Introduction
Amyloid beta deposition and microglial activation are two major pathophysiologic hallmarks of the progression of Alzheimer’s disease, often suggested to be associated with each other [
1]‐[
9]. Microglia are the resident phagocytes of the central nervous system. In the AD brain, microglia are found in a highly activated state often in association with senile plaques [
10]‐[
13]. Activated microglia present a different phenotype as compared to resting microglia, and are responsible for, in particular, pro-inflammatory cytokine secretions [
10]‐[
18]. Aβ accumulation in AD has been associated with these increases in pro-inflammatory markers [
19]‐[
25]. Aβ peptide has the ability to self-aggregate to form oligomers and fibrils. Both forms have been reported to have neurotoxic and gliotic actions
in vitro as well as
in vivo[
24]‐[
36]. Because of the close proximity of reactive microglia to deposited fibrils of Aβ, it is hypothesized that Aβ acts as a stimulus for microglial activation during disease to initiate or propagate the inflammatory changes observed. Therefore, there is a need to identify a therapeutically approachable microglial target for reducing inflammation in the brains of AD patients. Based upon the fact that recent data demonstrate that prolonged non-steroidal anti-inflammatory drug use decreases incidence of AD when taken by asymptomatic individuals, an anti-inflammatory approach to disease may be an effective prevention strategy [
37]. It is imperative to understand the mechanism(s) by which microglia become reactive to better design anti-inflammatory drug strategies.
Although data suggest a causative role of Aβ deposition for microgliosis, the underlying mechanisms involved are not fully resolved [
10,
18,
28,
38]‐[
41]. It has been demonstrated through a variety of studies that Aβ is capable of stimulating microglia
in vitro and
in vivo to increase protein phosphotyrosine levels. This correlates well with the reported increase in microglial phospho-tyrosine immunoreactivity in AD brains [
42]. These data have supported a hypothesis that the increase in phosphotyrosine immunoreactivity is due to either increased tyrosine kinase activity or decreased tyrosine phosphatase activity. It appears that both scenarios may be true. Microglia can use a multi-receptor complex for interacting with Aβ fibrils on the plasma membrane [
43]. Upon ligand binding, a specific signaling pathway is activated involving propagation and downstream increased activation of numerous non-receptor tyrosine kinases, including Src, Lyn, FAK and PYK2 [
24,
28,
39,
41,
44]‐[
47]. Based upon inhibition studies, the increased tyrosine kinase enzyme activities upon Aβ binding are absolutely critical for microgliosis to occur [
24,
28,
39,
41,
44]‐[
47]. In fact, it appears that not only increased tyrosine kinase activity is required for Aβ stimulation but also decreased tyrosine phosphatase activity [
48,
49]. Our prior work demonstrated that both oligomeric and fibrillar forms of Aβ stimulate increased protein phosphotyrosine levels
in vitro and
in vivo that correlated with activation of non-receptor tyrosine kinases [
38,
50]. Irrespective of the form of Aβ involved, one common mechanism of action appears to be involvement of tyrosine kinases leading to increased secretion of proinflammatory cytokines. This study tests whether inhibition of the Aβ fibril-stimulated signaling response, more precisely non-receptor tyrosine kinase activity, can attenuate microgliosis both
in vitro and
in vivo. The Src-Abl inhibitor, dasatinib, was used to treat primary murine microglia cultures
in vitro. In order to quantify effects of dasatinib in a more physiologically relevant form of disease, the drug was also administered to a transgenic mouse model of AD. This APP/PS1 mouse line expresses a Swedish mutation in APP and a deltaE9 mutation of presenilin 1 (PS1). The mice over-express human Aβ with a correlating high Aβ plaque immunoreactivity and microgliosis [
51,
52].
In this work, we demonstrated using primary murine microglia cultures that dasatinib was able to attenuate the Aβ-dependent increase in overall protein phospho-tyrosine levels and active levels of Src and Lyn non-receptor tyrosine kinases which correlated with decreased TNFα secretion. In addition to the in vitro analyses, dasatinib was able to reduce active Src but not Lyn levels as well as TNFα and microgliosis in the APP/PS1 mice following 28 days of subcutaneous infusion. Our study indicates that attenuation of specific non-receptor tyrosine kinase activities, in our case using an FDA approved cancer drug, dasatinib, may be therapeutically useful as a novel anti-inflammatory approach to AD.
Methods
Materials
Anti-Aβ, clones 6E10 and 4G8 were from Covance (Emeryville, CA, USA). The anti-Lyn antibody, anti-Src, anti-α-tubulin antibodies, FITC and Texas Red conjugated secondary antibodies, and horseradish peroxidase conjugated secondary antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Mouse TNF-α Elisa kit was obtained from R&D Systems (Minneapolis, MN, USA). Anti-phosphotyrosine (4G10) antibody was from EMD Millipore (Billerica, MA, USA) and anti-pLyn (Tyr 396) antibody was purchased from Abcam (Cambridge, MA, USA). Anti-Iba1 antibody was from Wako Chemicals USA, Inc (Richmond, VA, USA). Elite Vectastain ABC avidin and biotin and alkaline phosphatase kits, biotinylated anti-rabbit, anti-mouse, and anti-rat antibodies and the Vector VIP and Vector Blue chromogen kits were from Vector Laboratories Inc. (Burlingame, CA, USA). Anti-CD68 was obtained from Serotec (Raleigh, NC, USA). Anti-PSD95 and anti-pSrc (Tyr416) antibodies were purchased from Cell Signaling Technology (Danvers, MA, USA). Anti-APP antibody was purchased from Invitrogen (Camarillo, CA, USA). Synaptophysin and βIII tubulin antibodies were purchased from Chemicon International, Inc (Temecula, CA, USA). The non-receptor tyrosine kinase inhibitor, dasatinib, was obtained from LC Laboratories (Woburn, MA, USA). The transgenic mouse line, strain 005864 B6.Cg-Tg(APPswe,PSEN1dE9)85Dbo/J and wild type mouse line, C57BL/6 were obtained from the Jackson Laboratory (Bar Habor, ME, USA).
Animal use
All animal use was approved by the University of North Dakota Institutional Animal Care and Use Committee (UND IACUC). Mice were provided food and water ad libitum and housed in a 12 h light:dark cycle. The investigation conforms to the National Research Council of the National Academies Guide for the Care and Use of Laboratory Animals (8th edition).
BV2 cell line
Immortalized murine microglial BV2 cells were obtained from Dr. Gary E. Landreth, Cleveland, OH, USA. The cells were maintained at 3 × 106 cells/dish in 100-mm dishes in DMEM/F12 (Gibco RBL, Rockville, MD, USA) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (U.S. Biotechnologies Inc., Parkerford, PA, USA), 5% horse serum (Equitech-Bio, Inc., Kerrville, TX, USA), penicillin G (100 units/ml), streptomycin (100 mg/ml), and L-glutamine (2 mM) and incubated at 37°C in a humidified atmosphere containing 5% CO2 and 95% air.
Murine microglia culture
Microglia were derived, as described previously [
53], from the brains of postnatal day 1 to 3 C57BL/6 J mice. Briefly, meninges-free cortices were removed, trypsinized and triturated in microglia media (DMEM/F12 media containing L-Glutamine (Invitrogen, Carlsbad, CA, USA) and 20% heat inactivated FBS) and placed in T-75 flasks. The media in the flasks was replaced completely after 24 h and partially after 7 days with fresh media. The cells were ready to harvest and count after 14 days.
Cell stimulation
Aβ 1–42 fibrils (American Peptide, Sunnyvale, CA, USA) were prepared according to an established protocol [
54]. Microglia were stimulated by removing them from growth media into serum-free DMEM/F12 media containing Aβ fibrils. To assess dasatinib effects, BV2 or microglia were pretreated with the drug for 30 minutes before Aβ stimulations. For Western blot analyses, five- minute Aβ stimulations were performed and cells were lysed using radioimmunoprecipitation assay buffer (RIPA) (20 mM Tris, pH 7.4, 150 mM NaCl, 1 mM Na3VO4 10 mM NaF, 1 mM EDTA, 1 mM EGTA, 0.2 mM phenylmethylsulfonyl fluoride, 1% Triton X-100, 0.1% SDS, and 0.5% deoxycholate) with protease inhibitors (AEBSF 104 mM, Aprotinin 0.08 mM, Leupeptin 2.1 mM, Bestatin 3.6 mM, Pepstatin A 1.5 mM, E-64 mM). Protein concentrations were determined using the Bradford method [
55]. For ELISA and toxicity analyses, cells were stimulated with Aβ for 24 hours. To assess the level of proinflammatory cytokine, TNF-α, secreted after 24 h stimulation of microglia with Aβ fibril, the media from the cells was collected and analyzed using a commercially available ELISA kits (R & D Systems) according to the manufacturer’s protocol. Cell viability was assessed by the MTT reduction assay. After 24-hour stimulation of microglia, media was removed for ELISA and the cells were incubated with 3[4,5-dimethylthiazol-2-y1]-2,5-diphenyltetrazolium bromide (MTT, 100 μg/mL) for 4 hours. The media was aspirated and the reduced formazan precipitate was dissolved in isopropanol. Absorbance values were read at 560/650 nm via plate reader and averaged +/− SEM.
Western blot analysis of microglia cultures
Lysates from cell stimulation experiments were diluted into sample buffer and separated via 10% SDS-PAGE, transferred to polyvinylidene difluoride membrane and Western blotted using anti-pSrc (Tyr416), anti-pLyn (Tyr396), Src (loading control) and Lyn (loading control) and enhanced chemiluminescence for detection (GE Healthcare, Piscataway, NJ, USA). pSrc/pLyn optical densities (O.D.) from visualized Western blots were normalized to their respective loading controls (Src/Lyn) and averaged from five independent experiments.
Collection of brains from different age APP/PS1 mice
Brains from different aged APP/PS1 mice were collected for longitudinal analyses. 2-, 4-, 6- and 12-month old transgenic mice (n = 5 to 6) along with their age-matched C57BL/six wild type controls were euthanized and perfused with PBS-CaCl2. Brains were rapidly dissected and divided into left and right hemispheres, with right hemispheres fixed in 4% paraformaldehyde for sectioning. The left hemispheres were further dissected into different brain regions to obtain hippocampus, temporal cortex, frontal cortex and cerebellum, and flash frozen using liquid N2. The frozen tissue was lysed using RIPA with protease inhibitors and used for Western blot analysis.
Subcutaneous infusions of dasatinib into APP/PS1 mice
Dasatinib was infused subcutaneously into female APP/PS1 mouse at 13 months of age. Dasatinib was delivered via mini-osmotic pumps (model 1004, 0.25 μL/hour delivery rate, Alzet, Cupertino, CA, USA). Pumps delivered either vehicle (DMSO/Hepes) (n = 6) or dasatinib (500 ng/kg/day) (n = 7) for 28 days. At the end of the infusion period, mice were euthanized, brains perfused with PBS-CaCl2 and rapidly collected. Control untreated APP/PS1 animals were collected at a comparable age of completion, 14 months. The right hemispheres were collected for fixing in 4% paraformaldehyde and the left hemispheres were flash frozen in liquid nitrogen for biochemical analysis.
Immunostaining mouse brains
The paraformaldehyde fixed right hemispheres for different age mice or from dasatinib treated and control mice were cut using a freezing microtome. Briefly, paraformaldehyde fixed tissue was embedded in a 15% gelatin (in 0.1 M phosphate buffer) matrix and immersed in a 4% paraformaldehyde solution for two days to harden the gelatin matrix. The blocks were then cryoprotected through three cycles of 30% sucrose for three to four days each. The blocks were then flash frozen using dry ice/isomethylpentane, and serial 40 μm sections were cut using a freezing microtome. Serial sections were used for immunostaining with anti- pTyr (4G10) antibody at a dilution of 1:1,000, anti-Aβ (4G8) and anti-CD68 at a dilution of 1:500 to 1:1,000, anti-pSrc as 1:250, anti-Iba1 at 1:1,000, and anti-GFAP antibody at a dilution of 1:1,000, followed by incubation with biotinylated secondary antibodies (1:2,000 dilution) (Vector Laboratories Inc.) and avidin/biotin solution (Vector ABC kit). The immunoreactivity was observed using Vector VIP as chromogen. For pSrc/CD68 double staining, FITC and Texas Red conjugated secondary antibodies were used. For phosphotyrosine/CD68 or Iba1 double-labeling, the sections were first immunostained with anti-phosphotyrosine antibody 4G10 using Vector VIP as the chromogen. The tissue was then incubated in 0.2 N HCl to strip off antibodies then the tissue was double-labeled with either CD68 or Iba1 using Vector Blue as the chromogen. The slides were dehydrated and cover slipped using VectaMount (Vector Laboratories, Inc.) following a standard dehydrating procedure through a series of ethanol solutions and Histo-Clear (National Diagnostics, Atlanta, GA, USA). Images were taken using an upright Leica DM1000 microscope and Leica DF320 digital camera system (Leica Microsystems Inc., Buffalo Grove, IL, USA). Figures were made using Adobe Photoshop 7.0 software (Adobe Systems, San Jose, CA). For quantitation purposes, 1.25X images were taken from three consecutive serial sections, (960 μm apart) throughout the hippocampal region. Optical densities from the temporal cortex or CA1 regions from the same serial sections were measured using Adobe Photoshop software. All sections were immunostained simultaneously to minimize variability and background values in an unstained area of tissue for each section were set to zero using the curve tool before quantifying optical density values. The optical density of the entire temporal cortex region/CA1 region from a representative section was selected via marquee and the same size marquee was applied to all sections per condition to allow comparison of optical densities (O.D.) independent of area changes. The values for each section were averaged (three sections/brain, five to seven brains per condition) and plotted for Aβ, immunoreactivity for dasatinib infusion animals, pSrc immunoreactivities of dastinib treated mice, and Aβ and CD68 immunoreactivities for longitudinal study animals. For quantitating phospho-tyrosine immunostaining from different aged APP/PS1 and wild type mice, the serial sections were viewed under a microscope and the number of 4G10 positive plaques were viewed from the entire CA1 and temporal cortex regions for all the animals in each condition. The numbers of plaques were averaged (three sections/brain, five to seven brains per condition) and plotted. To insure reliability in comparison, only immunostains that were processed together were quantitatively compared to minimize any variability in staining processing from day-to-day. This allows an accurate assessment of relative comparisons within parallel processed conditions and samples although not necessarily a reflection of absolute values.
Western blot analyses of mouse brains
Hippocampus and temporal cortex were removed from flash frozen brains of treated mice, lysed, sonicated in RIPA buffer and quantitated using the Bradford method [
55]. The lysates were resolved using a custom-built 28-well comb and 10% SDS-PAGE and transferred to polyvinylidene difluoride membranes for Western blotting using anti-pTyr (4G10), anti-APP, anti-Aβ (6E10), anti-GFAP, anti-TNF-α and anti-CD68 antibodies with α-tubulin as their loading control, anti-pSrc (Tyr416), anti-pLyn (Tyr396) antibodies with anti-Src and anti-Lyn as their respective loading controls and anti-PSD95 and anti-synaptophysin with βIII-tubulin as the loading control. Antibody binding was detected using enhanced chemiluminescence for detection. Western blots were quantified using Adobe Photoshop software. Optical density (O.D.) of bands were normalized against their respective loading controls and averaged (+/−SEM).
T-maze
T-maze analysis was performed as described by Wenk, 1998 [
56]. Briefly, upon completion of the
in vivo infusion period, control and treated mice were placed into the starting arm, and the door was raised to allow animals to walk down the stem and choose an arm. Once the mice entered an arm with all four feet, they were returned to the starting arm and the door was closed. After 30 sec, the door was opened and the mice were allowed to choose an arm again. The process was repeated for nine trials with a 30 sec interval between each trial. The choice of arms was noted each time and the number of alternations between trials for each mouse was averaged and plotted.
Statistical analysis
Data are presented as mean +/− standard deviation. Values statistically different from controls were determined using one-way ANOVA or Student t-test when appropriate. The Tukey-Kramer multiple comparisons post-hoc test was used to determine P-values.
Discussion
Our findings demonstrated that Aβ fibrils stimulate microglia activation in vitro via a non-receptor tyrosine kinase, Src, associated pathway that results in increased secretion of the proinflammatory cytokine, TNF-α. It was possible to attenuate the Aβ-stimulated microglial phenotype change using a dual Src/Abl inhibitor, dasatinib. The in vitro observations were validated in vivo demonstrating that subcutaneous infusion of dasatinib into 13 month old APP/PS1 transgenic mice attenuated overall tyrosine phosphorylation and active Src levels, in particular, in the hippocampus. The drug did not affect Aβ plaque load but reduced microgliosis and TNF-α levels in these animals without altering synaptic markers in neurons. Moreover, dasatinib provided a significant increase in cognitive performance in correlation with this anti-inflammatory action. Collectively, these findings support the idea that Aβ fibrils can serve as a microglial activating ligand in disease contributing to their proinflammatory phenotype and use of selective non-receptor tyrosine kinase inhibitors is an effective strategy to limit microglial-mediated changes during disease.
Although it has been suggested that Aβ-stimulated microglial activation contributes to the pathophysiology of AD [
24]‐[
26,
28]‐[
30,
33,
67] and a broad range of microglial secreted inflammatory markers are elevated in AD brains, including IL-1α, IL-1β, TGF-β and TNF-α [
68,
69], enthusiasm for an anti-inflammatory approach to treating AD has decreased, in part, due to lack of drug efficacy of a number of human trials that targeted cyclooxygenase (Cox) activity in AD patients [
70]‐[
74]. Indeed, Cox inhibition during later stages of disease had adverse effects and only demonstrated protection when administered long-term to asymptomatic individuals [
37]. One possibility for the lack of efficacy of Cox inhibitors is the fact that Cox enzymes are expressed by multiple cell types in the brain and general drug inhibition has no cellular selectivity. Another possibility for the failed efficacy is that Cox 1 or 2 enzyme activities are simply not relevant targets for attenuating microglia-dependent changes. For this reason we focused on the direct signaling response initiated in microglia upon Aβ fibril stimulation. It has been reported from both AD brains [
75]‐[
77] and mouse models [
13,
78] that elevated protein phosphotyrosine levels are reliable markers of reactive microglia associated with plaques.
In vitro studies using monocytic lineage cells [
24,
44,
47] and microglia [
38,
41] have supported these data by demonstrating that fibrillar Aβ stimulates a specific increase in overall protein tyrosine phosphorylation. Based upon our
in vitro data, we targeted Src as a key enzyme activated downstream of Aβ fibril stimulation and demonstrated that a clinically available drug, dasatinib, was able to improve cognitive function while attenuating microglial activation and active Src levels in these cells. Importantly, this anti-inflammatory effect did not adversely affect Aβ plaque load in the mice. Therefore, a directed anti-inflammatory strategy targeting the particular enzymes involved in Aβ-stimulated microgliosis may be more relevant than broad-based Cox inhibition for testing during disease. Moreover, the fact that this tyrosine kinase inhibition strategy was effective even during advanced stages of disease in the mice suggests that this particular form of anti-inflammatory therapy is viable during advanced disease in contrast to Cox inhibition.
We are aware that the effect of decreasing microglial active Src and brain TNFα levels does not necessarily prove that these changes were responsible for the improved cognitive performance. However, our in vitro data clearly demonstrated that dasatinib treatment and Src inhibition led to attenuated TNFα secretion providing correlative evidence that Src inhibition in microglia in vivo contributed to the decrease in TNFα observed. Moreover, recent human data using TNFα neutralizing drugs demonstrated cognitive improvement in AD patients [
79] suggesting that diminished TNFα levels in the mice could have contributed to the cognitive improvements observed.
We also appreciate that dasatinib treatment may affect a number of other kinases
in vitro and
in vivo, and numerous cells express Src family kinases and Abl. Non-receptor Src family tyrosine kinases are expressed widely in the mammalian CNS and are known to play a role in proliferation and differentiation of the CNS [
80]‐[
89]. Indeed, Src family kinase activities are crucial for synaptic plasticity, including learning and memory [
90]‐[
94]. Additionally, there is compelling evidence that neuronal Abl activity can also mediate microgliosis
in vivo, suggesting that dasatinib may also work through this mechanism to exert its anti-inflammatory effects [
60]. It is also intriguing that Abl is able to phosphorylate tau protein on Tyr 394 identified from AD brains as well as the intracellular domain of APP to modulate signaling responses [
95]‐[
97]. Although we have not focused on Abl expression or activity in this study we appreciate that activity of this kinase is of great interest to not only the field of AD but myriad conditions in which oxidative stress-associated neuron death is involved, including Parkinson’s disease [
60,
96,
98]‐[
102]. However, this interest in Abl activity to regulate cell death [
60], parkin phosphorylation [
99,
100] and tau phosphorylation [
98,
101,
102] is all based upon neuronal expression of the kinase. Indeed, there is no reported expression of Abl in microglia, to the best of our knowledge. We expect that dasatinib actions in the brain will certainly include inhibition of not only microglial Src activity but additional Src family members expressed in microglia and other cells as well as Abl, which will have a broader target base than only a single kinase activity in microglia. This may have additional therapeutic benefits of not only anti-inflammatory actions on microglia but also direct neuroprotection. We do not rule out that some of the changes we observed are not also due to inhibition of Src family kinases or other non-receptor kinases, including Abl in our experiments. Certainly, our staining demonstrated additional vasculature phospho-tyrosine and pSrc immunoreactivity outside of the microglial immunoreactivity. Therefore, any strategy to manipulate activity of these tyrosine kinases in the brain should be carefully considered with regard to particular cellular targets. By focusing on microglia-Aβ interaction and demonstrating specificity of dasatinib for Src versus the related family member Lyn
in vivo as well as a clear improvement in cognitive performance, we suggest that reagents, such as dasatinib, at least be considered for anti-inflammatory human testing but, more importantly, for further drug development.
Our study intentionally focused on animals at 13 months of age with established plaques and reactive microglia to test the efficacy of our anti-inflammatory strategy in late-stage disease. However, it will be important in future work to determine if a strategy of kinase inhibition can attenuate or delay behavioral decline or microgliosis in earlier stage disease. Although our longitudinal assessment in these mice suggested that phosphotyrosine immunoreactive microglia correlated with increased plaque deposition, we have observed in earlier work that soluble oligomeric forms of Aβ are also potent stimuli of microglia responsible for initiating a unique type of tyrosine kinase-based signaling response [
38]. Therefore, fully determining the specific signaling pathways involved in different forms of Aβ stimulation of microglia may offer a strategy for inhibiting specific tyrosine kinase activities at different disease time points to maximally produce anti-inflammatory effects.
Conclusions
These data demonstrate that the mechanism underlying amyloid-dependent microgliosis in AD may involve increased Src family kinase activity. We targeted this specific signaling response with dasatinib, a dual Src/Abl inhibitor used for treatment of chronic myeloid leukemia [
103,
104] For the first time, we have found that dasatinib treatment not only attenuated microgliosis, TNFα levels, and active Src levels in the brains of APP/PS1 mice but also improved cognitive performance. This suggests that targeting the specific enzymes involved in Aβ-stimulated microglial activation may be efficacious therapeutically even during late stages of disease.
AD, Alzheimer’s Disease; Aβ, amyloid beta; APP, amyloid precursor protein; GFAP, glial fibrillary acidic protein; Iba1, ionized calcium binding adaptor molecule 1; PSD95, postsynaptic density protein 95; TNFα, tumor necrosis factor α.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
GD performed the majority of experiments and data analysis, and wrote the initial version of the manuscript. CC was involved in conceiving the study, performing a portion of the experiments and analysis, and coordinating the experiments. He was responsible for editing and revising the final version of the manuscript. All authors read and approved the final manuscript.