1020 • The Journal of Neuroscience, January 18, 2012 • 32(3):1020 –1034
Neurobiology of Disease
Cdk5/p25-Induced Cytosolic PLA2-Mediated
Lysophosphatidylcholine Production Regulates
Neuroinflammation and Triggers Neurodegeneration
Jeyapriya R. Sundaram,1,2 Elizabeth S. Chan,1,3 Charlene P. Poore,1,3 Tej K. Pareek,6 Wei Fun Cheong,1,3
Guanghou Shui,1,3 Ning Tang,1,2 Chian-Ming Low,1,2,5 Markus R. Wenk,1,3,4 and Sashi Kesavapany1,3
1Neurobiology and Ageing Program, Centre for Life Sciences, Yong Loo Lin School of Medicine, and Departments of 2Pharmacology, 3Biochemistry,
4Biological Sciences, and 5Anesthesia, National University of Singapore, Singapore 117456, and 6Department of Pediatrics, Case Western Reserve
University, Cleveland, Ohio 44106
The deregulation of cyclin-dependent kinase 5 (Cdk5) by p25 has been shown to contribute to the pathogenesis in a number of neurode-
generative diseases such as amyotrophic lateral sclerosis (ALS), Parkinson’s disease (PD) and Alzheimer’s disease (AD). In particular,
p25/Cdk5 has been shown to produce hyperphosphorylated tau, neurofibrillary tangles as well as aberrant amyloid precursor protein
processing found in AD. Neuroinflammation has been observed alongside the pathogenic process in these neurodegenerative diseases,
however the precise mechanism behind the induction of neuroinflammation and the significance in the AD pathogenesis has not been
fully elucidated. In this report, we uncover a novel pathway for p25-induced neuroinflammation where p25 expression induces an early
trigger of neuroinflammation in vivo in mice. Lipidomic mass spectrometry, in vitro coculture and conditioned media transfer experi-
ments show that the soluble lipid mediator lysophosphatidylcholine (LPC) is released by p25 overexpressing neurons to initiate astro-
gliosis, neuroinflammation and subsequent neurodegeneration. Reverse transcriptase PCR and gene silencing experiments show that
cytosolic phospholipase 2 (cPLA2) is the key enzyme mediating the p25-induced LPC production and cPLA2 upregulation is critical in
triggering the p25-mediated inflammatory and neurodegenerative process. Together, our findings delineate a potential therapeutic
target for the reduction of neuroinflammation in neurodegenerative diseases including AD.
Introduction brain tissue of AD patients (Patrick et al., 1999; Tseng et al.,
2002). Although this has been debated by a number of groups
Neurodegenerative diseases are characterized by a progressive (Takashima et al., 2001; Bian et al., 2002), the proteolytic cleavage
loss of neurons in the CNS, in which affected neurons often dis- of p35 to p25, which has been shown to hyperactivate Cdk5 is
play characteristic hallmarks of a particular disease process. Alz- without doubt. Deregulation of Cdk5 activity, accompanied by
heimer’s disease (AD) is characterized by the deposition of the accumulation of p25 has been implicated as a causative factor
␤-amyloid plaques, accumulation of phospho-tau containing in the pathogenesis of neurodegenerative diseases (Nguyen et al.,
neurofibrillary tangles, reactive astrogliosis and loss of neurons 2001; Smith et al., 2006). Although few studies debated the in-
and synapses in selected areas of the brain (Trojanowski et al., volvement of p25/Cdk5 in neurodegeneration (Hallows et al.,
1993). Cdk5, a member of the cyclin-dependent kinase (Cdk) 2006), large bodies of evidence show the involvement of p25/
family of Ser/Thr kinases, is a master regulator of the cytoarchi- Cdk5 hyperactivation in the development of pathological hall-
tecture of the brain (Meyerson et al., 1991). The Cdk5 activator marks of AD (Otth et al., 2002; Town et al., 2002; Cruz et al., 2003;
protein, p35 can be cleaved to a smaller fragment p25, by the Noble et al., 2003; Lopes et al., 2007; Saito et al., 2007).
calcium activated protease, calpain. Cdk5/p25 has a mislocalized
activity as well as a longer half-life (Patrick et al., 1999). Increased Numerous neurodegenerative diseases share the common fea-
levels of p25 protein have also been found in the postmortem ture of neuroinflammation, which is a complex cascade of self-
defense responses to injurious stimuli in the CNS. Increasing
Received Oct. 14, 2011; revised Nov. 16, 2011; accepted Nov. 22, 2011. evidence has confirmed a major role of reactive astrocytes and
Author contributions: T.K.P., C.-M.L., M.R.W., and S.K. designed research; J.R.S., E.S.C., C.P.P., T.K.P., W.F.C., and microglia in the initiation and exacerbation of CNS inflamma-
tion (Markiewicz and Lukomska, 2006). Although, recent report
N.T. performed research; J.R.S., E.S.C., T.K.P., G.S., and S.K. analyzed data; J.R.S. and S.K. wrote the paper. suggests that there is a close association between neurodegenera-
This work was supported by National Medical Research Council Grants WBS 183-000-179-213 and 184-000-180- tion and p25-mediated neuroinflammation, the precise mecha-
nism behind the link is not fully described (Muyllaert et al.,
213, and Singapore National Research Foundation Competitive Research Programme Award No. 2007-04. We thank 2008). Hence our study aims to investigate the mechanism be-
Dr. Paul MacAry, Immunology Program, Life Sciences Institute, National University of Singapore, for the kind gift of hind the early events that trigger the cellular alterations in neu-
antibodies CD4 and CD8. We also thank Noor Hazim Bin Sulaimee for technical assistance. roinflammation leading to neurodegeneration in p25-expressing
Correspondence should be addressed to Dr. Sashi Kesavapany, National University of Singapore, 28 Medical
Drive, # 04-21, Singapore 117456. E-mail: firstname.lastname@example.org.
Copyright © 2012 the authors 0270-6474/12/321020-15$15.00/0
Sundaram et al. • LPC/cPLA2 Mediates p25-Induced Neuroinflammation J. Neurosci., January 18, 2012 • 32(3):1020 –1034 • 1021
In this report, we have elucidated a pathway of p25-induced cultured on 15 mm poly-L-lysine-coated coverslips and placed in 3.0 m
neuroinflammation where p25 expression, at very early time tissue culture inserts (Greiner bio-one) and glia were plated either on
points, induces neuroinflammation in vitro and in vivo in the coverslips (18 mm) or at the bottom of 6 well plates and incubated for
absence of any amyloid or tau pathology. We show that p25 over- 48 h. Alternatively, supernatants from EV-LV/p25-LV virus-transduced
expressing neurons release a soluble lipid factor, lysophosphati- cortical neurons or 5 d induced p25Tg/control mice neurons were col-
dylcholine (LPC), through the upregulation of cytosolic PLA2 lected and centrifuged at 14,000 ϫ g for 15 min. The cell-free superna-
(cPLA2) which causes astrogliosis and increased proinflamma- tants were then transferred onto glia and incubated for 48 h. The glia
tory cytokine production. We also show that the inflammatory from coculture or supernatant transfer experiments were fixed for im-
component produced during p25 overexpression could trigger munocytochemistry or processed for Western blot analyses.
later pathological changes in the neurodegenerative process. Our
report unveils cPLA2 as a potential therapeutic target in p25- Factor removal experiments. Cell-free supernatants from EV-LV/p25-
mediated neuroinflammation and neurodegeneration. LV-transduced cortical neurons or 5 d induced p25Tg/control mice neu-
rons were treated with DNase (8 g/ml), RNase (50 g/ml), and
Materials and Methods Proteinase K (50 g/ml) (Sigma) for 60 min at 37°C. The reactions were
stopped by boiling the supernatants at 95°C for 20 min. The enzyme-
Animal handling. All animal experimentation was performed according treated supernatants were then cooled down to 37°C and transferred to
to approved protocols by the Institutional Animal Care and Use Com- glia. To remove lipids, cell-free supernatants were passed through the
mittee of the National University of Singapore. solid phase extraction column (SPE-C18 column, Waters) under vac-
uum and eluted supernatants were transferred to glia. The samples were
Antibodies. Antibodies used for Western blot analyses were rabbit anti- collected for Western blotting and immunocytochemistry after 48 h.
GFAP (Sigma, 1:1000), rabbit C-8 (Santa Cruz Biotechnology, 1:500),
mouse monoclonal anti-cPLA2 (Santa Cruz Biotechnology, 1:200) Immunocytochemical analysis. Glial cells on coverslips (300,000 cell/18
and mouse anti-␣-tubulin (Sigma, 1:10,000). Secondary horseradish mm coverslip) were fixed with 4% formaldehyde for 20 min. The cells
peroxidase-conjugated antibodies (GE Healthcare) were used at 1:1000 were permeabilized with 0.1% Triton X-100 for 20 min and blocked with
dilutions. Antibodies used for immunohistochemistry were mouse anti- 5% FBS in PBS for 30 min. Cells were incubated with the primary anti-
GFAP (Sigma, 1:1000), mouse monoclonal AT8 (Pierce, 1:100), anti-␤- body in 5% FBS/PBS for 1 h at room temperature. The cells were washed
amyloid 1– 42 (Millipore, 1:100), anti-CD11b (Millipore, 1:200), rabbit three times with PBS followed by addition of the appropriate secondary
anti-mouse tPA (Molecular Innovations, 1:500), mouse anti-CD4 and antibody for 1 h at room temperature and nuclei were counterstained
anti-CD8 (BioLegend, 1:200). Secondary fluorescence-conjugated anti- with DAPI (Sigma). The coverslips were then washed with PBS and
bodies Alexa Fluor 488 and Alexa Fluor 594 (Invitrogen) were used at mounted on glass slides in fluorescence mounting medium (Dako). Flu-
1:200 dilutions. orescent confocal images were captured with a Zeiss LSM-510 laser-
scanning confocal microscope at 40ϫ magnification.
p25 transgenic mouse model. p25 transgenic (p25Tg) mice C57BL/6-Tg
(tetO-CDK5R1/GFP) 337Lht/J (The Jackson Laboratory) were mated Immunohistochemistry. Mice were perfused with 4% PFA/PBS and 16 m
with Camk2a transgenic mice B6; CBA-Tg (Camk2a-tTA) 1Mmay/J of brain cryosections were permeabilized, blocked and incubated with pri-
(The Jackson Laboratory) to generate bitransgenic offspring that induc- mary antibody overnight at 4°C in PBS with 5% FBS. Sections were washed
ibly overexpress human p25 gene under the control of the tetracycline- in PBS before incubation with secondary antibody for 1 h at room temper-
derivative, doxycycline. These mice were initially described by Cruz et al. ature and nuclei were stained with DAPI. The immunostained sections were
(2003). All mice in this study were conceived and raised in the presence of mounted on cover glasses in fluorescent mounting medium. Confocal im-
doxycycline (Sigma; 200 g/ml) in drinking water, for 6 weeks postnatal ages were taken at 20ϫ, 40ϫ, and 63ϫ magnifications.
to avert any potential developmental consequences from the expression
of p25. The expression of p25 was induced by removal of doxycycline and TUNEL assay. TUNEL staining was performed according to manufac-
brain samples were collected at different weeks of induction periods. turer’s instructions using In Situ Cell Death Detection Kit, TMR red
Littermates of either sex mice were used for the experiments. Same sex (Roche).
mice were used for comparison whenever possible.
Western blot analysis. Mouse brain lysates, total cell lysates and soluble
Mammalian cell culture. Primary mouse cortical neurons were cul- cell lysates were prepared as previously described (Kesavapany et al.,
tured as described previously (Kesavapany et al., 2004b). Briefly, mouse 2004a). Briefly, Samples were separated on 4 –20% (w/v) polyacrylamide
embryonic day 16 –18 (E16 –E18) cortical neurons were cultured in Neu- gels (Invitrogen) and transferred to nitrocellulose membranes and
robasal medium with B27 supplement (Invitrogen) containing 100 probed with primary antibodies overnight. The membranes were then
IU/ml penicillin, 100 g/ml streptomycin, and 2 mM glutamine. Primary washed in Tris-buffered saline–Tween 20 (TBST) and incubated with
glial cultures (mixed glial culture) using P0-P2 pups followed the iden- horseradish peroxidase-conjugated mouse or rabbit secondary antibod-
tical procedure and were cultured in DMEM supplemented with 2 mM ies for followed by three washes with TBST. Blots were developed using
penicillin streptomycin glutamine, 10% FBS and 10 mM sodium pyruvate enhanced chemiluminescence (GE Healthcare) according to the instruc-
at 37°C with 5% CO2. Human embryonic kidney (HEK) FT cell line was tions of the manufacturer.
obtained from the American Type Culture Collection and cultured in
DMEM supplemented with 2 mM penicillin streptomycin glutamine, In vitro kinase assay. In vitro kinase assays to investigate the changes in
10% FBS, and 1.5 g/ml sodium bicarbonate at 37°C with 5% CO2. Cdk5 activity were performed as published previously (Poore et al.,
Lentivirus production and transduction. Lentiviruses of empty vector
(EV) and p25-EGFP were prepared as described previously (Zheng et al., Inhibitor studies. BEL (0.3 M; Sigma) (iPLA2 inhibitor) or 10 M
2005). Briefly, HEK293-FT cells were co-transfected with p25-LV or AACOCF3 (Biomol) (cPLA2 inhibitor) was incubated with neurons
EV-LV and Virapower Packaging Mix using Lipofectamine 2000 and transduced with p25-LV. Neurons were fixed after 48 h and immunocy-
following the methods described in the ViraPower Lentiviral Expression tochemistry was performed using anti-GFAP antibody.
system (Invitrogen). Cell-free viral supernatants were collected after 48
and 72 h and filtered through a 0.45 m filter and then concentrated cPLA2 silencing. cPLA2 silencing was performed in neurons by trans-
ϳ40-fold by ultracentrifugation through Amicon Ultra-15 centrifugal duction with cPLA2 shRNA lentivirus (Santa Cruz Biotechnology) ac-
filter units with Ultracel-100 membranes (Millipore). Viral titers, ex- cording to the manufacturer’s instructions. Silencing was validated by
pressed as the percentage of total cells expressing the p25-GFP/EV-GFP reverse transcriptase (RT)-PCR and Western blot analyses.
(ϳ80%), were determined by transduction of 7 days in culture (DIC)
cortical neurons and GFP-fluorescence was visualized after 72 h. cPLA2 activity assay. cPLA2 activity was determined for the lysates
from neurons transduced with EV-LV or p25-LV using cPLA2 activity
Coculture and supernatant transfer experiments. EV-LV/p25-LV- Assay Kit (Cayman Chemical) according to the manufacturer’s instruc-
transduced cortical neurons or 5 d induced p25Tg/control neurons were tions. The results were normalized against protein concentration deter-
mination by BCA assays (Pierce Biotechnology).
Real-time PCR. Mice brains were homogenized in TRIzol and total RNA
was extracted using RNeasy Mini Kit (Qiagen) and then quantified using a
NanoDrop Spectrophotometer (Thermo Scientific). cDNA was synthesized
1022 • J. Neurosci., January 18, 2012 • 32(3):1020 –1034 Sundaram et al. • LPC/cPLA2 Mediates p25-Induced Neuroinflammation
according to manufacturer’s protocol using High capacity cDNA reverse cytokines and chemokines. This suggested that neuroinflamma-
transcriptase kits (Applied Biosystems). Quantitative real time PCR was per- tion is an early phenomenon in p25 overexpressed mice (data not
formed in a 96-well plate using ABI Prism 7900 HT Fast 9 Detection System shown). To confirm this, Western blot analyses (Fig. 1 A, B) and
(Applied Biosystems) as per manufacturer’s instructions. The 25 l PCR mix immunohistochemistry (Fig. 1C) were performed on brain sam-
included 2.5 l of 2 ng/l cDNA and 22.5 l of Power SYBR Green PCR ples of 1, 4, 8 and 12 week induced p25Tg/control mice (age
master mix (Applied Biosystems) with 0.2 M respective primers. The fol- matched noninduced p25Tg mice) using mouse anti-GFAP an-
lowing primers were used: MIP-1␣ (5Ј-TTGAGCCCCGGAACATTC-3Ј; tibody. Robust increases in GFAP expression was observed in all
5Ј-GCAGCAAACAGCTTATAGGAGATG-3Ј), TNF-␣ (5Ј-AGCACA- the induction periods of p25Tg mice compared with their con-
GAAAGCATGATCCG-3Ј; 5Ј-GGAGTAGACAAGGTACAACC-3Ј), trols even at 1 week of induction of p25 expression. We also
TGF-␤ (5Ј-CTTTAGGAAGGACCTGGGTT-3Ј; 5Ј-CAGGAGCGCA- performed quantitative (q)RT-PCR studies for the inflammatory
CAATCATGTT-3Ј), IL-1␤ (5Ј-ACCTGCTGGTGTGTGACGTTC-3Ј; cytokines and chemokines such as MIP-1␣, TNF-␣, TGF-␤ and
5Ј-CAGCACGAGGCTTTTTTGTTGT-3Ј), iPLA2 (5Ј-TAACCTGAAGC- IL-1␤ in the brain samples of 1, 4, 8 and 12 week induced p25Tg/
CACCGACTC-3Ј; 5Ј-TAGTGTTGATCTCTGATATG-3Ј) and cPLA2 (5Ј- control mice and found significant increases in these cytokines
CTGCAAGGCCGAGTGACA-3Ј; 5Ј-TTCGCCCACTTCTCTGCAA-3Ј). and chemokines in all p25Tg mice samples compared with the
The Ct values were determined using default threshold settings in the system controls (Fig. 1 D, E). Collectively, our results suggest that activa-
software. tion of glia and subsequent chemokine production are very early
events in p25Tg mice. To validate the involvement of Cdk5 in
Lipid extraction from cell culture media. Lipids were extracted from cell p25-mediated neuroinflammation, kinase assays were performed
culture media/supernatant using a modified protocol described previously on p25Tg mice brain samples and the results confirmed the hy-
(Bremer and Norum, 1982). In brief, lipids were first extracted using 6 ml of peractivation of Cdk5 in all the induction points of p25 expres-
1-butanol by incubating the cell culture media-butanol solution with agita- sion (Fig. 1 F). In addition, Cdk5 levels were found to be
tion at 160 rpm for 2 h at 4°C. After the break phase, the butanol extract at the unchanged using Western blot analyses (Fig. 1G).
upper layer was transferred to clean glass vial and 5 ml of chloroform was
then added to perform a second extraction. The chloroform extract at the p25-induced astrogliosis is an ␤-amyloid independent event
bottom phase was obtained after break phase and pooled with the butanol and occurs before microgloisis in p25Tg mice
extract from the first extraction. The lipid extracts were then dried under a Previous studies suggesting that amyloid plays a key role in the
nitrogen stream and kept at Ϫ80°C. induction of neuroinflammation (Combs et al., 2001; White et
al., 2005). To investigate the role of amyloid and phospho-tau in
Lipids analysis using high performance liquid chromatography/mass spec- the initiation of p25-mediated neuroinflammation, we examined
trometry. Separation and quantification of individual lipids was performed time points of ␤ amyloid production and phospho tau in p25Tg
using an Agilent 1200 high-performance liquid chromatography (HPLC) mice. Hyperphosphorylation of tau and amyloid pathology were
system and a 3200 Q-Trap mass spectrometer (Applied Biosystems) (Chan seen only in 4 week and 8 week induced p25Tg mice respectively
et al., 2008; Shui et al., 2010). The HPLC system is made up of an Agilent 1200 (Fig. 2 A). However, astrogliosis and chemokine production were
binary pump, an Agilent 1200 thermo sampler and an Agilent 1200 column found even in 1 week of p25 induction (Fig. 1). Our results indi-
oven HPLC conditions Luna 3u silica column (inner diameter 150 ϫ 2.0 cate that astrogliosis induced by p25 overexpression might be
mm); mobile phase A (chloroform/methanol/ammonium hydroxide, 89.5: independent of amyloid pathology. To study the microglial acti-
10:0.5), B (chloroform/methanol/ammonium hydroxide/water, 55:39:0.5: vation status in p25Tg mice, we performed immunohistochem-
5.5); flow rate 350 l/min; 5% B for 3 min, then linearly changed to 30% B in istry with the brain sections of 1, 4, 8 and 12 week induced p25Tg/
24 min and maintained for 5 min, and then linearly changed to 70% B in 5 control mice using anti-Cd11b antibody (a marker for activated
min and maintained for 7 min. Eluents were changed to the original ratio in microglia) (Matsuoka et al., 2001) (Fig. 2 B). Previous studies
5 min and maintained for 6 min. Multiple reaction monitoring transitions reported that microglial activation can be induced by neuronal
for individual phosphatidylethanolamine (PE), phosphatidylcholine (PC), tissue plasminogen activator (tPA), a serine protease that cata-
phosphatidylserine (PS), phosphatidylinositol (PI), ceramide, and sphingo- lyzes the conversion of inactive plasminogen to the active pro-
myelin (SM) species were set up for quantitative analysis. Levels of individual tease plasmin (Takahashi et al., 2010). To study the role of tPA in
lipid levels were quantified using spiked internal standards PC-14:0/14:0, the microglial activation, we have also performed the immuno-
PE14:0 –14:0, PS-14:0/14:0, ceramide, C12-SM and LPC 20:0 (4 M), which histochemistry with tPA antibody. These results mirror the
were obtained from Avanti Polar Lipids. Dioctanoyl phosphatidylinositol Cd11b findings (Fig. 2 B). Together, our results show that micro-
PI-8:0/8:0 was used for phosphatidylinositol quantitation (Echelon gloisis was absent at 1 week of p25 induction in p25Tg mice and
Biosciences). only became apparent at later time points. To further investigate
the involvement of peripheral cells in p25-induced neuroinflamma-
Stereotactic injection of lipids into mouse brain. Mice were placed in a tion, immunohistochemistry was performed in p25Tg/control mice
stereotactic frame after induction of anesthesia. The vertex area was ex- brain sections with CD4 and CD8 antibodies and we found dramatic
posed after midline incision of the scalp, and a small opening was made increases in these markers in p25Tg mice compared with control
with a dental drill according to the following coordinates: caudal to mice (Fig. 2C). Our results suggest that there is peripheral cell infil-
bregma 2 mm, 2 mm lateral to the midline and 1.8 mm from the surface tration during p25-mediated neuroinflammation.
of the cortex. A Hamilton microsyringe was inserted stereotactically
through the hole and 1 l of solution containing lipids (2 M) was in- p25-induced astrocyte activation is mediated by a soluble
jected at a constant rate for 5 min. The needle was withdrawn 10 min later factor
and the scalp sutured. Mice were monitored for 4 – 6 h after recovery To explore how p25 overexpression mediates neuroinflamma-
from anesthesia and were perfused with 4% PFA 48 h after injection. The tion and the nature of the factor that is responsible for astro-
brains were removed and sectioned for immunohistochemistry. gliosis, glia were cocultured with neurons transduced with
p25-LV/empty vector (EV-LV) and primary neurons from
Statistical analysis. All values are expressed as the mean of at least three p25Tg/control mice. Significant increases in GFAP staining
determinations Ϯ SEM. Data were analyzed by Student’s t test and p Ͻ
0.05 was considered to indicate statistical significance.
Astrogliosis is an early event in p25 transgenic mice
To investigate the early changes in gene expression by p25 over-
expression in vivo, we performed microarray analyses of 4 week
induced p25Tg mice forebrain samples and found robust eleva-
tion of inflammatory markers such as GFAP, proinflammatory
Sundaram et al. • LPC/cPLA2 Mediates p25-Induced Neuroinflammation J. Neurosci., January 18, 2012 • 32(3):1020 –1034 • 1023
Figure 1. Neuroinflammation is an early event in p25 transgenic mice. A, Immunoblot analyses were performed on the brain lysates of 1, 4, 8 and 12 week induced p25 transgenic mice (p25Tg) and their
respective age-matched noninduced p25Tg control mice (Ctrl) using anti-GFAP antibody (top). Equal amounts of protein loading were confirmed by reprobing the membrane with anti-tubulin antibody
(bottom). B, Quantification of immunoblot analyses in A by densitometric scanning (**p Ͻ 0.01 and ***p Ͻ 0.001). C, Representative confocal images of the cortex from 1, 4, 8 and 12 week induced p25Tg and
control mice brains. The sections were immunostained with anti-GFAP antibody (red) and DAPI (blue). Scale bars represent 50 m and images are representative of n ϭ 3 mice. D, E, Real-time PCR (RT-PCR)
results for cytokines and chemokines TNF-␣, MIP-1␣, TGF-␤ and IL-1␤ expression from 1, 4, 8 and 12 week induced p25Tg/control mice (***p Ͻ 0.001, *p Ͻ 0.05). F, Representative in vitro kinase assay graph
using active kinase (Cdk5) from p25Tg/control mice brain lysates to phosphorylate a NF-H peptide (***p Ͻ 0.001, *p Ͻ 0.05). G, Western blot analyses of the brain lysates from p25Tg/control mice using
anti-Cdk5 (C8) antibody. Error bars indicate ϮSEM.
1024 • J. Neurosci., January 18, 2012 • 32(3):1020 –1034 Sundaram et al. • LPC/cPLA2 Mediates p25-Induced Neuroinflammation
Figure 2. Initiation of neuroinflammation is independent of ␤ amyloid and tau phosphorylation in p25 transgenic mice. A–C, Representative confocal images of the cortex from 1, 4, 8 and 12
week induced p25Tg and control mice (age matched for 12 week time point). The brain sections were immunostained with (A) phospho-tau (AT8) and A␤ 1-42, (B) anti-Cd11b and anti-tPA, and
(C) CD4 and CD8 (red) antibodies. The nuclei were counterstained with DAPI (blue). White arrows indicate the region that is enlarged in high-magnification insets. Scale bars: (main panel), 20 m;
(insets), 10 m. Images are representative of n ϭ 3 mice.
was seen in glia that were cocultured with neurons transduced LV/p25-LV and the neurons from p25Tg/control mice for the
with p25-LV and neurons from p25Tg mice compared with inflammatory cytokines and chemokines such as MIP-1␣,
the respective controls. Our immunocytochemistry results TNF-␣, TGF-␤ and IL-1␤. Significant increases in cytokine
showed that the factor produced by p25 overexpression was and chemokine levels were seen in glia that were cocultured
soluble since the neurons and glia were not in contact with with p25 overexpressing neurons compared with the controls
each other (Fig. 3A). These results were confirmed by Western (Fig. 3C,D). Hyperactivation of Cdk5 due to p25 overexpres-
blots of glial samples that were cocultured with neurons trans- sion in p25-LV-transduced cortical neurons was validated us-
duced with EV-LV/p25-LV or neurons from p25Tg/control ing kinase assays and Western blot analyses (Fig. 3 E, F ). To
mice (Fig. 3B) where the outcome mirrored the immunostain- further confirm the soluble nature of the factor causing reac-
ing results described above. We also performed qRT-PCR on tive gliosis, we transferred cell-free supernatants from neu-
glia that were cocultured with neurons transduced with EV- rons transduced with EV-LV/p25-LV and neurons from
Sundaram et al. • LPC/cPLA2 Mediates p25-Induced Neuroinflammation J. Neurosci., January 18, 2012 • 32(3):1020 –1034 • 1025
Figure 3. A soluble factor mediates p25-induced neuroinflammation. A, 7-DIC cortical neurons transduced with EV-LV/p25-LV and primary neurons from control/5 d induced p25Tg mice were
cocultured with glia for 48 h using tissue culture inserts. Glia were fixed and immunostained with anti-GFAP antibody (green) and DAPI (blue). Scale bars, 20 m. B, Immunoblot analyses of lysates
from glia cocultured with EV-LV/p25-LV-transduced neurons and neurons from control/p25Tg mice. The samples were immunoprobed with anti-GFAP antibody (top). Tubulin serves as a loading
control (bottom). Bar graphs show the quantification of GFAP levels (***p Ͻ 0.001). C, D, RT-PCR results showing the expression of cytokines and chemokines MIP-1␣, TNF-␣, TGF-␤, and IL-1␤
in glia cocultured with cortical neurons transduced with EV-LV/p25-LV and cortical neurons from control/ p25Tg mice (***p Ͻ 0.001 and *p Ͻ 0.05). E, Representative in vitro kinase assay graph
using active kinase (Cdk5) from cortical neurons transduced with EV-LV/p25-LV to phosphorylate a NF-H peptide (**p Ͻ 0.01). F, Western blot analyses of the lysates from cortical neurons
transduced with EV-LV/p25-LV using anti-Cdk5 (C8) antibody. G, Cell-free supernatants from EV-LV/p25-LV-transduced neurons and neurons from control/p25Tg mice were transferred to glia for
48 h. The glia were fixed and immunostained with antibody to GFAP (green) and DAPI (blue). Scale bars represent 20 m. H, Western blot analyses of lysates of glia incubated with cell-free
supernatants from neurons transduced with EV-LV/p25-LV and control/p25Tg mice neurons using antibody to GFAP. Bar graph shows the quantification of GFAP levels (***p Ͻ 0.001). Error bars
indicate Ϯ SEM.
p25Tg/control mice to glia. The immunostaining and Western neurons transduced with EV-LV/p25-LV were treated with
blot analyses results supported our finding that a soluble sig- DNase, RNase and Proteinase K to systematically remove possi-
nal from p25 overexpressed neuron causes glial activation ble factors that could be secreted into solution during inflamma-
(Fig. 3G,H ). tion (Lauber et al., 2003). These enzyme-treated supernatants
were then transferred to glia and incubated for 48 h and glia were
p25 overexpressed neurons activate glia through a lipid signal processed for immunostaining (Fig. 4 A, B) and Western blot
To characterize the biochemical nature of the secreted factor analyses (Fig. 4C,D). The increase in GFAP levels in glia that
from the p25 overexpressed neurons, cell-free supernatants of received DNase-, RNase-, and Proteinase K-treated supernatants
1026 • J. Neurosci., January 18, 2012 • 32(3):1020 –1034 Sundaram et al. • LPC/cPLA2 Mediates p25-Induced Neuroinflammation
Figure 4. p25-induced neuroinflammation is mediated by a soluble lipid. A, Cell-free supernatants from EV-LV/p25-LV-transduced neurons treated with DNase (DN), RNase (RN) and Proteinase
K (PK) for 60 min or passed through SPE-C18 columns to remove lipids, were transferred to glia for 48 h. The nontreated (NT) and treated glia were fixed and immunostained with anti-GFAP antibody
(green) and DAPI (blue). Scale bars, 20 m. B, Quantification of GFAP expression in A (***p Ͻ 0.001; NS, p Ͼ 0.05). C, Lysates of glia that received same treatment as in A were separated by
SDS-PAGE and immunoprobed with anti-GFAP antibody. Tubulin acts as a loading control. D, Quantification of immunoblot analyses in C (***p Ͻ 0.001, NS, p Ͼ 0.05). E, RT-PCR results showing
the expression of cytokines and chemokines MIP-1␣, TNF-␣, TGF-␤, and IL-1␤ in glia that received the same treatment as in A (***p Ͻ 0.001,**p Ͻ 0.01, *p Ͻ 0.05, NS, p Ͼ 0.05). Error bars
from the p25-overexpressed neurons showed that neither validate our finding in vivo, we stereotactically injected the lipids
DNA, RNA, nor protein was likely to be responsible for p25- from p25-transduced neurons into mice brain and found a ro-
mediated neuroinflammation. To further elucidate the nature bust increase in GFAP immunoreactivity compared with mice
of the factor, we removed major species of lipid from the injected with lipids from empty vector-transduced neurons (Fig.
supernatants via SPE-C18 column elution and transferred 5D). To further determine the identity of the lipid found in the
these to glia. The significant reduction in GFAP levels in glia supernatants of the p25 overexpressed neurons that caused this
that received the column eluted supernatants suggested that astrogliosis, lipidomic mass spectrometric analysis was per-
the soluble factor from the p25 overexpressed cells could be lipid in formed on the lipids extracted from the supernatants of neurons
nature (Fig. 4A–D). qRT-PCR analyses for chemokines on the glia transduced with EV-LV/p25-LV. Among the major lipids, LPC
that received DNase-, RNase-, and Proteinase-treated supernatants levels showed a significant increase in lipids derived from p25-
or SPE-C18 column eluted supernatants from p25 overexpressing LV-transduced neurons compared with the EV-LV-transduced
neurons further confirmed our findings and dramatic reductions in neurons. Although LysoPI levels were also elevated, the differ-
chemokines were found in SPE-C18-treated supernatants (Fig. 4E). ence was not statistically significant ( p ϭ 0.223) (Fig. 5E). Lip-
idomic mass spectrometry using internal standards for various
LPC mediates the p25/Cdk5-induced inflammatory cascade LPC isoforms further identified LPC 16:0, 18:0 and 18:1 as the
To further confirm the lipid nature of the factor responsible for major lipids elevated in p25-LV derived lipids compared with
p25-induced astrogliosis, we extracted lipids from the superna- controls (Fig. 5F ). To determine which LPC isoform is more
tants of neurons transduced with EV-LV/p25-LV and transferred potent in the induction of GFAP expression, we treated glia with
the lipids onto glia for 24 and 48 h. Immunocytochemical stain- commercial LPC 16:0, 18:0 and 18:1. Our results show that LPC
ing (Fig. 5A) and Western blot analyses (Fig. 5 B, C) using anti- 18:1 was the most potent lipid to cause GFAP upregulation even
GFAP antibody showed significant increases in GFAP expression at 24 h (Fig. 5G–I ). The amount of LPC in the cell-free superna-
in the glia treated with lipids from p25-transduced neurons. To tants of p25 overexpressed cells were ϳ10-fold lower than the
Sundaram et al. • LPC/cPLA2 Mediates p25-Induced Neuroinflammation J. Neurosci., January 18, 2012 • 32(3):1020 –1034 • 1027
Figure 5. p25 overexpressing cells secrete LPC that mediates astrogliosis. A, Glia incubated with lipids extracted from supernatants of neurons transduced with EV-LV/p25-LV for 24 and 48 h were
fixed and immunostained with anti-GFAP antibody (green) and DAPI (blue). Scale bars, 20 m. B, Immunoblot analyses of lysates from the glia samples that received identical treatment as in A
using anti-GFAP antibody. C, Quantification of immunoblot analyses in B (***p Ͻ 0.001). D, Lipids from supernatants of neurons transduced with EV-LV/p25-LV (Figure legend continues.)
1028 • J. Neurosci., January 18, 2012 • 32(3):1020 –1034 Sundaram et al. • LPC/cPLA2 Mediates p25-Induced Neuroinflammation
amount of commercial LPC required to activate the glial cells. cPLA2 knock-down attenuates the p25-mediated glial
The form of LPC in the supernatant might be more potent and activation and chemokine production
form an active micelle more readily than the commercially avail- To investigate the effect of cPLA2 knock-down in p25-mediated
able LPC. Additive effect of other lipids in the supernatant of p25 LPC production, mass spectrometric analyses were performed
overexpressed neurons may also be responsible for this increase and results showed a significant decrease in LPC 18:0 and 18:1
potency. It would be hard to determine this additive effect due to levels in lipid extracts from the p25-LVϩcPLA2 shRNA-transduced
the combination of various other lipids where the permutations neurons compared with controls (Fig. 7A). To further study the
are numerous. Our finding was supported by a previous study effect of cPLA2 silencing in p25-mediated neuroinflammation,
describing 100-fold differences between the amount of LPC in supernatants from p25-LVϩcPLA2 shRNA/p25-LVϩcontrol shRNA-
supernatants and commercial LPC required for chemotactic ac- transduced neurons were transferred to glia. Western blot analyses
tivity (Lauber et al., 2003). To investigate the in vivo relevance of and immunostaining were performed using anti-GFAP antibody
LPC in astrogliosis, stereotactic injections of commercial LPC after 48 h of incubation. Marked reductions in GFAP staining were
18:1 and vehicle were performed in mouse brains (Fig. 5J ) and observed in glia that were incubated with the supernatants from
the results mirror the in vitro experiment findings. Although p25-LVϩcPLA2 shRNA-transduced neurons compared with con-
some studies used PBS to dissolve commercial LPC (Sheikh et al., trols (Fig. 7B–D). We also performed qRT-PCR with the glia treated
2009), Chloroform/Methanol (1:2) was used as a vehicle in our with cell-free supernatants from p25-LVϩcPLA2 shRNA or p25-
experiments to keep the solvent constant for the commercial LPC LVϩcontrol shRNA-transduced neurons for MIP-1␣, TNF-␣,
as well as the lipid extract treatments. Lipid extract did not dis- TGF-␤ and IL-1␤. Decreases in chemokine and cytokine expression
solve in PBS due to the poor hydrating nature of the lipids other levels in glial samples that received supernatants from p25-
than LPC. Lipidomic mass spectrometric analysis on p25Tg mice LVϩcPLA2 shRNA-transduced neurons compared with controls
brain samples further validated our findings where LPC levels (Fig. 7E). To further confirm the effect of cPLA2 upregulation in the
were significantly elevated compared with controls (Fig. 6 A). induction of neuroinflammation, glia were incubated with lipids
Collectively, our in vivo and in vitro results revealed that LPC is extracted from supernatants of p25-LVϩcontrol shRNA or p25-
one of the major lipid secreted by the p25 expressing neurons to LVϩcPLA2 shRNA-transduced neurons for 48 h. Immunostaining
result in astrogliosis. and Western blot analyses results showed 4 to fivefold decreases in
GFAP expression in glia treated with lipids from p25-LVϩcPLA2
p25-induced upregulation of cPLA2 causes LPC production shRNA-transduced neurons compared with controls (Fig. 7F–H).
We investigated the changes in cPLA2 expression and activity in We also injected lipids from the supernatants of p25-LVϩ
p25 overexpressed neurons to determine the mechanism behind control shRNA and p25-LVϩcPLA2 shRNA-transduced neu-
p25-mediated LPC production. RT-PCR results showed robust rons into mice brains and immunostaining results showed
increases in cPLA2 expression in different induction periods of robust reductions in GFAP expression in mice injected with
p25Tg mice brain samples (Fig. 6 B). We also found significant lipids from p25-LVϩcPLA2 shRNA-transduced neurons (Fig.
increases in cPLA2 expression as well as cPLA2 activities in p25- 7I ). Together, our results show that cPLA2 upregulation is a
LV-transduced neurons and 5 d induced p25Tg mice neurons critical event in p25-mediated neuroinflammation.
(Fig. 6C–F ). The involvement of other PLA2 isoforms in p25-
induced LPC production was investigated using RT-PCR studies p25-induced inflammatory mediators trigger
for iPLA2 expression, inhibitor studies using BEL (iPLA2 inhib- neurodegeneration
itor) and AACOCF3 (cPLA2 inhibitor). In these experiments, To investigate the importance of p25-mediated neuroinflamma-
cPLA2 was found to be the principle enzyme involved in p25- tion in the induction of neurodegeneration, we treated cortical
mediated LPC production (Fig. 6G,H ). To further confirm the neurons with conditioned media from glia that received lipids
involvement of p25-induced upregulation of cPLA2 in LPC pro- from EV-LV, p25-LVϩcontrol shRNA and p25-LVϩcPLA2
duction, cPLA2 gene silencing experiments using cPLA2 shRNA shRNA-transduced neurons. Robust increases in neurodegen-
lentiviral particles in p25-LV-transduced neurons were per- erative markers such as phospho-tau (AT8) and intra cellular
formed. Silencing was validated by qRT-PCR, cPLA2 activity as- A␤ 1– 42 accumulation in neurons that received supernatants
says and Western blots (Fig. 6 I, J ). from glia treated with lipids from p25 overexpressed neurons
compared with glia treated with lipids purified from cPLA2
4 silenced p25 overexpressing neurons were seen (Fig. 8 A). Re-
sults from in vivo injections of lipids from EV-LV, p25-
(Figure legend continued.) were stereotactically injected into mice brains and the brain sec- LVϩcontrol shRNA and p25-LVϩcPLA2 shRNA-transduced
tions were immunostained with anti-GFAP antibody (red) and DAPI (blue). Scale bars: (20ϫ, neurons into mice brains show identical findings (Fig. 8 B).
top), 50 m; 20 m (40ϫ, bottom), 20 m. E, Mass spectrometric analyses were performed We found significant increases in cell death in neurons that
on lipids from supernatants of neurons transduced with EV-LV/p25-LV. Results were normal- were treated with supernatants from glia incubated with lipids
ized against the internal standards of the respective lipids (***p Ͻ 0.001). F, Mass spectromet- from p25 overexpressed neurons (Fig. 8C,D). In contrast, we
ric analyses were performed on lipids from samples as in E and the results were normalized did not observe any cell death in our in vivo experiments (data
against the internal standards of the respective LPC isoforms (**p Ͻ 0.01; NS, p Ͼ 0.05). G, Glia not shown). This could be because of a single dose of lipids
were treated with media containing 20 M LPC of 16:0, 18:0, or 18:1 for 24 and 48 h. Immuno- injected into mice and subsequent clearance of lipids as well as
cytochemistry of glia was performed using anti-GFAP antibody (green) and DAPI (blue). Scale the abnormal protein species (tau and amyloid). In general,
bars, 20 m. H, Western blot analyses of lysates from the samples as in G using anti-GFAP the crucial mechanism behind the neuronal death is the re-
antibody. I, Quantification of immunoblot analyses in H (***p Ͻ 0.001; NS, p Ͼ 0.05). J, LPC duced clearance capability in neurons, probably attributed to
18:1/vehicle was injected into mice brains and immunostaining was performed using anti- the reduction in function of the proteosomal and autophagic
GFAP antibody (red) and DAPI (blue). Scale bars: (20ϫ, top), 50 m; (40ϫ, bottom), 20 m. machinery (Keller et al., 2000; Komatsu et al., 2006). Addi-
Error bars indicate ϮSEM. tionally, this trigger has to be sustained either by p25 produc-
tion or by other toxic insults to cause neuronal damage.
Sundaram et al. • LPC/cPLA2 Mediates p25-Induced Neuroinflammation J. Neurosci., January 18, 2012 • 32(3):1020 –1034 • 1029
Figure 6. p25 expression causes an increase in cPLA2 expression and activity. A, Mass spectrometric analyses results for the brain samples from p25Tg/control mice. Results were normalized
against the internal standards of LPC (**p Ͻ 0.01, *p Ͻ 0.05; NS, p Ͼ 0.1). B, Quantitative real-time PCR results of cPLA2 gene expression in 1, 4, 8, and 12 week induced p25Tg mice and age
matched control mice (***p Ͻ 0.001). C, RT-PCR results of cPLA2 gene expression in neurons from p25Tg/control mice and neurons transduced with EV-LV/p25-LV (***p Ͻ 0.001). D, cPLA2 activity
assays were performed with lysates from neurons of p25Tg/control mice and neurons transduced with EV-LV/p25-LV (***p Ͻ 0.001). E, Western blot analyses were performed on lysates from the
samples same as in D using anti-cPLA2 antibody. F, Quantification of immunoblot analyses in E (***p Ͻ 0.001). G, Real-time PCR results of iPLA2 gene expression in cortical neurons transduced with
EV-LV/p25-LV (NS, p Ͼ 0.1). H, Glia were treated with supernatants from cortical neurons transduced with EV-LV/p25-LV or p25-LVϩcPLA2 inhibitor (AACOCF3)/p25-LVϩiPLA2 inhibitor (BEL).
Immunocytochemistry was performed on the glia with anti-GFAP antibody (green) and DAPI (blue). Scale bars, 20 m. I, RT-PCR results of cPLA2 gene expression in 7-DIC cortical neurons
transduced with p25-LVϩcPLA2 shRNA or p25-LVϩcontrol (ctrl) shRNA (***p Ͻ 0.001). Graph in the bottom panel shows the results of cPLA2 activity assays in neurons transduced with
p25-LVϩcPLA2 shRNA or p25-LVϩcontrol shRNA (**p Ͻ 0.01). J, Western blot analyses were performed on samples from neurons transduced with p25-LVϩcPLA2 shRNA or p25-LVϩcontrol
shRNA using anti-cPLA2 antibody. Tubulin acts as a loading control for Western blot analyses. Quantification of immunoblots was shown in the bottom (***p Ͻ 0.001). Error bars indicate ϮSEM.
However, an earlier study showed significant reduction of ferent neuronal subtypes to a toxic insult. Together, our re-
neurons after LPC injection (Sheikh et al., 2009). This differ- sults show that an inflammatory component produced during
ential observation could be due to the differences in the form the early event of p25-induced neuroinflammation triggers
of commercial LPC, injection site and susceptibilities of dif- neuropathological changes found in AD.
1030 • J. Neurosci., January 18, 2012 • 32(3):1020 –1034 Sundaram et al. • LPC/cPLA2 Mediates p25-Induced Neuroinflammation
Figure 7. cPLA2 knock-down regulates p25-induced neuroinflammation. A, Mass spectrometric analysis of lipids from the supernatants of p25-LVϩcontrol shRNA/p25-LVϩcPLA2 shRNA-
transduced neurons and results were expressed as normalized intensities against internal standards of the respective LPC isoforms (**p Ͻ 0.01, *p Ͻ 0.05). B, Immunocytochemistry on glia treated
with supernatants from neurons transduced with p25-LVϩcPLA2 shRNA or p25-LVϩcontrol shRNA using anti-GFAP antibody (green) and DAPI (blue). Scale bars, 20 m. C, Western blot analyses
were performed on glia that received identical treatment as in B using anti-GFAP antibody. D, Quantification of immunoblot analyses in C (***p Ͻ 0.001). E, RT-PCR results showing the expression
of inflammatory cytokines and chemokines MIP-1␣, TNF-␣, TGF-␤ and IL-1␤ in samples from glial cell that received same treatment as B (***p Ͻ 0.001; NS, p Ͼ 0.05). F, Glia treated with lipids
from the supernatants of p25-LVϩcontrol shRNA/p25-LVϩcPLA2 shRNA-transduced neurons were fixed and immunostained with anti-GFAP antibody (green) and DAPI (blue). Scale bars, 20 m.
G, Western blot analyses were performed on glia that received identical treatment as in F using anti-GFAP antibody. H, Quantification of immunoblots in G (**p Ͻ 0.01). I, Lipids from the
supernatants of p25-LVϩcontrol shRNA/p25-LVϩcPLA2 shRNA-transduced neurons were injected stereotactically into mice brains and immunostaining was performed in brain sections with
anti-GFAP antibody (red) and DAPI (blue). Scale bars: (20ϫ, top) 50 m, (40ϫ, bottom), 20 m. Error bars indicate ϮSEM.
Discussion The loss of Cdk5 activity (Takahashi et al., 2010) and hyperacti-
vation of Cdk5 (Kitazawa et al., 2005) lead to neuronal loss and in
Our findings detail a novel mechanism of neuroinflammation either situation neuroinflammation has been reported to be one key
as a result of p25 overexpression. We show that cPLA2- biochemical event linked with this process. However, the molecular
mediated LPC production plays a major role in p25 induced mechanism behind hyperactivation of Cdk5, and its relevance to
neuroinflammation and trigger the neuropathological changes neurodegenerative disease pathogenesis has not been elucidated.
reminiscent of AD both in vitro and in vivo. Inhibition of this
trigger reduces astrogliosis, tau phosphorylation, and subsequent Although a number of neurodegenerative models exhibit neu-
amyloid accumulation. roinflammation and p25 elevation (Cruz et al., 2003; Muyllaert et
Sundaram et al. • LPC/cPLA2 Mediates p25-Induced Neuroinflammation J. Neurosci., January 18, 2012 • 32(3):1020 –1034 • 1031
Figure 8. p25-mediated neuroinflammation is a trigger for neurodegeneration. A, In vitro treatment of 7-DIC cortical neurons with cell-free supernatants from glia that treated with lipids
extracted from EV-LV, p25-LVϩcontrol shRNA and p25-LVϩcPLA2 shRNA-transduced neurons and also from glia treated with LPC18:1/vehicle (chloroform/Methanol) for 48 h. The neurons were
then fixed and immunostained with antibodies specific to phospho-tau (AT8) and A␤ 1– 42. Nuclei were counterstained with DAPI. Scale bars, 20 m. B, In vivo injections of LPC18:1, vehicle and
lipids extracted from EV-LV, p25-LVϩcontrol shRNA and p25-LVϩcPLA2 shRNA-transduced neurons into mice brain. The animals were perfused after 4 d and the brain sections were immuno-
stained with antibodies specific to phospho-tau and A␤ 1– 42 (red). Nuclei were counterstained with DAPI (blue). Scale bars, 20 m. C, TUNEL staining of cortical neurons incubated with
supernatants from glia treated with vehicle, LPC 18:1 and lipids extracted from EV-LV, p25-LVϩcontrol shRNA and p25-LVϩcPLA2 shRNA-treated neurons. Scale bars, 20 m. D, Percentage cell
death in C was calculated by counting the TUNEL-positive cells normalized with DAPI from 10 independent fields (***p Ͻ 0.001). Error bars indicate ϮSEM.
al., 2008), this is the first report to provide a link between the two induction. However, microgliosis was not observed at this time
processes using in vivo p25Tg mice and in vitro p25 overexpress- point and was evident only during later induction periods in
ing neurons as well as identifying a specific lipid factor responsi- p25Tg mice. In contrast, we did not observe any microglial activa-
ble for triggering one phase of the neurodegenerative process. We tion in our mixed glia culture treated either with the conditioned
determined that neuroinflammation is an early event in p25Tg media or with the lipid extracts from the p25 overexpressed neurons
mice, where marked increases in GFAP and proinflammatory for 48 h (data not shown). This might possibly be because of the less
cytokines levels in the brain were observed even after 1 week of abundance of microglia in our in vitro system compared with in vivo.
1032 • J. Neurosci., January 18, 2012 • 32(3):1020 –1034 Sundaram et al. • LPC/cPLA2 Mediates p25-Induced Neuroinflammation
Moreover, conditioned media transfer
experiments between microglia and p25
overexpressed neurons did not show any
significant changes in microglial activa-
tion (data not shown). Our finding of as-
trogliosis preceding microgliosis in p25-
mediated neuroinflammation is also
supported by various other studies where
chemokines released by astrocytes attract
microglia which further express proin-
flammatory products contributing to the
additional damage to neurons (Hurwitz et
al., 1995; Tuppo and Arias, 2005). Neuro-
inflammation has been previously re-
ported to be initiated by amyloid
formation in various neurodegenerative
models (Combs et al., 2001; White et al.,
2005). However, our data suggest an alter-
nate trigger for astrogliosis that occurs
even before any evidence of amyloid pa-
thology in p25Tg mice which becomes ev-
ident after 8 weeks of p25 expression. Our
finding is also supported by studies where
elevations of GFAP were not caused by the
consequence of amyloid accumulation
(Ingelsson et al., 2004; Zhu et al., 2008)
and the data presented in this report sug-
gests that the alternate pathway could be
p25-mediated LPC production. Al-
though, the involvement of CD4ϩ and
CD8ϩ T cell recruitment in neuroinflam-
mation has been previously reported in
neurodegenerative diseases (Brisebois et
al., 2006; Brochard et al., 2009), this is the Figure 9. Model proposed to explain the mechanism behind neurodegeneration caused by p25/Cdk5-mediated neuroinflam-
first report showing that p25 overexpres- mation. p25/Cdk5 hyperactivation releases extracellular soluble LPC through the upregulation of cPLA2. LPC activates glia to
sion may initiate the peripheral cell re- produce proinflammatory cytokines and chemokines, which then causes infiltration of leukocytes. The inflammatory components
cruitment into the mice brain to produced by p25 overexpression then triggers neurodegenerative disease progression by causing the accumulation of phospho-
exacerbate neuroinflammation. LPC has tau and ␤ amyloid.
previously been shown to be a chemoat-
tractant for T cells (Ousman and David, 1984). Recent studies suggest that increased PLA2 activity and
2000; Zhang et al., 2007), therefore it is reasonable to believe that PLA2-generated proinflammatory mediators play a central role
p25-mediated LPC production is responsible for this peripheral in inflammatory responses associated with neurological disor-
cell recruitment. ders such as ischemia, AD, PD, and MS (Farooqui et al., 2006). To
date, cytosolic PLA2 (cPLA2-IV), Ca 2ϩ-dependent PLA2
To gain insight into the mechanism of p25-mediated neuroin- (iPLA2-VI) and secretory PLA2 (sPLA2-II) are the major PLA2
flammation, we performed coculture as well as conditioned media types that were involved in inflammatory-mediated neurodegen-
transfer experiments to identify the soluble factor that is released eration. Although, sPLA2 isoforms X, V and III produce LPC, the
from p25 expressing neurons to activate astrocytes. We then eluci- neuroinflammatory role of these isoforms has not been eluci-
dated the nature of the soluble signal by factor removal experiments dated (Dennis, 1997; Sun et al., 2004). The isoform sPLA2-II is
using mass spectrometry lipidomics to identify that the lipid was absent in our mouse model and primary neuronal cultures be-
lysophosphatidylcholine (LPC). Our findings have been supported cause of an inbred gene deletion in the C57BL/6 mice strain (Ken-
by previous studies where, increased LPC levels have been observed nedy et al., 1995). The isoform iPLA2 is generally observed as a
in multiple sclerosis and in aged human brain (Andreoli et al., 1973; housekeeping enzyme for the maintenance of membrane phos-
Wender et al., 1988). However, the precise mechanism behind LPC pholipids (Balsinde and Dennis, 1997). Our qRT-PCR and PLA2
production has not been fully elucidated. We characterized the par- inhibitor studies using AACOCF3 (cPLA2 inhibitor) and BEL
ticular subtype of LPC responsible for astrogliosis as LPC 18.1. Pre- (iPLA2 inhibitor) in the p25 overexpressing neurons, confirm
vious studies suggest that the potency of LPC may vary based on the
relative length, position, and unsaturation of acyl chain. The potency
of LPC is also determined by the ability to form the active micelles that p25 overexpression does not affect iPLA2 levels. AACOCF3
and the unsaturated LPC (18:1) could readily form the active mi- is 500-fold more potent inhibitor of cPLA2 than sPLA2. AA-
celles compared with the other saturated isoforms (Lauber et al., COCF3 inhibits bovine brain cPLA2 and iPLA2 in a dose-
2003; Ojala et al., 2007). dependent manner with IC50 values of 1.5 and 6.0 M,
LPC is produced by the hydrolysis of phosphatidylcholine via respectively. BEL is Ͼ1,000-fold selective for iPLA2 than cPLA2
the action of phospholipase A2 (PLA2) (Steinbrecher et al., (Riendeau et al., 1994; Jenkins et al., 2002; Farooqui et al., 2006).
Sundaram et al. • LPC/cPLA2 Mediates p25-Induced Neuroinflammation J. Neurosci., January 18, 2012 • 32(3):1020 –1034 • 1033
We show that p25 overexpression increases LPC levels with a in humans could therefore be a viable therapeutic intervention to
concomitant upregulation of cPLA2 expression and activity in all treat neuroinflammation in neurodegenerative diseases.
the p25 overexpressing systems including p25Tg mice, primary
neurons from the p25Tg mice and lentiviral-transduced neurons. References
Increased cPLA2 activity has been observed in AD brains in asso-
ciation with A␤ toxicity (Stephenson et al., 1996; Farooqui et al., Andreoli VM, Maffei F, Tonon GC, Zibetti A (1973) Significance of plasma
1997; Kriem et al., 2005). However, our studies in p25Tg mice lysolecithin in patients with multiple sclerosis: a longitudinal study.
show that upregulation of cPLA2 occurred before any phospho- J Neurol Neurosurg Psychiatry 36:661– 667.
tau and amyloid formation (Figs. 2 A, 6 B). Overexpression of p25
may mediate this response via the transcriptional regulation of Balsinde J, Dennis EA (1997) Function and inhibition of intracellular
cPLA2 gene expression, since p25/Cdk5 activity has been ob- calcium-independent phospholipase A2. J Biol Chem 272:16069 –16072.
served in the nucleus of neurons (Patrick et al., 1999). To
investigate whether the reduction of p25-mediated cPLA2 up- Bian F, Nath R, Sobocinski G, Booher RN, Lipinski WJ, Callahan MJ, Pack A,
regulation could ameliorate the pathological progression, we si- Wang KK, Walker LC (2002) Axonopathy, tau abnormalities, and dys-
lenced cPLA2 gene expression and found dramatic reductions in kinesia, but no neurofibrillary tangles in p25-transgenic mice. J Comp
LPC 18:1 and significant decreases in the expression of neuroinflam- Neurol 446:257–266.
matory markers (Fig. 7). To delineate whether the p25 induced in-
flammatory mediators production can have the ability to trigger Bremer J, Norum KR (1982) Metabolism of very long-chain monounsatu-
neurodegeneration, we performed conditional media transfer ex- rated fatty acids (22:1) and the adaptation to their presence in the diet.
periments between the activated glia and healthy cortical neurons. J Lipid Res 23:243–256.
Our results suggest that p25-induced neuroinflammation may be an
initial trigger for neurodegeneration. However, absence of TUNEL Brisebois M, Zehntner SP, Estrada J, Owens T, Fournier S (2006) A patho-
positive cells in the in vivo studies suggests that there is a possibility of genic role for CD8ϩ T cells in a spontaneous model of demyelinating
involvement other pathways in the progression of neurodegenera- disease. J Immunol 177:2403–2411.
tion. Marked reductions in neuropathological markers expression in
experiments with lipid from cPLA2 silenced p25 overexpressing Brochard V, Combadie`re B, Prigent A, Laouar Y, Perrin A, Beray-Berthat V,
neurons further confirm our findings. Together, our results indicate Bonduelle O, Alvarez-Fischer D, Callebert J, Launay JM, Duyckaerts C,
that cPLA2 is upregulated when p25 is produced, resulting in subse- Flavell RA, Hirsch EC, Hunot S (2009) Infiltration of CD4ϩ lympho-
quent neuroinflammation and neurodegeneration. Recent studies cytes into the brain contributes to neurodegeneration in a mouse model
support our findings where, CDP-choline (a cPLA2 inhibitor) pro- of Parkinson disease. J Clin Invest 119:182–192.
tected neurons from glutamate-mediated neurotoxicity (Mir et al.,
2003). Chan R, Uchil PD, Jin J, Shui G, Ott DE, Mothes W, Wenk MR (2008)
Retroviruses human immunodeficiency virus and murine leukemia virus
We propose that Cdk5/p25 hyperactivation causes cPLA2 up- are enriched in phosphoinositides. J Virol 82:11228 –11238.
regulation and releases soluble LPC which activates glia to pro-
duce proinflammatory chemokines and cytokines. LPC and Combs CK, Karlo JC, Kao SC, Landreth GE (2001) beta-Amyloid stimula-
subsequent inflammatory mediator also triggers recruitment of tion of microglia and monocytes results in TNFalpha-dependent expres-
peripheral immune cells into the brain, which could then initiate sion of inducible nitric oxide synthase and neuronal apoptosis. J Neurosci
another phase of neuroinflammatory changes in the CNS causing 21:1179 –1188.
further damage. Finally, the inflammatory mediators trigger the
initiation of AD-like pathological outcomes (Fig. 9). Cruz JC, Tseng HC, Goldman JA, Shih H, Tsai LH (2003) Aberrant Cdk5
activation by p25 triggers pathological events leading to neurodegenera-
There is no optimal approach to date supporting the treat- tion and neurofibrillary tangles. Neuron 40:471– 483.
ment of AD patients with anti-inflammatory drugs (Imbimbo,
2009) and it would be interesting to investigate the effect of spe- Dennis EA (1997) The growing phospholipase A2 superfamily of signal
cific inhibition of cPLA2 in neurodegenerative disease progres- transduction enzymes. Trends Biochem Sci 22:1–2.
sion. Specific knock down of cPLA2 in p25Tg mice would be an
interesting experiment to investigate p-tau, A␤ 1– 42 and neuro- Farooqui AA, Rapoport SI, Horrocks LA (1997) Membrane phospholipid
nal death in these mice especially in the absence of tau and amy- alterations in Alzheimer’s disease: deficiency of ethanolamine plasmalo-
loid pathology early in the disease process. However, knockdown gens. Neurochem Res 22:523–527.
of cPLA2 using cPLA2 shRNA lentivirus in p25Tg mice would be
challenging due to the sustained inflammatory trigger by con- Farooqui AA, Ong WY, Horrocks LA (2006) Inhibitors of brain phospho-
stant p25 expression and cPLA2 upregulation in p25Tg mice. lipase A2 activity: their neuropharmacological effects and therapeutic im-
Repeated injections with shRNA virus would be required to get portance for the treatment of neurologic disorders. Pharmacol Rev
the optimal result which is detrimental to the mice. Crossing 58:591– 620.
cPLA2 knock-out mice with p25Tg mice would be another pos-
sible study to further expand knowledge of the mechanisms in Hallows JL, Iosif RE, Biasell RD, Vincent I (2006) p35/p25 is not essential for
inflammatory-mediated neurodegenerative disease progression. tau and cytoskeletal pathology or neuronal loss in Niemann-Pick type C
disease. J Neurosci 26:2738 –2744.
In conclusion, we demonstrate for the first time that LPC is a
molecular signal released from neurons to mediate glial activa- Hurwitz AA, Lyman WD, Berman JW (1995) Tumor necrosis factor alpha
tion and the recruitment of inflammatory mediators in p25Tg and transforming growth factor beta upregulate astrocyte expression of
mice contributing to AD-like neuropathology. We also identify monocyte chemoattractant protein-1. J Neuroimmunol 57:193–198.
cPLA2 and the specific LPC subtype responsible for the pathological
effects in the brain. The identification of a specific inhibitor of cPLA2 Imbimbo BP (2009) An update on the efficacy of non-steroidal anti-
inflammatory drugs in Alzheimer’s disease. Expert Opin Investig Drugs
Ingelsson M, Fukumoto H, Newell KL, Growdon JH, Hedley-Whyte ET,
Frosch MP, Albert MS, Hyman BT, Irizarry MC (2004) Early Abeta ac-
cumulation and progressive synaptic loss, gliosis, and tangle formation in
AD brain. Neurology 62:925–931.
Jenkins CM, Han X, Mancuso DJ, Gross RW (2002) Identification of calcium-
independent phospholipase A2 (iPLA2) beta, and not iPLA2gamma, as the me-
diator of arginine vasopressin-induced arachidonic acid release in A-10 smooth
muscle cells. Enantioselective mechanism-based discrimination of mammalian
iPLA2s. J Biol Chem 277:32807–32814.
Keller JN, Hanni KB, Markesbery WR (2000) Possible involvement of pro-
teasome inhibition in aging: implications for oxidative stress. Mech Age-
ing Dev 113:61–70.
Kennedy BP, Payette P, Mudgett J, Vadas P, Pruzanski W, Kwan M, Tang C,
Rancourt DE, Cromlish WA (1995) A natural disruption of the secre-
tory group II phospholipase A2 gene in inbred mouse strains. J Biol Chem
Kesavapany S, Li BS, Amin N, Zheng YL, Grant P, Pant HC (2004a) Neuro-
nal cyclin-dependent kinase 5: role in nervous system function and its
specific inhibition by the Cdk5 inhibitory peptide. Biochim Biophys Acta
1034 • J. Neurosci., January 18, 2012 • 32(3):1020 –1034 Sundaram et al. • LPC/cPLA2 Mediates p25-Induced Neuroinflammation
Kesavapany S, Amin N, Zheng YL, Nijhara R, Jaffe H, Sihag R, Gutkind JS, Riendeau D, Guay J, Weech PK, Laliberte´ F, Yergey J, Li C, Desmarais S,
Takahashi S, Kulkarni A, Grant P, Pant HC (2004b) p35/cyclin- Perrier H, Liu S, Nicoll-Griffith D, et al. (1994) Arachidonyl trifluorom-
dependent kinase 5 phosphorylation of ras guanine nucleotide releasing ethyl ketone, a potent inhibitor of 85-kDa phospholipase A2, blocks pro-
factor 2 (RasGRF2) mediates Rac-dependent Extracellular Signal- duction of arachidonate and 12-hydroxyeicosatetraenoic acid by calcium
regulated kinase 1/2 activity, altering RasGRF2 and microtubule- ionophore-challenged platelets. J Biol Chem 269:15619 –15624.
associated protein 1b distribution in neurons. J Neurosci 24:4421– 4431.
Saito T, Konno T, Hosokawa T, Asada A, Ishiguro K, Hisanaga S (2007)
Kitazawa M, Oddo S, Yamasaki TR, Green KN, LaFerla FM (2005) p25/cyclin-dependent kinase 5 promotes the progression of cell death in
Lipopolysaccharide-induced inflammation exacerbates tau pathology by nucleus of endoplasmic reticulum-stressed neurons. J Neurochem
a cyclin-dependent kinase 5-mediated pathway in a transgenic model of 102:133–140.
Alzheimer’s disease. J Neurosci 25:8843– 8853.
Sheikh AM, Nagai A, Ryu JK, McLarnon JG, Kim SU, Masuda J (2009) Ly-
Komatsu M, Waguri S, Chiba T, Murata S, Iwata J, Tanida I, Ueno T, Koike sophosphatidylcholine induces glial cell activation: role of rho kinase.
M, Uchiyama Y, Kominami E, Tanaka K (2006) Loss of autophagy in the Glia 57:898 –907.
central nervous system causes neurodegeneration in mice. Nature
441:880 – 884. Shui G, Guan XL, Gopalakrishnan P, Xue Y, Goh JS, Yang H, Wenk MR
(2010) Characterization of substrate preference for Slc1p and Cst26p in
Kriem B, Sponne I, Fifre A, Malaplate-Armand C, Lozac’h-Pillot K, Koziel V, Saccharomyces cerevisiae using lipidomic approaches and an LPAAT ac-
Yen-Potin FT, Bihain B, Oster T, Olivier JL, Pillot T (2005) Cytosolic tivity assay. PLoS One 5:e11956.
phospholipase A2 mediates neuronal apoptosis induced by soluble oli-
gomers of the amyloid-beta peptide. FASEB J 19:85– 87. Smith PD, Mount MP, Shree R, Callaghan S, Slack RS, Anisman H, Vincent I,
Wang X, Mao Z, Park DS (2006) Calpain-regulated p35/cdk5 plays a
Lauber K, Bohn E, Kro¨ber SM, Xiao YJ, Blumenthal SG, Lindemann RK, central role in dopaminergic neuron death through modulation of the
Marini P, Wiedig C, Zobywalski A, Baksh S, Xu Y, Autenrieth IB, Schulze- transcription factor myocyte enhancer factor 2. J Neurosci 26:440 – 447.
Osthoff K, Belka C, Stuhler G, Wesselborg S (2003) Apoptotic cells in-
duce migration of phagocytes via caspase-3-mediated release of a lipid Steinbrecher UP, Parthasarathy S, Leake DS, Witztum JL, Steinberg D (1984)
attraction signal. Cell 113:717–730. Modification of low density lipoprotein by endothelial cells involves lipid
peroxidation and degradation of low density lipoprotein phospholipids.
Lopes JP, Oliveira CR, Agostinho P (2007) Role of cyclin-dependent kinase Proc Natl Acad Sci U S A 81:3883–3887.
5 in the neurodegenerative process triggered by amyloid-Beta and prion
peptides: implications for Alzheimer’s disease and prion-related enceph- Stephenson DT, Lemere CA, Selkoe DJ, Clemens JA (1996) Cytosolic phos-
alopathies. Cell Mol Neurobiol 27:943–957. pholipase A2 (cPLA2) immunoreactivity is elevated in Alzheimer’s dis-
ease brain. Neurobiol Dis 3:51– 63.
Markiewicz I, Lukomska B (2006) The role of astrocytes in the physiology
and pathology of the central nervous system. Acta Neurobiol Exp (Wars) Sun GY, Xu J, Jensen MD, Simonyi A (2004) Phospholipase A2 in the cen-
66:343–358. tral nervous system: implications for neurodegenerative diseases. J Lipid
Matsuoka Y, Picciano M, Malester B, LaFrancois J, Zehr C, Daeschner JM,
Olschowka JA, Fonseca MI, O’Banion MK, Tenner AJ, Lemere CA, Duff K Takahashi S, Ohshima T, Hirasawa M, Pareek TK, Bugge TH, Morozov A,
(2001) Inflammatory responses to amyloidosis in a transgenic mouse Fujieda K, Brady RO, Kulkarni AB (2010) Conditional deletion of neu-
model of Alzheimer’s disease. Am J Pathol 158:1345–1354. ronal cyclin-dependent kinase 5 in developing forebrain results in micro-
glial activation and neurodegeneration. Am J Pathol 176:320 –329.
Meyerson M, Faha B, Su LK, Harlow E, Tsai LH (1991) The cyclin-
dependent kinase family. Cold Spring Harb Symp Quant Biol 56:177–186. Takashima A, Murayama M, Yasutake K, Takahashi H, Yokoyama M,
Ishiguro K (2001) Involvement of cyclin dependent kinase5 activator
Mir C, Clotet J, Aledo R, Durany N, Argemí J, Lozano R, Cervo´ s-Navarro J, p25 on tau phosphorylation in mouse brain. Neurosci Lett 306:37– 40.
Casals N (2003) CDP-choline prevents glutamate-mediated cell death
in cerebellar granule neurons. J Mol Neurosci 20:53– 60. Town T, Zolton J, Shaffner R, Schnell B, Crescentini R, Wu Y, Zeng J, Delle-
Donne A, Obregon D, Tan J, Mullan M (2002) p35/Cdk5 pathway me-
Muyllaert D, Terwel D, Kremer A, Sennvik K, Borghgraef P, Devijver H, diates soluble amyloid-beta peptide-induced tau phosphorylation in
Dewachter I, Van Leuven F (2008) Neurodegeneration and neuroin- vitro. J Neurosci Res 69:362–372.
flammation in cdk5/p25-inducible mice: a model for hippocampal scle-
rosis and neocortical degeneration. Am J Pathol 172:470 – 485. Trojanowski JQ, Mawal-Dewan M, Schmidt ML, Martin J, Lee VM (1993)
Localization of the mitogen activated protein kinase ERK2 in Alzheimer’s
Nguyen MD, Larivie`re RC, Julien JP (2001) Deregulation of Cdk5 in a disease neurofibrillary tangles and senile plaque neurites. Brain Res
mouse model of ALS: toxicity alleviated by perikaryal neurofilament in- 618:333–337.
clusions. Neuron 30:135–147.
Tseng HC, Zhou Y, Shen Y, Tsai LH (2002) A survey of Cdk5 activator p35
Noble W, Olm V, Takata K, Casey E, Mary O, Meyerson J, Gaynor K, LaFran- and p25 levels in Alzheimer’s disease brains. FEBS Lett 523:58 – 62.
cois J, Wang L, Kondo T, Davies P, Burns M, Veeranna, Nixon R, Dickson
D, Matsuoka Y, Ahlijanian M, Lau LF, Duff K (2003) Cdk5 is a key factor Tuppo EE, Arias HR (2005) The role of inflammation in Alzheimer’s dis-
in tau aggregation and tangle formation in vivo. Neuron 38:555–565. ease. Int J Biochem Cell Biol 37:289 –305.
Ojala PJ, Hirvonen TE, Hermansson M, Somerharju P, Parkkinen J (2007) Wender M, Adamczewska-Goncerzewicz Z, Szczech J, Godlewski A (1988)
Acyl chain-dependent effect of lysophosphatidylcholine on human neu- Myelin lipids in aging human brain. Neurochem Pathol 8:121–130.
trophils. J Leukoc Biol 82:1501–1509.
White JA, Manelli AM, Holmberg KH, Van Eldik LJ, Ladu MJ (2005) Dif-
Otth C, Concha II, Arendt T, Stieler J, Schliebs R, Gonza´lez-Billault C, Mac- ferential effects of oligomeric and fibrillar amyloid-beta 1– 42 on
cioni RB (2002) AbetaPP induces cdk5-dependent tau hyperphospho- astrocyte-mediated inflammation. Neurobiol Dis 18:459 – 465.
rylation in transgenic mice Tg2576. J Alzheimers Dis 4:417– 430.
Zhang Z, Lee YC, Kim SJ, Choi MS, Tsai PC, Saha A, Wei H, Xu Y, Xiao YJ,
Ousman SS, David S (2000) Lysophosphatidylcholine induces rapid recruit- Zhang P, Heffer A, Mukherjee AB (2007) Production of lysophosphati-
ment and activation of macrophages in the adult mouse spinal cord. Glia dylcholine by cPLA2 in the brain of mice lacking PPT1 is a signal for
30:92–104. phagocyte infiltration. Hum Mol Genet 16:837– 847.
Patrick GN, Zukerberg L, Nikolic M, de la Monte S, Dikkes P, Tsai LH (1999) Zheng YL, Kesavapany S, Gravell M, Hamilton RS, Schubert M, Amin N, Albers W,
Conversion of p35 to p25 deregulates Cdk5 activity and promotes neuro- Grant P, Pant HC (2005) A Cdk5 inhibitory peptide reduces tau hyperphos-
degeneration. Nature 402:615– 622. phorylation and apoptosis in neurons. EMBO J 24:209–220.
Poore CP, Sundaram JR, Pareek TK, Fu A, Amin N, Mohamed NE, Zheng YL, Zhu M, Gu F, Shi J, Hu J, Hu Y, Zhao Z (2008) Increased oxidative stress and
Goh AX, Lai MK, Ip NY, Pant HC, Kesavapany S (2010) Cdk5-mediated astrogliosis responses in conditional double-knockout mice of
phosphorylation of delta-catenin regulates its localization and GluR2- Alzheimer-like presenilin-1 and presenilin-2. Free Radic Biol Med 45:
mediated synaptic activity. J Neurosci 30:8457– 8467. 1493–1499.