Introduction
Despite considerable academic and pharmaceutical investments during the last three decades, there is still no effective treatment for Alzheimer’s disease (AD). AD neuropathology is mostly defined by the accumulation of amyloid-beta (Aβ) plaques and neurofibrillary tangles (NFTs) in patient brains. Extracellular Aβ plaques are mainly formed by the aggregation of amyloid peptides (Aβ
40 and Aβ
42) whose production and accumulation are key elements in AD development [
15]. Intracellular NFTs are composed of aggregated hyper- and abnormal phosphorylated Tau proteins. Accumulating data from preclinical and clinical studies have established that several immune system-mediated factors, mainly driven by glial cells, also contribute to AD pathogenesis. Astrocytes and microglia surround amyloid plaques and release cytokines leading to inflammatory processes whose dysregulation contributes to AD pathology [
19,
37,
38].
Dual-specificity tyrosine-phosphorylation-regulated kinase 1A (DYRK1A), encoded by a gene localized in the Down syndrome critical region of chromosome 21, is a serine/threonine protein kinase which contributes to various biological processes in the embryonic and adult central nervous systems [
39]. In AD, DYRK1A is especially known to phosphorylate Tau at several sites including Thr181, Thr212, and Thr231, which are all observed in NFTs of AD brains [
29,
33,
46] and the amyloid precursor protein (APP) at Thr-668 [
34] or the presenilin 1 (PS1) at Thr-354 [
36]. DYRK1A also phosphorylates several immune response mediators associated with AD, including calcineurin-nuclear factor of activated T cells (NFAT) [
1] and signal transducer and activator of transcription-3 (STAT3) [
26]. Recent study showed that DYRK1A inhibition reduces APP phosphorylation and insoluble Tau phosphorylation and thereby reverse cognitive deficits in AD mice [
3]. However, previous contradictory studies have been published [
11,
23] and further study are required to confirm the DYRK1A protein levels in brain of individuals with AD. Furthermore, it is emphasized that level of its kinase activity is still unknown. Thus, the relevance of inhibiting kinase activity of DYRK1A in AD remains a matter of debate.
Here, we show that DYRK1A is truncated in the AD context. This increase of truncated forms of DYRK1A (DYRK1A
T) is associated with a decrease of full-length form of DYRK1A (DYRK1A
FL) thus confirming previous report by Jin and colleagues [
23]. This was observed in hippocampus from AD patients but also in APP/PS1 mice, an amyloid mouse model of AD [
22]. We demonstrated for the first time that this proteolysis is occurring in astrocytes and is not associated with a modification of the global DYRK1A kinase activity in AD. In vitro, we show that, compared to DYRK1A
FL, DYRK1A
T exhibit stronger affinity toward STAT3ɑ. We identified Leucettine L41, derived from the marine sponge alkaloid Leucettamine B [
8,
41], as an appropriate compound to inhibit DYRK1A proteolysis. To decipher the effects of DYRK1A proteolysis and its inhibition in vivo, we treated APP/PS1 mice with the leucettine L41. We show in the present study that L41 prevents DYRK1A proteolysis and reduces STAT3ɑ phosphorylation in APP/PS1 mice. Neuroinflammation, amyloid plaque load, synaptic plasticity and cognitive functions are improved. Altogether, our results confirm the involvement of DYRK1A
T in AD pathology and demonstrate the relevance of inhibitors of DYRK1A cleavage as a potentially relevant therapeutic strategy.
Material and methods
Animals
Fourteen APP
swe/PS1
ΔE9 mice (referred to as APP/PS1; Jackson Laboratories) and 12 age-matched littermate control mice were used for behavioral (Morris Water Maze), pathology, and biochemistry studies. A second cohort composed of 14 APP/PS1 and nine littermates were used for behavioral (Y-maze) and electrophysiological analysis. APP/PS1 mice express the human APP gene carrying the
Swedish double mutation (K595 N/M596 L). In addition, they express the human PS1∆E9 variant lacking exon 9 [
22]. Only male mice were used. The ages at treatment and analysis/sacrifice are given in the Results section. All experiments were conducted in accordance with the ethical standards of French, German, and European regulations (European Communities Council Directive of 24 November 1986). The supervisor of in vivo studies (J Braudeau) received official authorization from the French Ministry of Agriculture to carry out research and experimentation on animals (authorization number APAFIS#4449–2,016,031,012,491,697).
Tissue collection and sample preparation
Mice were anesthetized with ketamine/xylazine (100 and 10 mg/kg respectively) and decapitated. One hemisphere was post-fixed by incubation for 72 h in 4% PFA, cryoprotected in 30% sucrose in PBS, and cut into 40 μm sections with a freezing microtome (Leica) for histological analyses. The contralateral hemisphere was dissected for hippocampus isolation. Samples were homogenized in a lysis buffer (150 mM NaCl and 1% Triton in Tris-buffered saline) containing phosphatase (Pierce) and protease (Roche) inhibitors and centrifuged for 20 min at 15000 x g. The same procedure was applied to human samples.
Leucettine L41 treatment
The pre-weighed compound was dissolved in DMSO/PEG300/water (5/35/60) to a final concentration of 2 mg/mL for a dose of 20 mg/kg. The formulation was prepared on the day of the in vivo experiment. The mice received five intraperitoneal injections per week for four weeks.
DYRK1A in vitro proteolysis
Human hippocampus tissue and 4 months-aged mouse (C57Bl6) hippocampus were homogenized in 9 volumes of buffer consisting of 50 mM Tris-HCl (pH 7.4), 8.5% sucrose, 10 mM β-mercaptoethanol, 2.0 mM EDTA, followed by centrifugation at 16,000×g at 4 °C for 10 min. The supernatants were incubated in the presence or absence of various concentrations of Ca2+ with or without Harmine, Leucettine LeuI or Leucettine L41 at various concentrations (0.1; 1.0 or 2.0 μM) for 10 min at 30 °C. The reactions were terminated by the addition of 4-fold concentrated SDS-PAGE sample buffer, followed by heating in boiling water for 5 min. The products of proteolysis were analyzed by Western blots developed with antibody to DYRK1A.
Identification of DYRK1A interactions
Homogenized total proteins from mouse hippocampus tissue were incubated with 2 mM EDTA, 0 or 2 mM of Ca2+ and 0 or 1 μM of Leucettine L41 during 10 min at 30 °C. 200 μg of total proteins were incubated with 2 μg of α-DYRK1A-Nter antibody (DYRK1A D1694) overnight at 4 °C. The proteins interacting with DYRK1A were revealed by Western blots developed with STAT3 (1/1000, Cell Signaling), NFATc1 (1/1000, Cell Signaling), APP, Tau and PS1.
Western blotting
Equal amounts of protein (30 μg) were separated by electrophoresis in NuPAGE Bis-Tris Gels (Life Technologies) and transferred to nitrocellulose membranes. The membranes were hybridized with various primary antibodies (DYRK1A 7D10 (1/250, Abnova), DYRK1A D1694 (1/500, Sigma), GFAP (DAKO, 1/4000), Vimentin (1/1000, Abcam), pSTAT3(Y705) (1/1000, Cell Signaling), STAT3 (1/1000, Cell Signaling), IBA1 (1/2000, Wako), CD68 (1/1000, BioLegend), IDE (1/200, Santa Cruz), TREM2 (R&D Systems, 1/500), GAPDH (1/1000, Abcam)). Various secondary antibodies were also used (ECL Anti-mouse horseradish peroxidase linked, 1/2000, GE Healthcare; ECL Anti-rabbit horseradish peroxidase linked, 1/2000, GE Healthcare). Membranes were developed using enhanced chemiluminescence (Thermo Fisher Scientific). Signals were detected with Fusion FX7 (Vilber Lourmat) and analyzed and quantified using ImageJ (NIH).
Elisa
Inflammatory cytokines and interleukin concentrations were measured using the MSD Mouse V-PLEX Plus Proinflammatory Panel 1 kit (Meso Scale Diagnostics). Extracted soluble Aβ forms were quantified with the MSD Human and rodents V-Plex Plus Aβ Peptide Panel 1 (4G8) (Meso Scale Diagnostics). All ELISA were performed according to each kit manufacturer’s instructions.
Calpain activity
Calpain activity was measured using the fluorogenic peptide N-Succinyl-Leu-Tyr-7-Amido-4-Methylcoumarin as described by Kohli et al. [
25]. Briefly, 60 μg brain extract in a final volume of 40 μL was added to 160 μL 50 μM N-Succinyl-Leu-Tyr-7-Amido-4-Methylcoumarin dissolved in dimethyl sulfoxide and Tris buffer (100 mM Tris-HCl, 145 mM NaCl at pH 7.3). Proteolysis of the substrate was monitored for 21 min at room temperature with a FlexStation3 multi-mode microplate reader (excitation: 380 nm, emission: 460 nm; Molecular Devices) either in the presence of 10 mM Ca
2+ or 10 mM EGTA to determine calcium-independent activity, thus excluding cathepsin activity.
DYRK1A kinase activity assay
Catalytic activity of total and endogenous DYRK1A was assessed using high-performance liquid chromatography (HPLC), based on the separation and quantification of specific fluorescent peptides (substrate and phosphorylated product), as previously described [
4]. Both negative (without protein) and positive controls (DYRK1A recombinant protein) have been used to determine background and to check the specificity of the kinase activity assay.
Immunohistochemistry and image acquisition
Cryosections were washed with 0.25% Triton in 0.1 M PBS, incubated in an 88% formic acid solution for 15 min (antigen retrieval), and saturated by incubation in 0.25% Triton in 0.1 M PBS/5% goat serum. They were then incubated with primary antibodies (DYRK1A 7D10 (1/100, Abnova), DYRK1A D1694 (1/100, Sigma), GFAP (DAKO, 1/2000), APP 4G8-Biotin, (1/1000, Covance), IBA1 (1/500, Wako)). For non-fluorescent immunostaining, endogenous peroxidase was quenched with PBS containing 3% H2O2 for 5 min, followed by amplification using the ABC system (VECTASTAIN Elite ABC HRP Kit, Vector Laboratories, Burlingame, CA, USA). Horseradish peroxidase conjugate and 3,3′-diaminobenzidine were then used according to the manufacturer’s manual (Vector® DAB, Vector Laboratories, Burlingame, CA, USA). The sections were mounted, dehydrated by passing twice through ethanol and toluol solutions, and cover-slipped with Eukitt (O. Kindler). Images were captured with a Leica DM6000 microscope and analyzed using ImageJ software (NIH). For fluorescent immunostainings, slices were incubated with secondary Alexa Fluor conjugated antibodies (Invitrogen). Slices were stained with DAPI (1/5000, Sigma), mounted in Vectashield fluorescent mounting media (Vector laboratories) and conserved at 4 °C. Images were captured with a Leica SP8 confocal microscope (Leica) and analyzed using ImageJ software (NIH). Laser power, numeric gain, and magnification were kept constant between animals to avoid potential technical artefacts. Images were first converted to 8-bit gray scale and binary thresholded to highlight positive staining. At least three sections per mouse (between − 1.7 mm to − 2.3 mm caudal to Bregma) were quantified. The average value per structure was calculated for each mouse. For quantification of Iba1 and GFAP immunoreactivity around plaques, a region of interest (ROI) was drawn around the center of the plaque. The diameter of the circular ROI was set to two times the diameter of the plaque. Mean fluorescence intensity values were measured for either DYRK1A (7D10) (α-DYRK1A-Cter), DYRK1A (D1694) (α-DYRK1A-Nter) Iba1, or GFAP immunoreactivity and were processed using Icy software (Institut Pasteur, Paris, France). Experiments and data analysis were performed blind with respect to treatments and genotypes.
Behavioral assessment
Y-maze. Experiments were performed in a maze consisting of three transparent plastic arms, 46 × 11 × 25 cm each, set at a 120° angle relative to each other. During the first trial, mice could freely explore two arms, called familiar arms (FA), for 3 min, whereas the third arm was blocked by an opaque door. Assignment of arms was counterbalanced randomly within each experimental group to avoid any preference-related bias. Mice were then returned to their home cage for 5 min. Finally, mice were returned to the maze and allowed to explore all three arms, including the novel arm (NA), for 3 min. The maze was carefully cleaned with a 70% ethanol solution between each exploration phase to remove any olfactory cues. EthoVision software was used for recording and analysing each exploration trial.
Morris water maze
Experiments were performed in a tank 180 cm in diameter and 50 cm deep, filled with opaque water maintained at 21 °C, equipped with a platform of 18 cm in diameter, submerged 1 cm below the surface of the water. Visual clues were positioned around the pool and luminosity was maintained at 350 lx. The mice were initially exposed to a learning phase, which consisted of daily sessions (three trials per session) on five consecutive days. The starting position varied pseudo-randomly, between the four cardinal points. There was a mean interval of 20 min between trials. The trial was considered to have ended when the animal reached the platform. Long-term spatial memory was assessed 72 and 120 h after the last training trial in a probe trial in which the platform was no longer available. Animals were monitored with EthoVision software.
Ex vivo electrophysiology
Slice preparation. Mice were anesthetized with halothane and decapitated. The brain was rapidly removed from the skull and placed in chilled (0–3 °C) artificial cerebrospinal fluid (ACSF) containing 124 mM NaCl, 3.5 mM KCl, 1.5 mM MgSO4, 2.5 mM CaCl2, 26.2 mM NaHCO3, 1.2 mM NaH2PO4, and 11 mM glucose. Transverse slices (300–400 μm thick) were cut with a vibratome and placed in ACSF in a holding chamber, at 27 °C, for at least one hour before recording. Each slice was individually transferred to a submersion-type recording chamber and submerged in ACSF continuously superfused and equilibrated with 95% O2, 5% CO2.
Electrically induced long-term potentiation (LTP) was studied. Theta-burst stimulation (TBS), mimicking the natural stimulation at the theta frequency from the medial septum to the hippocampus, consisting of five trains of four 100 Hz pulses each, separated by 200 ms, was applied at the test intensity. The sequence was repeated three times, with an interburst interval of 10s. Testing with a single pulse was then performed for 60 min (TBS) or 75 min (3 × 100 Hz), to determine the level of LTP.
Statistical analysis
Data are expressed as the mean ± SEM. Statistical analyses were performed with GraphPad Prism (GraphPad Soft-ware, La Jolla, CA, USA) software. One-way ANOVA followed Tukey’s post-hoc tests were used to determine the significance of differences between groups. Student’s t test was used when only two groups were analyzed. All values are given as mean ± SEM. Statistical significance was set to a P-value < 0.05 for all tests.
Discussion
The present study provides compelling evidence for DYRK1A involvement in AD and describes a new mechanism through which DYRK1A modulation contributes to AD pathology. We first describe a proteolytic processing of DYRK1A in the hippocampus of AD patients and APP/PS1 mice reducing level of full-length form of DYRK1A (DYRK1AFL) and producing truncated forms (DYRK1AT). The increase of DYRK1A kinase activity was suspected to contribute to AD. However, we demonstrate that this proteolysis is not associated with modification of the global DYRK1A kinase activity but affect its specificity. DYRK1AT forms exhibit increased affinity toward STAT3α, an activator of neuroinflammation. We then show that inhibiting DYRK1A proteolysis through the peripheral administration of Leucettine L41 in APP/PS1 mice is associated with increased number of phagocytic-microglia around amyloid-β deposits and reduction of the plaque load. This is associated to improved synaptic plasticity and reduced cognitive and memory deficits in APP/PS1 mice. Specific inhibitors of DYRK1A proteolysis could be therapeutic interest for AD.
The
DYRK1A gene is located on chromosome 21 (21q22.2), a region known as the Down- Syndrome Critical Region (DSCR) [
9]. People living with Down Syndrome (DS) have higher prevalence to develop AD pathology primarily due to overexpression of the
APP gene on chromosome 21 [
17]. In addition, various evidence supports DYRK1A as a potential key player in AD progression and as a valid therapeutic target for AD [
3,
35]. However, no direct link has been shown between the kinase activity of DYRK1A and AD. Recently, decreased DYRK1A
FL and increased DYRK1A
T forms have been reported in the frontal and temporal cortex of AD patients (Braak V-VI/Tangles score = 14) through upregulated calpain I activity, the major calpain isoform in brain [
23]. We confirm this observation in hippocampus from severely affected AD patients (Braak V-VI, Thal IV-V). We show for the first-time decreased levels of DYRK1A
FL forms in APP/PS1 mice, together with increased DYRK1A
T forms and increased calpain activity.
DYRK1A has a long kinase domain, followed by a PEST region, and histidine-repeat and serine/threonine-rich domains [
24]. PEST sequences are rich in the amino acids proline (P), glutamate (E), serine (S) and threonine (T), and their presence is correlated with rapid protein turnover due to proteasome-mediated destruction [
32]. Proteolysis of DYRK1A by calpains probably occurs within the PEST domain [
23] as shown in other proteins [
32]. Beyond the PEST sequence, the C-terminus domain is the target of other proteins which negatively modulate DYRK1A kinase activity [
49]. In numerous protein kinases, non-catalytic domains participate in the kinase specificity [
44]. In our study, we provide evidences that the deletion of the C-terminal region does not affect kinase activity of DYRK1A but increases its affinity toward STAT3α.
Several molecules able to reduce DYRK1A kinase activity have been developed [
31]. One of them, Leucettine L41 (L41) is a synthetic analogue of the marine sponge alkaloid Leucettamine B, identified as an inhibitor of DYRKs/CLKs [
8,
41]. Although L41 prevents DYRK1A proteolysis in APP/PS1 mice hippocampus, we showed that this compound does not alter DYRK1A (and Calpain) activity. Thus, several hypotheses can be considered to explain L41 effect on DYRK1A: (i) L41 could prevent the interaction between DYRK1A and calpains by inducing a conformational change in the kinase, (ii) the DYRK1A/L41 complex could change its intracellular location and thereby be isolated from calpains, or (iii) the catalytic activity of DYRK1A could be required for calpain-mediated cleavage. Further experiments are still needed to better understand the action of L41 on DYRK1A regulation.
Consequently to the prevention of DYRK1A, we showed a decrease of STAT3α phosphorylation (pSTAT3α) in APP/PS1 mice. STAT3ɑ is a transcription factor and a major regulator of cytokine production [
18]. The tyrosine phosphorylation is required for its activation and STAT3α are remarkably activated in APP/PS1 mice [
7]. Inhibition of STAT3α phosphorylation attenuates Aβ-induced neuronal death [
45]. Our results indicate that a normalization of pSTAT3ɑ levels by L41 restores pSTAT3α/STAT3α ratio and may participate to the following events: (i) decreased release of key inflammatory mediators such as IL-1β, TNF-
α and IL-12, (ii) increased microglial cells recruitment around amyloid plaques, and (iii) decrease of the amyloid-β burden. The immune system, driven by inflammatory mediators, influences AD progression [
19]. In particular, TNF-α has been described to have a major impact in AD. Indeed, increased TNF-α in serum is associated with a worsen cognitive decline in AD [
21] and elevated concentrations of TNF-α in CSF increase the probability to evolve from a mild cognitive impairment (MCI) stage to dementia [
42]. In addition, these molecules have a strong impact on microglial dysfunction [
5,
40]. Emerging evidences suggest that microglial activation plays a crucial role in AD [
16]. Activated microglia have receptors that can uptake and clear Aβ and this may limit the formation of plaques through phagocytosis of Aβ species [
38]. In APP/PS1 mice, proinflammatory cytokines (IL-1β and TNF-α) increase in concentration with age and down regulate genes involved in amyloid-β clearance [
20]. Therefore, microglia become progressively dysfunctional and display altered activation as the disease progresses. Our data show that L41 treatment in APP/PS1 mice promotes a moderate reduction of the amyloid load, which may be explained by the induction of an activated microglia phenotype expressing increased levels of TREM2 (triggering receptor expressed on myeloid cells 2) and IDE (insulin degrading enzyme), two microglial proteins that have been demonstrated to regulate Aβ deposition in AD mouse models [
12,
14,
48]. A previous study reported that inhibition of calpains mitigates AD pathology and cognitive decline in 3xTgAD mice [
30]. We show that L41 has no effect on calpains activity which remains elevated in APP/PS1 mice. Interestingly, selective effect on DYRK1A proteolysis by Leucettine L41 improved synaptic plasticity measured by LTP and rescued spatial learning, working memory and long-term memory impairments in APP/PS1 mice tested with the Y-maze and the Morris water maze tasks at 14 months of age. These findings suggest that preventing DYRK1A proteolysis is sufficient to observe disease-modifying effects in this mouse model. This is supported by comparative evaluation of another synthetic analogue of Leucettamine B, the LeuI in APP/PS1 mice at the same age. As showed in vitro (see Fig.
1b), LeuI is unable to prevent DYRK1A proteolysis. Using similar experimental conditions, we compared LeuI and L41-treated APP/PS1 mice (Additional file
5: Figure 5A-F). No rescue of DYRK1A
FL levels and DYRK1A
T in astrocytes or no significant decrease of pro-inflammatory cytokines were observed in LeuI treated animals (Additional file
5: Figure 5A-C). Moreover, no decrease of the amyloid plaque burden and no improvement of the spatial / long-term memory were measured (Additional file
5: Figure 5D-F). These data confirm the role of DYRK1A proteolysis in AD progression and the potential interest of this mechanism as a new therapeutic target to counteract the disease.