Background
Current therapeutics and drugs for improving memory loss in Alzheimer’s disease (AD) patients only marginally ameliorate cognitive deficits, and provide patients with only restricted symptomatic relief [
1‐
3]. Many compounds under therapeutic development are years away from benefiting AD patients. Thus, there is a critical need for rapid development of safe, effective therapeutics against AD. Melatonin analogs may provide such a therapeutic.
Melatonin (MEL) is known to modulate many physiological functions [
4‐
6]. With advancing age and certain age-related diseases, the endogenous secretion of MEL drops markedly, and MEL supplementation can ameliorate sleep disorders in the aged [
7]. Declines in blood and CSF MEL levels in Alzheimer disease patients have been reported to parallel the progression of neuropathology [
8,
9]. Although MEL has been given routinely to AD patients to suppress sundowning, few publications have investigated the cognitive effects of MEL administration on AD patients. MEL stabilized cognitive function in AD patients over a 2–3 year period [
10] and improved cognitive performance in mild cognitive impairment (MCI) individuals [
11]. A 3-year course of MEL treatment to one of a pair of monozygotic AD twins resulted in milder cognitive impairment for the treated twin [
12]. Despite these epidemiologic and anecdotal reports, all of which involved low MEL doses (≤9 mg/day), no controlled clinical studies of MEL effects on cognition in AD patients have been published. MEL is amphiphilic, thus it is able to penetrate all cellular compartments and freely enter the brain, especially from the CSF [
13,
14]. The pharmacokinetics and oral bioavailability of exogenously-administered MEL and analogs have been well-established in preclinical and clinical studies [
15].
Considerable
in vitro evidence supports the premise that MEL exerts an anti-amyloid-β (Aβ) aggregation effect [
16‐
18]. MEL has been shown to protect against Aβ-induced neurotoxicity
in vitro and
in vivo [
17‐
22]. We and others have also demonstrated significantly reduced amyloid plaque burden in AD mice treated for several months [
23,
24]. Interestingly, the neuroprotection afforded by MEL in AD mice appears to be age-dependent [
25] in as much as treating mice from 4 to 8 months was not found to be significantly beneficial, whereas MEL from 8 to 12 months of age (or from 4 to 12 months of age) significantly preserved cognition while reducing amyloid plaque load in these animals. In another study [
26], a very low dose of MEL (0.08 mg/day) was administered to aged Tg2576 (APP
swe) mice beginning at 14–18 months of age. Neither soluble Aβ levels nor Aβ deposition was affected in cortex, leading the authors to conclude that MEL is unlikely to be a treatment for already established AD. However, the low dose of MEL utilized in their study, the very late onset of treatment, the lack of cognitive evaluation, and non-assessment of any other key markers, are clearly in contrast to our previous studies and the present study.
MEL has often been reported to have anti-inflammatory (and occasionally pro-inflammatory) properties in many species, including humans [
27‐
29]. It is noteworthy that MEL administration lessens Aβ-induced pro-inflammatory cytokine levels in rat and mouse brains [
23,
30]. Indeed, MEL may represent a new class of anti-inflammatory agent [
31], with accumulating evidence for a significant role in reducing neuroinflammation via diverse mechanisms (Hardeland et. 2015 ibid). Melatonin offers neuroprotection at the level of mitochondrial function [
32,
33]. Consistent with this idea, our published work points to reduced oxidative stress in a mouse model of AD (AβPP
swe/PSEN1dE9) after administration of MEL for ≥ 1 month [
23,
33] and a MEL-mediated decrease in COX activity in the striatum of our double AD mice. Finally, evidence demonstrates that MEL can decrease tau hyperphosphorylation in cell cultures [
34]. Another mechanism through which MEL may protect against cognitive impairment is through stabilization and enhancement of dendritic structure. Prior studies have shown that MEL is capable of preventing loss of dendritic length and number for pre-frontal cortical neurons of rats subjected to global ischemia [
35‐
37]. MEL has also been reported to promote dendritogenesis in the hippocampus [
38]. Thus, MEL appears to exert multiple complementary mechanisms of action in the brain and hence may be an excellent therapeutic against AD.
Despite these consistent and significant actions of MEL on the cognition and pathology of the AD mouse brain, the mechanisms of MEL action remain unclear. In two recent reports [
39,
40] the cognitive function of transgenic AD mice was assessed after treatment with the specific, nonselective MTNR ligand, Ramelteon® (Takeda Pharmaceuticals), for up to 6 months. Ramelteon® is a commercially available, clinically tested (for insomnia), highly specific agonist at both MTNRs, having no direct intracellular activity [
41]. Intriguingly, and despite evidence for reduced hippocampal protein oxidation [
40], Ramelteon® in these two studies was ineffective in lowering amyloid plaque load or preserving cognitive functions. The most parsimonious interpretation of these findings is that MEL acts via MTNR-independent mechanisms to cognitively protect the amyloid-afflicted brain.
In an effort to provide clarity on this matter, we have generated transgenic AβPPswe/ PSEN1dE9 (2xAD) mice that lack both known MTNRs (MTNR-) in order to determine whether MEL has neuroprotective capabilities that are independent of these receptors. Cognitive performance, Aβ hippocampal plaque load, blood Aβ1–40 and Aβ1–42 levels, and key markers of brain oxidative stress were assessed in 2xAD mice with or without MTNRs that received long-term MEL. Compared to NonAD mice, the 2xAD animals developed significant cognitive deficits that were lessened by MEL. In some cognitive domains this neuroprotection was seen even in the absence of MTNRs. Similarly, amyloid plaque load in the hippocampus and frontal cortex as well as circulating levels of Aβ1–42 were significantly lowered by MEL in an MTNR-independent manner. These results strongly suggest that MEL provides neuroprotective effects in the AD brain in a manner that is to some degree independent of the two known membrane receptors.
Discussion
Transgenic mice models for studying the cellular and molecular basis of AD have been used for many years, but rarely for investigations into the therapeutic potential of MEL to slow the progression of this disease [
24,
50,
51]. Our 2009 publication [
23], detailed the neuroprotective effects of long-term oral MEL administration (~0.5 mg/day) on cognitive performance, brain Aβ levels/deposition, and antioxidant enzyme expression in the AβPP
swe/PSEN1dE9 (2xAD) mouse model. Subsequent recent studies continue to demonstrate that MEL has significant and reproducible prophylactic properties in several mouse models of AD [
33,
52,
53].
The primary goal of the present study was to determine the role of the cognate G-protein coupled melatonin receptors in mediating the neuroprotective action of MEL in the 2xAD mouse model. As reported by the supplier (Jackson Laboratories) the 2xAD animals begin developing cognitive deficits after 12 months [
54]. We saw no significant cognitive deficits at 12 months of age when using three common behavioral tests (data not shown). However, at 15 months the picture was markedly different. Using the novel object recognition test (NORT), which assesses short-term memory in a non-spatial task, the cognitive protection afforded the AD mouse brain by MEL was equivalent in both the mice with MTRs as well as in the KO mice lacking MTNRs (Fig.
3). Whereas both vehicle-treated AD groups performed poorly (i.e. they were unable to recognize novelty), the MEL-treated 2xAD mice were able to perform at the level of the non-transgenic (NonAD) vehicle-treated control mice irrespective of whether they were with or without the melatonin receptors. These results are consistent with the view that MEL preserves non-spatial cognitive performance in 2×AD mice even in the absence of G-protein coupled membrane MTNRs.
Additional cognitive testing was conducted with the circular platform (Barnes) test for spatial reference learning and memory (Fig.
4). In this test, 15 month-old 2xAD mice without melatonin receptors that had received MEL had lower latencies to the escape hole on day 1, while the 2xAD mice with receptors learned to be even faster over the course of the 8-day testing paradigm when compared to vehicle-treated controls (Fig.
4a). The day 1 differences likely reflect enhanced long-term memory in MEL-treated mice, as the same animals were tested earlier at 12 months of age. Fig.
4b depicts comparative performance on the Barnes test between 12 months and 15 months of age. Significant improvements with age were seen in the 2×AD mice that had received MEL, with more pronounced improvements in the MTNR+ mice.
Mice were also tested with the Morris Water Maze, which includes elements of spatial learning as well as assessment of working memory. As would be expected, the NonAD/MTNR+ mice performed well in both domains (Fig.
5), while the vehicle-treated 2×AD groups performed poorly on both. Notably, cognitive performance of the 2xAD/MTNR+ mice that received MEL was similar to the NonAD/MTNR+ control mice for both testing components. 2×AD/MTNR- mice that received MEL did not show neuroprotection, as their performance levels were comparable to the vehicle-treated 2×AD/MTNR- mice. Thus, the results of these cognitive tests confirm that MEL is capable of preserving memory in 2×AD mice in both non-spatial and spatial domains, as shown previously by several research groups [
23,
25,
40]. Additionally, our new data reveal that MEL effects are dependent on the MTNRs in the case of hippocampal-dependent spatial learning tasks, such as the Barnes Maze and the Morris Water Maze, but independent of MTNRs for non-spatial learning.
Acute MEL administration has been reported to have anxiolytic effects in both animal models [
55,
56] and in human trials [
57,
58]. In addition, Ochoa-Sanchez and colleagues [
59] reported that the novel MT2- selective agonist, UCM765, showed anxiolytic properties in the Open Field test and Elevated Plus Maze. Recently, Di Paolo et al. [
60] reported that MEL had anxiolytic effects in both NonAD and Tg2576 (AβPP
swe) mice. In the present study, we could confirm that the anxiolytic effects of MEL depended on the presence of the MTNRs (Fig.
2). MEL trended to influence anxiety in the 2xAD mice, but only in the presence of melatonin receptors.
For the sake of completeness we also assessed sensorimotor function at 12 and 15 months of age in 2xAD and control mice receiving MEL. No evidence of sensorimotor deficits was seen on the Rota-rod or Platform Recognition tests at any of the ages tested (Fig.
1) and there were no statistical differences between any genotypes or treatment groups. The slight improvements in both tests at 15 months vs. 12 months are likely to be due to the effects of experience. Furthermore, despite their C3H background, these mice clearly had no visual deficits, as was confirmed by the behavioral performance on various tests requiring spatial learning (see Figs.
4 and
5). Evaluation of open field activity at 12 months and at 15 months of age revealed no significant differences in any groups (Fig.
1). The slightly increased activity levels seen in the 2xAD/MTNR- mice at 12 months is likely a consequence of genetic deletion of the MTNRs, which we have recently demonstrated to be associated with mild hyperactivity [
42].
As a neuropathological correlate of AD progression in these animals we assessed amyloid plaque load by immunohistochemistry in the hippocampi and frontal cortex of 15 month-old mice. A significant reduction of plaque area was seen in MEL-treated mice as compared to vehicle (Fig.
7), with MTNR+ mice that received MEL having 33.5 % less plaques than controls (P < 0.001), while MTNR- mice that received MEL having 18 % less plaques than controls (P = 0.0034), thus comprising the majority of the protective effect of the MEL treatment on plaque deposition. Identical plaque loads were seen in control 2xAD mice, indicating that the genetic deletion of the melatonin receptors does not predispose the animals to more severe neuropathology as they age. Indeed, in our recent comprehensive characterization of the NonAD mice [
42] we demonstrated that the MTNR- genotype slightly, but significantly, did just the opposite, i.e. it enhanced motor performance, cognitive function and long-term potentiation as measured in a hippocampal slice preparation. The plaque results from the current study (Fig.
7) confirm that the neuroprotective effects of MEL are largely independent of melatonin receptors, although maximal protection was achieved when melatonin receptors are expressed. In other words, both receptor signaling pathways as well as direct intracellular effects of MEL appear to contribute to its plaque-inhibiting mechanism of action with direct, receptor-independent actions predominating.
Although not the central focus of the current investigation, it is perhaps worthy to note that the 2×AD mouse model has been described by the commercial supplier as having a significant degree of mortality due to seizures in early life [
61]. While we saw no evidence of seizures for the mice expressing the melatonin receptors, we did notice that the 2×AD mice without melatonin receptors succumbed in significant numbers in the first 6–9 months of life (Fig.
6). Brains from these animals showed no signs of plaque formation at 4 months and furthermore, at the completion of the study cognition in these animals with melatonin treatment was similarly protected as in the 2xAD/MTNR+ mice (Figs.
4 and
5). Administration of MEL to this group only marginally slowed the loss of the 2xAD/MTNR- mice. It is not entirely clear why the genetic deletion of the melatonin receptors is associated with poor early survival in the 2×AD mice, although an alteration in the balance between inhibitory and excitatory circuits in mice lacking the melatonin receptors [
42] might be a possibility worth further testing.
The processing of the amyloid precursor protein in the AD brain generates the pathogenic peptides, Aβ
1–40 and Aβ
1–42, with the latter thought to be the major neurotoxic species [
62,
63]. Although the blood amyloid levels in this mouse model of AD reflect both central and peripheral APP processing and clearance, we felt that as an indirect measure of MEL’s potential to alter these activities, it would be informative to assess plasma levels of these peptides in our 2xAD mice during the course of plaque development. Thus, using commercial ELISAs we determined plasma levels of Aβ
1–40 and Aβ
1–42 at ages 9 (cognitively presymptomatic) and 14 months (Fig.
8). At 9 months of age – when plaques are beginning to accumulate – treatment of 2xAD mice with MEL led to lower plasma levels of Aβ
1–40 and Aβ
1–42 in the MTNR+ mice and a similar trend in the MTNR- mice. By 14 months of age (by which time cognitive symptoms are fully expressed), MEL-treated 2xAD mice of the MTNR+ genotype had a significant reduction in plasma Aβ
1–42 as well as the Aβ
1–42/Aβ
1–40 ratio (data not shown), concomitant with significantly lower amyloid load in the brain (Fig.
7). In view of evidence that Aβ
1–40 and Aβ
1–42 may be cleared from the brain by different mechanisms [
64,
65], our overall results are consistent with a general reduction of amyloidogenic APP processing (rather than Aβ clearance) in response to MEL administration. In agreement with this hypothesis [
25] measured hippocampal levels of Aβ
1–40 and Aβ
1–42 in the Tg2576 AD mouse and reported significant reductions following treatment with MEL. Additionally, this group demonstrated that MEL treatment decreases Aβ processing via reductions of hippocampal presenilin and β-secretase levels. More recently, Shukla et al. 2015 reported that MEL upregulates alpha-secretase protein levels and subsequent catalytic cleavage of beta-APP to nonamyloidogenic products in cultured neuronal and non-neuronal cells [
66]. Thus, our current results showing melatonin receptor-independent MEL-induced decreases in plasma Aβ
1–42 levels (Fig.
8), concomitant with protection of non-spatial cognitive performance (Fig.
3) and reduced amyloid plaque load following MEL (Fig.
7) are consistent with the hypothesis that MEL’s neuroprotective activities are at least in part due to its long-term effects on Aβ generation.
Brain mitochondrial function, which is very sensitive to oxidative stress and is impaired in AD [
67‐
69], is protected by MEL through numerous mechanisms. For example, we reported recently that chronic MEL treatment protected in a dose-dependent manner against Aβ-mediated mitochondrial dysfunction at multiple levels in 2xAD mice and that this protection could be blocked by use of specific melatonin receptor antagonists [
33]. As a further exploration of this phenomenon, we assessed the brain activity of the mitochondrial enzyme, cytochrome c oxidase (complex IV), in 2xAD mice possessing or lacking MTNRs. Striatal samples were assessed, as all of the hippocampal and cortical tissues from these animals were used for other purposes. However, in our previous study [
47] we found that mitochondrial respiratory activity in all three of these brain regions of 2xAD mice were similar.
Unexpectedly, we found increased complex IV activity caused by AβPP
swe/PS1 expression from 13 to 16 months of age, not decreased expression as we and others have observed when studying AβPP
swe Alzheimer’s mice of different genetic backgrounds [
33,
70]. We have confidence in our complex IV activity measurements in this report as it is well established that there is an aging-related decline in complex IV activity in rodent brain [
71]. This increased complex IV activity in 16-month 2xAD mouse brain may be mediated by mitochondrial proliferation or increased expression of oxidative phosphorylation complex subunits in response to oxidative stress [
72,
73]. If the oxidative stress is mild or of short duration, mitochondrial proliferation or increased expression of oxidative phosphorylation complex subunits may increase energy reserves to maintain cellular homeostasis, but when the oxidative stress is high or chronic such as in AD, mitochondria become damaged and these compensatory mechanisms become ineffective in restoring energy levels leading to cell and tissue dysfunction. The hypothesis that the increased complex IV activity in the 2xAD mice is caused by oxidative stress is supported by the observation that MEL treatment of the 2xAD/MTNR+ mice completely prevented the increased complex IV activity. MEL likely accomplishes this by both directly scavenging ROS and by decreasing the production of ROS from the electron transport chain [
74]. Melatonin can reduce electron leakage from complexes I and III of the ETC. This can importantly lead to reduced formation of nitric oxide [
75], peroxynitrite, and peroxynitrite-derived free radicals such as hydroxyl radicals and nitrogen dioxide, which prevent increased NADPH oxidase and iNOS activities, with an overall effect of decreasing neural inflammation [
29].
Consistent with our current findings of increased complex IV activity in the 2×AD mice, another group also found increased complex IV activity in AβPP-expressing mice [
76]. They found increased COX activity in the ventral striatum of AβPP23 mice partially backcrossed onto a C57BL/6 background. This report showed increased complex IV activity only in specific regions of the brain. In another report, complex IV activity increased in Tg2576 mice at 5 months of age compared to controls [
77]. Another group found similar results with these mice at 7 months of age [
78]. Consistent with both of these reports, Tg2576 mice show upregulation of mitochondrial electron transport genes [
79]. Increased complex IV activity has also been found in neurons from Alzheimer’s patients [
80]. This increased electron transport chain complex activity is consistent with the Inverse Warburg Hypothesis of Alzheimer’s disease, which states that aged, energetically stressed neurons attempt to upregulate oxidative phosphorylation to use the lactate produced by adjacent astrocytes as a respiratory substrate to maintain cellular ATP levels [
81]. However, several other groups have measured decreased complex IV activity in various brain regions of Tg2576 mice [
70,
82‐
84]. In addition, reduced complex IV activity has also been measured in double and triple transgenic mouse models of AD combining overexpression of presenilin-1 and/or tau with mutant APP overexpression [
85,
86].
A recent report has shown that the Aβ peptide directly inhibits complex I activity of the ETC. [
87]. The inhibition of complex IV activity was found to be an indirect result from damage of the mitochondrial phospholipid cardiolipin, required for complex IV activity. Cardiolipin was found to be oxidized as a result of the complex I inhibition, and not from a direct inhibition of complex IV by Aβ as has previously been suggested [
49]. Cardiolipin peroxidation is most frequently catalyzed by the cardiolipin peroxidase activity of a cytochrome c-cardiolipin complex [
88]. Adding back cardiolipin to aged mitochondria has been shown to restore complex IV activity [
43]. In addition, Aβ has been shown to decrease transcription of specific complex IV subunits in certain cell types that may also contribute to the decreased complex IV activity [
89]. In the 2xAD mice used in this report, altered electron transport chain function in response to increased amyloid-beta levels may have led to decreased activity of the cardiolipin peroxidase activity of cytochrome c, preserving or even increasing cardiolipin levels resulting in the increased complex IV activity measured.
Another possible reason as to why we were able to measure increased complex IV activity in the 2xAD mice is because we used brain extracts where increased mitochondrial proliferation can be detected, in contrast to when measurements are made using isolated mitochondria where increased mitochondrial proliferation is likely missed unless paying close attention to the mass of mitochondria isolated. It is also possible that complex IV activity was upregulated in the 2xAD brain without increases in mitochondrial proliferation. This could be accomplished through expression of alternative complex IV subunits [
90] or through post-translational modification of the complex [
91]. When damaged cardiolipin limits complex IV activity slowing the rate of electron transport, oxygen has more time to bind the electrons producing superoxide. Under these conditions increasing complex IV activity would be an effective strategy to increase the rate of electron transport to increase ATP levels while also decreasing mitochondrial superoxide production. In summary, our data show that MEL administration protects MTNR+ mice from AD-induced upregulation of complex IV activity. However, MEL administration did not completely restore the increased complex IV activity in the 2xAD/MTNR- mice suggesting both receptor-dependent and independent effects are important for protection. It is of interest to perform similar experiments using additional brain regions and a different genetic background of mice to determine if these restorative effects of MEL are observed when complex IV activity declines as a result of APP
swe expression.
Substantial data have accumulated that link oxidative stress to AD pathogenesis [
92‐
94]. It is thus significant to note that numerous studies have reported MEL effects on markers of oxidative stress [
95‐
98]. Acutely, MEL has been reported to elevate in many tissues the expression of antioxidant enzymes as part of a defense mechanism against free radicals [
14,
99]. Previously, we have shown significant effects of long-term MEL on SOD1, GPx-1, and CAT mRNA expression in the brain of 10-month-old 2xAD/MTNR+ mice [
23]. Also, in the present study, we evaluated antioxidant expression at 13 months of age. As seen in Fig.
10, long-term MEL treatment was associated with lower mRNA expression of Nrf2, SOD1, GPx-1 and CAT in the brain (frontal cortex) of 13 month-old mice in both 2xAD and age-matched NonAD mice. A lower expression of antioxidant mechanisms is the logical outcome for brains that have reduced levels of oxidative stress due to chronic MEL treatment. These results are entirely consistent with our previous study [
33] demonstrating MTNR-dependent effects of MEL in the 2xAD brain. Thus, in view of our current data demonstrating that the effects of MEL on some cognitive functions and on amyloid pathogenesis in 2xAD mice are independent of MTNRs (Figs.
3,
7, and
8), the clear implication of our current study is that the potent effect of MEL on antioxidant gene expression in the 2xAD brain is not the key mechanism for its remarkable neuroprotective capabilities.