Tissue inhibitor of metalloproteinases-3 mediates the death of immature oligodendrocytes via TNF-α/TACE in focal cerebral ischemia in mice.

Abstract

Oligodendrocyte (OL) death is important in focal cerebral ischemia. TIMP-3 promotes apoptosis in ischemic neurons by inhibiting proteolysis of TNF-α superfamily of death receptors. Since OLs undergo apoptosis during ischemia, we hypothesized that TIMP-3 contributes to OL death.
Middle cerebral artery occlusion (MCAO) was induced in Timp-3 knockout (KO) and wild type (WT) mice with 24 or 72 h of reperfusion. Cell death in white matter was investigated by stereology and TUNEL. Mature or immature OLs were identified using antibodies against glutathione S-transferase-π (GST-π) and galactocerebroside (GalC), respectively. Expression and level of proteins were examined using immunohistochemistry and immunoblotting. Protein activities were determined using a FRET peptide.
Loss of OL-like cells was detected at 72 h only in WT ischemic white matter where TUNEL showed greater cell death. TIMP-3 expression was increased in WT reactive astrocytes. GST-π was reduced in ischemic white matter of WT mice compared with WT shams with no difference between KO and WT at 72 h. GalC level was significantly increased in both KO and WT ischemic white matter at 72 h. However, the increase in GalC in KO mice was significantly higher than WT; most TUNEL-positive cells in ischemic white matter expressed GalC, suggesting TIMP-3 deficiency protects the immature OLs from apoptosis. There were significantly higher levels of cleaved caspase-3 at 72 h in WT white matter than in KO. Greater expression of MMP-3 and -9 was seen in reactive astrocytes and/or microglia/macrophages in WT at 72 h. We found more microglia/macrophages in WT than in KO, which were the predominant source of increased TNF-α detected in the ischemic white matter. TACE activity was significantly increased in ischemic WT white matter, which was expressed in active microglia/macrophages and OLs.
Our results suggested that focal ischemia leads to proliferation of immature OLs in white matter and that TIMP-3 contributes to a caspase-3-dependent immature OL death via TNF-α-mediated neuroinflammation. Future studies will be needed to delineate the role of MMP-3 and MMP-9 that were increased in the Timp-3 wild type.

Full text

Page 1

RESEARCHOpen Access
Tissue inhibitor of metalloproteinases-3 mediates
the death of immature oligodendrocytes via TNF-
a/TACE in focal cerebral ischemia in mice
Yi Yang1, Fakhreya Y Jalal1, Jeffrey F Thompson1, Espen J Walker5, Eduardo Candelario-Jalil1, Lu Li2,
Ross R Reichard4, Chi Ben6, Qing-Xiang Sang6, Lee Anna Cunningham2and Gary A Rosenberg1,2,3*
Abstract
Background and Purpose: Oligodendrocyte (OL) death is important in focal cerebral ischemia. TIMP-3 promotes
apoptosis in ischemic neurons by inhibiting proteolysis of TNF-a superfamily of death receptors. Since OLs
undergo apoptosis during ischemia, we hypothesized that TIMP-3 contributes to OL death.
Methods: Middle cerebral artery occlusion (MCAO) was induced in Timp-3 knockout (KO) and wild type (WT) mice
with 24 or 72 h of reperfusion. Cell death in white matter was investigated by stereology and TUNEL. Mature or
immature OLs were identified using antibodies against glutathione S-transferase-π (GST-π) and galactocerebroside
(GalC), respectively. Expression and level of proteins were examined using immunohistochemistry and
immunoblotting. Protein activities were determined using a FRET peptide.
Results: Loss of OL-like cells was detected at 72 h only in WT ischemic white matter where TUNEL showed greater
cell death. TIMP-3 expression was increased in WT reactive astrocytes. GST-π was reduced in ischemic white matter
of WT mice compared with WT shams with no difference between KO and WT at 72 h. GalC level was significantly
increased in both KO and WT ischemic white matter at 72 h. However, the increase in GalC in KO mice was
significantly higher than WT; most TUNEL-positive cells in ischemic white matter expressed GalC, suggesting TIMP-3
deficiency protects the immature OLs from apoptosis. There were significantly higher levels of cleaved caspase-3 at
72 h in WT white matter than in KO. Greater expression of MMP-3 and -9 was seen in reactive astrocytes and/or
microglia/macrophages in WT at 72 h. We found more microglia/macrophages in WT than in KO, which were the
predominant source of increased TNF-a detected in the ischemic white matter. TACE activity was significantly
increased in ischemic WT white matter, which was expressed in active microglia/macrophages and OLs.
Conclusions: Our results suggested that focal ischemia leads to proliferation of immature OLs in white matter and that
TIMP-3 contributes to a caspase-3-dependent immature OL death via TNF-a-mediated neuroinflammation. Future
studies will be needed to delineate the role of MMP-3 and MMP-9 that were increased in the Timp-3 wild type.
Background
Oligodendrocytes (OLs) undergo a complex pattern of
death in cerebral ischemia with an early loss from gluta-
mate excitotoxicity and a slower death due to apoptosis
[1-4]. The vulnerability of OLs to ischemia was shown
morphologically [5] and the molecular mechanism was
related to excitotoxicity via the glutamate AMPA
receptors [6,7]. Ischemia results in apoptosis of OLs, but
the mechanisms involved are less well understood. The
pro-inflammatory factor tumor necrosis factor-a (TNF-
a) and TNF death receptors play an important role in
OL death as shown in vitro and in vivo [1,8,9]. Cerebral
ischemia results in an inflammatory response consisting
of microglia activation, gliosis, and cell death. Activated
microglia and reactive astrocytes produce TNF-a and
interleukin-1b (IL-1b), which are associated with peri-
ventricular white matter damage in hypoxic neonatal
brain [10]. Extravasation of inflammatory cells into the
* Correspondence: Grosenberg@salud.unm.edu
1Department of Neurology, University of New Mexico Health Sciences
Center, Albuquerque, NM 87131, USA
Full list of author information is available at the end of the article
Yang et al. Journal of Neuroinflammation 2011, 8:108
http://www.jneuroinflammation.com/content/8/1/108
JOURNAL OF
NEUROINFLAMMATION
© 2011 Yang et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons
Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in
any medium, provided the original work is properly cited.

Page 2

central nervous system (CNS) is facilitated by extracellu-
lar activities of matrix metalloproteinases (MMPs) that
are regulated, in part, by the endogenous tissue inhibi-
tors of metalloproteinases (TIMPs) [11].
Cell surface sheddases, including TNF-a converting
enzyme (TACE) and stromelysin-1 (MMP-3) regulate
the TNF superfamily of death receptors by activating
the ligands and removing the death receptors from the
cell surface. Tissue inhibitor of metalloproteinases-3
(TIMP-3) plays a central role in this process by inhibit-
ing TACE and MMP-3. In an early study, TIMP-3
blocked the release of the TNF death receptor by
TACE, promoting apoptosis [12]. More recently, we
observed that TIMP-3 expression in ischemic neurons
was associated with APO-1 (Fas/CD95) and DNA frag-
mentation [13]. Using neuronal cell cultures, we have
demonstrated that TIMP-3 promoted apoptosis by pre-
venting the shedding of Fas receptors from the neuronal
cell surface during oxygen-glucose deprivation, and that
the Timp-3 knockout (KO) mouse showed reduced neu-
ronal death after a middle cerebral artery occlusion
(MCAO) with reperfusion compared to the wild type
(WT) [14,15].
Because of TIMP-3’s unique role as a cell surface inhi-
bitor of TACE and MMP-3, it could contribute to cell
death in OLs by promoting retention of TNF death
receptors [16]. To investigate the role of TIMP-3 in OL
cell loss, we induced a middle cerebral artery occlusion
(MCAO) in Timp-3 knockout (KO) and wild type (WT)
mice followed by 24 and 72 h of reperfusion. We
hypothesized that TIMP-3 and Fas/CD95 expression
occurred early in neurons and preceded DNA fragmen-
tation, but that in white matter TIMP-3 may contribute
to OL death that is associated with TNF-a receptors.
We report that TIMP-3 deficiency protects the imma-
ture OLs from apoptotic death in white matter induced
by ischemia-reperfusion injury. The immature OL death
in the WT was associated with increased expression of
MMPs, TIMP-3, TNF-a/TACE in reactive astrocytes
and/or infiltrating macrophages/microglia in white
matter.
Methods
MCAO with reperfusion
The study was approved by the University of New Mex-
ico Animal Care Committee and conformed to the
National Institutes of Health Guide for the Care and
Use of Animals in research. Eighty-nine male Timp-3
KO (n = 44) and WT (n = 45) mice weighing 20-30 g
(age 10-12 weeks) were employed in this study. Fifty-
nine animals were subjected to 90 min MCAO using
the intraluminal thread method followed by 24 or 72 h
reperfusion. MCAO was performed as previously
described [15,17]. Mice underwent reperfusion via the
circle of Willis. Thirty sham-operated animals were
anesthetized and had a suture introduced into the com-
mon carotid artery and immediately removed.
Primary antibodies and dilutions used in
immunohistochemistry
Glial fibrillary acidic protein (GFAP) (1:400; Sigma), Iba-
1 (1:200; Wako Pure Chemical Industries), GST-π
(1:600, Abcam), GalC (1:300, Millipore), Fas (1:400;
Santa Cruz Biotechnology), Fas ligand (1:400; Santa
Cruz Biotechnology), MMP-2 (1:250; Chemicon), MMP-
3 (1:1500; Chemicon), MMP-9 (1:1000; Chemicon), cas-
pase-3 (1:1000, Cell Signaling), TIMP-3 (1:1000; Chemi-
con), TNF-a (15 μg/ml; R&D Systems), TNF Receptor I
(1:1000; Abcam), von Willebrand Factor (vWF) (10 μg/
ml; Chemicon), myeloperoxidase (MPO; 1:250; Santa
Cruz Biotechnology), and several markers for oligoden-
droglia; RIP (1:200,000; Chemicon), CNPase (15 μg/ml;
Chemicon), and CC-1 (APC; 10 μg/ml; Calbiochem).
Immunohistochemistry
Brain tissues fixed with 2% paraformaldehyde, 0.1 M
sodium periodate, 0.075 M lysine in 100 mM phosphate
buffer at pH 7.3 (PLP) were used for immunohistochem-
ical analysis. For immunofluorescence, 10 μm PLP-fixed
brain sections were treated with acetone and blocked
with 5% normal serum. Primary antibodies were incu-
bated for two nights at 4°C. Slides were incubated for 90
min at room temperature with secondary antibodies
conjugated with FITC or Cy-3 (Jackson). 4’-6-diami-
dino-2-phenylindole (DAPI; Molecular Probes) was used
to label cell nuclei.
For brightfield immunolabeling, 10 μm PLP-fixed sec-
tions were rehydrated through alcohol series and incu-
bated in phosphate buffered saline and 0.1% Tween-20
pH 7.3 (PBT) prior to blocking with 5% normal serum
in PBT. Slides were incubated with primary antibodies
for 48 h at 4°C followed by 1 h incubation with biotiny-
lated secondary antibodies, Vectastain Elite ABC
reagents (Vector Laboratories) and reacted with 3,3’-dia-
minobenzidine (DAB) was used for visualization of
immunolabeling. Cresyl violet acetate (CVA, Sigma-
Aldrich) was used for counterstaining.
Immunohistochemistry negative controls were incu-
bated without the primary antibody or with normal
(non-immune) IgGs and did not exhibit specific immu-
nolabelling. All immunohistochemically stained slides
were viewed on an Olympus BX-51 bright field and
fluorescence microscope (Olympus America Inc.)
equipped with an Optronics digital camera and Picture
Frame image capture software (Optronics). Dual immu-
nofluorescence slides were also imaged using high reso-
lution confocal microscopy to verify co-localization of
antigens (Zeiss LSM 510, Carl Zeiss Microimaging).
Yang et al. Journal of Neuroinflammation 2011, 8:108
http://www.jneuroinflammation.com/content/8/1/108
Page 2 of 14

Page 3

Immunofluorescence was quantified using ImageJ soft-
ware (NIH software).
Stereological Counts
The cellular composition of the white matter is com-
prised of > 90% OLs [18] consisting of OL progenitors,
immature, and mature OLs. Antibodies used to identify
mature OLs, including CC-1, RIP, and CNPase, either
labeled too few OL cell bodies or obscured the cell
bodies in the processes. To measure white matter
damage in mouse focal ischemia, we first assessed loss
of OL-like cells in white matter [2] using multiple-anti-
body immunostaining [19]. Labeling cells with a Nissl
substance stain, such as cresyl violet acetate (CVA), all
cells are clearly observable and an assessment of OL
density can be determined stereologically. To distinguish
OLs from astrocytes and microglia, we stained slides
with GFAP for astrocytes and Iba-1 for microglia, using
the DAB method. All slides were counterstained with
CVA. By eliminating DAB-stained astrocytes and micro-
glia, small CVA-positive cell bodies that were negative
for GFAP and Iba-1 immunoreactivity were identified as
OL-like cells and quantified using unbiased stereology
with StereoInvestigator (Version 6, MBF Bioscience)
software controlling a motorized stage equipped Olym-
pus BX-51 microscope [19]. Employing the optical frac-
tionator function of StereoInvestigator, we estimated
OL-like cell number in the medial corpus callosum (CC)
and external capsule (EC) regions of white matter over a
total distance of 3.0 mm rostral and caudal to bregma.
Tissue sections were blinded to the investigator as to
animal identity and reperfusion time. A subset of slides
was stained with myeloperoxidase or von Willebrand
factor to identify neutrophils and endothelial cells.
These could be distinguished from the OL-like cells
based on morphology (data not shown).
Western blots
Western blot was performed to determine protein levels
in ischemic white matter. Proteins were extracted in
RIPA buffer from punch biopsies of white matter in
ischemic and nonischemic external capsule regions of
Timp-3 KO and WT mice. 50 μg total protein was sepa-
rated on 10% or 12% gels. The proteins were transferred
to polyvinylidene fluoride (PVDF) membranes. The
membranes were then incubated with primary antibo-
dies: MBP (1:2000, Millipore), GST-π (1:5000, Abcam),
GalC (1:500, Millipore), MMP-3 (1:1000, Epitomics),
both inactive (pro-) and active (cleaved) caspase-3
(1:1000; Cell Signaling Technology), and TNFR1
(1:1000; Abcam). The membranes were incubated with
the respective secondary antibodies and blots were
developed using the West Pico Detection System
(Pierce). Protein bands were visualized on X-ray film.
Semiquantitation of target protein intensities was done
with the use of Scion image software (Scion), and Brilli-
ant Blue R (Sigma-Aldrich) staining on the same PVDF
membranes was used to normalize protein loading and
transfer. The results are reported as normalized band
intensity for quantifying relative protein expression.
TUNEL assay
Terminal deoxynucleotidyl transferase-mediated dUTP
nick end labeling (TUNEL) assay was done to identify
the extent of DNA fragmentation. Using the Neuro-
TACS II kit (Trevigen) brain sections were treated fol-
lowing the procedure specified by the manufacturer. For
dual immunofluorescence, the brain sections were then
exposed to the primary antibodies and streptavidin
fluorescein and Cy-3 secondary antibodies. For positive
controls the nuclease provided with the kit was added
to the labeling reaction mix and induced TUNEL posi-
tivity in all cells. As negative controls, slides were pre-
pared in a labeling reaction mix without the TdT
enzyme resulting in no TUNEL positivity (data not
shown).
Enzymatic assay of MMP-3 activity using a fluorescence
resonance energy transfer (FRET) peptide
MMP-3 enzymatic activity was assessed using a peptide-
based FRET assay. In the intact FRET peptide, Mca-
Arg-Pro-Lys-Pro-Val-Glu-Nva-Trp-Arg-Lys(Dnp)-NH2,
obtained from Bachem (Torrance, CA) the fluorescence
of Mca (7-methoxycoumarin-4-acetyl) is quenched by
Dnp (2,4-dinitrophenyl). Upon cleavage into two sepa-
rate fragments by the MMP-3 present in the sample, the
fluorescence of Mca was recovered, and monitored at
lex= 328 nm and lem= 393 nm using a spectrofluori-
meter (Perkin-Elmer LS 50B).
Brain tissue was homogenized in lysis buffer contain-
ing 50 mM Tris-HCl pH 7.6, 150 mM NaCl, 5 mM
CaCl2, 0.05% Brij-35, 0.02% NaN3, and 1% Triton X-
100, and centrifuged at 12,000 × g for 5 min at 4°C.
Equal amount of total protein (6-10 μg) was mixed with
assay buffer containing 50 mM HEPES buffer pH 7.5,
200 mM NaCl, 10 mM CaCl2, 0.01% Brij-35, and fluoro-
genic substrate (1 μM final concentration) to a final
volume of 200 μl. Fluorescence was monitored for 15
min. The relationship between fluorescence units and
nM of product produced was determined from the
fluorescence values obtained when all the substrate was
hydrolyzed.
Fluorometric assay of TNF-a converting enzyme (TACE)
activity
TACE activity was measured using the fluorometric
SensoLyte 520 TACE (a-Secretase) Activity Assay Kit
(AnaSpec) according to the manufacturer’s instructions.
Yang et al. Journal of Neuroinflammation 2011, 8:108
http://www.jneuroinflammation.com/content/8/1/108
Page 3 of 14

Page 4

SensoLyte™ 520 TACE Activity Assay Kit uses a 5-
FAM (fluorophore) and QXL™ 520 (quencher) labeled
FRET peptide substrate for continuous measurement of
enzyme activity. In the intact FRET peptide, the fluores-
cence of 5-FAM is quenched by QXL™ 520. Upon clea-
vage of the FRET peptide by the active enzyme, the
fluorescence of 5-FAM is recovered, and can be con-
tinuously monitored at lex= 490 nm/520 nm. TACE
activity was measured in the brain tissue homogenized
as described before for MMP-3 activity. Equal amount
of total protein was mixed with assay buffer (AnaSpec)
and TACE substrate to a final volume of 100 μl. Fluor-
escence was monitored for 60 min with data recorded
every 5 min. Fluorescence was expressed as relative
fluorescence units (RFU). Some brain samples from KO
and WT mice had 10 μl of 10 μM TACE inhibitor,
TAPI-0, added as a control.
Data analysis
Statistical comparisons among groups were done using
ANOVA with post-hoc analysis for multiple t- tests
(Prism 4.0, GraphPad Software Inc.). All data are pre-
sented as mean ± standard error of the mean (S.E.M.).
Statistical significance was set at p < 0.05.
Results
Timp-3 KO mice were protected from delayed immature
OL death
Two regions of ischemic white matter were analyzed:
the corpus callosum (CC) and the external capsule (EC).
No OL-like cell loss was seen in ischemic CCs in both
KO and WT mice at 24 and 72 h (data not shown) or
in ischemic EC at 24 h (Figure 1A, top two panels).
Representative regions from the nonischemic and the
ischemic EC at 72 h showed the normal appearing OL-
like cells in the nonischemic sides and the reactive
astrocytosis and/or microglia/macrophages in the
ischemic ECs in KO and WT (Figure 1A. bottom two
panels). The ischemic EC in KO showed no observable
loss of OL-like cells, while the ischemic EC in WT had
marked loss of OL-like cells (Figure 1A, bottom two
panels).
Next, the loss of OL-like cells in ischemic white mat-
ter was quantified by unbiased stereology. At 24 h, KO
and WT mice showed similar cell counts in the
ischemic and the nonischemic hemispheres (Figure 1B).
By 72 h, there was a significant loss of OL-like cells in
the ischemic EC of WT compared to KO mice (p <
0.01). Furthermore, there was a significant loss of cells
in the ischemic compared to the nonischemic hemi-
sphere in WT only (p < 0.05). In the CC, no significant
reduction of OL-like cells was detected in either WT or
KO ischemic CC at 72 h (data not shown). Next, we
investigated apoptotic cell death in ischemic EC at 72 h
of reperfusion by TUNEL assay. More TUNEL-positive
cells were detected in ischemic WT hemisphere includ-
ing the EC region compared with KO (Figure 1C). The
difference in the distribution of TUNEL-positive cells
between the ischemic KO and WT hemispheres is con-
sistent with the patterns of the neuronal degeneration
revealed by Fluoro-Jade staining (Figure 1D). This result
suggested that ischemia-reperfusion injury induced a
delayed cell loss in white matter in which TIMP-3 may
be involved.
We then studied the expression of TIMP-3 in astro-
cytes and OLs in WT mice. At 24 h, very little TIMP-3
was seen in the white matter. We observed a new popu-
lation of GFAP-positive astrocytes closely surrounding
the core infarct areas at 72 h (data not shown), which
has been reported by others [20]. Maximal TIMP-3
expression occurred at 72 h in GFAP-positive astrocytes
in ischemic EC (Figure 2A). Most of the astrocytes were
co-labeled with TIMP-3 and co-localization of TIMP-3
with CC1, an OL marker, was also seen by 72 h (Figure
2B). This temporal and spatial relationship of TIMP-3
expression and the loss of OL-like cells in ischemic EC
indicated the involvement of TIMP-3 in the white mat-
ter damage in focal cerebral ischemia.
To further investigate the role of TIMP-3 in OL death,
we examined the expression of myelin basic protein
(MBP) and glutathione S-transferase π (GST-π, or GST-
pi) [21], markers of mature OL, in ischemic ECs of KO
and WT mice at 72 h using Western blot analysis (Fig-
ure 3A). Unexpectedly, we found no significant changes
of MBP level either in ischemic ECs between KO and
WT mice or between ischemic ECs and sham ECs.
However, we observed a significant decrease of GST-π
level in WT ischemic EC compared with sham WT
mice. Next, we tested the expression of galactocerebro-
side (GalC), a marker of immature OL [22,23]. We
found that a significant increase in GalC was seen in
ischemic ECs of WT and KO by 72 h reperfusion after
ischemia compared with sham WT (p < 0.001) and
sham KO (p < 0.05), respectively. Most importantly,
TIMP-3 KO mouse shows significantly higher levels of
GalC expression in white matter than WT mouse (p <
0.01). These findings suggested that TIMP-3-mediated
delayed OL death mainly occurs in immature OLs in
white matter after ischemic stroke with reperfusion
injury.
To confirm the observation that TIMP-3 contributed
to delayed immature OL death in ischemic white matter,
we next performed immunohistochemistry of GST-π
and GalC on brain tissue from KO and WT mice with
90 min-MCAO and 72 h reperfusion (Figure 3B). Dou-
ble-immunostaining of GST-π and GalC demonstrated
stronger immunohistochemical signals of GST-π and
GalC in ischemic KO EC than WT EC. Furthermore, we
Yang et al. Journal of Neuroinflammation 2011, 8:108
http://www.jneuroinflammation.com/content/8/1/108
Page 4 of 14

Page 5

Figure 1 White matter injury and OL-like cell loss in EC at 72 h in Timp-3 KO and WT mice. A. Representative regions from the
nonischemic and the ischemic EC at 24 and 72 h. Iba-1 (microglia/macrophage) and GFAP (astrocyte) were both stained with DAB and
counterstained with cresyl violet acetate (CVA). The intact linear rows of dark-staining nuclei by CVA were formed by OL-like cells (arrows) in
nonischemic ECs at 24 and 72 h, as well as in ischemic ECs at 24 h. Ischemic EC shows massive DAB-positive glial cells at 72 h. A marked loss of
CVA positive cells was seen in ischemic EC in WT mouse compared with that in KO mouse. Scale bars = 50 μm. B. Stereological counting of OL-
like cells in ischemic EC at 24 and 72 h reperfusion shows OL-like counts were similar at 24 h between Timp-3 KO and WT ischemic. A significant
loss of OL-like cells occurred at 72 h in the WT ischemic EC compared to KO ischemic and to the WT nonischemic EC (p < 0.01 and 0.05,
respectively, n = 6). From 24 to 72 h, OLs in ischemic KO EC remained unchanged. The brain cartoon shows location of regions used for OL
counting (shaded areas). CC: corpus callosum; EC: external capsule. Micrographic image shows Iba-1 and GFAP (arrows) visualized by DAB in
higher magnification. C. TUNEL assay demonstrates apoptotic cell death in ischemic EC at 72 h reperfusion. TUNEL-positive cells were seen in
ischemic WT EC. Scale bar = 100 μm. D. Distribution of neuronal degeneration in ischemic hemispheres in Timp-3 KO and WT mice at 24 and 72
h of reperfusion following 90 min transient MCAO. Coronal sections were stained for degenerating neurons using Fluoro-Jade (FluoJ). Number of
FluoJ-positive cells was much greater in WT than in KO at both tiom points. La-ctx: lateral cortex, stri: striatum, pcf: piriform cortex. Scale bar =
100 μm.
Yang et al. Journal of Neuroinflammation 2011, 8:108
http://www.jneuroinflammation.com/content/8/1/108
Page 5 of 14