Targeted Deletion of CCR2 Impairs Deep Vein ThombosisResolution in a Mouse Model1,2

Peter K. Henke,3 Charles G. Pearce, Daria M. Moaveni, Andrea J. Moore, Erin M. Lynch,Christopher Longo, Manu Varma, Nicholas A. Dewyer, K. Barry Deatrick,Gilbert R. Upchurch, Jr., Thomas W. Wakefield, Cory Hogaboam, and Steven L. Kunkel

CCR2 is required for monocyte recruitment in many inflammatory processes, as well as conferring Th1 lymphokine responses.Deep vein thrombosis (DVT) resolution represents a specific inflammatory response whereby the thrombus must be dissolved forrestoration of blood flow. Using a stasis model of DVT in the mouse, we investigated the role of CCR2 on DVT resolution. Geneticdeletion of CCR2 (CCR2�/�) was associated with larger thrombi at early and later time points, increased thrombus collagen, fewerthrombus monocytes (F4/80), and significantly impaired neovascularization. IL-2 and IFN-� were significantly reduced in earlyCCR2�/� thrombi, whereas MCP-1 was significantly increased, and Th2 lymphokines were unaffected. Supplementation ofCCR2�/� mice with IFN-� normalized early thrombus resolution without increasing monocyte influx. Neither Ab depletion ofIFN-� nor genetic deletion of IFN-� impaired early DVT resolution. Early fibrinolysis was not impaired in CCR2�/� mice, buta significant reduction in both matrix metalloproteinase (MMP)-2 and MMP-9 activity was observed. However, only MMP-9activity was restored with administration of IFN-�. We conclude that an early CCR2-dependent Th1 lymphokine responsepredominates in normal DVT resolution, mediates this in part by MMP-9 activation, but is not solely dependent on IFN-�. TheJournal of Immunology, 2006, 177: 3388–3397.

D eep vein thrombosis (DVT)4 is a major clinical problem,the incidence of which has not decreased over the last 20years (1). Indeed, the U.S. Surgeon General has named

DVT a national health problem in 2005. Anticoagulant therapy isefficacious when used prophylactically, but it carries a significantrisk of major bleeding when used for long term therapy (2). Directthrombolysis has not been established as a standard of care, givenunremarkable clinical benefits (3). The mechanism of thrombusresolution is complex and involves plasmin-mediated fibrinolysis,as well as proteinase-mediated collagen breakdown (4–6).

Prior studies have suggested an important role for chemokinesin DVT resolution. Chemokines are ubiquitous peptide inflamma-tory mediators that act as leukocyte chemotactic agents as well ascell activators. Although four main chemokine families exist, theones most relevant to DVT resolution are the cysteine-cysteine(CC) and cysteine-X-cysteine (CXC) chemokine types. For exam-ple, both CXC and CC chemokines are present within a resolvingthrombus (4), and exogenous administration of IL-8 (CXC che-mokine) and MCP-1 (a CC chemokine) have been shown to ac-celerate thrombus resolution in experimental models (7, 8). Many

pathophysiological processes involve CCR2 signaling, includingperitoneal and pulmonary inflammation, and atherogenesis (9–11).It is likely that chemokine-directed leukocytes primarily mediatethe DVT resolution. Monocytes are particularly important, andMCP-1 drives monocyte chemotaxis via CCR2 signaling (12–14).Similarly, monocytes expressing CXCR2 were also shown to beessential for early thrombus resolution, possibly by promoting fi-brinolysis (5).

The CCR2 receptor also directs lymphokine responses via theTh1-Th2 axis. A typical Th1 lymphokine is IFN-�, whereas Th2lymphokines are exemplified by IL-4 and IL-13 (15, 16). Thesemediators are produced by activated monocytes as well as certainlymphocytes. Numerous prior studies have confirmed that CCR2gene deletion (CCR2�/�) confers a loss of Th1 lymphokine re-sponse and thus can act as a functional model for such (10, 17).

Restoration of blood flow in a thrombosed vein also involvesneovascularization with functional flow channels, thrombus retrac-tion, and fibrinolysis (7, 18, 19). Histological studies have con-firmed the appearance of intrathrombus clefts and neovascularchannels that evolve mostly within the first 4 wk (7, 18, 20). Pleo-tropic angiogenic growth factors released by polymorphonuclearneutrophils (PMNs) and monocytes are known to be present in theresolving thrombus and include vascular endothelial growth factor(VEGF) and basic fibroblast growth factor (bFGF) (21–23). Thepathobiology of angiogenesis also involves CCR2 signaling (13,24, 25). Fibrinolysis of venous thrombi is thought to be primarilymediated by urokinase plasminogen activator (uPA) from mono-cytic cells (26), as evidenced from impaired venous thrombus res-olution in mice lacking uPA but not tissue-type plasminogen ac-tivator (27). Linked to the plasminogen activators are the matrixmetalloproteinases (MMP), particularly MMP-2 and MMP-9, andwhich modulate collagen and matrix turnover in vessel wall re-modeling and early wound healing (6, 28, 29).

The aim of this study was to characterize CCR2� leukocytekinetics over time within the thrombus, delineate the role of the

University of Michigan Health System, Ann Arbor, MI 48109

Received for publication February 28, 2006. Accepted for publication June 21, 2006.

The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby marked advertisement in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.1 This work was supported by National Institutes of Health Grant 675317 (to P.K.H.).2 Presented in part at the 66th Annual Meeting of the Society of University Surgeons,February 11, 2005, Nashville, TN.3 Address correspondence and reprint requests to Dr. Peter K. Henke, University ofMichigan Health System, 1500 East Medical Center Drive, 2210 Taubman HealthCare Center, Ann Arbor, MI 48109-0329. E-mail address: henke@umich.edu4 Abbreviations used in this paper: DVT, deep vein thrombosis; PMN, polymorpho-nuclear neutrophil; VEGF, vascular endothelial growth factor; bFGF, basic fibroblastgrowth factor; uPA, urokinase plasminogen activator; MMP, matrix metalloprotein-ase; WT, wild type; IVC, inferior vena cava; vWF, von Willebrand factor; hpf, highpower field; PAI-1, plasminogen activator inhibitor-1.

The Journal of Immunology

Copyright © 2006 by The American Association of Immunologists, Inc. 0022-1767/06/$02.00

early lymphokine response, and correlate this with thrombus res-olution. We document that early DVT resolution is associated witha Th1 lymphokine response and that deficiency of CCR2 signalingis associated with impaired DVT resolution, in part because ofreduced MMP-9 activity, and not fibrinolysis, suggesting a recip-rocal relation between the plasminogen system and the MMPs.Furthermore, although exogenous administration of the Th1 lym-phokine IFN-� reverses the early CCR2�/�-impaired DVT reso-lution, it is not necessary for this process.

Materials and MethodsMouse model

The rodent model of stasis-induced DVT has been well described (4, 5, 7,30) and used B6/129 (wild-type (WT) controls, F2 hybrids; The JacksonLaboraory) and B6/129 CCR2�/� mice (10) weighing 20–30 g, between 4and 6 wk old. The mice underwent general inhalational anesthesia and, viaa laparotomy, the inferior vena cava (IVC) was ligated with a 6-0 polypro-pylene suture. This consistently produces a stasis thrombus for tissue anal-ysis. Mice were sacrificed at 2, 4, 8, 12, and 21 days. At harvest, thethrombosed IVC was weighed and measured, snap frozen, and stored at�70°C or placed in formalin, followed by ethanol, for histological analy-sis. Baseline comparisons of serum clotting factors with measurement ofactivated partial thromboplastin time and total clotting time of WT andCCR2�/� mice was done, as was peripheral differential leukocyte counts,using an automated system as described (5).

In a separate set of experiments, a group of WT mice underwent ligationand treated with i.p. saline, control nonspecific IgG, or 1 ml of 10 �g/mlanti-murine IFN-� antisera (31). Another group, CCR2�/� mice, weregiven exogenous IFN-� (R&D Systems; 50,000 IU i.p.) (32). All agentswere administered at 24 h pre- and postligation. These groups were allharvested at 4 days. Finally, WT C57/BL6 and the corresponding IFN-��/� mice (all from The Jackson Laboratory) underwent the same proce-dure of IVC ligation and harvest at 4 days. Genetic deletion of IFN-� wasconfirmed by RT-PCR genotyping using tail snip sample analysis per pro-tocol (data not shown; www.med.umich.edu/tamc/pcr.html). The Univer-sity of Michigan Committee on Use and Care of Animals approved thisresearch protocol.

Laser doppler of IVC

The Lisca laser doppler (Lisca) was used to assess in vivo microvascularIVC blood flow, based on techniques described (33). After laparotomy, thepre-IVC ligation, immediate postligation, and harvest blood flow throughthe exposed IVC region of interest was assessed; a 5-s scan with 30-sinterval, over four scanning cycles, was performed. Depth was constant andwas adjusted for each mouse to ensure the best estimation of the midcoro-nal IVC section. These scans were saved, and accompanying image soft-ware was used to estimate the mean color flow by using a standardized areaof analysis. The intensities were reported as percent of baseline bloodflows, specific to each animal to ensure consistency.

Histological analysis/immunohistochemical staining/neovascularchannel quantification/Trichrome staining

Immunohistochemical staining was performed on the paraffin-embeddedtissue sections (10 �m) as described (7, 18, 19, 34). Anti-PMN (1/1000;Accurate Chemical and Scientific), anti-F4/80 (1/100; Serotec), anti-CCR2(1/100; Santa Cruz Biotechnology), anti-uPA (1/250; Santa Cruz Biotech-nology), and anti-von Willebrand factor (vWF; 1/100; Serotec) were used.Secondary Ab and color development was done using a Vecta ABC kitmanufacturer’s instructions. In a blinded manner, positively stained cells orneovascular channels in five high power fields (hpf; �1000) radially withinthe thrombus were counted and totaled. Neovessels were counted only if anucleus and cellular morphology were present (5). Other sections wereprocessed per standard protocol for Trichrome staining (5, 7, 19).

Chemokine/cytokine/vascular growth factor ELISA

After thrombus-vein wall separation, the thrombus was placed in completelysis buffer at 0°C (Boehringer Mannheim), homogenized, sonicated for10 s, and centrifuged at 10,000 � g for 5 min, after which the supernatantwas collected. Quantification of peptide mediators was normalized to totalprotein in the sample, by a modified Bradford assay per manufacturer’sinstructions (Pierce) with serial dilutions of BSA (Sigma-Aldrich) as stan-dards. Tissue homogenate ELISAs for mouse MIP-1�, RANTES, MCP-1,TGF�, IFN-�, IL-2, IL-4, IL-5, and IL-13 were performed with species-

specific primary Abs quantified using a double ligand technique, as hasbeen described for similar chemokines (4, 5, 35). Mouse bFGF and VEGFwere analyzed by using a commercial ELISA according to the manufac-turer’s instructions (R&D Systems).

Collagen assay

Thrombus collagen content was estimated by a commercially available kitaccording to the manufacturer’s instructions (Bicolor), as described (19).Collagen amount was then corrected to thrombus weight for each sample(micrograms per milliliter per milligram of thrombus).

Fibrin staining protocol

Fibrin staining with anti-human fibrinogen Ab (1/100; Dako A/S), whichcross-reacts with rat and mouse fibrin, was used for quantitative fibrincontent analysis (5, 36). Computer imaging analysis was done to determinerelative staining intensities as described.

Real time PCR

RNA was isolated from the thrombus homogenate using the Trizol reagentas described (6, 37). Expression of uPA, CCR2, MCP-1, and �-actinmRNA was determined from primer sequences derived using premier soft-ware (Premier Biosoft International) based on primary cDNA sequencesfrom GenBank (http://www.ncbi.nlm.nih.gov-/GenBank/): �-actin: sense,GAGGGAAATCGTGCGTGACAT; antisense, AGAAGGAAGGC-TGGAAAAGAG; CCR2: sense, GGAGTGGGAAGAAGTATGT; antisense,TCAACCTTGGCAAGATAA; MCP-1: sense, TCCCTGTCATGCTTCTGG; antisense, CTGCTGGTGATCCTCTTGTA; uPA: sense, TATGCAGCCCCAC-TACTATGGCTC; antisense, GAAGTGTGAGACCCTCGTGTAGAC. For quantification of mRNA levels, expression of the targetgene in ratio to �-actin expression was calculated by the formula: targetgene expression/�-actin expression � 2�(�Ct) (38).

Gelatin SDS-PAGE substrate zymography

As described (39), precast 10% SDS-polyacrylamide gels containing 1mg/ml gelatin were used to determine MMP-2 and MMP-9 activity (unlessotherwise stated, all zymography supplies were from Novex) of the throm-bus and vein wall homogenate. Separate densitometry analysis was per-formed using a FOTO/Analyst CCD CAMERA (Fotodyne) and GEL-ProAnalyzer software version 3.1 (Media Cybernetics). Total activity (pro-and active) MMP activity was summed and normalized to thrombusweights (milligrams).

Western immunoblots

Protein was extracted from vein wall thrombus samples by overnight in-cubation in 1% SDS (6, 37). Western blotting for uPA and plasminogenactivator inhibitor-1 (PAI-1) was performed as previously described (6).Primary Abs (uPA, Santa Cruz Biotechnology; PAI-1, BD Pharmingen)were diluted in Tris-buffered saline with Tween 20 to a dilution of 1/5000,and after secondary Ab application, the labeled proteins were detected viathe application of ECL chemiluminescence agents (Detection reagents andfilm; Amersham Biosciences). Images were detected and densitometry per-formed as described for zymography.

Statistical analysis

Data were reported as mean � SEM. All data were subjected to statisticalanalysis by t test to compare experimental groups with appropriate con-trols. Analysis was performed with Sigma Stat version 2.0 (SPSS), andgraphic representations were created using Sigma Plot. p � 0.05 was as-signed as significant.

ResultsDVT resolution is impaired in CCR2�/� mice

Venous thrombi were significantly larger at 4, 12, and 21 days inthe CCR2�/� mice, as compared with control (n � 13–15; Fig.1A). At these time points, thrombus size was �25–40% larger inCCR2�/� than in controls. Although WT DVT sizes decreasedlinearly over 21 days, this was delayed in the CCR2�/� mice.Serum clotting parameters of activated partial thromboplastin timeand total clotting time were similar between the WT and CCR2�/�

mice (data not shown), suggesting no intrinsic alteration in clottingfactor production or activation.

3389The Journal of Immunology

When evaluating the thrombus proper (for which consistent ex-traction is possible before 8 days), total collagen levels were sig-nificantly elevated by 30–60% in CCR2�/� thrombi at all time

points, as compared with WT controls (Fig. 1B, n � 5, p � 0.001).Trichrome staining of thrombosed IVC sections qualitatively con-firmed that increased thrombus collagen was present at these early

FIGURE 1. A, Thrombus resolu-tion is reflected by the thrombosedIVC weight divided by length (mg/cm). The thrombus size was signifi-cantly greater in CCR2�/� mice ascompared with controls at 4, 12 and21 days. �, p � 0.05; n � 13–15. B,Total thrombus collagen was as-sessed by a colorimetric assay. Atdays 2–8, collagen was significantlygreater in the CCR2�/� thrombi thanin controls. �, p � 0.05, n � 5. C,Representative histological section ofthe thrombosed IVC (�400). Notethe central portion of the thrombus isblue-white consistent with the pres-ence of collagen. Less of this stainingappearance is noted in the WT con-trols. C, clot. D, Laser doppler (LD)imaging of the pre- and harvest IVCcomparisons of blood flow showsless IVC blood flow at later timepoints, 12 and 21 days, in theCCR2�/� mice as compared withcontrols. �, p � 0.05; n � 5–13. E,Thrombus neovascular channels werequantified by positive vWF staining.No significant staining is present be-fore 8 days. Note significantly fewerchannels in 12 and 21 days ofCCR2�/� thrombi as compared withcontrols. �, p � 0.05; n � 4–5. F,Immunohistological representativeexample of a 12-day thrombosed IVCsection (�400) stained for vWF.Note significantly more positivechannels in the control thrombi thanin CCR2�/� thrombi. Arrows, Chan-nels; W, wall.

3390 CCR2 AND DVT RESOLUTION

time points in CCR2�/� mice (Fig. 1C). Immunostaining of sep-arate thrombi sections showed the collagen subtype to be primarilytype III, with essentially no type I collagen staining (data notshown).

Later venous thrombus resolution is associated with neovascu-larization (5, 20, 23). Laser doppler imaging detects blood flowwithin and around the thrombosed IVC (33). In WT controls,blood flow returns to nearly 70% of normal nonthrombosed IVCflow by 21 days. However, 20–30% less blood flow was observedat later time points (days 12 and 21) in CCR2�/� thrombosed IVCas compared with WTs (Fig. 1D) (n � 5–9, p � 0.05). Consistentwith these observations were �3-fold fewer vWF-positive chan-nels at days 12 and 21 in CCR2�/� mice thrombi than in WTcontrols (Fig. 1, E and F; n � 4–6, p � 0.05). This was particularlystriking at day 21 between the groups. Because thrombus neovas-cularization was less at days 12 and 21 in the CCR2�/� thrombithan in WT controls, the proangiogenic factor VEGF was assessedby ELISA. VEGF levels were low (days 2–8; range, 15–75 pg/mgthrombus) within the thrombus and were not significantly differentbetween the CCR2�/� and WT mice at any time point.

CCR2�/� mice have fewer thrombus monocytes anduPA-positive cells

CCR2 signaling directs monocyte chemotaxis and activity to areasof inflammation (9, 11, 13). In WT thrombus, F4/80� cell influxpeaked at 8 days. As compared with WT, F4/80�-staining mono-cytes were significantly reduced at all time points in CCR2�/�

thrombi except at day 21 (Fig. 2, A and B; n � 3–4, p � 0.05). InWT only, CCR2�-staining cells increased linearly from days 2 to12, but absolute cell numbers at all times were less than the F4/80monocytes (Fig. 2, C and D). Phenotypically, CCR2� cells were

mononuclear type cells. As a corollary measure, thrombus CCR2gene expression rose from day 2 and peaked at day 12 (2 days,7.3 � .3 � 10�5; 12 days, 5.0 � 3 � 10�4 gene/�-actin, n � 4).No CCR2 gene expression was detectable in CCR2�/� mice.

No differences were observed in PMN number within the throm-bus between the WT and CCR2�/� mice, with highest numbers at2 days (26–34 cells/5 hpf) and lowest at 8 days (4–5 cells/5 hpf).Peripheral leukocyte counts were not significantly different com-paring WT controls with CCR2�/� (8.2 � 1 vs 6.5 � 0.6 � 103

cells/ml; n � 11–17, p � 0.20). The number of circulating mono-cytes were less in the CCR2�/� mice (0.31 � 0.04 vs 0.51 �0.08 � 103 cells/ml; n � 11–17, p � 0.02).

Prior experiments have shown a correlation between uPA�-staining cells and thrombus resolution (5, 27). No difference inuPA�-staining cells was present before day 8 in WT andCCR2�/� thrombi. However, �40% fewer uPA� cells at 8 and 12days in the CCR2�/� thrombi than in WT controls was found (Fig.2, E and F; n � 5–6, p � 0.05). However, analysis of the thrombushomogenate did not show significant differences in uPA gene ex-pression at these time points (8 days, 4.3 � 2 vs 0.96 � 0.1 �10�7; n � 5, p � 0.3; 12 days, 5.2 � 3 vs 5.8 � 0.4 � 10�8, n �5, p � 0.6).

DVT in CCR2�/� mice have an altered chemokine milieu andan impaired Th1 lymphokine response

Thrombus mediators are produced both in the thrombus and by thesurrounding vein wall cells, given the intimate physical proximityof the thrombus-vein wall interface (5, 18, 33). To evaluate thethrombus milieu, only the thrombus (and not wall) was evaluatedat 2, 4, and 8 days after DVT (for which complete thrombus ex-traction is possible) for analysis (Fig. 3A). MCP-1 protein levels by

FIGURE 2. A, Quantification ofF4/80� cells within the thrombus.Significantly fewer positive cells arepresent in CCR2�/� thrombi sectionsthan in WT controls through 12 days.�, p � 0.05; n � 3–4 per time point.B, Representative IVC histology sec-tion of 8-day WT and CCR2�/� F4/80� cells (�400; arrows). Note theperithrombus distribution of cells,and significantly less positive stain-ing is noted in the CCR2�/� mice, aswell as overall less cellularity. C,Quantification of thrombus CCR2�

cell staining is shown. Note the linearincrease over time to level out after12 days. n � 3–4 per time point. D,Representative 8-day (8d) thrombusIVC section stained for CCR2� cells(�1000; arrows). The cellular mor-phology is of mononuclear cell type.Neg., Negative. E, Quantification ofuPA� cellular staining. Fewer uPA� cells are present at 8 and 12 daysin CCR2�/� thrombi as comparedwith WT controls. �, p � 0.05, n �4–5. F, Representative IVC sectionof uPA Ag immunostaining at 12 daytime point (�400), arrows. Note sig-nificantly fewer � staining cells inthe CCR2�/� thrombi and overall di-minished cellularity. C, Clot.

3391The Journal of Immunology

ELISA were significantly elevated at all time points within thethrombus (n � 5, p � 0.001) in CCR2�/� mice, peaking at 8 days.Thrombus MCP-1 gene expression was higher at 4 days inCCR2�/� mice by 7-fold (3.1 � 0.5 vs 0.43 � 0.03 � 10�2

gene-�-actin ratio; n � 4, p � 0.002). Levels of the CC chemo-kines MIP-1� and RANTES, which interact via CCR5 and notCCR2 (17), were not significantly different in the thrombus be-tween the two groups through 8 days (data not shown). The proin-flammatory mediator IL-1� was significantly reduced at days 2and 8 in the CCR2�/� thrombi (Fig. 3B). The profibrotic factorbFGF was elevated at days 2 and 4, but a reciprocal relation wasfound at 8 days in CCR2�/� thrombi as compared with WT con-trols (Fig. 3C), whereas TGF� levels were low and not signifi-cantly different between groups within the thrombus (range, 2–8ng/mg protein).

The Th1 lymphokine levels were reduced in CCR2�/� thrombias compared with WT controls. Thrombus IL-2 was significantlyreduced at 2 days in CCR2�/� as compared with WT controls andwas not detectable at other time points in either group (Fig. 3D).Thrombus IFN-� at days 2 and 4 was decreased by nearly 80%(Fig. 3E; n � 5–6, p � 0.01). In contrast, no increase in the Th2

lymphokines, IL-4, IL-5, and IL-13 were found in the CCR2�/�

thrombi as compared with WT controls, and no significant differ-ences were found between the groups (data not shown).

The Th1 lymphokine IFN-� is sufficient but not necessary forearly DVT resolution

Given the strong association between early impaired DVT resolu-tion, decreased IFN-� levels and decreased monocyte influx, the4-day time point was examined in detail. To better delineate therole of IFN-� in modulating early DVT resolution, an addition andsubtraction strategy was used. As shown previously, DVT sizeswere significantly greater in CCR2�/� mice than were WT con-trols (Fig. 4A). The saline and IgG controls were combined as onegroup for purposes of analysis, given that there were no observeddifferences in any of the DVT resolution parameters. The impairedDVT resolution in CCR2�/� mice was totally reversed by exog-enous IFN-�, with an actual nonsignificant reduction in DVT sizecompared with controls (Fig. 4A; n � 7). In contrast, Ab depletionof IFN-� of WT controls was not associated with an increase inDVT size.

FIGURE 3. Thrombus homoge-nate ELISA for peptide mediators,corrected for total protein in the sam-ple. A, MCP-1; B, IL-1�; C, bFGF;D, IL-2; E, IFN-� (gIFN). All arecorrected as picograms per milligramof protein. �, p � 0.05, n � 5 foreach time point.

3392 CCR2 AND DVT RESOLUTION

Collagen levels in CCR2�/� thrombi were also significantly re-duced by exogenous IFN-� administration to nearly WT levels,whereas Ab depletion of IFN-� was not associated with an ele-vated thrombus collagen (Fig. 4B; n � 5). Intrathrombus cellularanalysis showed that monocytes (F4/80) were not restored to WTlevels in those CCR2�/� that received IFN-�. Consistent with thefact that DVT sizes were similar, no reduction in F4/80� mono-cytes or CCR2� cells were found in those that received anti-IFN-�Ab (Fig. 4C).

From the prior experiments, thrombus MCP-1 levels showed aninverse relationship with monocyte influx and impaired DVT res-olution. Whereas MCP-1 levels were significantly elevated inCCR2�/� thrombi as compared with WT, exogenous IFN-� ad-ministration to CCR2�/� mice was associated with reducedMCP-1 levels similar to WT controls (Fig. 4D). Similarly, micethat received anti-IFN-� Ab had no elevation of thrombus MCP-1.To confirm the addition and subtraction strategies altered theIFN-� thrombus milieu, IFN-� levels were measured by ELISA.Importantly, IFN-� was elevated slightly above baseline WT levelsin those CCR2�/� mice that received IFN-� and was reduced toundetectable levels in WT mice receiving anti-IFN-� Ab (Fig. 4E).

To further confirm that IFN-� was not necessary for DVT res-olution, mice with targeted deletion of IFN-� (C57BL/6-IFN-��/�) and WT controls (C57BL/6) underwent IVC ligation to pro-duce DVT and were harvested at 4 days. Consistent with the Abdepletion experiments, no significant differences in thrombus size(Fig. 4F) or thrombus collagen (WT, 0.42 � 0.18 vs IFN-��/� �0.82 � 0.2 �g/mg thrombus, n � 3–4, p � 0.2) were noted in theIFN-��/� mice as compared with WT controls. By ELISA, nodetectable IFN-� was present, and MCP-1 levels were similar toWT levels (data not shown). Importantly, despite murine geneticbackground differences from the prior experiments, thrombus sizeat 4 days in C57BL/6 controls were not significantly different thanin B6/129 WT controls, suggesting no species-specific intrinsicdifferences in DVT resolution at this time point.

CCR2�/� mice have reduced local vein wall MMP-2 andMMP-9 activity, and MMP-9 activity is restored with IFN-�administration

The Th1 lymphokine IFN-� has multiple inflammatory activities,including MMP activation (28, 32). First, to exclude that differ-ential fibrinolysis might account for the observed difference in

FIGURE 4. A, DVT sizes were sig-nificantly greater only in CCR2�/�

mice, as compared with the othergroups. Administration of IFN-� re-stored DVT resolution, but inhibition ofIFN-� by Ab depletion did not impairDVT resolution in WT mice. n � 7 eachgroup. �, p � 0.05. B, Total collagen in4-day thrombus was greater only inCCR2�/� mice and was significantly re-duced with IFN-� (gIFN) administra-tion. �, p � 0.05, n � 5–7. C, Compar-ison of cellular thrombus content invarious conditions. Note that theCCR2�/� mice had significantly fewerF4/80� cells and were not altered withthe addition of IFN-�. Similarly, anti-IFN-� inhibition was not associated withreduced F4/80� cells or reduced CCR2�

cell influx. �, p � 0.05, n � 3–5. D,Thrombus MCP-1 levels as measured byELISA were elevated in CCR2�/�

thrombi, reduced in CCR2�/� when ex-ogenous IFN-� was administered, andnot affected by IFN-� inhibition. �, p �0.05, n � 5 each time point. E, Confir-mation of prior experiments showed re-duced levels of IFN-� in CCR2�/�

thrombi. This was restored by exoge-nous IFN-� administration and was sig-nificantly reduced in WT by anti-IFN-�Ab. n � 4–5 each time point. �, p �0.05. F, The same IVC ligation modelwas used in C57BL/6 and C57BL/6IFN-��/� mice to produce DVT. No sig-nificant difference in DVT sizes was ob-served. Note relative size of the WTC57BL/6 control thrombus weight (wt)/length was similar to the WT B6/129control thrombus weight/length shownin the other experiments.

3393The Journal of Immunology

DVT size, comparison of fibrin content by quantitative image anal-ysis was done (5). No difference in thrombus fibrin content wasfound between CCR2�/� and WT controls (Fig. 5A). This findingis also supported by the observation that the number of uPA� cellswere similar in both CCR2�/� and WT controls at the day 4 timepoint (Fig. 2E). To further assess possible differential plasminogensystem activation, Ag levels of uPA and PAI-1 were determined byWestern immunoblotting, and a ratio calculated to yield the fi-brinolytic to anti-fibrinolytic balance (6). Unexpectedly, CCR2�/�

thrombi had the greatest relative amount of uPA, with a ratio 5–8fold greater than WT, CCR2�/� � IFN-�, and WT-IFN-� murinethrombi (n � 4–5, p � 0.01; Fig. 5B).

Given that differences in fibrinolysis between WT and CCR2�/�

thrombi were unlikely the mechanism of impaired DVT resolution,investigation of the proteinases MMP-2 and MMP-9 was done.

Densitometric analysis of MMP-2 and MMP-9 total gelatinolyticactivity showed significantly reduced levels in the vein wall ho-mogenate and a trend toward less activity in the thrombus properin CCR2�/� as compared with WT controls. IFN-� administrationpartially restored MMP-2 activity but was still significantly lessthan WT controls. Anti-IFN-� was associated with a significantreduction in MMP-2 activity (Fig. 5C). Conversely, the CCR2�/�

mice that received IFN-� had a significant increase in MMP-9 (to60% of WT levels, p � 0.3), but not MMP-2 (Fig. 5D). Anti-IFN-�Ab administered to WT mice significantly reduced MMP-2 but notMMP-9 activity. Similarly, no difference in MMP-2 and MMP-9activity was found in the IFN-��/� as compared with WT controls(data not shown). These data also suggest that the primary collag-enolytic activity is from the vein wall itself, and not derived fromcells within the thrombus.

FIGURE 5. A, Comparison of 4-day fibrin content between WT and CCR2�/� showed no significant difference in positively stained area by imageanalysis, suggesting that fibrin content was not different between these groups. n � 5 for each group. B, Comparison of uPA and PAI-1 levels by Westernimmunoblotting were done to examine the relative fibrinolytic balance. Note that only CCR2�/� thrombi had a greater uPA content ratio whereas theCCR2�/� � IFN-� (gIFN), and WT-IFN-� were not different from WT alone. Representative Western immunoblot images are shown. KO, Knockout. �,p � 0.05, n � 4–5. C, IVC MMP-2 total activity was measured by zymography under conditions shown. As compared with WT controls, all groups hada significant reduction in MMP-2 activity. Note that CCR2�/� � IFN-� had a partial restoration of MMP-2 that was significantly greater than CCR�/�

� vehicle but did not reach control WT levels. Representative gel image is shown, and most zymographic activity is in the latent form. n � 4–5 mice pergroup. D, IVC MMP-9 activity as measured by zymography. Note that only CCR2�/� mice had a significant reduction in MMP-9 activity as comparedwith WT controls. Neither CCR2�/� � IFN-� nor WT � IFN-� MMP-9 activity was significantly less than WT controls. n � 4–5 mice per group.Representative gel image is shown, and most activity is in the latent form.

3394 CCR2 AND DVT RESOLUTION

DiscussionEfficient thrombus clearance is essential for allowing prograde re-turn of blood flow in the venous system after DVT and is clinicallyassociated with less risk of postphlebitic syndrome (1, 40).Whereas other studies have suggested the early DVT environmentis proinflammatory (4, 5, 8), the data herein support several mech-anisms associated with CCR2 signaling activity that directly af-fects thrombus resolution (Fig. 6). Five interrelated conclusionscan be drawn from the data: 1) a Th1 lymphokine response pre-dominates in early DVT resolution; 2) loss of CCR2 signaling isassociated with decreased thrombus neovascularization; 3) theMCP-1-CCR2 axis is important for monocyte influx into the re-solving thrombus, independent of other CC chemokines; 4) IFN-�is sufficient, but not necessary for DVT resolution; 5) vein wallMMP-9 may be more important than MMP-2 in early DVTresolution.

The data suggest that a CCR2-dependent Th1 lymphokine re-sponse occurs early in the resolving thrombus, and interferencewith this response is associated with impaired early and later DVTresolution. Specifically, a bimodal effect of the CCR2 signaling onleukocytes and surrounding vein wall cells after DVT seems tooccur; namely, early events are probably due to early monocytemediated events, such as thrombolysis and MMP activation,whereas later events may be related to other cellular healing pro-cesses, such as neovascularization and thrombus contraction. Thisnotion is supported by the fact that CCR2�/� mice, functionallydeficient in Th1 responses (10, 17), had larger DVT both early(day 4), and later, at days 12 and 21. The two primary Th1 lym-phokines, IFN-� and IL-2 (15, 41), were significantly decreased inthe CCR2�/� thrombi, which is consistent with other disease mod-els utilizing these mice (10, 14). How the loss of CCR2 signalingcauses an impaired Th1 response has not been fully delineated, butit is probably due to decreased monocyte trafficking with loss ofTh1-type lymphokine release (42). Overproduction of MCP-1, act-ing through a receptor independent of CCR2, may also inhibit theTh1 response, although this has been defined only in vitro (43). Itis unlikely that lymphocytes play much of a role in this responseas few are present in the thrombus beyond 2 days (4).

Later DVT resolution involves neovascularization (5, 20, 23).This was markedly impaired in CCR2�/� mice, with a significantreduction in thrombus neovascularization as measured by both invivo laser doppler imaging and immunohistochemical analysis ofneovascular channels as compared with WT mice. However, lessdifference in thrombus monocytes was found, including nearly

identical numbers at day 21. This suggests that other processes thatare not necessarily monocyte dependent may contribute to DVTresolution at the later time points, or that early monocyte presenceconfers long lasting effects on later DVT resolution. The proan-giogenic activity of MCP-1 is well documented but is dependenton a functional CCR2 receptor, and despite significantly elevatedthrombus MCP-1 in the CCR2�/� mice, impaired neovasculariza-tion was observed (13, 21). Others have found that CCR2�/� micehave impaired neovascular responses in various models of disease(12, 24). Although specific proangiogenic mediators were not eval-uated at days 12 and 21 because of difficulty in clean separation ofvein wall and thrombus, we speculate that reduced bFGF inCCR2�/� mice at 8 days (but not VEGF) may have contributed tothis impaired later neovascular response. Basic FGF is a well-established proangiogenic and profibrotic mediator associated withDVT neovascularization (5). MMPs, in particular MMP-9, haveproangiogenic activity by mediating basement membrane remod-eling for endotheliazation of a tissue matrix (28, 29). Indeed,MMP-9 inhibition is associated with impaired neovascularizationin an animal model of cardiac ischemia (44).

Monocytes are the most important leukocyte in modulation ofDVT resolution. Monocytes release and direct smooth muscle cellproangiogenic mediator release, uPA, as well as MMP activity (5,8, 26, 45). In support of this contention is that Ab inhibition andgenetic deletion of IFN-� did not impair DVT resolution, probablybecause these thrombi had a normal content of F4/80 and CCR2�

cells. We speculate that these monocytes may mediate DVT res-olution by promoting MMP release and activation from the adja-cent vein wall cells. Other disease models involving CCR2�/�

mice have also shown significant deficiencies in monocyte extrav-asation into inflammatory sites (9–11). Perhaps more importantphysiologically and consistent with our findings is that CCR2�

monocytes are critical for influx of other monocyte subtypes, suchas F4/80 cells (46). It was not surprising that DVT resolution wasnot impaired at 2 days, given that PMNs predominate in the throm-bus and are critical for very early thrombus resolution (19). It wasnot as clear why 8-day CCR2�/� and WT mice had similar throm-bus sizes, but other mechanisms may be more important at this timepoint and/or have attained a threshold of thrombus monocytes presentto allow normal resolution. Alternatively, the activity of CXCR2�

monocytic activity may predominate at this time point (5).The observation that both MCP-1 gene expression and protein

levels increases over time to peak at 8 days in WT controls andfollows a similar bell-shaped curve to the influx of monocytes

FIGURE 6. Proposed hypothesis of CCR2 signalingactivity after stasis DVT in the mouse. Early events areprimarily related to thrombolysis (collagenolysis) in partvia MMP-2 and MMP-9. Later activities are related tothrombus neovascularization. D4, Day 4; C, vein wallcells.

3395The Journal of Immunology

suggests a central signaling role for the MCP-1-CCR2 axis in DVTresolution. Deletion of CCR2 was associated with greatly in-creased levels of MCP-1, suggesting loss of a normal feedbackinhibition. Similar elevation of MCP-1 has been observed in otherdisease models with CCR2�/� mice (10, 35). Importantly, theother CC chemokines, such as RANTES and MIP-1�, presumablyacting via competent CCR5 signaling, were not significantly up-regulated in the CCR2�/� mice despite significantly fewer throm-bus monocytes. This suggests that these other monocyte chemo-tactic signals do not compensate for the loss of CCR2 signalingand are probably not integral for DVT resolution. Lastly, althoughothers have found that exogenous MCP-1 accelerated thrombusmaturation (8), we have not found any significant effect at earlytime points of 4 and 8 days (our unpublished data) and suggest thatother ligands of CCR2 such as MCP-2—MMP-5 may beimportant.

The presence of early collagen is consistent with a thrombusproviding a healing matrix scaffold for scar formation (5, 19).Markedly increased thrombus collagen was observed in theCCR2�/� mice at early time points and at 4 days was associatedwith reduced vein wall levels of MMP-2 and MMP-9, suggestingimpaired collagenolysis. This observation has also been describedin a model of pulmonary and cardiac fibrosis in CCR2�/� mice,showing significantly reduced MMP-2 and MMP-9 activity (11,47). Although MMP-9 has collagenolytic activity, it is primarilyagainst type I collagen, for which little was present in the throm-bus. More likely in this system is that MMP-9, and to a lesserextent, MMP-2 probably act against type IV collagen and laminin,both of which are present in the early thombus (Ref. 7 and ourunpublished observations). Alternatively, a different collagenase,such as MMP-13, may be decreased with CCR2 gene deletion;thus, MMP-2 and MMP-9 may only be surrogate markers.

The association of CCR2 deletion with organ fibrosis is mixedin other disease models. For example, in the bleomycin pulmonaryinjury model, CCR2 deletion is associated with less late collagenaccumulation (9, 11). In contrast, and more similar to the presentfindings, deletion of CCR2 was associated with significantlygreater organ fibrosis in a pulmonary model of infection and amodel of mechanical vascular injury (48, 49). Thus, the local en-vironment and mechanism of injury may be the driving factor ofhow CCR2 signaling contributes to the fibrotic response. Althoughthe Th2 responses are both anti-inflammatory and profibrogenic(15, 41), no elevation of Th2 lymphokines, such as IL-4 or IL-13,were found in the CCR2�/� thrombi to account for the increasedthrombus collagen. However, it is possible that a relative imbal-ance existed in the CCR2�/� mice (i.e., significantly less Th1lymphokines present) and that the Th2 lymphokines may have un-impeded activities.

Because deletion of CCR2 signaling is associated with impairedIFN-� production (14, 28, 50) and IFN-� is a known activator ofa local MMP release (16), we asked whether exogenous replace-ment of this lymphokine could restore normal DVT resolution.Indeed, the DVT resolution was normalized with exogenous IFN-�as measured by size, collagen content, and MCP-1 levels. Impor-tantly, exogenous IFN-� restored vein wall MMP-9, but notMMP-2, to nearly WT levels. This occurred despite significantlyreduced thrombus F4/80 cells and no CCR2� cells. Converselyand consistent with these observations is that IFN-� inhibition wasassociated with a significant reduction in MMP-2, but not MMP-9,and no impairment of DVT resolution. Although the presence ofnormal levels of MMP-9 may explain the lack of DVT inhibition,those WT mice that received anti-IFN-� had a normal thrombuscontent of F4/80 and CCR2� cells. Taken together, these datasupport the notion that IFN-� is sufficient but not necessary for

DVT resolution. Given that IFN-� Ab inhibition or genetic dele-tion did not impair DVT resolution, other signals including IL-1�and TNF-� may compensate for the loss of this signaling mediator(28). Consistently, thrombus IL-1� is present in the resolving WTthrombi and was significantly lower at 2 and 8 days after DVT inCCR2�/� mice. Finally, although IFN-� can directly decrease col-lagen synthesis (51), we believe this is not a likely mechanism asanti-IFN-�-treated mice had collagen levels similar to those ofcontrol WTs (Fig. 4B). Taken together, these findings suggest thata redundant system of MMP activation exists, and the most im-portant common factor may be CCR2 signaling that mediatescross-talk between the vein wall and adjacent thrombus cells.

The plasmin system, by activation of uPA, is a primary mech-anism of venous thrombolysis (27). It is now clear that arterial,pulmonary, and venous vascular beds may have differential throm-botic and fibrinolytic activities (52). Although we were not able tomeasure plasmin activity directly, the measures of uPA/PAI-1,quantification of thrombus uPA� cells and gene expression at 4days, and thrombus fibrin content all suggest that loss of CCR2signaling did not inhibit this process. Although similar numbers ofuPA� cells were present, we speculate that each cell may haveproduced more uPA per cell to account for this difference. Indeed,the ratio of thrombus uPA to PAI-1 was elevated in the CCR2�/�

mice but was normalized with the addition of IFN-� (possiblybecause of restored MMP-9 activity). We speculate that later dif-ferences in DVT resolution associated with significantly feweruPA� cells in the CCR2�/� mice (at 8 and 12 days) were probablysecondary to impaired monocyte influx, because monocytes areprimary sources of uPA (26, 27). It is also unlikely that theCCR2�/� mice had any intrinsic procoagulant or fibrinolytic de-fects, given that as baseline clotting parameters were similar.

These findings have clinical applications. First, the addition oflimited duration local pro-MMP-9 mediators may hasten throm-bolysis in addition to the native or exogenously delivered plasmin-ogen activators. More data are accruing that the longer a thrombusis in contact with the vein wall, the greater is the vein wall damage(40). Cellular specific modulation to hasten DVT resolution maybe particularly fruitful for rapid reduction of thrombus burdenwithout the complications of bleeding risk associated withanticoagulants.

DisclosuresThe authors have no financial conflict of interest.

References1. Heit, J. A., M. D. Silverstein, D. N. Mohr, T. M. Petterson, C. M. Lohse, W. M.

O’Fallon, and L. J. Melton, 3rd. 2001. The epidemiology of venous thromboem-bolism in the community. Thromb. Haemost. 86: 452–463.

2. Linkins, L. A., P. T. Choi, and J. D. Douketis. 2003. Clinical impact of bleedingin patients taking oral anticoagulant therapy for venous thromboembolism: ameta-analysis. Ann. Intern. Med. 139: 893–900.

3. Geerts, W. H., G. F. Pineo, J. A. Heit, D. Bergqvist, M. R. Lassen, C. W. Colwell,and J. G. Ray. 2004. Prevention of venous thromboembolism: the Seventh ACCPConference on Antithrombotic and Thrombolytic Therapy. Chest 126:338S–400S.

4. Wakefield, T. W., R. M. Strieter, C. A. Wilke, A. M. Kadell, S. K. Wrobleski,M. D. Burdick, R. Schmidt, S. L. Kunkel, and L. J. Greenfield. 1995. Venousthrombosis-associated inflammation and attenuation with neutralizing antibodiesto cytokines and adhesion molecules. Arterioscler. Thromb. Vasc. Biol. 15:258–268.

5. Henke, P. K., A. Varga, S. De, C. B. Deatrick, J. Eliason, D. A. Arenberg,P. Sukheepod, P. Thanaporn, S. L. Kunkel, G. R. Upchurch, Jr., andT. W. Wakefield. 2004. Deep vein thrombosis resolution is modulated by mono-cyte CXCR2-mediated activity in a mouse model. Arterioscler. Thromb. Vasc.Biol. 24: 1130–1137.

6. Deatrick, K. B., J. L. Eliason, E. M. Lynch, A. J. Moore, N. A. Dewyer,M. R. Varma, C. G. Pearce, G. R. Upchurch, T. W. Wakefield, and P. K. Henke.2005. Vein wall remodeling after deep vein thrombosis involves matrix metal-loproteinases and late fibrosis in a mouse model. J. Vasc. Surg. 42: 140–148.

7. Henke, P. K., T. W. Wakefield, A. M. Kadell, M. J. Linn, M. R. Varma,M. Sarkar, A. Hawley, J. B. Fowlkes, and R. M. Strieter. 2001. Interleukin-8

3396 CCR2 AND DVT RESOLUTION

administration enhances venous thrombosis resolution in a rat model. J. Surg.Res. 99: 84–91.

8. Humphries, J., C. L. McGuinness, A. Smith, M. Waltham, R. Poston, andK. G. Burnand. 1999. Monocyte chemotactic protein-1 (MCP-1) accelerates theorganization and resolution of venous thrombi. J. Vasc. Surg. 30: 894–899.

9. Belperio, J. A., M. P. Keane, M. D. Burdick, J. P. Lynch, 3rd, Y. Y. Xue,A. Berlin, D. J. Ross, S. L. Kunkel, I. F. Charo, and R. M. Strieter. 2001. Criticalrole for the chemokine MCP-1/CCR2 in the pathogenesis of bronchiolitis oblit-erans syndrome. J. Clin. Invest. 108: 547–556.

10. Boring, L., J. Gosling, S. W. Chensue, S. L. Kunkel, R. V. Farese, Jr.,H. E. Broxmeyer, and I. F. Charo. 1997. Impaired monocyte migration and re-duced type 1 (Th1) cytokine responses in C-C chemokine receptor 2 knockoutmice. J. Clin. Invest. 100: 2552–2561.

11. Okuma, T., Y. Terasaki, K. Kaikita, H. Kobayashi, W. A. Kuziel, M. Kawasuji,and M. Takeya. 2004. C-C chemokine receptor 2 (CCR2) deficiency improvesbleomycin-induced pulmonary fibrosis by attenuation of both macrophage infil-tration and production of macrophage-derived matrix metalloproteinases.J. Pathol. 204: 594–604.

12. Weber, C., A. Schober, and A. Zernecke. 2004. Chemokines: key regulators ofmononuclear cell recruitment in atherosclerotic vascular disease. Arterioscler.Thromb. Vasc. Biol. 24: 1997–2008.

13. Charo, I. F., and M. B. Taubman. 2004. Chemokines in the pathogenesis ofvascular disease. Circ. Res. 95: 858–866.

14. Peters, W., M. Dupuis, and I. F. Charo. 2000. A mechanism for the impairedIFN-� production in C-C chemokine receptor 2 (CCR2) knockout mice: role ofCCR2 in linking the innate and adaptive immune responses. J. Immunol. 165:7072–7077.

15. Liew, F. Y. 2002. Th1 and Th2 cells: a historical perspective. Nat. Rev. Immunol.2: 55–60.

16. Schroder, K., P. J. Hertzog, T. Ravasi, and D. A. Hume. 2004. Interferon-�: anoverview of signals, mechanisms and functions. J. Leukocyte Biol. 75: 163–189.

17. Sato, N., W. A. Kuziel, P. C. Melby, R. L. Reddick, V. Kostecki, W. Zhao,N. Maeda, S. K. Ahuja, and S. S. Ahuja. 1999. Defects in the generation of IFN-�are overcome to control infection with Leishmania donovani in CC chemokinereceptor (CCR) 5-, macrophage inflammatory protein-1�-, or CCR2-deficientmice. J. Immunol. 163: 5519–5525.

18. Wakefield, T. W., M. J. Linn, P. K. Henke, A. M. Kadell, C. A. Wilke,S. K. Wrobleski, M. Sarkar, M. D. Burdick, D. D. Myers, and R. M. Strieter.1999. Neovascularization during venous thrombosis organization: a preliminarystudy. J. Vasc. Surg. 30: 885–892.

19. Varma, M. R., A. J. Varga, B. S. Knipp, P. Sukheepod, G. R. Upchurch,S. L. Kunkel, T. W. Wakefield, and P. K. Henke. 2003. Neutropenia impairsvenous thrombosis resolution in the rat. J. Vasc. Surg. 38: 1090–1098.

20. Sevitt, S. 1973. The vascularisation of deep-vein thrombi and their fibrous resi-due: a post mortem angiographic study. J. Pathol. 111: 1–11.

21. Pepper, M. S. 1997. Manipulating angiogenesis: from basic science to the bed-side. Arterioscler. Thromb. Vasc. Biol. 17: 605–619.

22. Sunderkotter, C., K. Steinbrink, M. Goebeler, R. Bhardwaj, and C. Sorg. 1994.Macrophages and angiogenesis. J. Leukocyte Biol. 55: 410–422.

23. Waltham, M., K. G. Burnand, M. Collins, and A. Smith. 2000. Vascular endo-thelial growth factor and basic fibroblast growth factor are found in resolvingvenous thrombi. J. Vasc. Surg. 32: 988–996.

24. Heil, M., T. Ziegelhoeffer, S. Wagner, B. Fernandez, A. Helisch, S. Martin,S. Tribulova, W. A. Kuziel, G. Bachmann, and W. Schaper. 2004. Collateralartery growth (arteriogenesis) after experimental arterial occlusion is impaired inmice lacking CC-chemokine receptor-2. Circ. Res. 94: 671–677.

25. Hansson, G. K. 2001. Immune mechanisms in atherosclerosis. Arterioscler.Thromb. Vasc. Biol. 21: 1876–1890.

26. Northeast, A. D., K. S. Soo, L. G. Bobrow, P. J. Gaffney, and K. G. Burnand.1995. The tissue plasminogen activator and urokinase response in vivo duringnatural resolution of venous thrombus. J. Vasc. Surg. 22: 573–579.

27. Singh, I., K. G. Burnand, M. Collins, A. Luttun, D. Collen, B. Boelhouwer, andA. Smith. 2003. Failure of thrombus to resolve in urokinase-type plasminogenactivator gene-knockout mice: rescue by normal bone marrow-derived cells. Cir-culation 107: 869–875.

28. Visse, R., and H. Nagase. 2003. Matrix metalloproteinases and tissue inhibitorsof metalloproteinases: structure, function, and biochemistry. Circ. Res. 92:827–839.

29. Whatling, C., W. McPheat, and E. Hurt-Camejo. 2004. Matrix management: as-signing different roles for MMP-2 and MMP-9 in vascular remodeling. Arterio-scler. Thromb. Vasc. Biol. 24: 10–11.

30. Thanaporn, P., D. D. Myers, S. K. Wrobleski, A. E. Hawley, D. M. Farris,T. W. Wakefield, and P. K. Henke. 2003. P-selectin inhibition decreases post-thrombotic vein wall fibrosis in a rat model. Surgery 134: 365–371.

31. Chiu, B. C., C. M. Freeman, V. R. Stolberg, E. Komuniecki, P. M. Lincoln,S. L. Kunkel, and S. W. Chensue. 2003. Cytokine-chemokine networks in ex-perimental mycobacterial and schistosomal pulmonary granuloma formation.Am. J. Respir. Cell Mol. Biol. 29: 106–116.

32. Xiong, W., Y. Zhao, A. Prall, T. C. Greiner, and B. T. Baxter. 2004. Key rolesof CD4� T cells and IFN-� in the development of abdominal aortic aneurysms ina murine model. J. Immunol. 172: 2607–2612.

33. Varma, M. R., D. M. Moaveni, N. A. Dewyer, A. J. Varga, K. B. Deatrick,S. L. Kunkel, G. R. Upchurch, Jr., T. W. Wakefield, and P. K. Henke. 2004. Deepvein thrombosis resolution is not accelerated with increased neovascularization.J. Vasc. Surg. 40: 536–542.

34. Henke, P. K., L. A. DeBrunye, R. M. Strieter, J. S. Bromberg, M. Prince,A. M. Kadell, M. Sarkar, F. Londy, and T. W. Wakefield. 2000. Viral IL-10 genetransfer decreases inflammation and cell adhesion molecule expression in a ratmodel of venous thrombosis. J. Immunol. 164: 2131–2141.

35. Hogaboam, C. M., C. L. Bone-Larson, S. Lipinski, N. W. Lukacs, S. W. Chensue,R. M. Strieter, and S. L. Kunkel. 1999. Differential monocyte chemoattractantprotein-1 and chemokine receptor 2 expression by murine lung fibroblasts derivedfrom Th1- and Th2-type pulmonary granuloma models. J. Immunol. 163:2193–2201.

36. Sullivan, V. V., A. E. Hawley, D. M. Farris, B. S. Knipp, A. J. Varga,S. K. Wrobleski, P. Thanapron, M. J. Eagleton, D. D. Myers, J. B. Fowlkes, andT. W. Wakefield. 2003. Decrease in fibrin content of venous thrombi in selectin-deficient mice. J. Surg. Res. 109: 1–7.

37. Eliason, J. L., K. K. Hannawa, G. Ailawadi, I. Sinha, J. W. Ford,M. P. Deogracias, K. J. Roelofs, D. T. Woodrum, T. L. Ennis, P. K. Henke, et al.2005. Neutrophil depletion inhibits experimental abdominal aortic aneurysm for-mation. Circulation 112: 232–240.

38. Ginzinger, D. G. 2002. Gene quantification using real-time quantitative PCR: anemerging technology hits the mainstream. Exp. Hematol. 30: 503–512.

39. Upchurch, G. R., J. W. Ford, S. J. Weiss, B. S. Knipp, D. A. Peterson,R. W. Thompson, M. J. Eagleton, A. J. Broady, M. C. Proctor, and J. C. Stanley.2001. Nitric oxide inhibition increases matrix metalloproteinase-9 expression bya rat aortic smooth muscle cells in vitro. J. Vasc. Surg. 34: 76–83.

40. Prandoni, P., A. W. Lensing, M. H. Prins, E. Bernardi, A. Marchiori, P. Bagatella,M. Frulla, L. Mosena, D. Tormene, A. Piccioli, P. Simioni, and A. Girolami.2002. Residual venous thrombosis as a predictive factor of recurrent venousthromboembolism. Ann. Intern. Med. 137: 955–960.

41. Wynn, T. A. 2004. Fibrotic disease and the Th1/Th2 paradigm. Nat. Rev. Immu-nol. 4: 583–594.

42. Mack, M., J. Cihak, C. Simonis, B. Luckow, A. E. Proudfoot, J. Plachy, H. Bruhl,M. Frink, H. J. Anders, V. Vielhauer, et al. 2001. Expression and characterizationof the chemokine receptors CCR2 and CCR5 in mice. J. Immunol. 166:4697–4704.

43. Omata, N., M. Yasutomi, A. Yamada, H. Iwasaki, M. Mayumi, and Y. Ohshima.2002. Monocyte chemoattractant protein-1 selectively inhibits the acquisition ofCD40 ligand-dependent IL-12-producing capacity of monocyte-derived dendriticcells and modulates Th1 immune response. J. Immunol. 169: 4861–4866.

44. Ducharme, A., S. Frantz, M. Aikawa, E. Rabkin, M. Lindsey, L. E. Rohde,F. J. Schoen, R. A. Kelly, Z. Werb, P. Libby, and R. T. Lee. 2000. Targeteddeletion of matrix metalloproteinase-9 attenuates left ventricular enlargement andcollagen accumulation after experimental myocardial infarction. J. Clin. Invest.106: 55–62.

45. Vos, C. M., S. Gartner, R. M. Ransohoff, J. C. McArthur, L. Wahl, L. Sjulson,E. Hunter, and K. Conant. 2000. Matrix metalloprotease-9 release from mono-cytes increases as a function of differentiation: implications for neuroinflamma-tion and neurodegeneration. J. Neuroimmunol. 109: 221–2466.

46. Peters, W., J. G. Cyster, M. Mack, D. Schlondorff, A. J. Wolf, J. D. Ernst, andI. F. Charo. 2004. CCR2-dependent trafficking of F4/80dim macrophages andCD11c dim/intermediate dendritic cells is crucial for T cell recruitment to lungsinfected with Mycobacterium tuberculosis. J. Immunol. 172: 7647–7653.

47. Kaikita, K., T. Hayasaki, T. Okuma, W. A. Kuziel, H. Ogawa, and M. Takeya.2004. Targeted deletion of CC chemokine receptor 2 attenuates left ventricularremodeling after experimental myocardial infarction. Am. J. Pathol. 165:439–447.

48. Blease, K., B. Mehrad, T. J. Standiford, N. W. Lukacs, J. Gosling, L. Boring,I. F. Charo, S. L. Kunkel, and C. M. Hogaboam. 2000. Enhanced pulmonaryallergic responses to Aspergillus in CCR2�/� mice. J. Immunol. 165: 2603–2611.

49. Schober, A., A. Zernecke, E. A. Liehn, P. von Hundelshausen, S. Knarren,W. A. Kuziel, and C. Weber. 2004. Crucial role of the CCL2/CCR2 axis inneointimal hyperplasia after arterial injury in hyperlipidemic mice involves earlymonocyte recruitment and CCL2 presentation on platelets. Circ. Res. 95:1125–1133.

50. Traynor, T. R., W. A. Kuziel, G. B. Toews, and G. B. Huffnagle. 2000. CCR2expression determines T1 versus T2 polarization during pulmonary Cryptococcusneoformans infection. J. Immunol. 164: 2021–2027.

51. Amento, E. P., N. Ehsani, H. Palmer, and P. Libby. 1991. Cytokines and growthfactors positively and negatively regulate interstitial collagen gene expression inhuman vascular smooth muscle cells. Arterioscler. Thromb. 11: 1223–1230.

52. Mackman, N. 2005. Tissue-specific hemostasis in mice. Arterioscler. Thromb.Vasc. Biol. 25: 2273–2281.

3397The Journal of Immunology