Altered distribution of the EphA4 kinase in hippocampal brain tissue of patients with Alzheimer’s disease correlates with pathology

Human brain tissue was obtained from the Netherlands Brain Bank (NBB, Amsterdam, The Netherlands). Prior to death, all donors gave written informed consent according to the Declaration of Helsinki for the use of their brain tissue and medical records for research purposes. This work was approved by the ethics committee of the NBB. Dementia status at death was determined on the basis of clinical information available during the last year of life. Neuropathological diagnosis was performed using histochemical analysis of formalin-fixed, paraffin-embedded tissue from different parts of the brain, including the frontal cortex (F2), temporal pole cortex, parietal cortex (superior and inferior lobule), occipital pole cortex and the hippocampus (essentially CA1 and entorhinal area of the parahippocampal gyrus) with routine stainings (hematoxilin and eosin, periodic acid Schiff-Luxol fast blue). Hippocampus and cortical areas were also stained with methenamine silver [[23]] and using the Bodian staining. Immunohistochemistry was performed using antibodies raised against hyperphosphorylated tau (AT8) and Abeta. Staging of AD pathology was evaluated according to modified assessment of Braak and Alafuzoff [[24]]. Cases with and without clinical neurological disease diagnosis were processed identically. Patients with co-morbidities like Parkinson’s disease or Lewy-body disease were excluded from the study. Age, sex, clinical diagnosis and Braak score for neurofibrillary tangles (NFTs) of all cases used in study are listed in Table 1. Post mortem delay of all cases was between 2 ½ and 9 hours. In total, 29 patients with AD and 19 controls were included (see Table 1), with an average age at death of 84 years (range 57 to 100 years old); of these, 22 (45.8%) were men.

Frozen hippocampal tissue slides of patients in all Braak stages (I-VI, n = 29) were cut and lysed with M-PER (pH 7.6, Thermo Scientific) containing protease and phosphatase inhibitors (Roche). After incubation for 30 min on ice and subsequent centrifugation (2 × 10 min, 4°C, 15682 × g), protein concentrations of the supernatants were determined with the standard Bradford Lowry Assay (Bio-Rad Protein Assay).

Protein lysates were re-suspended in sample buffer (Thermo Scientific) and heated for 5 minutes at 95°C. Proteins were resolved by SDS-PAGE using 8–16% gradient polyacrylamide gels (mini-PROTEAN®TGX™, Bio-Rad) in running buffer (25 mM Tris, 192 mM glycine, 0.1% SDS, pH 8.3, Bio-Rad) and electrophoretically transferred onto a nitrocellulose membrane (0.2 μm; Whatman, Protran™) in transfer buffer (25 mM Tris, 192 mM glycine, 20% methanol, Bio-Rad). Membranes were blocked for 1 hour in Tris-buffered saline (50 mM Tris pH 7.5, 0.15 M NaCl, 0.1% Tween-20) containing 5% BSA (Roche Diagnostics) and incubated over night with primary anti-EphA4 antibody (1:500). A mouse monoclonal antibody directed at the extracellular (c-terminal) domain of the Ephrin-A4 receptor (amino acids 379–472; BD Transduction Laboratories) and a rabbit polyclonal EphA4 antibody directed at the intracellular domain (amino acids 890–904; Abcam) were utilised. Incubation with a secondary antibody linked to horseradish peroxidase ([HRP]-anti-rabbit IgG or HRP-anti-mouse IgG, 1:1000, DAKO) for 1 hour followed. Immunoreactive bands were detected with an enhanced chemiluminescence reagent (ECL Plus, GE Healthcare). Actin (mouse anti-actin AC-40, Sigma) was used as a loading control. The intensity of the resulting protein bands was quantified using MacBiophotonics ImageJ (version 1.47 k). Data is expressed as relative signal intensities (EphA4/Actin) of the individual samples (Figure 1). Statistical analyses were performed by t-test analysis for independent samples using GraphPad Prism version 6.0 (San Diego, CA). Furthermore, whole protein extracts were analysed by Immuno-blot analysis with SDS-PAGE using precast stainfree gradient gels (Bio-Rad; Additional file 1: Figure S6). After UV activation of the gel, proteins were visualised and relative protein amounts were determined and used for normalization. Recombinant proteins EphA1 (75 kDa, aa 569–976; Pro Quinase), EphA4 (72 kDa, aa 586–986; Carna Biosciences) and EphB2 (73 kDa, aa 581–987; Carna Biosciences) were used to proof specificity of the rabbit polyclonal EphA4 antibody. The EphA4 mouse antibody binds extracellularly (aa 379–472) which is outside of the binding domain for recombinant EphA4, therefore only the polyclonal rabbit antibody is shown (Additional file 2: Figure S8).

For single immunohistochemical analysis, formalin fixed (4%, 24 hours) paraffin embedded 5 μm thick sequential sections were mounted on coated glass slides (Menzel Gläser Superfrost PLUS, Thermo Scientific, Braunschweig Germany), dried over night at 37°C and deparaffinised. In order to block the endogenous peroxidase, the sections were incubated in 0.3% (v/v) H₂O₂ in methanol (100%) for 30 min at RT. Sections were boiled in 10 mM sodium citrate buffer (pH 6.0) for 10 minutes for antigen retrieval. They were incubated over night with the primary antibody at room temperature. Mouse monoclonal anti-EphA4 [[25]–[27]] (1:200, BD Transduction Laboratories™); mouse monoclonal anti-phospho-tau [[28]] (AT8 1:200, Pierce Biotechnology, Rockford IL, USA); mouse monoclonal anti-Aβ (IC16 1:500, Prof. C. Korth, Heinrich Heine University Düsseldorf, Germany) and mouse monoclonal anti-Aβ 1–17 (VU17 [[29]] 1:1000, Dr. R. Veerhuis, VU University Medical Center, Amsterdam, NL) were used. As a second step, envision mouse/rabbit HRP (DAKO) was used. Sections were stained using the EnVision method (incubation with Dako REAL™ EnVison™ HRP rabbit/mouse antibody for 30 min). Subsequently, colour was developed using 3,3-diaminobenzidine (EV-DAB; DAKO) as chromogen. Sections were counterstained with hematoxylin and mounted using Quick-D mounting medium (Klinipath B.V., Duiven, The Netherlands) (Figures 2 and 3). Between all incubation steps, sections were washed extensively with phosphate buffered saline (PBS, pH 7.4), which also served as a negative control by omission of primary antibodies.

To determine the co-localization of EphA4 with amyloid beta and phosphorylated tau double immunohistochemistry was performed. Hippocampal brain slices were pre-incubated with serum-free protein blocking (SFPB, DAKO) for 10 min and subsequently incubated with mouse monoclonal anti-EphA4 [[25]–[27]] (1:200, IgG1, BD Transduction Laboratories™) for 1 hour. After washing with PBS, slices were incubated with EnVision solution (goat anti-rabbit HRP, undiluted, DAKO) for 30 minutes. Colour was developed using DAB as chromogen. Sections were treated with 10 mM sodium citrate buffer pH 6.0 heated at 180 Watt for 10 minutes. After pre-incubation with SFPB for 10 minutes, sections were incubated with anti-Aβ VU 1–17 antibody [[29]] (1:1000, IgG2a, VU medical center) or anti-pTau AT8 antibody (dilution 1:200, IgG1) for 1 hour. Sections were washed with PBS and incubated with either GαMIgG2a^HRP (dilution 1:100, Southern Biotech) for the detection of VU 1–17 or GαMIgG1^AF (dilution 1:100, Southern Biotech) for the detection of AT8 for 1 hour at RT. Colour was developed using Liquid Permanent Red (LPR, DAKO) as chromogen. Slices were counterstained with hematoxylin and mounted using Aquamount (BDH Laboratories Supplies). The Nuance™ spectral imaging system (CRi, Woburn, MA) was used for the analysis of double stained specimens. Spectral imaging data cubes were taken from 460–660 nm at 10 nm intervals and analyzed with the Nuance™ software. Spectral libraries of single-brown (DAB), single-red (LPR) and hematoxylin were obtained from control slides. The resulting library was applied to the double stained slides and the different reaction products were then spectrally unmixed into individual black-and-white images, representing the localization of each of the reaction products, and reverted to fluorescence-like images composed of pseudo-colours using the Nuance™ software (Figure 4).

Presence of EphA4 in hippocampal subregions CA1 and subiculum were analysed in sections immunostained with EphA4 (BD Transduction Laboratories) by an observer blinded to the clinical and pathological diagnosis. Microscopic fields (n = 4, magnification 100x) were analysed (Table 2). The number of EphA4 depositions was scored as follows: 0, none; 1, one to ten depositions; 2, eleven to twenty depositions; 3, more than twenty. The presence of Aβ plaques (plaque severity) was quantified in a similar manner (0, none; 1, occasional plaque; 2, several plaques scattered throughout the field; 3, abundant presence of plaques). The presence of pTau-tangles was scored: 0, none; 1, mild (occasional immunoreactivity); 2, moderate (scattered throughout the field); 3, severe (abundant presence of plaques). Additional file 3: Figure S7 shows examples for scoring of EphA4, Aβ and pTau immunoreactivity.

Pairwise correlation analyses for non-parametric ordinal data were conducted using IBM SPSS Statistics 22.0 (Figure 5A-D). Spearman’s correlation coefficients and Kendall’s tau coefficients were calculated. Correlations were considered significant in the 95% confidence interval when p < 0.05 (Table 3).

In order to investigate protein levels of EphA4 we analysed frozen hippocampal brain tissue lysates of AD patients (n = 18) and non-demented controls (n = 11) by Western Blotting. Using two different EphA4 antibodies i.e. a rabbit polyclonal antibody raised against the intracellular part of the EphA4 and a mouse monoclonal antibody to detect the extracellular domain of EphA4. The rabbit antibody detected a protein (fragment) of approx. 108 kDa whereas the mouse antibody showed a band that correlates to a molecular weight of 130 kDa (Figure 1 A,C). Actin was used as a loading control (molecular weight is ~ 45 kDa) and to determine relative signal intensities for EphA4 (Figure 1B,D). We observed no significant difference in the levels of the intracellular and extracellular domain of EphA4 between AD cases and controls (Figure 1). We analysed the levels of EphA4 at different stages of AD and found no significant differences (results not shown). Since EphA4 is an integral membrane protein we analysed whole tissue lysates (containing soluble and insoluble fraction). No difference in average levels was observed between AD and control hippocampus (see Additional file 1: Figure S6). In conclusion we observed no differences in EphA4 levels in hippocampal tissue at different stages of AD. Recombinant proteins EphA1, EphA4 and EphB2 were used to prove specificity of the rabbit polyclonal EphA4 antibody. The mouse anti-EphA4 antibody detects EphA4 at aa 379–472. This sequence is not present in recombinant EphA4 which represents a shorter fragment of EphA4. Therefore mouse anti-EphA4 antibody is not able to detect recombinant EphA4 (Additional file 2: Figure S8). No bands were observed for recombinant EphA1 and EphB2 whereas a band for recombinant EphA4 was detected at 72 kDa for all three concentrations (1:50, 1:100, 1:200). This suggests highly specific binding of the rabbit EphA4 antibody.

Since we observed no difference in the levels of EphA4 we wondered whether the localisation of EphA4 might change in AD versus non-demented controls. The mouse monoclonal antibody directed at the extracellular (c-terminal) domain of EphA4 was used for detection of the protein. In non-demented control cases (without AD pathology), a low intense immune-reactive signal of EphA4 was detected in the parenchyma. The signal was evenly distributed throughout the hippocampus (Figure 2A,C). In contrast, in AD cases EphA4 immuno-reactivity was observed in plaque-like structures (Figure 2B,D). Interestingly, EphA4 immuno staining of these plaque-like structures was observed more frequently in the higher Braak stages (V and VI), but some of the samples representing low stages (three in Braak I and 5 in Braak II) showed the same pattern. The EphA4-positive plaque-like structures showed considerable variations in shape and size (5–50 μm) and were detected in all subregions of the hippocampus.

Intense EphA4 immunostaining was observed in the cytoplasm of granular neurons in the dentate gyrus of AD but not in the dentate gyrus of non-demented control cases (Figure 2E,F). In non-demented controls, EphA4 immunoreactivity was found as a fine punctate line in the plexiform layers of the dentate gyrus (Figure 2G). In AD patients, EphA4-positivity was frequently decorating the dentate gyrus showing a brown band in the perforant path target zone of the outer molecular layer of the dentate gyrus, forming a frame around the granular layer (Figure 2F,H). This staining pattern was not observed in control cases. EphA4-positivity was not restricted to specific subareas of the hippocampus but was also found in and around neurons of the granular layer and in the stratum lacunosum. The presence of EphA4 immunoreactive plaque-like deposits in the CA regions of the hippocampus and entorhinal cortex in combination with the intense EphA4 immunostaining of the molecular layer was most prominent in demented individuals. Interestingly, EphA4 immunoreactive plaques were also observed in the hippocampi of non-demented controls (Braak II, see scoring Table 2) which points towards an early involvement of EphA4 in AD pathology.

The detection of EphA4 immunoreactivity plaque-like structures suggests a co-localization with neuritic plaques. In order to investigate the co-occurrence of EphA4 with neuritic plaques, adjacent brain tissue sections were stained for EphA4, pTau and Aβ (Figure 3). Consecutive brain slices were stained for Aβ, pTau and EphA4 and partial co-localization in the hippocampus of AD patients was revealed.

In order to corroborate the possible co-localization of EphA4 with either Aβ or pTau, double-immunostaining was performed. We found confirmation of partial co-localization of EphA4 with pTau and Aβ-immunoreactive plaques in AD patients (Figure 4A). EphA4-immunoreactivity was characteristically found at the ends of dystrophic neurites around central Aβ plaques (Figure 4B). EphA4 immunoreactivity is less intense compared to that of Aβ or pTau which could reflect differences in the antigen binding properties of the different antibodies. So EphA4 localization is clearly linked to the deposition of both Aβ and pTau.

In order to investigate whether the number of EphA4 depositions increases with Braak stage and correlates with the hallmarks of AD, immunoreactivity for EphA4 was analysed in a new second cohort representing all Braak stages (I-VI) (Table 2). Adjacent tissue sections were stained and scored for EphA4 depositions, Aβ plaques, pTau-positive plaques and pTau tangles (see Additional file 3: Figure S7 for scoring example).

To examine the relationship between EphA4 and the hallmarks of AD, Spearman’s correlation [r_s] and Kendall’s correlation [τ] analyses for non-parametric ordinal data were performed (Table 3). Spearman’s correlation coefficients are used for correlation analyses of non-parametric data sets. Kendall’s correlation coefficients are preferred for small sample sizes with a large number of tied ranks. The analyses revealed a significant correlation between EphA4 positive depositions and pTau-positive tangles (Figure 5A; r_s = .420, p < .05; τ = .393, p < .05) and between EphA4 and pTau-positive plaques (Figure 5B; r_s = .378, p < .05; τ = −.329, p < .05). No significant correlation was found between EphA4-load and Aβ plaques (Figure 5C; r_s = .282, ns; τ = .259, ns). However, the correlation between EphA4 deposits and Braak stage was evident (Figure 5D; r_s = .486, p < .05; τ = .421, p < .05). Table 3 summarizes the obtained correlation coefficients and p-values.