Critical Role of TNF-a-Induced Macrophage VEGF and iNOS
Production in the Experimental Corneal
Peirong Lu,*,1–3Longbiao Li,1Gaoqin Liu,1Tomohisa Baba,3Yuko Ishida,4Mizuho Nosaka,4
Toshikazu Kondo,4Xueguang Zhang,2and Naofumi Mukaida*,3
(TNF)-a in alkali-induced corneal neovascularization (CNV).
evaluated the roles of tumor necrosis factor
METHODS. CNV was induced by alkali injury and compared in
wild-type (WT) BALB/c mice, and TNF receptor 1-deficient
(TNF-Rp55 KO) counterparts, or in mice treated with TNF-a
antagonist and recombinant TNF-a. Angiogenic factor expres-
sion and leukocyte accumulation in the early phase after injury
were quantified by real-time PCR and immunohistochemical
RESULTS. Alkali injury augmented the intraocular mRNA
expression of TNF-a and its receptor, together with a transient
macrophage and neutrophil infiltration. Compared to WT
mice, TNF-Rp55 KO mice exhibited reduced CNV. Intraocular
F4/80-positive macrophages and Ly-6G-positive neutrophils
infiltration did not change in KO mice compared to WT mice
after the injury. Alkali injury induced a massively increased
intraocular mRNA expression of angiogenic factors, including
vascular endothelial growth factor (VEGF), inducible nitric
oxide synthase (iNOS), interleukin (IL)-6, E-selectin, and
intercellular adhesion molecule (ICAM)-1 in WT mice, whereas
these increments were retarded severely in KO mice.
Immunofluorescence analysis demonstrated that F4/80-posi-
tive cells expressed VEGF and iNOS. Moreover, TNF-a
enhanced VEGF and iNOS expression by peritoneal macro-
phage from WT, but not KO mice. Topical application of TNF-a
antagonist reduced CNV, while topical application of recom-
binant TNF-a enhanced it.
CONCLUSIONS. TNF-Rp55-KO mice exhibited impaired alkali-
induced CNV through reduced intracorneal infiltrating macro-
phage VEGF and iNOS expression. (Invest Ophthalmol Vis Sci.
essential to visual acuity, and depends on the balance between
angiogenic and anti-angiogenic molecules.2–6Corneal neovas-
cularization (CNV) arises from various causes, including
corneal infections, misuse of contact lenses, chemical burns,
and inflammation,7–9and frequently can lead to impaired
vision. Before the onset of CNV, a large number of neutrophils
and, to a lesser degree, monocytes/macrophages infiltrate into
the cornea. We proved previously that experimental CNV can
occur independently of granulocyte infiltration.10Moreover,
we observed that infiltrated macrophages exert complicated
roles, by using different chemokine receptor signals in the
development of CNV.11–13
A proinflammatory cytokine, tumor necrosis factor-a (TNF-
a), is produced by a variety of cell types, including neutrophils,
macrophages, lymphocytes, and endothelial cells.14TNF signals
through two distinct cell surface receptors, TNF-receptor (TNF-
R)p55, and TNF-Rp75. TNF-Rp55 mediates the major biological
activities of TNF-a.15Based on their essential roles in inflamma-
tory responses, anti-TNF-a mAbs or TNF antagonists have been
used to treat inflammatory ocular diseases and uveitis.16
Although evidence is accumulating to indicate the essential
roles of the TNF-a-TNF-Rp55 axis in inflammatory responses,
the involvement of this axis in angiogenesis is defined poorly.
Saika et al. observed that alkali-induced CNV was more severe
in TNF-a-deficient mice than wild-type (WT) controls.17They
demonstrated further that endogenous TNF-a can counteract
the activities of transforming growth factor (TGF)-b and
vascular endothelial growth factor (VEGF) on vascular endo-
thelial cells, thereby inhibiting corneal neovascularization.18
On the contrary, TNF-a can induce endothelial cell migration in
vitro and angiogenesis in vivo, when implanted into corneas,
chorioallantoic membranes, or sponge implants.19–22In line
with these observations, Shi et al. reported that a TNF
antagonist, etanercept, and TNF antibody reduced laser-
induced choroidal neovascularization.23An anti-TNF mAb,
infliximab, was reported to be effective against age-related
macular degeneration (AMD),24which is characterized by
choroidal neovascularization. These conflicting observations
prompted us to investigate further the roles of TNF-a on ocular
neovascularization, by using a frequently used ocular neovas-
cularization model, alkali injury-induced CNV.4–6,10–13
he cornea is characterized by an absence of blood vessels
under physiological conditions.1Corneal avascularity is
From the1Department of Ophthalmology and2Clinical Immu-
nology Key Laboratory of Jiangsu Province, the First Affiliated
Hospital of Soochow University, Suzhou City, China; the3Division of
Molecular Bioregulation, Cancer Research Institute, Kanazawa
University, Kanazawa, Japan; and the
Medicine, Wakayama Medical University, Wakayama, Japan.
Supported by International Cooperative Program of Kanazawa
University (NM), National Natural Science Foundation in China
(NSFC) Grants 30771978 and 30972712(NSFC) Grants 30771978
and 30972712, Qing-Lan Project of Education Bureau of Jiangsu
Province, and Jiangsu Province’s Key Medical Talents Program Grant
Submitted for publication March 19, 2010; revised September
26, 2010, March 8, June 7, August 8, and December 4, 2011, and
March 27, 2012; accepted April 15, 2012.
Disclosure: P. Lu, None; L. Li, None; G. Liu, None; T. Baba,
None; Y. Ishida, None; M. Nosaka, None; T. Kondo, None; X.
Zhang, None; N. Mukaida, None
*Each of the following is a corresponding author: Peirong Lu,
Clinical Immunology Key Laboratory of Jiangsu Province, the First
Affiliated Hospital of Soochow University, 188 Shizi Street, Suzhou
215006, China PR; email@example.com.
Naofumi Mukaida, Division of Molecular Bioregulation, Cancer
Research Institute, Kanazawa University, 13-1 Takara-machi, Kana-
zawa 920-0934, Japan; firstname.lastname@example.org.
4Department of Forensic
Investigative Ophthalmology & Visual Science, June 2012, Vol. 53, No. 7
Copyright 2012 The Association for Research in Vision and Ophthalmology, Inc.
MATERIALS AND METHODS
Reagents and Antibodies
A specific TNF antagonist, etanercept, was purchased from Wyeth
Pharmaceutical (Osaka, Japan). Recombinant mouse interferon (IFN)-c
(catalog No. 485-MI/CF) and mouse VEGF ELISA Kit (MMV00) were
obtained from R&D Systems (Minneapolis, MN). Rat anti-mouse F4/80
(clone A3-1) mAb was acquired from Serotec (Oxford, UK). Rat anti-
mouse CD31 (MEC13.3), anti-mouse-Ly-6G (Clone IA8, catalog No.
551495) mAbs, and rabbit anti-inducible nitric oxide synthase (iNOS)
(catalog No. 610332) polyclonal antibodies, were purchased from BD
Pharmingen (San Diego, CA). Rabbit anti-mouse TNF-a (ab34674) and
anti-mouse TNF-Rp55 (ab19139) polyclonal antibodies were supplied
by Abcam (Cambridge, UK). Goat anti-mouse VEGF (sc-1836) polyclon-
al antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA).
VECTASHIELD Mounting medium with DAPI (H-1200) was obtained
from Vector Laboratories (Burlingame, CA). Recombinant mouse TNF-a
and IL-1b, and neutralizing rabbit anti-mouse TNF-a IgG were prepared
as described previously.25Lipopolysaccharide (LPS, Cat No. L8274) was
purchased from Sigma-Aldrich Chemical Co. (St. Louis, MO). Alexa
Fluor 594 or 488-labeled donkey anti-rat IgG Ab, Alexa Fluor 594 or 488-
labeled donkey anti-rabbit IgG Ab, and donkey anti-goat IgG Ab were
purchased from Invitrogen Life Technologies (Carlsbad, CA).
Pathogen-free BALB/c mice were obtained from Clea Japan (Yokohama,
Japan), and were designated as WT mice in the present experiments.
TNF-Rp55-deficient mice were backcrossed to BALB/c mice from eight
to 10 generations,26,27and bred under specific pathogen-free
conditions at the Animal Research Center of Kanazawa University
(Kanazawa, Japan). Seven- to 8-week-old male mice were used for the
experiments. All animal experiments were performed under specific
pathogen-free conditions in the Institute for Experimental Animals,
Kanazawa University, in accordance with the ARVO Statement for the
Use of Animals in Ophthalmic and Vision Research, and complied with
the standards set out in the Guidelines for the Care and Use of
Laboratory Animals of Kanazawa University.
Alkali-Induced Corneal Injury Model
Corneal injury was induced by placing a 2-mm filter disc saturated with
1 N NaOH onto the left eye of the mouse for 45 seconds as described
previously.10–13TNF-a, etanercept, and anti-TNF-a antibodies were
dissolved in 0.2% sodium hyaluronate (Sigma-Aldrich) immediately
before the topical application. In some experiments, the alkali-treated
eyes received 5 lL of TNF-a antagonist preparation dissolved in 0.2%
sodium hyaluronate at a concentration of 0.5 or 1 mg/mL, or 5 lL of
0.2% sodium hyaluronate as vehicle twice a day for 7 days immediately
after the alkali injury. In another series of experiments, the eyes were
treated with alkali for 40 seconds, and received 5 lL of recombinant
mouse TNF-a preparation or neutralizing rabbit anti-mouse TNF-a IgG
at a concentration of 100 lg/mL, or 5 lL of 0.2% sodium hyaluronate as
vehicle twice a day for 7 days immediately after the alkali injury. At the
indicated time intervals, mice were sacrificed and whole eyes were
removed. The eyes were snap-frozen in ornithine carbamoyltransferase
(OCT) compound for histological analysis, or the corneas were
removed and placed immediately into RNALate (Qiagen, Tokyo, Japan)
and kept at ?868C until total RNA extraction was performed. Each
experiment was repeated at least three times.
Histological and Immunohistochemical Analysis
The OCT-embedded tissues were cut into 8-lm thick slices, and the
fixed cryosections then were subjected to hematoxylin and eosin
staining. The sections were incubated overnight at 48C with rat anti-
mouse F4/80 antibody (1 lg/mL) and rat anti-mouse-Ly-6G (2.5 lg/mL)
to identify macrophages and neutrophils, respectively. Tissue sections
then were incubated with biotin-conjugated anti-rat immunoglobulin
antibodies as secondary antibodies. Other sets of sections were
incubated overnight at 48C with rabbit anti-TNF-a, rabbit anti-TNF-
Rp55, and goat anti-VEGF pAbs, to detect TNF-a, TNF-Rp55, and VEGF
expression, respectively. These slides were incubated with biotin-
conjugated anti-rabbit or anti-goat immunoglobulin antibodies as the
secondary antibodies. The immune complexes were detected by using
an ABC kit and a DAB Substrate Kit from Vector Laboratories, Inc.
according to the manufacturer’s instructions. Slides then were
counterstained with hematoxylin and mounted. The numbers of F4/
80- or Ly-6G-positive cells were counted at 200-fold magnification in
five randomly chosen fields of corneal sections from each animal,11–13
by an examiner with no prior knowledge of the experimental
procedures. The numbers of positive cells per mm2were calculated.
A Double-Color Immunofluorescence Analysis
A double-color immunofluorescence analysis was performed to
determine VEGF or iNOS expression by infiltrating macrophages.
Briefly, the fixed cryosections (8 lm thick) were incubated with PBS
containing 10% normal donkey serum and 1% BSA to reduce
nonspecific reactions. Thereafter, the sections were incubated with
the combinations of rat anti-Ly-6G and goat anti-VEGF, rat anti-F4/80
and goat anti-VEGF, rat anti-F4/80 and rabbit anti-iNOS, or rabbit anti-
TNF-a and goat anti-VEGFAbs overnight at 48C. After being rinsed with
PBS, for a double-immunofluorescence analysis using anti-VEGF, the
sections then were incubated with a combination of Alexa Fluor 594
donkey anti-rat IgG and Alexa Fluor 488 donkey anti-goat IgG (1/100)
for 45 minutes at room temperature in the dark. For double-
immunofluorescence analyses to detect macrophage iNOS expression,
the sections were incubated further with the combination of Alexa
Fluor 594 donkey anti-rat IgG and Alexa Fluor 488 donkey anti-rabbit
IgG (1/100) for 45 minutes at room temperature in the dark. For
double-immunofluorescence analyses using anti-TNF-a and anti-VEGF,
the sections were incubated further with the combination of Alexa
Fluor 594 donkey anti-rabbit IgG and Alexa Fluor 488 donkey anti-goat
IgG (1/100) for 45 minutes at room temperature in the dark. Finally, the
sections were washed with PBS and mounted with VECTASHIELD
Mounting medium with DAPI, and immunofluorescence was visualized
with a fluorescence microscope (Olympus, Tokyo, Japan). Images were
processed by using graphics software (Adobe Photoshop, version 7.0;
Adobe, Mountain View, CA).
Enumeration of Corneal Neovascularization
Corneal whole mount staining was performed and blood vessels in the
corneas were measured according to previous reports.2,28Corneal flat
mounts were rinsed in PBS, fixed in acetone, rinsed in PBS, blocked in
2% BSA, stained with rat anti-mouse CD31 (1:100; BD Pharmingen) at
48C overnight, and washed, and the corneas then were incubated with
Alexa Fluor 594 or Alexa Fluor 488 donkey anti-rat IgG (1/100) for 2
hours at room temperature in the dark and analyzed by microscope.
Digital pictures of the flat mounts were taken. Then, the area covered
by CD31 was measured morphometrically on these flat mounts using
NIH Image software (National Institutes of Health, Bethesda, MD). The
total corneal area was outlined using the innermost vessel of the limbal
arcade as the border. The total area of neovascularization then was
normalized to the total corneal area, and the percentage of the cornea
covered by vessels calculated. In another series of experiments, the
fixed cryosections (8 lm thick) were stained with rat anti-mouse CD31
(1:100; BD Pharmingen) at 48C overnight, washed, and then incubated
with biotin-conjugated anti-rat immunoglobulin antibody as secondary
antibodies. The immune complexes were detected by using an ABC kit
and a DAB Substrate Kit from Vector Laboratories, Inc. according to the
manufacturer’s instructions. Slides then were counterstained with
hematoxylin and mounted. The numbers and sizes of the CNV were
determined as described previously,11–13by an examiner with no
IOVS, June 2012, Vol. 53, No. 7
TNF in Corneal Neovascularization 3517
knowledge of the experimental procedures. Briefly, images were
captured with a digital camera and imported into Adobe Photoshop.
Then, the numbers of neovascular tubes per mm2, and the proportions
of CNV in the hot spots were determined using NIH Image analysis
software.11–13Most sections were taken from the central region of the
cornea. The numbers and areas of corneal neovascularization were
evaluated on at least two sections from each eye.
RNA Isolation and Real-Time PCR
Total RNAs were extracted from the corneas or cultured macrophages
with the use of RNeasy Mini Kit (Qiagen). The resultant RNA
preparations were treated further with ribonuclease-free deoxyribonu-
clease (DNase) I (Life Technologies Inc., Gaithersburg, MD) to remove
genomic DNA. The PCR solution contained 2 lL cDNA, the specific
primer set (0.2 lM final concentration), and 12.5 lL of SYBR Premix
Ex TaqTM(SYBR Premix Ex Taq Perfect Real Time PCR Kit; Takara) in a
final volume of 25 lL. The sequences of the PCR primer pairs are listed
in the Table. Quantitative PCR was performed on iCycler iQ Multi-Color
Real Time PCR Detection System (170-8740; Bio-Rad, Shanghai, China).
PCR parameters were initial denaturation at 958C for 1 minute,
followed by 40 circles of 958C for 5 seconds, and 608C for 30 seconds.
mRNA expression was normalized to the levels of b-actin mRNA.
Isolation and Culture of Murine Peritoneal
Peritoneal macrophages from WTor TNF-Rp55 KO mice were obtained
as described previously.11,12The cells were suspended in antibiotic-free
RPMI medium containing 10% fetal bovine serum (FBS), and incubated
in a humidified incubator at 378C in 5% CO2in 6-well cell culture plates
(Nalge Nunc International Corp., Naperville, IL). Two hours later, non-
adherent cells were removed, and the medium was replaced. The cells
then were stimulated with the indicated concentrations of murine TNF-
TNF-Rp55 (B) mRNA normalized to the levels of b-actin mRNA as described in Materials and Methods, and are shown with mean 6 SEM.
Representative results from 3 independent experiments are shown. *P < 0.05; **P < 0.01 vs. D0. (C) Whole eyes were obtained at 0, 2, 4, and 7
days after alkali injury, and processed for immunohistochemical analysis using anti-TNF-a (upper panels) and anti-TNF-Rp55 (lower panels)
antibodies. Representative results from 3–5 animals of each time point are shown. Arrows indicate the positive cells. Original magnification ·400.
Scale bar 50 lm.
The expression of TNF-a and its receptor, TNF-Rp55, in cornea after alkali injury. Real-time PCR was conducted to detect TNF-a (A) and
TABLE 1. Specific Sets of Primers of Real-Time PCR
Names Primer Sequences
F, forward primer; R, reverse primer.
3518 Lu et al.
IOVS, June 2012, Vol. 53, No. 7
a for 12 hours or the indicated concentrations of LPS, IFN-c or IL-1b for
24 hours. Total RNAs were extracted from the cultured cells and
subjected to real-time PCR as described above. In another series of
experiments, murine macrophages were seeded onto 12-well-plates at
5 · 105cells/well. After adhesion, the cells were stimulated with the
indicated concentrations of murine TNF-a for 12 hours in a 378C
incubator with 5% CO2. Supernatants were collected to determine
VEGF concentrations using a Mouse VEGF ELISA Kit (R&D Systems),
according to the manufacturer’s instructions.
The means and SEM were calculated for all parameters determined in
the study. Data were analyzed statistically using one-way ANOVA, or
two-tailed Student’s t-test. A value of P < 0.05 was considered
Intracorneal Expression of TNF-a, and its
Receptor, TNF-Rp55 after Alkali-Induced Corneal
We first examined the expression of TNF-a and its receptor,
TNF-Rp55, in corneas after alkali-induced corneal injury. TNF-a
mRNA barely was detectable in untreated eyes, but was
increased markedly after alkali injury (Fig. 1A). Concomitantly,
TNF-a protein was detected immunohistochemically in infil-
trating cells after alkali injury, but not in untreated eyes (Fig.
1C, upper panels). Moreover, alkali injury augmented markedly
the mRNA expression of TNF-Rp55 (Fig. 1B). Furthermore,
immunohistochemical analysis demonstrated that the infiltrat-
ed leukocytes and corneal epithelial cells expressed TNF-Rp55
(Fig. 1C, lower panels). The enhanced intracorneal expression
of TNF-a and TNF-Rp55 suggests the possible involvement of
the TNF-a-TNF-Rp55 interactions in alkali-induced CNV.
Impaired Alkali-Induced CNV in TNF-Rp55 KO Mice
We next explored the effects of genetic ablation of TNF-Rp55
on alkali-induced CNV. CNV was evident macroscopically in
WT mice 2 weeks after the injury, consistent with our previous
reports.11–13CNV at 2 weeks after the injury was attenuated
significantly in TNF-Rp55 KO mice compared to those of WT
mice as evidenced by corneal whole mount method (Figs. 2A,
2B). Immunohistochemical analysis using anti-CD31 antibodies
revealed similar tendencies in WTand TNF-Rp55 KO mice even
at microscopic levels (Figs. 2C, 2D). These observations
indicate that the TNF-a-TNF-Rp55 axis was indispensable for
2 weeks after alkali injury, stained by anti-CD31 antibodies to visualize blood vessels. Original magnification ·25. (B) Neovascularized corneal area
of total corneal area percent measured by whole mount staining 2 weeks after alkali injury. Value represents mean and SEM (n¼5–7 animals). **P <
0.01. (C) Corneal tissues were obtained at 2 and 4 weeks after the injury from WT (left panels) and TNF-Rp55-deficient (right panels) mice. Tissues
were stained with hematoxylin and eosin staining (upper panels) or immunostained with anti-CD31 antibodies (middle and lower panels), and
representative results from three independent experiments are shown. Arrows indicate the CD31-positive newly formed vessels. Original
magnification ·400. Scale bar 50 lm. (D) CNV numbers per mm2in whole section (left panel), CNV numbers per mm2in hot spots (middle panel),
and percent CNV areas in hot spots (right panel) were determined on corneas obtained from WT or KO mice 2 and 4 weeks after the injury. Each
value represents mean and SEM (n ¼ 5 animals). *P < 0.05; **P < 0.01 WT versus TNF-Rp55 KO mice.
Alkali-induced CNV. (A) Representative photographs of corneal flat mounts of WT (left panel) and TNF-Rp55-deficient (right panel) mice
IOVS, June 2012, Vol. 53, No. 7
TNF in Corneal Neovascularization 3519
Marginal Effects of TNF-Rp55 Deficiency on Alkali
Injury-Induced Intraocular Leukocyte Infiltration
We observed previously that Ly-6G-positive granulocytes and
F4/80-positive macrophages infiltrated injured corneas, reach-
ing their peak levels 2 and 4 days after the injury in WT mice,
respectively.11–13Monocytes/macrophages can be a rich
source of angiogenic factors,29–33while infiltrated granulocytes
have few roles in alkali-induced CNV.10Hence, we examined
the effects of TNF-Rp55 deficiency on macrophage and
granulocyte infiltration into the wounded corneas. Rare F4/
80-positive macrophages were observed in untreated corneas
as observed previously.11Ly-6G-positive granulocytes (data not
shown) and F4/80-positive macrophages infiltrated into
corneas to similar extents in WT mice compared to TNF-
Rp55 KO mice after the injury (Fig. 3). We observed further
that mRNA of a major macrophage-tropic chemokine, CCL2,
was enhanced in corneas of WT and TNF-Rp55 KO mice to
similar extents after alkali injury (Fig. 3C). Thus, the lack of
TNF-Rp55 has few apparent effects on the intraocular
infiltration of leukocytes in the present alkali injury-induced
CNV, probably because it has few effects on the expression of a
potent macrophage-tropic chemokine, CCL2.
Reduced Pro-Angiogenic and Adhesion Molecule
Expression in TNF-Rp55 KO Mice after Alkali
The balance between angiogenic and anti-angiogenic factors
determines the outcome of angiogenesis processes in various
situations. Hence, we examined the mRNA expression of
angiogenic and anti-angiogenic factors in corneas after alkali
injury. Alkali injury increased intraocular mRNA expression of
an angiogenic factor, TGF-b, and an anti-angiogenic molecule,
thrombospondin (TSP)-1 in WT and TNF-Rp55 KO mice to
similar extents (Fig. 4). After alkali injury, WT mice exhibited
augmented intraocular mRNA expression of other potent
angiogenic molecules, including VEGF, basic fibroblast growth
factor (bFGF), iNOS, interleukin (IL)-6, and adhesion mole-
cules, E-selectin and intercellular adhesion molecule (ICAM)-1,
and the enhanced expression of these molecules was
attenuated markedly in TNF-Rp55 KO mice (Fig. 4). On the
contrary, the expression of IL-1a and IL-1b was enhanced to
similar extents in WT and KO mice. Consistently with our
previous observations,10immunohistochemical analysis dem-
onstrated that infiltrating leukocytes and repopulating epithe-
lial cells after alkali injury expressed VEGF in the early phase
(Fig. 5A). A double immunofluorescence analysis demonstrated
that infiltrating macrophages (Fig. 5B), but not the Ly-6G-
positive granulocytes, mainly expressed VEGF. Moreover,
double-color immunofluorescence analysis detected the cells
expressing simultaneously VEGF and TNF-a (Fig. 5B), indicat-
ing that the infiltrating macrophages expressed VEGF and TNF-
a. We further revealed that F4/80-positive macrophage also
expressed iNOS (Fig. 5B). Thus, intraocularly infiltrated
macrophages expressed VEGF and iNOS, two potent angio-
Enhanced VEGF and iNOS Expression by Murine
Macrophages with TNF-a Stimulation
We observed that alkali injury enhanced intracorneal TNF-a
mRNA expression to similar extents in WT and TNF-Rp55 KO
mice (Fig. 4). Hence, we next examined the effects of TNF-a
on the expression of VEGF and iNOS. TNF-a enhanced
markedly the mRNA expression of VEGF and iNOS by
peritoneal macrophages from WT mice in a dose-dependent
manner (Figs. 6A, 6B). Moreover, TNF-a augmented VEGF
(left panels) and TNF-Rp55-deficient mice (right panels), two days (upper panels), four days (middle panels), and seven days (lower panels) after
the injury. Tissues were immunostained with rat anti-F4/80 mAb. Representative results from 5 individual animals are shown. Arrows indicate the
positive cells. Original magnification ·400. Scale bar 50 lm. (B) The numbers of infiltrated F4/80-positive macrophages were determined, as
described in Materials and Methods, and the mean and SEM are shown here (n ¼ 5). The data were judged as statistically insignificant when
compared between WT and TNF-Rp55 KO mice. (C) The mRNA of CCL2 to b-actin of WT (black bars) and TNF-Rp55-deficient mice (open bars)
were determined by quantitative RT-PCR analysis. Values represent means and SEM (n ¼ 3–5 animals).
Macrophage recruitment in the injured corneas of WT and TNF-Rp55-deficient mice. (A) Corneal tissues were obtained from WT mice
3520Lu et al.
IOVS, June 2012, Vol. 53, No. 7
protein production by macrophages (Fig. 6C), but TNF-Rp55
deficiency abrogated the responsiveness of macrophages to
TNF-a in terms of VEGF and iNOS expression (Figs. 6A–C).
Moreover, the intracorneal numbers of VEGF and F4/80
double-positive macrophages were reduced in TNF-Rp55 KO
mice compared to WT mice after injury (Fig. 5B). However,
LPS, IFN-c, or IL-1b induced WT- or TNF-Rp55 KO-derived
macrophages to express VEGF or bFGF to similar extents
(Figs. 6D–F), indicating that TNF-Rp55 deficiency did not
impair the capacity of macrophages to express VEGF or bFGF
as a whole. These observations would indicate that TNF-a can
induce macrophages to express potent angiogenic molecules,
VEGF, iNOS, and bFGF, solely through the interaction with
Effects of TNF-a and its Inhibitors on Alkali-
The genetic deletion of TNF-Rp55 gene may have compound
effects on the phenotypes observed after alkali injury. To
exclude this possibility, we first examined the effects of topical
application of TNF-a antagonist, etanercept, on alkali-induced
CNV of WT mice. Etanercept reduced alkali-induced CNV level
compared to control mice (Fig. 7). Because etanercept is a
fusion protein between the extracellular portion of human
TNF-Rp75 and human IgG Fc portion, it can block TNF-a and
TNF-b.16To clarify the roles of TNF-a more definitively, we
next examined the effects of recombinant mouse TNF-a or
neutralizing anti-mouse TNF-a topical application on alkali-
induced CNV of WT mice. Topical administration of mouse
recombinant TNF-a enhanced alkali-induced CNV when it was
administered locally in the early phase after alkali injury at the
concentration of 100 lg/mL (Fig. 8). Neutralizing anti-mouse
TNF-a consistently attenuated alkali-induced CNV when given
topically in the early phase after alkali injury (Fig. 8). These
observations indicate further that TNF-a may promote alkali-
induced CNV by activating the infiltrated macrophages to
produce potent angiogenic factors, VEGF and iNOS.
TNF-a is a proinflammatory cytokine produced by a variety of
cell types including neutrophils, macrophages, lymphocytes,
and endothelial cells.14There are two distinct types of TNF
receptors, TNF-Rp55 and TNF-Rp75. TNF-Rp55 is expressed
ubiquitously by most types of cells, whereas TNF-Rp75 usually
is restricted to some cell types and its expression should be
induced.16Thus, most of its biologic activities are mediated by
mice. Representative results from three independent experiments are shown. The mRNAs of VEGF, bFGF, TGF-b, TSP-1, iNOS, IL-6, E-selectin, ICAM-
1, IL-1a, IL-1b, and TNF-a to b-actin of WT (black bars) and TNF-Rp55-deficient mice (open bars) were determined as described in Materials and
Methods. All values represent means and SEM (n ¼ 3–5 animals). *P < 0.05;#P < 0.01 WT versus KO mice.
Quantitative RT-PCR analysis of pro-angiogenic and anti-angiogenic gene expression in the injured corneas of WT and TNF-Rp55 KO
IOVS, June 2012, Vol. 53, No. 7
TNF in Corneal Neovascularization 3521
The roles of TNF-a in neovascularization remain elusive.
This cytokine can induce endothelial cell migration in vitro,
and angiogenesis in vivo, when implanted into corneas,
chorioallantoic membranes, or sponge implants.19–22In
contrast, the deficiency of TNF-a gene can aggravate ocular
neovascularization after alkali injury or central cauteriza-
tion.17,18TNF-Rp55 KO mice exhibited no abnormalities in
physiologic retinal neovascularization, but showed defects in
pathologic retinal neovascularization.34We observed that a
higher dose of TNF-a (500 lg/mL) did not have any apparent
effects on alkali-induced CNV when applied topically (data not
shown). Thus, the effects of TNF-a may be dependent on its
local concentration, exposure duration, and the type and
growth state of the target cells.35,36
These discrepancies prompted us to define the roles of the
TNF-a-TNF-Rp55 axis in alkali-induced CNV. Here, we demon-
strated that genetic ablation of TNF-Rp55 gene, and the
administration of TNF antagonist or anti-TNF-a, attenuated
alkali-induced CNV. Moreover, in contrast to the effects of a
higher dose of TNF-a (500 lg/mL), topical administration of
mouse TNF-a (100 lg/mL) aggravated alkali-induced CNV.
These results were inconsistent with the previous observation
on TNF-a-deficient mice.17,18These discrepancies may be
explained by the differences in the mouse strains used in the
anti-VEGF antibodies. Representative results from four individual animals are shown. (B) Double-color immunofluorescence analysis of VEGF or
iNOS expression on F4/80-positive macrophages or Ly-6G-positive granulocytes, or TNF-a and VEGF expression from corneas 4 days after the injury.
The samples were immunostained with combinations of anti-VEGF and anti-Ly-6G, anti-VEGF and anti-F4/80, anti-iNOS and anti-F4/80, or anti-TNF-a
and anti-VEGF antibodies, as described in Materials and Methods, and observed with fluorescence microscopy. Signals were merged digitally in the
right panels. Representative results from 5 individual animals are shown. Arrows indicate the double-positive cells. Original magnification ·400.
Scale bar 50 lm.
(A) Whole eyes were obtained at the indicated time intervals after alkali injury and processed for immunohistochemical analysis using
3522Lu et al.
IOVS, June 2012, Vol. 53, No. 7
previous studies17,18and our present work. Moreover, in the
previous study, expression of other TNF-related genes, and the
effects of TNF antagonists or anti-TNF-a antibodies were not
examined. Thus, it cannot be excluded that genetic ablation of
TNF-a gene can result in augmentation of TNF-b expression. If
so, augmented TNF-b expression can induce neovasculariza-
tion by acting on TNF-Rp55. On the contrary, if a higher dose
of TNF-a might reduce TNF-b expression by a negative
feedback mechanism, reduced TNF-b expression might negate
the effects of TNF-a.
Normal corneas lack any vasculature, and physiological
corneal avascularity is maintained by the net balance between
pro-angiogenic and anti-angiogenic factors.2–6Alkali treatment
b, IL-6, and iNOS, as well as E-selectin and ICAM-1, the adhesion
molecules that have crucial roles in ocular neovasculariza-
tion.37,38Except TGF-b, the augmented expression of the
angiogenic factors and adhesion molecules was depressed in
TNF-Rp55-deficient mice compared to WT mice. On the
contrary, the expression of a potent anti-angiogenic factor,
time PCR was conducted as described in Materials and Methods. Representative results of macrophage VEGF (A) and iNOS (B) expression
normalized to the levels of b-actin mRNA after recombinant TNF-a stimulation from three independent experiments are shown here. (C) Murine
macrophages were stimulated with the indicated concentrations of TNF-a for 12 hours. VEGF production in the supernatants was detected with
ELISA as described in Materials and Methods. (D) Real-time PCR results of macrophage VEGF expression normalized to the levels of b-actin mRNA
after 24 hours of IFN-c (500 U/mL), LPS (100 ng/mL), or the combination of LPS and IFN-c stimulation. The representative results from three
independent experiments are shown. Each value represents the mean and SEM (n ¼ 3). *P < 0.05; **P < 0.01 compared to untreated. NS, no
significant difference for WT versus KO mice. (E) and (F) Real-time PCR results of macrophage VEGF and bFGF expression normalized to the levels
of b-actin mRNA after 24 hours of IL-1b (100 ng/mL) stimulation. All values represent means 6 SEM (n ¼ 6 experiments). *P < 0.05, versus
untreated WT macrophages; #P < 0.05 versus untreated KO macrophages.
VEGF and iNOS expression by murine peritoneal macrophages from WT mice or TNF-Rp55 KO mice after stimulation. (A) and (B) Real-
IOVS, June 2012, Vol. 53, No. 7
TNF in Corneal Neovascularization 3523
mice. Thus, it is likely that TNF-a induced CNV by augmenting
the expression of angiogenic factors and adhesion molecules.
Several lines of evidence indicate that macrophages can be
pro-angiogenic by producing angiogenic factors in ocular
neovascularization.29–33TNF-a and IL-1 can activate similar
intracellular signaling pathways, such as NF-jB and AP-1, and
can induce the expression of chemokines and adhesion
molecules, which are presumed to have crucial roles in the
recruitment of leukocytes, including monocytes/macrophages.
We recently observed that IL-1ra-deficient mice exhibited
enhanced intracorneal macrophage recruitment after alkali
injury.39Thus, enhanced IL-1 signal can augment macrophage
recruitment. Actually, we found that alkali injury increased
intraocular IL-1a and IL-1b mRNA expression in WT and TNF-
Rp55 KO mice to similar extents. Moreover, similar observa-
tions were obtained on mRNA expression of CCL2, a chemo-
kine crucially involved in macrophage infiltration. We assumed
at first that TNF-a may promote neovascularization by inducing
macrophage recruitment as we observed in the colon
carcinogenesis model.40In contrast to our expectation,
absence of TNF-Rp55 gene did not reduce alkali injury-induced
intracorneal macrophage infiltration, probably due to sustained
CCL2 expression. Thus, the TNF-a-TNF-Rp55 interactions did
at 2 weeks after alkali injury, as shown by the CD31-positive areas in the corneal flat mounts (Li, limbal vascular arcade). Original magnification
·100. Representative results from five animals from each group are shown. (B) Corneal tissues were obtained two weeks after the injury from WT
mice applied topically with TNF-a antagonist or vehicle, and were immunostained with anti-CD31 antibody. Representative results from three
independent experiments are shown. Arrows indicate the CD31-positive newly formed vessels. Original magnification ·200. Scale bar 50 lm. (C)
and (D) The CNV area measured by corneal whole mount staining (C), while percent CNV areas in hot spots measured by immunohistochemical
staining from corneal cryosections (D) were determined. Each value represents mean and SEM (n¼5 animals). *P < 0.05; **P < 0.01 compared to
The effects of topical TNF-a antagonist application on CNV. (A) Representative CNVof WT mice applied topically with TNF-a antagonist,
CNV. Neovascularized corneal area of total corneal area percent of WT
mice applied topically with TNF-a or anti-TNF-a at 2 weeks after alkali
injury measured by whole mount staining. Each value represents mean
and SEM. Representative results from 5–7 animals from each group are
shown here. *P < 0.05; **P < 0.01 compared to vehicle. Our study
demonstrates that TNF-Rp55-KO exhibited impaired alkali-induced
corneal neovascularization through reduced expression of VEGF and
iNOS by infiltrating macrophages. Anti-TNF therapy may be useful as an
angiogenic regulator for treating corneal diseases that exhibit
The effects of topical TNF-a and anti-TNF-a application on
3524Lu et al.
IOVS, June 2012, Vol. 53, No. 7
not have apparent profound effects on the infiltration of
macrophages, a rich source of angiogenic factors, in this
We observed that intraocularly infiltrated F4/80-positive
macrophages expressed two potent angiogenic factors, VEGF
and iNOS. Lee et al. reported that TNF-a can stimulate
macrophage to express VEGF.41Likewise, we observed that
TNF-a induced murine peritoneal macrophages from WT, but
not TNF-Rp55-deficient mice, to express VEGF and iNOS.
Consistently, VEGF expression was depressed in TNF-Rp55 KO-
derived F4/80-positive macrophages compared to WT-derived
ones. Moreover, because we detected TNF-Rp55 mRNA
expression in murine peritoneal macrophages (data not
shown), these observations would indicate that the TNF-a-
TNF-Rp55 axis is crucial to VEGF and iNOS production by TNF-
Rp55-expressing macrophages. However, TNF-Rp55 deficiency
failed to abrogate CNV completely, suggesting the contribution
of other mediators to CNV. The candidate molecules may be IL-
1a and IL-1b, because IL-1a and IL-1b can induce the
expression of VEGF and bFGF to similar levels in WT and
TNF-Rp55 KO-mouse-derived macrophages. This assumption is
supported further by the observation that alkali injury
enhanced, to similar extents, their intracorneal expression in
WT and TNF-Rp55 KO mice.
Double-color immunofluorescence analysis revealed that
the infiltrating F4/80-positive macrophages expressed VEGF
and TNF-a. M1 macrophages are a main producer of TNF-a,
while M2 macrophages can promote angiogenesis by produc-
ing VEGF.42However, in selected pathological conditions,
mixed types of macrophage phenotypes have been observed,42
particularly in resolution phase of tissue injury.43As we
conducted double-color immunofluorescence analysis 4 days
after the injury, when corneal lesions were in resolution phase,
M1-M2 mixed types of macrophages can be detected in the
Corneal epithelial cell migration also is indispensable to
healing after alkali injury by suppressing inflammatory corneal
angiogenesis.28Okada et al. observed that TNF-a inhibited
corneal epithelial cell migration.44Thus, decreased TNF-a
signal in TNF-Rp55-deficient mice can promote epithelial cell
migration, and eventually reduce CNV.
Corneal clarity is necessary for normal vision. Corneal
neovascularization frequently can lead to loss of corneal
transparency and impaired vision. Our study demonstrates
that TNF-Rp55-KO exhibited impaired alkali-induced CNV
through reduced expression of VEGF and iNOS by infiltrating
macrophages. Moreover, we observed that the treatment with
TNF-a antagonist or anti-TNF-a antibodies significantly de-
creased alkali-induced mouse corneal neovascularization. Thus,
anti-TNF therapy may be useful as an angiogenic regulator for
treating corneal diseases that exhibit neovascularization.
However, this requires further investigation.
1. Ambati BK, Nozaki M, Singh N, et al. Corneal avascularity is
due to soluble VEGF receptor-1. Nature. 2006;443:993–997.
2. Cursiefen C, Masli S, Ng TF, et al. Roles of thrombospondin-1
and -2 in regulating corneal and iris angiogenesis. Invest
Ophthalmol Vis Sci. 2004;45:1117–1124.
3. Gao G, Ma J. Tipping the balance for angiogenic disorders.
Drug Discov Today. 2002;7:171–172.
4. Zhang SX, Ma JX. Ocular neovascularization: implication of
endogenous angiogenic inhibitors and potential therapy. Prog
Retin Eye Res. 2007;26:1–37.
5. Edelman JL, Castro MR, Wen Y. Correlation of VEGF
expression by leukocytes with the growth and regression of
blood vessels in the rat cornea. Invest Ophthalmol Vis Sci.
6. Lai CM, Spilsbury K, Brankov M, Zaknich T, Rakoczy PE.
Inhibition of corneal neovascularization by recombinant
adenovirus mediated antisense VEGF RNA. Exp Eye Res.
7. Chang JH, Gabison EE, Kato T, Azar DT. Corneal neovascular-
ization. Curr Opin Ophthalmol. 2001;12:242–249.
8. Epstein RJ, Stulting RD, Hendricks RL, Harris DM. Corneal
neovascularization. Pathogenesis and inhibition. Cornea.
9. Klintworth GK, Burger PC. Neovascularization of the cornea:
current concepts of its pathogenesis. Int Ophthalmol Clin.
10. Lu P, Li L, Mukaida N, Zhang X. Alkali-induced corneal
neovascularization is independent of CXCR2-mediated neu-
trophil infiltration. Cornea. 2007;26:199–206.
11. Lu P, Li L, Kuno K, et al. Protective roles of the fractalkine/
CX3CL1-CX3CR1 axis against alkali-induced corneal neovas-
cularization through enhanced anti-angiogenic factor expres-
sion. J Immunol. 2008;108:4283–4291.
12. Lu P, Li L, Wu Y, Mukaida N, Zhang X. Essential contribution of
CCL3 to alkali-induced corneal neovascularization by regulat-
ing vascular endothelial growth factor production by macro-
phages. Mol Vis. 2008;14:1614–1622.
13. Lu P, Li L, Liu G, Van Rooijen N, Mukaida N, Zhang X. Opposite
roles of CCR2 and CX3CR1 macrophages in alkali-induced
corneal neovascularization. Cornea. 2009;28:562–569.
14. Baud V, Karin M. Signal transduction by tumor necrosis factor
and its relatives. Trends Cell Biol. 2001;11:372–377.
15. Chen G, Goeddel DV. TNF-R1 signaling: a beautiful pathway.
16. Rodrigues EB, Farah ME, Maia M, et al. Therapeutic
monoclonal antibodies in ophthalmology. Prog Retin Eye
17. Saika S, Ikeda K, Yamanaka O, et al. Loss of tumor necrosis
factor alpha potentiates transforming growth factor beta-
mediated pathogenic tissue response during wound healing.
Am J Pathol. 2006;168:1848–1860.
18. Fujita S, Saika S, Kao WW, et al. Endogenous TNFalpha
suppression of neovascularization in corneal stroma in mice.
Invest Ophthalmol Vis Sci. 2007;48:3051–3055.
19. Leibovich SJ, Polverini PJ, Shepard HM, Wiseman DM, Shively
V, Nuseir N. Macrophage-induced angiogenesis is mediated by
tumour necrosis factor a. Nature. 1987;329:630–632.
20. Mahadevan V, Hart IR, Lewis GP. Factors influencing blood
supply in wound granuloma quantitated by a new in vivo
technique. Cancer Res. 1989;49:415–419.
21. Gao B, Saba TM, Tsan MF. Role of a(v)b(3)-integrin in TNF-a-
induced endothelial cell migration. Am J Physiol Cell Physiol.
22. Nakao S, Kuwano T, Ishibashi T, Kuwano M, Ono M.
Synergistic effect of TNF-alpha in soluble VCAM-1-induced
angiogenesis through alpha 4 integrins. J Immunol. 2003;170:
23. Shi X, Semkova I, M¨ uther PS, Dell S, Kociok N, Joussen AM.
Inhibition of TNF-alpha reduces laser-induced choroidal
neovascularization. Exp Eye Res. 2006;83:1325–1334.
24. Theodossiadis PG, Liarakos VS, Sfikakis PP, Vergados IA,
Theodossiadis GP. Intravitreal administration of the anti-tumor
necrosis factor agent infliximab for neovascular age-related
macular degeneration. Am J Ophthalmol. 2009;147:825–830.
25. Fujioka N, Mukaida N, Harada A, et al. Preparation of specific
antibodies against murine IL-1ra and the establishment of IL-
1ra as an endogenous regulator of bacteria-induced fulminant
hepatitis in mice. J Leukoc Biol. 1995;58:90–98.
IOVS, June 2012, Vol. 53, No. 7
TNF in Corneal Neovascularization3525
26. Ishida Y, Kondo T, Tsuneyama K, Lu P, Takayasu T, Mukaida N. Download full-text
The pathogenic roles of tumor necrosis factor receptor p55 in
acetaminophen-induced liver injury in mice. J Leukoc Biol.
27. Mori R, Kondo T, Ohshima T, Ishida Y, Mukaida N. Accelerated
wound healing in tumor necrosis factor receptor p55-deficient
mice with reduced leukocyte infiltration. FASEB J. 2002;16:
28. Cursiefen C, Chen L, Saint-Geniez M, et al. Nonvascular VEGF
receptor 3 expression by corneal epithelium maintains
avascularity and vision. Proc Natl Acad Sci USA. 2006;103:
29. Sakurai E, Anand A, Ambati BK, Van Rooijen N, Ambati J.
Macrophage depletion inhibits experimental choroidal neo-
vascularization. Invest Ophthalmol Vis Sci. 2003;44:3578–
30. Grossniklaus HE, Ling JX, Wallace TM, et al. Macrophage and
retinal pigment epithelium expression of angiogenic cytokines
in choroidal neovascularization. Mol Vis. 2002;21:119–126.
31. Tsutsumi C, Sonoda KH, Egashira K, et al. The critical role of
ocular-infiltrating macrophages in the development of choroi-
dal neovascularization. J Leukocyte Biol. 2003;74:25–32.
32. Espinosa-Heidmann DG, Suner IJ, Hernandez EP, Monroy D,
Csaky KG, Cousins SW. Macrophage depletion diminishes
lesion size and severity in experimental choroidal neovascu-
larization. Invest Ophthalmol Vis Sci. 2003;44:3586–3592.
33. Oh H, Takagi H, Takagi C, et al. The potential angiogenic role
of macrophages in the formation of choroidal neovascular
membranes. Invest Ophthalmol Vis Sci. 1999;40:1891–1898.
34. Kociok N, Radetzky S, Krohne TU, Gavranic C, Joussen AM.
Pathological but not physiological retinal neovascularization is
altered in TNF-Rp55-receptor-deficient mice. Invest Ophthal-
mol Vis Sci. 2006;47:5057–5065.
35. Regatieri CV, Dreyfuss JL, Melo GB, Lavinsky D, Farah ME,
Nader HB. Dual role of intravitreous infliximab in experimen-
tal choroidal neovascularization. Effect on the expression of
sulfated glycosaminoglycans. Invest Ophthalmol Vis Sci. 2009;
36. Fiers W, Beyaert R, Boone E, et al. TNF-induced intracellular
signaling leading to gene induction or to cytotoxicity by
necrosis or by apoptosis. J Inflamm. 1995;47:67–75.
37. Moromizato Y, Stechschulte S, Miyamoto K, et al. CD18 and
ICAM-1-dependent corneal neovascularization and inflamma-
tion after limbal injury. Am J Pathol. 2000;157:1277–1281.
38. Zhu SN, Dana MR. Expression of cell adhesion molecules on
limbal and neovascular endothelium in corneal inflammatory
neovascularization. Invest Ophthalmol Vis Sci. 1999;40:1427–
39. Lu P, Li L, Liu G, Zhang X, Mukaida N. Enhanced experimental
corneal neovascularization along with aberrant angiogenic
factor expression in the absence of IL-1 receptor antagonist.
Invest Ophthalmol Vis Sci. 2009;50:4761–4768.
40. Popivanova BK, Kitamura K, Wu Y, et al. Blocking TNF-alpha in
mice reduces colorectal carcinogenesis associated with
chronic colitis. J Clin Invest. 2008;118:560–570.
41. Lee H, Baek S, Joe SJ, Pyo SN. Modulation of IFN-gamma
production by TNF-alpha in macrophages from the tumor
environment: significance as an angiogenic switch. Int
42. Sica A, Mantovani A. Macrophage plasticity and polarization: in
vivo veritas. J Clin Invest. 2012;1222:787–795.
43. Bystrom J, Evans I, Newson J, et al. Resolution-phase
macrophages possess a unique inflammatory phenotype that
is controlled by cAMP. Blood. 2008;112:4117–4127.
44. Okada Y, Ikeda K, Yamanaka O, et al. TNFalpha suppression of
corneal epithelium migration. Mol Vis. 2007;13:1428–1435.
3526Lu et al.
IOVS, June 2012, Vol. 53, No. 7