophysin to identify presynaptic terminals; class III beta-tubulin (TUBB3) that stains neuronal cell bodies, dendrites, and axons; neuronal nuclear protein (NeuN) to stain the neuronal somas; and post-synaptic density protein-95 (PSD-95). We also stained for the glial fibrillar acidic protein (GFAP) to identify astrocytes, and for the Ca2+-binding adaptor molecule-1 (IBA-1) to identify microglia (Fig. 3i). The results showed co-localization of VDAC1 in the rings only with synaptophysin and TUBB3, but not with the astrocytic or microglial markers. The result suggested that the overexpressed VDAC1 was in neuronal terminals surrounding Aβ plaques.
Given the effect of VBIT-4 in improving learning and memory in the AD mice, we next analyzed the effect of VBIT-4 on the levels of Aβ plaques and proteins implicated in AD pathology. The hippocampus and cortical areas occupied by Aβ plaques were analyzed following thioflavin-S staining and immunostaining for Aβ (Fig. 4a–d). For both hippocampal and cortical Aβ staining, there was about a 20% decrease in the area occupied by Aβ plaques in VBIT-4-treated relative to untreated 5 × FAD mice.
Fig. 4 Effect of VBIT-4 treatment on the levels of Aβ plaques, and the expression of p-Tau and PrPc in the 5 × FAD brain. a Brain sections from 5 × FAD mice were immunostained for Aβ using anti-Aβ antibodies. The cerebral cortex and hippocampal formation that were analyzed in this study are enlarged in (i) and (ii) panels. The CA1 (cornu ammonis subfield 1), ML (molecular layers), GCL (granule cell layer), and DG (dentate gyrus) are indicated. b Representative thioflavin-S (Thio-s) staining of Aβ plaques in cortical and hippocampal sections from VBIT-4-treated and untreated 5 × FAD mice. c, d Areas occupied by Aβ plaques in the cortex (c) and hippocampus (d), as analyzed from Thio-s or anti-Aβ antibodies (anti-Aβ), are expressed as mean ± SE (n = 5–9 as indicated). e–h IF staining and quantification of p-Tau and VDAC1 (e, f), and of VDAC1 and PrPc (g, h) in cortical sections from WT and VBIT-4-treated and untreated 5 × FAD mice. and their quantification. Results show means ± SEM (n = 3 animals for each group, with IF was performed 2–3 times for each group), **P < 0.01, ***P < 0.001, ****P < 0.0001. NS, not significant. p-value in blue color represents the significance of VBIT-4-treated 5 × FAD mice relative to untreated mice Full size image
We then examined the effects of VBIT-4 on other proteins implicated in AD pathology and found that the level of p-Tau was increased (2.7-fold) in 5 × FAD, but VBIT-4 treatment had no effect on its level (Fig. 4e, f).
The cellular prion protein (PrPc) [58] with proposed neuroprotective effects [59] has been found to be lower in AD than in non-AD individuals [60]. Here, we found about 2.5-fold lower expression of PrPc in 5 × FAD mouse brains than in WT brains, while VBIT-4 treatment increased the PrPc level in 5 × FAD mice to be even higher than that in WT mice (Fig. 4g,h).
Finally, we analyzed the level of islet amyloid polypeptide amylin. Amylin is a peptide hormone synthesized and co-secreted with insulin by pancreatic β cells [61], and mediates toxic effect via mitochondrial dysfunction [62]. The results showed that the amylin expression levels were higher in both the cortex and hippocampus of 5 × FAD mice compared to WT mice, but VBIT-4 treatment had no significant effect on its expression levels (Additional file 1: Fig. S3b–e).
Thus, among the four tested proteins whose expression levels are greatly modified in the context of AD, PrPc and Aβ expression levels were altered by VBIT-4, while p-Tau and amylin were not. The decrease in Aβ level by VBIT-4 may result from the prevention of plaque formation/growth or its increased removal.
VBIT-4 protects against AD-related neuronal loss
To determine the effect of VBIT-4 on neuronal survival in 5 × FAD mice and to further identify the cell compartments surrounding the Aβ plaques that overexpress VDAC1, we used four neuron-specific markers: synaptophysin, TUBB3, NeuN, and PSD-95 (Fig. 5a). IF staining for synaptophysin in both cortex and hippocampus showed threefold decrease in the 5 × FAD mice relative to the WT mice, while the VBIT-4-treated 5 × FAD mice did not exhibit this decrease (Fig. 5b,c). The prevention of synaptophysin decrease by VBIT-4 treatment was confirmed by q-RT-PCR (Fig. 5d).
Fig. 5 Overexpressed VDAC1 around the Aβ-plaques is localized to neurons. a Schematic presentation of four neuronal markers localized to different compartments within a neuron. b–g Co-immunostaining for VDAC1 and synaptophysin (b), TUBB3, (e) of cortical sections from WT, untreated- and VBIT-4-treated 5 × FAD mice. (i) and (ii) are enlargements to show co-localization of synaptophysin or TUBB3 with VDAC1. Quantitative analysis of synaptophysin IF staining (c) and its mRNA levels (d) and of TUBB3 (g). f Cortical sections from VBIT-4-treated 5 × FAD mouse IF with anti-TUBB3 and VDAC1 showing neurons with their terminals reaching the Aβ plaque. Nuclei were stained with DAPI. h, i Cortical sections from untreated and VBIT-4 treated 5 × FAD mice immunostained for PSD-95 and VDAC1 (h) and PSD-95 quantification analysis (i). Arrows point to dendrites with no overexpressed VDAC1. Results show means ± SEM (n = 3–4 mice), **P < 0.01, ****P < 0.0001. P-value in blue color represents the significance of VBIT-4-treated relative to untreated 5 × FAD mice. NS, not significant Full size image
TUBB3 was highly expressed in neurons in WT mice, but was decreased by 2.5 folds in the 5 × FAD mice, and the decrease was prevented by VBIT-4 treatment (Fig. 5e–g), suggesting that VBIT-4 prevented the neuronal loss in the 5 × FAD mice.
Next, staining for the neuronal soma protein NeuN, an RNA-binding protein specific for post-mitotic neurons predominantly associated with cell nuclei [63], was lacking in the VDAC1-overexpressing neuropils surrounding Aβ plaques (Additional file 1: Fig. S4a). In addition, in the 5 × FAD mice, the structures that were stained with either DAPI or NeuN appeared smaller than those in the WT, which might represent apoptotic cells (Additional file 1: Fig. S4a). In VBIT-4-treated 5 × FAD mice, the nuclei and somas of neurons had similar sizes as those of the WT.
Next, we analyzed the effect of VBIT-4 on the expression level of PSD-95, a scaffolding protein involved in the assembly and function of the post-synaptic density complex. This protein is involved in anchoring receptors and ion channels, and plays an indispensable role in signal transmission and, hence, in cognition [64]. PSD-95 expression was decreased in 5 × FAD mice, but the decrease was prevented by VBIT-4 treatment (Fig. 5h, i, Additional file 1: Fig. S4d). Interestingly, most but not all PSD-95-expressing compartments in the Aβ plaques showed co-localization with the overexpressed VDAC1 (Fig. 5h, Additional file 1: Fig. S4d, white arrows). Similar results were obtained in the hippocampus (Additional file 1: Fig. S4b–d).
Taken together, in 5 × FAD mice, the co-localization of overexpressed VDAC1 with synaptophysin and TUBB3 [see Fig. 5b(i), e(ii)] suggests that the “ring” structures surrounding the Aβ plaques contain neuronal terminals overexpressing VDAC1, leading to cell death, and thereby neuronal loss. Treatment with VBIT-4 protected against synaptic and neuronal loss both in the cortex and in the hippocampus.
VBIT-4 inhibits apoptosis in 5 × FAD mice
The overexpression of VDAC1 in the synaptic terminals surrounding Aβ plaques found in this study and previous report of association of VDAC1 overexpression with apoptotic cell death [34] suggest that VDAC1 may induce apoptotic neuronal death [34, 51]. Thus, apoptosis was evaluated by TUNEL staining and IF staining for activated caspase-3 expression, which was shown to be elevated in the brains of severe AD cases [65]. Relative to WT mice, the number of TUNEL-stained cells in the 5 × FAD mice was increased over 3 folds, while in the VBIT-4-treated 5 × FAD mice, the number was significantly reduced (Fig. 6a, b). Activated caspase-3 levels in the 5 × FAD mice were increased by 2.5 folds in both the cortex and the hippocampus, and the levels were greatly reduced in VBIT-4-treated 5 × FAD mice (Fig. 6c, d, Additional file 1: Fig. S5a).
Fig. 6 VBIT-4 treatment of 5 × FAD mice protects against cell death. a Representative TUNEL staining of cortical sections from WT, untreated-, and VBIT-4-treated-5 × FAD mice, with a magnification of the selected area (i). The arrows point to apoptotic cells stained green/yellow; red represents propidium iodine-stained nuclei. b Average number of TUNEL-stained cells per mm2. c–f Confocal images of cortical sections from WT, untreated- and VBIT-4-treated-5 × FAD mice co-immunostained for VDAC1 and activated caspase-3 (c) with a magnification of the selected area (ii), or p53 (e) and their quantifications (d, f). Results show means ± SEM (n = 3 mice), *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. P-value in blue color represents the significance of VBIT-4-treated relative to untreated 5 × FAD mice. NS, not significant Full size image
Consistent with our results from the primary neuronal cultures, the expression level of p53, which regulates cell cycle, apoptosis, and senescence [66], was higher in the cortex and hippocampus of 5 × FAD mice than in WT mice, and that of VBIT-4-treated mice was similar to the WT group (Fig. 6e, f, Additional file 1: Fig. S5b). The p53 level also increased in some Aβ plaque-surrounding compartments, and was co-localized with the overexpressed VDAC1 (Fig. 6e, Additional file 1: Fig. S5b), suggesting mitochondrial localization [67].
VBIT-4 prevents dysregulated metabolism in 5 × FAD mice
Neuro-metabolic dysfunctions leading to neurodegeneration, are associated with impaired glucose transport and metabolism, brain insulin resistance, and age-induced mitochondrial dysfunction [10, 36, 68, 69]. Considering impaired metabolism in AD [10, 36, 68] and VDAC1 regulation of metabolism [21, 24], we evaluated the expression of several metabolism-related proteins in 5 × FAD mice and the effects of VBIT-4 on their expression.
The glucose transporters (Gluts) are differentially expressed in the brain, with Glut-1 expressed in astrocytes, Glut-2 in microglia and neurons, and Glut-3 and insulin-regulated Glut-4 in neurons [70]. Glut-1 and Glut-3 are downregulated in AD [71, 72].
We found that the expression levels of Glut-1, Glut-2, and Glut-4 were downregulated in 5 × FAD mice, but not in VBIT-4-treated 5 × FAD mice (Fig. 7a, b, Additional file 1: Fig. S6). As expected, Glut-1 is expressed in the astrocytic dendritic end-feet near blood vessels, mediating glucose uptake across the blood–brain barrier endothelial cells (Fig. 7a(i), Additional file 1: Fig. S6a (i)). The expression of Glut-1 in the non-blood vessel compartment showed a two-fold decrease in 5 × FAD mice, and this was prevented by VBIT-4 treatment (Fig. 7a, b). Similar results were obtained at the mRNA level (Fig. 7c).
Fig. 7 VBIT-4 treatment of 5 × FAD mice protects against cell metabolic impairments. a Confocal images of cortical sections from WT, untreated- and VBIT-4-treated-5 × FAD mice co-immunostained for glucose transporters: Glut-1 and GFAP, showing localization in astrocytes’ dendritic end-feet touching the blood vessels (white arrows), magnified in (i), Glut-2 co-stained with IBA-1 and Glut-4. b Glu-1,2,4 quantifications. (c) q-RT-PCR analysis of Glut-1 mRNA levels. d–g Cortical sections from the three groups co-stained for VDAC1 and HK-I, with a magnification of the selected area (ii), CS, or ATP synthase (ATPsyn5a) and their quantification in cortex and hippocampus (e–g). h, i Co-staining of Na,K-ATPase and VDAC1 and their quantification. Results show means ± SEM (n = 3), *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. P-value in blue color represents the significance of VBIT-4-treated relative to untreated 5 × FAD mice. NS, not significant Full size image
Glut-2, expressed in the microglia and other neuronal cells, was reduced in 5 × FAD mice, but not when treated with VBIT-4 (Fig. 7a, b, Additional file 1: Fig. S6b). Glut-4 is expressed in neurons, and its level was also reduced in 5 × FAD mice, which was prevented by VBIT-4 treatment (Fig. 7a, b). Similar results were obtained for the levels of Glut-1, Glu-2, and Glut-4 in the hippocampus (Additional file 1: Fig. S6c–f).
Considering impaired metabolism in AD [10, 36, 68] and VDAC1 regulating metabolism [21, 24], we also evaluated the expression of several metabolism-related proteins in 5×FAD mice and the effects of VBIT-4 on their expression levels (Fig. 7d–g). The expression levels of the glycolytic enzyme, hexokinase-I (HK-I), in the cortex and hippocampus were increased in the 5 × FAD mice, but the increase was prevented by VBIT-4 treatment (Fig. 7d, e, Additional file 1: Fig. S7). HK-I punctate staining was co-localized with VDAC1, including in the neuropils surrounding the Aβ plaques (Fig. 7d(ii)), suggesting that HK-I was mitochondrial bound. The expression levels of the Krebs cycle enzyme citrate synthase (CS) and ATP synthase (ATPsyn5a) were highly decreased in both the cortex and the hippocampus of 5 × FAD mice, but the decreases were prevented by VBIT-4 treatment (Fig. 7d, f, g, Additional file 1: Figs. S8, S9). Interestingly, in contrast to HK-I, which was co-localized with the overexpressed VDAC1 in the Aβ-plaque area, the mitochondrial proteins CS and ATPsyn5a were co-localized with VDAC1 in healthy neurons, but not in the neuropils around Aβ plaques (Additional file 1: Figs. S8, S9). This points to dysfunctional mitochondria around the Aβ-plaque areas, consistent with reduced metabolic activity [18].
Some VDAC1 was previously found to be localized to the plasma membrane (pl-VDAC1) [21], so we tested whether this might be the case also in the Aβ-plaque area by analyzing the co-localization of VDAC1 with the plasma membrane protein, Na,K-ATPase (Fig. 7h, i). The results showed that in the 5 × FAD mouse cortex, most of the VDAC1 around the Aβ-plaques was not co-localized with Na,K-ATPase (Fig. 7h).
The results also showed that the Na,K‐ATPase expression was decreased by about three folds (Fig. 7i), consistent with the reported decrease in AD patients and in a transgenic mouse AD model [73, 74]. Moreover, the decrease of Na,K‐ATPase staining in the VDAC1-overexpressing neuronal terminals surrounding the Aβ plaques was largely prevented by VBIT-4 treatment (Fig. 7h, i, Additional file 1: Fig. S10, circled plaque area). Similar results were obtained for the hippocampus (Additional file 1: Fig. S10c, d). Considering the function of Na,K-ATPase in maintaining the Na+ and K+ gradient across the plasma membrane, which is essential for maintaining resting membrane potential and hence neuronal excitability [75], the decrease in its expression levels points to decreased neuronal excitability in the 5 × FAD mice.
VBIT-4 changes phenotypic properties of astrocytes and microglia
Astrocytes support neurons by shuttling metabolites, secreting trophic factors, and regulating ion balance and pH [45]. Reactive gliosis has been shown in numerous models of AD and in AD patients [76]. In 5 × FAD mice, gliosis begins to occur around 2 months, and develops in parallel with plaque formation [45]. Indeed, IF staining showed that GFAP and glutamine synthase (GS), both expressed mainly in astrocytes [77], were increased, with GFAP increased by nine and three folds in the cortex and hippocampus of 5 × FAD mice, respectively, compared to the levels in WT (Fig. 8a, b; Additional file 1: S11a–c).
Fig. 8 VBIT-4 treatment of 5 × FAD mice improves astrocyte and microglia morphology and activates microglia. a Confocal images of cortical sections from WT, untreated-, and VBIT-4-treated-5 × FAD mice co-immunostained for VDAC1 and GFAP. b Quantification of GFAP intensity in cortical and hippocampal sections. c–e Spinning disk microscopy 3D imaging of 50 μm cortical sections from 5 × FAD, and VBIT-4-treated-5 × FAD mice co-immunostained for GFAP and VDAC1 (c), analyzed using Imaris software for astrocyte 3D structures (c), number of branching points as a function of the distance from the soma (d) and number of processes for each branch order (e). f Cortical sections from WT, VBIT-4-treated, and untreated 5 × FAD mice were co-immunostained for VDAC1 and IBA-1. Higher magnifications of selected areas are shown. g Quantitative analysis of IBA-1 expression levels in and outside the Aβ plaques. h–l Spinning disk microscopy 3D imaging of 50 μm cortical sections from VBIT-4-treated- and untreated-5 × FAD mice stained for IBA-1 shown in 3D, as analyzed using Imaris software (h). Representative microglia structures in the Aβ plaques are shown at the bottom, and number of processes for each branch order (i) and the number of branching points as a function of the distance from soma (j). k, l Confocal images of cortical sections from WT, VBIT-4-treated, and untreated-5 × FAD mice co-immunostained for TSPO and VDAC1 (k) and quantification of staining intensity in the Aβ plaques (l). Results show means ± SEM (n = 3), ***P < 0.001, ****P < 0.0001. P-value in blue color represents the significance of VBIT-4-treated relative to untreated 5 × FAD mice. NS, not significant Full size image
Since 3 × FAD mice show astroglia atrophy, as manifested by decreased surface areas and volumes of GFAP-positive cells relative to WT [78], we reconstructed astrocytes by confocal imaging of 50-μm-thick sections, and analyzed the GFAP-stained images within the Aβ plaques using Imaris. The results showed that astrocyte morphology in 5 × FAD mice was highly modified in comparison to the VBIT-4-treated mice (Fig. 8c–e). Astrocytes in the VBIT-4-treated mice had more processes with greater surface area, more branches, and more branching points along the processes (Fig. 8d, e; Additional file 1: Fig. S11d, e). The result suggested that astroglia distraction leads to early synaptic disorders, resulting in cognitive deficits in AD [78].
In AD, the microglia play important roles in Aβ clearance and neuroinflammatory response via secretion of pro-inflammatory cytokines [79]. In 5 × FAD mice, IF immunostaining showed that IBA-1, involved in phagocytosis by activated microglia, was threefold and sevenfold higher levels of IBA-1 outside and inside the Aβ plaques relative to WT (Fig. 8f-h, Additional file 1: Fig. S11f). Upon VBIT-4 treatment, IBA-1 levels were further increased about 11-fold in the Aβ plaques (Fig. 8g). No co-localization of IBA-1 and VDAC1 was observed, indicating that the cells overexpressing VDAC1 were not microglia.
Activated microglia undergo morphological changes and migrate to the site of injury [80]. 3D images of microglia within the Aβ plaques were reconstructed from IBA-1 images with Imaris. In the 5 × FAD mice, microglia had short and thick processes with an amoeboid-shape, whereas in the VBIT-4-treated mice the microglia were larger and had more and longer processes (Fig. 8h–j, Additional file 1: Fig. S11g, h). This finding suggests that VBIT-4 prevents damage to the microglia.
To further determine the effect of VBIT-4 on microglial activation, we analyzed the expression levels of the mitochondrial translocator protein (TSPO), as its upregulation is often accompanied with microglial activation and secretion of cytokines, and it is considered to be a marker of neuroinflammation [81] and AD severity [82]. The TSPO expression was found to be redistributed to be mainly in microglia around the Aβ plaques (Fig. 8k) and was increased in the 5 × FAD cortex and hippocampus (Fig. 8l, Additional file 1: Fig. S11i). Indeed, in 5 × FAD mice, TSPO was present mainly in Aβ plaques as visualized by VDAC1 staining, and its level was sevenfold higher in VBIT-4-treated mice (Fig. 8l).
Astrocytes play an important role in brain energy metabolism, mediating glucose uptake from blood vessels to neurons (Additional file 1: Fig. S6(i)) and microglial phagocytosis, which require a large amount of energy. The decreased expression of several metabolism-related enzymes (CS and cytochrome c oxidase [COX-IV]) in both astrocytes and microglia in 5 × FAD mice, was restored by VBIT-4 (Additional file 1: Figs. S12, S13), indicating that the astrocytic metabolic functions were restored in VBIT-4-treated mice.
VBIT-4 treatment prevents exaggerated neuroinflammation
Next, we tested the effects of VBIT-4 on neuroinflammation associated with AD [83]. The transcription factor nuclear factor kappa B (NF-κB) functions in inflammation, and is implicated as a risk factor in AD [84]. Immunostaining showed that the phosphorylated, activated NF-κB-p65 (p-NF-κB-p65) was significantly increased in 5 × FAD mice compared to WT (Fig. 9a, b). In the cortex, p-NF-κB-p65 was mainly abundant in the cytoplasm and not as expected in the nuclei of neurons, particularly around neuronal nuclei with few stained nuclei (Fig. 9a, yellow and blue arrows) [85]. NF-κB-p65 mRNA level was dramatically increased (~ 150-fold) in the 5 × FAD brain, and VBIT-4 prevented this increase (Fig. 9c).
Fig. 9 VBIT-4 reduces inflammation signaling and induces anti-inflammatory neuroprotective astrocytes and microglia in 5 × FAD mice. a, b, d Confocal images of cortical sections from WT, untreated-, and VBIT-4-treated-5 × FAD mice co-immunostained for p-NF-kB-p65, TNF-α, NRLP3, caspase-1 or IL1-β (a), and their quantifications (b, d, f). c, e, g, i q-RT-PCR analysis of NF-kB-p65, IL-1β, caspase-1 and IL-4 mRNA levels. h Confocal images of cortical sections from WT, untreated- and VBIT-4-treated 5 × FAD mice co-immunostained for GFAP and IL-4 or TGF-β, and their quantifications (j, k). P-value in the blue color represents the significance of VBIT-4-treated relative to untreated mice. Results show means ± SEM (n = 3), *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. P-value in blue color represents the significance of VBIT-4-treated relative to untreated 5 × FAD mice. NS, not significant Full size image
As in the cortex, in the hippocampus, p-NF-κB-p65 staining was stronger in both the molecular and the granular layers where it was concentrated around the nuclei (Additional file 1: Fig. S14(i), arrows). In both the cortex and hippocampus of the 5 × FAD mice, VBIT-4 treatment reduced p-NF-κB-p65 expression to a level comparable to that in WT mice (Fig. 9a, b; Additional file 1: Fig. S14a,b). The activated p-NF-κB-p65 was found in astrocytes (Additional file 1: Fig. S14c), but also in neurons as shown for AD patients [86], where it co-localized with TUBB3, except in some of their processes (Additional file 1: Fig. S14d(ii) (iii)).
The levels of cytokines interleukin-1β (IL-1β) and tumor necrosis factor-α (TNF-α) regulated by NF-κB, are known to be increased in AD brains [87]. Consistent with this, our results showed increased TNF-α expression in 5 × FAD mice, while VBIT-4 prevented this elevation (Fig. 9a, b, Additional file 1: Fig. S15). Similarly, IL-1β expression was significantly increased in both cortex and hippocampus of 5 × FAD mice, while VBIT-4 prevented or attenuated this increase (Fig. 9a, d; Additional file 1: Fig. S15). Consistently, result of q-RT-PCR showed that IL-1β mRNA expression in the 5 × FAD mouse cortex was about 70-fold higher than that in WT mice, and VBIT-4 treatment greatly reduced this level (Fig. 9e). IL-1β expression levels were also increased in the hippocampus (Additional file 1: Fig. S15b).
NLRP3 (NOD-, LRR-, and pyrin domain-containing protein 3) acts as a sensor molecule, and together with the adaptor protein ASC (apoptosis-associated speck-like protein containing CARD) and pro-caspase-1 forms the NLRP3 inflammasome. The NLRP3 inflammasome is critical for the innate immune system [88], and is associated with neuroinflammation in AD [89]. We found that NLRP3 was highly expressed in both the cortex and hippocampus of 5 × FAD mice, while VBIT-4 treatment prevented the increase (Fig. 9a, f; Additional file 1: Fig. S16a,b,f).
In the activated inflammasome, caspase-1 is activated, converting proinflammatory cytokines such as pro-IL-1β into active forms [90]. The expression level of caspase-1 was highly increased in the cortex and hippocampus of 5 × FAD mice, but not in VBIT-4-treated mice (Fig. 9a, f;. Additional file 1: Figs. S16 and S17). This was also confirmed at the mRNA level (Fig. 9g). Thus, VBIT-4 protects against neuroinflammation.
The increase in activated microglia and decrease in pro-inflammatory agents induced by VBIT-4 treatment led us to consider the transition of microglia and astrocytes from a pro-inflammatory/neurotoxic (M1) to an anti-inflammatory/neuroprotective (M2) phenotype [91]. As the neuroprotective astrocytes and microglia are promoted by IL-4, IL-13, IL-10, and TGF-β [91], we compared the expression levels of IL-4 and TGF-β in VBIT-4-treated and untreated 5 × FAD mice. The results show that in 5 × FAD mice IL-4 levels were as in the WT in the microglia of both cortex and hippocampus, while TGF-β was decreased over 2 folds in the cortex and increased 3 folds in the hippocampus of 5 × FAD mice. However, upon VBIT-4 treatment, IL-4 and TGF-β were increased about 2- and 5-folds in the cortex and hippocampus, respectively (Fig. 9h–k, Additional file 1: Figs. S18, S19).