Sildenafil for the Treatment of Alzheimer’s Disease: A Systematic Review

INTRODUCTION

Alzheimer’s disease (AD) is the leading cause of dementia worldwide, and AD patients and their families urgently require novel therapeutics to prevent and slow the progression of this devastating disorder. Hallmarks of AD include amyloid-β (Aβ) peptide secretion and deposition into neuritic plaques, tau protein hyperphosphorylation and neurofibrillary tangle formation, metal ion dyshomeostasis [1–9], oxidative stress and lipid, nucleic acid, and protein damage [10–13], abortive cell cycle reentry [14–26], neuroinflammation and microbial dysbiosis [27–33], insulin resistance [34, 35], cerebrovascular dysfunction [36–38], synaptic dysfunction [39, 40], neuronal loss, endoplasmic reticulum stress [41–44], and mitochondrial dysfunction [45–48].

Sildenafil (Viagra) is a drug used to treat erectile dysfunction and pulmonary arterial hypertension that inhibits phosphodiesterase 5 (PDE5) (Fig. 1). PDE5 degrades cyclic guanosine monophosphate (cGMP). Upstream of cGMP, normally, the amino acid L-arginine is converted by three varieties of the enzyme nitric oxide synthase (NOS) into nitric oxide (NO). NO is a small cell-permeable gas molecule that diffuses across the plasma membrane and activates soluble guanylyl cyclase (sGC). sGC converts guanosine triphosphate (GTP) into cGMP [49]. However, in AD, the activity of the NOS/NO/cGMP pathway is severely impaired. NOS activity is significantly decreased in AD patients’ superior frontal gyri and hippocampi compared to age-matched controls [50], even though aberrant neuronal NOS (nNOS) protein expression has been observed in a subpopulation of isocortical pyramidal neurons in AD patients’ brains [51, 52] and the intensity of astrocyte endothelial NOS (eNOS) and inducible NOS (iNOS) expression had increased in AD patients’ deep cortical layers [52, 53]. NO-induced soluble sGC (but not basal sGC or particulate guanylyl cyclase) activity was found to be decreased by 50% in AD patients’ superior temporal cortices compared to controls [54]. cGMP levels were found to be significantly lower in AD patients’ cerebrospinal fluid (CSF) compared to controls, and decreases in levels of cGMP correlated with CSF Aβ₄₂ levels [55], comorbid depression [56], and cognitive decline as measured by Mini-Mental State Examination (MMSE) [55, 57].

OPEN IN VIEWER

Fig. 1

Sildenafil mechanism of action.

Potentially contributing to this cGMP depletion, PDE5 protein levels have been found to be significantly upregulated in AD patients’ temporal cortices [55], and PDE5 mRNA levels are significantly elevated in AD patients’ entorhinal cortices compared to controls according to a meta-analysis of mRNA datasets (p = 0.001, FDR = 0.018; Fig. 2) [58]. Granted, low PDE5 mRNA expression in the human CNS has been reported relative to peripheral tissue [59–61], and no specific hybridization signal was observed for PDE5 mRNA in the brains of aged and AD patients using radioactive in situ hybridization histochemistry in one study, drawing skepticism as to whether PDE5 inhibition has relevance to AD [62–64]. Regardless though, the weight of the evidence suggests that PDE5 is expressed (albeit relatively less than in peripheral tissues) in the normal human brain [58–61] and that, moreover, it is upregulated in the AD entorhinal and temporal cortices [55, 58], supporting the notion that PDE5 may be an important therapeutic target in AD.

OPEN IN VIEWER

Fig. 2

PDE5A mRNA levels upregulated in the AD entorhinal cortex.

Low and high but not moderate cGMP levels activate PGC1α?

Consistent with the dose-dependent effect of sildenafil and cGMP on PGC1α, 10μM of 8-Br-cGMP treatment for 24 hours upregulated PGC1α mRNA expression and mitochondrial biogenesis in renal proximal tubular cells [70]. By contrast, in endothelial cells, 100μM 8-Br-cGMP treatment for 6–24 hours downregulated PGC1α [80, 83]. In addition, treatment for less than 12 hours with NO donors decreased PGC1α expression, whereas treatment for 24 hours or more increased PGC1α expression [83]. NO-triggered PGC1α downregulation and consequent reactive oxygen species (ROS) production were required for endothelial cell migration, an effect that was mediated by NO-activated PI3K/Akt signaling and Akt phosphor-inhibition of Foxo3a [80].

Surprisingly though, much higher dosages of cGMP than 100μM upregulate PGC1α like lower doses do: for example, PGC1α was upregulated in brown adipocytes treated with 1 mM 8-Br-cGMP for 4 days [73]. In human monocytic U937 cells, rat L6 myotubes, and rat PC12 neurosecretory cells, 3 mM 8-Br-cGMP treatment every day for 6 days upregulated PGC1α, as well as Nrf1 and Tfam, mitochondrial proteins Cox IV and Cytochrome C, and mitochondrial DNA content [66]. Furthermore, 6 days of 3 mM 8-Br-cGMP in these three cell types resulted in increased oxidative phosphorylation-coupled oxygen consumption [66].

Therefore, there appears to be a U-shaped dose-response curve between cGMP concentrations and PGC1α expression, with PGC1α being downregulated by 100μM cGMP for 6–24 hours [80, 83] but upregulated by 10μM cGMP or 1–3 mM cGMP for 24 hours or longer (Fig. 3) [66, 73]. This might be because 100μM cGMP corresponds to the 10–100μM dosage of sildenafil that is sufficient to activate PDE2 and lower cAMP [76], whereas much higher dosages of cGMP and sildenafil nevertheless stimulate mitochondrial biogenesis and PGC1α transcription [66, 73] because supraphysiological cGMP signaling is sufficient to overcome the cAMP depletion to induce robust PGC1α activity independently of cAMP, perhaps via cAMP response element binding protein (CREB) phosphorylation and SIRT1 activation (see next subsection). If so, this would not be unprecedented: in fibroblasts, PDE2 overexpression was sufficient to lower cAMP levels and induce a transition into a pro-fibrotic myofibroblast phenotype, an alternation that cGMP elevating agents reversed, suggesting that sufficiently high cGMP levels can overcome the deficits to cellular signaling induced by PDE2-mediated cAMP depletion [77]. Therefore, it appears that, for optimal PGC1α expression, mitochondrial biogenesis, and BACE1 downregulation, either sufficiently low or high sildenafil dosages must be used, or sildenafil must be co-administered with a PDE2 inhibitor.

OPEN IN VIEWER

Fig. 3

Multiphasic dose response of PGC1α expression to cGMP concentration.

OPEN IN VIEWER

Fig. 4

How sildenafil and cGMP may regulate PGC1α.

In addition to this complex but clinically relevant dose-specific effect, the mechanism by which NO/cGMP typically upregulates PGC1α remains unresolved [67]. Two potential mechanisms involving either CREB or SIRT1 will now be described.

How sildenafil and cGMP activate PGC1α

In the hippocampus during long-term potentiation and in other neural tissues, both the cGMP/PKG and the cAMP/PKA pathways contribute to CREB phosphorylation [85–90], and PKA-mediated CREB phosphorylation promotes PGC1α transcription [67], so cGMP/PKG/CREB signaling might promote PGC1α expression as well. In addition, cGMP may contribute to the post-translational regulation of PGC1α. Once transcribed and translated, the stability, subcellular localization, and co-activator activity of PGC1α proteins are regulated by multiple post-translational modifications. For example, PGC1α can be inhibited via phosphorylation by Akt [81, 82], S6 Kinase [91], or glycogen synthase kinase 3β (GSK3β) [74, 92], or via acetylation by general control of amino acid synthesis 5 (GCN5) [81, 93–95] or p300 [96]. Conversely, PGC1α can be activated via phosphorylation by adenosine monophosphate activated protein kinase (AMPK) [97–101] or p38 mitogen activated protein kinase (MAPK) [101, 102], via methylation by protein arginine methyltransferase 1 (PRMT1) [103], or via deacetylation by SIRT1 (Fig. 5) [69, 104–107]. Interestingly given the SIRT1-mediated activation of PGC1α, sildenafil has been shown to upregulate SIRT1 in heart, cardiomyocytes [108], serum, and subcutaneous adipose tissue [109]. cGMP analogues upregulated SIRT1 expression in white adipose tissue [110, 111] and in osteoblasts [112]. Mice with osteoblast-specific PKGII overexpression exhibited increased SIRT1 expression [112]. The PKG inhibitor KT-5823 blocked the relief of spinal allodynia induced by resveratrol, a SIRT1 activator [113, 114]. In hypoxic myocardial cells, 1μM sildenafil treatment decreased PGC1α protein acetylation [115, 116]. Therefore, it is possible that sildenafil may promote PGC1α deacetylation in some contexts via cGMP/PKG/SIRT1 signaling (Fig. 5).

OPEN IN VIEWER

Fig. 5

PGC1α post-translational regulation.

MATERIALS AND METHODS

To determine the current progress in studying sildenafil and AD, we searched PubMed for “sildenafil Alzheimer’s.” Both preclinical and clinical studies were reviewed. Results that were not about the effect of sildenafil on patients or preclinical models with AD (e.g., studies about the interaction between cGMP and Aβ in long-term potentiation) were discarded. To ensure clinical relevancy, studies about derivates of sildenafil were also discarded.

RESULTS

As per the methods section, two in vitro studies, ten rodent studies, one systematic review, and two pilot studies in patients were included. Overall, all the studies supported the use of sildenafil in AD (see Table 1 for a summary of results).

HT-22 hippocampal neuronal cells treated with Aβ _25 - 35 exhibited mitochondrial calcium overload, which was associated with ATP depletion, ROS generation, permeability transition pore opening, caspase-9 activation, and cell death. Sildenafil prevented these effects by promoting the opening of ATP-sensitive K ⁺ channels [140]. In cultured HT-22 hippocampal neuronal cells, exposure to advanced glycation end products [141] (a risk factor for AD [142]) induced mitochondrial ROS generation, depleted intracellular ATP, opened the mitochondrial permeability transition pore, released cytochrome C, activated caspase-3, and initiated apoptosis. Treatment with sildenafil upregulated heme oxygenase 1 (HO1), protected mitochondria from permeability transition pore opening and cytochrome C release, and decreased caspase-3 activation and apoptosis [141]. HO1 expression was required for sildenafil-induced protection of mitochondrial reductive capacity and permeability transition pore opening [141], suggesting that sildenafil protected mitochondria via HO1 upregulation.

In mice with cholinergic dysfunction mimicking AD due to scopolamine injection, sildenafil rescued maze performance, with 3 mg/kg appearing to be more efficacious than the 1.5 or 4.5 mg/kg dosages [143].

In hippocampal slices from transgenic amyloid precursor protein (APP)/presenilin 1 (PS1) AD mice, 50 nM sildenafil rescued the deficits in tetanus-induced long-term potentiation in the Schaffer collateral pathway caused by the APP/PS1 genotype [144]. In vivo sildenafil treatment produced similar results. In APP/PS1 mice, a one-time dosage of 3 mg/kg sildenafil rescued contextual fear memory. Daily intraperitoneal dosages for 2-3 weeks of 3 mg/kg sildenafil partially rescued spatial working memory deficits on the radial arm water maze test. Similar benefits were found for daily dosages of 3 mg/kg sildenafil 9–12 weeks after treatment cessation, indicating a long-term benefit that persists even after treatment cessation. Sildenafil also rescued long-term spatial reference memory on the Morris water maze and the probe trial. Sildenafil treatment rescued tetanus-induced CREB phosphorylation in the CA1 hippocampus to normal levels. 3 weeks of daily 3 mg/kg sildenafil was sufficient to reduce Aβ ₄₀ and Aβ ₄₂ levels in cerebral cortex samples [144].

In APP/PS1 mice administered sildenafil 6 mg/kg intraperitoneally daily for 3 months, significant improvements were documented in behavioral tests (nesting behavior, arm entries in the Y maze, Morris water maze escape latency and path length), as well as immunoreactivity of inflammatory microglial and astrocytic markers ionized calcium binding adaptor molecule 1 (Iba1) and glial fibrillary acidic protein (GFAP), neurogenesis as shown by NeuN-positive neurons and doublecortin (DCX)-positive cells in dentate gyrus, and amyloid plaque burden [145].

In APP/PS1 AD mice, 2 mg/kg sildenafil twice daily for 4 months rescued cognition as shown by spontaneous alternation and escape from electrical stimulation in the Y-maze test, decreased amyloid pathology as shown by decreased cortical and hippocampal AβPP, Aβ ₄₀, and Aβ ₄₂ levels, decreased PDE5 expression, and increased nNOS, eNOS, iNOS, NO, and cGMP levels [146].

Sub-chronic intraperitoneal sildenafil treatment in APP/PS1 mice was found to improve memory as shown by novel object recognition preference, downregulated proinflammatory cytokines interleukin-1β (IL-1β), intereukin-6 (IL-6), and tumor necrosis factor-α (TNF-α) in the hippocampus, decreased hippocampal soluble Aβ ₄₀ and Aβ ₄₂ expression, and increased CREB phosphorylation [147].

A recent systematic review found that, on the Morris water maze and the T-maze, sildenafil improved spatial memory retention but not acquisition in aged mice [64]. Interestingly, this review reported that PDE5 is not located in brain structures critical for AD based on the lack of specific hybridization of a PDE5 mRNA probe [62–64], but as noted, other groups have found increased PDE5 mRNA or protein expression in AD patients’ entorhinal [58] or temporal cortices [55], respectively, and PDE5 mRNA expression has been found by multiple groups in the normal human brain, albeit at lower levels than in peripheral tissues [58–61].

In J20 AD mice, 15 mg/kg sildenafil daily for 10 weeks in drinking water resulted in improved performance on the Morris water maze, decreased tau hyperphosphorylation, and increased Akt and GSK3β phosphorylation, but it did not alter prefrontal cortex Aβ ₄₂ levels [61].

In Tg2576 AD mice, 15 mg/kg/day intraperitoneal sildenafil significantly rescued learning and memory deficits as shown by the Morris water maze and fear conditioning tasks, reduced hippocampal tau hyperphosphorylation, GSK3β activity, and the CDK5 p25/p35 ratio, upregulated hippocampal p-CREB and c-Fos following fear conditioning training, and increased hippocampal expression of brain-derived neurotrophic factor (BDNF) and activity-regulated cytoskeletal-associated protein (Arc) (an immediate early response gene involved in memory encoding) [148]. However, sildenafil did not affect total Aβ ₄₂ levels in the frontal cortex [148].

In aged mice, 3 mg/kg intraperitoneal sildenafil daily for 3 weeks decreased double-stranded DNA breaks and apoptotic cells as visualized by the terminal deoxyuridine triphosphate nick end labeling (TUNEL) assay in the CA1 hippocampus, downregulated proapoptotic proteins caspase-3 and B-cell lymphoma 2-associated X (Bax), upregulated antiapoptotic B-cell lymphoma protein-2 (Bcl2) and BDNF, downregulated AβPP expression, and suppressed the age-associated shift in the Aβ ₄₂/Aβ ₄₀ ratio [149].

In the senescence accelerated mouse-prone 8 (SAMP8) mouse model of accelerated aging and sporadic AD, 7.5 mg/kg sildenafil for 4 weeks improved cognitive performance as shown by the Morris water maze and the passive avoidance test [145, 151], tau hyperphosphorylation [145, 151], inflammation as shown by GFAP downregulation, and amyloid pathology as shown by downregulation of BACE1, cathepsin B, and hippocampal Aβ ₄₂ [150]. Sildenafil also activated Akt and inhibited GSK3β, calpain, cyclin-dependent kinase 5 (CDK5) [145, 150], and c-Jun N-terminal kinase (JNK) [151].

DISCUSSION

As predicted in the introduction, preclinical studies that tested these parameters have found that sildenafil rescued CREB phosphorylation, long-term potentiation, and learning and memory [61, 151], increased neurogenesis [145], and decreased neuroinflammation [145, 150]. In addition, these studies consistently found that sildenafil decreased tau hyperphosphorylation and related parameters [61, 151]. This might be in part because the MnSOD downregulation in AD hippocampal neurons contributes to tau hyperphosphorylation [45,153, 45,153], and low-dose sildenafil appears to upregulate MnSOD via PGC1α activation [69–72]. The relatively high 15 mg/kg dosages appear to have resulted in decreased tau hyperphosphorylation predominantly because high-dose sildenafil activated the cGMP/PKG/PI3K/Akt pathway, leading to increased inhibitory Ser9 phosphorylation of tau kinase GSK3β [61, 148].

Discrepancies have been reported, however, regarding the effect of sildenafil on Aβ levels: most of the studies showed that sildenafil decreases Aβ levels [144–147, 150], but the two studies using the highest dosages (15 mg/kg) reported no effect on frontal cortex Aβ levels [61, 148]. This can be understood through the lens of the U-shaped dose-response curve of sildenafil and cGMP on PGC1α signaling documented in the introduction: low-dose sildenafil and cGMP appear to activate PGC1α and suppress BACE1 expression [65, 76], whereas high-dose sildenafil and 100μM cGMP appear to inhibit PGC1α due to crosstalk with cAMP and PDE2 signaling, failing to suppress BACE1 expression. In other words, it is possible that the 15 mg/kg sildenafil studies may not have decreased Aβ levels significantly [61, 148] because that dosage activates PDE2, depletes cAMP [76], and inhibits cAMP-signaled PGC1α activation and BACE1 repression [65, 67]. Another intriguing observation is that only the 15 mg/kg studies reported Akt activation and/or GSK3β inhibition [61, 148], suggesting the possibility that the 15 mg/kg dosages activated Akt [61, 148] and may have consequently repressed PGC1α expression and its anti-amyloidogenic properties (Fig. 4) [80–82].

Interestingly, in vitro studies found that sildenafil protected mitochondria from Aβ or AGEs via ATP-sensitive K ⁺ channels or HO1 upregulation, respectively [140, 141], suggesting that sildenafil may promote mitochondrial function via multiple mechanisms, some of which may be independent of PGC1α.

In patients, an especially promising finding is that 50 mg sildenafil increased the cerebral metabolic rate of oxygen [127]. This effect might be accounted for totally by the increases in the cerebral blood flow [127], or it may have been partially mediated as well by PGC1α-regulated mitochondrial biogenesis. However, none of the studies reviewed measured PGC1α mRNA, protein, or acetylation levels, nor other markers of mitochondrial biogenesis, making it impossible to evaluate this possibility. Nor was the effect of sildenafil on SIRT1 activity or the SIRT1/PGC1α pathway explored in any of these studies. Nor did these studies examine the effect of sildenafil on markers of insulin resistance or antioxidant enzyme expression. Future preclinical studies in transgenic AD mice should address these points directly to assess the possible role of SIRT1 and PGC1α activity in sildenafil-induced Aβ suppression, mitochondrial biogenesis, and antioxidant enzyme expression in AD, as well as the putative effect of sildenafil on insulin resistance. Future studies should also assess the potential of combining sildenafil with a PDE2 inhibitor to increase sildenafil’s maximal effective dosage continuously throughout the treatment duration. This would bypass the dose-limiting effect of sildenafil on PDE2 activation and cAMP depletion [76], allowing for robust simultaneous cAMP and cGMP signaling and therefore maximal PGC1α activity, mitochondrial biogenesis, antioxidant enzyme expression, and BACE1 repression [48, 154–156]. The best candidate for this role would be propentofylline, a potent and broad-spectrum PDE inhibitor and methyl xanthine derivate like caffeine that is particularly effective at inhibiting cGMP-stimulated PDE2 activity and PDE4 [157]. Propentofylline would be superior to other PDE2 inhibitors primarily because 300 mg of it taken thrice daily one hour before meals has been tested and found to be safe and effective in mild to moderate AD patients in 5 phase III clinical trials [158–164]. Intriguingly, a recent review by Heckman and colleagues opined that, based on the preclinical evidence, inhibition of PDE2, PDE4, and PDE5 seemed to hold the most promise for the treatment of AD [165], and sildenafil and propentofylline administered together would potently inhibit these three therapeutic targets simultaneously [157].

Ultimately, a randomized control trial of sildenafil should be undertaken in AD patients to assess the clinical benefits of long-term sildenafil administration in this population compared to elderly controls. This RCT should use the 50 mg/day dosage [127]. As outcome measures, the RCT should test cognition on the MMSE and the Alzheimer’s Disease Assessment Scale-Cognitive Subscale, comorbid depression on the Geriatric Depression Scale [56], amyloid and tau pathology binding with 2-(1-6-[(2-[F- 18]fluoroethyl)(methyl)amino]-2-naphthylethylidene)malononitrile positron emission tomography (FDDNP-PET) [166, 167], cerebral blood flow with MRI [127], the cerebral metabolic rate of oxygen with MRI [127, 169], the cerebral metabolic rate of glucose with ¹⁸F-fluoro-deoxyglucose positron emission tomography (FDG-PET) [34, 170–174], inflammation with CSF IL-1β, IL-6, and TNF-α [124–126, 175–182], NOS/NO/sGC/cGMP/PGC1α pathway dysfunction with CSF cGMP [55, 56], and antioxidant enzyme activity with urine 8-oxo-2’-deoxyguanosine as an indirect biomarker [45, 183].

CONFLICT OF INTEREST

The authors have no conflict of interest to report.