Loss of c-Jun N-terminal kinase-interacting protein-1 does not affect axonal transport of the amyloid precursor protein or Aβ production

Disruption to axonal transport is an early pathological feature in Alzheimer's disease. The amyloid precursor protein (APP) is a key axonal transport cargo in Alzheimer's disease since perturbation of its transport increases APP processing and production of amyloid-β peptide (Aβ) that is deposited in the brains of Alzheimer's disease patients. APP is transported anterogradely through axons on kinesin-1 motors. One favoured route for attachment of APP to kinesin-1 involves the scaffolding protein c-Jun N-terminal kinase-interacting protein-1 (JIP1), which has been shown to bind both APP and kinesin-1 light chain (KLC). However, direct experimental evidence to support a role of JIP1 in APP transport is lacking. Notably, the effect of loss of JIP1 on movement of APP through axons of living neurons, and the impact of such loss on APP processing and Aβ production has not been reported. To address these issues, we monitored how siRNA mediated loss of JIP1 influenced transport of enhanced green fluorescent protein (EGFP)-tagged APP through axons and production of endogenous Aβ in living neurons. Surprisingly, we found that knockdown of JIP1 did not affect either APP transport or Aβ production. These results have important implications for our understanding of APP trafficking in Alzheimer's disease.


INTRODUCTION
The correct transport of protein and organelle cargoes through axons (axonal transport) is an essential requirement for proper functioning of neurons. Since most proteins are synthesized in neuronal cell bodies, axonal and synaptic cargoes need to be transported anterogradely through axons to their final destinations, whereas trophic factors and other cargoes are transported retrogradely from the synapse to the cell body. Indeed, perturbation of axonal transport is now known to contribute to the pathogenic process in some neurodegenerative diseases (see for reviews (1)(2)(3)(4)).
Alzheimer's disease is the most common form of dementia, but the molecular mechanisms that cause disease are not fully understood. However, altered metabolism of amyloid precursor protein (APP) leading to changes in production of amyloid-b peptide (Ab) is believed to be central to the disease process.
Ab is an approximately 40-42 amino acid peptide that is derived from APP by proteolytic processing and which is deposited in the brains of Alzheimer's disease patients within amyloid plaques. Generation of Ab involves processing by b-site APP cleaving enzyme-1 (BACE1) and g-secretase that cleave APP at the N-and C-termini, respectively, of the Ab sequence. In addition, APP is processed by aand g-secretase but since a-secretase cleaves within the Ab sequence, this route precludes Ab production (see for review (5)). A large body of evidence suggests that at least some forms of Ab have pathogenic properties (see for reviews (6,7)). As such, understanding the mechanisms that control APP processing and Ab production has major relevance to Alzheimer's disease research.
One route for altering APP processing involves changes to its transport through axons. Thus, experimental disruption of anterograde axonal transport of APP increases production of Ab † These two authors contributed equally to this study. ‡  (8 -10). Moreover, since Ab itself disrupts axonal transport (11 -17), one suggestion is that damage to axonal transport promotes a toxic cycle of events. In this scenario, disruption to APP transport increases Ab production which, in turn, induces more damage to axonal transport to further elevate Ab production (8). As such, APP represents a key axonal transport cargo in Alzheimer's disease.
APP is transported anterogradely through axons on kinesin-1 molecular motors (8,(18)(19)(20). Most functional kinesin-1 is a heterotetramer of two kinesin-1 motor proteins and two kinesin-1 light chains (KLCs). Kinesin-1 contains the motor domain, whereas the KLC is mainly involved in binding of cargoes (21). KLC is required for APP transport (8,22), but the mechanism by which APP attaches to KLC is not clear.
However, much of the data supporting a role of JIP1 in axonal transport and processing of APP involve analyses of fixed cells or cell lines overexpressing JIP1 and/or APP (27,32,35,36). Notably, the effect of loss of JIP1 on axonal transport of APP in living neurons and how such loss affects production of endogenous Ab have not been reported. Clearly, a proper analysis of the role of JIP1 in APP transport requires dynamic studies of APP movement. Here, we address these issues by analyses of neurons in which JIP1 expression is downregulated with siRNAs.

RESULTS
siRNA loss of JIP1 does not influence Ab production, APP processing at the aor b-secretase sites or phosphorylation of APP at threonine-668 We first tested the role of JIP1 in the production of endogenous Ab in rat cortical neurons. To do so, we utilized siRNAs to knock down JIP1. In the neurons, JIP1 migrated as two major species which is consistent with previous reports and is probably the result of differential phosphorylation since JIP1 is a known phosphoprotein (42,43). Two different siRNAs and a mixture of both siRNAs all reduced JIP1 to undetectable levels (Fig. 1A). However, Ab ELISAs revealed that this reduction in JIP1 did not affect secretion of either Ab (1 -40) or Ab(1 -42) into the media ( Fig. 1B and C). For comparison and as controls, we also treated the neurons with siRNAs for calsyntenin-1 and with the g-secretase inhibitor N-[N-(3,5-difluorophenacetyl)-1alanyl]-S-phenylglycine-t-butyl ester (DAPT). Loss of calsyntenin-1 has been shown to disrupt axonal transport of APP to increase Ab production, whereas DAPT inhibits g-secretase cleavage of APP to decrease Ab production (9,10,44). In agreement with these findings, siRNA knockdown of calsyntenin-1 using a pool of siRNAs previously shown to reduce calsyntenin-1 to levels ,5% of that seen in control neurons (9) increased whereas DAPT decreased production of Ab(1-40) and Ab(1-42) ( Fig. 1B and C).
We next enquired whether siRNA knockdown of JIP1 influenced APP expression or its phosphorylation on threonine-668. JIP1 has been reported to preferentially bind to APP phosphorylated on threonine-668 (32). Since the two individual and mix of both JIP1 siRNAs all reduced JIP1 levels to the same extent ( Fig. 1A), we utilized the siRNA mix for these experiments. siRNA loss of JIP1 did not alter total APP levels or the levels of APP phosphorylated on threonine-668 in the neurons ( Fig. 2A).
We also enquired whether siRNA loss of JIP1 influenced APP processing at the a-secretase and BACE1 sites. Processing of APP by a-secretase and BACE1 generates N-terminal secreted APP fragments (sAPPa and sAPPb), which can be detected in conditioned media from cells using antibodies that specifically recognize each sAPP fragment. Immunoblotting of conditioned media from control or JIP1 siRNA knockdown neurons with antibody 22C11 (that detects total sAPP) and antibodies that detect sAPPa and sAPPb revealed that loss of JIP1 did not influence the levels of total sAPP, sAPPa or sAPPb ( Fig. 2A). These findings are consistent with the Ab assays, which also suggest that JIP1 knockdown does not affect APP processing.
Finally, we enquired whether loss of JIP1 influenced calsyntenin-1 levels in the neurons since APP is transported through axons by kinesin-1 on calsyntenin-1 carriers and loss of calsyntenin-1 affects APP processing (9,10,25). However, loss of JIP1 had no discernible effect on calsyntenin-1 levels ( Fig. 2A). Thus, loss of JIP1 in neurons does not influence endogenous APP levels, phosphorylation of endogenous APP at threonine-668, processing of endogenous APP at the a-secretase or BACE1 sites or production of endogenous Ab.
siRNA loss of JIP1 does not noticeably influence axonal transport of APP To examine the role of JIP1 in the axonal transport of APP, we quantified movement of enhanced green fluorescent protein (EGFP)-tagged APP (APP-EGFP) through axons of transfected living rat cortical neurons following siRNA knockdown of JIP1 using time-lapse microscopy. To avoid any possible artifacts produced by high levels of APP-EGFP expression, we chose for study, cells expressing low levels of APP-EGFP as judged by brightness of the EGFP signal. This approach is line with our previous studies (9,29,45). Kymographs were generated from the time-lapse movies, and these were used to calculate the velocities and flux rates of APP-EGFP movement in both anterograde and retrograde directions. Flux rates were determined by counting the numbers of APP-EGFP particles that crossed a defined line within the mid-section of axons as previously described by us and others (8,9,46). In control siRNA treated neurons, APP-EGFP moved through axons in a predominantly anterograde direction at fast speeds of between 1-9 mm/s and 1-5 mm/s in the anterograde and retrograde directions, respectively. These velocities and the bias towards anterograde movement are similar to those described previously for movement of fluorescent protein-tagged APP in neurons (8,9,18,19,(46)(47)(48). siRNA knockdown of JIP1 did not significantly alter the velocities of APP-EGFP movement in either anterograde or retrograde directions, the bias towards anterograde movement, or the flux rates of anterograde or retrograde movements (Fig. 3) (and see Supplementary data). Thus, loss of JIP1 has no noticeable effect on axonal transport of APP through axons of living rat cortical neurons.

The levels of JIP1 are not altered in Alzheimer's disease brains
To begin to determine whether JIP1 is altered in Alzheimer's disease, we monitored the levels of JIP1 protein in control and Braak stage V-VI Alzheimer's cortical brain samples (see Table 1, for details of patient samples). The age at death and postmortem delay was not significantly different between the control and Alzheimer's disease samples. However, we found no differences in the levels of JIP1 between the two sample sets (Fig. 4).

DISCUSSION
APP is transported anterogradely through axons by kinesin-1 motors (8,(18)(19)(20), but the mechanisms that mediate attachment of APP to kinesin-1 are not properly understood. JIP1 is a ligand for KLC and via its PTB domain also binds to the C-terminus of APP (27,32,36 -41). As such, JIP1 has been favoured as a scaffold to connect APP to KLC/kinesin-1 for anterograde axonal transport (27,32,35,36). Here, we directly tested the role of JIP1 in axonal transport of APP by quantifying APP-EGFP movement in JIP1 siRNA knockdown neurons. Despite highly efficient knockdown of JIP1, we detected no changes in APP movement in either anterograde or retrograde directions following loss of JIP1.
Disruption to axonal transport of APP influences its processing and production of Ab (8 -10). Others have shown that overexpression of JIP1 decreases Ab production but these studies were conducted in non-neuronal or neuroblastoma cells overexpressing APP (27,41). Our analyses revealed no effect of loss of JIP1 on endogenous Ab production in bona fide neurons, and these findings are consistent with the absence of an effect of JIP1 loss on axonal transport of APP in the same neurons. Of course, it is possible that JIP1 facilitates axonal transport of a proportion of APP on kinesin-1 motors but that in the JIP1 siRNA knockdown neurons, this effect is not noticeable due to switching of APP transport to other kinesin-1 carriers, such as those containing calsyntenin-1 so as to rescue any phenotype. Calsyntenin-1 has recently been shown to mediate transport of APP by kinesin-1 (9,10). Thus, while our results cannot formally eliminate any role of JIP1 in kinesin-1 transport of APP, the absence of any noticeable effect of loss of JIP1 on APP movement, APP processing or Ab production together argue against a major role for JIP1 in axonal transport of APP at least in the cortical neurons utilized here.
In contrast to an absence of effect of JIP1, siRNA loss of calsyntenin-1 reduces axonal transport of APP to increase Ab production (9,10) (and see Fig. 1). Calsyntenin-1 is a type-1 membranespanning protein that attaches via its intracellular C-terminus to KLC (26)(27)(28). Calsyntenin-1 also binds to the adaptor protein X11b (also known as X11-like and munc-18-interacting protein-2) (49). There is evidence that X11b functions as a coat protein to direct loading of cargoes into particular vesicle sub-types (50). X11b also binds to the C-terminus of APP (51,52) and like calsyntenin-1, modulating X11b expression affects APP processing and Ab production (53)(54)(55). Thus, X11b may load APP onto calsyntenin-1 containing vesicles for post Golgi transport by kinesin-1 through axons and disruption to this transport either via loss of calsyntenin-1 or X11b may promote amyloidogenic processing of APP. JIP1 has been reported to be mislocalised in Alzheimer's disease (56) but we detected no changes in the total levels of JIP1 in Alzheimer's disease brains. This contrasts with calsyntenin-1 where both decreased protein levels and altered processing have been reported (9,57,58). Thus, defective axonal transport of APP in Alzheimer's disease may be more closely linked to defects in calsyntenin-1 rather than JIP1 mediated cargo transport.

Cell culture and transfection
Cortical neurons were obtained from embryonic day 18 rat embryos and cultured in Neurobasal medium containing B27 supplement, 100 IU/ml penicillin, 100 mg/ml streptomycin and 2 mM L-glutamine (Invitrogen). For siRNA studies, neurons at DIV4 were treated with 1 mM of each siRNA (control, JIP1 or calsyntenin-1) for 96 h and then analysed at DIV8. For transfection studies, neurons were cultured on poly-L-lysine-coated glass cover slips in 12-well plates. Neurons were transfected on a magnetic plate for 30 min using 1 mg plasmid DNA and 2 ml magnetic nanoparticles (NeuroMag, OZ Biosciences) as described by the manufacturer. Neurons were transfected on DIV7 and analysed on DIV8.

Measurement of Ab production
To determine how loss of JIP1 affects production of endogenous Ab (1 -40) and Ab , conditioned medium from rat cortical neurons at DIV4 cultured in 24-well plates (2.5 × 10 5 cells per well) was removed and replaced with 400 ml fresh Neurobasal/ B27 medium containing 1 mM control, JIP1 or calsyntenin-1 siRNAs. The neurons were then returned to the incubator for a further 96 h and the conditioned media then assayed for Ab (1 -40) and Ab (1 -42). Some neurons were also treated with 2 mM of the g-secretase inhibitor DAPT. For Ab assays, the conditioned media was removed and centrifuged at 100000 × g for 30 min at 48C to remove any insoluble material. Levels of Ab (1 -40)

SDS-PAGE and immunoblotting
Cells were harvested for sodium dodecyl sulphate -polyacrylamide gel electrophoresis (SDS -PAGE) by washing with phosphate buffered saline pre-warmed at 378C and scraping into SDS-PAGE sample buffer and immediately heating to 1008C. Total sAPP, sAPPa and sAPPb levels in conditioned media were likewise prepared for SDS-PAGE by addition of sample buffer at heating to 1008C. Human brain samples were prepared as 20% homogenates and equal protein concentrations then prepared for SDS-PAGE by addition of SDS -PAGE sample buffer and heating to 1008C. Samples were separated on either 8 or 10% (w/v) acrylamide gels; separated proteins were then transferred to Protran nitrocellulose membranes (Schleicher & Schuell) using a Transblot system (BioRad) and following blocking, probed with primary antibodies. Following washing, the blots were further probed with horseradish peroxidase-conjugated goat anti-mouse or antirabbit Igs and developed using an enhanced chemiluminescence system (GE Healthcare).

Human tissues
All human tissue collection and processing were carried out under the regulations and licensing of the Human Tissue Authority and in accordance with the Human Tissue Act, 2004.

Time-lapse microscopy
Live microscopy of APP-EGFP was performed with an Axiovert S100 microscope (Zeiss) equipped with a Lambda LS Xenon-Arc light source (Sutter Instrument Company, Novato, USA), GFP filter set (Chroma Technology Corp., Rockingham, USA), 100× Plan-Apochromat 1.4 N.A. objective (Zeiss), Lambda 10-3 filter wheel (Sutter Instrument Co.) and a Photometrics Cascade-II 512B High Speed EMCCD camera (Photometrics, Tuscon, USA). Twenty-four hour post-transfection, the cells were transferred to a Ludin imaging chamber (Life Imaging Systems, Basel, Switzerland) mounted on the stage of the microscope. Neurons were maintained in a HEPES-buffered extracellular neuronal solution (composition in mM: naCl, 140; KCl, 5; NaHCO 3 , 5; MgCl 2 .6H 2 0, 1; CaCl 2 , 1.2; Na 2 HPO 4 , 1.2; Glucose, 10; Hepes, 20; pH 7.4) at 378C using an objective heater (IntraCell, Royston, UK) and 'The Box' Microscope Temperature Control System (Life Imaging Systems). Vesicle movements were recorded for 2 min with a 1 s time-lapse interval using MetaMorph software (Molecular Devices). Image analysis was performed with ImageJ. The flux rates for APP-EGFP transport in both anterograde and retrograde directions were determined essentially as described (8,9,46). This involved quantifying the numbers of APP-EGFP particles that crossed a defined line in mid-axons per minute. APP-EGFP velocities were determined from kymographs, essentially as described for other cargoes (62). Statistical significance was determined by Student's t-test.