Tyrosine kinase inhibition increases functional parkin-Beclin-1 interaction and enhances amyloid clearance and cognitive performance

Tyrosine kinase inhibitors (TKIs) are effective therapies for leukaemia. Alzheimer is a neurodegenerative disease characterized by accumulation of β-amyloid (plaques) and hyper-phosphorylated Tau (tangles). Here we show that AD animals have high levels of insoluble parkin and decreased parkin-Beclin-1 interaction, while peripheral administration of TKIs, including Nilotinib and Bosutinib, increases soluble parkin leading to amyloid clearance and cognitive improvement. Blocking Beclin-1 expression with shRNA or parkin deletion prevents tyrosine kinase (TK) inhibition-induced amyloid clearance, suggesting that functional parkin-Beclin-1 interaction mediates amyloid degradation. Isolation of autophagic vacuoles (AVs) in AD mouse brain shows accumulation of parkin and amyloid, consistent with previous results in AD brains, while Bosutinib and Nilotinib increase parkin-Beclin-1 interaction and result in protein deposition in the lysosome. These data suggest that decreased parkin solubility impedes parkin-Beclin-1 interaction and amyloid clearance. We identified two FDA-approved anti-cancer drugs as potential treatment for AD. Two FDA-approved tyrosine kinase inhibitor drugs, Bosutinib and Nilotinib, are shown to ameliorate Alzheimer's disease pathology in mouse models by increasing soluble parkin and leading to amyloid clearance and cognitive improvement.

AD is an aging disorder characterized by deposition of extracellular b-amyloid (Ab) plaques and intraneuronal tangles
Beclin-1 levels were increased (144 AE 37, mean AE sd, p < 0.02, N ¼ 7) relative to actin in parkin À/À mice compared to control ( Fig 1D, first blot) and in Tg-APP mice treated with Bosutinib (138 AE 42, mean AE sd, p ¼ 0.02) and Nilotinib (144 AE 43, mean AE sd, p ¼ 0.02) relative to actin (N ¼ 7). There was no difference in Beclin-1 between DMSO treated control and Tg-APP mice relative to actin. Beclin-1 was immunoprecipitated from wild type, parkin À/À and Tg-APP mice (Fig 1D,  A. Graph represents brain levels of TKIs over a 24 hr period after IP injection with Imatinib, Nilotinib, Bosutinib and DMSO (N ¼ 7). [Correction added after publication on 4 July 2013: The y axis of the graph shown in Fig. 1A has been corrected from "concentration (mM)" to "concentration (nM)"] B. WB in Tg-APP total brain lysates on 4-12% NuPAGE SDS gel (Invitrogen) show total Abl (top blot) T412 Abl (2 nd blot), soluble parkin level (3 rd blot), and LC3 (4 th blot) relative to MAP-2 (N ¼ 9). C. Quantitative ELISA showing soluble (STEN extract) and insoluble (4 M urea extract) mouse parkin in Tg-APP (N ¼ 9). Parkin À/À brains were used as a specificity control. D-F. (D) Blots represent immunoprecipitated Beclin-1 in mice probed with parkin antibody, and (E) control IgG in parallel with immunoprecipitates, (F) immunoprecipitated Beclin-1 probed with parkin and the reverse experiment in post-mortem human AD cortex analyzed on 4-12% SDS-NuPAGE gel. G,H. In situ proximity ligation assay (PLA) shows endogenous parkin-Beclin-1 complexes in (G) WT C57BL/6 mice (N ¼ 5) and (H) parkin À/À as control. I-K. PLA in Tg-APP mice IP injected once daily for 3 weeks with (I) DMSO (J) 5 mg/kg Bosutinib and (K) 10 mg/kg Nilotinib (N ¼ 5). L-Q. PLA in human post-mortem brains in the (L) cortex of a normal subject and (M) cortex of an AD patient; the hippocampus of (N) a normal subject and (O) an AD patient; the caudate of (P) a normal subject and (Q) an AD patient. R. Graph represents human Ab 1-42 ELISA in rat B35 neuroblastoma cells transfected with human cDNA Ab 1-42 (or LacZ) or Beclin-1 shRNA for 24 hr, and then treated with 1 mM Bosutinib for an additional 24 hr (N ¼ 12). Ã Significantly different to control or as indicated, Mean AE SEM, ANOVA with NeumannKeuls multiple comparison. Research Article www.embomolmed.org TKI ameliorates functional parkin-beclin-1 interaction input) and probed with parkin. As expected no Beclin-1-parkin interaction was detected in parkin À/À compared to wild type mice ( Fig 1D, second blot, N ¼ 5), and this interaction was decreased in Tg-APP mice treated with DMSO, but stronger parkin bands were detected with Bosutinib and Nilotinib, suggesting increased interaction between parkin and Beclin-1. Control experiments were conducted using Beclin-1 immunoprecipitation from brain lysates and analysed by WBs with anti-Beclin-1 (Fig 1E, top blot) and anti-parkin (bottom blot) antibodies along with IgG control and parkin À/À . HRP-secondary antibodies showed no bands in IgG lanes and no parkin in Parkin À/À mice, indicating specificity of Beclin-1 and parkin bands. Beclin-1 was also increased (Fig 1F,  161 AE 48, mean AE sd, N ¼ 5, p ¼ 0.01) in post-mortem AD cortex relative to tubulin compared to control subjects. Significantly decreased levels were detected when immunoprecipitated Beclin-1 was probed with parkin ( Fig 1F, 28 AE 6.5, mean AE sd, p < 0.03, N ¼ 5) and inversely when immunoprecipitated parkin was probed with Beclin-1 (46 AE 18, mean AE sd, N ¼ 5), suggesting decreased parkin-Beclin-1 interaction in AD.

Bosutinib decreases Ab levels and reduces plaques in AD models
Staining of 20 mm brain sections shows plaque formation in Tg-APP mice treated with DMSO (Fig 2A, B representing different animals, N ¼ 7), confirmed by thioflavin-S staining (Fig 2C), though plaque staining was reduced in the Bosutinib group after 3-week treatment (Fig 2D-F). Higher magnification shows endogenous parkin staining in Tg-APP ( Fig 2G) and plaque deposition (Fig 2H and I) in the hippocampus. Bosutinib increased endogenous parkin ( Fig 2J) and reduced plaque load (Fig 2K and L). Using different parkin antibodies to show parkin ( Fig 2M) and plaques ( Fig 2N and O), Bosutinib increased parkin levels ( Fig 2P) and reduced plaques (Fig 2Q and R) in the cortex. Quantification of plaque load in Tg-APP mice was performed by a blind investigator using ImageJ by drawing a line around individual plaques within 1 mm 2 radius of 6 randomly selected hippocampal and cortical regions in 6E10 stained slides (N ¼ 9). The average number of plaques was significantly reduced in Bosutinib ( Fig 2S, 64 AE 10, mean AE sd per mm 2 , p ¼ 0.02, N ¼ 9) compared to DMSO ( Fig 2S, 198 AE 49, mean AE sd per mm 2 , p ¼ 0.02, N ¼ 9) treated mice. WB analysis showed no difference between control and Tg-APP mice with either Bosutinib or Nilotinib and DMSO (WT data not shown) in the level of APP cleavage enzymes ( Fig 2T, N ¼ 4), including b-secretase (BACE-1, first blot), a-secretase (ADAM-10, second blot) and g-secretase (Presenilin-1, third blot), suggesting that TKI-induced decrease in Ab levels is unlikely to be mediated by changes of expression of APP-cleaving secretases. No change in total APP was detected but CTFs were decreased ( Fig 2U, 74 AE 21, mean AE sd, p ¼ 0.04, N ¼ 9) in Tg-APP þ Bosutinib compared to DMSO (100 AE 36, mean AE sd). We further examined whether the reduction of Ab level in Tg-APP mice was due to clearance mechanisms via lentiviral injection of Ab 1-42 that is not derived from APP cleavage (Fig 4). Bosutinib also decreased phospho-tyrosine proteins level (Fig 2U), indicating that it is not a specific Src-Abl inhibitor.  . Chronic treatment with TKI alters brain amyloid level. Ã Significantly different to control, Mean AE SEM, ANOVA with Neumann Keuls multiple comparison, p < 0.05. A,B. Graphs represent ELISA of human Ab 1-42 in (A) brain levels and (B) blood Ab 1-42 levels in 8-month old Tg-APP mice injected I.P. every other day for 6 weeks with Bosutinib or Nilotinib (N ¼ 10). C-E. ELISA of human soluble and insoluble brain (C) Ab 1-42 and (D) Ab 1-40 levels as well as (E) mouse p-Tau levels in 8-month old Tg-APP mice injected IP daily for 3 weeks with 5 mg/kg Bosutinib (N ¼ 9). F.

Research Article
www.embomolmed.org TKI ameliorates functional parkin-beclin-1 interaction ELISA of human soluble and insoluble Ab 1-42 at 6 weeks post-injection of lentiviral Ab 1-42 and daily treatment with 5 mg/kg Bosutinib (N ¼ 9) for 3 weeks. Ã Significantly different to control or as indicated, Mean AE SEM, ANOVA with Neumann Keuls multiple comparison.
F. WB on 4-12% NuPAGE SDS gel of total brain lysates in 1 year old wild type and parkin À/À mice treated with 5 mg/kg Bosutinib for 3 weeks on 4-12% NuPAGE SDS gel (Invitrogen) showing parkin (1 st blot)total Abl (2 nd blot), T412 Abl (3 rd blot), Beclin-1 (4th blot) and LC3 (5 th blot) relative to actin (N ¼ 7).  Chronic treatment with TKI alters brain amyloid level TKIs are pleiotropic drugs that affect a wide range of phosphotyrosine proteins; therefore, a lower dose may prevent some of the off-site effects of the drugs. IP injection every second day for 6 weeks resulted in significant reductions in brain Ab 1-42 levels in 8-month old Tg-APP mice with 5 mg/kg (154 ng/mL) or 1 mg/kg (162 ng/mL) Bosutinib compared to DMSO (207 ng/mL) treated mice (Fig 3A, p < 0.03, N ¼ 10). Nilotinib (5 mg/kg) also reduced Ab 1-42 (131 ng/mL) in the brain (Fig 3A, p ¼ 0.03, N ¼ 10) but lower dose (1 mg/kg) did not have any effects. High levels of Ab 1-42 (175 ng/mL) was detected in the blood (Fig 3B, p ¼ 0.04, N ¼ 10) in Tg-APP mice compared to control, but drug treatment did not change this level. However, daily 5 mg/kg Bosutinib injection for 3 weeks (Fig 3C)  ) and elimination of serine-199 (AT8) p-Tau relative to total Tau compared to DMSO (100 AE 31, mean AE sd, p ¼ 0.0001). No phosphotyrosine Tau was detected with the commercially available antibody (4G10, Millipore) and immunoprecipitation of total Tau (Tau-5 antibody) and probing with total phospho-tyrosine did not show any difference between control and Tg-APP mice (data not shown), suggesting that Tau phosphorylation at Ser and Thr may affect phosphorylation at tyrosine residues at later stages of Tau pathology.

DISCUSSION
These studies evaluated the effects of TKI on parkin-Beclin-1 interaction and modulation of autophagic amyloid clearance in AD models. Here we show novel mechanisms of parkin-Beclin-1 3 Figure

Research Article
www.embomolmed.org TKI ameliorates functional parkin-beclin-1 interaction interaction, which is dependent on parkin stability as increased levels of insoluble parkin in AD and Tg-APP mice lead to loss of parkin-Beclin-1 interaction, perhaps impairing autophagic amyloid clearance. These data, along with the identification of brain penetrant FDA-approved drugs, are novel mechanistic and translational findings. These results demonstrate the impact of decreased parkin solubility (Lonskaya et al, 2012b;Lonskaya et al, 2013), which co-localizes with intraneuronal Ab 1-42 in post-mortem AD brains (Lonskaya et al, 2012c), suggesting failure to facilitate amyloid clearance. We previously reported that exogenous parkin mediates autophagic clearance (Burns et al, 2009;Khandelwal et al, 2011;Lonskaya et al, 2012b;Lonskaya et al, 2013;Rebeck et al, 2010) and here we delineate the mechanisms related to parkin function via interaction with Beclin-1 to facilitate autophagosome maturation (Lonskaya et al, 2012b;Lonskaya et al, 2013), suggesting that parkin stability affects its protein clearance ability. Although the effects of osmotic pump delivery of TKIs on microgliosis (Dhawan & Combs, 2012), Ab pathology (Cancino et al, 2008) and parkin relationship with amyloid accumulation (Perucho et al, 2010) were previously reported in AD models, our results identified novel mechanisms involving TKI-mediated autophagic clearance of intraneuronal Ab and Tau and demonstrated the effects of brain-penetrant TKIs (Bosutinib and Nilotinib) in improving amyloid pathology and cognition. These novel findings potentially have high medical impact due to lack of effective drug treatment for AD and other neurodegenerative diseases, involving intraneuronal accumulation of proteins, including the Tauopathies and a-Synucleinopathies. Additionally, penetration of well tolerated TKIs into the brain to clear intraneuronal amyloid and reduce plaque load, contrasts with anti-Ab vaccine therapies that may reduce extracellular plaques but fail to rescue neurons from intracellular amyloid stress, leading to progression of cell death. Furthermore, the current findings show that TKImediated autophagy may reduce p-Tau, indicating that autophagy may clear free unbound p-Tau, which can potentially lead to toxic intracellular inclusions, and spare Tau that may be bound to microtubule. The decrease in p-Tau at serine and threonine residues may be due to increased autophagic clearance of this protein, but lack of detection of tyrosine phosphorylated Tau suggests that tyrosine phosphorylation of Tau may occur at a later stage of Tau pathology. These data provide TKI as a therapeutic strategy to reduce p-Tau in a number of human Tauopathies. Beclin-1 levels were reported to decrease in AD brain (Pickford et al, 2008), and autophagic defects result in amyloid accumulation due to lack of autophagosome clearance (Nixon & Yang, 2011). However, our results suggest that Beclin-1 levels are increased in AD brains, perhaps due to different stages of disease pathology and sample extraction between our studies and those reported by (Pickford et al, 2008), but this increase in Beclin-1 is not associated with interaction with parkin, whose solubility is decreased in AD (Lonskaya et al, 2012c). Our results show that blocking Beclin-1 expression or deleting parkin impairs amyloid clearance, while others showed that lentiviral Beclin-1 expression activates autophagy in AD models (Pickford et al, 2008). TKIs stimulate autophagy (Salomoni & Calabretta, 2009), and decrease the level of insoluble parkin, leading to amyloid clearance in a parkin-dependent manner. Mutations in the gene coding for the E3-ubiquitin ligase parkin (Park2) are associated with inherited PD (Kitada et al, 1998). Parkin solubility is affected with many non-familial PD-linked stressors, including MPpþ, rotenone, 6-hydroxydopamine and dopamine (Wang et al, 2005), and protein aggregates alter parkin solubility (Kawahara et al, 2008). Furthermore, decreased parkin solubility is associated with alteration of its activity via increased phosphorylation by several kinase activities, including Abl (Imam et al, 2011;Ko et al, 2010). Therefore, alteration in parkin solubility suggests that parkin activity is affected in AD.
TKs, including Abl, are activated in neurodegeneration (Imam et al, 2011;Jing et al, 2009;Ko et al, 2010;Tremblay et al, 2010) and may lead to alteration of parkin function (Imam et al, 2011;Ko et al, 2010). Therefore, TK activity may be manipulated to induce functional parkin-Beclin-1 interaction to stimulate autophagic clearance in neurodegeneration. Abl encodes a protein TK that is distributed in the nucleus and the cytoplasm of proliferating cells and is involved in a wide range of functions, including control of cell cycle and apoptosis (Wang, 2000). Ab activates Abl and induces p-Tau suggesting that Abl participates in Ab-induced p-Tau, while Imatinib reduces AT8 and PHF1 levels of p-Tau (Cancino et al, 2011) and reverses cognitive decline (Cancino et al, 2008;Cancino et al, 2011) in AD mice. These findings are in agreement with our data showing that TKI leads to clearance of amyloid proteins and improvement of cognitive function in AD models. Accumulation of autophagosomes in neurodegeneration may be due to reduced autophagic flux (Gonzalez-Polo et al, 2005). Parkin, Ab and p-Tau accumulate in AVs compared to lysosomes in Tg-APP and lentiviral Ab 1À42 expressing mice, while TKI increases protein levels in the lysosomes, indicating autophagic flux. Autophagosome accumulation could be due to lack of maturation, leading to inefficient fusion with lysosomes and decreased autophagic clearance (He & Klionsky, 2009;Iwata et al, 2005;Kovács et al, 1982). The accumulation of LC3-II in the brain of AD models suggests autophagic defects, including autophagosome accumulation, leading to decreased amyloid clearance. TKI decreases LC3-II levels and subcellular fractionation show a decrease in Ab and p-Tau in autophagosomal fractions (AV10 and AV20) and deposition in lysosomes, indicating that TKI facilitates autophagosome clearance. Additionally, amyloid clearance in the brain, but not blood, suggest that TKI targets intracellular amyloid. The lentiviral model confirms the role of parkin in autophagic degradation of intracellular Ab 1À42 , leading to decreased plaques, while parkin deletion leads to more plaque formation due to lack of parkinmediated clearance of intraneuronal Ab 1À42 . Failure of TKIs to alter the contents of AVs and deposit amyloid in the lysosome in parkin À/À mice suggest that functional parkin plays an essential role in autophagosome maturation, leading to lysosomal degradation (He & Klionsky, 2009;Iwata et al, 2005;Kovács et al, 1982). Parkin modulates Beclin-1-LC3 mediated autophagy (Chen et al, 2010) and loss of parkin function (T240R) is associated with lack of autophagosome clearance in a-Synuclein
Although no parkin mutations are found in AD, manipulation of parkin activity can be a disease modifying therapy that would provide an alternative approach to prevent progression from mild cognitive impairment (MCI) to AD. Nilotinib and Bosutinib are FDA approved and enter the brain at sufficient concentrations to inhibit TKs, therefore they may be re-purposed to treat MCI. TKIs are pleiotropic drugs used in late stage CML, but AD does not involve a single pathway that may be efficiently treated with a single drug, therefore a reduction of drug dose may prevent the pleiotropic effects of TKIs. Progression from MCI to AD is a slow neurodegenerative process, and Bosutinib and Nilotinib are effective in young and aged AD mice with a lower drug dose over a longer time period, providing some proof of concept that lower dose of TKIs may be useful to halt the slow progression from MCI to AD. Although TKIs do not accumulate in the brain longer than 8 (Nilotinib) to 12 h (Bosutinib), it is possible that these drugs stimulate autophagic clearance to clear amyloid proteins that have accumulated between different treatments. Despite the promising effects of TKI in AD models, several studies previously reduced amyloid plaques in transgenic mice, but failed to halt AD progression in humans. Therefore, phase II clinical trials are needed to demonstrate the efficacy of TKI on human pathology and dementia. This is a novel mechanism involving TKI enhancement of autophagic degradation of amyloid proteins in AD mice. This approach suggests that autophagic degradation of intracellular Ab and p-Tau can reduce extracellular plaques, leading to cognitive improvement.

Human postmortem brain tissues
Human postmortem samples were obtained from John's Hopkins University brain bank. Patients' description and sample preparation are summarized in (Lonskaya et al, 2012c). Data were analysed as mean AE SEM, using two-tailed t-test.

Stereotaxic injection
Lentiviral constructs encoding LacZ, or Ab 1-42 (Rebeck et al, 2010) were stereotaxically injected 1 Â 10 6 m.o.i bilaterally into the CA1 hippocampus of 1 year old C57BL/6 or parkin À/À . A Total of 6 mL lentiviral stocks were delivered at a rate of 0.2 mL/min and. All procedures were approved by the Georgetown University Animal Care and Use Committee (GUACUC).

TKI treatment
TKIs were dissolved in DMSO and a total volume of 30mL were IP injected once a day for 3 weeks. Half the animals received DMSO and the other half received TKI in DMSO.

Statistical analysis
All statistical analysis was performed using a GraphPad Prism, version 5.0 (GraphPad software, Inc, San Diego, CA). The number (N) indicates the number of independent experiments (cell culture) or number of individual animals. Asterisks designate significantly different to DMSO or as indicated, all data are presented with Mean AE SEM, with actual p-values obtained using ANOVA with Neumann Keuls multiple comparison.

Stereological methods
Stereological methods were applied by a blinded investigator using unbiased stereology analysis (Stereologer, Systems Planning and Analysis, Chester, MD) as described in (Lonskaya et al, 2012c;Rebeck et al, 2010).

Proximity ligation assay (PLA)
Primary 1:100 mouse anti-parkin (PRK8, above) and rabbit 1:100 anti-Beclin-1 (above) antibodies were applied to 20 mm thick sections of mouse brain or de-parrafanized PPE human brains overnight at 4°C. Duolink In Situ Red Starter Kit (Cat#92101-KI01) containing speciesspecific secondary antibodies or PLA probes, each with a unique short DNA strand attached to it (Axxora, LLC, Farmingdale, NW) was used as described in manufacturer's protocol. When the PLA probes are in close proximity, the DNA strands interact through a subsequent addition of two other circle-forming DNA oligonucleotides. After joining of the two added oligonucleotides by enzymatic ligation, they are amplified via rolling circle amplification using a polymerase to highlight the interaction. Fluorescence in each single-molecule amplification product is easily visible as a distinct bright spot when viewed with a fluorescence microscope.

Ab and p-Tau ELISA
Ab and p-Tau enzyme-linked immunosorbent assay (ELISA) using specific p-Tau, Ab 1-40 and Ab 1-42 ELISA and caspase-3 activity were performed according to manufacturer's protocol as described in (Lonskaya et al, 2012c;Rebeck et al, 2010).
Subcellular fractionation to isolate autophagic vacuoles 0.5 g of Frozen human or animal brains were homogenized at low speed (Cole-Palmer homogenizer, LabGen 7, 115 Vac) in 1ÂSTEN buffer and centrifuged at 1000 Â g for 10 min to isolate the supernatant from the pellet. The pellet was re-suspended in 1ÂSTEN buffer and centrifuged once to increase the recovery of lysosomes. The pooled supernatants were then centrifuged at 100,000 rpm for 1 h at 4°C to extract the pellet containing AVs and lysosomes. The pellet was then resuspended in 10 mL (0.33 g/mL) 50% Metrizamide and 10 mL in cellulose nitrate tubes. A discontinuous Metrizamide gradient was constructed in layers from bottom to top as follows: 6 mL of pellet suspension, 10 mL of 26%; 5 mL of 24%; 5 mL of 20% and 5 mL of 10% Metrizamide (Marzella et al, 1982). After centrifugation at 10,000 rpm for 1 h at 4°C, the fraction floating on the 10% layer (Lysosome) and the fractions banding at the 24%/20% (AV 20) and the 20%/10% (AV10) Metrizamide inter-phases were collected by a syringe and examined.

Cell culture and transfection
Human neuroblastoma M17 or rat B35 cells were grown in 24 well dishes (Falcon) as previously described (Lonskaya et al, 2012c;Rebeck et al, 2010). Transient transfection was performed with 3 mg Ab 1-42 cDNA, or 3 mg LacZ cDNA for 24 h. Cells were treated with 10 mM Nilotinib for 24 h. Cells were harvested 48 h after transfection. Cells were harvested one time with STEN buffer and centrifuged at 10,000 Â g for 20 min at 4°C, and the supernatant was collected.

Parkin ELISA
Parkin ELISA was performed on brain soluble brain lysates (in STEN buffer) or insoluble brain lysates (4 M urea) using mouse specific parkin kit (MYBioSource) in 50 mL (1 mg/mL) of brain lysates detected with 50 mL primary antibody (3 h) and 100 mL anti-rabbit antibody (30 min) at RT. Extracts were incubated with stabilized Chromogen for 30 min at RT and solution was stopped and read at 450 nm, according to manufacturer's protocol.

Morris water maze
All animals were pre-trained (trials) to swin for 90 s in a water maze containing a platform submerged in water (invisible) for 4 consecutive days once a day. The pretraining trials 'teach' the swimming animals that to 'escape' , they must find the hidden platform, and stay on it. The water maze 'test' was performed on day 5 (D' Hooge and De Deyn, 2001), when the platform was removed and mice have to swin and find it, thus assessing acquisition and retention. All parameters, including distance travelled to reach platform, speed to get to the platform, latency or time spent on platform and platform entry were digitally recorded on a computer and analysed by a blind investigator.

The paper explained
PROBLEM: Alzheimer's disease (AD) is an aging disorder that leads to memory loss. AD is characterized by intraneuronal tangles containing hyperphosphorylated Tau (p-Tau) and extracellular b-amyloid (Ab) plaques, derived from amyloid precursor protein (APP) cleavage and accumulation of Ab. Abl is found within neuritic plaques and neurofibrillary tangles (NFTs) and activated (via phosphorylation) in AD post-mortem brains. Src tyrosine kinase (TK) is also recognized in AD pathology via interaction with Tau. Abl inhibition prevents Ab 1-42 fibrils and hydrogen peroxide (H 2 O 2 )-induced cell death, and hippocampal injection of Ab fibrils leads to an increase of Abl levels. These data led to the hypothesis that tyrosine kinase inhibitors (TKIs) will activate parkin and facilitate autophagic amyloid clearance, thus preventing cognitive decline in AD models. We used several TKIs, including Bosutinib and Nilotinib, which penetrate the brain and induce autophagy in AD models and evaluated the effects of TKIs on parkin-mediated autophagic amyloid clearance in AD models.

RESULTS:
We identified two FDA-approved drugs, Bosutinib and Nilotinib, as potential therapies for AD. Our results indicate that decreased parkin solubility decreases functional interaction with a key autophagy molecule, Beclin-1, while TKIs reverse parkin-Beclin-1 interaction in AD models, leading to autophagic clearance of amyloid proteins. TKI-induced decrease of intraneuronal Ab 1-42 led to plaque disappearance while lentiviral production of intraneuronal Ab 1-42 resulted in plaque formation. Importantly, parkin is required for autophagosome maturation and autophagic clearance and TKIs enhance amyloid clearance and cognitive performance in a parkindependent manner. IMPACT: Our studies show that TKI could be a therapeutic strategy, via degradation of amyloid proteins, in neurodegenerative diseases, including AD. We identified two FDA-approved drugs as potential therapies for AD. TKIs are well tolerated in human leukaemia patients and could be used with a smaller dose over a longer period of time to prevent progression from mild cognitive impairment (MCI) to AD.

Quantification of plaque load
Quantification of plaque load or counting plaque number was performed by a blind investigator using ImageJ by drawing a line around individual plaques within 1 mm 2 radius of 6 randomly selected hippocampal and cortical regions in 6E10 stained slides. The number of plaques was averaged per mm 2 and compared between treatment conditions.

Immunoprecipitation
Mouse brains or human post-mortem tissues were homogenized in 1XSTEN buffer and the soluble fraction was isolated as indicated above. The lysates were pre-cleaned with immobilized recombinant protein G agarose (Pierce #20365), and centrifuged at 2500 Â g for 3 min at 4°C . The supernatant was recovered, and quantified by protein assay and a total of 100 mg protein was incubated for 1 h at 4°C with primary 1:100 mouse anti-parkin (PRK8, above) and rabbit 1:100 anti-Beclin-1 (above) antibodies in the presence of sepharose G and an IgG control with primary antibodies. Parkin À/À mouse brain lysates were also used to determine IP specificity. The immunoprecipiates were collected by centrifugation at 2500 Â g for 3 min at 4°C, washed 5Â in PBS, with spins of 3 min, 2500 Â g using detergent-free buffer for the last washing step and the proteins were eluted according to Pierce instructions (Pierce #20365). After IP, the samples were sizefractionated on 4-12% SDS-NuPAGE and transferred onto 20 mm nitrocellulose membranes. The primary antibodies used for WB analysis of the parkin and Beclin-1 were the same as those used for IP. WB detection of the parkin and Beclin-1 was then performed using either HRP conjugated secondary antibodies.

Author contributions
IL performed prepared lentivirus, IHC and Morris Water Maze, MLH performed ELISA, parkin activity assays and Western blots, NMD and AF performed WB. CE-HM injected the animals, oversaw the studies and wrote the manuscript.