Analysis of the Relationship Between Metalloprotease-9 and Tau Protein in Alzheimer’s Disease

Brain tissue

Brain tissue samples were obtained from National Dementia BioBank, Mexico in accordance with the institutional bioethics’ guidelines. The pathological severity of AD samples was determined according to the protocols described by Basurto-Islas and colleagues [30]. In brief, using the immunoperoxidase technique, AD cases were processed and the average total densities of NFTs in the entorhinal cortex and hippocampal formation for each case were grouped according to their corresponding Braak stage (BST). The density of both tau-C3 and mAb 423-labeled NFTs was significantly correlated with the BST. For this study, cases classified as BST I were considered as early AD, BST IV as moderate AD and BST VI as late AD. After immunohistochemical analysis, three AD cases were selected from each pathologic category: early (AD1, AD4, and AD6); moderate (AD2, AD3, and AD5); and late (AD7, AD8, and AD9) (Table 1) to analyze the relevance of metalloproteases in the development of AD.

Additionally, all the brain samples used were diagnosed with AD. First, selected hippocampus slices, stained using the modified Bielschowsky technique [31, 32], Thiazine red dye (TR) [33, 34], and double-stained with 423/pThr231 and thiazine red/tauC3 antibodies. These were used for histological identification of pathological deposits of amyloid plaques and NFTs of AD.

Protein extraction

Hippocampal tissues (100 mg) were triturated and sonicated in homogenization solution (20 mM Tris, 1 mM EGTA, 1 mM EDTA, 10 mM sucrose % (w/v), pH 7.4), supplemented with protease inhibitors (200 mM dithiothretiol, 100 mM phenylmethylsulfonyl fluoride) at 4°C [35]. The homogenates were then centrifuged at 10,000×g for 10 min at 4°C and the total protein of the supernatant was quantified by Bradford protein assay.

Western blot analysis

Total protein extract (30 μg) was separated by SDS-PAGE using 10% separating gels and transferred to nitrocellulose membranes. Membranes were blocked with 5% non-fat dried milk in phosphate-buffered saline (PBS) pH 7.2 containing 0.1% Tween 20 and incubated overnight at 4°C with primary monoclonal antibodies against MMP-9 (1 : 1,000; Abcam) or MMP-2 (1 : 1,000; Abcam). Next, membranes were washed and incubated with secondary antibody conjugated with horseradish peroxidase (1 : 5,000; Invitrogen) for 2 h at 4°C. Finally, immunoreactive bands were visualized using ECL detection reagent. Autoradiograms were scanned and labeled bands quantified using the Image J software (NIH, USA). A protein extract of MCF-7 breast cancer cells, stimulated with phorbol 12, 13-dibutyrate (PDBu), was included as a control.

Gelatin-substrate zymography

MMP-9 and MMP-2 levels in AD brains and controls were monitored by gelatin zymography assay. Equal volumes of non-heated total protein or 30 μg total protein extract were mixed with sample buffer (2.5% SDS, 1% sucrose, 4 μg/mL phenol red) without reducing agents, and resolved using 8% polyacrylamide gels copolymerized with gelatin (1 mg/ml) [36]. Gels were washed three times with 2.5% Triton X-100, and then incubated in assay buffer (50 mM Tris-HCl, pH 7.4, 5 mM CaCl₂) at 37°C for 24 h. Gels were fixed and stained with 0.25 % Coomassie Brilliant Blue G-250 in 10% acetic acid and 30% methanol. Proteolytic activity was detected as clear bands against background stain of undigested gelatin. The mobility of the gelatinolytic bands corresponded to the molecular mass for MMP-9 and MMP-2 and were measured by software analysis. Conditioned medium from MCF-7 cells stimulated with PDBu or ethanol were also included.

Immunofluorescence and confocal microscopy

Free-floating sliding brain tissue sections (50-μm thick) were processed using a sliding microtome (Leica TS). Antibody epitopes were retrieved with a citrate buffer (0.1 M citric acid, 0.1 M sodium citrate, pH 6.0) at 100°C during 10 min. Membrane permeabilization was achieved by incubation in ethanol-water (70%, v/v) for 5 min. The auto fluorescence of tissue was eliminated by incubation in Sudan Black B (SBB) solution (0.1% SBB in 70% ethanol) for 10 min.

Sections were then incubated with a solution of 0.2% IgG-free albumin (Sigma Chemical Co) for 1 h at room temperature and incubated overnight at 4°C with the primary antibodies against tau (mAb 423 and tau-C3) combined with anti-MMP-9. Additionally, anti-cathepsin D (Calbiochem) was used to identify neurons in early stages of degeneration, and TR dye to identify insoluble PHF-tau protein [34]. Sections were then incubated for 1 h at 37°C with a FITC-tagged goat-anti-rabbit IgG secondary antibody (for MMP-9) and TRITC-tagged goat-anti-mouse IgGγ secondary antibody (for 423 and tau-C3) (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA). Samples were mounted with Vectashield (Vector Laboratories) and examined using a Leica TCS SP2 confocal microscope. Observations were performed in 14 planes from the top to the bottom of each sample, and the distance between scanning planes were at 0.8–1.0 μm intervals for two or three channels throughout the z-axis of the sample. The resultant stack of images was projected and analyzed onto the two-dimensional plane using a pseudocolor display of green (FITC) and red (TRITC). Fluorochromes were excited at 488 nm (for FITC) and 540 nm (for TRITC).

Generation of theoretical 3D models

In order to obtained the 3D structure of tau and MMP9 proteins the sequences NP_005901 and NP_004985.2 respectively (retrieved from NCBI database) were submitted to I-TASSER server, where the starting model was selected according to the highest C-score http://zhanglab.ccmb.med.umich.edu/I-TASSER/) [37]. I-TASSER employed the 3D crystal structures from Protein Data Bank (PDB) as templates (http://www.rcsb.org).

The templates used to build the tau 3D structure on I-TASSER were 2MZ9, the structure of tau (267-312) bound to microtubule; 2NBI. The structure of the PSCD-region of the cell wall protein pleuralin-1; 1W0R, a solution structure of dimeric form of propending by X-ray solution scattering and analytical ultracentrifugation; 1DUR, the X-ray crystal structure of full-length type II human plasminogen; and 1ZIW, the human toll-like receptor 3 extracellular domain structure. Whereas for MMP-9 we employed as template the crystal structure 1L6J from PDB database which contained 425 residues from MMP-9, obtained the whole protein (707 residues) in I-TASSER. The most energetically stable 3D models were selected and for tau 3D model their quality evaluated by Ramachandran plots on RAMPAGE server (http://mordred.bioc.cam.ac.uk/ rapper/rampage.php). To determine whether the Molecular Dynamics simulation (MDS) should be run in a lipid or an aqueous environment, we submitted the 3D conformations in the OPM database [38]. Based upon the results, MDS for both proteins was undertaken in an aqueous environment.

The 3D structure prediction for tau protein shows a major region lacking secondary structure. For intrinsically disordered proteins, the disordered region was acquired using the DISOPRED3.1 server (http://bioinf.cs.ucl.ac.uk/psipred/).

Molecular dynamics simulation

MDS were performed with NAMD 2.8 software [39] through the use of GPU-CUDA with video cards graphics NVIDIA Tesla C2070/Tesla C2075. The CHARMM22 force fields [40] were used to create the topologies for proteins; whereas the waters molecules and the hydrogen atoms were added using the psfgen on the VMD software [41], and then water molecules and ions were randomly added to neutralize the system. We added 142511 water molecules and 2 chloride ions for tau protein and 26650 waters and 12 sodium ions for MMP-9. The systems were minimized for 1000 steps followed by equilibration under constant temperature and pressure (NPT) conditions for 1 ns with protein and lipid atoms restrained. Then, 20 ns-long MDS were run, considering these proteins as soluble, without position restraints under periodic boundary conditions and using an NPT ensemble at 300 K. The electrostatic interactions were calculated via the PME method [42] and a 9 Å cut off was used for the van der Waals interactions. The temperature was maintained employing Langevin dynamics while pressure was kept constant at one bar using a Nosé-Hoover-Langevin piston [43]. The time step was set to 2.0 fs and the coordinates were saved for analyses every one ps.

Trajectory and docking analysis

Carma software [44] was employed to get the root mean square deviation (RMSD), the root mean square fluctuation (RMSF), the radius of gyration (Rg), and the snapshots used for docking analysis. Molecular graphics were performed in SigmaPlot 12.0 and protein visualization was performed using VMD. The protein-protein docking studies were developed in a ClusPro server.

Statistical analysis

The data for western blot and zymography assays are presented as mean±standard error (SEM) where n = 4 samples for each AD case for the results. For comparison between cases, we used t-test, with the significance level of 5% (p value <0.05).

Histopathological characterization of cases with AD

Among the methods commonly used to demonstrate the abundance of lesions in the hippocampus and temporal and frontal cortex, is the impregnation with silver salts (Bielschowky’s method), and the fluorescent marker called thioflavin-S, which can be detected by exciting the sample at 488 nm. We have used the red fluorescent dye or TR, which has an affinity for the same structures as thioflavin-S.

The silver impregnation technique has a high affinity for fibrillar structures. In Fig. 1, a large number of fibrillar plaques and NFTs (arrows) are observed at low magnification.

In Fig. 1A, neurofibrillary plaques and tangles are seen by the Bielschowky’s methods. Fibrillar state and beta folded conformation is shown with TR dye (Fig. 1B).

OPEN IN VIEWER

Fig. 1

Characterization of Alzheimer’s diseases cases. Fixed hippocampus sections from AD cases stained with Bielschowsky (A), thiazine red (B), double-stained with 423/pThr231 (C) and thiazine red/tauC3 (D). Neurofibrillary tangles (rows), neuritic plaques (Aβ), and expression of tau truncated at Glu-391, as detected with tau-423 antibody, were corroborated.

NFTs are exclusively observed in the entorhinal cortex II (Fig. 1C) with antibodies directed against the tau protein truncated at Glu-391 (green channel) and phosphorylated at Thr-231 (pT231).

In Fig. 1D, double staining is observed with the TauC-3, an antibody which recognizes truncation in Asp421, and the TR dye. Amyloid plaques (red channel) and NFTs (blue channel) are observed. NFTs colocate mostly with the TR dye.

Metalloproteinases 2 and 9 are elevated in brains with late-stage AD pathology

To confirm the presence of tau and metalloproteinases in samples of different pathological severity, we analyzed the protein levels for tau, MMP-9, and MMP-2 using specific antibodies against these proteins. Extracts were separated by SDS-PAGE and the levels of tau, MMP-2, and MMP-9 were analyzed by western blotting. AD samples revealed bands corresponding to tau proteins truncated at Glu-391, with molecular weight from 15 to 75 kDa (Fig. 2A). Control samples did not show immunoreactivity with this antibody (Fig. 2A).

OPEN IN VIEWER

Fig. 2

Increased expression of MMP-9 in AD brains. A) Characterization of proteins reactive with tau-423, a generic marker of AD, in hippocampal samples. B) Western blot analysis with 30 μg of protein extract from hippocampus of AD patients having early, moderate, and late stage severity revealed with anti-MMP-9 and MMP-2. C, D) Zymographic analysis of MMP-9 and MMP-2 in the same series of brains; the assay was carried out with either constant concentration of protein (30 μg) (C) or constant volume of supernatant (10 μl) (D). C, control samples; AD, samples from individuals with AD; PDB, supernatant of cancer cells stimulated with phorbol-12,13-dibutyrate (PDBu) or ethanol (EtOH). Anti-actin was used as a loading control. The relative intensities of the bands detected by western blot were analyzed by densitometry. E, F) MMPs were normalized to actin and the fold change in AD samples over control was plotted (n = 4; *p<0.05).

The analysis of matrix metalloproteases shows that the bands were more intense in brain tissue from cases with AD compared with samples from healthy subjects (C1, C2, C3, and C4) (Fig. 2B). MMP-2 bands were prominent in late-stage AD cases (Fig. 2B, cases AD8 and AD9). The intensity for MMP-2 in early AD cases (AD1, AD4, and AD6) or those of moderate severity (AD2, AD3, and AD5) did not differ significantly from those observed in the healthy controls (Fig. 2B, E).

On the other hand, the bands recognized by anti-MMP-9 increased with increasing severity of AD (Fig. 2B). Samples with early AD (cases AD1, AD4, and AD6) showed MMP-9 levels similar to the controls, but the levels were increased significantly in both moderate AD (Fig. 2B, cases AD2, AD3, and AD5) and late AD (Fig. 2B, AD7, AD8, and AD9 and Fig. 2F).

Samples with early AD and controls had similar low levels of MMP-9, with the exception that control case C3 had significant MMP-9. Densitometric analysis showed that the MMP-9 expression in moderate and late AD was significantly increased by up to nearly 5-fold (Fig. 2F). Actin levels were similar for all samples analyzed (Fig. 2B). The specificity of the MMP-2 and MMP-9 antibodies was verified against protein extracts of MCF-7 cells stimulated with PDB (Fig. 2B).

In summary, the amount of MMP-9 is low in cases with early severity of AD, increases slightly in brains with moderate severity and is even greater in those with high severity; i.e., the level of expression of MMP-9 increases as the pathology progresses.

Enzymatic activity of MMP-9 increased in cases with moderate and late-stage AD pathology

Having determined the presence of MMP-9 in brain tissue and detected increased amounts in cases with moderate and late AD, the presence of the MMP-2 and MMP-9 activities was confirmed in the same cases using in situ gelatinase assays. Gelatin-substrate zymography revealed that MMP-9 and MMP-2 levels increased with the pathological severity of AD. The test was performed under two different conditions: constant mass and constant volume. In both experimental conditions, the same results were obtained. As found with the western blot assays, MMP-9 activity was greater than MMP-2 (Fig. 2C, D). Cases with moderate severity (Fig. 2C, E, cases AD2, AD3, and AD5) and late-stage AD (Fig. 2B, cases AD7, AD8, and AD9) showed abundant MMP-9.

MMP-9 and PHFs-tau co-localized at early stage of NFT formation

To determine whether MMP-9 expression was related to the progression of NFTs and if there was an interaction with tau, we performed double-immunolabelling confocal microscopy with antibodies directed against cleaved tau (tau-C3) and MMP-9. TR dye was used to identify fibrillary forms of tau aggregates labelled with MMP-9 (Fig. 3A). Extracellular NFTs (E-NFTs) were observed as typical pyramidal cells with dense accumulations of PHF-tau distributed throughout the axon, dendrites and cell body (Fig. 3A, red channel). Additionally, the anti-cathepsin D was used to identify cases with an elevated number of neurons initial stages of AD degeneration (Fig. 3A, green channel, arrows). Alterations in cell morphology and increase in proteins of the endosome-lysosome pathway is considered one of the first events caused by the presence of AD [45, 46]. In contrast, intracellular NFTs (I-NFTs) (Fig. 3B), showed tau-C3 immunoreactive products from the cleavage of tau present in the axon and terminal dendrites, that were absent from the cytoplasm (Fig. 3B, red channel, arrows). Such neurons with early neuropathological stages showed MMP-9 immunoreactivity present mainly throughout the cytoplasm (Fig. 3B, green channel). In contrast, co-localization of the two markers was mainly associated with dot-like structures located in axon, terminal dendrites and cytoplasm (Fig. 3B, arrowhead).

OPEN IN VIEWER

Fig. 3

Increased MMP-9 in I-NFTs and co-localization with PHF-tau. Double immunolabelling of tangles and plaques was observed for MMP-9 and tau protein. A) Intracellular accumulation of fibrillary tau displayed an intense fluorescence with thiazine red dye (red channel). Also shown some neurons in the early stages of degeneration detected using anti-cathepsin D (green channel, arrows). B) Double immunolabelling with MMP-9 (green channel) and tau-C3 (red channel, arrows) identified I-NFT and dystrophic neurites. In both structures, there is colocalization between MMP-9 and tau protein (arrowhead). C) Extracellular NFTs, with abundant tau protein in filaments (thiazine red, red channel), contain very low levels of MMP-9 (green channel, arrows). D) NFTs detected with TR (red channel) surrounded by other cells such as astrocytes or microglia have high levels of MMP-9 (green channel, arrow).

We also performed MMP-9 immunolabeling (green channel) and TR counterstaining in E-NFTs (red channel). As illustrated in Fig. 3C, these E-NFTs showed a few cytoplasmic clusters or small aggregates of MMP-9 (Fig. 3C, green channel, arrow). In contrast, elevated expression of PHF-tau was observed in the cell body for all NFT-bearing neurons (Fig. 3C, red channel). No colocalization between MMP-9 and PHF-tau was observed in these neurons.

In the AD samples studied (early, moderate, and late stages), we observed NFTs surrounded with other types of cells, probably microglia or astrocytes, cells that have abundant cytoplasmic MMP-9 immunoreactivity (Fig. 3D).

Generation of a theoretical model of the Tau-441 and MMP-9 protein 3D structures

The 3D models of microtubule-associated protein tau isoform 2 showed an extended protein (Fig. 4A, B). The model retrieved from the I-TASSER server presented thirteen β-strands (Table 2), with the rest of the residues in random coil structures. This model was submitted to Ramachandran plot and the results showed 68.8% of the amino acids residues on the favored region, 23.2% inside the allowed region and 8% on the outlier region (Fig. 5A). The structural analysis after 20 ns suggest at RMSD the looking for the equilibrium of tau protein, a finding in agreement with tau being an intrinsically disorder protein (Fig. 6A). After MDS at 20 ns by NAMD software, the 3D structure exhibited twelve β-strands (Fig. 4B). The Ramachandran plot after 20 ns of MDS showed 73.8% of the amino acid residues on the favored region, 22.3% inside the allowed region, and only 4% on the outlier region (Fig. 5B). These findings indicate that the angles of certain amino acids were refined through the trajectory of MDS. The RGB analysis (Fig. 6B) suggest the search for system equilibrium. In order to understand the behavior of the RMSD, we looked for the presence of disordered regions using the DisProt server, and found that most of the amino acids correspond to an intrinsically disordered protein (Fig. 6C). These finding are in agreement with the RMSD. The RMSF for tau protein measures the movements of the alpha carbon of the backbone of the protein and, in Fig. 6D, the results show extensive movement of the whole protein, perhaps because of the disordered regions of this protein.

OPEN IN VIEWER

Fig. 4

3D homology models of MMP-9 and tau proteins. A) 3D model of tau retrieved from I-TASSER server. B) 3D model of tau after 20 ns of MDS using NAMD software. C) 3D model of MMP-9 protein obtained from I-TASSER server. D) 3D model of MMP-9 after 20 ns of MDS. COOH, C-terminus. N, N-terminus. Axes: x, red; y, green; z, blue.

OPEN IN VIEWER

Fig. 5

Ramachandran plots of tau. Ramachandran plots of 3D model proteins before and after 20 ns of MDS. A) Ramachandran plot of tau from I-TASSER 3D model. B) Ramachandran plot of tau after 20 ns MDS. Red dots: residues in outlier regions.

OPEN IN VIEWER

Fig. 6

Structural analysis of tau. The trajectory of MDS was performed using Carma software and the following parameters were evaluated: A) normalized RMSD, B) Rg, C) disordered state and binding, and D) normalized RMSF.

For MMP-9, we used as templates the crystal structure of MMP-9 1L6J:A, which contains 425 amino acids residues; the rest of the chain was built by I-TASSER (Fig. 4C, D). The 3D model showed 12 α-helices and 32 β-strands (Tables 3 and 4). Whereas, the RMSD for MMP-9 shows this protein reach the equilibrium at 10 ns of MDS (Fig. 7A). The RG shows a compaction of the protein (Fig. 7B); this compaction is visible in Fig. 4C and 4D. The RG values for MMP-9 indicate that this protein has an expansion at the first ns of the trajectory, and it remains without changes after 10 ns (Fig. 7B), which is in concordance with the RMDS analysis, and suggests equilibrium of the system. The RMSF values for MMP-9 (Fig. 7C) suggest the region near to the 415–522 as the major movements.

Docking analysis predicts a complementary binding between MMP-9 and Tau-441

MMP-9 and tau protein reached by MDS during 20 ns was used to carry out protein-protein docking using the ClusPro server. Docking analysis between MMP-9 and tau showed interactions mediated by 6 saline bonds and surrounded by 50 hydrogen bonds (Fig. 8 and Tables 5 and 6). MMP-9 binds to tau mainly through a region located at Arg95-Lys 638 residues. In this complex, the distances between interacting residues were <2.61 Å with a ΔG of –1182.9 for MMP-9-tau complex (Fig. 8, Tables 5 and 6). Six saline bridges bound is formed between residues tau with MMP-9 with an interaction distance of 2.61 Å (Table 6). Interestingly, this molecular complex is stabilized by 50 hydrogen bonds (Table 5). The hydrogen interactions are very close to the six salt bridges: three of the residues of the tau protein forming the salt bridges with the residues Arg95 from MMP-9 to form hydrogen bonds. Interestingly the residues of MMP-9 participating on the interaction with tau protein, exhibited high fluctuation values on the RMSF plot (Fig. 7C). For example, the residue Pro196 presented a fluctuation of 3.97 Å, Gly197 with 3.37 Å, pro110 with 2.73 Å, pro294 with 2.36 Å, Tyr179 with 2.23 Å, Phe 107 with 2.02 Å, Arg95 with 2.29 Å, Arg 96 with 2.00 Å, Lys 559 with 2.02 Å, Lys 638 with 4.20 Å, and Arg 621 with 3.10 Å. This fluctuation could suggest that these regions need to be attached to others protein to confer the stabilization for this movement.

OPEN IN VIEWER

Fig. 7

Structural analysis of MMP-9. The trajectory of MDS was performed using Carma software and the following parameters were evaluated: A) RMSD, B) Rg, and C) RMSF.

OPEN IN VIEWER

Fig. 8

Heterodimer complex of tau and MMP-9. Protein-protein docking between tau and MMP-9. A) Complementary binding between tau and MMP-9. B) Magnification of the interacting residues in the association of tau and MMP-9. Tau is shown in green, MMP-9 in red. ΔG, binding energy. Amino acids in black letters belong to tau and in red letters to MMP-9. COOH, C-terminus. N, N-terminus. Axes: x, red; y, green; z, blue.

In summary, molecular docking analysis indicates that MMP-9 can bind to the C-terminal end of tau protein by means of 6 salt bridge and 50 hydrogen bonds. Interestingly, the binding zone of tau coincides with those regions predicted to be binding zones. The DisProt server predicted both the disordered region and the possible binding zone of tau protein, a region localized to the C-terminus between the residues Ile-260 and Leu-441 (Table 5). The amino acids of MMP-9 participating in the interaction correspond with the start of the active enzyme (Arg95) (Tables 5 and 6), which might suggest that the interaction between tau and MMP-9 are initiated when MMP-9 is active.