In the present study, network analysis was used to explore from the systems biology perspective, the molecular connections among multifactorial complex diseases with the shared clinical symptoms of dementia, which could suggest related disease mechanisms. A number of diseases were considered, both common (e.g. Alzheimer’s disease) and rare disorders (e.g. amyotropic lateral sclerosis with parkinsonism and dementia) that have as a common and major symptom a progressive and permanent loss of cognitive and mental performance (Table
While previous systems biology studies on disease focus on the disease gene network or drug target network, separately, the method proposed in the current study presented an integrated methodology that can take advantage of both these data, providing further insight into the interactome related to dementia.
Among the most connected proteins (with more than 100 interactions in the network; Table
3) the first 2 proteins in the list were PRKAA1 and PRKAA2, subunits of AMP-regulated kinase (AMPK). AMPK is a central regulator of energy homeostasis controlling neuronal maintenance in response to metabolic stress. Latest research support an involvement of AMPK in Alzheimer
[30, 31] and, in our previous study on Alzheimer’s disease, on a separate set of data and with a very different systems biology methodological approach, AMPK-related genes were also found to be strongly associated to the disease
. Moreover, abnormal neuronal accumulation of activated AMPK (pAMPK) has been described in different tauopathies including PSP, AD, Pick’s disease, and CBD
. Thus, the present findings support once more the proposed hypothesis of an alteration of metabolic functions and energy regulation in dementia.
Considering the complete list of mediator proteins, Gene ontology (GO) enrichment analysis confirmed a significant involvement of metabolic signaling regulating energy homeostasis, lipid and glucose metabolism (Figure
3). Metabolic disturbances have been strongly associated to or considered a predisposing factor in AD and a metabolic/signal transduction hypothesis for AD and other tauopathies has been suggested by Iqbal et al.
. Amongst the metabolic-related terms, a role for autophagy and regulation of autophagy was highlighted (Figures
4). Although autophagy has been known for decades, its relevance in neurons and glial physiology has been demonstrated only recently
. Autophagy is involved in the intracellular turnover of proteins and cell organelles
[35, 36] and AMPK is one of its main regulator
. In neurons, it is involved in cellular homeostasis and cellular protein clearance pathway and for the remodeling of terminals in support of neuronal plasticity
. In glial cells, autophagy is implicated in the elimination not only of glial proteins, but also of those secreted by neurons, which otherwise would accumulate in the extra-neuronal space
, and it is activated in astrocytes following injury
. Thus, it is not surprising that neurodegenerative dementia diseases, which have been linked to the abnormal accumulation of proteins and to alteration of synaptic plasticity, have been associated to the autophagic system
. Moreover, a potential role of autophagy in dementia is also suggested by the expression profile extracted from Mantra (http://mantra.tigem.it/), a transcriptional response database of FDA approved drugs, of 2 drugs clinically in use for the treatment of Alzheimer’s disease: galantamine and memantine. Several genes are modulated including AMBRA1, GABARAPL1, CLN3, SQSTM1, and AMPK subunits.
Detailed examination of autophagy-related genes in the mediator list, showed a preferential association to tauopathies, as demonstrated also by the GO enrichment study in the subset of mediators linked to dementia disease characterized by Tau protein inclusions (Figure
4). Autophagy process consists of several sequential steps including protein kinase network regulating the system, vesicle elongation, autophagosome assembly, and vesicle nucleation (Figure
 and specific autophagy dysfunctions could explain the diverse pathological course of the disorders. Analyzing in more detail these autophagy mediators and the molecular link to the specific disease genes, AD and FTD-related mediator proteins appears to be mainly associated to the initiation complex of the macroautophagy cascade, involving mainly beclin 1 interactome: B-cell CLL/lymphoma 2 (BCL2), BCL2-like 1 (BCL2L1), and Atg14 (Figure
6). Beclin 1 interactome contains stimulating and suppressive components regulating the initiation of the autophagosome formation and, recently, Beclin 1 has been found to be down-regulated in AD brain. Moreover, suppression of Beclin 1 in cultured neurons and transgenic mice induces the deposition of amyloid-β peptides, whereas its overexpression reduces its accumulation
. Beclin 1 protein also assembles two core complexes, Atg14L or UVRAG complexes, and with Atg14L protein induces the phagophore formation and, thus, stimulates autophagocytosis, whereas the UVRAG/Beclin 1 complex controls other Beclin 1-dependent functions, e.g. phagocytosis. The subcellular compartmentation of Beclin 1 is regulated by different molecules including BCL-2
 and mTOR
 which are represented in the mediator list. Several proteins can control the activation of beclin complex including two protein kinases included in our mediator list: CKD2 and CDK5
 (Additional file
1: Table S1). Atg5 protein, a mediator related to both AD and FTD disease genes (Figures
6), is also essential for the autophagy process and, in a conjugated form with Atg12 and Atg8 (LC3), is involved in the early stages of autophagosome formation (Figure
7C). Taken together these results suggest impairment in the early stages of autophagy complex essential for autophagosome formation
[46, 47], including protein kinase network regulating autophagosome assembling. The hypothesis that autophagy regulation and, in particular, its induction could contribute to AD pathology is also supported by recent evidence suggesting that the synthesis of components of the lysosome is up-regulated at the transcriptional and translational levels in the AD brain and AD mouse models
[48–54]. Moreover, Lipinski et al.
 recently reported an up-regulation of the transcription of genes stimulating autophagy and a down-regulation of the negative regulators of autophagy in the brains of AD affected subjects. In the dementia network, this interactome is connected to presenilin (Figure
5) whose mutation underlies the majority of familial Alzheimer’s disease cases
[56–58] and whose role in autophagy has been shown to be central
, presenilin 1 being essential for lysosome acidification, and proteolysis during autophagy
Frontotemporal dementia-related mediator proteins seem to be involved not only in the vesicle elongation and autophagosome assembling process, but also, and exclusively, to vesicle nucleation procedure (Figure
7C). This process includes WIPI proteins (WD-repeat protein interacting with phosphoinosides), WIPI-1 and WIPI-2, evolved from the yeast ancestral autophagy protein Atg18 (Proikas-Cezanne T, 2004; Polson HE, 2010) as membrane components of autophagosomes (Mauthe 2011,
). Both WIPI-1 and WIPI-2 specifically bind PtdIns(3)P and localize at autophagosomal membranes (phagophore)
TARDPB (TDP-43) appears to be a central protein in our autophagy-related sub-network (Figure
5). TDP-43 is a DNA/RNA-binding protein with multicellular functions. Several mutations of its gene have been reported in cases of frontotemporal lobar degeneration (FTLD)
. It is processed and degraded by both autophagy and the ubiquitin-proteasome systems
. Activation of autophagy by rapamycin plays an active role in the clearance of TDP-43 deficits in mouse model with proteinopathies of the TAR DNA-binding protein 43
. Depletion of TDP-43 induces a down-regulation of the major autophagy component Atg7, causes impairment of autophagy and facilitates the accumulation of polyubiquitinated proteins which could be rescued by overexpression of the protein, with a feedback regulatory loop between TDP-43 and autophagy
. In our network, TDP-43 is linked to AMPK subunit PRKAA2 and a functional link between these two proteins has been suggested in pathological conditions showing that activated AMPK adversely affects mutant TDP-43-induced motor neurons diseases
. In addition, it is related to other central autophagy proteins such as ATG5 and ATG16L, which create a multimeric complex playing an essential role in autophagosome formation, a system highly conserved in all eukaryotes
. The other proteins linked to TDP-43 are WIPI1 and 2 (Figure
5). Thus, these findings could suggest a therapeutic modulation of autophagy involving approaches that functionally target WIPI proteins and ATG5-ATG16 complex for the treatment of FTD and other diseases involving mutations in TDP-43 gene.
Apart from the metabolic-associated biological processes terms, cell surface receptor signaling pathway-related terms were also highly significant enriched, in particular in proteins associated to the Wnt pathway (Figures
4). Several proteins in the mediator list are represented in the pathway (see Additional file
4: Figure S4) including GSK-3β, a tau kinase that was also included in the most connected mediator proteins list (Table
3) and in the autophagy-related proteins (Additional file
3: Table S3). Several preclinical and clinical data strongly link GSK-3β to dementia: different inhibitors of GSK3B activity block neurodegeneration in vitro, and GSK-3β -mediated Wnt signaling can mediate amyloid peptide toxicity in vitro
[68, 69]. Finally, in human postmortem brain, this protein is physically associated with neurofibrillary tangles, one of the pathologic hallmarks of AD
. WNT pathway has also been recently linked to autophagy. In fact, autophagy negatively regulates Wnt signalling by promoting Dishevelled (Dvl) degradation, with a role for Von Hippel–Lindau protein-mediated ubiquitylation
, both of them present in the dementia network mediator list.
In our dementia network, among the drug targets associated to the autophagy-related mediators, the highest represented proteins are subunits of the Protein phosphatase 2A (PP2A; Figure
5), a serine/threonine-specific protein phosphatase consisting of A, B and C subunits that plays multiple roles in different signaling pathways and regulates diverse cellular processes. Among the six PP2A proteins, three proteins (PPP2R2B, PPP2CA, and PPP2R1A) are also listed in the highly ranked proteins in the dementia network (Table
3), demonstrating their centrality. A recent study confirms that PP2A blockade inhibits autophagy potentially through activation of AMPK
. A role of PP2A in dementia is further demonstrated by the evidence that okadaic acid and calyculin A, two potent PP2A inhibitors
, are able to induce tauopathy and cognitive deficiency in rats
[74, 75]. Thus, PP2A subunits could be considered as a potential therapeutic target for AD.
In our drug targets list related to autophagy mediators (Figure
6), other molecular targets could be considered suitable for therapeutic intervention including AMPK-related proteins, a highly ranked protein in our network (Table
3 and Additional file
3: Table S3) and a target which has been already considered for the treatment of Alzheimer’s disease. In fact, pioglitazone, an antidiabetic drug which acts also by activating AMPK
, has been proven to reverse pathological conditions in an animal model of the disease
 and it is in clinical trial for Alzheimer’s disease (http://www.clincaltrial.gov).
In more general terms, a direct action on the regulation of autophagy, potentially an activation of the autophagic process should be considered to the development of optimal therapeutics, although autophagy can function both as a cytoprotective mechanism, but it also has the capacity to cause cell death.