Introduction
Alzheimer’s disease (AD) is the most common neurodegenerative disorder and has an increasing effect on our ageing population. Pathological hallmarks of AD are extracellular amyloid beta (Aβ) deposits and intracellular accumulation of hyper-phosphorylated tau protein leading to the formation of neurofibrillary tangles (NFTs) [[
1]]. In addition, progressive synaptic dysfunction is thought to occur in early stages of the disease and has been found to correlate closely with cognitive deficits observed in patients with AD [[
2]–[
4]]. There is emerging evidence that the erythropoietin-producing hepatocellular (Eph) receptors and their ligands, the so-called ephrins, are involved in aberrant synaptic functions associated with cognitive impairment in AD [[
5]]. Eph/ephrin signaling is required for a wide range of biological processes both during embryogenesis and adult life and involves the Eph receptors which form the largest of the 20 subfamilies of human receptor kinases. Eph/ephrin signaling plays a role not only during synapse formation and maturation and synaptic plasticity [[
6]–[
8]] in the brain but also in directing cell positioning and migration, axon guidance [[
9],[
10]], control of tissue morphogenesis, patterning, tumour invasion and metastasis, immune function [[
11],[
12]], haematopoiesis and blood clotting [[
13]] and tissue repair and maintenance.
Eph receptors and their ligands are exclusively membrane-bound and hence cell-cell contact is required for activation of the kinase through oligomerisation and transphosphorylation [[
14]]. EphA4 is the Eph receptor family member that is most highly expressed in the adult hippocampus where it plays a role in adult synaptic plasticity and learning [[
15],[
16]]. The EphA4 kinase is pre- and post-synaptically expressed on dendritic spines of pyramidal neurons and axon terminals [[
17]]. Emerging evidence supports a critical role for EphA4/ephrin A3 signaling in the regulation of spine morphology in the hippocampus. Activation of EphA4 upon binding to its glia-derived ligand ephrin A3 was found to induce spine retraction and to trigger the reduction of dendritic spines and synaptic proteins, whereas inhibiting those interactions led to distorted spine shape and organization in the murine hippocampus. These findings suggest an essential role for EphA4 in the elimination of excitatory synapses [[
18]–[
20]].
Two major forms of Aβ coexist in the brain: a shorter form with 40 amino acid residues and a longer form with 42 amino acids. The longer form is extremely toxic and can self-aggregate to form oligomers (amyloid beta oligomers, AβOs). Increased levels of EphA4 in cultured neurons and synaptoneurosomes was reported to be crucially involved in synaptic damage induced by AβOs [[
21]]. Interestingly, reduced expression of the EphA4 receptor has been linked to cognitive impairment in a transgenic mouse model for AD overexpressing the human amyloid beta precursor protein (APP) [[
22]]. Loss of synapses is an early event in AD pathogenesis. It has therefore been suggested that changes in hippocampal EphA4 signaling might precede the onset of memory decline in AD. Whether EphA4 levels and activation are altered in human AD brain is not known.
In the present study we are the first group to report the involvement of EphA4 in AD pathology. We have investigated EphA4 expression levels and localization in human brain tissue of patients with AD and non-demented controls. An association of EphA4 with the hallmarks of AD was investigated using sequential single stainings and double-labelling with phosphorylated tau and amyloid beta.
Discussion
Synaptic loss is one of the major pathological hallmarks of AD. This is considered to be an early event in the pathogenesis of the disease. Synaptic failure correlates with cognitive decline and is observed in patients with mild cognitive impairment (MCI) and incipient AD [[
3]]. The molecular mechanism of synaptic dysfunction in AD remains elusive.
The tyrosine kinase receptor EphA4 is essential for synaptic function as it is involved in dendritic spine morphogenesis, synapse formation and maturation [[
5]]. EphA4 is highly expressed in the adult hippocampus, where it is known to play a role in adult synaptic plasticity and learning [[
16],[
30]]. Of all cortical areas, the hippocampus appears to be most severely affected by the loss of synaptic proteins in AD (44 to 55%) [[
31]]. Interestingly, analysis of synaptoneurosomes from AD patients revealed a ~2-fold increase in EphA4 mRNA [[
32]], suggesting a role in synaptotoxicity for EphA4 Therefore we investigated EphA4 expression and localization in the hippocampus of patients at various Braak stages.
In this study the total EphA4 protein levels were similar in AD patients compared to control cases. However, immunohistochemical localization of EphA4 revealed an altered distribution in AD compared to control hippocampus. EphA4 partially co-localized with neuritic amyloid beta plaques. An aberrant function of EphA4 might be the underlying cause for the altered distribution of EphA4 in different Braak stages. Thus, EphA4 could be contributing to synaptic dysfunction which is considered an early event in AD.
So far, reports about changes in EphA4 protein levels in AD hippocampus have been contradictory. Simón et al. reported a reduction of EphA4 (20%) in hippocampal tissue of three patients with very mild cognitive deficits (Braak stage II and III) compared to three control subjects [[
33]]. Matsui et al. showed that total protein levels of EphA4 in AD brains were not altered compared to controls. Furthermore they reported a decrease of membrane associated EphA4 (intracellular domain) in frontal lobes of AD cases while the amount of full-length EphA4 was unchanged [[
34]]. In the present study we investigated a large cohort (n = 35) covering all Braak stages using two different antibodies directed at different epitopes of EphA4. EphA4 fragments with different molecular weights (108 kDa vs. 130 kDa) were detected. This discrepancy can be explained by the different binding sites of the used antibodies. Overall, we were not able to detect changes in EphA4 protein levels between control and AD groups and between different Braak stages when we used Western Blot analysis.
In contrast, at the immunohistochemical level we observed differences in the staining patterns of EphA4 when comparing control and AD cases. We showed that depositions of the EphA4 protein kinase are present in all subareas of the hippocampus in AD patients. The number of EphA4 deposits increases with Braak stage and those deposits partially co-localize with neuritic plaques and tangles. Moreover, EphA4 immunoreactive plaques are already present in Braak stage II which points towards an involvement of EphA4 in the early stages of AD pathology. The increased occurrence of EphA4 deposits with AD pathology in the absence of changes in total EphA4 protein levels indicate an altered distribution of EphA4. The important role of EphA4 in synaptic dysfunction, an early event in AD, has been reported [[
35]]. Pathological changes happen in the brain even decades before the first clinical symptoms emerge. Therefore, it is likely that the aberrant EphA4 staining in part of the control cases (Braak stage II and III) poses an early event in AD pathogenesis and is therefore specific.
Mostly, investigations into AD related synaptic changes have focused on the toxic effects of Aβ. Like the amyloid precursor protein (APP), EphA4 is a substrate of γ-secretase [[
34]]. The EphA4 intracellular domain (EICD) that remains after cleavage is known to enhance the formation of dendritic spines via activation of the Rac signaling pathway [[
34]]. It has been suggested that aberrant γ-secretase activity followed by hindered cleavage of EphA4 results in reduced formation of dendritic spines and may be the major cause of synaptic failure in AD [[
36]]. In addition, reduced EphA4 levels have been reported in whole-cell lysates of hAβPP
swe-ind mice and Tg2576 mice compared to non-transgenic mice [[
33]]. Those hAβPP
swe-ind mice show amyloid-related pathology but no accumulation of pTau in tangles. These data strengthen the possible link between Aβ and aberrant EphA4 signaling.
The emerging view is that toxic amyloid-β oligomers (AβOs) are an important pathological factor in early neurodegenerative events in AD [[
37]]. Two major forms of Aβ coexist in the brain: a shorter form with 40 amino acid residues and a longer form with 42 amino acids. The longer form is extremely toxic and can self-aggregate to form oligomers (amyloid beta oligomers, AβOs) [[
2]]. Those oligomers accumulate into Aβ deposits in patients with AD. It has been reported recently that activation of the Abelson non-receptor tyrosine kinase c-Abl, a kinase downstream of EphA4, mediates synaptic loss and long term potentiation in dendritic spines of cultured rat hippocampal neurons. The co-localization of c-Abl with amyloid plaques, neurofibrillary tangles and granulovacuolar degeneration in the hippocamups and entorhinal cortex of AD patients has been reported in 2009 by Jing et al. [[
38]]. AβOs activate the c-Abl kinase and thereby induce synaptic loss [[
21]]. Concomitantly, EphA4 tyrosine phosphorylation is increased in these cultured neurons and in synaptoneurosomes exposed to AβOs. EphA4/c-Abl activation is a key-signaling event mediating the synaptic damage induced by AβOs. EphA4/c-Abl signaling could hence be a relevant pathway involved in the early cognitive decline observed in AD patients [[
21]].
In human brain, a stronger relation between EphA4 and pTau positive plaques exists. In AD, c-Abl is detected in neurofibrillary tangles [[
38]] and phosphorylates tau directly [[
39]] and through activation of Cdk5 [[
40]]. In the present study, we show a significant correlation between EphA4 positive depositions and pTau-positive plaques and an almost significant correlation with pTau-positive tangles in AD. Also Matsui et al. previously reported a correlation between the intracellular domain of EphA4 (EICD) and tau phosphorylation, although the correlation did not reach statistical significance. In contrast, the level of EICD did not correlate with the level of Aβ [[
34]].
Tau may be involved in synaptic dysfunction in dementia [[
36]], strengthening the association with EphA4 [[
5]]. A key question remains whether the association with EphA4 is causally related to tau phosphorylation and aggregation. Either aberrant EphA4 signaling promotes phosphorylation of tau contributing to the formation of neurofibrillary tangles and neuronal dystrophy, or on the contrary, aggregating tau affects the subcellular distribution of key proteins involved in synaptic function such as the EphA4 receptor. So far several groups have reported that Eph/ephrin signaling up-regulates tau expression and phosphorylation [[
41],[
35]]. When EphA4 is activated by ephrin A1 it recruits and phosphorylates cyclin-dependent kinase 5 (CDK5) [[
35]]. CDK5 is a tau kinase and is increased in AD brain [[
42]]. Increased CDK5 immunoreactivity is observed in pretangle neurons supporting its involvement in early stages of AD pathogenesis [[
43]]. The significance of a dysregulation of CDK5 by EphA4 in pathological conditions remains elusive.
In our study, EphA4 depositions partially co-localize (~30%) with neuritic Aβ plaques in human hippocampal brain tissue. Manczak et al. recently reported a physical interaction between Aβ and phosphorylated Tau [[
44]]. They found an abnormal interaction of oligomeric Aβ with pTau in neurons of post mortem brains from AD patients. This interaction may be involved in neuronal and synaptic damage, leading to cognitive decline in incipient AD patients [[
44]]. Those findings may explain that EphA4 is related to both Aβ and pTau and suggests an underlying common pathway. Further studies are necessary to shine light on the relation of those proteins and the significance in the progression of AD.
Four recent large late-onset AD (LOAD) genome-wide association studies (GWAS) have identified EPHA1 as a genetic factor. Like EphA4, EphA1 is suggested to be important for synaptic function [[
12],[
45]]. This supports the emerging evidence that the Eph receptors and their ligands, the ephrins, are involved in aberrant synaptic functions associated with cognitive impairment in AD [[
5]]. In this study we show that EphA4 co-localizes with neuritic plaques in human brain tissue of patients with AD. EphA4 is associated with both Aβ and pTau. The altered distribution of EphA4 in AD hippocampus may reflect a decreased function of EphA4, which is likely to contribute to synaptic dysfunction that occurs in the early stages of AD. Pathological changes happen decades before the first clinical symptoms emerge explaining EphA4-positivity in early Braak stages (II and III, see Table
2).
Competing interests
The authors have no conflicts of interests to declare.
Authors’ contributions
AFNR coordinated the study and was responsible for writing the manuscript. AFNR and JJMH designed and performed experiments. AJMR was responsible for the autopsy material and neuropathological evaluation. SMvdV, AJMR, WvdF, PS and JJMH made intellectual contributions to experimental design and discussion. All authors read and approved the final manuscript.