Towards Alzheimer's root cause: ECSIT as an integrating hub between oxidative stress, inflammation and mitochondrial dysfunction

Hypothetical role of the adapter protein ECSIT in familial and sporadic Alzheimer's disease pathogenesis



Here we postulate that the adapter protein evolutionarily conserved signalling intermediate in Toll pathway (ECSIT) might act as a molecular sensor in the pathogenesis of Alzheimer's disease (AD). Based on the analysis of our AD-associated protein interaction network, ECSIT emerges as an integrating signalling hub that ascertains cell homeostasis by the specific activation of protective molecular mechanisms in response to signals of amyloid-beta or oxidative damage. This converges into a complex cascade of patho-physiological processes. A failure to repair would generate severe mitochondrial damage and ultimately activate pro-apoptotic mechanisms, promoting synaptic dysfunction and neuronal death. Further support for our hypothesis is provided by increasing evidence of mitochondrial dysfunction in the disease etiology. Our model integrates seemingly controversial hypotheses for familial and sporadic forms of AD and envisions ECSIT as a biomarker to guide future therapies to halt or prevent AD.

Editor's suggested further reading in BioEssays Binding of amyloid peptides to domain-swapped dimers of other amyloid-forming protein may prevent their neurotoxicity Abstract

Please, also watch the Video Abstract


, amyloid beta; ABAD, amyloid binding alcohol dehydrogenase; AD, Alzheimer's disease; APOE, apolipoprotein E; APP, amyloid beta A4 protein; BMP, bone morphogenetic pathway; BMPr1a, bone morphogenetic protein receptor type-1A; BMP4, bone morphogenetic protein 4; BMP6, bone morphogenetic protein 6; CYP2C18, cytochrome P450 2C18; ESCIT, evolutionarily conserved signalling intermediate in toll pathway; FAD, familial AD; GCDH, glutaryl-CoA dehydrogenase; IFIT5, interferon-induced protein with tetratricopeptide repeats 5; IP3, inositol 1,4,5-triphosphate; LONP1, lon protease homolog 1; LOXL4, lysyl oxidase-like 4; MAPK, mitogen-activated protein kinase; mt, mitochondrial; NADPH, nicotinamide adenine dinucleotide phosphate; NF-κB, nuclear factor kappa-light-chain-enhancer of activated B cells; NFT, neurofibrillary tangle; NMDA, N-methyl-D-aspartate; PRDX2, peroxiredoxin-2; PSEN1, presenilin 1; PSEN2, presenilin 2; PTP, phosphate transport protein; ROS, reactive oxygen species; SAD, sporadic AD; SMAD, mothers against decapentaplegic homolog; tau, microtubule-associated protein tau; TGF-β, transforming growth factor-beta; TLR, toll-like receptor; TLR4, toll-like receptor 4; TRAF6, TNF receptor-associated factor 6.


Alzheimer's disease (AD) is a progressive, incurable neuro-degenerative disorder, neuro-pathologically characterised by the extra-cellular accumulation of amyloid-beta (Aβ) plaques, and the intra-cellular accumulation of hyper-phosphorylated tau protein in the form of neurofibrillary tangles (NFTs). These aggregations likely lead to subsequent neuronal degeneration in the brain regions important for memory and cognition. The increase in adult life expectancy during the past century has made AD the most common form of dementia in humans 1.

There are two known types of AD: early-onset familial AD (FAD), which is a rare form diagnosed before the age of 65, and late-onset sporadic AD (SAD), the most common form, accounting for about 95% of cases and usually occurring after the age of 65.

Familial AD genetics has shed light on early-onset pathogenesis (Table 1), but the etiology of SAD remains largely unclear, since it is probably influenced by both genetic and environmental risk factors 2.

Table 1. Alzheimer's disease genes
Gene (and protein)Chromosomal locationPrevalence in ADRelevance to AD pathogenesisReferences
  1. Genes and encoded proteins involved in early-onset familial (FAD) or late-onset (SAD) Alzheimer's disease. The prevalence of gene mutations and their relevance in AD are also indicated.

APP (amyloid precursor protein)21q21.3 2910–15% of FADIncrease in Aβ production or Aβ42/Aβ40 ratio


PSEN1 (presenilin 1)14q24.330–70% of FADIncrease in Aβ42/Ab40 ratio; enzymatic role in γ-secretase complex


PSEN2 (presenilin 2)1q31-42<5% of FADIncrease in Aβ42/Ab40 ratio; enzymatic role in γ-secretase complex


APOE (apolipoprotein E)19q13.32Allele ε4 increases risk 20–90% of SADAccumulation of Aβ senile plaques; reduction of choline-acetyltransferase in synaptic transmission


The amyloid cascade hypothesis first proposed accumulation of Aβ as the cause of the disease 3 (Fig. 1). More recently, new evidence has suggested that oligomeric Aβ, rather than extracellular fibrillary Aβ, may be the most toxic entity 4, 5. Nevertheless, the poor correlation between Aβ deposits and SAD has prompted a re-examination of the etiological significance of Aβ, since Aβ-lowering therapies have failed to produce any tangible benefit; neither diminishing disease prevalence nor progression 6, 7.

Figure 1.

Amyloid-beta formation and amyloid cascade hypothesis. A: Processing of APP protein by the sequential cleavage of β- and γ-secretase complexes, resulting in the formation of Aβ peptides of different lengths. APOE, besides its primary function in lipid transport, is also involved in the deposition or clearance of the extracellular/intracellular Aβ pool, although the underlying mechanism is still unclear 93 B: The sequence of pathogenic events leading to AD proposed by the amyloid cascade hypothesis 3.

Alternatively, instead of a linear chain of events, diverse pathways could initiate and drive AD, in which some alterations would be reactionary or protective but not pathologic. Along this line of reasoning, several competing hypotheses have been proposed, implicating critical ageing-related processes as the disease trigger 8. It is still under debate whether any of these factors lead to amyloid deposition and tauopathy in humans 1.

In our quest to uncover potential causative/susceptibility genes related to the disease, we identified evolutionarily conserved signalling intermediate in Toll pathway (ECSIT) as a promising candidate underlying AD mechanisms 9. It encodes for an adapter protein that intersects the immune toll-like receptor (TLR) and the homeostatic bone morphogenetic pathway (BMP)/transforming growth factor-beta (TGF-β) signalling pathways 10. ECSIT functions as a cytoplasmic, and to a lesser extent a nuclear, signalling protein; although, an N-terminal signal might also direct ECSIT to mitochondria. Indeed, ECSIT has been recently implicated in the assembly and stability of the mitochondrial respiratory complex 11. Nevertheless, the exact mechanism by which ECSIT mediates this cross-talk still remains uncertain.

Here we discuss the role of ECSIT as a nexus among signalling pathways and we hypothesise its implication in the etiology of AD.

Early events in AD pathology

The innate immune response: Inflammation

Even though neuro-inflammation is an early feature of AD, it still remains unclear whether inflammation initiates or accelerates AD or simply represents a defensive response to protect the damaged brain 12.

Amyloid-beta plaques and NFTs are frequently associated with reactive astrocytes and activated microglia 13, which can stimulate the production of reactive oxygen species (ROS) in response to Aβ-deposition 14. However, Aβ could also have a pathological role in early phases of the disease 15, since amyloid peptides can be considered as damage-associated molecular patterns (DAMPs) that specifically recognise innate immune receptors (i.e. TLRs). Such receptors can initiate an intracellular signal that induces pro-inflammatory responses 16. Furthermore, other studies indicate that inflammatory factors can be perturbed at the genetic level, which would also imply a more causal involvement in AD 12. Intriguingly, neurons are also capable of locally synthesising inflammatory cytokines 17.

In summary, different factors may synergise and lead to inflammation, either as a primary or secondary event, which emphasises the complexity of AD patho-physiology.

Mitochondrial dysfunction: Oxidative stress

Mitochondrial dysfunction is another of the earliest and most prominent features of AD 18. Mitochondria play essential functions in cell survival or death in order to maintain a healthy mitochondrial population, in particular in long-lived cells such as neurons 19. Mitochondria generate energetic potential through respiratory complexes I to IV, which constitute the electron transport chain (ETC) (Fig. 2). Compared to other organs, bio-energetic production is critical in the brain, since neurons are extremely energy-demanding but have limited glycolysis, making them highly dependent on aerobic oxidative phosphorylation (OXPHOS) 20. Moreover, the ability of mitochondria to move within cells is also critical in highly polarised cells, such as neurons 21.

Figure 2.

ECSIT as an integrating hub and a biological switch regulator of Aβ pathology. The adaptor protein ECSIT is a key component in multiple signalling pathways as well as in the assembly of the mitochondrial respiratory chain (ETC, complexes I to IV). With the additional complex V, the OXPHOS system becomes functional, which is crucial for cell bioenergetics. Interactions in our AD-PIN network 9 revealed that ECSIT is also functionally connected with well-established AD causative genes (see Table 1). The position of ECSIT in this molecular circuitry indicates that it functions as an integrating hub that may constitute a molecular switch to soothe/exacerbate amyloid pathology (see the legend for details of symbols. Biological processes affected by AD are highlighted in red). Albeit the molecular mechanisms by which ECSIT mediates the cross-talk remain uncertain, we suggest that a fine balance between ubiquitination, alternative splicing and degradation could be responsible for different levels of temporal/spatial intervention.

Thus, it is not surprising that mitochondrial dynamics can be greatly affected in response to physiological or environmental alterations, with severe consequences for neuronal function and survival. Indeed, altered levels of OXPHOS enzymes, particularly down-regulation of respiratory complex I genes, have been found to be directly responsible for a decrease in energy production in the brains of AD patients 22, 23. As a consequence, ROS production is significantly increased, which results in Ca2+ buffering impairment and the release of pro-apoptotic proteins. This triggers the activation of the caspase cascade and the alteration of mitochondrial dynamics such as mitochondrial fission/fusion or mitophagy 24. In addition, Ca2+ storage in the endoplasmic reticulum (ER) is also affected through the mitochondria-associated ER membranes (MAMs) 25, compromising neuronal Ca2+ homeostasis and, hence, neuro-transmission in synaptic terminals 24, 26.

Swerdlow et al. 27 and Swerdlow and Khan 28 proposed the mitochondrial cascade hypothesis to explain late-onset SAD. This hypothesis implies that AD phenotypic expression may involve two factors: a predisposing mutation and an age-related factor that causes a decline in mitochondrial functioning that amplifies the inherited defect. In fact, oxidative damage may cause somatic mtDNA mutations, which have been shown to increase in abundance with age in the mtDNA of both human and mouse brains 29.

Nevertheless, the causal role of mitochondrial dysfunction in the disease is still controversial; although it is clear that mitochondrial dysfunction, abnormal mitochondrial dynamics and degradation by mitophagy occur during the disease process 19.

Amyloid-beta, cause or effect of mitochondrial dysfunction?

Amyloid-beta progressively accumulates within mitochondria of AD brain cells, which correlates with abnormal synaptic morphology and deficits in long-term synaptic plasticity 30. Aβ can directly affect respiration, with a consequent increase in ROS production. Interactions of Aβ with mitochondrial proteins such as amyloid binding alcohol dehydrogenase (ABAD) exacerbate the toxic effects on mitochondrial and neuronal functioning 31 (Fig. 2).

However, the origin of mitochondrial Aβ is still under debate. Aβ accumulation might derive from endosomes or the ER/golgi subcellular compartments 32. Indeed, it has been hypothesised that oligomeric Aβ has the ability to permeabilise cellular membranes and lipid bilayers and can thus enter organelles such as mitochondria 33. Conversely, it might derive from mitochondria-associated amyloid beta A4 protein (APP), since there is experimental evidence of localised Aβ production by a mitochondrial γ-secretase complex (Fig. 1A) 34.

Overall, these findings provide compelling evidence that mitochondria-associated and/or intra-mitochondrial Aβ may directly cause neurotoxicity. This is in accordance with the modified model of the amyloid cascade hypothesis that suggests that the toxic Aβ species may be soluble intracellular aggregates instead of extracellular insoluble plaques (Fig. 1B) 35. Nevertheless, what seems certain is the importance of the regulation/modulation balance in the Aβ pool between intracellular and extracellular compartments under patho-physiological conditions (Fig. 1A) 36.

Exploiting network biology to explore molecular mechanisms underlying AD: The multi-functional role of ECSIT

Our current understanding of the complexity underlying AD is insufficient to address the unsolved issues and controversies mentioned above, making the identification of new approaches to tackle the disease challenging. With the advent of genomic and proteomic technologies, network and systems biology now offer a holistic paradigm to explore the molecular mechanisms of AD, beyond individual genes or proteins 37–39.

Identification of ECSIT at a late-onset AD susceptibility chromosomal region

Interestingly, gene linkage analyses have suggested that the 19p13.2 chromosomal region contains susceptibility loci involved in the etiology of SAD or FAD with unknown genetic cause 40, 41. However, besides the APOE-ε4 gene, very few associations could be established with specific genes 42. Therefore, in an attempt to identify novel genes in this locus, we applied an interaction discovery approach to generate an AD-associated protein interaction network (AD-PIN), based on our previous findings that AD causative and susceptibility genes tend to be physically connected 9. We discovered 200 novel protein-protein interactions between AD causative and susceptibility gene products. Among these, we identified ECSIT as a central node in the AD-associated network and, since it is located at this susceptibility region 43, MIM:608907, it emerged as a potential candidate implicated in AD mechanisms.

Role of ECSIT as a nexus among inflammation, homeostasis and mitochondrial dysfunction

Toll-like receptors are located at the cell surface and can recognise a variety of extracellular ligands such as pathogen- or damage-associated molecules. Upon binding, they lead to the activation of the NF-κB transcription complex and, in turn, the induction of mediators of the innate immune response 44. Additionally, the TLR pathway can activate the mitogen-activated protein kinase (MAPK) phosphorylation cascade, which is involved in gene regulation of a wide variety of key processes 45. Engagement of a subset of TLRs results in the recruitment of the signalling adaptor TNF receptor-associated factor 6 (TRAF6) to specifically bind ECSIT, which then determines the activation of the NF-κB or MAPK signalling pathways (Fig. 2) 46. Notably, after stimulation of the toll-like receptor 4 (TLR4) receptor with its agonist, the lipopolysaccharide (LPS), knockdown ECSIT cells show severely compromised TLR signalling 47. Intriguingly, TLR4 receptors are present in neurons and are particularly activated in response to brain injury 48.

The BMP is required for normal embryonic development or adult tissue homeostasis and is mechanistically connected to TGF-β signalling 49, which regulates cell differentiation, proliferation and apoptosis via mothers against decapentaplegic homolog (SMAD) proteins 50. In addition, TGF-β also plays anti-inflammatory and immuno-suppressive roles 44, 51. As expected, targeted disruption of ECSIT causes early mouse embryonic lethality, but interestingly, knockdown ECSIT stem cells display reduced proliferation and generate poorly differentiated neural cells. Moreover, ECSIT ablation completely abrogates bone morphogenetic protein receptor type-1A (BMPr1a) receptor signalling and precludes the formation of SMAD complexes (Fig. 2) 47.

Earlier studies had indicated that ECSIT could be potentially modified during TLR signalling 10, although it is not functional in the BMP pathway. Notably, recent findings have shown that interaction with TRAF6 leads to ECSIT ubiquitination 52, suggesting that this modification could thereby account for the observed antagonism between the TLR and BMP/TGF-β pathways 53. Alternatively, ECSIT could also be sequestered in nuclear SMAD complexes to activate TGF-β signalling 47, since LPS up-regulates cytosolic inhibitory SMADs through the TLR pathway 54. An additional layer of complexity is provided by alternative splicing of the ECSIT gene (data from the ENSEMBL genome browser, ENSG00000130159). Indeed, the ECSIT-2 isoform has been shown to localise both to the cytoplasm and nucleus, where it associates with SMADs in a BMP-inducible manner 47.

Furthermore, it has recently been reported that, after exposure to LPS, macrophages show ECSIT enrichment at the mitochondrial periphery, resulting in augmentation of mitochondrial ROS and a higher sensitivity of ECSIT to protease digestion 52. Therefore, ECSIT might modulate energetic requirements upon inflammation by regulating the OXPHOS complex, enhancing redox signalling and, in turn, the generation of effector responses (Fig. 2) 11, 52.

In summary, ECSIT arises as an essential component of three signalling pathways and several studies provide further evidence that it might act as a biological switch in determining which pathway is functional at a particular time 47. Even though it undergoes tight regulation both spatially and temporally, it is unclear which factors determine whether ECSIT is targeted to the mitochondrion, the nucleus or the cytosol. As a result, the exact mechanism by which ECSIT mediates this cross-talk still remains uncertain. Nevertheless, a fine balance between ubiquitination, alternative splicing and degradation could be responsible for the remarkable complexity of ECSIT functions at different temporal/spatial levels of intervention.

Role of ECSIT as a central node in the AD-protein interaction network

In support of previous findings, our AD-PIN revealed several functional modules that also point to ECSIT as an interacting factor with proteins involved in redox signalling and immune responses (Fig. 2) 9. In particular, we observed the association of ECSIT with the mitochondrial proteins lon protease homolog 1 (LONP1), which is required for intra-mitochondrial proteolysis as a cellular response to oxidative stress, and glutaryl-CoA dehydrogenase (GCDH), which is involved in redox signalling. Furthermore, ECSIT also interacts with other ER redox proteins, like lysyl oxidase-like 4 (LOXL4) and the cytochrome P450 2C18 (CYP2C18), involved in nicotinamide adenine dinucleotide phosphate (NADPH)-dependent electron transport, which in turn modulates Ca2+ release channels (e.g. IP3 and ryanodine receptors) on ER membranes 55. Notably, ECSIT associates with peroxiredoxin-2 (PRDX2), which becomes oxidatively inactivated when ROS levels exceed the detoxifying abilities of the antioxidant systems, as appears to be the case in mouse AD brains 56. Conversely, expression of interferon-beta-induced proteins like interferon-induced protein with tetratricopeptide repeats 5 (IFIT5) is up-regulated in AD brain tissues. Our findings that ECSIT physically interacts with both PRDX2 and IFIT5 gives insights into the mechanisms by which ECSIT modulates energetic requirements upon oxidative stress and inflammatory responses (Fig. 2).

Most interestingly, we have observed a novel association of ECSIT with apolipoprotein E (APOE) 9, which has been shown to bind Aβ 57. Thus, ECSIT could be indirectly implicated in Aβ extracellular clearance (Fig. 2). Moreover, in AD affected neurons, APOE undergoes proteolysis to generate an N-terminal fragment that associates with NFT-like structures and a C-terminal truncation that binds to mitochondrial respiratory complexes, reducing their activities 58–60. These results give insights into the association of APOE with cytoskeletal components in order to induce NFT-like inclusions containing phosphorylated tau and phosphorylated neurofilaments of high molecular weight in neurons 58. However, it still remains unclear how the APOE fragment associates with mitochondria. The direct interaction of APOE with ECSIT provides an association mechanism that would place APOE on the mitochondrial membrane for respiratory complex inhibition in affected cells 9. Furthermore, the additional interactions of ECSIT with the APP-cleaving enzymes, presenilin 1 (PSEN1) and presenilin 2 (PSEN2) 9, suggest that ECSIT might also be directly involved in Aβ production. FAD mutations in presenilins result in loss of ER Ca2+ buffering, which is independent of their γ-secretase activity and thus APP processing 61. Intriguingly, APOE has also been associated with synaptic Ca2+ signalling 62. These additional roles might contribute to the tight compartmentalisation and regulation of Ca2+ signalling and, therefore, ECSIT could act as an indirect calcium regulator to maintain cellular homeostasis and synaptic transmission.

Hypothesis: The role of ECSIT as a double-edge sword in AD pathogenesis

Taken together, these observations indicate that ECSIT acts as a molecular sensor that is required for the interpretation of multiple signalling pathways, the integration of these signalling inputs and the generation of precise cellular responses in a temporal/spatial pattern. This ability to orchestrate these effects has prompted us to elaborate a hypothesis for its mechanistic roles in AD pathogenesis (Fig. 3).

Figure 3.

The hypothetical ECSIT signalling mechanism in AD pathogenesis. The orchestrating ability of ECSIT has prompted us to elaborate a hypothesis as to its mechanistic roles in AD pathogenesis. ECSIT may serve as a molecular sensor of damage signals, both extracellularly- or intracellularly-derived. These initiatory signals can originate from multiple sources, i.e. from gene mutations, unbalanced calcium/redox systems or extracellular pathogens. In an attempt to protect the cell, ECSIT could increase bioenergetics and/or protective molecular mechanisms like immune signalling. Concomitantly, these processes would unleash parallel processes like inflammation or oxidative stress, which in turn would result in the accumulation of Aβ oligomers. These oligomers could trigger their toxicity through specific cell receptors or by inducing mitochondrial dysfunction, progressively enhancing the response mechanisms with the aim of repairing the neuronal damage. Nevertheless, failure to repair would switch ECSIT into a detrimental factor by activating an apoptotic cascade to trigger neuronal death. Proposed mechanisms in (A) familial early-onset AD, and (B) in sporadic late-onset AD. Potential initiatory signals are highlighted in bold. mt, mitochondrial; DAMPs, damage associated molecular pattern molecules; PAMPs, pathogen associated molecular pattern molecules.

The origin of initiatory pathogenic molecules in early-onset versus late-onset AD

Our hypothetical model begins with the well-established etiogenesis for early-onset FAD, where initiatory signals of genetic origin generate an Aβ burden in the extracellular space and/or in the mitochondria (see Table 1 and Fig. 1A; reviewed in ref. 1).

These altering signals would be more heterogeneous in the sporadic late-onset form, in accordance with reported observations 63. Nevertheless, even though recent studies propose 19, 28 that they could originate predominantly in the mitochondria, extracellular ligands could also exacerbate initial deregulation (Fig. 3). Such pathogenic signals would emerge from events such as an age-associated accumulation of oxidative radicals, inherited mitochondrial DNA mutations, carriage of the APOE-ε4 risk factor, among others 8, 27, 64. These presumptive inducers are not necessarily in accordance with the mitochondrial cascade hypothesis 27, since they might also include other competing hypotheses like amyloid accumulation, brain injury or infection 65, 66.

Neuro-protective role of ECSIT from extracellular pathogens: Activation of the immune system

Extracellular damage-initiating signals would stimulate cell-surface TLR4 receptors (Fig. 2), which would activate TLR intracellular signalling and, thereby, induce ubiquitination of ECSIT 52. Ubiquitinated ECSIT could then trigger an inflammatory cascade 10, leading to an increase in immune response factors, such as interferon-induced proteins (IFIT), and redox-regulating factors like PRDX2 to balance cytosolic ROS levels. These would become progressively more inefficient until they no longer could protect against oxidative damage.

Intriguingly, Aβ is highly hydrophobic and therefore could also induce innate immune responses similar to those triggered by the TLR4 receptor agonists. Interestingly, microglia activation due to LPS administration seems to mimic some features of AD such as increased Aβ levels 14, suggesting that TLR4 very likely plays a role in extracellular Aβ uptake. In fact, TLR4 expression increases in neurons when exposed to Aβ42 67 and, furthermore, a TLR4-attenuated polymorphism has been associated with protection against late-onset AD development 68. Therefore, an initial accumulation of extracellular Aβ could also be a pathogenic signal, as proposed in the amyloid cascade hypothesis 3.

Neuro-protective role of ECSIT from intracellular pathogens: Induction of oxidative stress

Alternatively, endogenous molecules could trigger their effect by challenging the mitochondrial redox-surveillance system (Fig. 3). This would eventually become saturated in an attempt to down-regulate excessive ROS production, leading to mitochondrial oxidative stress and subsequently to an increase in mitochondrial Aβ levels 28, 69, 70. The stress response would promote the mitochondrial translocation of ECSIT to enhance bioenergetics but, concomitantly, it would also increase mitochondrial ROS as a by-product. This, in turn, would lead to the up-regulation of ATP-dependent proteases such as LONP1 to control the selective turnover of mitochondrial proteins 71, 72. Our observed interactions with ECSIT (Fig. 2) provide evidence that LONP1 can eventually target ubiquitinated ECSIT towards proteosomal degradation and, hence, precludes the assembly of the ETC complex 72. As a result, mitochondria would induce anaerobic glycolysis, producing even higher levels of ROS that would amplify metabolic oxidation. Exposure to high levels of oxidants, especially in the presence of calcium, could then induce opening of the non-selective ion channels (i.e. phosphate transport protein, PTP) between the mitochondrial matrix and the cytoplasm 73. This would enable the diffusion of ions and ROS to the cytosol, thereby triggering the release of damage-related signals (Fig. 2) 52. Additionally, the increase in intracellular free calcium would directly affect the efflux through ER Ca2+-receptors in synaptic terminals, compromising synaptic transmission 74, 75 and exacerbating Aβ-mediated excitotoxicity through N-methyl-D-aspartate (NMDA) receptors 76.

Neuro-reparative role of ECSIT for cell damage

In any case, amyloid-driven damage would converge to a common pathological cascade, ramping up TLR signalling and, in turn, activating ECSIT to regulate this paracrine loop (Fig. 3). The direct binding of ECSIT to PSEN and APOE 9 suggests that it could simultaneously regulate the extracellular/intracellular Aβ pool and neuronal Ca2+ homeostasis. As a consequence, this initial paracrine feedback would progressively expand as an autocrine loop to other neuronal cell types like microglia, with the effect of enhancing the immune response to repair the neuronal damage 77.

Moreover, as the cell goes anaerobic, mitochondrial dysfunction initiates a dedifferentiation response, as observed in AD neurons 28, 78. Intriguingly, hippocampal AD cells show increased expression of BMP4 levels, which signals through BMPr1a receptors and ECSIT, resulting in negative modulation of the proliferation and further inhibition of neurogenesis 79. The ubiquitination of ECSIT could provide a mechanistic explanation of this process, since it would inactivate the differentiation pathways by precluding its binding to SMAD complexes and impairing their translocation to the nucleus 47, 52, 80. Additionally, ECSIT could switch on the MAPK pathway, unleashing a phosphorylation cascade that would result in hyperphosphorylation of tau 81. Very interestingly, it has recently been reported that TGF-β signalling can be disrupted through the sequestration of SMADs by phosphorylated tau in the cytoplasm 82, which provides further support for the role of ECSIT as a mechanistic signalling switch.

Neuro-apoptotic role of ECSIT in irreversible damage

Nevertheless, failure to repair would eventually trigger the induction of an apoptotic response via the activation of BMP/TGF-β signalling, probably by destabilising SMAD-tau complexes from microtubule networks 81. The release of phosphorylated tau could then favour the formation of insoluble oligomers and aggregates. Thus, acute up-regulation of BMP expression would start as a compensatory response, but it would become deleterious once chronic, as has been observed for BMP6 levels upon sustained treatment with exogenous Aβ 83. Furthermore, high-doses of Aβ would also behave as a specific TLR4 antagonist, inhibiting immune signalling 84. Once the pro-apoptotic signal is generated, secondary events could amplify the disease process, ultimately leading to neuronal death 85.

Conclusions and prospects

We provide a mechanistic hypothesis that integrates the multiple abilities of ECSIT as an integrating hub and biological switch regulator, placing it as a key factor in Aβ pathology (Fig. 2). Indeed, ECSIT may serve as a molecular sensor of multiple damage signals, as a molecular effector through the augmentation of bioenergetics or the specific regulation of protective molecular mechanisms, and as a molecular switch by triggering an apoptotic cascade when repair processes fail (Fig. 3).

Our hypothesis integrates seemingly controversial hypotheses, since although accumulation of Aβ is a main pathological player in both AD forms, it is causative in FAD whereas it is a consequence of SAD, as recently discussed 86. Hence, in support of previous postulations, the nature of the response, not the nature of the injury, determines the outcome of AD 87.

This hypothetical ECSIT signalling mechanism would also solve the debate as to whether mitochondrial function alters Aβ dynamics or Aβ alters mitochondrial functioning 28 and would provide further insights into the coupling of mitochondria with immune signalling 88, where ECSIT could act as the required energetic checkpoint.

Furthermore, it reasons that Aβ oligomers trigger their toxicity through specific neuronal (or glial) receptors rather than indiscriminately perturbing diverse receptor or channel proteins 4, suggesting that ECSIT could conduct the beneficial or detrimental effects that Aβ might execute at particular stages of the disease.

Taken together, ECSIT is postulated to be an early pathogenic sensor in AD, and it is therefore envisioned as a promising biomarker to detect AD in pre-symptomatic stages. The combination of ultra-sensitive analytical methods such as bionanotechnology with molecular imaging tools could help to further elucidate ECSIT dynamics and to evaluate its diagnostic accuracy. Additionally, functional assays in AD-like phenotype cell and animal models will be required to better characterise the physiological functions of ECSIT as an integrating hub, which could guide the identification of novel drug targets capable of halting or preventing the disease.


This work was partially supported by the Spanish Ministerio de Ciencia e Innovación (PSE-010000-2009-1; BIO2010-22073) and the European Commission under FP7 Grant Agreement 223101(AntiPathoGN).