cells exposed to sublethal oxidative stress selectively attract monocytes/macrophages via scavenger...

of June 18, 2013.This information is current as

MyD88-Mediated Signalingvia Scavenger Receptors andSelectively Attract Monocytes/Macrophages Cells Exposed to Sublethal Oxidative Stress

RachmilewitzSmith, Dawn M. Bowdish, Gabriel Nussbaum and Jacob Anat Geiger-Maor, Inbar Levi, Sharona Even-Ram, Yoav

http://www.jimmunol.org/content/188/3/1234doi: 10.4049/jimmunol.1101740January 2012;

2012; 188:1234-1244; Prepublished online 4J Immunol 

MaterialSupplementary

0.DC1.htmlhttp://www.jimmunol.org/content/suppl/2012/01/04/jimmunol.110174

Referenceshttp://www.jimmunol.org/content/188/3/1234.full#ref-list-1

, 25 of which you can access for free at: cites 57 articlesThis article

Subscriptionshttp://jimmunol.org/subscriptions

is online at: The Journal of ImmunologyInformation about subscribing to

Permissionshttp://www.aai.org/ji/copyright.htmlSubmit copyright permission requests at:

Email Alertshttp://jimmunol.org/cgi/alerts/etocReceive free email-alerts when new articles cite this article. Sign up at:

Print ISSN: 0022-1767 Online ISSN: 1550-6606. Immunologists, Inc. All rights reserved.Copyright © 2012 by The American Association of9650 Rockville Pike, Bethesda, MD 20814-3994.The American Association of Immunologists, Inc.,

is published twice each month byThe Journal of Immunology

by guest on June 18, 2013http://w

ww

.jimm

unol.org/D

ownloaded from

The Journal of Immunology

Cells Exposed to Sublethal Oxidative Stress SelectivelyAttract Monocytes/Macrophages via Scavenger Receptorsand MyD88-Mediated Signaling

Anat Geiger-Maor,* Inbar Levi,* Sharona Even-Ram,† Yoav Smith,‡ Dawn M. Bowdish,x

Gabriel Nussbaum,{ and Jacob Rachmilewitz*

The innate immune system responds to endogenous molecules released during cellular stress or those that have undergone mod-

ifications normally absent in healthy tissue. These structures are detected by pattern-recognition receptors, alerting the immune

system to “danger.” In this study, we looked for early signals that direct immune cells to cells undergoing stress before irreversible

damage takes place. To avoid detecting signals emanating from apoptotic or necrotic cells we exposed fibroblasts to sublethal

oxidative stress. Our results indicate that both nonenzymatic chemical reactions and aldehyde dehydrogenase-2–mediated enzy-

matic activity released signals from fibroblasts that selectively attracted CD14+ monocytes but not T, NK, and NKT cells or

granulocytes. Splenocytes from MyD882/2 mice did not migrate, and treatment with an inhibitory peptide that blocks MyD88

dimerization abrogated human monocyte migration. Monocyte migration was accompanied by downmodulation of CD14 expres-

sion and by the phosphorylation of IL-1R–associated kinase 1, a well-known MyD88-dependent signaling molecule. The scavenger

receptor inhibitors, dextran sulfate and fucoidan, attenuated monocyte migration toward stressed cells and IL-1R–associated

kinase 1 phosphorylation. Surprisingly, although monocyte migration was MyD88 dependent, it was not accompanied by inflam-

matory cytokine secretion. Taken together, these results establish a novel link between scavenger receptors and MyD88 that

together function as sensors of oxidation-associated molecular patterns and induce monocyte motility. Furthermore, the data

indicate that MyD88 independently regulates monocyte activation and motility. The Journal of Immunology, 2012, 188: 1234–1244.

Since the introduction of the “Danger Model” (1), it is clearthat innate immunity can be stimulated not only by viral orbacterial (nonself) components but also by nonmicrobial

endogenous “danger signals.” The identity of the danger signals,and how the immune system senses stress or damage, is the focusof numerous studies that primarily center on the way the immunesystem discriminates between modes of cell death (2–4).According to this proposal, trauma is associated with tissue

damage that is recognized through the detection of “prepacked”intracellular molecules (5). These prepacked molecules includenucleic acids, lipids, and molecules such as HGMB1, ATP, and

uric acid, which are normally hidden within the cell and are re-leased from cells killed by acute abnormal death (3, 6), serving asligands to an array of receptors primarily of the TLR family (7).Recognition by TLRs activates several signaling pathways pri-marily through the common MyD88 adaptor that in turn triggersan inflammatory response. Hence, the prevailing view is that im-mune cells are both attracted to wounds and are activated bydanger signals released from damaged or necrotic cells.In contrast to necrotic cell death resulting from trauma, pro-

grammed cell death is a continuous physiological process occurringin healthy tissue. Apoptotic cells are rapidly engulfed by phago-cytes without triggering an inflammatory response, and in fact,apoptotic cell phagocytosis tends to be an anti-inflammatory triggerthat may even promote tolerance (5). Interestingly, cells dying byphysiological apoptotic death take an active role in their ownengulfment by phagocytic cells (8). They have been shown tosignal to immune cells through a variety of secreted “find-me” andexposed “eat-me” signals. Viable cells, in contrast, lack these signalsand even express “do-not-eat-me” signals. Accordingly, the com-mon features of putative signaling components that alert the im-mune system to danger is that they should not be sent by healthycells or by cells undergoing normal physiological death and shouldfulfill two critical activities; first, they should attract immune cells tothe site of damage, and second, they should activate the recruitedcells that in turn will contribute to the clearance of cell debris andcoordinate inflammation and wound repair (5).Although the outcome of many insults is cell death, in the daily

course of life, tissues and cells are constantly exposed to sublethalstressful conditions that do not necessarily end in physical damageor cause a noticeable effect. One such prevailing insult is oxidativestress. The sources of oxidants can be extrinsic or cell intrinsic,resulting from an imbalance in the oxidation/reduction process thatgenerates reactive oxygen species. These highly reactive free

*Goldyne Savad Institute of Gene Therapy, Hadassah-Hebrew University MedicalCenter, 91120 Jerusalem, Israel; †Hadassah Human Embryonic Stem Cell ResearchCenter, Goldyne Savad Institute of Gene Therapy, Hadassah-Hebrew UniversityMedical Center, 91120 Jerusalem, Israel; ‡Genomic Data Analysis Unit, HebrewUniversity Medical School, Jerusalem, Israel; xDepartment of Pathology and Molec-ular Medicine, McMaster University, Hamilton, Ontario, Canada; and {Institute ofDental Sciences, Faculty of Dental Medicine, Hadassah-Hebrew University MedicalCenter, 91120 Jerusalem, Israel

Received for publication June 17, 2011. Accepted for publication November 30,2011.

This work was supported by Israel Science Foundation Grant 337/08.

The expression data presented in this article have been deposited at the Gene Ex-pression Omnibus (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE33631)under accession number GSE33631.

Address correspondence and reprint requests to Dr. Jacob Rachmilewitz, GoldyneSavad Institute of Gene Therapy, Hadassah University Hospital, P.O. Box 12000,91120 Jerusalem, Israel. E-mail address: rjacob@cc.huji.ac.il

The online version of this article contains supplemental material.

Abbreviations used in this article: 4-HNE, 4-hydroxy-2-nonenal; IRAK, IL-1R–as-sociated kinase; LDL, low-density lipoprotein; MARCO, macrophage receptor withcollagenous structure; MDA, malondialdehyde; SCARA, scavenger receptor class A;SDF-1, stromal cell-derived factor 1; SR, scavenger receptor; WT, wild-type.

Copyright� 2012 by TheAmericanAssociation of Immunologists, Inc. 0022-1767/12/$16.00

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1101740

by guest on June 18, 2013http://w

ww

.jimm

unol.org/D

ownloaded from

radicals can cause damage to lipids, proteins, and nucleic acids (9–11). In turn, these oxidatively modified cellular components can actas endogenous ligands of pattern recognition receptors, in partic-ular, members of the scavenger receptor (SR) family (12–14).Whereas the mechanisms and signals involved in the way the

immune system deciphers between modes of cell death are wellstudied, the immune response to stress before potential damagetakes place have received little attention. The failure to assist cellssuffering from sublethal stress and to remove toxic substances andcounteract their harmful effects may often lead to tissue dys-function and disease. Therefore, insights into the mechanisms thatallow immune cells to respond to early signs of stress is funda-mental to our understanding of how immune cells maintain ho-meostasis. To study this pivotal issue of how the immune systemsenses and responds to stress and to determine the danger signalsexpressed by cells undergoing sublethal stress, we exposed primarycultures of human fibroblasts to oxidative stress and tested theirability to attract immune cells. Most previous studies dealing withimmune responses to oxidative stress involved experimental con-ditions that induce tissue damage or result in cell death. In the cur-rent study, we applied sublethal conditions to detect stress signalsas such and to avoid signals released from dying cells (either apopto-tic or necrotic cell death).In this article, we present evidence that normal healthy cells

exposed to sublethal oxidative stress release chemotactic signals thatselectively attract monocytes. Nonenzymatic chemical reactions andspecifically, the derivative of lipid peroxidation malondialdehyde(MDA), play a role in the generation of attractive signals.We furtheridentified the mitochondrial enzyme ALDH-2 as an additionalplayer in the process that generates attraction signals. These earlysignals induce selective and rapid attraction of monocytes in an SRand MyD88-dependent manner. However, although one would ty-pically have expected this MyD88-dependent monocyte migrationto be part of an inflammatory cycle induced by oxidized fibro-blasts, our data indicate that the stressed fibroblasts do not in factprovoke monocyte proinflammatory cytokine secretion.

Materials and MethodsCells

Primary human fibroblasts were provided by Dr. G. Zamir from the De-partment of Surgery, Hadassah Hebrew-University Medical Center (Jer-usalem, Israel). Cells were maintained (not more than eight passages) ina high-glucose DMEM (Life Technologies) supplemented with 10% heat-inactivated FCS, 1% sodium pyruvate, and 1% penicillin/streptomycin(Biological Industries, Beit-Haemek, Israel) at 37˚C and 5% CO2. Foroxidative stress induction, fibroblasts were washed with PBS (containingCa2+ and Mg2+) and then incubated with either H2O2 diluted in PBS or withPBS alone at 37˚C. After 1 h, fresh medium was added, and cells werewashed and then supplemented with fresh media. PBMC were purified fromthe venous blood of healthy donors using Ficoll gradient (Histopaque 1077;Sigma-Aldrich, St. Louis, MO). CD14+ cells were isolated by negativeselection using the RosetteSep enrichment mixture (StemCell Technolo-gies, Vancouver, BC, Canada), according to the manufacturer’s instructions.For MyD88 inhibition, PBMC or isolated CD14+ cells were incubated witheither the inhibitor peptide or its scrambled control peptide (15) for 30 minat 37˚C prior to use. Alda-1 (16, 17) was kept in 70% ethanol at a con-centration of 4 mM and was added to fibroblasts in a final concentrationof 20 mM. Splenocytes were isolated using Ficoll gradient from 8 to 12C57BL/6 wild-type (WT) and MyD882/2 mice, age and sex matched.

Transwell migration assay

A total of 0.53 106 PBMC were placed in the upper chamber of Transwellmembranes (5-mm pore size) and placed in wells containing either mock-treated or peroxide-treated fibroblasts or media alone. PBMC were allowedto migrate at 37˚C and 5% CO2 for 1 h. After 1 h, cells were harvestedfrom both the upper and lower chambers, immunostained as indicated, andthen analyzed and counted by flow cytometry. The data are presented asthe percentage of migrating cells calculated for each cell type by dividing

the number of cells in the lower chamber by the total number of cells(number cells in the upper and lower chambers).

Flow cytometry

Cells were washed twice in PBS and stained using FITC-conjugated anti-CD16, PE-conjugated anti-CD56, and allophycocyanin-conjugated anti-CD3 or FITC-conjugated anti-CD14 or FITC-conjugated anti-CD16 andallophycocyanin-conjugated anti-CD14 (IQProducts, Groningen, The Neth-erlands) for 30 min on ice. Cells were washed twice in PBS, and the im-munostained cells were counted and analyzed by a FACSCalibur flow cy-tometer (BD Biosciences, San Jose, CA) using Cell Quest software.

Peritoneal in vivo infiltration assay

WT C57BL/6 or MyD882/2, 8- to 12-wk-old mice were injected with 1 mlPBS, PBS supplemented with 500 mM H2O2, or 1 3 106 peroxide- ormock-treated fibroblasts directly into the peritoneal cavity, and 4 d later,the peritoneum was washed with PBS with heparin (5 U/ml), and cellswere collected and counted by hemocytometer. The number of F4/80+

(eBioscience, San Diego, CA) cells in each group was determined by flowcytomery. As positive control, mice were injected with thioglycollate (BD,Le Pont de Claix, France).

Cell lysis and immunoblotting

A total of 0.5 3 106 monocytes were placed on the upper chamber ofTranswell membranes (0.4-mm pore size) placed in wells containing eithermock-treated or peroxide-treated fibroblasts or medium alone and incu-bated at 37˚C for the indicated time points. At each time point monocyteswere collected from the Transwell and whole-cell extracts were preparedby lysing the cells with lysis buffer (0.5% Nonidet P-40, 50 mM Tris-HCl[pH 8], 100 mM NaCl, 1 mM PMSF, 1 mM sodium orthovanadate, 10 mg/ml leupeptin, and 10 mg/ml aprotinin) for 30 min on ice. Lysates wereseparated by electrophoresis on 10% SDS-PAGE gels and then transferredto polyvinylidene difluoride membranes (Bio-Rad, Hercules, CA). Theblots were probed with anti-phosphorylated IL-1R–associated kinase(IRAK)1 Ab (Ser376; Acris Abs, Herford, Germany), processed with ECLplus Western blotting detection system (Amersham Biosciences), and ex-posed to Chemiluminescence BioMax Light Film (Kodak Industries, Paris,France). Following stripping, the membranes were reprobed with anti-Actin mAb (MP Biomedicals, Illkirch, France).

Confocal microscopy

For confocal microscopy analysis, 0.53 106 monocytes were plated on theupper chamber of Transwell membranes (0.4-mm pore size) and placedin wells containing either mock-treated or peroxide-treated fibroblastsor medium alone for 10 min. In experiments using the MyD88 inhibitorpeptide and its scramble control, monocytes were pretreated for 30 minprior to placing them on the Transwell membranes. At the end of the in-cubation time, Transwell membranes were removed and washed with icecold PBS, and then, the cells were fixed with 4% PFA for 20 min, at roomtemperature, followed by permeabilization with 0.5% Triton X-100 and4% PFA in PBS containing 5% sucrose for 5 min. Cells were then stainedwith Alexa Fluor 488-conjugated-phalloidin (Invitrogen, Eugene, OR) andDAPI (Calbiochem, Darmstadt, Germany). Transwell membranes weremounted on cover slides with FluoroGel mounting buffer (EMS). Imageswere obtained by a laser scanning confocal microscope system (Fluoview-1000; Olympus) with a 360 Uplan-SAPO objective.

RT-PCR analysis for SRs expression. Total RNAwas isolated from isolatedmonocytes using PerfectPure RNA Cultured Cell Kit (5 PRIME, Gai-thersburg, MD). Specific primers for SR-AI, SR-AII, macrophage receptorwith collagenous structure (MARCO) and scavenger receptor class A(SCARA)-5, SCARA-3, and SCARA-4, and RT-PCR protocols wereaccording to DeWitte-Orr et al. (18).

Cytokine production

Cultures containing 0.15 3 106 fibroblasts, mock or peroxide treated, withor without 0.6 3 106 monocytes were plated in 24-well plates (Corning,Corning, NY). As positive controls, monocytes were incubated with 10 ng/ml LPS (Sigma-Aldrich) or with 0.15 3 106 necrotic fibroblasts. Cellswere incubated for 24 h, and conditioned media were collected. Mediumwas also collected from monocytes cultured alone. TNF-a and IL-1blevels in the conditioned media were assayed by ELISA (R&D Systems).

Gene array

Cultures containing 0.15 3 106 fibroblasts, mock or peroxide treated,with or without 0.6 3 106 monocytes were plated in 24-well plates

The Journal of Immunology 1235

by guest on June 18, 2013http://w

ww

.jimm

unol.org/D

ownloaded from

(Corning). As control, monocytes were incubated alone. After 24 h, RNAwas extracted using the PerfectPure RNA Cultured Cell Kit (5 PRIME,Hamburg, Germany). Amplified and biotinylated sense-strand DNAwere prepared according to the standard Affymetrix protocol from 200ng total RNA (Expression Analysis Technical Manual; Affymetrix).Following fragmentation, 5.5 mg biotinylated ssDNA was hybridizedfor 17 h at 45˚C on GeneChip Human Gene 1.0 ST Array. GeneChipwere washed and stained in the Affymetrix Fluidic Station 450. Gen-eChips were scanned using Affymetrix GeneChip Scanner 3000. Ex-pression data were deposited at the Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE33631; accession numberGSE33631).

ResultsCell motility is essential to the function of immune cells, whichcirculate throughout the body, seeking out signs of infection ortissue damage. After detecting find-me signals, immune cells canquickly and efficiently localize at site of damage or stress. To lookfor early find-me signals that alert immune cells to stress, we ex-posed fibroblasts to sublethal oxidative stress (Supplemental Fig.1). This approach enabled us to detect signals emanating fromstressed cells per se and to avoid any signals released from apo-ptotic or necrotic dying cells. We chose oxidative stress as a modelbecause generation of reactive oxygen species is a downstream ef-fect common to many biochemical processes and cellular stressors(19). A Transwell assay was established in which human PBMCwere allowed to migrate toward fibroblasts that were pretreatedwith hydrogen peroxide. The pretreatment strategy enabled us tolook for soluble find-me signals emanating from fibroblasts thatwere exposed to stress and to avoid any direct effect of the per-oxide on the immune cells that may directly induce their motility(specifically given that hydrogen peroxide gradients have beenshown to induce leukocyte migration (20)).Using the Transwell assay described above, we found that

PBMCs migrated much more avidly toward stressed fibroblaststhan toward control fibroblasts or fresh media (Fig. 1A). Cell typeanalysis revealed that a relatively high percentage of CD14+ cellsmigrated specifically toward the stressed cells (Fig. 1B), whereasonly ∼2–5% of CD3+ T cells, NKT cells, NK cells, or dendriticcells migrated nonspecifically to the lower chamber (no differencebetween migration toward stressed versus mock-treated cells).Interestingly, no granulocytes were observed in the lower chamberin experiments where whole blood was used instead of PBMCs(data not shown). These results demonstrate selective mobilizationof CD14+ cells toward sublethally stressed cells. The extent ofmonocyte migration correlated with the concentration of peroxideadded (Fig. 1C), and interestingly, monocyte migration towardstressed cells was evident as early as 15 min after placing thePBMC in the Transwell (Fig. 1D) and increased with time. Themonocytes accounted for 60–70% of migrating leukocytes withmaximal accumulation at 1–2 h. This selective migration contraststo PBMC migration toward the chemokine stromal cell-derivedfactor 1 (SDF-1), which is characterized by migration of bothmonocyte and lymphocyte populations (data not shown). Of note,similar results were obtained whether migration assays were per-formed in the presence of serum in the media or in its absence(data not shown). In humans, circulating monocytes are dividedinto two major subsets on the basis of CD14 and CD16 expression(21). The two subsets are CD14highCD162 monocytes andCD14lowCD16+ monocytes that represent ∼80–90 and ∼10–20%of circulating monocytes, respectively. Subset analysis of mono-cytes migrating in response to fibroblasts treated with peroxiderevealed that the CD14highCD162 but not CD14lowCD16+ subsetmigrated toward stressed cells (Fig. 1E).A closer look at CD14 staining reveals that surface CD14 ex-

pression was downregulated on monocytes that migrated toward

oxidized fibroblasts (i.e., those found in the lower chamber).Moreover, this downregulation was observed only in cells thatmigrated toward oxidized fibroblasts and not in monocytes thatremained in the upper chamber or cells that migrated toward mock-treated fibroblasts or to SDF-1 (Supplemental Fig. 2A), pointing toa possible role for CD14 in monocyte migration toward suble-thally stressed fibroblasts. Because CD14 is a coreceptor forseveral TLRs and was shown to directly bind several pathogen-associated molecular patterns as well as danger-associated mo-lecular patterns, we first tested whether CD14 is involved in therecognition of attractive signals by adding a soluble form of CD14to the Transwell system. Soluble CD14 has been shown to inhibitLPS-mediated responses (22–25). However, addition of solubleCD14 did not change the percentage of monocytes in the lowerchamber of both mock-treated and stressed fibroblasts, suggestingit is not involved in ligand recognition and monocyte attraction(Supplemental Fig. 2B). Another component of the TLR4 complexis MD2. MD2 expression is required for cell surface expression ofTLR4 and for optimal ligand binding. For example, MD2 extractsLPS from circulating CD14–LPS complexes and transfers the LPSinto a ternary complex with TLR4 (26, 27). The addition of asoluble chimeric TLR-4/MD2 protein had no effect on monocytemigration as well (data not shown). Thus, the CD14/TLR-4/MD2complex does not seem to play a role in driving monocyte mi-gration toward stressed cells in our system.Almost all TLRs use the adaptor molecule MyD88, which plays

a crucial role in the signaling pathways triggered by IL-1 and TLRs.To test whether the IL-1/TLR family plays a role in the migratoryresponse of monocytes toward oxidized fibroblasts, we blockedMyD88-mediated signaling using a peptide that inhibits MyD88homodimerization (15). Pretreatment of PBMC with the MyD88–homodimerization inhibitory peptide strongly reduced monocytemigration induced by oxidized fibroblasts, whereas a scrambledpeptide with the same amino acid composition used as control wasinactive (Fig. 2A). This indicates that MyD88-mediated signalingis an essential step for induction of monocyte migration by signalsemanating from cells exposed to oxidative stress.To further study the role of MyD88 in monocyte motility in re-

sponse to stress, we turned to a murine cell system. Splenocytesisolated from C57BL/6 mice migrated in a Transwell system towardoxidized human fibroblasts, albeit at a lower efficiency whencompared with human monocytes (Fig. 2B, upper panel). In con-trast to human PBMC, the cells that migrated to the lower chamberconsisted of both F4/80+ cells and lymphocytes (data not shown).In contrast, splenocytes derived from MYD882/2 mice failed tomigrate toward oxidized fibroblasts (Fig. 2B, lower panel), againsupporting the role of this signaling adaptor in monocyte migration.We also tested the ability of oxidized fibroblasts to recruit

macrophages in a peritoneal in vivo infiltration model. To this end,peroxide- and mock-treated fibroblasts were injected directly intothe peritoneal cavity, and 4 d later, the peritoneumwas washed withPBS, and cells were collected and counted. Supporting our previousin vitro results, there was considerably more recruitment of cellsafter injection of oxidized fibroblasts compared with mock-treatedcells (∼3-fold and higher) or to PBS alone or peroxide diluted inPBS (Fig. 2C). The chemotactic activity of oxidized fibroblastswas comparable (although lower) to that seen after injection ofthioglycollate used as a positive control because of its potentability to recruit macrophages. Furthermore, flow cytometricanalysis revealed that, like thioglycollate-stimulated peritonealcells, the recovered cells were mostly F4/80+. We repeated theexperiment above, this time counting specifically F4/80+ cellsusing flow cytometry. These experiments corroborated the resultsabove whereby oxidized fibroblasts recruited significantly more

1236 MyD88-DEPENDENT MACROPHAGE MOTILITY

by guest on June 18, 2013http://w

ww

.jimm

unol.org/D

ownloaded from

F4/80 cells as compared with mock-treated cells or PBS alone(Fig. 2D). In contrast, injection of peroxide-treated fibroblasts tothe peritoneum of MyD882/2 mice did not attract significantnumbers of macrophages (Fig. 2D). Thus, these data underline thebiological relevance of our findings and suggest that oxidativestress-related attraction signals play a role in MyD88-dependentmacrophage recruitment in vivo.To gain further insight into MyD88-mediated monocyte motility

in response to oxidized cells, we examined the morphology ofmonocytes exposed to the various treatments. Monocytes wereplaced on a Transwell membrane over either media or mock-treatedversus H2O2-treated fibroblasts in the lower chamber. After 10min, monocytes were fixed and stained. Confocal microscopyimages of actin and nuclei (phalloidin/DAPI)-stained monocytesplaced over stressed fibroblasts demonstrated a rounded shapewith cells protruding into the pores in contrast to a more flattenedmorphology of control cells (i.e., placed over media or mock-treated cells) (Fig. 3A, 3C). Monocytes placed over stressed fi-broblasts and treated with MyD88 inhibitory peptide acquired amore rounded shape, as compared with cells treated with scram-bled control peptide, with fewer membrane projections (Fig. 3B,3C). These morphological changes are illustrated in Fig. 3D.MyD88 inhibitory peptide had no significant effect on the mor-phology of the control cells (data not shown). These data clearlydemonstrate specific and MyD88-dependent changes in monocytemorphology in response to soluble factors released from peroxide-treated cells. The unique morphology acquired by cells treatedwith MyD88 inhibitory peptide may be an outcome of the partial

response of these cells to the attractive signal emanating fromstressed fibroblasts.To further establish the link between MyD88 and monocyte

motility, we investigated the induction of a downstream MyD88signaling event in response to signals emanating from stressedcells. Interaction of MyD88 with the intracellular domain of anactivated TLR initiates a signaling cascade. Once recruited by thereceptor’s TIR domain, MyD88 associates with the IRAK4. Thisevent leads to recruitment and phosphorylation of IRAK1 (28).IRAK1 phosphorylation was evaluated in cell extracts frommonocytes that were either exposed to mock- or H2O2-treatedfibroblasts. To detect signaling events within the monocytes thatare the consequence of the soluble find-me signal released by thefibroblasts, the monocytes were placed on a 0.4-mm Transwellmembrane, and either mock-treated or H2O2-treated fibroblastswere placed in the lower chamber. At various time points, mono-cytes were collected and lysed, and cell extracts were fractionatedon SDS-PAGE and immunoblotted using anti–phospho-IRAK1Ab. Monocytes placed over peroxide-treated fibroblasts dis-played a bimodal pattern starting with rapid IRAK1 phosphory-lation at 1–2 min, followed by a slight decline that increased againat 10–15 min after stimulation. In contrast, only a faint signal wasobserved in monocytes that were placed over mock-treated fi-broblasts (Fig. 4A). In monocytes stimulated with LPS or necro-tic fibroblasts, used as positive controls for MyD88-mediated sig-naling, a strong and gradual increase in IRAK1 phosphorylationwas detected, peaking at 5–10 min following stimulation (Fig.4B). We further tested the phosphorylation of downstream targets

FIGURE 1. CD14+ cells are selectively attracted

toward H2O2-treated fibroblasts. Fibroblasts were

either treated with PBS supplemented with the in-

dicated concentrations of H2O2 (100–500 mM) or

mock treated with PBS alone (0). After 1 h, the

cells were washed and fresh culture media was

added; then PBMC were placed over a Transwell

(5.0 mm) and allowed to migrate for 1 h. Then

Transwells were removed and cells were collected

from both the upper and lower chambers. Cells

were immunostained and then analyzed and coun-

ted by flow cytometry. The data are presented as

percentage of migrating cells. A, The average of

four separate experiments is graphed as percentage

of migrating PBMC. B, The average of four separate

experiments is graphed as percentage of CD14+ cells

in the lower chamber. C, Migration of CD14+ cells

toward fibroblasts treated with increasing concen-

trations of H2O2 is presented. One of four repre-

sentative experiments is shown. D, CD14+ cell mi-

gration can be observed as early as 15 min. Similar

results were obtained in three independent experi-

ments. E, Peroxide-treated fibroblasts selectively

attract CD14highCD162 human monocyte subset.

Dot plot flow cytometric analysis of monocyte

subset distribution based on CD16 and CD14 ex-

pression in PBMCs (input) and in cells collected

from the lower chamber following migration toward

peroxide-treated fibroblasts (migrating). Similar re-

sults were obtained in four independent experi-

ments. *p , 0.05, **p , 0.005.

The Journal of Immunology 1237

by guest on June 18, 2013http://w

ww

.jimm

unol.org/D

ownloaded from

p38 and AKT as well as NF-kb nuclear translocation. Interest-ingly, there was no significant change in the level of activity ofthese intermediates in monocytes incubated in the presence ofperoxide-treated fibroblasts as compared with monocytes incu-bated alone or in the presence of mock-treated fibroblasts (Sup-plemental Fig. 3). These data demonstrate that soluble factorsreleased from cells undergoing oxidative stress, but not mock-treated cells, induce MyD88-dependent signaling. It is temptingto suggest that the observed bimodal pattern of IRAK1 phos-phorylation is a reflection of multiple molecular signals that arereleased at various time points from peroxide-treated cells.Having shown that soluble factors released from oxidized cells

can selectively attract monocytes in a MyD88-dependent manner,we next carried out several lines of experiments to characterize thenature of the attraction signal(s). Our data suggest that at least someof the attractive signals originate from pre-existing molecules thatare either hidden and then revealed or changed as a result ofoxidation and then released through the action of nonenzymaticchemical reactions following exposure to the oxidative agent (Sup-plemental Fig. 4A), whereas other chemotactic factors are gener-ated by the enzymatic activity of ALDH-2 (Supplemental Fig.4B).Protein, carbohydrate, and lipid peroxidation are well estab-

lished mechanisms of cellular injury with the latter used as anindicator of oxidative stress in cells and tissues. Lipid peroxides,derived from polyunsaturated fatty acids, are unstable and de-compose to form complexes with a series of compounds that arereleased from the cell surface, which include reactive carbonylcompounds, such as 4-hydroxy-2-nonenal (4-HNE) andMDA (26).

We therefore assessed MDA and 4-HNE as potential find-mesignals in a migration assay. Because these compounds are ex-tremely unstable, we used in the case of MDA a stable analogcalled MDA-Bis (dimethyl acetal). MDA-bis is used as standard inthe thiobarbituric acid reactive substance assay used to measureMDA levels in the serum. MDA-bis added to the lower chamberinduced selective migration of monocytes similar to that observedwith oxidized fibroblasts (Fig. 5A). Interestingly, 4-HNE failed toinduce monocyte migration at all concentrations tested, possiblybecause of the instability of this compound. Furthermore, althoughmacrophages from splenocytes isolated from WT mice migratedin response to MDA-bis, macrophages derived from MYD882/2

mice failed to do so (Fig. 5B). In addition, treatment of humanmonocytes with MDA-bis induced weak but rapid and significantIRAK1 phosphorylation in a dose- and time-dependent manner(Fig. 5C). Taken together, these data demonstrate that MDA-bisselectively induces monocyte migration in a MyD88-dependentmanner, replicating, at least partially, the effects seen with oxi-dized cells. Although in these experiments, we have used an an-alog of MDA, the data establish that compounds such as MDA orone of its derivatives, may represent at least part of the attractivesignal released from oxidized cells.Although our data established that monocyte motility in response

to signals released from cells undergoing oxidative stress is de-pendent on MyD88 activity, the search using blocking Ab or cellsfrom knockout mice for specific TLRs that mediate binding of find-me signals originating from oxidized cells did not yield conclusiveresults. It seems likely that this is due to some level of redundancyin the receptors and/or the ligands. It is noteworthy that, in general,

FIGURE 2. Human CD14+ cells and mouse mac-

rophage migration toward H2O2-treated fibroblasts is

MyD88 dependent. A, PBMC were pretreated for 30

min with either MyD88 inhibitory peptide (gray bars)

or scrambled peptide (black bars) and then were

placed in a Transwell over either mock (0)- or per-

oxide-treated fibroblasts and allowed to migrate for

1 h. Cells were analyzed as in Fig. 1. The percentage

of CD14+ cell migration is shown. One of five repre-

sentative experiments is shown. B, Cell migration was

performed as above using splenocytes derived from

either WT or MyD882/2 mice instead of human

PBMCs. The percentage of macrophage migration is

shown. Similar results were obtained in three inde-

pendent experiments. C, In vivo cell recruitment into

the peritoneal cavity in response to oxidized fibroblasts

was tested. Mice were injected as described in Mate-

rials and Methods. The average number of cells in the

peritoneal cavity of each group of mice is shown. One

of three representative experiments is shown. The

majority of cells recovered from the peritoneal cavity

of mice injected with peroxide treated fibroblasts was

identified as F4/80+ and are similar to those recovered

from mice injected with thioglycollate used as a posi-

tive control. D, Either PBS or peroxide- and mock-

treated fibroblasts were injected directly into the

peritoneal cavity of either WT or MyD882/2 mice and

4 d later, the cells were collected as described and the

number of F4/80+ cells in each group was counted by

flow cytometry. Comparable results were obtained in

three independent experiments.

1238 MyD88-DEPENDENT MACROPHAGE MOTILITY

by guest on June 18, 2013http://w

ww

.jimm

unol.org/D

ownloaded from

ligands of >TLRs require presentation via a coreceptor. The cor-eceptor CD14 has been implicated in facilitating TLR-mediatedresponses to bacterial lipopeptides by enhancing the physical pro-ximity of the ligand to the TLR heterodimers, without binding di-rectly to the receptor complex (30–32). However, as demonstratedabove, soluble CD14 did not inhibit cell migration in our system

(Supplemental Fig. 2B), suggesting that CD14 is not likely to playa role as coreceptor in our system.SRs are another group of receptors that may bind ligands in our

system. SRs bind a range of ligands of endogenous and exogenousorigin with relatively low affinity. These ligands include, amongothers, oxidized lipids. By scavenging oxidized lipids from the en-vironment, these receptors protect against oxidants and decrease in-flammation (33). In addition to their role in clearance of these alteredendogenous molecules, SRs have been recently identified as com-ponents that coordinate and enhance the response to TLR ligands(18, 34, 35). These observations led us to investigate whether SRsmay be involved in the attraction of monocytes toward stressed cells.Monocyte migration toward stressed cells was abrogated by

treatment with SR-specific competitive ligands, fucoidan anddextran sulfate (DxSO4), but not their noncompetitive counterparts,fetuin, and chondroitin sulfate (ChSO4) (Fig. 6A, 6B). Monocytemigration in response to SDF-1, used as control, was not attenu-ated by the addition of either DxSO4 or fucoidan (data not shown).In addition, both DxSO4 and fucoidan inhibited IRAK1 phos-phorylation induced by stressed cells (Fig. 4A), thus linking SRs toMyD88 signaling. Given that these competitive inhibitors arespecific for SR class A receptors (36), we next profiled class A SRexpression in human monoyctes using RT-PCR with previouslydescribed gene-specific primers (18). Human monocytes werefound to express SR-AI, SR-AII, MARCO, and SCARA-5.SCARA-3 and SCARA-4 transcripts were undetectable (Fig.6C). Treatment of monocytes with anti–SR-AI and AII-neu-tralizing Abs did not attenuate migration toward stressed fibro-blasts (data not shown), suggesting the involvement of the otherSR-A members (i.e., either MARCO or SCARA-5) to which spe-cific blocking Abs are not available. Taken together, these re-sults indicate that the class A SRs play an important role in the li-gand binding event that induces MyD88-dependent signaling andmonocyte migration.Finally, we examined whether another type of stress, hypoxia,

can induce monocyte migration in a similar manner. To that end, weevaluated PBMC migration toward fibroblast cultures exposed to

FIGURE 3. MyD88-dependent changes in monocyte morphology in response to oxidized fibroblasts. Monocytes were either left untreated or were

pretreated for 30 min with either MyD88 inhibitory peptide or scrambled peptide. Then the monocytes were placed in a Transwell (5.0 mm) over fibroblasts

that were either mock treated or peroxide treated for 10 min. After 10 min, monocytes were fixed, stained, and evaluated for cell morphological char-

acteristics using confocal microscopy. A, Dual-color fluorescent upper view three-dimensional images of monocytes demonstrating morphological changes

after incubation with oxidized fibroblasts (H2O2; right panel) compared with monocytes incubated with mock-treated fibroblasts (PBS; left panel). B,

Three-dimensional images of monocytes after incubation with MyD88 inhibitory peptide (right panel) or scrambled peptide (left panel) that were cultured

over oxidized fibroblasts. C, Three-dimensional composite of Z-stack images taken by confocal microscopy showing side views of monocytes placed over

fibroblasts. Dotted lines represent the Transwell membrane. The data are representative of four independent experiments. Original magnification 360 (A–

C). D, An illustration depicting changes in monocytes morphology with the various treatments.

FIGURE 4. Phosphorylation of the MyD88-dependent signaling mole-

cule, IRAK1, is induced in monocytes by oxidized fibroblasts. A, Fibro-

blasts were plated in a 24-well dish and were either mock or peroxide

treated as above. Monocytes were then placed in the upper chamber of

a 0.4-mm Transwell and incubated for the indicated time. In parallel wells,

either fucoidan (100 mg/ml) or DxSO4 (100 mg/ml) was added, as indi-

cated. Following incubation, the Transwells were removed and the

monocytes were collected, washed with cold PBS, and lysed. B, As

a positive control, monocytes were incubated with LPS (10 ng/ml) or

necrotic fibroblasts for the indicated times and then washed and lysed. Cell

extracts were subjected to SDS-PAGE and anti-phosphorylated IRAK1

immunoblotting (upper panels). Anti–b-actin immunoblotting was used to

detect the relative amounts of protein in each lane (lower panels). Similar

results were obtained in three independent experiments.

The Journal of Immunology 1239

by guest on June 18, 2013http://w

ww

.jimm

unol.org/D

ownloaded from

continuous hypoxia for 2–24 h, as described previously. Similar tooxidative stress, we observed selective mobilization of CD14+

cells toward fibroblasts that were exposed to hypoxia. The numberof monocytes increased with the duration of hypoxia, peaking at18 h (Fig. 7A). Surprisingly, in contrast to monocyte migrationtoward peroxide-treated fibroblasts, treatment of PBMC with theMyD88 inhibitory peptide did not inhibit monocyte migration to-ward fibroblasts exposed to hypoxia (Fig. 7B), suggesting thatdifferent attraction signals/receptors are being used.Inflammatory cytokine secretion is a predictable response fol-

lowing monocyte and macrophage recruitment to sites of damage.Moreover, MyD88 is a well characterized signal transducer of in-nate immune responses that lead to monocyte and macrophage re-lease of proinflammatory mediators. Next, we asked whether fol-lowing the migration of monocytes toward oxidized fibroblasts theyinteract with the latter and respond by proinflammatory cytokinesecretion. For that end, monocytes were incubated either alone or

were cocultured with mock- or H2O2-treated fibroblasts for 24 h,and the levels of cytokines were determined in the conditionedmedia. No secretion of either TNF-a or IL-1b was detected inconditioned media of monocytes alone or their coculture witheither oxidized or mock-treated cells. In contrast, exposure ofmonocytes to LPS or necrotic fibroblasts resulted in significantTNF-a and IL-1b release (Fig. 8A). Gene array analysis corrob-orated these findings by revealing that there is no significantchange in the expression of genes associated with inflammation orphagocytosis in cocultures of monocytes and peroxide-treatedfibroblasts as compared with monocytes cocultured with mock-treated fibroblasts (Fig. 8B), suggesting that overall, inflammationis not induced by exposure to sublethally stressed fibroblasts,despite the attraction of monocytes.Whereas most studies attribute a proinflammatory role to reactive

oxygen species and specifically to oxidized phospholipids (37, 38),more recent studies demonstrate that some products of phospho-

FIGURE 5. MDA-bis induces monocyte motility and

MyD88 signaling. A, PBMCs were placed in a 5.0-mm

Transwell in the absence or presence of the indicated

concentrations of either 4-HNE or MDA-bis in the lower

chamber. After 1 h, cells were analyzed as in Fig. 1. The

percentage of migrating CD14+ cells is shown. No other

cell type analyzed including T, NK, and NKT cells and

granulocytes significantly migrated in response to MDA-

bis. One of three representative experiments is shown. B,

Cell migration was performed as above using spleno-

cytes derived from either WTor MyD882/2mice instead

of human PBMCs. The percentage of migrating F4/80+

macrophages in the absence (black bars) or presence of

10 mM MDA-bis (gray bars) is shown. Similar results

were obtained in three independent experiments. C,

Monocytes were incubated for various time points with

10 or 100 mM MDA-bis. At the indicated time point,

monocytes were washed with cold PBS and lysed. Cell

extracts were subjected to SDS-PAGE and immuno-

blotted with anti-phosphorylated IRAK1 (upper pan-

els). Anti–b-actin immunoblotting reveals the relative

amounts of protein in each lane (lower panels). Similar

results were obtained in three independent experiments.

FIGURE 6. Class A SRs mediate MyD88 signaling

and monocyte motility in response to peroxide-trea-

ted fibroblasts. Fibroblasts were either treated with

PBS supplemented with the indicated concentrations

of H2O2 (500 mM) or mock treated with PBS only

(0). After 1 h, the cells were washed and fresh culture

media was added; then PBMC were placed over

a Transwell (5.0 mm) and allowed to migrate for 1 h

as above in the presence of SR inhibitors or their

controls: ChSO4 (black bars) or DxSO4 (gray bars)

(A) and fetuin (black bars) or fucoidan (gray bars)

(B). Transwells were then removed and cells were

collected from both the upper and lower chambers.

Cells were immunostained and then analyzed and

counted by flow cytometry. The data are presented as

the percentage of migrating CD14+ cells. C, Total

RNAwas isolated from monocytes and was subjected

to RT-PCR using specific primers to SR-AI, SR-AII,

MARCO, SCARA3, SCARA4, and SCARA5 tran-

scripts. Similar results were obtained with monocytes

from three different donors.

1240 MyD88-DEPENDENT MACROPHAGE MOTILITY

by guest on June 18, 2013http://w

ww

.jimm

unol.org/D

ownloaded from

lipid oxidation may in fact exhibit anti-inflammatory properties(38–40). Specifically, these studies demonstrated that oxidizedphospholipids inhibit TLR signaling in response to agonist stim-ulation (40–44). To test whether oxidized cells induce monocyteunresponsiveness, we stimulated monocytes with either LPS ornecrotic fibroblasts in the absence or presence of oxidized fibro-blasts. Apoptotic fibroblasts were used as a positive control for theability to inhibit the monocyte proinflammatory response. As ex-pected, apoptotic cells significantly inhibited TNF-a and IL-1bsecretion in response to both LPS and necrotic cells; however,oxidized fibroblasts had no effect, ruling out any inhibitory activity(Fig. 8C). In aggregate, although MyD88 clearly plays a role inmonocyte migration toward cells exposed to sublethal oxidativestress, it does not activate a downstream proinflammatory cytokineresponse, suggesting that MyD88 independently regulates thesetwo functions.

DiscussionNecrotic cells lose membrane integrity and release endogenousmolecules that serve as danger signals. Once revealed, these hiddensignals alert immune cells to the site of cell death and tissue dam-age and stimulate the generation of proinflammatory mediators. Bymaintaining membrane integrity, at least initially, apoptotic cellsare removed by phagocytes without the induction of an inflam-matory response. Previous studies have extensively studied the waythe immune system discriminates between live, necrotic, and ap-

optotic cells. Specifically, the molecular nature of how dying cellsalert the immune system to danger has been characterized (3). Bysubjecting normal fibroblasts to sublethal oxidative stress, in thecurrent study, we were able to examine the response of immunecells to cellular stress per se, and the molecular identity of thesignals emitted during stress. Our data show that fibroblasts ex-posed to sublethal doses of peroxide selectively attract CD14high

CD162 human monocytes as well as mouse F4/80 macrophages.In addition to the attraction of monocytes and macrophagesin vitro, oxidized fibroblasts also induced the attraction of mac-rophages upon i.p. injection in vivo. Chemoattraction of murinemacrophages (both in vitro and in vivo) to oxidized human fi-broblasts suggests that the signals and mechanism involved in thisprocess are conserved across these species.Interestingly, we observed that, following migration of mono-

cytes in response to stressed fibroblasts, they expressed lower levelsof CD14 on their cell surface. However, CD14 expression does notappear to be directly responsible for “sensing” signals released fromstressed cells, because cell migration could not be competed byadding excess soluble CD14. Truman et al. (45) demonstrated thatmacrophages that had transmigrated in response to apoptotic Bur-kitt’s lymphoma cells showed increased CD14 expression. Despitethe altered surface expression, CD14 in their study was also notinvolved in detecting the chemotactic signal. The authors suggestedthat this change in CD14 expression levels might represent readi-ness of the migrating monocytes for apoptotic cell clearance, asdemonstrated previously (46).We found that blocking MyD88-dependent signaling abrogated

monocyte migration both in vitro and in vivo, demonstrating thatsignals released from stressed cells do not act through conventionalchemotactic receptors but rather initiate signal transduction for mi-gration through a yet-to-be defined sensor that signals throughMyD88. In support of MyD88-dependent monocyte motility, Coteet al. (47) have demonstrated that in MyD882/2 mice exposed tothe toxin MPTP, infiltration of monocytes/macrophages to theCNS was significantly lower as compared with WT mice. It isnoteworthy that MPTP can lead to mitochondrial dysfunction andincreased oxidative stress (48).Several lines of evidence argue against the possibility that in-

hibition of MyD88 signaling causes a generalized defect in cellmigration. First, thioglycollate-induced macrophage infiltration inMyD882/2 mice is comparable to that observed in wild-typemice. Second, human monocytes treated with MyD88–homo-dimerization inhibitory peptide migrated normally in response toSDF-1 or in response to cells exposed to hypoxia. Hence, the roleof MyD88 in monocyte/macrophage migration toward oxidizedcells appears specific to cues emanating from oxidized cells.Furthermore, in contrast to the MyD88-dependent motility ofmonocytes in response to oxidatively stressed cells, activation ofa MyD88-mediated pathway by either LPS or necrotic fibroblastsinduced a strong inflammatory response but did not result insignificant monocyte migration (data not shown). Hence, our datareveal a role for MyD88 in cell motility independent of its role inthe inflammatory response to classic pathogen-associated molec-ular patterns or danger-associated molecular patterns. Furthermore,this MyD88-mediated monocyte response is specific to signalsemanating from oxidized cells and does not participate in the re-sponse to other chemoattractants (SDF-1 and thioglycollate).Although the issue of how MyD88 signaling differentially

regulates motility or inflammation is beyond the scope of the cur-rent study, it is tempting to speculate that the differences may be re-lated to the duration and magnitude of IRAK1 phosphorylation.Significantly, within minutes of exposure to soluble factors releasedfrom oxidized fibroblasts but not mock-treated cells, monocytes

FIGURE 7. Selective attraction of CD14+ cells toward fibroblasts ex-

posed to hypoxic conditions. Fibroblasts were grown under normoxic or

hypoxic (for 2–24 h) conditions. Then PBMC were placed over a Trans-

well (5.0 mm) and allowed to migrate for 1 h and were analyzed as in Fig.

1. A, The data are presented as the percentage of migrating CD14+ cells.

No other cell type analyzed including T, NK, and NKT cells and gran-

ulocytes significantly migrated in response to hypoxic fibroblasts. Similar

results were obtained in three independent experiments. B, Cell migration

was performed as above using PBMC that were pretreated for 30 min with

either MyD88 inhibitory peptide (gray bars) or scrambled peptide (black

bars). One experiment of two independent experiments is shown.

The Journal of Immunology 1241

by guest on June 18, 2013http://w

ww

.jimm

unol.org/D

ownloaded from

demonstrated increased IRAK1 phosphorylation, indicating acti-vation of the MyD88 signaling pathway. However, in contrast toLPS and necrotic cells that induced a gradual and sustainedMyD88-dependent signaling, the response to oxidized cells wasbiphasic and transient. In contrast, LPS and necrotic cells induced astrong proinflammatory response measured by cytokine produc-tion, whereas oxidized fibroblasts failed to induce any detectableresponse. Divergent roles have also been shown for MAPK sig-naling, which regulates diverse physiological functions includingcell proliferation, differentiation, transformation, and survival. It isnow clear that the duration andmagnitude of signaling through theMAPK pathway, as well as the spatial restriction of MAPK activity,play a key role in determining the diverse physiological outcomes(49). Differences in IRAK1 phosphorylation kinetics or othersignaling differences may skew the MyD88 response toward the

Src family kinase/focal adhesion kinase axis, which has beenshown to induce cytoskeleton reorganization and cell motility(50–52). A more detailed study into the signaling pathways andkinetics will be required to resolve this issue. Whatever the ex-planation for this phenomenon, the data suggest that ligands re-leased from oxidized cells signal through MyD88-linked receptorsin a “nonconventional” way that induces cell migration but notinflammation.Two different mechanisms concerning the generation of factors

able to trigger monocyte motility have been revealed in the currentstudy. First, the local release of chemotactic factors seems to bemediated by oxidation of phospholipids and other cellular com-ponents that is initiated by reactive oxygen species and propagatedvia the classical mechanism of lipid peroxidation chain reaction(29). One such potential candidate for a danger signal has been

FIGURE 8. Peroxide-treated fibroblasts do not induce or inhibit monocyte inflammatory responses. A, Monocytes were either cultured alone or

cocultured with mock- or peroxide-treated fibroblasts. As a control, monocytes were cocultured with apoptotic fibroblasts (induced by staurosporin),

necrotic fibroblasts, or stimulated with LPS (10 ng/ml). After 24 h, conditioned media were collected and the levels of TNF-a and IL-1b secreted to the

media were determined by ELISA. The data represent the mean values of triplicate samples and SDs. Data represent one of three independent experiments.

B, Heat map of 98 genes associated with inflammation and phagocytosis showing gene expression levels in monocytes cocultured with either mock-treated

fibroblasts (right column) or peroxide-treated fibroblasts (left column). In all genes, there was no significant change that exceeded 1.5-fold in the level of

expression. C, To test whether oxidized fibroblasts inhibit monocyte response to stimuli, monocytes were either left untreated or were stimulated with either

LPS or necrotic fibroblasts in the absence (black bars) or presence of either peroxide-treated fibroblasts (white bars) or apoptotic fibroblasts (gray bars).

After 24 h, conditioned media were collected and the levels of TNF-a and IL-1b secreted to the media were determined by ELISA. The data represent the

mean values of triplicate samples and SDs. Data represent one of three independent experiments.

1242 MyD88-DEPENDENT MACROPHAGE MOTILITY

by guest on June 18, 2013http://w

ww

.jimm

unol.org/D

ownloaded from

tested, namely MDA, a major product of lipid peroxidation. Spe-cifically, our data demonstrated that addition of a stable analog ofMDA by itself recapitulated, although to a lesser extent, some ofthe effects seen with oxidized fibroblasts; namely, it selectivelyattracted monocytes in a MyD88-dependent manner and inducedIRAK1 phosphorylation. Second, chemotactic factors are gener-ated by the enzymatic activity of ALDH-2. Our initial hypothesiswas that activation of ALDH-2, an enzyme that acts to convert anddetoxify reactive lipid peroxidation products, such as MDA, willdecrease the level of these products and thus will reduce cellmigration. In contrast, we found that ALDH-2 enzymatic activityenhanced cell migration, suggesting that the enzymatic metabo-lites generated by ALDH-2 also contribute to monocyte motility.In a similar fashion, the enzymatic activity of caspase-3 in apo-ptosis also releases the phopspholipid lysophosphatidylcholine,that in turn acts as a putative attraction signal to induce the mi-gration of macrophages (53).Detection of stress signals, based on altered cellular components,

is advantageous because the signals are generated immediately.Thus, immune cells can respond rapidly to enzymatic and non-enzymatic changes that occur during oxidative stress, without thedelay required for gene transcription or increased protein pro-duction (1, 5). In accord with this proposal, monocytes haveevolved to directly detect a complex series of modified moleculesthat behave like a “beacon” of environmental stress, all of whichcan be released immediately from stressed cells and have somecapacity for chemoattraction of monocytes.The SR family consists of structurally unrelated receptors with

potential functions in binding and clearance of apoptotic cells andendogenous molecules (54). SRs are known to bind lipids andtherefore may serve as the sensing receptors for monocytes to re-spond to newly generated lipid moieties released from oxidizedcells. Indeed, our data suggest monocytes detect and respond tosignals emanating from oxidized cells through class A SRs thatalso directly or indirectly transmit their intracellular signalthrough MyD88.It is important to note that although our data support a role for

oxidation metabolites and SR in recruitment of macrophages, aprevious report demonstrated that oxidized low-density lipoprotein(LDL), but not native LDL, inhibited macrophage migration,a process that was suggested to contribute to macrophage trappingin the arterial intima. Interestingly, this inhibition was associatedwith generation of reactive oxygen species and the enhancement offocal adhesion kinase phosphorylation and actin polymerizationthat was dependent on signaling through a member of the SR Bfamily, CD36 (55). The extent and type of oxidative alteration tonative molecules may determine the outcome of monocyte re-sponse (i.e., whether monocytes migrate toward the insult) or re-main in close proximity to the damaged cells.The demonstration that SR-As play a role in monocyte attraction,

and the ability of SR-A inhibitors to inhibit MyD88-depen-dent signaling does not exclude the involvement of TLRs in thissystem. Indeed, recent studies have shown that particular TLRligands also depend on SRs. For example, CD36, a class B SR, hasbeen shown to signal via a TLR2/6 herterodimer upon binding ofmicrobial diacylglycerides (56). Furthermore, altered endogenousCD36 ligands, oxidized LDL and amyloid-b trigger sterile in-flammation through TLR4/6 heterodimerization and MyD88-dependent signaling (35). In addition, MARCO, a class A SR,triggers macrophage cytokine responses via TLR2 and CD14 inresponse to a mycobacterial cell wall component, trehalosedimycolate. Because CD14 was shown to bind not only LPS butalso phospholipids (57), a function shared with SRs (14), it wassuggested to be involved in extracting and presenting MARCO-

bound ligand to TLR2 (34). SR, therefore, may represent a prin-ciple monocyte sensor of exogenous and endogenous signals. Inour system, these receptors seem to specifically recognize oxida-tively modified lipids and other yet unknown ligands and functionas coreceptors (with or without CD14 used as a tethering receptor)that trigger monocyte migration via a MyD88 signaling pathway.The lack of an inflammatory response or an inhibitory effect on

monocyte activation observed upon coculturing monocytes withoxidized fibroblasts confirms the sublethal nature of the treatmentused in this study (given that necrotic cells are strong monocyteactivators, and apoptotic cells are strong inhibitors of monocyteactivation) and also raises the question of what is the purpose ofmonocyte migration in this setting. On the basis of the preliminarydata (A. Geiger-Maor and J. Rachmilewitz, unpublished obser-vations), we suggest that monocytes resolve the oxidative cellulardamage, thereby limiting more extensive tissue injury.In conclusion, we believe that the experimental approach used in

this study of testing various “intermediate” conditions within thefull continuum of cellular conditions rather than the extremeconditions (i.e., cell death) will allow us to explore the potentialenormous diversity of immune responses, far beyond the canoni-cal inflammatory response.

AcknowledgmentsWe thank Drs. D. Mochly-Rosen and C.H. Chen (Department of Chemical

and Systems Biology, Stanford University, Stanford, CA) for providing

Alda-1.

DisclosuresThe authors have no financial conflicts of interest.

References1. Matzinger, P. 1994. Tolerance, danger, and the extended family. Annu. Rev.

Immunol. 12: 991–1045.2. van der Most, R. G., A. J. Currie, B. W. Robinson, and R. A. Lake. 2008.

Decoding dangerous death: how cytotoxic chemotherapy invokes inflammation,

immunity or nothing at all. Cell Death Differ. 15: 13–20.3. Kono, H., and K. L. Rock. 2008. How dying cells alert the immune system to

danger. Nat. Rev. Immunol. 8: 279–289.4. Rock, K. L., and H. Kono. 2008. The inflammatory response to cell death. Annu.

Rev. Pathol. 3: 99–126.5. Gallucci, S., and P. Matzinger. 2001. Danger signals: SOS to the immune system.

Curr. Opin. Immunol. 13: 114–119.6. Skoberne, M., A. S. Beignon, and N. Bhardwaj. 2004. Danger signals: a time and

space continuum. Trends Mol. Med. 10: 251–257.7. Bianchi, M. E. 2007. DAMPs, PAMPs and alarmins: all we need to know about

danger. J. Leukoc. Biol. 81: 1–5.8. Grimsley, C., and K. S. Ravichandran. 2003. Cues for apoptotic cell engulfment:

eat-me, don’t eat-me and come-get-me signals. Trends Cell Biol. 13: 648–656.9. Valko, M., M. Izakovic, M. Mazur, C. J. Rhodes, and J. Telser. 2004. Role of

oxygen radicals in DNA damage and cancer incidence. Mol. Cell. Biochem. 266:

37–56.10. Dean, R. T., S. Fu, R. Stocker, and M. J. Davies. 1997. Biochemistry and pa-

thology of radical-mediated protein oxidation. Biochem. J. 324: 1–18.11. Cooke, M. S., M. D. Evans, M. Dizdaroglu, and J. Lunec. 2003. Oxidative DNA

damage: mechanisms, mutation, and disease. FASEB J. 17: 1195–1214.12. Sambrano, G. R., and D. Steinberg. 1995. Recognition of oxidatively damaged

and apoptotic cells by an oxidized low density lipoprotein receptor on mouse

peritoneal macrophages: role of membrane phosphatidylserine. Proc. Natl. Acad.

Sci. USA 92: 1396–1400.13. Hazen, S. L., and G. M. Chisolm. 2002. Oxidized phosphatidylcholines: pattern

recognition ligands for multiple pathways of the innate immune response. Proc.

Natl. Acad. Sci. USA 99: 12515–12517.14. Hazen, S. L. 2008. Oxidized phospholipids as endogenous pattern recognition

ligands in innate immunity. J. Biol. Chem. 283: 15527–15531.15. Loiarro, M., C. Sette, G. Gallo, A. Ciacci, N. Fanto, D. Mastroianni,

P. Carminati, and V. Ruggiero. 2005. Peptide-mediated interference of TIR do-

main dimerization in MyD88 inhibits interleukin-1-dependent activation of NF-

kB. J. Biol. Chem. 280: 15809–15814.16. Chen, C. H., G. R. Budas, E. N. Churchill, M. H. Disatnik, T. D. Hurley, and

D. Mochly-Rosen. 2008. Activation of aldehyde dehydrogenase-2 reduces is-

chemic damage to the heart. Science 321: 1493–1495.

The Journal of Immunology 1243

by guest on June 18, 2013http://w

ww

.jimm

unol.org/D

ownloaded from

17. Perez-Miller, S., H. Younus, R. Vanam, C. H. Chen, D. Mochly-Rosen, andT. D. Hurley. 2010. Alda-1 is an agonist and chemical chaperone for the commonhuman aldehyde dehydrogenase 2 variant. Nat. Struct. Mol. Biol. 17: 159–164.

18. DeWitte-Orr, S. J., S. E. Collins, C. M. Bauer, D. M. Bowdish, andK. L. Mossman. 2010. An accessory to the “Trinity”: SR-As are essentialpathogen sensors of extracellular dsRNA, mediating entry and leading to sub-sequent type I IFN responses. PLoS Pathog. 6: e1000829.

19. Malhotra, J. D., H. Miao, K. Zhang, A. Wolfson, S. Pennathur, S. W. Pipe, andR. J. Kaufman. 2008. Antioxidants reduce endoplasmic reticulum stress andimprove protein secretion. Proc. Natl. Acad. Sci. USA 105: 18525–18530.

20. Niethammer, P., C. Grabher, A. T. Look, and T. J. Mitchison. 2009. A tissue-scale gradient of hydrogen peroxide mediates rapid wound detection in zebrafish.Nature 459: 996–999.

21. Passlick, B., D. Flieger, and H. W. Ziegler-Heitbrock. 1989. Identification andcharacterization of a novel monocyte subpopulation in human peripheral blood.Blood 74: 2527–2534.

22. Wright, S. D., R. A. Ramos, P. S. Tobias, R. J. Ulevitch, and J. C. Mathison.1990. CD14, a receptor for complexes of lipopolysaccharide (LPS) and LPSbinding protein. Science 249: 1431–1433.

23. Pugin, J., C. C. Schurer-Maly, D. Leturcq, A. Moriarty, R. J. Ulevitch, andP. S. Tobias. 1993. Lipopolysaccharide activation of human endothelial andepithelial cells is mediated by lipopolysaccharide-binding protein and solubleCD14. Proc. Natl. Acad. Sci. USA 90: 2744–2748.

24. Pugin, J., R. J. Ulevitch, and P. S. Tobias. 1993. A critical role for monocytes andCD14 in endotoxin-induced endothelial cell activation. J. Exp. Med. 178: 2193–2200.

25. Kitchens, R. L., and P. A. Thompson. 2005. Modulatory effects of sCD14 andLBP on LPS-host cell interactions. J. Endotoxin Res. 11: 225–229.

26. Re, F., and J. L. Strominger. 2003. Separate functional domains of human MD-2mediate Toll-like receptor 4-binding and lipopolysaccharide responsiveness. J.Immunol. 171: 5272–5276.

27. Kennedy, M. N., G. E. Mullen, C. A. Leifer, C. Lee, A. Mazzoni, K. N. Dileepan,and D. M. Segal. 2004. A complex of soluble MD-2 and lipopolysaccharideserves as an activating ligand for Toll-like receptor 4. J. Biol. Chem. 279: 34698–34704.

28. Suzuki, N., S. Suzuki, and W. C. Yeh. 2002. IRAK-4 as the central TIR signalingmediator in innate immunity. Trends Immunol. 23: 503–506.

29. Kohen, R., and A. Nyska. 2002. Oxidation of biological systems: oxidative stressphenomena, antioxidants, redox reactions, and methods for their quantification.Toxicol. Pathol. 30: 620–650.

30. Brightbill, H. D., D. H. Libraty, S. R. Krutzik, R. B. Yang, J. T. Belisle,J. R. Bleharski, M. Maitland, M. V. Norgard, S. E. Plevy, S. T. Smale, et al. 1999.Host defense mechanisms triggered by microbial lipoproteins through Toll-likereceptors. Science 285: 732–736.

31. Manukyan, M., K. Triantafilou, M. Triantafilou, A. Mackie, N. Nilsen,T. Espevik, K. H. Wiesmuller, A. J. Ulmer, and H. Heine. 2005. Binding oflipopeptide to CD14 induces physical proximity of CD14, TLR2 and TLR1. Eur.J. Immunol. 35: 911–921.

32. Nakata, T., M. Yasuda, M. Fujita, H. Kataoka, K. Kiura, H. Sano, and K. Shibata.2006. CD14 directly binds to triacylated lipopeptides and facilitates recognitionof the lipopeptides by the receptor complex of Toll-like receptors 2 and 1 withoutbinding to the complex. Cell. Microbiol. 8: 1899–1909.

33. Dahl, M., A. K. Bauer, M. Arredouani, R. Soininen, K. Tryggvason,S. R. Kleeberger, and L. Kobzik. 2007. Protection against inhaled oxidantsthrough scavenging of oxidized lipids by macrophage receptors MARCO andSR-AI/II. J. Clin. Invest. 117: 757–764.

34. Bowdish, D. M., K. Sakamoto, M. J. Kim, M. Kroos, S. Mukhopadhyay,C. A. Leifer, K. Tryggvason, S. Gordon, and D. G. Russell. 2009. MARCO,TLR2, and CD14 are required for macrophage cytokine responses to myco-bacterial trehalose dimycolate and Mycobacterium tuberculosis. PLoS Pathog. 5:e1000474.

35. Stewart, C. R., L. M. Stuart, K. Wilkinson, J. M. van Gils, J. Deng, A. Halle,K. J. Rayner, L. Boyer, R. Zhong, W. A. Frazier, et al. 2010. CD36 ligandspromote sterile inflammation through assembly of a Toll-like receptor 4 and 6heterodimer. Nat. Immunol. 11: 155–161.

36. Acton, S. L., P. E. Scherer, H. F. Lodish, and M. Krieger. 1994. Expressioncloning of SR-BI, a CD36-related class B scavenger receptor. J. Biol. Chem. 269:21003–21009.

37. Leitinger, N. 2003. Oxidized phospholipids as modulators of inflammation inatherosclerosis. Curr. Opin. Lipidol. 14: 421–430.

38. Feige, E., I. Mendel, J. George, N. Yacov, and D. Harats. 2010. Modifiedphospholipids as anti-inflammatory compounds. Curr. Opin. Lipidol. 21: 525–529.

39. Kronke, G., V. N. Bochkov, J. Huber, F. Gruber, S. Bluml, A. Furnkranz,A. Kadl, B. R. Binder, and N. Leitinger. 2003. Oxidized phospholipids induceexpression of human heme oxygenase-1 involving activation of cAMP-responsive element-binding protein. J. Biol. Chem. 278: 51006–51014.

40. Bochkov, V. N., A. Kadl, J. Huber, F. Gruber, B. R. Binder, and N. Leitinger.2002. Protective role of phospholipid oxidation products in endotoxin-inducedtissue damage. Nature 419: 77–81.

41. Walton, K. A., A. L. Cole, M. Yeh, G. Subbanagounder, S. R. Krutzik,R. L. Modlin, R. M. Lucas, J. Nakai, E. J. Smart, D. K. Vora, et al. 2003. Specificphospholipid oxidation products inhibit ligand activation of Toll-like receptors 4and 2. Arterioscler. Thromb. Vasc. Biol. 23: 1197–1203.

42. Ma, Z., J. Li, L. Yang, Y. Mu, W. Xie, B. Pitt, and S. Li. 2004. Inhibition of LPS-and CpG DNA-induced TNF-a response by oxidized phospholipids. Am. J.Physiol. Lung Cell. Mol. Physiol. 286: L808–L816.

43. Bluml, S., S. Kirchberger, V. N. Bochkov, G. Kronke, K. Stuhlmeier, O. Majdic,G. J. Zlabinger, W. Knapp, B. R. Binder, J. Stockl, and N. Leitinger. 2005.Oxidized phospholipids negatively regulate dendritic cell maturation induced byTLRs and CD40. J. Immunol. 175: 501–508.

44. Erridge, C., S. Kennedy, C. M. Spickett, and D. J. Webb. 2008. Oxidizedphospholipid inhibition of Toll-like receptor (TLR) signaling is restricted toTLR2 and TLR4: roles for CD14, LPS-binding protein, and MD2 as targets forspecificity of inhibition. J. Biol. Chem. 283: 24748–24759.

45. Truman, L. A., C. A. Ogden, S. E. Howie, and C. D. Gregory. 2004. Macrophagechemotaxis to apoptotic Burkitt’s lymphoma cells in vitro: role of CD14 andCD36. Immunobiology 209: 21–30.

46. Devitt, A., O. D. Moffatt, C. Raykundalia, J. D. Capra, D. L. Simmons, andC. D. Gregory. 1998. Human CD14 mediates recognition and phagocytosis ofapoptotic cells. Nature 392: 505–509.

47. Cote, M., J. Drouin-Ouellet, F. Cicchetti, and D. Soulet. 2011. The critical role ofthe MyD88-dependent pathway in non-CNS MPTP-mediated toxicity. BrainBehav. Immun. 25: 1143–1152.

48. Rieker, C., D. Engblom, G. Kreiner, A. Domanskyi, A. Schober, S. Stotz,M. Neumann, X. Yuan, I. Grummt, G. Schutz, and R. Parlato. 2011. Nucleolardisruption in dopaminergic neurons leads to oxidative damage and parkinsonismthrough repression of mammalian target of rapamycin signaling. J. Neurosci. 31:453–460.

49. Karlsson, M., M. Mandl, and S. M. Keyse. 2006. Spatio-temporal regulation ofmitogen-activated protein kinase (MAPK) signalling by protein phosphatases.Biochem. Soc. Trans. 34(Pt. 5): 842–845.

50. Sanjuan, M. A., N. Rao, K. T. Lai, Y. Gu, S. Sun, A. Fuchs, W. P. Fung-Leung,M. Colonna, and L. Karlsson. 2006. CpG-induced tyrosine phosphorylationoccurs via a TLR9-independent mechanism and is required for cytokine secre-tion. J. Cell Biol. 172: 1057–1068.

51. Maa, M. C., M. Y. Chang, Y. J. Chen, C. H. Lin, C. J. Yu, Y. L. Yang, J. Li,P. R. Chen, C. H. Tang, H. Y. Lei, and T. H. Leu. 2008. Requirement of induciblenitric-oxide synthase in lipopolysaccharide-mediated Src induction and macro-phage migration. J. Biol. Chem. 283: 31408–31416.

52. Maa, M. C., M. Y. Chang, J. Li, Y. Y. Li, M. Y. Hsieh, C. J. Yang, Y. J. Chen,Y. Li, H. C. Chen, W. E. Cheng, et al. 2011. The iNOS/Src/FAK axis is critical inToll-like receptor-mediated cell motility in macrophages. Biochim. Biophys.Acta 1813: 136–147.

53. Lauber, K., E. Bohn, S. M. Krober, Y. J. Xiao, S. G. Blumenthal,R. K. Lindemann, P. Marini, C. Wiedig, A. Zobywalski, S. Baksh, et al. 2003.Apoptotic cells induce migration of phagocytes via caspase-3–mediated releaseof a lipid attraction signal. Cell 113: 717–730.

54. Bowdish, D. M., and S. Gordon. 2009. Conserved domains of the class Ascavenger receptors: evolution and function. Immunol. Rev. 227: 19–31.

55. Park, Y. M., M. Febbraio, and R. L. Silverstein. 2009. CD36 modulates migra-tion of mouse and human macrophages in response to oxidized LDL and maycontribute to macrophage trapping in the arterial intima. J. Clin. Invest. 119:136–145.

56. Hoebe, K., P. Georgel, S. Rutschmann, X. Du, S. Mudd, K. Crozat, S. Sovath,L. Shamel, T. Hartung, U. Zahringer, and B. Beutler. 2005. CD36 is a sensor ofdiacylglycerides. Nature 433: 523–527.

57. Yu, B., E. Hailman, and S. D. Wright. 1997. Lipopolysaccharide binding proteinand soluble CD14 catalyze exchange of phospholipids. J. Clin. Invest. 99: 315–324.

1244 MyD88-DEPENDENT MACROPHAGE MOTILITY

by guest on June 18, 2013http://w

ww

.jimm

unol.org/D

ownloaded from