Monocyte chemoattractant protein 1 mediates retinaldetachment-induced photoreceptor apoptosisToru Nakazawa*†, Toshio Hisatomi*, Chifuyu Nakazawa*, Kosuke Noda*, Kazuichi Maruyama*, Haicheng She*,Akihisa Matsubara*, Shinsuke Miyahara*, Shintaro Nakao*, Yuqin Yin‡, Larry Benowitz‡, Ali Hafezi-Moghadam*,and Joan W. Miller*§

*Angiogenesis Laboratory, Massachusetts Eye and Ear Infirmary, Department of Ophthalmology, and ‡Department of Neurosurgery and NeurobiologyProgram, Children’s Hospital, Harvard Medical School, Boston, MA 02114; and †Department of Ophthalmology, Tohoku University School of Medicine,Sendai, Miyagi 980-8574, Japan

Edited by Jeremy Nathans, Johns Hopkins University School of Medicine, Baltimore, MD, and approved December 11, 2006 (received for reviewSeptember 15, 2006)

Photoreceptor apoptosis is a major cause of visual loss in retinaldetachment (RD) and several other visual disorders, but the under-lying mechanisms remain elusive. Recently, increased expression ofmonocyte chemoattractant protein 1 (MCP-1) was reported in vitre-ous humor samples of patients with RD and diabetic retinopathy aswell as in the brain tissues of patients with neurodegenerativediseases, including Alzheimer’s disease and multiple sclerosis. Herewe report that MCP-1 plays a critical role in mediating photoreceptorapoptosis in an experimental model of RD. RD led to increased MCP-1expression in the Muller glia and increased CD11b� macrophage/microglia in the detached retina. An MCP-1 blocking antibody greatlyreduced macrophage/microglia infiltration and RD-induced photore-ceptor apoptosis. Confirming these results, MCP-1 gene-deficientmice showed significantly reduced macrophage/microglia infiltrationafter RD and very little photoreceptor apoptosis. In primary retinalmixed cultures, MCP-1 was cytotoxic for recoverin� photoreceptors,and this toxicity was eliminated through immunodepleting macro-phage/microglia from the culture. In vivo, deletion of the geneencoding CD11b/CD18 nearly eliminated macrophage/microglia infil-tration to the retina after RD and the loss of photoreceptors. Thus,MCP-1 expression and subsequent macrophage/microglia infiltrationand activation are critical for RD-induced photoreceptor apoptosis.This pathway may be an important therapeutic target for preventingphotoreceptor apoptosis in RD and other CNS diseases that share acommon etiology.

macrophage recruitment � neuroprotection

Photoreceptor apoptosis is the basis for permanent visual lossin a number of retinal disorders, including macular degen-

eration (1), retinal detachment (RD) (2, 3), diabetic retinopathy(4), retinopathy of prematurity (5), and retinitis pigmentosa (6).Physical separation of the photoreceptors from the underlyingretinal pigment epithelium occurs in rhegmatogenous and in trac-tional and exudative RD, as well as in neovascular macular degen-eration and central serous chorioretinopathy. In these conditionsphotoreceptors are highly vulnerable and undergo apoptosis (2, 3).In a rodent model of RD we have shown that electroretinogramssensitively reflect RD-induced functional changes and that theseelectroretinogram changes correlate highly with the amount ofphotoreceptor apoptosis (7). Although surgery is carried out forrhegmatogenous RD, visual acuity is not always restored because ofphotoreceptor apoptosis (2, 3). In the other conditions mentioned,serous RD may persist despite treatment, and vision loss progressesbecause of photoreceptor apoptosis. Therefore, new insights aboutthe mechanisms that underlie photoreceptor apoptosis in RDwould be of clinical interest and could lead to new treatments.Previously we demonstrated that caspase activation is associatedwith RD-induced photoreceptor apoptosis (8). However, suppres-sion of caspases alone is not sufficient to prevent photoreceptorapoptosis (7), because caspase-independent pathways also appear

to be involved, although the detailed mechanisms are currentlyunclear (9).

Monocyte chemoattractant protein 1 (MCP-1, CCL-2) contrib-utes to the recruitment of leukocytes to sites of injury (10) in thepathogenesis of atherosclerosis (11), lung infection (12), angiogen-esis (13, 14), and various CNS diseases (15–17). Vitreous samplesfrom patients with RD contain significantly higher levels of MCP-1than samples from patients with other retinal conditions, such asmacular hole or idiopathic premacular fibrosis (18, 19). The MCP-1receptor CCR2 is expressed on leukocytes, endothelium (20), glialcells, and neurons in the brain (21). Correspondingly, newly dis-covered functions of MCP-1 include neurodegeneration (22), neu-roprotection (23), and increased vessel permeability (20). However,the mechanisms of MCP-1’s effect on neurons, and the contributionof direct vs. indirect effects, for instance via macrophage/microgliarecruitment, remain to be investigated.

Previously, we and others demonstrated the presence of activatedbone marrow-derived monocytes/macrophages and differentiatedtissue macrophages, namely microglia, in the detached retina afterRD (24, 25). Bone marrow-derived macrophages accumulatedsubretinally after RD and removed the debris of apoptotic photo-receptors (24). However, the chemoattractant that causes theaccumulation of macrophage/microglia in the detached retina andits role in the neurodegeneration have remained unclear. Using anexperimental model of RD, the current study demonstrates thatMCP-1 plays a critical role in photoreceptor apoptosis by causingmacrophage/microglia accumulation and generation of oxidativestress in the injured retina.

ResultsMCP-1 Expression After RD. To investigate the possible role ofMCP-1 in RD we examined the expression of MCP-1 mRNA andprotein in the retina 72 h after RD using quantitative PCR(TaqMan probe) and ELISAs, respectively. This time point waschosen because the number of TUNEL� photoreceptors peaks by72 h after RD (7, 26). Quantitative PCR revealed that MCP-1

Author contributions: T.N., L.B., A.H.-M., and J.W.M. designed research; T.N., T.H., C.N.,K.N., H.S., A.M., S.M., and S.N. performed research; K.M. and Y.Y. contributed newreagents/analytic tools; T.N., L.B., A.H.-M., and J.W.M. analyzed data; and T.N., L.B.,A.H.-M., and J.W.M. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS direct submission.

Abbreviations: RD, retinal detachment; MCP-1, monocyte chemoattractant protein 1; OPL,outer plexiform layer; IHC, immunohistochemistry; ONL, outer nuclear layer; TEM,transmission EM.

See Commentary on page 2033.

§To whom correspondence should be addressed at: Angiogenesis Laboratory, Massachu-setts Eye and Ear Infirmary, Department of Ophthalmology, Harvard Medical School, 243Charles Street, Boston, MA 02114. E-mail: joan�miller@meei.harvard.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0608167104/DC1.

© 2007 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0608167104 PNAS � February 13, 2007 � vol. 104 � no. 7 � 2425–2430

NEU

ROSC

IEN

CESE

ECO

MM

ENTA

RY

Dow

nloa

ded

by g

uest

on

Apr

il 22

, 202

0

mRNA levels increased 84-fold (Fig. 1A), whereas protein levelsincreased 10-fold over baseline (Fig. 1B). Whereas MCP-1 immu-noreactivity was very weak in the normal controls (Fig. 1C),spindle-shaped cells, which morphologically resembled Muller glia,were strongly MCP-1-immunoreactive in the outer plexiform layer(OPL) after RD (Fig. 1 D and F). To verify the identity of theMCP-1� cells, double immunostaining was performed by usingantibodies against MCP-1 and glutamine synthetase, a Muller cellmarker. The two signals colocalized in Muller cells in the OPL (Fig.1H, arrowheads). These data demonstrate a dramatic increase ofMCP-1 mRNA and protein in the mouse retina after RD, specif-ically in the Muller glial cells.

Acute Blockade or Genetic Deletion of MCP-1 Prevents RD-InducedPhotoreceptor Loss. To investigate whether MCP-1 is involved inRD-induced photoreceptor apoptosis, we injected a functionallyblocking anti-MCP-1 F(ab�) fragment (0.1 �g/�l) subretinally at thetime of RD induction. Photoreceptor apoptosis was quantified byTUNEL at 72 h (7, 26). MCP-1 blockade almost completelysuppressed the appearance of TUNEL� cells in the outer nuclearlayer (ONL), whereas a control IgG1 F(ab�) fragment had no effect(Fig. 2). To understand the role of MCP-1 in this process better, weexamined the consequences of RD in MCP-1-deficient (MCP-1�/�)mice. In the absence of RD, the general appearance of the retinaand the thickness of the ONL were similar in MCP-1�/� and WTmice [supporting information (SI) Fig. 7F]. Seven days after induc-ing RD, the thickness of the ONL decreased significantly in WTanimals (SI Fig. 7 B and F). In contrast, in MCP-1�/� mice, thethickness of the ONL remained unchanged from baseline after RD(SI Fig. 7 D and F). As an additional parameter for quantifying therole of MCP-1 in RD-induced photoreceptor apoptosis, we per-formed TUNEL staining 72 h after inducing RD in WT andMCP-1�/� mice. Under normal conditions, TUNEL� cells are notdetected in the ONL of WT or MCP-1�/� mice (data not shown).After RD, TUNEL� photoreceptors were detected in both groups,

although the amount of cell death in MCP-1�/� mice was �80% lessthan in WT mice (Fig. 3). Transmission EM (TEM) demonstratedcellular shrinkage, chromatin condensation, and apoptotic bodyformation, signs that were far more prevalent in the ONL of WT

Fig. 1. MCP-1 expression is up-regulated after retinal detachment (RD) inmice. (A–D) Up-regulation of MCP-1 after RD. (A) Quantitative real-time PCRdata for MCP-1 mRNA 72 h after RD (n � 6). (B) ELISA to detect MCP-1 protein72 h after RD (n � 6). **, P � 0.01. (C–E) Immunoreactivity of MCP-1 in controlretina (C) or after RD in WT mice (D) or in MCP-1�/� mice (E). (F–H) MCP-1localization in Muller glia. Double IHC was carried out with antibodies againstMCP-1 (F) and glutamine synthetase, a marker for Muller cells (G). Arrowsindicate colocalization. (H) Overlay. (Scale bar: 50 �m.)

Fig. 2. An MCP-1 blocking antibody prevents RD-induced photoreceptorloss. (A and B) TUNEL in retinal sections with subretinal injection of controlantibody (A) or MCP-1 blocking antibody (B). (Scale bar: 100 �m.) (C) Quan-tification of TUNEL� photoreceptors 72 h after RD (n � 8 each). **, P � 0.01.

Fig. 3. Cytotoxic effect of MCP-1 on RD-induced photoreceptor apoptosis. (Aand B) TUNEL 72 h after RD in WT mice (A) and MCP-1�/� mice (B). (Scale bar:50 �m.) (C and D) TEM photomicrographs through the ONL 72 h after RD in WTmice (C) and MCP-1�/� mice (D). Note the increased presence of apoptoticphotoreceptors in WT mice compared with MCP-1�/� mice (arrows). (Scale bar:10 �m.) (E) Quantification of TUNEL� cells (n � 8 each). *, P � 0.05.

2426 � www.pnas.org�cgi�doi�10.1073�pnas.0608167104 Nakazawa et al.

Dow

nloa

ded

by g

uest

on

Apr

il 22

, 202

0

mice than in MCP-1�/� mice (Fig. 3 C, arrows, and D). These datashow that MCP-1 plays a critical role in RD-induced photoreceptorapoptosis.

Role of CD11b� Macrophage/Microglia in the Detached Retina. Usingan anti-CD11b antibody, immunohistochemistry (IHC) confirmedthe previously reported accumulation of CD11b� macrophage/microglia in the subretinal space and in the OPL 72 h after RDinduction in WT (SI Fig. 8 C and F) (24, 25). Some CD11b�

macrophage/microglia were detected in the inner nuclear layer ofnormal mice even without RD, and the number of these cells didnot changed after RD (data not shown). Double IHC revealed thatMCP-1� and most CD11b� cells were in close apposition in theOPL (SI Fig. 8F). Confocal microscopy confirmed this colocaliza-tion and further revealed that macrophage/microglia processeswere entwined around MCP-1� Muller cells (SI Fig. 8G). Tovisualize the relationship between macrophage/microglia infiltra-tion in the OPL and TUNEL� photoreceptors in the ONL, wecompared the number of CD11b� macrophage/microglia andTUNEL� cells at various time points after RD (SI Fig. 8I). Thesestudies showed that the peak of macrophage/microglia infiltrationcoincided with the peak of TUNEL� cells at 72 h. CD11b�

macrophage/microglia were seen to extend their processes into theONL and to engulf TUNEL� photoreceptors (SI Fig. 8H). Thesedata suggest that the pathogenesis of RD involves MCP-1 expres-sion, infiltration of CD11b� macrophage/microglia, and RD-induced photoreceptor apoptosis.

MCP-1 Contributes to Macrophage/Microglia Infiltration and Disrup-tion of the OPL in the Detached Retina. To examine the role ofMCP-1 in macrophage/microglia infiltration in more detail, weinvestigated whether deletion of the MCP-1 gene affects macro-phage/microglia infiltration after RD. RD-induced infiltration ofmacrophage/microglia was strongly suppressed in MCP-1�/� micecompared with WT mice (Fig. 4). Ultrastructural studies by TEMrevealed that invading macrophage/microglia, which were muchmore prevalent in WT than in MCP-1�/� mice, could be distin-

guished from photoreceptors by virtue of their larger somata andless electron-dense nuclei (Fig. 4E, white arrows). Synaptic struc-tures, including cone pedicles and rod spherules between photo-receptors and bipolar cells, were more severely disrupted in thedetached retinas of WT than MCP-1�/� mice (Fig. 4 E and F, blackarrows). These data suggest that MCP-1 is critical for the infiltrationof macrophage/microglia to the subretinal space and the OPL afterRD, and for the subsequent structural and functional disruption ofthese retinal layers.

Cytotoxic Effect of MCP-1 on Cultured Photoreceptors. To investigatewhether the increase in MCP-1 after RD contributes to photore-ceptor apoptosis directly, we performed experiments using primaryadult retinal cultures. Because these cultures contained a variety ofretinal cells (60% photoreceptor), photoreceptors were identifiedby immunocytochemistry with an antibody against recoverin, acommonly used marker for photoreceptors in vitro (27). Primarycultures also contained CD11b� macrophage/microglia (1%) (Fig.5C, arrows). To remove the activated macrophage/microglia fromthese cultures, we carried out immunopanning in dishes precoatedwith rat anti-mouse CD11b antibody (Fig. 5B). After macrophage/

Fig. 4. Markedly reduced number of CD11b� cells in MCP-1�/� mice. (A andB) IHC with an antibody against CD11b 72 h after RD in WT mice (A) orMCP-1�/� mice (B). Macrophage/microglia was recruited in the IPL (shortarrows) and subretinal space (arrows). (Scale bar: 50 �m.) (C and D) Quanti-fication of CD11b� cells in the OPL (C) and in the subretinal space (D) (n � 8each). *, P � 0.05. (E and F) TEM photomicrographs in the ONL 72 h after RDin WT mice (E) and MCP-1�/� mice (F). Apoptotic photoreceptors (short whitearrow) and invading cells (long white arrows) are more prevalent in WT micethan in MCP-1�/� mice. The disruption of synaptic structures such as conepedicles and rod spherules was more severe in WT mice than in MCP-1�/� mice(black arrows). (Scale bar: 10 �m.)

Fig. 5. CD11b� cells mediate the cytotoxic effect of MCP-1 on culturedphotoreceptors. (A and B) Recoverin� photoreceptors in retinal primary cul-ture with (B) or without (A) MCP-1. (C and D) CD11b� cells before (C) or after(D) depletion by immunopanning. Arrows indicate CD11b� cells. (E and F)Recoverin� photoreceptors with (F) or without (E) MCP-1 (1 ng/ml) afterdepletion of CD11b� cells. (G and H) Recoverin� photoreceptors (red) andperipheral macrophage (green) in culture in the presence (H) or absence (G) ofMCP-1 (1 ng/ml) after depletion of CD11b� cells. (Scale bar: 100 �m.) (I)Dose–response curve of MCP-1 cytotoxicity on the cultured photoreceptors inthe presence or absence of resident CD11b� cells. * (P � 0.05) and ** (P � 0.01)represent the significance when compared with controls without MCP-1.Catalase (2 �g/ml) suppresses MCP-1-induced photoreceptor loss. (J) Dose–response curve of MCP-1 cytotoxicity on cultured photoreceptors after addedperipheral macrophage (PM).

Nakazawa et al. PNAS � February 13, 2007 � vol. 104 � no. 7 � 2427

NEU

ROSC

IEN

CESE

ECO

MM

ENTA

RY

Dow

nloa

ded

by g

uest

on

Apr

il 22

, 202

0

microglia depletion from the cultures, MCP-1 had no effect onphotoreceptor survival (Fig. 5 E, F, and I). In contrast, withoutdepletion of CD11b� cells form retinal cultures, the number ofrecoverin� photoreceptors declined progressively with increasingMCP-1 concentration (Fig. 5 A, B, and I) and MCP-1 concentra-tions as low as 0.1 ng/ml had significant cytotoxic effects (Fig. 5I).These data suggest that MCP-1’s cytotoxicity is mediated throughresident macrophage/microglia but not through direct interactionwith the cultured photoreceptors. Next, to investigate whetherMCP-1’s cytotoxicity for cultured photoreceptors was related tooxidative stress, a known cause of photoreceptor damage (28, 29),catalase was added to the culture medium. Catalase reducesoxidative stress through decomposition of hydrogen peroxide, areactive oxygen species, into water and oxygen (28, 29). Thecytotoxic effect of MCP-1 was significantly suppressed with theaddition of catalase (Fig. 5I). To examine whether peripheralmacrophages also have a cytotoxic effect for cultured photorecep-tors, we added peripheral macrophages into retinal cultures afterdepletion of the resident macrophage/microglia with or withoutMCP-1. Addition of peripheral macrophages at a ratio of 1% ofcultured cells restored MCP-1’s cytotoxicity through oxidativestress (Fig. 5 F and G). Interestingly, peripheral macrophagesderived from Mac-1 (CD11b/CD18) gene-deficient (Mac-1�/�)mice did not have cytotoxic effects even after MCP-1 stimulation(Fig. 5J). These data suggest that the cytotoxic effect of MCP-1 oncultured photoreceptors is dose-dependent, is mediated throughactivated macrophage/microglia, and is likely due to oxidativestress.

Mac-1�/� Mice Are Protected Against RD-Induced PhotoreceptorApoptosis. To examine whether the cytotoxic effects of MCP-1 aremediated by macrophage/microglia in vivo, we induced RD in micedeficient for the Mac-1 integrin, a critical receptor for leukocyterecruitment and activation (30, 31). Without RD, the number andmorphology of CD45� cells in the OPL was similar in Mac-1�/� andWT mice (SI Fig. 9 A, B, and E). However, 72 h after RD,significantly fewer CD45� cells were found in Mac-1�/� comparedwith WT mice (SI Fig. 9). Interestingly, the morphology of thesemacrophage/microglia after RD resembles that of resting cells. Incontrast to WT mice, Mac-1�/� mice did not show a decrease inONL thickness (SI Fig. 10) or an increase in the number ofTUNEL� photoreceptors (Fig. 6), suggesting an important role forMac-1-mediated infiltration and activation of macrophages/microglia during RD-induced injury.

Role of Oxidative Stress in RD-Induced Injury in Vivo. To examine therole of oxidative stress on RD-induced photoreceptor apoptosis,the antioxidant PBN (100 mg/kg per day) was administered for 3days after RD. RD-induced MCP-1 expression was not changed byPBN treatment, nor was the number of recruited macrophage/microglia in the OPL (data not shown). i.p. administration of PBNsignificantly suppressed RD-induced photoreceptor degeneration(P � 0.007) (Fig. 6F), suggesting that the RD-induced photore-ceptor apoptosis in vivo is likely due to oxidative stress.

DiscussionUsing an experimentally induced model of RD in mice, we showthat MCP-1 is a critical mediator of photoreceptor apoptosis, amajor cause of visual loss in several retinal disorders. MCP-1 levelsrapidly rise in Muller glial cells after RD and lead to an increasednumber of macrophage/microglia into the site of the injury. Acuteblockade of MCP-1 with a functionally blocking antibody ordeletion of its gene in mice almost completely eliminates RD-induced photoreceptor apoptosis. We further show that the cyto-toxic effect of MCP-1 on cultured photoreceptors is not direct, buta consequence of oxidative stress produced by activated macro-phage/microglia. Deletion of the gene for either MCP-1 or Mac-1(CD11b/CD18) in mice almost completely eliminated infiltration of

macrophage/microglia after RD and protected photoreceptorsfrom RD-induced apoptosis.

Increased MCP-1 expression has been reported in several otherretinal disease models, including light damage (32), uveitis (33),diabetic retinopathy (34), retinitis pigmentosa (35), and ischemia–reperfusion (36). Thus, MCP-1 may be an important factor formacrophage/microglial responses during various acute and chronicretinal disorders. In the retinal ischemia–reperfusion model,MCP-1 up-regulation is detected only in the inner retina (36),corresponding to the area of retinal injury (37). In the current studyMCP-1 protein expression was detected in Muller cells, especiallyin the OPL. The OPL has a rich vascular capillary bed, which wouldfacilitate extravasation of leukocytes (38). The colocalization ofMCP-1� Muller cells and activated macrophage/microglia in theOPL, along with the decreased number of infiltrated macrophage/microglia in MCP-1�/� mice after RD, indicates that increasedMCP-1 in Muller cells attracts macrophage/microglia toward theouter retina. This finding is consistent with the fact that the outerretina is the main site of injury after RD. Shen et al. (39) have shownthat Matrigel-induced angiogenesis and associated photoreceptordegeneration can be severe even in the absence of MCP-1, pre-sumably because of the harmful effect of pathologic angiogenesis onphotoreceptor viability. That study indicates an absence of macro-phage/microglia recruitment in the absence of MCP-1, consistentwith our findings.

Endothelin2 has previously been shown to play an important rolein Muller cell activation in various types of retinal injury (40). Wehave confirmed that the expression of endothelin2 mRNA in-creased 4.5-fold 6 h after RD, although not at 3 h. Our recent datashow that expression of MCP-1 mRNA was already elevated 1 h

Fig. 6. Deletion of the Mac-1 gene prevents RD-induced photoreceptor apo-ptosis. (A and B) TUNEL in retinal sections of WT mice (A) or Mac-1�/� mice (B).(Scale bar: 100 �m.) (C and D) TEM photomicrographs in the ONL 72 h after RD inWT mice (C) and Mac-1�/� mice (D). Apoptotic photoreceptors (arrow) are moreprevalent in WT mice than in Mac-1�/� mice. (Scale bar: 10 �m.) (E) Quantificationof TUNEL� photoreceptors 72 h after RD (n � 8 each). (F) Quantification ofTUNEL� photoreceptors with or without PBN treatment. **, P � 0.01.

2428 � www.pnas.org�cgi�doi�10.1073�pnas.0608167104 Nakazawa et al.

Dow

nloa

ded

by g

uest

on

Apr

il 22

, 202

0

after RD and continued to be elevated over 3 days (41). These datasuggest that the increased expression of MCP-1 occurs earlier thanthat of endothelin2, although it is possible that these two moleculesact together to alter retinal responses after RD.

Photoreceptor apoptosis and degeneration of the ONL and OPLwere nearly eliminated after RD when MCP-1 was blocked or itsgene was deleted. These data indicate that suppression of MCP-1has beneficial effects on RD-induced photoreceptor death andretinal degeneration in vivo. In line with our in vivo results, inprimary retinal culture experiments we showed that MCP-1’scytotoxicity is mediated through macrophage/microglia activationand not through a direct effect on photoreceptors. MCP-1’s cyto-toxicity in this system was seen at a concentration as low as 0.1 ng/mland peaked at 1 ng/ml (Fig. 5). A low concentration of MCP-1similar to that which we find cytotoxic (i.e., 1 ng/ml) was also shownto be chemotactic for macrophage/microglia (42). In contrast,markedly higher concentrations of MCP-1 are required for othereffects reported in vitro, including the modulation of Ca2� dynamicsin neurons (200 ng/ml) (43), degeneration or protection of neurons(10–100 ng/ml) (22, 23), endothelial cell migration (10–100 ng/ml),or a decrease in transendothelial cell electrical resistance (1 �g/ml)(20). These data suggest that macrophage/microglia are moresensitive to MCP-1 than other cells and that the cytotoxic effect ofMCP-1 occurs primarily through activated macrophage/microgliaboth in vivo and in vitro.

The origin of the macrophage/microglia seen in the OPL afterRD remains largely unknown. It is possible that resident macro-phage/microglia translocate and proliferate in the OPL, based onprevious reports showing the importance of resident monocytes inhuman, cat, rabbit, and ground squirrel RD (25). These cells mayalso represent newly recruited peripheral blood monocytes, con-sistent with our previous finding that bone marrow-derived mac-rophages are recruited to the subretinal space after RD (24). Innormal mice we found that resident macrophage/microglia weredetected mainly in the inner retina (GCL and IPL) but not in theOPL. Even after RD, the number of resident macrophage/microgliaremained unchanged in the inner retina and very few macrophage/microglia, which had just migrated from the IPL to OPL throughthe inner nuclear layer, were detected in the inner nuclear layer(data not shown). Furthermore, after RD, some of the CD11b�

cells in the OPL showed immunoreactivity for CD68 and F4/80,markers for peripheral macrophages (data not shown). In vitro, bothresident macrophage/microglia and isolated peripheral bloodmonocytes were cytotoxic to cultured photoreceptors after MCP-1stimulation. Thus, although the exact origin of the CD11b� mac-rophage/microglia in the OPL cannot be pinpointed, both residentand recruited macrophage/microglia may mediate MCP-1’s cyto-toxicity. However, our experiments show that CD11b� macro-phage/microglia are necessary for MCP-1’s cytotoxic effect oncultured photoreceptors.

To further confirm the role of macrophage/microglia in RD-induced photoreceptor apoptosis, we studied mice with a deficit ofmacrophage/microglia recruitment. Mac-1 (CD11b/CD18) is a �2integrin with an established role in monocyte recruitment toperipheral nonocular tissues (44). Mac-1�/� mice show a decreasedsusceptibility to brain ischemia–reperfusion injury relative to WTmice (45). Activated macrophage/microglia express high amountsof Mac-1 in the retina (46). Here we show that Mac-1�/� mice havesignificantly fewer macrophage/microglia and fewer signs of mor-phological activation both in the OPL and in the subretinal spacethan WT mice (SI Fig. 9). These data suggest that Mac-1 is animportant adhesion molecule for infiltration and activation ofmacrophage/microglia after RD. Macrophage/microglia have re-cently been reported to promote apoptosis of developing Purkinjecells by engulfing and terminating these apoptotic cells producingsuperoxide ions (47, 48). Consistent with these findings, in thecurrent study oxidative stress was found to mediate the cytotoxiceffect of MCP-1 both in vitro (Fig. 5) and in vivo (Fig. 6F).

Generally, oxidative stress is one of the major cytotoxic factors forphotoreceptor death in various pathological conditions (28, 29). Incontrast to the neurotoxic effects of oxidative free radicals, molec-ular oxygen (O2) was found to be neuroprotective when given even1 day after inducing RD in cats (49). Although the mechanismunderlying the protective effect of molecular oxygen was notelucidated, it is possible that it prevented hypoxia-induced MCP-1expression in retinal glia (50); this in turn would prevent theactivation of macrophage/microglia and generation of oxidativefree radicals, the final effectors of photoreceptor cell death. Thus,suppression of macrophage/microglia or antioxidant treatment maythus represent alternative strategies for neuroprotection againstphotoreceptor degeneration.

In conclusion, we demonstrate that MCP-1 up-regulation plays acritical role in inducing photoreceptor apoptosis after RD. Thecytotoxic effect of MCP-1 on photoreceptors is mediated throughits chemotactic properties and possibly macrophage/microglia-generated oxidative stress. Blockade of MCP-1 may open newtherapeutic avenues to treat photoreceptor death in the setting ofvarious retinal disorders as well as other CNS disorders that sharea common etiology.

Materials and MethodsAnimals. All animal procedures were performed in accordance withthe statement of the Association for Research in Vision andOphthalmology and the protocol approved by the Animal CareCommittee of the Massachusetts Eye and Ear Infirmary. Adultmale MCP-1�/� mice, Mac-1�/� mice (C57BL/6 background,20–25 g; The Jackson Laboratory, Bar Harbor, ME), and age- andsex-matched C57BL/6 mice were housed in covered cages. A totalof 28 MCP-1�/� mice, 28 Mac-1�/� mice, and 128 WT mice wereused for this study.

Surgical Induction of RD. Induction of RD and subretinal injectionof MCP-1 blocking antibody [11K2, mouse F(ab�) IgG1, 0.1 �g/�l;Biogen, Cambridge, MA] or isotype control (Southern Biotechnol-ogy Associates, Birmingham, AL) were performed as previouslydescribed (Fig. 4E) (7–9, 24, 41). The blocking effect for anti-MCP-1 F(ab�) antibody has been established in a mouse arterio-sclerosis model (51). RD was created only in the right eye of eachanimal, with the left eye serving as a control.

RNA Extraction, RT-PCR, and Quantitative Real-Time PCR. Total RNAextraction and reverse transcription were performed as previouslyreported (41, 52). PCR primers for MCP-1 used in this study are asfollows: mMCP1 forward, 5�-ACTCACCTGCTGCTACTCAT-TCACC-3�; mMCP1 reverse, 5�-CTACAGCTTCTTTGGGA-CACCTGCT-3�; and mMCP1, VIC-ATC CCA ATG AGT AGGCTG GAG AGC TAC AAG AGG ATC-TAMRA. For relativecomparison of each gene, we analyzed the Ct value of real-timePCR data with the ��Ct method normalizing by an endogenouscontrol (18S ribosomal RNA) (41, 52, 53).

ELISA. The posterior lens capsule, vitreous and neural retina com-bined, was collected 72 h after RD. Protein extraction and ELISAwere performed as previously described (41). One hundred micro-grams of total protein was used for ELISA of MCP-1 (BioSource,Camarillo, CA).

IHC. IHC was performed as previously reported (41, 54). Rabbitanti-MCP-1 (1:100; PeproTech, Rocky Hill, NJ), rat anti-mouseCD11b (1:50; Serotec), rat anti-mouse CD45 (1:50; Pharmingen),or mouse anti-glutamine synthetase (1:100; Chemicon, San Diego,CA) were used as primary antibodies.

TUNEL. TUNEL and quantification of TUNEL� cells were per-formed as previously described (41) by using the ApopTag Fluo-rescein In Situ Apoptosis Detection Kit (S7110; Chemicon Inter-

Nakazawa et al. PNAS � February 13, 2007 � vol. 104 � no. 7 � 2429

NEU

ROSC

IEN

CESE

ECO

MM

ENTA

RY

Dow

nloa

ded

by g

uest

on

Apr

il 22

, 202

0

national, Temecula, CA). The center of the detached retina wasphotographed, and the number of TUNEL� cells in the ONL wascounted in a masked fashion. The area of the ONL was measuredwith OpenLab software.

TEM. TEM was performed as previously described (7, 24). Eyes werefixed in 1% glutaraldehyde/1% paraformaldehyde in PBS andpostfixed in veronal acetate buffered osmium tetroxide (2%),dehydrated in ethanol and water, and embedded in Epon. Ultrathinsections were cut from blocks and mounted on copper grids. Thespecimens were examined by using a Philips CM10 electronmicroscope.

Adult Mouse Retinal Primary Cultures. Adult primary retinal cultureswere prepared as previously described (54) with minor modifica-tions. Neural retinas were incubated at 37°C for 20 min in a CO2incubator in digestion solution containing papain (10 units/ml;Worthington) and L-cysteine (0.3 mg/ml; Sigma, St. Louis, MO) inHanks’ buffered saline solution (HBSS). Cell density was adjustedto 3.5 � 104 cells per well of an eight-well chamber (Nunc) withNeurobasal A medium (Invitrogen) containing B27 supplementwithout antioxidants (NBA/B27AO�; Invitrogen) and 1 �g/mlinsulin, 2 mM L-glutamate, and 12 �g/ml gentamicin. One hourlater, MCP-1 at specified concentrations was added to culturemedium, and incubation was continued for 24 h. To assess theviability of photoreceptors, we performed immunocytochemistrywith rabbit anti-recoverin antibody (1:500 dilution, AB5585;Chemicon). The number of recoverin� photoreceptors wascounted at 10 random fields per well in a blind fashion by using

ImageJ software. Values are given as the mean SEM of fourreplicate wells. An immunopanning dish was prepared by incubat-ing 10-cm culture dishes (Falcon) with 50 �g/ml rat anti-CD11bantibody (Serotec) in 4 ml of HBSS overnight. The panning dish wasblocked with 4 ml of HBSS/0.1% BSA for 1 h, and dissociated cellswere incubated in the dish for 30 min, with slow agitation of the dishevery 10 min. Peripheral CD11b� macrophage was collected aspreviously described (55) and added to retinal primary culture afterdepletion of resident macrophage/microglia.

Statistical Analysis. The statistical significance of RT-PCR andELISA results was determined by using unpaired t tests. The datafrom the TUNEL and in vitro survival assays were analyzed with theScheffe post hoc test by using StatView 4.11J software for Macin-tosh (Abacus Concepts, Berkeley, CA). The significance level wasset at P � 0.05 (* in figures) and P � 0.01 (** in figures). The datarepresent mean SD except for primary culture results.

We thank Thaddeus Dryja for thoughtful comments on the manuscript. Wealso thank Norman Michaud and Sreedevi Mallemadugula (MassachusettsEye and Ear Infirmary) for technical assistance and Biogen-Idec for thegenerous gift of the MCP-1 antibody (11K2). This work was supported byan Alcon Research Award (to J.W.M.), a Bausch & Lomb VitreoretinalFellowship (to T.N.), National Institutes of Health Grant AI50775 (toA.H.-M.), and National Eye Institute Grants EY014104 (Massachusetts Eyeand Ear Infirmary Core Grant) and EY05690 (to L.B.). We thank theMassachusetts Lions Foundation for generous funds provided for labora-tory equipment used in this project and Research to Prevent Blindness forunrestricted funds awarded to the Department of Ophthalmology at Har-vard Medical School.

1. Dunaief JL, Dentchev T, Ying GS, Milam AH (2002) Arch Ophthalmol 120:1435–1442.

2. Cook B, Lewis GP, Fisher SK, Adler R (1995) Invest Ophthalmol Vis Sci36:990–996.

3. Arroyo JG, Yang L, Bula D, Chen DF (2005) Am J Ophthalmol 139:605–610.4. Barber AJ, Lieth E, Khin SA, Antonetti DA, Buchanan AG, Gardner TW (1998)

J Clin Invest 102:783–791.5. Fulton AB, Hansen RM, Petersen RA, Vanderveen DK (2001) Arch Ophthalmol

119:499–505.6. Doonan F, Cotter TG (2004) Curr Neurovasc Res 1:41–53.7. Hisatomi T, Sakamoto T, Goto Y, Yamanaka I, Oshima Y, Hata Y, Ishibashi T,

Inomata H, Susin SA, Kroemer G (2002) Curr Eye Res 24:161–172.8. Zacks DN, Hanninen V, Pantcheva M, Ezra E, Grosskreutz C, Miller JW (2003)

Invest Ophthalmol Vis Sci 44:1262–1267.9. Hisatomi T, Sakamoto T, Murata T, Yamanaka I, Oshima Y, Hata Y, Ishibashi

T, Inomata H, Susin SA, Kroemer G (2001) Am J Pathol 158:1271–1278.10. Matsushima K, Larsen CG, DuBois GC, Oppenheim JJ (1989) J Exp Med

169:1485–1490.11. Reape TJ, Groot PH (1999) Atherosclerosis 147:213–225.12. Huffnagle GB, Traynor TR, McDonald RA, Olszewski MA, Lindell DM, Herring

AC, Toews GB (2000) Immunopharmacology 48:231–236.13. Ishida S, Yamashiro K, Usui T, Kaji Y, Ogura Y, Hida T, Honda Y, Oguchi Y,

Adamis AP (2003) Nat Med 9:781–788.14. Nakao S, Kuwano T, Tsutsumi-Miyahara C, Ueda S, Kimura YN, Hamano S,

Sonoda KH, Saijo Y, Nukiwa T, Strieter RM, et al. (2005) J Clin Invest 115:2979–2991.

15. Fenoglio C, Galimberti D, Lovati C, Guidi I, Gatti A, Fogliarino S, Tiriticco M,Mariani C, Forloni G, Pettenati C, et al. (2004) Neurobiol Aging 25:1169–1173.

16. Mahad DJ, Ransohoff RM (2003) Semin Immunol 15:23–32.17. Nishimura M, Kuno S, Mizuta I, Ohta M, Maruyama H, Kaji R, Kawakami H

(2003) Movement Disorders 18:953–955.18. Abu el-Asrar AM, Van Damme J, Put W, Veckeneer M, Dralands L, Billiau A,

Missotten L (1997) Am J Ophthalmol 123:599–606.19. Capeans C, De Rojas MV, Lojo S, Salorio MS (1998) Retina 18:546–550.20. Stamatovic SM, Keep RF, Kunkel SL, Andjelkovic AV (2003) J Cell Sci 116:4615–

4628.21. Banisadr G, Queraud-Lesaux F, Boutterin MC, Pelaprat D, Zalc B, Rostene W,

Haour F, Parsadaniantz SM (2002) J Neurochem 81:257–269.22. Kalehua AN, Nagel JE, Whelchel LM, Gides JJ, Pyle RS, Smith RJ, Kusiak JW,

Taub DD (2004) Exp Cell Res 297:197–211.23. Eugenin EA, D’Aversa TG, Lopez L, Calderon TM, Berman JW (2003) J Neu-

rochem 85:1299–1311.24. Hisatomi T, Sakamoto T, Sonoda KH, Tsutsumi C, Qiao H, Enaida H, Yamanaka

I, Kubota T, Ishibashi T, Kura S, et al. (2003) Am J Pathol 162:1869–1879.25. Lewis GP, Sethi CS, Carter KM, Charteris DG, Fisher SK (2005) Mol Vis

11:491–500.26. Yang L, Bula D, Arroyo JG, Chen DF (2004) Invest Ophthalmol Vis Sci 45:648–

654.27. Wiechmann AF (1996) J Neurochem 67:105–110.

28. Carmody RJ, McGowan AJ, Cotter TG (1999) Exp Cell Res 248:520–530.29. Rotstein NP, Politi LE, German OL, Girotti R (2003) Invest Ophthalmol Vis Sci

44:2252–2259.30. Ehlers MR (2000) Microbes Infect 2:289–294.31. Patarroyo M (1994) Immunobiology 191:474–477.32. Chen L, Wu W, Dentchev T, Zeng Y, Wang J, Tsui I, Tobias JW, Bennett J,

Baldwin D, Dunaief JL (2004) Exp Eye Res 79:239–247.33. Crane IJ, McKillop-Smith S, Wallace CA, Lamont GR, Forrester JV (2001) Invest

Ophthalmol Vis Sci 42:1547–1552.34. Meleth AD, Agron E, Chan CC, Reed GF, Arora K, Byrnes G, Csaky KG, Ferris

FL, III, Chew EY (2005) Invest Ophthalmol Vis Sci 46:4295–4301.35. Zeng HY, Zhu XA, Zhang C, Yang LP, Wu LM, Tso MO (2005) Invest Ophthalmol

Vis Sci 46:2992–2999.36. Jo N, Wu GS, Rao NA (2003) Invest Ophthalmol Vis Sci 44:4054–4060.37. Nakazawa T, Shimura M, Tomita H, Akiyama H, Yoshioka Y, Kudou H, Tamai

M (2003) Curr Eye Res 26:55–63.38. Miyahara S, Kiryu J, Miyamoto K, Katsuta H, Hirose F, Tamura H, Musashi K,

Honda Y, Yoshimura N (2004) Invest Ophthalmol Vis Sci 45:4197–4201.39. Shen D, Wen R, Tuo J, Bojanowski CM, Chan CC (2006) Ophthalmic Res 38:71–73.40. Rattner A, Nathans J (2005) J Neurosci 25:4540–4549.41. Nakazawa T, Matsubara A, Noda K, Hisatomi T, She H, Skondra D, Miyahara S,

Sobrin L, Thomas KL, Chen DF, et al. (2006) Mol Vis 12:867–878.42. Ashida N, Arai H, Yamasaki M, Kita T (2001) J Biol Chem 276:16555–

16560.43. van Gassen KL, Netzeband JG, de Graan PN, Gruol DL (2005) Eur J Neurosci

21:2949–2957.44. Hutchings P, Rosen H, O’Reilly L, Simpson E, Gordon S, Cooke A (1990) Nature

348:639–642.45. Soriano SG, Coxon A, Wang YF, Frosch MP, Lipton SA, Hickey PR, Mayadas TN

(1999) Stroke 30:134–139.46. Nakazawa T, Tomita H, Yamaguchi K, Sato Y, Shimura M, Kuwahara S, Tamai

M (2002) Curr Eye Res 24:114–122.47. Marin-Teva JL, Dusart I, Colin C, Gervais A, van Rooijen N, Mallat M (2004)

Neuron 41:535–547.48. Mallat M, Marin-Teva JL, Cheret C (2005) Curr Opin Neurobiol 15:101–107.49. Lewis GP, Talaga KC, Linberg KA, Avery RL, Fisher SK (2004) Am J Ophthalmol

137:1085–1095.50. Yoshida S, Yoshida A, Ishibashi T (2004) Graefes Arch Clin Exp Ophthalmol

242:409–413.51. Lutgens E, Faber B, Schapira K, Evelo CT, van Haaften R, Heeneman S, Cleutjens

KB, Bijnens AP, Beckers L, Porter JG, et al. (2005) Circulation 111:3443–3452.52. Nakazawa T, Morii H, Tamai M, Mori N (2005) Brain Res 1061:97–106.53. Livak KJ, Schmittgen TD (2001) Methods 25:402–408.54. Yin Y, Henzl MT, Lorber B, Nakazawa T, Thomas TT, Jiang F, Langer R,

Benowitz LI (2006) Nat Neurosci 9:843–852.55. Maruyama K, Ii M, Cursiefen C, Jackson DG, Keino H, Tomita M, Van Rooijen

N, Takenaka H, D’Amore PA, Stein-Streilein J, et al. (2005) J Clin Invest115:2363–2372.

2430 � www.pnas.org�cgi�doi�10.1073�pnas.0608167104 Nakazawa et al.

Dow

nloa

ded

by g

uest

on

Apr

il 22

, 202

0