Background
Alzheimer’s disease is the most common age-related dementia and a major and increasing burden for our society [
1]. This disorder is currently without treatment, and therapies to ameliorate alterations in amyloid beta (Aβ) or tau metabolism, initiated too late in the disease course, have proven disappointing [
2]. For this reason the search for early pathogenic mechanisms susceptible to therapeutic intervention is a medical necessity.
Emerging evidence indicates a role for multifactorial mechanisms involved in the early stages of the disease and preceding diagnosis. These mechanisms include genetic factors, possibly of cumulative impact [
3], risk factors, such as cardiovascular disease and diabetes [
4,
5], alterations in the microvasculature [
4‐
6], chronic dysregulated inflammation [
6,
7] and glutamate-mediated excitotoxicity [
3,
8‐
10].
In the present report we focused on glutamate excitotoxicity. Glutamate is the predominant excitatory neurotransmitter in the mammalian brain and participates in mechanisms associated with long-term potentiation and synaptic plasticity [
11]. Excessive production and release of glutamate, however, leads to neuronal injury and is a major pathogenic factor in many acute and chronic brain conditions, including Alzheimer’s disease [
3,
8‐
10,
12].
In our search for novel compounds with neuroprotective effects against glutamate neurotoxicity, we focused in a class of compounds, the angiotensin receptor blockers (ARBs) or sartans, that effectively blocks the physiological AT1 receptor (AT1R) and therefore the effects of angiotensin II, the main active factor of the renin–angiotensin system [
13] both in the periphery and the brain [
14]. Excessive peripheral AT1R activity associates with hypertension, heart and kidney failure, peripheral vascular and tissue inflammation, and metabolic abnormalities such as insulin resistance [
15‐
17]. Consequently, the use of sartans, because of their beneficial effects on inflammatory and metabolic alterations beyond their effect on blood pressure control, has become a cornerstone for the treatment of cardiovascular and chronic kidney disease [
18]. In turn, increased brain AT1R stimulation associates with brain ischemia, blood–brain barrier breakdown, Aβ production and toxicity, brain inflammation, traumatic brain injury and glutamate excitotoxicity, risk factors leading to neuronal injury, cognitive decline, and the incidence and progression of neurodegenerative diseases [
19‐
26]. For this reason it is not surprising that sartans have been found to be effective neuroprotective compounds. In vitro experiments demonstrated that sartans ameliorate neuronal injury produced by glutamate excitotoxicity and high levels of interleukin (IL)-1β, and microglia activation as a result of systemic administration of bacterial endotoxin (lipopolysaccharide (LPS)) [
25,
27‐
29]. In rodent models of Alzheimer’s disease, sartans (candesartan, losartan, valsartan and telmisartan) ameliorate all risk factors for human Alzheimer’s disease, including protecting cerebral blood flow and cognition during stroke, decreasing inflammation and Aβ neurotoxicity, and reducing traumatic brain injury [
24,
26,
27,
30‐
37]. Furthermore, clinical studies indicate that ARBs protect cognition after stroke and during aging [
15,
22,
38,
39], and cohort analyses reveal that treatment of hypertension with sartans significantly reduces the incidence and progression of Alzheimer’s disease [
40,
41].
To clarify the role of glutamate excitotoxicity, we used primary rat cerebellar granule cells (CGCs) in vitro. This is a well-characterized and reliable primary neuronal model to analyze mechanisms and excitotoxic neuronal damage and neuroprotection [
42,
43]. Although in humans CGCs are not primary targets for Alzheimer’s disease, rat CGCs are very sensitive to glutamate excitotoxicity, a major early injury factor in this illness, and are extensively used in Alzheimer’s disease research [
44‐
46]. We selected the ARB candesartan for our study because of its demonstrated neuroprotective effects on cultured primary cortical neurons, microglia and cerebrovascular endothelial cells, and its amelioration of brain inflammation in vivo [
27] including reducing glutamate-induced apoptosis in cultured CGCs [
25].
Our study was initially designed to provide mechanistic insight into the potential targets and pathways that may underlie glutamate-induced cell injury and its possible reversal by the neuroprotective action of candesartan. To this aim, we performed genome-wide expression analysis and evaluated the data with several pathway analysis programs: ingenuity pathway analysis (IPA), gene set enrichment analysis (GSEA) [
47,
48] and Kyoto encyclopedia of genes and genomes (KEGG).
The strong correlation of our findings with many signal transduction mechanisms and pathways associated with Alzheimer’s disease prompted us to determine whether there was an association between the changes in gene expression in our study with those found in postmortem brain samples from patients who suffered from Alzheimer’s disease. To this end, we compared our results with alterations in gene expression published in two independent microarray studies of hippocampal samples obtained postmortem from brains of patients diagnosed with Alzheimer’s disease. Because of evidence of cerebrovascular endothelial dysfunction in Alzheimer’s disease, we wanted to establish which of the genes altered in Alzheimer’s disease patients were predominantly expressed in cerebrovascular endothelial cells or in neurons. To clarify this point, we compared gene expressions altered in published Alzheimer’s disease patients with published analysis of predominant gene expression in human cerebrovascular endothelial cells and neurons obtained by laser capture microdissection from postmortem dorsolateral prefrontal cortex samples and then we looked at the effect of candesartan on these gene signatures in our CGC study.
Discussion
The overall goal of the study was to determine whether glutamate-induced alterations in gene expression in our primary neuronal culture were normalized by candesartan, and whether these changes correlated with alterations in gene expression in postmortem hippocampus of Alzheimer’s disease patients. We hypothesized that, if present, significant correlations would provide major preclinical evidence of beneficial therapeutic effects of candesartan.
There were several major findings in our study. Based on our results, we propose that candesartan may be neuroprotective on neuronal glutamate-induced injury. There were multiple functionally annotated genes strongly associated with Alzheimer’s disease and impressively correlated with alterations in gene expression in autopsy samples from Alzheimer’s disease hippocampus. We found novel functions differentially associated with genes predominantly expressed in neurons and in cerebrovascular endothelial cells.
Candesartan profoundly influenced glutamate-induced neuronal injury, since candesartan prevented glutamate-induced alterations in gene expression in about 800 of the over 1100 transcripts upregulated or downregulated by glutamate (Additional file
2: Table S2). Candesartan effects were unrelated to the proposed stimulation of angiotensin II (AT2) receptors by AT1R blockade, since AT2 receptors are not expressed in CGCs [
25].
Using qPCR, we confirmed glutamate-induced upregulation, normalized by candesartan, of a number of these genes, including several factors with fundamental roles in APP metabolism and Alzheimer’s disease, such as ADAM metallopeptidase domain 17 [
65‐
67] and APOE [
68‐
72] (Fig.
2; Additional file
3: Figure S1, Additional file
4: Figure S2 and Additional file
5: Figure S3).
Inflammation plays a significant role in the pathogenesis of Alzheimer’s disease [
6,
7,
73,
74]. Glutamate excitotoxicity upregulated many pro-inflammatory genes associated with Alzheimer’s disease [
25,
75‐
89] and were normalized by candesartan (Fig.
2; Additional file
3: Figure S1, Additional file
4: Figure S4 and Additional file
5: Figure S3). Glutamate also upregulated the expression of some genes involved in anti-inflammatory processes, and candesartan prevented these changes (Fig.
2; Additional file
3: Figure S1, Additional file
4: Figure S2 and Additional file
5: Figure S3) [
31,
90‐
93]. We hypothesize that while glutamate increases inflammation, at the same time it sets in motion a powerful anti-inflammatory response that is not necessary when the inflammatory response is prevented by candesartan.
Under the conditions of our experiments, we found that, when added after glutamate injury, candesartan does not protect neurons from cell injury [
25]. We interpreted that candesartan administration, although it may not reverse glutamate-induced cell injury which has already occurred, will prevent further glutamate-induced injury. Since glutamate excitotoxicity is a long-term process during progression of Alzheimer’s disease [
3,
8‐
10,
12], we believe our results are translationally relevant. The IPA analysis of the list of functionally annotated genes with their expression altered by glutamate and normalized when compared with the glutamate + candesartan group (over 400 genes) supported the proposed key role of inflammation in the pathogenesis of Alzheimer’s disease, [
6,
7], agreed with the demonstrated major anti-inflammatory effect of candesartan [
22,
27], and revealed many additional and novel diseases and functions, such as cell death and lesion formation, diabetes and glucose metabolism and vascular disease main risk factors for Alzheimer’s disease [
4,
5] (Table
1; Additional file
6: Table S3).
Furthermore, IPA analysis of upstream regulators of these genes included APP, APOE and retinoic acid (Tretionin), which play major roles in Alzheimer’s disease [
65‐
72] (Table
2; Additional file
7: Table S4) and revealed five kinase inhibitors, PD98059, SB203580, U0126, SP600125 and LY294002 (Table
2), that are part of the mitogen-activated protein kinase kinase/c-Jun N-terminal kinase/extracellular regulated kinase/p38-mitogen activated kinase/TGFβ-1 (MEK/JNK/ERK1/2/p38/TGFβ) pathways, reduce inflammation and toxicity, and have been associated with Alzheimer’s disease [
85,
94‐
98]. We found that the influence of PD98059 and SB203580 over inflammatory genes was similar to that revealed by candesartan in our study [
94,
95]. In support of the present findings, we have earlier reported that AT1R blockade prevents glutamate-induced ERK1/2, JNK and c-Jun activation [
25,
29], demonstrating that the effect of candesartan is upstream of ERK1/2-p38MAPK.
GSEA supported the findings revealed by IPA. Inflammatory, chemokine signaling, focal adhesion, actin cytoskeleton, apoptosis and extracellular matrix receptor interaction pathways were most relevant, and the expression of the associated genes, upregulated by glutamate, was normalized by candesartan (Fig.
3; Additional file
8: Table S5). Conversely, candesartan prevented the glutamate-induced downregulation of genes associated with neuronal function (Additional file
8: Table S5).
Many genes (19 out of 53) in the KEGG Alzheimer’s disease reference pathway were altered in postmortem Alzheimer’s disease patients and by glutamate and normalized by candesartan. The pathways included mitochondrial dysfunction, APP processing, including β-secretase (BACE1), apoptosis, DNA damage, lipid peroxidase, Ca
2+ signaling pathway and Ca
2+ overload (Table
3).
Most remarkably, GSEA showed a striking association between changes observed in our neuronal culture and those observed in published datasets of hippocampal samples obtained from Alzheimer’s disease patients. Genes up- or downregulated in Alzheimer’s disease hippocampus [
59‐
61] strongly correlated with genes up- or downregulated in neurons exposed to glutamate and prevented by candesartan (Fig.
4 and Table
3; Additional file
8: Table S5 and Additional file
9: Table S9).
Our results indicate that although the primary neurons studied here, CGCs, are not the primary targets for Alzheimer’s disease [
44‐
46], upon glutamate injury they exhibited multiple mechanisms closely associated with those revealed in human hippocampal autopsy samples. Some of the glutamate-induced injury mechanisms observed in CGCs have been replicated in primary cortical neuronal cultures [
25]. While analysis of postmortem samples has limitations because of the premortem agonal process and postmortem changes in glutamate metabolism, there was a striking correlation in alterations in gene expression between the two independent published datasets evaluated in our study. Furthermore, the normal controls used for the Alzheimer’s disease postmortem samples were also postmortem samples normalized for age and gender. Moreover, there were impressive correlations between our neuronal culture findings and those revealed in a mouse model of Alzheimer’s disease (Fig.
5), supporting the validity of our comparative analysis.
The predominant cellular expression of the genes altered in Alzheimer’s disease hippocampus and in our neuronal culture revealed two different pathological processes (Fig.
6). Multiple genes, upregulated in Alzheimer’s disease hippocampus and by glutamate in our neuronal culture, and normalized by candesartan, were predominantly expressed in human cerebrovascular endothelial cells when compared to neurons. IPA analysis of these genes revealed cellular movement/migration, extracellular matrix proteins, apoptosis, angiogenesis and vasculogenesis as principal functions controlled by these genes, and their most significant upstream regulators were TGFβ1 and beta-estradiol. TGFβ1 has been strongly associated with microvascular alterations in Alzheimer’s disease [
99]. There is substantial evidence for a role of beta-estradiol, and in particular hippocampus-synthesized 17β-estradiol in synaptic plasticity and cognition [
100,
101] and for neuroprotective effects of nonfeminizing estrogens [
102]. The glutamate-induced upregulation of genes selectively overexpressed in cerebrovascular endothelial genes strongly supports the proposed role of alterations in the microvasculature in Alzheimer’s disease, not only as a risk factor but also playing a major role in its pathogenesis [
4,
103‐
108].
Conversely, pathway analysis of genes predominantly expressed in human neurons when compared to human cerebrovascular endothelial cells and downregulated in Alzheimer’s disease hippocampus and by glutamate in our neuronal cultures, normalized by candesartan (Fig.
6), revealed neurological diseases, neurodegeneration, neuronal apoptosis and disorders of basal ganglia as principal related diseases. For these genes, the most significant upstream regulator was the NFE2L2 or Nrf2 gene that has been associated with the early stages of Alzheimer’s disease [
109]. These results are concordant with the well-known loss of neural function in Alzheimer’s disease.
It is tempting to speculate that pathological processes in Alzheimer’s disease may be based on two sequential and/or concomitant processes: enhanced inflammation in microvascular endothelial cells and neuronal injury. Although candesartan is a drug that was designed to work on the hypertensive endothelial vascular system, our data indicates that candesartan may directly protect neurons from injury, a proposal supported by a previous observation [
27].
Our report adds novel findings to the substantial body of evidence strongly suggesting that blockade of AT1R is a new avenue for the treatment of Alzheimer’s disease [
22,
110]. Preclinical experiments indicate that excessive brain angiotensin II activity through overactivation of brain AT1R leads to cognitive loss associated with hippocampal long-term potentiation blockade, inhibition of the cholinergic system and stimulation of Aβ production and tau phosphorylation [
22,
26,
111]. Of note, AT1R gene expression is upregulated by glutamate, and this change is normalized by candesartan (Additional file
2: Table S2).
Conversely, in preclinical models, AT1R blockade ameliorates hypertension, traumatic brain injury, brain ischemia and diabetes, the main modifying risk factors for Alzheimer’s disease, effects that include reduction of cognitive loss [
22]. In addition, AT1R blockade ameliorates cognitive loss in most of the rodent models of Alzheimer’s disease by reducing brain inflammation, excessive oxidative stress and in some cases decreasing Aβ production, oligomerization, tau phosphorylation and reducing blood flow [
22,
26,
31,
110‐
114].
Supporting the role of enhanced AT1R activity in Alzheimer’s disease, there was a correlation between alterations in gene expression in the APPswe mouse model of Alzheimer’s disease treated with captopril, an ACEI reducing angiotensin II formation, and those found in our study [
64] (Fig.
5). ARBs reduce inflammation in human circulating monocytes exposed to LPS [
28,
115], and prevent glutamate-induced neuronal apoptosis [
25]. Clinical studies demonstrate that AT1R blockers reduce major risk factors for Alzheimer’s disease, [
22,
110,
116]; observational and cohort studies reported that AT1R blockade delayed development of Alzheimer’s disease and protect cognition [
22,
117]. There are increasing calls to conduct randomized controlled trials to effectively test the hypothesis that AT1R blockade may be a novel therapeutic approach for the treatment of Alzheimer’s disease [
22,
117‐
119], and in particular including patients at the very early stages of the disease [
120].
The mechanism of candesartan neuroprotection from glutamate excitotoxicity has been associated with blockade of the glutamate NMDA receptor [
25]. In addition, candesartan neuroprotection may involve an increase in glutamate uptake into the cell [
121]. Furthermore, AT1R blockade may not be the only mechanism responsible for the neuroprotective effect of candesartan. Some ARBs, in particular telmisartan and candesartan, are powerful activators of a major neuroprotective mechanism, the peroxisome proliferator-activated receptor gamma (PPARγ) [
22,
25,
28,
33], and PPARγ activation plays a significant role in neuroprotection from glutamate excitotoxicity in cultured CGCs [
25].
Our gene analysis revealed major associations of the gene alterations reported here with Parkinson’s disease, neurological diseases and neurodegeneration. These observations support the hypothesis that ARB neuroprotection may not only be effective in Alzheimer’s disease, but also in other neurodegenerative diseases [
22,
23,
110].
Authors’ contributions
AGE participated in the design of the study, performed all gene and statistical analysis, analyzed data and wrote the draft and final manuscript. RH performed all cell culture and qPCR assays and helped to draft the final manuscript. JMS participated in the design of the study, supervised cell culture and qPCR assays, and wrote the draft and final manuscript. All authors read and approved the final manuscript.