NF-κB-regulated, proinflammatory miRNAs in Alzheimer's disease

Abundant neurochemical, neuropathological, and genetic evidence suggests that a critical number of proinflammatory and innate immune system-associated factors are involved in the underlying pathological pathways that drive the sporadic Alzheimer's disease (AD) process. Most recently, a series of epigenetic factors - including a select family of inducible, proinflammatory, NF-κB-regulated small noncoding RNAs called miRNAs - have been shown to be significantly elevated in abundance in AD brain. These upregulated miRNAs appear to be instrumental in reshaping the human brain transcriptome. This reorganization of mRNA speciation and complexity in turn drives proinflammatory and pathogenic gene expression programs. The ensuing, progressively altered immune and inflammatory signaling patterns in AD brain support immunopathogenetic events and proinflammatory features of the AD phenotype. This report will briefly review what is known concerning NF-κB-inducible miRNAs that are significantly upregulated in AD-targeted anatomical regions of degenerating human brain cells and tissues. Quenching of NF-κB-sensitive inflammatory miRNA signaling using NF-κB-inhibitors such as the polyphenolic resveratrol analog trans-3,5,4'-trihydroxystilbene (CAY10512) may have some therapeutic value in reducing inflammatory neurodegeneration. Antagonism of NF-κB-inducing, and hence proinflammatory, epigenetic and environmental factors, such as the neurotrophic herpes simplex virus-1 and exposure to the potent neurotoxin aluminum, are briefly discussed. Early reports further indicate that miRNA neutralization employing anti-miRNA (antagomir) strategies may hold future promise in the clinical management of this insidious neurological disorder and expanding healthcare concern.


Introduction
Although intensively studied for well over 100 years, the biological factors that initiate and drive the Alzheimer's disease (AD) process remain incompletely understood [1][2][3]. Anti-AD therapies directed solely against amyloid beta (Aβ) peptides have generally proved extremely disappointing, although therapeutic strategies targeted against multiple AD biomarkers -such as amyloid and tau abundance and processing dysfunction and neuroinfl ammation -have more recently shown greater promise [1][2][3][4].
As one recent example, the experimental drug posiphen, a chirally pure positive enantiomer of phenserine and βamyloid precursor protein (βAPP) synthesis inhibitor, has shown a signifi cantly improved effi cacy against multi ple AD-relevant targets, at least in proof-of-princi pal phase I testing [4]. Interestingly, this drug has been shown not only to attenuate Aβ42 peptide levels but also to lower the infl ammatory biomarkers complement factor C3 and monocyte chemotactic protein in the cere brospinal fl uid of patients suff ering from mild cognitive impairment [4]. Indeed, signifi cant increases in infl am ma tory biomarkers such as cytokines, chemokines, com ple ment factors, chemotactic proteins and C-reactive protein, mitochondrial-mediated upregulation of reactive oxygen species (ROS), and the proinfl ammatory actions of Aβ peptides have long been thought to be involved in a brain-specifi c infl ammatory process as AD initiates and progresses throughout the limbic system of the brain [4][5][6][7][8][9][10][11][12][13].
In the central nervous system (CNS), macrophages and glial cells -as the primary immune cells in the brain's privileged immune compartment -function primarily, by a variety of phagocytic and digestive mechanisms, to promote host defense by maintaining tissue homeostasis through the destruction of invading pathogens, through sequestering and eliminating deleterious debris via the cytoplasmic multi-protein infl ammasome complex, and by promoting tissue repair . On the contrary, sustained, uncontrolled activation of brain macrophages and glial cells can lead to excess production of various pathogenic factors that contribute to neuronal injury, including the signifi cant and dramatic upregulation of proinfl ammatory chemokines, cytokines and ROS. Th ese in turn are capable of activating infl ammatory transcription factors such as NF-кB and proinfl ammatory gene expression programs that drive cellular fate towards CNS dys-homeostasis, compromised neuronal function and, ultimately, apoptosis and brain cell death [2,3,[38][39][40][41][42][43][44][45][46][47][48].
A strong association between infl ammation and AD has been suggested for almost 50 years, and to date at least 2,750 peer-reviewed papers have appeared on the contribution of infl ammation to the AD process [11][12][13][14]. Some of these infl ammatory processes may be necessary in an attempt to regain brain cell homeostasis in early AD, but the integration of these processes into AD proliferation and the progression to late-stage AD is not well understood [15][16][17][18]. Over the last year there have been at least half-a-dozen excellent reviews on this area of research on the AD-infl ammation connection so this topic will not be covered in depth here [5,[15][16][17][18][19][20].
Briefl y, AD is characterized neuropathologically by at least fi ve heterogeneous features, all of which support the progressive generation of abnormal tau and amyloid, neural and synaptic defi cits and proinfl ammatory signaling to various degrees. Th ese features include: the appearance of hyperphosphorylated tau-protein containing intracellular neurofi brillary tangles; amyloido genesis -the progressive, age-related generation, aggregation and accumulation of Aβ peptides into dense, insoluble, proinfl ammatory and pathogenic deposits of senile plaque; reduced synaptic densities and synaptic protein assemblies; signifi cant neuronal loss in the temporal lobe and hippocampal regions that, as AD progresses, radiates into the more distal parietal, frontal and occipital poles of the brain; and a unique, chronic and progressive smoldering infl ammation of the neocortex and limbic system of the brain, especially in the middle to late stages of AD [11][12][13][14][15][16][17][18][20][21][22][23][24][25].
During upregulation of infl ammatory processes in the CNS and retina there appears to be a signifi cant parallel upregulation of the dimeric DNA-binding protein NF-кB (as the p50/p65 complex) [40][41][42][43][44][45][46][47][48]. Indeed, originally described in 1986, NF-кB has emerged as a ubiquitous transcription factor that controls diverse biological functions including infl am matory and immune functions in both the central and peripheral nervous systems [40][41][42][43][44][45]. NF-кB may be singularly important in regulating genetic responses to nervous system stress through the innate immune res ponse because it belongs to the category of pre-existing primary transcription factors that are already present in cells in an inactive-sensory state and do not require new protein synthesis to be activated [40][41][42][43][44][45]. Th at the NF-кB p50 and p65 subunits belong to an expanding family of more than 25 NF-кB subunits indicates that the subunit composition of NF-кB is variable and may be tailored by the cell to accommodate various infl ammatory signaling needs [40,41,[44][45][46][47][48][49]. Interestingly, compared with inter leukin-1 receptor-associated kinase (IRAK)-1, the more chronic and persistent activation of the NF-кB p50/p65 complex via the IRAK-2 signaling pathway in AD has recently been described [49]. Importantly, NF-кB activation and binding in the promoters of NF-кB-sensitive genes, including miRNA precursors (see below), leads to the facilitated transcription of many hundreds of potentially pathogenic genes, and therefore has the capacity to completely overwhelm the cell's anti-oxidant and antiinfl ammatory defenses while at the same time altering the functional properties of nervous system cells [40][41][42][43][44][45][46][47][48][49].
Common to aged, degenerating brain and retina are signifi cant upregulation of miRNA-125b and miRNA-146a, and their increases positively correlate with AD progression [38,59]. As discussed further below, up regulation of these miRNAs has been shown to be involved with a defi cit in synaptic and neurotrophic signaling, synaptogenesis and the induction of amyloidogenesis and infl ammatory signaling due to their selective targeting of several brain mRNA 3'-UTRs, including a critical downregulation of 15-lipoxygenase (15-LOX), synapsin-2 (SYN-2), IRAK-1, CFH and tetraspanin-12 (TSPAN12) gene expression [38,55,56,. Interestingly, the miRNA-mediated downregulation of certain brain mRNAs, and hence the impairment in their expression, contributes downstream to AD-relevant defi cits. For example, the miRNA-146a-mediated downregulation of TSPAN12 impairs the disintegrin and metallo proteinase-10 activity, thus shunting βAPP processing activities into more amyloidogenic and proinfl ammatory Aβ42generating pathways (  Table 1 displays some of the integrated neurobiological eff ects of just two of the most consistently upregulated NF-кB-induced miRNAs in AD brain and in stressed HNG cells in primary culture: miRNA-125b and miRNA-146a. Several of their multiple AD-relevant mRNA targets, the function of those mRNAs and conse quences of their defi cits, and original key references are shown.
More recently, interrelated and independent studies further suggest the sensitivity of human miRNA-9, miRNA-34a and miRNA-155 to AD-relevant stress and neuropathology as NF-кB-mediated miRNAs (Figure 1) [38,40,45,59,74,83]. While the neurological activities of miRNA-9, miRNA-34a and miRNA-155 are currently under active investigation by multiple laboratories, miRNA-125b and miRNA-146a and several of their mRNA targets, and the implications, are further discussed in the following sections.
Although miRNA-146a is a much less basally abundant miRNA when compared with miRNA-125b, it has been found to be the most inducible and upregulated miRNA in AD brain compared with all other NF-кB-regulated species so far indentifi ed ( Figure 1 and Table 1). Th e reason why miRNA-146a is one of the most rapidly induced of all brain miRNAs may be due to the presence of three cannonical tandem NF-кB binding sites in the pre-miRNA-146a promoter located at chromosome 5q34 [38,70,78]. Disease-related upregulation of miRNA-146a has also been observed in human prion disease and in infl ammatory processes associated with epilepsy, but no increase in miRNA-146a has been associated with multiple sclerosis, Huntington's disease, schizophrenia, and in certain grades of glioblastoma where the actions of other upregulated miRNAs may predominate [84][85][86]. Table 1, upregulation in brain-abundant miRNA-125b is associated with downregulation of the cell cycle inhibitor CDKN2A and glial cell proliferation, a pathological feature of AD gliosis, glioma and glioblastoma [57,58,72]. Upregulated miRNA-126b also downregulates the synaptic vesicle-associated neuronalenriched phosphoprotein (which associates with the cyto plasmic surface of synaptic vesicles) and neurotransmitter release regulator SYN-2 [61][62][63][64][65], as well as the 15-LOX enzyme essential for the conversion of the essential omega-3 fatty acid docosahexaenoic acid into the potent docosahexaenoic acid derivative and neuroprotectant NPD1 [66][67][68][69]. Defi cits in 15-LOX correlate with NPD1 defi cits in AD brain [66][67][68]. Similarly, a miRNA-146a-regulated CFH is a key negative regulator of the innate immune system, and miRNA-146a upregulation associates with decreased CFH and a chronic infl ammatory neural degeneration [38,53,56,87].

As indicated in
Similarly, the mRNA encoding the four-time membrane spanning integral membrane protein TSPAN12 is also a target for miRNA-146a, and upregulated miRNA-146a contributes to the downregulation of TSPAN12 as is observed in AD brain and in cytokine and Aβ peptidestressed human brain cells [8,80,81]. Just as suffi cient TSPAN12 appears to be required for the neurotrophic cleavage of the βAPP, insuffi cient TSPAN12 is associated with the induction of amyloidogenesis [8,80,81].
Th e integrated miRNA-mRNA interactions of as few as two human brain miRNAs (miRNA-125b and miRNA-146a) may hence in part explain not only the observed downregulation of CDKN2A, 15-LOX, SYN-2, CFH, IRAK-1 and TSPAN12, but also progressive, pathogenic defi ciencies in innate and immune signaling, neurotrophic support, and synaptogenesis and amyloidogenesis in the AD brain.

Herpes simplex virus 1
While only about 5% of all AD cases are genetic and 95% of all AD cases are of sporadic (idiopathic, or unknown) origin, a signifi cant epigenetic contributor to sporadic AD may well be of extraneural or environmental origin [1][2][3]. Two independent factors that have long been thought to contribute to infl ammatory aspects of AD are neuro tropic viral infection, specifi cally by HSV-1, and the abun dant neurotoxin aluminum in the environment .
Treatment of AD with antiviral agents -such as the already US Food and Drug Administration-approved acyclovir (brand name Zovirax®; GlaxoSmithKline, London, UK), penciclovir, valacyclovir (brand name Valtrex®; GlaxoSmithKline) or foscarnet -has been suggested as a possible effi cacious or adjunct treatment for AD [94][95][96] (unpublished observations). Th e pharmacological strategy here is that HSV-1 infection in the brain induces the accumulation of key pathogenic proteins, such as Aβ42 peptides, abnormally phosphorylated tau, and proinfl ammatory miRNAs, and that these antiviral agents have been shown to greatly reduce the abundance of Aβ42 peptides, phosphorylated tau and proinfl ammatory miRNA-146a accumulation in human brain cells previously infected with HSV-1 [73,88,[94][95][96].
Aluminum has also been shown to aggregate Aβ42 peptides into a much more neurotoxic, immunogenic and proinfl ammatory fi brillar form, as observed within the end-stage senile plaques in advanced AD brain [98,100]. For example, when tested for the ability to induce ROS and NF-кB activation in vitro, comparison of aluminum, cadmium, copper, iron, mercury, gallium, magnesium, manganese, nickel, lead, tin or zinc (as sulfates) at 50 nmol concentrations in HNG cell cocultures (using the novel, mixed isomer, fl uorescent indicator 5-(and-6)-carboxy-2' ,7'-dichlorofl uorescein diace tate) found aluminum to have by far the strongest ROS-inducing, NF-кB-inducing and infl ammatory gene expression-inducing capacity of any trace metal tested [10,109,110].
While antiviral therapeutic strategies have been advocated for the clinical treatment of AD [94,95], a single clinical trial using the actinobacterial siderophore desferrioxamine (mesylate) as an anti-oxidant, ROS scavenger and aluminum chelator has proved to be one of the most effi cacious treatments yet for mild-to-severe AD [105][106][107]. Th is is also in line with the idea that drugs such as desferrioxamine (mesylate) and posiphen that target multiple pathogenic molecules or processes in AD brain may hold the best promise in the clinical management of this complex and multifactorial neurological disorder [1][2][3][4][5]105,108].

Anti-miRNA (antagomir) strategies
Using perfectly complimentary ribonucleotide anti-sense (anti-miRNA; antagomir) sequences to lower the ambient abundance of upregulated miRNA in the brain is a logical approach to neutralizing the pathogenic gene expression eff ects of some overly expressed miRNAs, and attenuating their eff ects on selective mRNA abundance. Th is neutralization has been demonstrated in primary human brain cell tissue co-culture for both miRNA-125b and miRNA-146a [38,75,78,79,83,84]. Th e structure of these small, single-stranded therapeutic anti-miRNAs can be chemically modifi ed to increase their stability within the cell in vitro, and as little as 5 nM locked nucleic acidstabilized anti-miRNA per million human brain cells in primary tissue culture has been shown to have a dramatic quenching eff ect on both the target miRNA and proinfl ammatory gene expression induction patterns when analyzed using DNA and miRNA arrays and LED-Northern analytical techniques [6,7,55,75,79].
While it is not at the present time clear whether these anti-miRNA strategies can be translated into human therapies for infl ammatory degeneration, these kinds of RNA silencing approaches have shown recent promise in the treatment of glioblastoma, the most lethal form of primary malignant tumor in the human CNS [58,[83][84][85].

Conclusion
Th e six main conclusions from the research work presented in this review are as follows: the miRNAmediated downregulated expression of several bioinformatics and experimentally confi rmed mRNAs are targeted by increases in AD brain-relevant miRNA; stressors known to induce NF-кB also transactivate specifi c NF-кB-sensitive brain cell miRNAs; single miRNAs, such as miRNA-125b and miRNA-146a, have the potential to regulate multiple mRNA abundances relevant to the AD process; epigenetic and environmental factors such as HSV-1 infection and bioavailable aluminum may be highly relevant to the AD process, as they are both exceedingly strong inducers of NF-кB and pro infl ammatory miRNAs; specifi c antiviral, trivalent metal chelation, NF-кB inhibitors, or anti-miRNA strategies may be able to quench pathogenic miRNA over abundance and restore homeostasis to the AD brain, as is seen in models of AD in vitro; and HNG cells in primary co-culture are a proven, reliable, and human brain disease-relevant in vitro cell model to study the mecha nism of transcription factor-mediated miRNA activation and speciation, and infl ammatory signaling under normal aging, and physiologically relevant stress conditions.
Whether these mechanisms are operative in Tg-AD murine models, whether antiviral, anti-aluminum, anti-NF-кB or anti-miRNA strategies operate mechanistically in the same way in Tg-AD or cell culture models, or whether Tg-AD results can be extrapolated into human clinical trials are currently not known, and are all very active areas of independent research investigation. Since multiple mRNA targets are known to associate with neurodegenerative disease, and participate in complex positive or negative NF-кB-mediated feedback and signal ing loops, these miRNA-mRNA linkage studies and their functional interpretations in disease may be more complex than initially anticipated, especially when multiple epigenetic or environmental factors are involved . Importantly, the signifi cant overabundance of NF-кB and miRNA in specifi c anatomical regions in AD neocortex and hippocampus strongly implicates an NF-кB-mediated, miRNA-regulated infl ammatory disease mechanism that appears to selectively downregulate diff erent pathology-associated brain gene transcripts during the sporadic AD process, including those AD-relevant miRNA-mRNA pairings and the pathogenic consequences depicted in Table 1 [38,78].
In summary, AD is a complex neurodegenerative disease caused by the dysregulation of numerous brain cell functions and multiple neurobiological networks [1][2][3]5,[34][35][36][37][109][110][111][112][113][114]. A wiser therapeutic strategy may therefore be to consider the use of drugs or drug combinations that have multiple pathogenic targets, with minimal off -target and negligible peripheral toxic eff ects [1][2][3][4][5][6][7]32,[109][110][111][112][113][114]. Th ese eff ects include the imple mentation of novel drug delivery systems [111][112][113][114]. As an important step to achieve this goal we currently need to better understand the role of brain chromatin-mediated transcription mechanisms in AD and how these compare with normally aging brain, to better understand the role of ancillary DNA-binding proteins and proinfl ammatory transcription factors such as NF-кB in these processes, and to better understand features of other related epigenetic mechanisms on specifi c miRNA-mRNA recog nition, activation, and signaling pathways. Yet another layer of miRNA-mediated genetic complexity in the brain appears to be the role of miRNA nucleases and the relatively rapid turnover of specifi c miRNAs, which ultimately modulates the ability of miRNAs to impact pathogenic signaling [38,78,87,115,116]. Paradoxically, certain infl ammatory responses may prove to be neuroprotective or benefi cial, so it will be important to quantify both the individual contribution, and integration, of each of these proinfl ammatory signaling pathways to the AD process. Eventually, their net impact on the neurogenetics of brain cell function in healthy aging and in infl ammatory neurodegenerative disease will be elucidated, yielding advanced therapeutic strategies and combi natorial approaches that have not yet been considered.

Competing Interests
The author declares that he has no competing interests.