Embed Size (px)
Citation preview
Trans-arachidonic acids generated during nitrativestress induce a thrombospondin-1–dependentmicrovascular degenerationElsa Kermorvant-Duchemin1,9, Florian Sennlaub1,2,9, Mirna Sirinyan1,3, Sonia Brault1,3, Gregor Andelfinger1,Amna Kooli1,3, Stephane Germain4, Huy Ong5, Pedro d’Orleans-Juste6, Fernand Gobeil Jr6, Tang Zhu1,Chantal Boisvert1, Pierre Hardy1, Kavita Jain7, J Russel Falck8, Michael Balazy7 & Sylvain Chemtob1,3
Nitrative stress has an important role in microvascular degeneration leading to ischemia in conditions such as diabetic
retinopathy and retinopathy of prematurity. Thus far, mediators of nitrative stress have been poorly characterized. We recently
described that trans-arachidonic acids are major products of NO2�-mediated isomerization of arachidonic acid within the cell
membrane, but their biological relevance is unknown. Here we show that trans-arachidonic acids are generated in a model of
retinal microangiopathy in vivo in a NO�-dependent manner. They induce a selective time- and concentration-dependent apoptosis
of microvascular endothelial cells in vitro, and result in retinal microvascular degeneration ex vivo and in vivo. These effects are
mediated by an upregulation of the antiangiogenic factor thrombospondin-1, independently of classical arachidonic acid
metabolism. Our findings provide new insight into the molecular mechanisms of nitrative stress in microvascular injury and
suggest new therapeutic avenues in the management of disorders involving nitrative stress, such as ischemic retinopathies
and encephalopathies.
Ischemia-induced proliferative retinopathies such as diabetic retino-pathy and retinopathy of prematurity are characterized by microvas-cular degeneration and retinal ischemia, which can lead to secondaryaberrant neovascularization, hemorrhages and blindness1–3. Severalmechanisms involved in the loss of retinal vasculature have beenidentified. A downregulation of the proangiogenic vascular endothelialgrowth factor (VEGF) by hyperoxia or an inhibition of its receptorsignaling leads to microvascular degeneration4–7. Antiangiogenic fac-tors, such as thrombospondin (TSP)-1 are also important becausedeletion of TSP-1 protects against oxygen-induced vaso-obliteration8.Oxygen-derived free radicals contribute to microvascular injury2,9,and antioxidants inhibit microvascular degeneration in models ofdiabetic retinopathy10 and oxygen-induced retinopathy (OIR)2,11,12.These reactive oxygen species can react with nitric oxide (NO
�)
to generate highly reactive nitrogen species13, including peroxynitrite,nitrogen dioxide (NO2
�) and dinitrogen trioxide (N2O3),
whose deleterious effects on cell function are generally referred to asnitrative stress14. Several studies have indicated a crucial role for NO
�
and ensuing nitrative stress in ischemic retinopathies. In particular, ithas been shown that retinal microvascular degeneration is associated
with increased expression of endothelial nitric oxide synthase(eNOS)15–17, increased generation of nitrites, nitrates and peroxy-nitrite, and protein tyrosine nitration16–19, and can be preventedby pharmacological inhibition of NOS16,18 or deletion of Nos3,which encodes eNOS18. The mechanisms by which NO
�-derived
reactive species participate in microvascular injury, however, are notfully characterized.
We previously described a peroxidation process mediated by NO2�
that results in cis-trans-isomerization of arachidonic acid. This reac-tion produces four stable trans-arachidonic acid (TAA) monoisomers,5E-AA, 8E-AA, 11E-AA and 14E-AA20, two of them (5E-AA and8E-AA) being endogenous and not found in diet21. Furthermore, weshowed in a model of septic shock that isomerization of arachidonicacid by NO2
�occurs in vivo22. Thus far, the pathological relevance and
specific biological effects of these TAAs are unknown. We reasonedthat the increased nitrative stress upon oxygen exposure in OIR16,18
may predispose to isomerization of arachidonic acid, and thereforeinvestigated the generation of TAAs in an oxygen-induced model ofmicrovascular degeneration and their influence on retinal vasculatureand endothelial cell signaling and survival.
Received 21 January; accepted 1 November; published online 27 November 2005; doi:10.1038/nm1336
1Department of Pediatrics, Ophthalmology and Pharmacology, Research Center, Hopital Ste-Justine, 3175 Cote de Ste-Catherine, Montreal, Quebec, H3T1C5, Canada.2Inserm, Unite 598, Institut Biomedical des Cordeliers, 75006 Paris, France. 3Department of Pharmacology and Therapeutics, McGill University, 3655 PromenadeSir-William-Osler, Montreal, Quebec, H3G1Y6, Canada. 4Inserm, Unite 36, College de France, 75006 Paris, France. 5Faculty of Pharmacy, C.P. 6128, Succ. Centre-Ville, Universite de Montreal, Montreal, Quebec, H3C3J7, Canada. 6Department of Pharmacology, Universite de Sherbrooke, 3001 12e Avenue Nord Sherbrooke,Quebec, J1K2R1, Canada. 7Department of Pharmacology, New York Medical College, Valhalla, New York 10595, USA. 8Department of Biochemistry, University of TexasSouthwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, Texas 75390, USA. 9These authors contributed equally to this work. Correspondence should beaddressed to S.C. ([email protected]), otherwise to M.B. ([email protected]).
NATURE MEDICINE VOLUME 11 [ NUMBER 12 [ DECEMBER 2005 1339
ART ICL ES©
2005
Nat
ure
Pub
lishi
ng G
roup
ht
tp://
ww
w.n
atur
e.co
m/n
atur
emed
icin
e
RESULTS
TAAs increase after nitrative stress
NO2�
is a highly reactive free radical that originates in a number ofways in vivo, including through aerobic oxidation of NO
�(refs. 21,23)
and homolytic cleavage of peroxynitrite, which is formed byreaction of NO
�with superoxide (O2
�–)21,23,24. NO�, O2
�– and 3-nitrotyrosine, markers of peroxynitrite- and NO2
�-induced nitration
of proteins23,24, are abundantly generated in endothelial cells exposedto hyperoxia15 and in oxygen-induced retinopathy16,18. To evaluate thehypothesis that these conditions may favor isomerization of arachi-donic acid, we exposed postnatal day (P) 7 rat and mouse pups to 75–80% oxygen or room air for 24 h, and measured TAA levels in theretinas. Hyperoxia led to a significant increase in the concentration offree (unesterified to phospholipids, P o 0.001, n ¼ 6–13 animals pergroup) as well as total (free plus phospholipid-bound, P o 0.01, n ¼3–4 animals per group) TAA, reaching levels of 1.65 mM and 4.72 mM,respectively, in O2-exposed retinas (calculated from estimated retinalvolumes; Fig. 1a,b). To ascertain that the TAAs originated from anNO
�-dependent pathway, we treated oxygen-exposed animals with the
specific NO�
synthase (NOS) inhibitor N-nitro-L-arginine-methylester (L-NAME)18. Intraocular injection of L-NAME beforeoxygen exposure prevented the increase in TAA levels (Fig. 1a),indicating that their formation depends on NO
�and ensuing nitrative
stress in this model. Because eNOS is mainly responsible for thegeneration of nitrative stress in hyperoxia16,18, as neuronal NOS doesnot increase and inducible NOS is not expressed16, we tested the
contribution of eNOS on generation of TAAs in this model ofmicrovasculopathy using C57BL/6 mice with a disrupted Nos3 gene.TAA levels did not increase in retinas of Nos3–/– mice in contrast totheir wild-type counterparts (Fig. 1b).
TAAs inhibit angiogenesis and induce vaso-obliteration
We next evaluated the hypothesis that the nitrative stress–evokedincrease in TAA concentrations could be involved in the retinalmicrovascular damage observed in OIR. Concordant with the obser-vations on changes in TAA concentrations in OIR and the role ofeNOS presented above, we corroborated that NOS inhibition (usingL-NAME in rats) preserves retinal vasculature in the model of oxygen-induced vaso-obliteration in oxygen-exposed rat pups (Fig. 1c, P o0.01). Using concentrations of TAAs in the same range as those foundin OIR (Fig. 1a), we determined their effects on physiologic retinalvascularization in vivo as well as ex vivo on retinal explants. In rodents,retinal angiogenesis starts soon after birth25, and superficial micro-vessels reach the edges of the retina by P9 (ref. 25). We assessed theeffects of TAAs on normal angiogenesis using intraocular injections inP6 rat pups. Eyes injected with TAAs showed a reduced vascularizedretinal area (P o 0.01, n ¼ 14 eyes per group from three differentexperiments) and vascular density (P o 0.01) compared to vehicle-treated eyes, clearly showing that TAAs hinder physiological angiogen-esis and cause vaso-obliteration (Fig. 1d), similar to the changesobserved in the hyperoxic models16,18,26. To investigate the effects ofTAAs on microvascular degeneration in situ, we exposed newborn pigretinal explants27 to TAAs and assessed vascular density. Again,exposure to TAAs resulted in a significant decrease in vascular densitycompared to controls (Po 0.001, n ¼ 14–15 explants per group fromthree separate experiments; Fig. 1e).
TAAs induce selective cell death by apoptosis
To clarify the cell-type susceptibility thought to be involved inmicrovascular degeneration, we assessed effects of TAAs on thesurvival of different types of cultured neuromicrovascular cells.TAAs induced a concentration- and time-dependent microvascularendothelial cell death (Fig. 2a,b; EC50 at 24 h: 3–8 mM, consistent within vivo concentrations; Fig. 1b,c); concentrations of TAAs within thisrange were used thereafter. In contrast, smooth muscle, humanumbilical vein endothelial cells (HUVECs) and astrocytes were unaf-fected by exposure to TAAs (Fig. 2c–e), suggesting that TAAsspecifically lead to microvascular endothelial cell death.
We examined the nature of microvascular endothelial cell deathusing TUNEL staining for detection of nuclear DNA fragmentationand a caspase chromogenic substrate to detect caspase activity, bothof which are hallmarks of apoptosis. DNA fragmentation (P o0.0001) and caspase activity (P o 0.0001) were significantly
20
10
TA
As
(ng/
mg
prot
ein)
0
45
30
15
TA
A (
ng/m
g pr
otei
n)
0
a b
c
e
d
21% O280% O2
Ctl L-NAME
80% O2
Ctl L-NAME
CTL 14EP6
P9
0
0
25
20
40
60
80
100
120
50
75
100
75% O2
Nos3+/+
Nos3–/–
21% O2
Nos3+/+ Nos3
–/–
****
Vas
cula
r de
nsity
Vas
cula
r de
nsity
*
L-NAMEControl
Control
Ctl
14E-AA
Control 14E-AA
14E-AA
0
40
60
70
80
90
80
120
Vas
cula
r de
nsity
Ves
sel a
rea
Ctl 14E-AA
**
****
**
**
Figure 1 TAA levels rise secondary to nitrative stress and induce
retinal microvascular degeneration. (a) Free (unesterified) retinal
TAAs in room air–raised and oxygen-exposed rats treated with L-NAME.
Experiment performed on P8. (b) Total (free + bound (phospholipid-
esterified)) retinal TAAs in room air–raised and oxygen-exposed Nos3+/+ or
Nos3–/– mice. Experiment performed on P8. (c) Effect of NOS inhibitor
L-NAME on vascular density in oxygen-induced retinal microvascular
degeneration in the rat. Experiment performed on P9. Scale bar, 500 mm.
(d) In vivo effects of 14E-AA on retinal vascularization and vascular density.
Experiment performed on P9. Scale bar, 500 mm. (e) Ex vivo effects of
14E-AA on vascular density of retinal explants. Experiment performed on
P4. Scale bar, 500 mm. Values in histograms are mean ± s.e.m. *P o 0.05;
**P o 0.01. Ctl, control; 14E, 14E-AA.
ART ICL ES
1340 VOLUME 11 [ NUMBER 12 [ DECEMBER 2005 NATURE MEDICINE
©20
05 N
atur
e P
ublis
hing
Gro
up
http
://w
ww
.nat
ure.
com
/nat
urem
edic
ine
increased in endothelial cells treated with TAAs (Fig. 2f,g), indi-cative of apoptotic endothelial cell death. Consistently, the nuclearmorphology of Hoechst-stained endothelial cells treated withTAAs was altered, showing a pycnotic appearance in TUNEL-positivecells (Fig. 2g).
TAAs induce apoptosis through upregulation of TSP-1
To investigate the mechanisms underlying TAA-induced endothelialcell apoptosis, we first determined whether arachidonic acid metabolicpathways2 are involved; notably, roles for leukotrienes and prostanoidsin angiogenesis and microvascular damage are well documented2,28,29.Inhibition of cyclooxygenase 1, cyclooxygenase 2 and leukotrienereceptor did not prevent TAA-induced cell death (SupplementaryFig. 1 online). TAAs only slightly increased concentrations of throm-boxane (Supplementary Fig. 1 online), reported to exert retino-vascular endothelial cytotoxicity28; correspondingly, inhibitors ofthromboxane A2 synthase and receptor did not affect TAA-inducedendothelial cell death (Supplementary Fig. 1 online). Agonists ofprostanglandin I2 and E2 receptors, previously shown to possessprosurvival properties2,29, did not counter the cytotoxic effects ofTAAs (Supplementary Fig. 1 online). Furthermore, inhibition of thecytochrome P450 monooxygenase pathway, which can metabolizeTAAs30, did not influence TAA-evoked cytotoxicity (SupplementaryFig. 1 online). Thus, pro–cell-death effects of TAAs seemed to beindependent of classical major arachidonic acid metabolic pathways.
Because pro- and antiangiogenic factors have been shown to havean important role in the pathogenesis of oxygen-induced retinopathy3,we determined whether TAAs influenced their expression in endothe-lial cells. TAAs caused a rapid (within 2–6 h) increase in the expressionof the antiangiogenic TSP-1 in cultured endothelial cells (Fig. 3a).
Expression of VEGF receptor 2 (VEGFR2), shown to contribute toendothelial cell apoptosis and vessel loss in ischemic retinopathies4–7,was found to exhibit a relatively delayed (B12 h) time-dependentdecrease in response to TAA (Fig. 3a). We investigated the effect ofTSP-1 on VEGFR2 expression; TSP-1 depressed expression ofVEGFR2 after exposure for 4 h, probably by inducing proteaseactivation and its degradation31 (Fig. 3b). Therefore, the delayeddecrease in expression of VEGRF2 is likely to occur secondarily tothe TAA-induced upregulation of TSP-1. TSP-1 is a large matricellularprotein that is synthesized and secreted by several cell types includingendothelial cells32. TSP-1 inhibits angiogenesis by inducing micro-vascular endothelial cell apoptosis31–33 and inhibiting the migration ofthese cells32. The ability of TSP-1 to trigger endothelial cell apoptosisrequires the activation of its transmembrane receptor CD36 (refs.34,35) and is essential to its antiangiogenic activity in vivo35. Toestablish a cause-and-effect relationship between TSP-1 upregulationand TAA-induced endothelial cell apoptosis, we used function-block-ing antibodies directed against TSP-1 and CD36. Both TSP-1–specificantibody and antibody to CD36 that specifically blocks the TSP-1binding site fully prevented TAA-induced endothelial cell death(Fig. 3c); in contrast, cell death elicited by TAAs was not altered byan antibody to CD36 that blocks the binding site of oxidized low-density lipoprotein (Fig. 3c). We also tested the involvement of TSP-1in retinal explants treated with TAAs. The vaso-obliterative effects ofTAA under these conditions were prevented by both TSP-1– andCD36-specific antibodies (Fig. 4a). To further corroborate the anti-angiogenic role of TSP-1 and CD36 in response to TAAs, we tested theeffects of TAAs on vascular sprouting from Matrigel-embedded aorticrings isolated from TSP-1– and CD36-knockout mice; sproutingaortic endothelium (von Willebrand factor (vWF) positive) containedimmunoreactivity to CD36 (Fig. 4b). TAAs interfered with vascularsprouting in aortas of wild-type mice, whereas this effect was virtuallyundetected in aortas of mice with disruptions in the genes encodingTSP-1 and CD36 (Fig. 4b, P o 0.01). In line with these observations,the absence of an effect of TAAs on HUVECs (Fig. 2d) is consistentwith their lack of CD36 receptors36. Taken together, the increase inTSP-1 expression observed before endothelial cell death, the apoptoticnature of endothelial cell death induced by TAAs and abrogation ofcytotoxic effects of TAAs with antibodies to either TSP-1 or CD36 aswell as in mice deficient in these proteins strongly point to a TSP-1–dependent pathway.
We also conducted experiments to determine mechanisms under-lying TAA-induced upregulation of TSP-1. Involvement of the mito-gen-activated protein kinase (MAPK) pathway has been shown inmany apoptotic processes37. Recent studies have indicated that a tran-sient activation of ERK1/2 (also known as MAPK p42/44) contributesto induction of apoptosis in several cell types, whereas basal, sustainedactivity is required for the maintenance of cell survival38,39. We foundERK1/2 to be transiently phosphorylated in response to exposure to
f
g
edc
a b
TAAs (log mol/l)
**
Cel
l via
bilit
y
120
80
40
0−10 −9 −8 −7 −6 −5 −4
Ctl 14E-AA
Pos
itive
cel
ls
40
20
0
**
Ctl 14E-AA
Pos
itive
cel
ls40
20
0
TAAs (log mol/l)
Cel
l via
bilit
y
120
80
40
0−10 −9 −8 −7 −6 −5 −4
TAAs (log mol/l)
Cel
l via
bilit
y
120
80
40
0−10 −9 −8 −7 −6 −5 −4
TAAs (log mol/l)
Cel
l via
bilit
y120
80
40
0−10 −9 −8 −7 −6 −5 −4
TAA exposure (h)
Cel
l via
bilit
y
120
80
40
0483624120
Control 14E-AA
Control 14E-AA
5E-AA8E-AA11E-AA14E-AA
Figure 2 TAAs induce selective neuromicrovascular endothelial cell death
by apoptosis. Concentration-dependent effect of TAAs (5E-, 8E-, 11E- and
14E-AA) on cell viability: (a) microvascular endothelial cells, (c) smooth
muscle cells, (d) HUVECs and (e) astrocytes. Time-dependent effect (b) of
TAA (14E) on microvascular endothelial cell viability. Values (percent of
control) are mean ± s.e.m. of five to six experiments (a,b) and three separate
experiments (c–e), each performed in triplicate. (f) Caspase activity assay of
14E-AA–treated microvascular endothelial cells. (g) TUNEL assay of 14E-
AA–treated microvascular endothelial cells. Inset shows morphological
changes in TUNEL-positive cells. Scale bar, 250 mm. Values in histograms
are mean ± s.e.m. of percent total cell count from three different
experiments performed in duplicate. Ctl, control. **P o 0.01.
ART ICL ES
NATURE MEDICINE VOLUME 11 [ NUMBER 12 [ DECEMBER 2005 1341
©20
05 N
atur
e P
ublis
hing
Gro
up
http
://w
ww
.nat
ure.
com
/nat
urem
edic
ine
TAAs in endothelial cells (Fig. 3d). Furthermore, blockage of theMEKK-ERK1/2 pathway with the specific MAPK kinase inhibitorPD98059 markedly suppressed TAA-induced upregulation of TSP-1 aswell as TAA-induced endothelial cell death (Fig. 3e,f); inhibition ofthe other two MAPK pathways involved in stress and inflammation,namely p38 and JNK, had no effect on these parameters (data notshown). Hence, ERK1/2 seems to be required for TAA signaling.
TAAs induce TSP-1 ensuing vaso-obliteration in vivo
Oxygen-induced retinopathy is associated with vaso-obliteration,resulting in part from nitrative stress16,18 (Fig. 1a), which increases
formation of TAAs21,22 (Fig. 1b,c). TAAs in turn cause retinalendothelial degeneration in vitro, ex vivo and in vivo (Figs. 1d,e and2a,f,g), through induction of TSP-1 (Figs. 3c and 4a,b). We thereforeinvestigated in this oxygen-induced model of ischemic retinopathy theinvolvement of TSP-1 and CD36 in microvascular degeneration. Wefirst showed that intravitreal injection of TAAs indeed caused anupregulated expression of TSP-1 in vivo, which was specificallylocalized on the endothelium, as shown by immunohistochemistry(Fig. 5a); on the other hand, there was no change in the expression oftotal retinal VEGFR2 expression, which is found in numerous celltypes at that stage of development7. An increase in TSP-1 expression(protein and mRNA) was also detected in retinas of rats and miceexposed to hyperoxia for 24 h (Fig. 5b,c) along with the rise in TAAlevels (Fig. 1b,c); TSP-1 was localized on endothelium in this modelalso (Fig. 5d). Correspondingly, this increase in TSP-1 (Tsp1 mRNAexpression detected by quantitative PCR and corrected for that ofVwf) was significantly attenuated in Nos3–/– compared to wild-typeoxygen-exposed mice (Fig. 5c; P o 0.01, n ¼ 6 per group). Finally,to verify that oxygen-induced changes in TAAs (Fig. 1b,c) and ensuingupregulation of TSP-1 (Fig. 5b) are physiologically relevant, westudied retinal vascularization in hyperoxia-exposed mice deficientin CD36; vaso-obliteration was significantly attenuated in Cd36–/–
a
c
e f
d
b
Ctl
Ctl
Ctl
Ctl 14E
PD98059 + 14E
5′ 10′ 15′ 20′ 30′
Ctl
Ctl14E
TSP-1
VEGFR2
TSP-1
TSP-1
Ctl
β-actin
β-actin
β-actin
pERK
ERK
5E 8E 11E 6 h 2 h
1 h 1 h 4 h 4 h
6 h 12 h 18 h 24 h 24 h14E
VEGFR2
****
*
Cel
l via
bilit
y
Cel
l via
bilit
y
100
80
60
40
20
0Ctl 14E
80
120
40
0TSPAb CD36Ab
PD98059Vehicle14E-AA
TSP-1oxLDL
+14E-AA
+––+–++–
++––
Figure 3 TAAs induce endothelial cell death by an ERK 1/2-dependent
upregulation of TSP-1. (a) Representative western blot of TSP-1 expression
in endothelial cells exposed to each TAA for 18 h (left), and of TSP-1 and
VEGFR2 immunoreactivity in response to 14E-AA for different durations
(right). (b) Representative western blot of VEGFR2 expression in TSP-1–
exposed endothelial cells. (c) Effects of TSP-1– (TSPAb) and two CD36-
specific (CD36Ab) antibodies (blocking either the TSP-1 or oxidized
low-density lipoprotein (oxLDL) binding sites of CD36) on TAA-induced
endothelial cell viability. Values in histogram are mean ± s.e.m. of three
separate experiments performed in triplicate. **P o 0.01 compared with
control (Ctl) and other values. (d) Representative western blot of ERK 1/2
phosphorylation in endothelial cells exposed to 14E-AA. (e) Representative
western blot of TSP-1 expression in the presence or absence of 14E-AA and
the MEKK-ERK1/2 pathway inhibitor PD98059. Immunoblots for a,b,d and
e are representative of three separate experiments. (f) Effect of PD98059 onTAA-induced endothelial cell death. Values are mean ± s.e.m. of three
separate experiments performed in triplicate. *P o 0.05 compared with
other values.
a b
**
Ctl 14E
+14E
TSPAb CD36Ab
14E +TSP-1Ab 14E+CD36Ab 14E Tsp1−/− 14E Cd36−/−
Vas
cula
r de
nsity
100
150
50
0
Control 14E ControlvWF CD36 Merge
14E
***
Ctl 14E Ctl 14E Ctl 14EAre
a of
gro
wth
(da
ys 4
−6)
1.5
2.0
1.0
0.5
0.0
Tsp1−/− Cd36−/−+/+
Figure 4 TAAs induce a TSP-1–dependent microvascular degeneration and inhibit of angiogenesis in tissue explants. (a) Representative retinal explants from
newborn pigs treated with 14E-AA in the presence or absence of TSP-1– (TSP-1Ab) and CD36-specific (CD36Ab) antibodies. Scale bar, 250 mm. Values in
histogram are mean ± s.e.m. of four to five explants per treatment group. **P o 0.01. (b) Representative microvascular sprouting from Matrigel-embedded
aortic rings from wild-type, Tsp1–/– and CD36–/– mice exposed to 14E-AA (left panels). Scale bar, 1 mm. CD36 expression in microvascular sprouts (right
colored panels); representative of two experiments. Scale bar, 100 mm. Values in histograms are mean ± s.e.m. of 19–20 explants per group. ***P o 0.01
compared with control (Ctl) and other values.
ART ICL ES
1342 VOLUME 11 [ NUMBER 12 [ DECEMBER 2005 NATURE MEDICINE
©20
05 N
atur
e P
ublis
hing
Gro
up
http
://w
ww
.nat
ure.
com
/nat
urem
edic
ine
mice (Fig. 5e, P o 0.001, n ¼ 8 per group); a similar attenuation wasreported in hyperoxia-exposed Tsp1–/– mice8.
DISCUSSION
The recently described markers of nitrative stress and major productsof NO2
�-mediated isomerization of arachidonic acid, TAAs, represent
a new aspect of NO2�-induced toxicity. We investigated whether and
how endogenous TAAs contribute to endothelial cell degeneration. Wefound that TAAs are generated in vivo in an experimental model ofoxygen-induced microvascular degeneration, reaching concentrationsin the micromolar range. The nitrative stress dependency of TAAformation was shown by using the NOS inhibitor L-NAME andNos3–/– animals, which prevented the increase of TAAs upon exposureto oxygen and associated vaso-obliteration. Moreover, TAAs werefound to induce microvascular endothelial cell death in vitro, ex vivoand in vivo.
Previous studies have shown that isomerization of arachidonic acidoccurs in vivo. TAAs are found to circulate as free acids in humanplasma40 and have been shown to increase in a model of septicshock22, a condition that is correlated with nitrative stress41,42. Ourdata provide an additional pathological condition in which TAAs aregenerated in substantial amounts. We now show that these uniquelipids are biologically active molecules and important contributingfactors to microvascular damage as seen in ischemic retinopathies.This was demonstrated by showing that TAAs induce selectiveapoptosis of microvascular endothelial cells in vitro, and exert specific,species-independent cytotoxicity on retinal microvessels in bothex vivo and in vivo models.In vivo, TAAs derive solely from the reaction of the radical NO2
�
with arachidonic acid21,22, probably involving the binding of NO2�
to
arachidonic acid followed by a rearrangement of the nitroarachidonylradical, loss of NO2
�and formation of a trans bond21. Thus, they are
specifically produced under conditions associated with nitrative stress,such as inflammation and ischemia; this inference is supported in ourstudy using pharmacological inhibitors of NOS and Nos3–/– mice.Under these pathological conditions, NO
�-mediated cell death is
thought to be caused by the reaction of NO�-derived free radicals,
in particular peroxynitrite, N2O3, and NO2�, with biological targets
such as DNA, lipids and proteins13. The proposed mechanismsinclude inactivation of mitochondrial respiration43, modulationof gene expression44, nitration of DNA and protein13,24 and peroxida-tion of lipids13,23. Although all the previously described effectsof NO
�-derived free radicals concentrated on direct modifications of
target molecules such as protein and DNA, which may altertheir function, no secondary signaling molecule has so far beenidentified. By showing that TAAs are released during nitrativestress and induce microvascular damage mediated by upregulationTSP-1 in retinal microvascular endothelial cells, our resultsidentify TAAs as new mediators of nitrative stress in vivo and unravela mechanism by which reactive nitrogen species contribute tomicrovascular degeneration in ischemic retinopathies. Because gen-eration of NO
�-derived free radicals has been suggested to be involved
in the pathophysiology of numerous diseases41, we proposethat our findings could possibly be extrapolated to explain micro-vascular damage occurring in diabetes, ischemic encephalopathiesand inflammation.
Our data indicate that TAA-induced cell death and microvasculardegeneration are mediated by TSP-1; in addition, upregulation of thelatter seems to depend on activation of ERK1/2. Although ERK1/2 isusually regarded as a cytoprotective kinase, whereas the two othermain MAPKs, p38 and JNK, are generally considered to be mediatorsof cell death37, a transient activation of ERK1/2 has been shown toinduce apoptosis, whereas sustained activity favors cell survival38,39.Thus, our results provide additional evidence that ERK1/2 can haveopposing effects on cell death and survival, depending on the kineticsand amplitude of its activation and the cellular environment, andprovide new insight on the regulation of TSP-1 expression. Participa-tion of TSP-1 in oxygen-induced microvascular injury has been shownby the demonstration that TSP-1–deficient mice are less susceptible tooxygen-induced vaso-obliteration8. In agreement with this finding, wefound that retinal expression of TSP-1 is enhanced after 24 h ofhyperoxia in OIR, before considerable vascular degeneration wouldnormally be observed in this model25, and mice deficient in CD36show less vaso-obliteration upon exposure to hyperoxia. Together, ourdata indicate TAA-induced expression of TSP-1 as an important
*
Nos3 +/+ Nos3–/–
75% O2
O2
Tsp
1-1/
Vw
f 2.0
1.5
2.5
1.0
0.5
0.0
Merge
Merge
Anti -TSP
Anti -TSP
Lectin
Lectin
β-actin
TSP-1
6 h
21% 80% 21% 80%
24 h
β-actin
TSP-1
VEGFR2
**
Cd36+/+ Cd36 –/–
Cd36 +/+ Cd36 –/–
75% O2
Cap
illar
y dr
op-o
ut
0.3
0.2
0.4
0.1
0.0
Ctl 14Ea
b
d
e
c
Figure 5 TAA-induced upregulation of TSP-1 in vivo and importance in
hyperoxia-induced retinal microvascular degeneration. (a) Representative
western blots of TSP-1 and VEGFR2 expression in rat pup retinas injected
with TAAs (left). TSP-1 was localized in microvessels (detected by lectin
staining) in retinas exposed to TAAs. Scale bar, 250 mm. TAAs were
injected on days 1 and 4 and retinas were isolated on day 5 for western blot
and immunofluorescent staining. Experiments were performed three times.
(b) Western blot of TSP-1 in rat pups exposed to hyperoxia or normoxia for
6 and 24 h; representative of four rats in each group and time. (c) Tsp1
mRNA expression corrected for Vwf (as a measure of endothelialization) in
retinas of wild-type and Nos3–/– mice exposed to hyperoxia for 24 h.
(d) Microvascular localization of TSP-1 in retinas of wild-type mouse pups
exposed to hyperoxia for 24 h. Panel representations of two experiments
are similar to those in a. Scale bar, 125 mm. (e) Vaso-obliteration in Cd36+/+
and Cd36–/– mice exposed to hyperoxia for 24 h. Scale bar, 1 mm.*P o 0.05; **P o 0.01.
ART ICL ES
NATURE MEDICINE VOLUME 11 [ NUMBER 12 [ DECEMBER 2005 1343
©20
05 N
atur
e P
ublis
hing
Gro
up
http
://w
ww
.nat
ure.
com
/nat
urem
edic
ine
pathway responsible for subsequent microvascular damage, and alsoprovide evidence for a previously undescribed link between nitrativestress and TSP-1, notably involving ERK1/2.
The signaling pathway leading from TAA-induced activation ofMAPK to upregulation of TSP-1 remains undefined. We found thatexpression of p53 and PPARg, which are thought to be involved inregulation of TSP-1 (refs. 45,46), were not modulated by TAAs. On theother hand, previous studies suggest that trans-fatty acids alter cellmembrane function by modifying its fluidity and permeability, lead-ing to activation of membrane-bound channels and enzymes21.Nevertheless, the specificity of the effect of TAAs on expression ofTSP-1 suggests that they might exert their effect through an as yetunknown receptor. Further characterization of the molecular mechan-isms leading to TAA-evoked induction of TSP-1 is required and iscurrently under investigation.
The formation of TAAs is subject to the redox potential of thetissue. Accordingly, an immature subject would be more vulnerable asa result of its decreased antioxidant capacity47–49. We have recentlyshown that nitration, which leads to generation of TAAs20, occurswhen the redox state of the developing tissue is shifted toward anoxidative environment16. Our current findings, primarily observed innewborn animals, concur with this inference. There is currently noavailable effective treatment to prevent microvascular damage inischemic retinopathies and microangiopathies. By identifying a newmechanism of microvascular injury, our data provide new therapeuticstrategies in these seriously debilitating diseases associated withnitrative stress. Thus, interventions targeting the formation of TAAsor NO2
�may have potential therapeutic applications in conditions
such as ischemic retinopathies and other microangiopathies.
METHODSAnimals. All animals were used according to a protocol of the Hopital Sainte-
Justine Animal Care Committee. We purchased 1–3-d-old piglets from Fermes
Menard and Sprague-Dawley rats from Charles River. Nos3–/–, Cd36–/– and
Tsp1–/– mice and corresponding wild-type mice were provided by P. D’Orleans-
Juste (Universite de Sherbrooke), M. Febbraio (Weill Medical College, Cornell
University) and J. Lawler (Beth Israel Deaconess Medical Center), respectively.
Animal preparation and quantitation of retinal vascularization. We main-
tained P7 rat pups, Nos3+/+ and Nos3–/– mice, and Cd36+/+ and Cd36–/– mice of
C57BL/6 background in room air or exposed them to 80% (rats) or 75%
(mice) oxygen for 24 h (Oxycycler A82OCV, Biospherix) to generate an
oxygen-induced retinopathy. We intravitreally injected rat pups with L-NAME
(Sigma; 10–5 M final intraocular concentration based on estimated eye
volume29) before exposure to oxygen. We intravitreally injected other rat pups
at P6, P7 and P8 with either vehicle or 14E-AA (estimated final concentration
of 5 � 10–6 M)29.
Explants of retinal tissue can be cultured for several days and retain their
proliferation and differentiation capacities and their histological architecture27.
We cultured retinas of newborn (1–2 d old) pigs for 3 d on a Nucleopore
polycarbonate Track Etch membrane (pore size 0.03 mm, Whatman) placed on
DMEM (Gibco) containing 1% FBS and 1% penicillin-streptomycin, in the
presence or absence of 14E-AA (5 � 10–6 M), CD36-specific (clone FA6-152, 10
mg/ml, Immunotech) or TSP-1–specific (clone A4.1, 30 mg/ml, Neomarkers)
antibodies. We performed immunohistochemistry for whole retinas and
explants using thrombospondin Ab-3 antibody (1:100, Oncogene) and lectin
Griffonia simplicifolia conjugated to TRITC (1:100, Sigma) as previously
described29. We flat-mounted and photographed retinas and explants; we
measured vascular density as capillary length per retinal surface area
(mm/mm2); we measured vascularized areas and expressed them as percent
of total retinal area16,29.
TAA measurement. We dissected retinas and placed them into a 1:1 chloro-
form/methanol solution containing butylated hydroxytoluene to extract lipids.
We measured TAA (both esterified to lipids and free) content by mass
spectrometry analysis20,21 and normalized it to protein content.
Cell cultures. We cultured endothelial cells and astrocytes from newborn pig
brain and retina28,50. We purchased VSMCs and HUVECs from Clonetics.
Assessment of cell viability. We exposed cells to each TAA for 24 h at indicated
concentrations or to 14E-AA (5 � 10–6 M) for different durations and assessed
cell viability by MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium
bromide, Sigma) assay28. In another set of experiments, we pretreated cells
with TSP-1–specific antibody (clone A4.1, 30 mg/ml, Neomarkers), CD36-
specific antibodies (clones FA6-152 (which blocks TSP-1 binding), 10 mg/ml,
Immunotech, and JC63.1 (which blocks binding of oxidized low-density
lipoprotein), 2 mg/ml, Cayman Chemicals) and PD98059 (MEKK inhibitor,
10–5 M, Calbiochem) 30 min before exposing them to 14E-AA (5 � 10–6 M) for
24 h. We calculated cell viability as percent of control values.
Apoptosis assays. We cultured endothelial cells for 18 h in the presence or
absence of 14E-AA (5 � 10–6 M). We performed detection of DNA fragmenta-
tion with a commercial kit based on TUNEL (TACS TdT kit, R&D Systems).
We assessed caspase activity using a commercial kit (Sulforhodamine Multi-
caspase Activity kit, Biomol Research Laboratories). We calculated values as
positive cells as a percent of total cells.
Western blots. We exposed lysates of endothelial cells to 14E-AA (5 � 10–6 M)
in the presence or absence of PD98059 (10–5 M) or to TSP-1 (10–7 M, provided
by H. Ong, Universite de Montreal) and we used the membrane fraction of rat
pup retinas for western blot analysis of protein expression28. We used the
following antibodies: monoclonal TSP-1–specific (clone Ab-1, 1:400,
Oncogene), VEGFR2-specific (1:250, Chemicon International), polyclonal
ERK 1/2- and phosphorylated ERK 1/2-specific (1:5,000, Promega), and mono-
clonal b-actin–specific (1:60,000, Abcam).
Microvascular sprouting from aortic explants. We prepared aortae from
10-week-old male C57BL/6 wild-type, Tsp1–/– and Cd36–/– mice of the same
background. We covered aortic rings with 50 ml Matrigel and cultured them for
4 d in EGM-2 medium (Clonetics). We exposed explants to 14E-AA (5 � 10–6
M) from day 4 to 6 of culture. We took photomicrographs of individual
explants at day 4 and day 6 and measured microvascular sprouting as the
surface covered at day 6 minus that at day 4. We characterized outgrowing cells
by double-labeling them with monoclonal vWF-specific (1:100, Dako) and
polyclonal CD36-specific (1:100, Santa Cruz) antibodies after acetone fixation.
Real-time quantitative RT-PCR. We exposed Nos3+/+ and Nos3–/– mice of
C57BL/6 background to 75% oxygen from P7 to P8. We dissected retinas and
performed RNA extraction using the Qiagen RNA extraction kit (Qiagen). For
cDNA synthesis, we used 2 mg mRNA in all reactions. We performed real-time
quantitative PCR on MJ4000 and MJ3005 thermocyclers (Stratagene), using
5¢-GTTTGTGCAGCAGAGGAACA-3¢ and 5¢-CATTCACCCTGGCTCTTCTC-3¢for Vwf and 5¢-GCCATGTCCTCTCACAGGGC-3¢ and 5¢-CTGTGGCCACTGG
GAGATTAGC-3¢ for Tsp1. We standardized RNA levels to parallel measure-
ments of S16 RNA.
Data analysis. Data were analyzed by Kruskall-Wallis and Mann-Whitney tests
according to the number of groups compared. P o 0.05 was considered to be
statistically significant.
Note: Supplementary information is available on the Nature Medicine website.
ACKNOWLEDGMENTSThe authors wish to thank H. Fernandez for her technical skills and help. Thiswork was supported by grants from the Canadian Institutes of Health Research,the March of Dimes Birth Defects Foundation, the Heart and Stroke Foundationof Quebec, the Fonds de la Recherche en Sante du Quebec, Le Reseau deRecherche en Sante de la Vision and La Fondation du NO. E.K.-D. is recipient ofa fellowship award from the ‘Association des Juniors en Pediatrie/Gallia’ (France).F.S. and S.C. are recipients of fellowship and scientist awards, respectively, fromthe Canadian Institutes of Health Research. S.B. and M.S. are recipients ofstudentships from the Canadian Institutes of Health Research and Heart andStroke Foundation of Canada, respectively. P.H. is supported by grants from the
ART ICL ES
1344 VOLUME 11 [ NUMBER 12 [ DECEMBER 2005 NATURE MEDICINE
©20
05 N
atur
e P
ublis
hing
Gro
up
http
://w
ww
.nat
ure.
com
/nat
urem
edic
ine
Hospital For Sick Children Foundation and Fonds de la Recherche en Sante duQuebec. M.B. is supported by grants from the US National Institutes of Health(R01 GM62453) and Philip Morris USA, Inc. S.C. also holds a Canada ResearchChair (perinatology). The authors thank M. Febbraio and J. Lawler, who providedthe CD36 and TSP-1 knockout animals, respectively.
COMPETING INTERESTS STATEMENTThe authors declare that they have no competing financial interests.
Published online at http://www.nature.com/naturemedicine/
Reprints and permissions information is available online at http://npg.nature.com/
reprintsandpermissions/
1. Lee, P., Wang, C.C. & Adamis, A.P. Ocular neovascularization: an epidemiologic review.Surv. Ophthalmol. 43, 245–269 (1998).
2. Hardy, P. et al. Oxidants, nitric oxide and prostanoids in the developing ocularvasculature: a basis for ischemic retinopathy. Cardiovasc. Res. 47, 489–509 (2000).
3. Campochiaro, P.A. Retinal and choroidal neovascularization. J. Cell. Physiol. 184,301–310 (2000).
4. Alon, T. et al. Vascular endothelial growth factor acts as a survival factor for newlyformed retinal vessels and has implications for retinopathy of prematurity. Nat. Med. 1,1024–1028 (1995).
5. Pierce, E.A., Foley, E.D. & Smith, L.E. Regulation of vascular endothelial growth factorby oxygen in a model of retinopathy of prematurity. Arch. Ophthalmol. 114, 1219–1228 (1996).
6. Smith, L.E. et al. Regulation of vascular endothelial growth factor-dependent retinalneovascularization by insulin-like growth factor-1 receptor. Nat. Med. 5, 1390–1395(1999).
7. Shih, S.C., Ju, M., Liu, N. & Smith, L.E. Selective stimulation of VEGFR-1 preventsoxygen-induced retinal vascular degeneration in retinopathy of prematurity. J. Clin.Invest. 112, 50–57 (2003).
8. Wang, S., Wu, Z., Sorenson, C.M., Lawler, J. & Sheibani, N. Thrombospondin-1-deficient mice exhibit increased vascular density during retinal vascular developmentand are less sensitive to hyperoxia-mediated vessel obliteration. Dev. Dyn. 228,630–642 (2003).
9. Spierer, A., Rabinowitz, R., Pri-Chen, S. & Rosner, M. An increase in superoxidedismutase ameliorates oxygen-induced retinopathy in transgenic mice. Eye 19, 86–91(2005).
10. Kowluru, R.A., Tang, J. & Kern, T.S. Abnormalities of retinal metabolism in diabetesand experimental galactosemia. VII. Effect of long-term administration of antioxidantson the development of retinopathy. Diabetes 50, 1938–1942 (2001).
11. Penn, J.S., Tolman, B.L. & Bullard, L.E. Effect of a water-soluble vitamin E analog,trolox C, on retinal vascular development in an animal model of retinopathy ofprematurity. Free Radic. Biol. Med. 22, 977–984 (1997).
12. Raju, T.N., Langenberg, P., Bhutani, V. & Quinn, G.E. Vitamin E prophylaxis to reduceretinopathy of prematurity: a reappraisal of published trials. J. Pediatr. 131, 844–850(1997).
13. Squadrito, G.L. & Pryor, W.A. Oxidative chemistry of nitric oxide: the roles ofsuperoxide, peroxynitrite, and carbon dioxide. Free Radic. Biol. Med. 25, 392–403(1998).
14. Kroncke, K.D. Mechanisms and biological consequences of nitrosative stress. Biol.Chem. 384, 1341 (2003).
15. Gu, X. et al. Hyperoxia induces retinal vascular endothelial cell apoptosisthrough formation of peroxynitrite. Am. J. Physiol. Cell Physiol. 285, C546–C554(2003).
16. Beauchamp, M.H. et al. Redox-dependent effects of nitric oxide on microvascularintegrity in oxygen-induced retinopathy. Free Radic. Biol. Med. 37, 1885–1894(2004).
17. El-Remessy, A.B., Abou-Mohamed, G., Caldwell, R.W. & Caldwell, R.B. High glucose-induced tyrosine nitration in endothelial cells: role of eNOS uncoupling and aldosereductase activation. Invest. Ophthalmol. Vis. Sci. 44, 3135–3143 (2003).
18. Brooks, S.E. et al. Reduced severity of oxygen-induced retinopathy in eNOS-deficientmice. Invest. Ophthalmol. Vis. Sci. 42, 222–228 (2001).
19. El-Remessy, A.B. et al. Experimental diabetes causes breakdown of the blood-retinabarrier by a mechanism involving tyrosine nitration and increases in expression ofvascular endothelial growth factor and urokinase plasminogen activator receptor. Am.J. Pathol. 162, 1995–2004 (2003).
20. Jiang, H. et al. Nitrogen dioxide induces cis-trans-isomerization of arachidonic acidwithin cellular phospholipids. Detection of trans-arachidonic acids in vivo. J. Biol.Chem. 274, 16235–16241 (1999).
21. Balazy, M. & Poff, C.D. Biological nitration of arachidonic acid. Curr. Vasc. Pharmacol.2, 81–93 (2004).
22. Balazy, M. Trans-arachidonic acids: new mediators of inflammation. J. Physiol.Pharmacol. 51, 597–607 (2000).
23. Kirsch, M., Korth, H.G., Sustmann, R. & de Groot, H. The pathobiochemistry ofnitrogen dioxide. Biol. Chem. 383, 389–399 (2002).
24. Prutz, W.A., Monig, H., Butler, J. & Land, E.J. Reactions of nitrogen dioxide in aqueousmodel systems: oxidation of tyrosine units in peptides and proteins. Arch. Biochem.Biophys. 243, 125–134 (1985).
25. Chan-Ling, T., Gock, B. & Stone, J. The effect of oxygen on vasoformative cell division.Evidence that ‘physiological hypoxia’ is the stimulus for normal retinal vasculogenesis.Invest. Ophthalmol. Vis. Sci. 36, 1201–1214 (1995).
26. Smith, L.E. Pathogenesis of retinopathy of prematurity. Growth Horm. IGF Res. 14Suppl A, 140–4 (2004).
27. Caffe, A.R. et al. Mouse retina explants after long-term culture in serum free medium.J. Chem. Neuroanat. 22, 263–273 (2001).
28. Beauchamp, M.H. et al. Role of thromboxane in retinal microvascular degeneration inoxygen-induced retinopathy. J. Appl. Physiol. 90, 2279–2288 (2001).
29. Sennlaub, F. et al. Cyclooxygenase-2 in human and experimental ischemic proliferativeretinopathy. Circulation 108, 198–204 (2003).
30. Roy, U., Loreau, O. & Balazy, M. Cytochrome P450/NADPH-dependent formationof trans epoxides from trans-arachidonic acids. Bioorg. Med. Chem. Lett. 14,1019–1022 (2004).
31. Nor, J.E. et al. Thrombospondin-1 induces endothelial cell apoptosis and inhibitsangiogenesis by activating the caspase death pathway. J. Vasc. Res. 37, 209–218(2000).
32. Armstrong, L.C. & Bornstein, P. Thrombospondins 1 and 2 function as inhibitors ofangiogenesis. Matrix Biol. 22, 63–71 (2003).
33. Guo, N., Krutzsch, H.C., Inman, J.K. & Roberts, D.D. Thrombospondin 1 and type Irepeat peptides of thrombospondin 1 specifically induce apoptosis of endothelial cells.Cancer Res. 57, 1735–1742 (1997).
34. Dawson, D.W. et al. CD36 mediates the in vitro inhibitory effects of thrombospondin-1on endothelial cells. J. Cell Biol. 138, 707–717 (1997).
35. Jimenez, B. et al. Signals leading to apoptosis-dependent inhibition of neovasculariza-tion by thrombospondin-1. Nat. Med. 6, 41–48 (2000).
36. Swerlick, R.A., Lee, K.H., Wick, T.M. & Lawley, T.J. Human dermal microvascularendothelial but not human umbilical vein endothelial cells express CD36 in vivo andin vitro. J. Immunol. 148, 78–83 (1992).
37. Wada, T. & Penninger, J.M. Mitogen-activated protein kinases in apoptosis regulation.Oncogene 23, 2838–2849 (2004).
38. Ishikawa, Y. & Kitamura, M. Dual potential of extracellular signal-regulated kinase forthe control of cell survival. Biochem. Biophys. Res. Commun. 264, 696–701 (1999).
39. Gauld, S.B., Blair, D., Moss, C.A., Reid, S.D. & Harnett, M.M. Differential roles forextracellularly regulated kinase-mitogen-activated protein kinase in B cell antigenreceptor-induced apoptosis and CD40-mediated rescue of WEHI-231 immature Bcells. J. Immunol. 168, 3855–3864 (2002).
40. Zghibeh, C.M., Raj Gopal, V., Poff, C.D., Falck, J.R. & Balazy, M. Determination oftrans-arachidonic acid isomers in human blood plasma. Anal. Biochem. 332, 137–144(2004).
41. Llorens, S. & Nava, E. Cardiovascular diseases and the nitric oxide pathway. Curr. Vasc.Pharmacol. 1, 335–346 (2003).
42. Liu, L. et al. Essential roles of S-nitrosothiols in vascular homeostasis and endotoxicshock. Cell 116, 617–628 (2004).
43. Radi, R., Rodriguez, M., Castro, L. & Telleri, R. Inhibition of mitochondrial electrontransport by peroxynitrite. Arch. Biochem. Biophys. 308, 89–95 (1994).
44. Tabuchi, A, Oh, E, Taoka, A, Sakurai, H, Tsuchiya, T & Tsuda. Rapid attenuation ofAP-1 transcriptional factors associated with nitric oxide (NO)-mediated neuronalcell death. J Biol. Chem. 271, 31061–7 (1996).
45. Dameron, K.M., Volpert, O.V., Tainsky, M.A. & Bouck, N. Control of angiogenesis infibroblasts by p53 regulation of thrombospondin-1. Science 265, 1582–1584 (1994).
46. Okuno, M., Arimoto, E., Nishizuka, M., Nishihara, T. & Imagawa, M. Isolation of up- ordown-regulated genes in PPARgamma-expressing NIH-3T3 cells during differentiationinto adipocytes. FEBS Lett. 519, 108–112 (2002).
47. Nielsen, J.C., Naash, M.I. & Anderson, R.E. The regional distribution of vitamins E andC in mature and premature human retinas. Invest. Ophthalmol. Vis. Sci. 29, 22–26(1988).
48. Flynn, J.T. et al. A cohort study of transcutaneous oxygen tension and the incidenceand severity of retinopathy of prematurity. N. Engl. J. Med. 326, 1050–1054 (1992).
49. Mann, R.M., Riva, C.E., Stone, R.A., Barnes, G.E. & Cranstoun, S.D. Nitric oxide andchoroidal blood flow regulation. Invest. Ophthalmol. Vis. Sci. 36, 925–930 (1995).
50. Lahaie, I. et al. A novel mechanism for vasoconstrictor action of 8-isoprostaglandin F2alpha on retinal vessels. Am. J. Physiol. 274, R1406–R1416 (1998).
ART ICL ES
NATURE MEDICINE VOLUME 11 [ NUMBER 12 [ DECEMBER 2005 1345
©20
05 N
atur
e P
ublis
hing
Gro
up
http
://w
ww
.nat
ure.
com
/nat
urem
edic
ine
