MT5-MMP is a new pro-amyloidogenic proteinase that promotes amyloid pathology and cognitive decline in a transgenic mouse model of Alzheimer’s disease
- 2.7k Downloads
Membrane-type 5-matrix metalloproteinase (MT5-MMP) is a proteinase mainly expressed in the nervous system with emerging roles in brain pathophysiology. The implication of MT5-MMP in Alzheimer’s disease (AD), notably its interplay with the amyloidogenic process, remains elusive. Accordingly, we crossed the genetically engineered 5xFAD mouse model of AD with MT5-MMP-deficient mice and examined the impact of MT5-MMP deficiency in bigenic 5xFAD/MT5-MMP−/− mice. At early stages (4 months) of the pathology, the levels of amyloid beta peptide (Aβ) and its amyloid precursor protein (APP) C-terminal fragment C99 were largely reduced in the cortex and hippocampus of 5xFAD/MT5-MMP−/−, compared to 5xFAD mice. Reduced amyloidosis in bigenic mice was concomitant with decreased glial reactivity and interleukin-1β (IL-1β) levels, and the preservation of long-term potentiation (LTP) and spatial learning, without changes in the activity of α-, β- and γ-secretases. The positive impact of MT5-MMP deficiency was still noticeable at 16 months of age, as illustrated by reduced amyloid burden and gliosis, and a better preservation of the cortical neuronal network and synaptophysin levels in bigenic mice. MT5-MMP expressed in HEKswe cells colocalized and co-immunoprecipitated with APP and significantly increased the levels of Aβ and C99. MT5-MMP also promoted the release of a soluble APP fragment of 95 kDa (sAPP95) in HEKswe cells. sAPP95 levels were significantly reduced in brain homogenates of 5xFAD/MT5-MMP−/− mice, supporting altogether the idea that MT5-MMP influences APP processing. MT5-MMP emerges as a new pro-amyloidogenic regulator of APP metabolism, whose deficiency alleviates amyloid pathology, neuroinflammation and cognitive decline.
KeywordsMMP-24 Synaptotoxicity Neuroprotection Knockout mouse Neurodegenerative disease
Alzheimer’s disease (AD) is the most common neurodegenerative disorder. No cure or prevention exists to date. Proteolytic control of amyloid precursor protein (APP) is central to AD pathogenesis. The sequential cleavage of APP by β-secretase (β-site APP-cleaving enzyme, BACE) and γ-secretase complex releases the amyloid beta peptide (Aβ), whose accumulation is paramount in AD . The alternative α-secretase cleavage of APP by a disintegrin and metalloproteinase-10 (ADAM-10) releases an N-terminal soluble APP fragment (sAPPα), thereby preventing Aβ formation . β- and γ-secretase inhibitors prevent Aβ accumulation and AD pathogenesis in animal models, but they also interfere with the physiological processing of other substrates [3, 4]. It is therefore crucial to identify new targets to restrain APP metabolism and Aβ production while preventing side effects.
Matrix metalloproteinases (MMPs) have pivotal roles in neuroinflammation, neuroplasticity and neural cell death, but their implication in AD remains elusive [5, 6]. MMP-2 and MMP-9 have putative Aβ-degrading activities [1, 5], although MMP-9 overexpression also upregulates sAPPα levels and decreases Aβ oligomers in 5xFAD mice [7, 8], suggesting that overall MMPs could convey beneficial effects in AD. This general assumption is challenged by recent data indicating that membrane-type 1-MMP (MT1-MMP) promotes the accumulation of Aβ and its neurotoxic precursor, the APP C-terminal fragment of 99 amino acids (C99) . Interestingly, MT1-MMP and other members of the MT-MMP subfamily such as MT3- and MT5-MMP can process APP in heterologous cell systems and release soluble APP N-terminal fragments, which are different from those generated by α- and β-secretases [10, 11]. MT5-MMP (also known as MMP-24) stands out as the only MMP essentially expressed in the nervous system [12, 13], whose functional diversity includes the modulation of post-lesion neuronal plasticity , neural stem cell proliferation , or neuro-immune interactions between neurons and mast cells . MT5-MMP has been detected in brain amyloid plaques of Alzheimer patients  and suggested to cooperate with γ-secretase to synergistically hamper synaptic function in neuronal cultures . Collectively, the alleged interactions of MT5-MMP with key components of the amyloidogenic pathway provide strong rationale to question the possible implication of this MMP in AD pathogenesis. Accordingly, we crossed the APP/PS1 transgenic 5xFAD mouse model of AD (Tg thereafter)  with mice deficient in MT5-MMP (MT5−/−)  to generate a new bigenic TgMT5−/− strain and evaluate the impact of MT5-MMP deficiency in this AD mouse model. Compared to Tg mice, TgMT5−/− mice exhibited at early stages of the pathology strong and durable reductions of Aβ load and neuroinflammation along with improved hippocampal LTP and spatial memory, while the activity of the main secretases remained unchanged. Combined in cellulo and in vivo work supported the ability of MT5-MMP to interact with APP and to promote C99 and Aβ accumulation as well as the alternative processing of APP resulting in sAPP95 release. MT5-MMP emerges as proteinase with pro-amyloidogenic features that could be targeted to reduce amyloidosis/neuroinflammation and their associated functional deficits.
Materials and methods
5xFAD transgenic mice (Tg) were purchased from the Jackson Laboratory (Bar Harbor, ME, USA) . They carry three mutations in the human APP gene and two mutations in the human presenilin 1 (PSEN1) gene; in the APP gene: Swedish (K670N/M671L), Florida (I716V) and London (V717I); in the PSEN1 gene: M146L and L286V. All transgenes are under transcriptional control of neuron-specific mouse Thy1 promoter. Tg mice (B6/SJL genetic background) were first derived in a C57BL6 genetic background. F1 male mice resulting from crossing Tg B6/SJL and C57Bl6 mice were then crossed with C57Bl6 females and the male offspring backcrossed eight times with C57Bl6 females. The resulting C57Bl6 Tg mice were then crossed with C57BL6 homozygous transgenic MT5-MMP−/− mice  to obtain 5xFAD/MT5-MMP+/− and then with MT5-MMP−/− mice to generate 5xFAD/MT5-MMP−/− bigenic mice (TgMT5−/−) in a C57BL6 background. Genotyping was performed by PCR analysis of tail DNA . All experimental procedures were conducted only on male mice in accordance with the National and European regulations (EU directive No. 2010/63) and in agreement with the authorization for animal experimentation attributed to the laboratory by the Prefecture des Bouches du Rhone (permit number: D 13 055 08). All efforts were made to minimize animal suffering and to reduce the number of mice used.
Immunohistochemistry, image analysis and quantifications
Male mice were deeply anesthetized with sodium pentobarbital and transcardially perfused with cold saline (NaCl 0.9 %), followed by 4 % paraformaldehyde in 0.1 M phosphate buffer, pH 7.4 (PFA). The brains were then post-fixed 24 h in the same fixative and stored at 4 °C in phosphate buffer saline (PBS), pH 7.4. Free-floating coronal sections (30 μm thick) were obtained using a vibratome (HM 650V, Thermo Scientific, MA, USA) and stored at −20 °C in a cryoprotectant solution (40 % phosphate buffer 0.5 M, pH 7.4, 30 % ethylene glycol and 30 % glycerol). Sections were pre-incubated in a blocking PBS solution containing 3 % BSA and 0.1 % Triton X-100, followed by overnight incubation at 4 °C with polyclonal anti-GFAP (1/300, Dako France, Trappes, France), monoclonal anti-GFAP (Millipore, Molsheim, France), polyclonal anti-Iba1 (1/200, Wako, Sobodia, Montbonnot-Saint-Martin, France), monoclonal anti-Aβ 6E10 (1/300, Covance, Eurogentec, Angers, France), polyclonal anti-IL-1β (1/200, Peprotech, Neuilly-sur-Seine, France), monoclonal anti-MAP-2 (1/500, Abcam, Paris, France) and monoclonal anti-synaptophysin (1/500, Millipore), and then the corresponding secondary Alexa Fluor® 488 and 568-coupled antibodies (1/800, Life Technologies, Saint Aubin, France). Nuclei were stained with 0.5 µg/ml DNA intercalant Hoechst #33258 (Life Technologies). Sections were mounted using Prolong Gold Antifading reagent (Life Technologies). Omission of primary antibodies was used as negative controls of immunostaining. Tissues were observed under LSM 700 confocal or Axiovert inverted microscopes (Zeiss, Jena, Germany). Images were analyzed using the Axiovision (Zeiss), Photoshop and ImageJ (NIH) softwares.
The number of plaques was blindly scored in two brain sections per animal by three investigators at bregma −1.9 and −2.9 and normalized by the surface of the brain region analyzed. Densitometric analysis of immunostained sections was performed using ImageJ software and binarization methods, applying the same threshold of immunofluorescence across experimental groups. Brain areas of interest were outlined and the percent of immunoreactive pixels scored. The number of glial cells in the vicinity of plaques was determined over a 0.01 mm2 area, with the plaque placed at the center of the square.
Thioflavin T staining
After 1 h in a blocking PBS solution containing 3 % BSA and 0.1 % Triton X-100, free-floating sections were rinsed with PBS and incubated with Hoechst #33258 for 20 min, washed in PBS and then incubated for 15 min under agitation with 10 µg/ml thioflavin T (ThT, Sigma-Aldrich, Saint-Quentin Fallavier, France). Sections were incubated in 70° EtOH for 5 min to neutralize ThT and then washed to remove EtOH. Sections were protected from light and kept at room temperature during the entire procedure and then mounted using Prolong Gold Antifade reagent.
Male mice were deeply anesthetized with sodium pentobarbital and transcardially perfused with cold saline (NaCl 0.9 %). The neocortex and hippocampus were microdissected and snap frozen for biochemical assays. Samples were homogenized in 25 % w/v of 50 mM Tris–HCl buffer, pH 7.5 containing 150 mM NaCl, 2 mM EDTA, 1 % Triton X-100, 0.05 % SDS and a proteinase inhibitor cocktail (Millipore) and centrifuged at 10,000×g for 10 min at 4 °C. This “soluble fraction” contained cytosolic proteins and proteins easy to solubilize. Cell pellets were resuspended in 25 % w/v of 50 mM Tris–HCl buffer, pH 7.5 containing 2 % SDS, then sonicated and centrifuged at 10,000×g for 10 min at 4 °C to obtain an “insoluble fraction” containing more insoluble proteins, including an important part of membrane proteins . Protein concentrations were determined using a Bio-Rad DC™ protein assay kit (Bio-Rad, Marnes-La-Coquette, France) and 50 μg of protein was run on 10–15 % SDS-PAGE gels and transferred onto nitrocellulose membranes (Amersham Bioscience, Velizy-Villacoublay, France) . After blocking, membranes were probed with the antibodies, monoclonal anti-APP N-terminal (22C11, 1/500, Millipore), monoclonal anti-Aβ (6E10, 1/200), polyclonal anti-APP C-terminal (APP-CTF, 1/2000, Sigma-Aldrich), polyclonal anti-MMP-2 (1/500, Millipore), polyclonal anti-MMP-9 (1/500, Millipore), monoclonal anti-synaptophysin (1/1000), monoclonal anti-N-cadherin C-terminal (1/1000, BD Biosciences, Le Pont de Claix, France), monoclonal anti-GAPDH (1/500, Millipore) and monoclonal anti-actin (1/5000, Sigma-Aldrich), and then incubated with horseradish peroxidase-conjugated secondary IgG antibodies (Jackson Immunoresearch, West Grove, PA, USA). MT5-MMP deficiency was confirmed by western blot in the soluble fraction of the hippocampus and cerebellum from Tg and TgMT5−/− mice using a specific rabbit polyclonal MT5-MMP antibody (1/500) developed in our laboratory (Online Resource 1). Immunoblot signals were visualized using the ECL chemiluminescence kit (GE Healthcare, Dutscher, Brumath, France) and quantified using the ImageJ software.
The secreted forms of human APP, hsAPPα and hsAPPβ were analyzed in soluble diethylamine (DEA)-extracted fractions generated according to previously reported protocols [21, 22], using monoclonal anti-hAPPs-α, clone 14D6 and anti-hAPPs-β swe, clone 192swe (kindly provided by Dale Schenk), respectively. Polyclonal anti-calnexin (Enzo Life Sciences Inc., Farmingdale, NY, USA) was used for sample normalization. Blots were developed using appropriate horseradish peroxidase-conjugated secondary IgG antibodies (anti-mouse/anti-rabbit, Promega Corp., Madison, WI, USA; anti-rat, Santa Cruz Biotechnology Inc., Dallas, TX, USA) and the ECL chemiluminescence system (GE Healthcare). Quantification of sAPPα and sAPPβ bands was carried out using Fuji Las-4000 software (Fuji Film inc., Minato, Tokyo, Japan).
Levels of PS1, Aph1, Pen2 and nicastrin were analyzed in solubilized cell membranes as reported previously . Briefly, intact cell pellets were resuspended in 10 mM Tris–HCl buffer, pH 7.5 with proteinases inhibitors (Sigma-Aldrich), subjected to repeated passages through a 25 G needle and first centrifuged at 800×g for 10 min at 4 °C. The resulting supernatants were subjected to an additional 20,000×g centrifugation for 1 h at 4 °C. Pellets containing the membranes were then resuspended in solubilization buffer, 150 mM sodium citrate, pH 6.4 with 3-[(3-cholamidopropyl) dimethylammonio]-2-hydroxy-1-propanesulfonate [CHAPSO, 1 % (v/v)] and proteinase inhibitors (Millipore). The solubilized membranes were diluted in solubilization buffer (1 mg/ml). The denatured samples were subjected to western blot using a polyclonal anti-PS1 N-terminal (1/1000, provided by Paul Fraser), polyclonal anti-Aph1 (1/1000), polyclonal anti-Pen2 (1/1000, Covance) and polyclonal anti-nicastrin (1/1000, Sigma-Aldrich) and normalized with tubulin using monoclonal anti-tubulin (1/5000, Sigma-Aldrich).
ELISA assay for Aβ and IL-1β quantifications
For detection of soluble Aβ38, Aβ40 and Aβ42 species in the DEA fractions, we used the MSD Aβ Triplex sandwich immunoassay (MesoScale Discovery, Rockville, MD, USA) . MSD C-terminal-specific antibodies were pre-spotted into each well. For detection, a ruthenylated 4G8 antibody was used. The concentrations of Aβ peptides were normalized to protein content and calculated using the MSD Discovery Workbench software.
For detection of IL-1β in the soluble fraction, we used the murine IL-1β ELISA Development Kit (PeproTech, Neuilly-sur-Seine, France), loading 5 μg of protein per well, according to the manufacturer’s recommendations.
The MT5-MMP cDNA was amplified by PCR from C57Bl6 mouse cerebellum and cloned using protocols previously described for other MMP cDNAs [24, 25, 26]. The following primers were used: MT5For ATA TAT GAA TTC GGA TGC CGA GGA GCC GGG GAG GCC GCG CTG and MT5Rev ATA TAT GTC GAC AGT ACC CAC TCC TGG ACC GGC CGC TTA TAG TAG and cloned into pEGFP-N1 (Clontech, Saint-Germain-en-Laye, France). The intracellular domain (ic), corresponding to the last 22 amino acids, was subcloned after GFP encoding sequence using the following primers: icFor TAA TTA CTG TAC AAG AAG AAC AAG GCG GGT CCT CAG C and icRev TAT ATA GCG GCC GCT CAT ACC CAC TCC TGG ACC GGC. This construction is referred to as MT5. Inactivation of MT5-MMP (E to Q mutation, referred to as MT5Δ) was performed using the Quick Change Lighting MultiSite-directed Mutagenesis kit (Agilent Technologies, Les Ulis, France) with the primer CCA TGA TAG CAC TGG GGT CAT TAG AGT GCT CCA AGC CCA GTG CAT GGC CCA GTT GAT GCA CGG CCA CCA GGA A, according to the manufacturer’s recommendations. GFP, MT5 and MT5Δ plasmids were amplified in Escherichia coli DH5α (Life Technologies) and purified using the NucleoBond Xtra Midi Plus EF (Macherey-Nagel, Hoerd, France) according to the manufacturer’s recommendations.
The TfR cDNA was amplified by PCR from human brain total RNA (Clontech), and cloned into pEGFP-C1 (Clontech). The following primers were used: TfRFor ATA TAT GAA TTC TAT GAT GGA TCA AGC TAG ATC AGC ATT CTC T and TfRev TTA ATT GTC GAC CTA AAA CTC ATT GTC AAG GTC CCA AAC GTC AC.
Cell culture and immunoprecipitation assay
Human embryonic kidney cells expressing APP carrying the Swedish mutation (HEKswe) were used to assess the interactions of MT5-MMP and its potential amyloidogenic properties in vitro. Like in the 5xFAD mice, the Swedish mutation increases the affinity of BACE1 for APP, which results in higher Aβ production . Cells were transiently transfected using the JetPei® transfection reagent according to the manufacturer’s instructions (Ozyme, Saint Quentin, France). Briefly, HEKswe cells were plated to 106 cells/ml for 24 h in six-well plates in DMEM Glutamax, 10 % FBS (fetal bovine serum) and 1 % penicillin/streptomycin (Life Technologies) and then transfected with 1 μg of GFP, MT5 or TfR plasmids for 16 h. The medium was replaced by OptiMEM containing ITS (Life Technologies) for 48 h. Cell lysates were prepared by sonication in RIPA buffer (Sigma-Aldrich) containing proteinase inhibitors (Millipore). Two hundred and fifty micrograms of proteins in a total volume of 500 μL of RIPA buffer was incubated overnight at 4 °C under agitation with 5 µg of non-specific mouse IgG antibodies (Jackson Immunoresearch) or mouse anti-GFP (Roche Diagnostics, Meylan, France) or anti-APP-CTF (Sigma-Aldrich) antibodies, and then pulled down for 2 h using protein G-coupled Dynabeads (Life Technologies). Samples were boiled and subjected to western blotting using 22C11 (1/1000) and APP-CTF (1/2000) antibodies, polyclonal anti-MT5-MMP (1/1000; developed in our laboratory) and polyclonal anti-transferrin receptor (1/250, Santa Cruz Biotechnology, Inc., Heidelberg, Germany). The latter was used as a putative negative control to demonstrate that overexpressed MT5-MMP/GFP does not inevitably interact in a non-specific manner with any endogenous or overexpressed transmembrane glycoprotein, such as the transferrin receptor (TfR; ~90 kDa).
Immunocytochemistry and co-localization in HEKswe cells
Double immunocytochemistry was performed on HEKswe cells to determine the colocalization of MT5-MMP/GFP and APP 48 h after transfection, as indicated above. Cells were fixed for 20 min with Antigenfix solution (Diapath, MM France, Francheville, France) at room temperature. After washing in PBS, cells were blocked for 1 h at room temperature using a PBS solution containing 0.1 % Triton X-100 and 3 % BSA. Cells were then incubated overnight at 4 °C with polyclonal anti-GFP (1/500) and 22C11 (1/500) antibodies, followed by incubation with appropriate Alexa Fluor®-coupled secondary antibodies for 1 h at room temperature (Life Technologies). Nuclei were stained with Hoechst #33258 (0.5 µg/ml, Life Technologies). Omission of the primary antibody was used as control and no immunostaining was observed. Coverslips were mounted using Prolong Gold Antifading reagent on Superfrost glass slides (Life Technologies). Images were taken and processed using a confocal microscope (LSM 700) and the Zen software (Zeiss). Co-localization analyses were performed using the Jacop plugin of ImageJ .
Aβ production assay
Aβ monomers were identified by immmunoprecipitation and western blot, as previously described . HEKswe cells were plated to 106 cells/ml for 24 h in six-well plates in DMEM Glutamax, 10 % FBS and 1 % penicillin/streptomycin (Life Technologies). The medium was discarded and replaced with OptiMEM containing 1 % FBS; then the cells were transfected with 1 μg of GFP and MT5-MMP/GFP by MT5 plasmids for 16 h . A hundred microliters of 100 mM Tris–HCl buffer, pH 8, containing 1.5 M NaCl and 50 mM EDTA was added to supernatants and then incubated overnight with polyclonal FCA40 antibody and protein A agarose beads (Life Technologies) for immunoprecipitation. Beads were washed twice, boiled and subjected to western blot. Aβ peptide was detected using the 6E10 antibody (1/200). Immunoblot signals were visualized using the ECL chemiluminescence kit (GE Healthcare, Dutscher, Brumath, France) and optical density was quantified using the ImageJ software.
Experiments were performed on hippocampal slices (300 µm) obtained from 4-month-old male mice. After decapitation, the brains were quickly dissected and placed in ice-cold buffer containing: 124 mM NaCl, 3.5 mM KCl, 25 mM NaHCO3, 1.25 mM NaH2PO4, 1 mM CaCl2, 2 mM MgSO4, 10 mM Glucose and 10 mM Hepes (bubbled with O2/CO2: 95/5 %). Slices were then prepared with a vibratome (VT1000S Leica, Germany) and maintained at room temperature for at least 1 h in the same buffer supplemented with 1 mM CaCl2, which was further used for recordings.
For electrophysiological recordings, slices were transferred to a multielectrode array (MEA60; Multi Channel Systems, Germany) comprising 60 extracellular electrodes [30, 31]. The interelectrode distance was 200 μm. Each individual electrode from the array could be used either as a recording or a stimulating electrode. A nylon mesh was positioned above the slice to obtain a satisfactory electrical contact with the electrode array. Slices were continually superfused with the extracellular medium described above (flow rate 2 ml/min) and kept at 32 °C. Stimulation was achieved with an external stimulator (STG 1004; Multi Channel Systems) by applying biphasic current pulses to one electrode of the array in the Schaffer collateral pathway of the hippocampus. Stimulation intensity (60–200 μA) and duration (70–150 μs) were adapted to avoid multiphasic responses due to an excessive stimulation . Selected electrodes of the array were then used to simultaneously record field excitatory postsynaptic potentials (fEPSPs) in the stratum radiatum of the CA1 hippocampal area. Test frequency to evoke fEPSPs was 0.066 Hz. Slices that displayed epileptic-like activity were discarded. Data storage and analysis were performed with MC Rack software (Multi Channel Systems). Changes in synaptic transmission were assessed by measuring peak amplitudes of fEPSPs, which were modified in a way similar to fEPSP slopes, as previously reported by others using MEAs . Applying a single train of stimulation at 100 Hz for 1 s triggered LTP.
Hippocampal-dependent spatial memory was assessed in a six-arm radial water maze (6-ARWM), using a protocol slightly modified with respect to the original method . The maze was inserted in a 120 cm diameter polycarbonate pool containing water at 20 °C, and distal cues of different shapes and sizes were placed on the walls of the testing room. Briefly, 10–12 mice/group were isolated 1 week before testing. Tests were performed during two consecutive days (15 trials/day). Each full entry in the non-goal arm (or 15 s without moving) was counted as an error, the maximal time to perform the task being 60 s. Once the platform was discovered or time had run out, the animal was allowed to rest on the platform for 15 s. Experimental groups of five to six mice were constituted and an hour rest period was allowed between each block of six trials. During the first day, the platform was alternately visible, and then hidden (2.5 cm below the water surface) for the first 12 trials. Then all trials were performed using the hidden platform configuration. For each mouse, the start location for each trial was determined randomly, while the goal arm remained constant for a given mouse. The sequence and total number of arms entered were recorded by an automatic video-tracking system (Viewpoint, Champagne au Mont d’Or, France).
Significant differences between groups were determined using the Kaleida Graph software. We used a one-way ANOVA followed by post hoc Fisher’s LSD or Kruskal–Wallis tests for multiple comparisons. Two-tailed unpaired Student’s t test was used to compare two experimental groups. Values represent the mean ± SEM of the indicated number of independent experiments/animals, and the level of significance was set for p < 0.05.
Reduced amyloid burden and neuro-inflammation in the brains of TgMT5−/− mice
Decreased C-terminal APP fragments and stable α-, β- and γ-secretase activities in the brains of TgMT5−/− mice
Long-lasting beneficial effects of MT5-MMP deficiency at advanced stages of the pathology
MT5-MMP interacts with APP, stimulates Aβ and C99 production and releases a soluble APP fragment of 95 kDa
These data, together with previous reports indicating the processing of APP by MT5-MMP , led us to hypothesize a possible interaction between MT5-MMP and APP not previously reported. The hypothesis was supported by two experimental evidences: (1) double immunocytochemistry experiments revealed partial colocalization of MT5-MMP and APP in MT5-MMP-transfected cells in perimembranous and cytosolic compartments (Fig. 7c); (2) MT5-MMP/GFP immunoprecipitated endogenous APP (Fig. 7d). Moreover, APP-CTF antibody immunoprecipitated MT5-MMP/GFP (Online resource 4a), but did not immunoprecipitate overexpressed TfR/GFP, which was used as negative control. As additional control of specific MT5-MMP/APP interactions, we showed that MT5-MMP/GFP did not immunoprecipitate the endogenous or overexpressed TfR (Online resource 4b). Altogether, these data provide the first evidence of MT5-MMP/APP interactions. To explore whether MT5-MMP deficiency could also affect putative sAPP95 levels in vivo, we analyzed the N-terminal fragments of APP using the 22C11 antibody. A fragment of 95 kDa was indeed detected in the soluble fraction of Tg brains at 2 months of age, and most interestingly its levels were reduced in the cortex (51 %, not shown) and hippocampus (55 %) of TgMT5−/− mice compared to Tg (Fig. 7e).
Preserved LTP in the hippocampus of TgMT5−/− mice
This protocol unveiled a deficiency in LTP in Tg mice at 4 months not previously detected using theta burst protocols . Indeed, CA1 LTP in Tg hipppocampi reached only 127 % of the basal fEPSP amplitude 50 min after HFS delivery (Fig. 8c). This weak LTP in Tg mice was in clear contrast with robust LTP exhibited by TgMT5−/− mice (195 %), indicating that MT5-MMP deficiency contributed to LTP preservation (Fig. 8c).
Improved performance of TgMT5−/− mice in the six-arm radial water maze
MT5-MMP is the only MMP predominantly expressed in the nervous system. Despite growing importance of MT5-MMP in nervous function and disease, and some evidence indicating functional interactions with key elements of the amyloidogenic process, its role in Alzheimer’s pathology remains unexplored. Accordingly, we have tested the hypothesis that MT5-MMP may influence amyloid-associated pathology and provide the first compelling evidence that this is the case in the 5xFAD mouse model of AD. We show in early stages of the pathology that MT5-MMP deficiency dramatically reduces amyloid load and C99 levels in the cortex and hippocampus of 5xFAD mice, which were concomitant with reduced neuroinflammation and the preservation of LTP and cognitive functions in bigenic mice. Long-lasting beneficial effects of MT5-MMP deficiency were also observed at advanced stages of the pathology, further highlighting the importance of MT5-MMP in the control of amyloidosis and associated neurodegeneration. Moreover, we discovered physical interactions between MT5-MMP and APP, and provide convergent data involving MT5-MMP in the processing of APP in vitro and in vivo. Together, the functional interactions between MT5-MMP and APP, without changes in the activity of the main secretases, may indicate that MT5-MMP acts somewhere along the amyloidogenic pathway to modulate APP trafficking/bioavailability and subsequently its processing.
The striking effects of MT5-MMP deficiency on Aβ load are reminiscent of those observed upon BACE1 deletion , or conditional PS1 knockout . Yet, the putative APP cleavage of MT5-MMP (and also MT1-MMP and MT3-MMP) of about 100 residues N-terminal upstream from the β-cleavage site  suggests that the MMP does not harbor a β-secretase-like activity, which could directly initiate Aβ generation. The stimulating effect of MT5-MMP on amyloidosis throughout our in vivo and in vitro experiments is in keeping with recent findings uncovering pro-amyloidogenic properties of MT1-MMP . It is thus possible that the MT5-MMP and MT1-MMP process APP in a similar manner. The degree of functional redundancy or specificity between these MMPs and how this affects AD pathogenesis remains to be determined. In any case, the pro-amyloidogenic features of MT5-MMP are in clear contrast with the Aβ-degrading activities classically attributed to soluble MMPs, i.e., MMP-2 and MMP-9 [9, 50, 51]. Such divergence could represent a fundamental difference in the way soluble and membrane-bound MMPs interact with APP/Aβ and contribute to their processing.
Simultaneously to reduced Aβ burden in TgMT5−/− brains, the decrease of glial reactivity and IL-1β levels illustrate a diminished inflammatory response in bigenic mice. Prominent IL-1β expression in reactive astrocytes of Tg mice is in agreement with similar astrocyte-restricted IL-1β expression found around amyloid plaques early after exogenous injection of Aβ1-42 . From a functional standpoint, the inhibition of IL-1β signaling has been reported to partly reduce fibrillar and oligomeric Aβ and to improve cognition in a mouse model of AD . It is also noteworthy that MT5-MMP deficiency alters IL-1β signaling in the peripheral nervous system in a non-AD context . Taken together, these findings advocate: (1) a role of astrocytes as key mediators of the Aβ-driven inflammatory response in early phases of the pathological process; and (2) a possible functional link between MT5-MMP and IL-1β-mediated inflammation, whereby MT5-MMP deficiency could interfere with IL-1β signaling. It is therefore plausible that a reduction in Aβ and IL-1β levels would jointly contribute to tune down microglial activation and homing around plaques, as observed in TgMT5−/− brains.
Small Aβ oligomers, including dimers and trimers, appear to be important in vivo Aβ assemblies , with neurotoxic properties that undermine memory [55, 56] and LTP [38, 57]. Accordingly, the remarkable reductions of Aβ trimers exhibited by TgMT5−/− brains correlate with the positive outcome of the pathology, possibly indicating a causal effect. Additional beneficial effects for TgMT5−/− mice may result from the strong reduction in the levels of Aβ precursor C99, which has been shown to be neurotoxic on its own . Interestingly, we have previously observed abnormal cortico-hippocampal accumulation of C99 in 5xFAD mice as early as 2 months of age, whereas Aβ trimers are first detected at 4 months . Drastic reductions in C99 upon MT5-MMP deficiency should be in principle coherent with changes in canonical APP or in the enzymatic systems that contribute to its generation (β-secretase) or conversion into Aβ (γ-secretase). However, none of these possibilities was supported by our data, which indicated stable levels of full-length APP and β-secretase activity between Tg and TgMT5−/− brains. Furthermore, unaltered PS1 autocatalysis or processing of the N-cadherin CTF1 fragment supported normal γ-secretase activity in TgMT5−/− brains, as well. This is in good agreement with the reported finding that another important substrate of γ-secretase, Notch, remains stable in MT5-MMP-deficient mice . In the absence of noticeable secretase dysfunctions, the differential impact of MT5-MMP deficiency on soluble and intracellular APP fragments suggest that MT5-MMP may influence the intracellular trafficking of APP and/or its targeting to subcellular compartments where C99 is degraded. Even though this hypothesis needs to be further investigated, it is consistent with previous reports showing that proteasome/lysosome activity accounted for significant degradation of C99 [59, 60]. Most interestingly, these studies suggested that decreases in C99 levels could be independent of γ-secretase activity, thus providing with alternative mechanisms to interpret changes in C99 turnover. Our experiments showing MT5-MMP/APP interactions and colocalization in perimembranous and intracellular compartments of HEKswe further support the idea that MT5-MMP could be involved in APP trafficking.
It is also noteworthy that MT5-MMP induces the release of a truncated APP fragment of 95 kDa, which is reminiscent of the N-terminal APP fragment reported in HEKWT cells transfected with either MT1-, MT3- or MT5-MMPs . In this study, the putative MT-MMP cleavage site was identified between residues 504–505 of recombinant APP. Our findings in HEKswe cells corroborate these previous observations in HEKWT cells, but most importantly provide the first in vivo evidence that the levels of a 95 kDa APP N-terminal fragment are reduced in TgMT5−/− brains compared to Tg. Altogether, these data support the idea that MT5-MMP can directly or indirectly influence APP processing in vivo. Wether there is a functional link between the MT5-MMP-mediated proteolytic event that releases sAPP95 and amyloidosis remains to be elucidated. In this context, one could expect that MT5-MMP activity generates a residual C-terminal APP fragment of ~25 kDa, but such fragment was not consistently detected in our experimental conditions, suggesting high instability. In the same vein, an unprecedentedly described APP C-terminal fragment of 25 kDa—named eta-CTF (η-CTF)—has been recently reported in various cell lines only after treatment with inhibitors of protein degradation . Interestingly, the generation of η-CTF is independent of BACE1 activity and is prevented by MMP inhibitors (A. Saunders, personal communication). Whether MT5-MMP or other MT-MMPs contribute to η-CTF production requires further investigation. Overall, gaining knowledge on alternative proteolytic APP processing should eventually help us to better understand the fate of about 50 % of APP processing that is not explained merely by the action of β-secretase, as reported .
In agreement with previous studies on 5xFAD mice  AMPA/kainate (GluA) receptor-mediated synaptic responses (basal synaptic transmission) and short-term presynaptic plasticity (paired-pulse facilitation) were not affected in our Tg or TgMT5−/− mice. This implies that MT5-MMP deficiency in vivo does not interfere with GluA functioning, as could be possibly inferred from functional interactions reported between MT5-MMP and GluA receptors in hippocampal neuronal cultures . Thus, the remarkable preservation of LTP in TgMT5−/− mice compared to Tg likely results from the robust reduction of Aβ burden, and in particular of Aβ trimers, which are considered to be powerful inhibitors of LTP . Overall, small Aβ oligomers are increasingly seen as a major trigger of synaptic dysfunction in AD  by altering NMDA receptors containing the GluN2B subunit . A plausible functional link between reduced amyloidosis and preserved synaptic plasticity is further supported by the improvement of LTP observed in bigenic 5xFAD/BACE1+/− mice, where strong decreases in Aβ levels are comparable to those found in our TgMT5−/− mice .
Using the 6-ARWM, we showed for the first time mild impairment of hippocampal-dependent associative spatial learning in 5xFAD mice at 4 months of age, in agreement with previous studies reporting that the first signs of behavioral disruption in 5xFAD mice occur between 4 and 5 months of age [19, 36]. Preliminary experiments failed to demonstrate significant alterations of learning using the classic Morris Water Maze test. This stresses the importance of using suitable tests to decipher subtle learning deficits characteristic of early phases of the degenerative process. Most importantly, we demonstrate that TgMT5−/− mice perform better than Tg mice in a context of preserved LTP. Similar positive correlations between learning and LTP have been reported in other AD mouse models, where 50 % reduction in Aβ levels was sufficient to improve LTP and spatial learning . Altogether, the age of 4 months in 5xFAD mice may represent a “prodromal-like” stage of the pathology, where we are able to detect subtle alterations of behavior. This highlights the potential therapeutic interest of modulating MT5-MMP activity before irreversible damage occurs.
We conclude that brain MT5-MMP is endowed with a regulatory role on processes that influence pathogenesis and cognitive decline in AD. MT5-MMP emerges as a proteinase that seems to exert multiple regulatory effects on APP metabolism, possibly mediated in a non-exclusive manner by protein–protein interactions and/or proteolytic processing. Further investigation will be required to gain insight into the nature of these interactions to better understand how MT5-MMP deficiency keeps Aβ load within a “safe” range in a durable manner without impacting on the activity of β- and γ-secretases. This could be of therapeutic relevance in the prospects of designing MT5-MMP-based strategies to modulate rather than inhibit secretase activities. Our data support the idea that MMPs are not only Aβ degrading enzymes and hence potentially beneficial, but on the contrary, some MMPs may promote Aβ/C99 accumulation and neurodegeneration. For these reasons, caution should be observed when targeting MMP activity in a non-specific manner. Accordingly, this study should lay the basis to encourage the design of therapeutic strategies to specifically modulate MT5-MMP activity and/or interactions with APP. Moreover, MT5-MMP expression being essentially restricted to neural tissues may further highlight the relevance of this proteinase as a therapeutic target to alleviate amyloid-mediated pathology in AD.
This work was supported by funding from the CNRS and Aix-Marseille Université, and by grants from the French National Agency for Research (ANR) to SR, MK and MV (TIMPAD, PREVENTAD and ADHOC, respectively) within the frame of the French “Fondation Plan Alzheimer”. The work was also supported by LECMA grants to SR, FC and MK, by grants to SR and MK from “Fonds Européen de Développement Régional” FEDER in PACA, by BMBF KNDD, SyNergy (DFG) and the Verum Foundation to SFL, by the Bavarian-French University Center to SR and SFL, by “Fondation pour la Recherche Médicale” and by the “Conseil Général des Alpes Maritimes” to FC. This work was also supported through the LABEX (excellence laboratory, program investment for the future) DISTALZ (Development of Innovative Strategies for a Transdisciplinary Approach to Alzheimer’s disease). KB was granted a research associate fellowship by the French “Fondation Plan Alzheimer”. YM was recipient of a EU Marie Curie international reintegration grant (IRG, NGINFAD) and a postdoctoral support from the CNRS. NAP and AEB were recipients of a doctoral fellowship from the French Ministry of Research. JMP was recipient of a doctoral fellowship from LECMA. The TfR plasmid and antibodies were kindly provided by Marion David. The authors thank Dr. G. Wenk and Dr. M. N. Simon for critical reading of the manuscript.
Compliance with ethical standards
Conflict of interest
The authors declare no financial conflict of interest that might be construed to influence the results or interpretation of the manuscript.
- 2.Kuhn PH, Wang H, Dislich B, Colombo A, Zeitschel U, Ellwart JW, Kremmer E, Rossner S, Lichtenthaler SF (2010) ADAM10 is the physiologically relevant, constitutive alpha-secretase of the amyloid precursor protein in primary neurons. EMBO J 29(17):3020–3032. doi: 10.1038/emboj.2010.167 PubMedPubMedCentralCrossRefGoogle Scholar
- 8.Girard SD, Jacquet M, Baranger K, Migliorati M, Escoffier G, Bernard A, Khrestchatisky M, Feron F, Rivera S, Roman FS, Marchetti E (2014) Onset of hippocampus-dependent memory impairments in 5XFAD transgenic mouse model of Alzheimer’s disease. Hippocampus 24(7):762–772. doi: 10.1002/hipo.22267 PubMedCrossRefGoogle Scholar
- 9.Py NA, Bonnet AE, Bernard A, Marchalant Y, Charrat E, Checler F, Khrestchatisky M, Baranger K, Rivera S (2014) Differential spatio-temporal regulation of MMPs in the 5xFAD mouse model of Alzheimer’s disease: evidence for a pro-amyloidogenic role of MT1-MMP. Front Aging Neurosci 6:247. doi: 10.3389/fnagi.2014.00247 PubMedPubMedCentralCrossRefGoogle Scholar
- 16.Folgueras AR, Valdes-Sanchez T, Llano E, Menendez L, Baamonde A, Denlinger BL, Belmonte C, Juarez L, Lastra A, Garcia-Suarez O, Astudillo A, Kirstein M, Pendas AM, Farinas I, Lopez-Otin C (2009) Metalloproteinase MT5-MMP is an essential modulator of neuro-immune interactions in thermal pain stimulation. Proc Natl Acad Sci USA 106(38):16451–16456. doi: 10.1073/pnas.0908507106 PubMedPubMedCentralCrossRefGoogle Scholar
- 17.Sekine-Aizawa Y, Hama E, Watanabe K, Tsubuki S, Kanai-Azuma M, Kanai Y, Arai H, Aizawa H, Iwata N, Saido TC (2001) Matrix metalloproteinase (MMP) system in brain: identification and characterization of brain-specific MMP highly expressed in cerebellum. Eur J Neurosci 13(5):935–948PubMedCrossRefGoogle Scholar
- 19.Oakley H, Cole SL, Logan S, Maus E, Shao P, Craft J, Guillozet-Bongaarts A, Ohno M, Disterhoft J, Van Eldik L, Berry R, Vassar R (2006) Intraneuronal beta-amyloid aggregates, neurodegeneration, and neuron loss in transgenic mice with five familial Alzheimer’s disease mutations: potential factors in amyloid plaque formation. J Neurosci 26(40):10129–10140. doi: 10.1523/JNEUROSCI.1202-06.2006 PubMedCrossRefGoogle Scholar
- 21.Colombo A, Wang H, Kuhn PH, Page R, Kremmer E, Dempsey PJ, Crawford HC, Lichtenthaler SF (2012) Constitutive alpha- and beta-secretase cleavages of the amyloid precursor protein are partially coupled in neurons, but not in frequently used cell lines. Neurobiol Dis 49:137–147. doi: 10.1016/j.nbd.2012.08.011 PubMedPubMedCentralCrossRefGoogle Scholar
- 22.Kuhn PH, Koroniak K, Hogl S, Colombo A, Zeitschel U, Willem M, Volbracht C, Schepers U, Imhof A, Hoffmeister A, Haass C, Rossner S, Brase S, Lichtenthaler SF (2012) Secretome protein enrichment identifies physiological BACE1 protease substrates in neurons. EMBO J 31(14):3157–3168. doi: 10.1038/emboj.2012.173 PubMedPubMedCentralCrossRefGoogle Scholar
- 24.Sbai O, Ferhat L, Bernard A, Gueye Y, Ould-Yahoui A, Thiolloy S, Charrat E, Charton G, Tremblay E, Risso JJ, Chauvin JP, Arsanto JP, Rivera S, Khrestchatisky M (2008) Vesicular trafficking and secretion of matrix metalloproteinases-2, -9 and tissue inhibitor of metalloproteinases-1 in neuronal cells. Mol Cell Neurosci 39(4):549–568PubMedCrossRefGoogle Scholar
- 27.Marambaud P, Chevallier N, Ancolio K, Checler F (1998) Post-transcriptional contribution of a cAMP-dependent pathway to the formation of alpha- and beta/gamma-secretases-derived products of beta APP maturation in human cells expressing wild-type and Swedish mutated beta APP. Mol Med 4(11):715–723PubMedPubMedCentralGoogle Scholar
- 31.Crouzin N, Baranger K, Cavalier M, Marchalant Y, Cohen-Solal C, Roman FS, Khrestchatisky M, Rivera S, Feron F, Vignes M (2013) Area-specific alterations of synaptic plasticity in the 5XFAD mouse model of Alzheimer’s disease: dissociation between somatosensory cortex and hippocampus. PLoS One 8(9):e74667. doi: 10.1371/journal.pone.0074667 PubMedPubMedCentralCrossRefGoogle Scholar
- 36.Girard SD, Baranger K, Gauthier C, Jacquet M, Bernard A, Escoffier G, Marchetti E, Khrestchatisky M, Rivera S, Roman FS (2013) Evidence for early cognitive impairment related to frontal cortex in the 5XFAD mouse model of Alzheimer’s disease. J Alzh Dis 33(3):781–796. doi: 10.3233/JAD-2012-120982 Google Scholar
- 39.Hung LW, Ciccotosto GD, Giannakis E, Tew DJ, Perez K, Masters CL, Cappai R, Wade JD, Barnham KJ (2008) Amyloid-beta peptide (Abeta) neurotoxicity is modulated by the rate of peptide aggregation: Abeta dimers and trimers correlate with neurotoxicity. J Neurosci 28(46):11950–11958. doi: 10.1523/JNEUROSCI.3916-08.2008 PubMedCrossRefGoogle Scholar
- 44.Jawhar S, Trawicka A, Jenneckens C, Bayer TA, Wirths O (2012) Motor deficits, neuron loss, and reduced anxiety coinciding with axonal degeneration and intraneuronal Abeta aggregation in the 5XFAD mouse model of Alzheimer’s disease. Neurobiol Aging 33(1):196.e129–196.e140. doi: 10.1016/j.neurobiolaging.2010.05.027
- 49.Saura CA, Chen G, Malkani S, Choi SY, Takahashi RH, Zhang D, Gouras GK, Kirkwood A, Morris RG, Shen J (2005) Conditional inactivation of presenilin 1 prevents amyloid accumulation and temporarily rescues contextual and spatial working memory impairments in amyloid precursor protein transgenic mice. J Neurosci 25(29):6755–6764. doi: 10.1523/JNEUROSCI.1247-05.2005 PubMedCrossRefGoogle Scholar
- 51.Yin KJ, Cirrito JR, Yan P, Hu X, Xiao Q, Pan X, Bateman R, Song H, Hsu FF, Turk J, Xu J, Hsu CY, Mills JC, Holtzman DM, Lee JM (2006) Matrix metalloproteinases expressed by astrocytes mediate extracellular amyloid-beta peptide catabolism. J Neurosci 26(43):10939–10948. doi: 10.1523/JNEUROSCI.2085-06.2006 PubMedCrossRefGoogle Scholar
- 52.Fragkouli A, Papatheodoropoulos C, Georgopoulos S, Stamatakis A, Stylianopoulou F, Tsilibary EC, Tzinia AK (2012) Enhanced neuronal plasticity and elevated endogenous sAPPalpha levels in mice over-expressing MMP9. J Neurochem 121(2):239–251. doi: 10.1111/j.1471-4159.2011.07637.x PubMedCrossRefGoogle Scholar
- 53.Carrero I, Gonzalo MR, Martin B, Sanz-Anquela JM, Arevalo-Serrano J, Gonzalo-Ruiz A (2012) Oligomers of beta-amyloid protein (Abeta1-42) induce the activation of cyclooxygenase-2 in astrocytes via an interaction with interleukin-1beta, tumour necrosis factor-alpha, and a nuclear factor kappa-B mechanism in the rat brain. Exp Neurol 236(2):215–227. doi: 10.1016/j.expneurol.2012.05.004 PubMedCrossRefGoogle Scholar
- 54.Kitazawa M, Cheng D, Tsukamoto MR, Koike MA, Wes PD, Vasilevko V, Cribbs DH, LaFerla FM (2011) Blocking IL-1 signaling rescues cognition, attenuates tau pathology, and restores neuronal beta-catenin pathway function in an Alzheimer’s disease model. J Immunol 187(12):6539–6549. doi: 10.4049/jimmunol.1100620 PubMedPubMedCentralCrossRefGoogle Scholar
- 57.Shankar GM, Li S, Mehta TH, Garcia-Munoz A, Shepardson NE, Smith I, Brett FM, Farrell MA, Rowan MJ, Lemere CA, Regan CM, Walsh DM, Sabatini BL, Selkoe DJ (2008) Amyloid-beta protein dimers isolated directly from Alzheimer’s brains impair synaptic plasticity and memory. Nat Med 14(8):837–842. doi: 10.1038/nm1782 PubMedPubMedCentralCrossRefGoogle Scholar
- 58.Ardiles AO, Tapia-Rojas CC, Mandal M, Alexandre F, Kirkwood A, Inestrosa NC, Palacios AG (2012) Postsynaptic dysfunction is associated with spatial and object recognition memory loss in a natural model of Alzheimer’s disease. Proc Natl Acad Sci USA 109(34):13835–13840. doi: 10.1073/pnas.1201209109 PubMedPubMedCentralCrossRefGoogle Scholar
- 60.Lauritzen I, Pardossi-Piquard R, Bauer C, Brigham E, Abraham JD, Ranaldi S, Fraser P, St-George-Hyslop P, Le Thuc O, Espin V, Chami L, Dunys J, Checler F (2012) The beta-secretase-derived C-terminal fragment of betaAPP, C99, but not Abeta, is a key contributor to early intraneuronal lesions in triple-transgenic mouse hippocampus. J Neurosci 32(46):16243a–16255a. doi: 10.1523/JNEUROSCI.2775-12.2012 CrossRefGoogle Scholar
- 61.Nunan J, Shearman MS, Checler F, Cappai R, Evin G, Beyreuther K, Masters CL, Small DH (2001) The C-terminal fragment of the Alzheimer’s disease amyloid protein precursor is degraded by a proteasome-dependent mechanism distinct from gamma-secretase. Eur J Biochem 268(20):5329–5336PubMedCrossRefGoogle Scholar
- 62.Page RM, Munch A, Horn T, Kuhn PH, Colombo A, Reiner O, Boutros M, Steiner H, Lichtenthaler SF, Haass C (2012) Loss of PAFAH1B2 reduces amyloid-beta generation by promoting the degradation of amyloid precursor protein C-terminal fragments. J Neurosci 32(50):18204–18214. doi: 10.1523/JNEUROSCI.2681-12.2012 PubMedCrossRefGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.