A CCR2 macrophage endocytic pathway mediates extravascular fibrin clearance in vivo

Conversion of fibrinogen into the insoluble polymer, fibrin, stems blood loss after vessel rupture. Furthermore, fibrin deposited in extravascular space forms a provisional matrix that supports cell migration during tissue repair and is critical for controlling initial stages of bacterial infection.^1-5 

Because of its potent proinflammatory properties, the rate of deposition and removal of extravascular fibrin must be carefully coordinated. This is illustrated by the inflammation-associated multiorgan pathology and impaired tissue regenerative capacity of humans and mice deficient in the key fibrinolytic protease zymogen, plasminogen,^6-17  as well as by the capacity of extravascular fibrin to exacerbate the morbidity of a range of chronic human diseases, including multiple sclerosis, tissue fibrosis, muscular dystrophy, and rheumatoid arthritis.^18-24 

Plasminogen is a serine protease zymogen present in plasma and extravascular fluids that is converted to the active protease plasmin by endoproteolytic cleavage by the closely related trypsin-like serine proteases urokinase plasminogen activator (uPA) and tissue plasminogen activator (tPA).^25,26 

Four pathways for plasminogen activation are known in the context of physiological fibrinolysis: (1) fibrin-dependent tPA-mediated plasminogen activation, in which fibrin binds plasminogen and tPA to bring the two molecules in close apposition to favor plasminogen activation^27-30 ; (2) cell-dependent, tPA-mediated plasminogen activation, which involves the receptor-mediated binding of tPA and plasminogen to the cell surface^31-38 ; (3) cell-dependent, uPA-mediated plasminogen activation, which involves the binding of uPA to the uPA receptor (uPAR) and receptor-mediated binding of plasminogen to the cell surface^39-44 ; and (4) a poorly understood uPAR-independent, uPA-mediated plasminogen activation pathway, which may be cell dependent or cell independent.^15,17,45-54  Although mechanistically distinct, these pathways display considerable functional redundancy in extravascular fibrin surveillance.^15,17,45-53 

The enzymatic pathways that facilitate productive plasmin formation are well defined, but the cellular and molecular mechanisms by which fibrin ultimately is cleared from extravascular space are poorly investigated. Plasmin digestion of fibrin ex vivo results in the release of fibrin degradation products of high molecular weight.⁵⁵  Extravascular fibrin deposits are infiltrated by leukocytes,^{15,39,51,53,56}  and cultured primary macrophages, human peripheral blood mononuclear cells, and monocytoid cell lines all can endocytose soluble fibrin monomer.^57,58  Furthermore, early electron microscopy studies reported an abundance of fibrillar material morphologically consistent with fibrin in leukocytes associated with extravascular fibrin deposits in rheumatoid arthritis.^59-61  This suggests that extravascular fibrin degradation may be orchestrated at the cellular level and include an intracellular lysosomal step.

To gain insight into the process of extravascular fibrin degradation, we used intravital imaging with subcellular resolution to directly visualize the dissolution of fibrin matrices placed within subcutaneous space and to identify the cell types, enzymes, and receptors involved. We report that fibrin is degraded predominantly by a C-C chemokine receptor type 2 (CCR2)-positive subpopulation of macrophages via a plasmin-dependent endocytic mechanism that is functional in the absence of the established fibrin(ogen) receptors αMβ2 (Mac-1, CD11b/CD18) and intercellular adhesion molecule 1 (ICAM-1) or the integrity of the major integrin-binding site on fibrin.

Animal procedures were performed in an Association for Assessment and Accreditation of Laboratory Animal Care–accredited vivarium under approved protocols. Mouse strain and genotyping details are in supplemental Table 1 (available on the Blood Web site).

Citrated plasma in Tris-buffered saline, 5 mM benzamidine, and 200 U/mL aprotinin was precipitated twice with 25% saturated (NH₄)₂SO₄ at room temperature for 15 min. The precipitate was resuspended in 50 mM imidazole and 100 mM NaCl, dialyzed against HEPES-buffered saline, and stored at −80°C. Fibrinogen (1 mg) was fluorescently conjugated using the Alexa Fluor 488 protein-labeling kit (Life Technologies) as per the manufacturer’s recommendations.

Fluorescent fibrin gels were made by polymerizing plasminogen-depleted, FXIII-containing fibrinogen with thrombin in the presence of Alexa-conjugated human fibrinogen. The detailed procedure is found in supplemental Data.

A 1-cm dorsal incision was made on the left side of the spine of isoflurane-anesthetized mice. Labeled fibrin gel was placed into a subcutaneous pocket made under the left flank. The pocket was filled with a 250-µL solution of 0.4 mg/mL Alexa Fluor 647 (10 000 MW), anionic, fixable dextran or Texas red (10 000 MW), lysine fixable dextran (Life Technologies) in phosphate-buffered saline (PBS), and the incision was closed with wound clips. Mice were administered Hoechst 33342 intravenously as previously described⁶²  and killed by CO₂ inhalation 3 to 4 hours later. Immediately thereafter, the location of the fibrin gel was determined with a NightSea DFP-1 Dual Fluorescent Protein Flashlight, and the dermis over the detected gel was marked and excised with a 10-mm margin around the edge. The tissue was immediately imaged by confocal microscopy or whole-mount immunostained. Confocal imaging was performed using an inverted confocal microscope (IX81; Olympus) equipped with a scanning head (FluoView 1000; Olympus) as described previously.⁶²  Details on imaging are found in supplemental Data.

Three to five 30-µm z-stacks were collected from each sample, first identifying with the EPI scope the point of highest fibrin signal and acquiring stacks 1500 µm above, left, below, and right of this point. Cell number in each z-stack was determined by counting all nuclei stained with Hoechst dye. Internalization was scored as the presence of fluorescently labeled fibrin in vesicular structures around Hoechst-labeled nuclei. The percentage of cells positive for internalization or the percentage of CCR2-RFP–, Cx3Cr1-GFP-, or Col1a1-GFP–positive cells was calculated by dividing the number of cells scored positive for that signal by the total number of cells. Imaris 3D reconstruction software (Bitplane, South Windsor, CT) was used for image generation.

Whole-mount staining of tissues was performed as described previously.⁶²  Details on antibodies and incubation conditions are found in supplemental Data.

Alexa Fluor 488–labeled fibrin gels were implanted subcutaneously as described above. For each time point, at least 3 control gels in PBS containing 0.5% sodium azide were left in the dark at 37°C as reference. Mice were killed, and fibrin gels were visualized and removed using the NightSea flashlight. The gels were weighed and dissolved overnight at 37°C in 1.5 mL 5 mg/mL trypsin, 4.8 mM EDTA. Vortexed samples were centrifuged at 16 000g for 2 minutes, dispensed in 100-µL aliquots into Costar 96-well black clear-bottom plates, and read with a Synergy Neo HTS multimode microplate reader (BioTek) with the absorption wavelength at 488 nm and emission wavelength at 520 nm. Alexa Fluor 488–conjugated fibrinogen in PBS was used to produce standard curves.

Flow cytometry was performed using cells isolated from fibrin implantation zones. Details on analysis are found in supplemental Data.

Real-time PCR analysis was done on cells isolated by fluorescence-activated cell sorting using TaqMan gene expression arrays. Details on analysis are found in supplemental Data.

CCR2-depleter^+/0 and littermate control mice were implanted with fibrin gels. The mice received daily intraperitoneal injections with 50 ng/g body weight diphtheria toxin, starting 1 day prior to gel implantation and continuing until termination at 24 hours after fibrin implantation.

To explore if the degradation of extravascular fibrin involves endocytic uptake, we developed an assay to directly visualize the fate of fibrin gels placed in extravascular space. We generated fibrin gels ex vivo by polymerizing fibrinogen with thrombin. The gels incorporated fibrinogen labeled with Alexa Fluor 488, 594, or 647, which are pH insensitive, photostable fluorescent dyes that can be visualized in various extracellular and intracellular compartments during in vivo imaging procedures^62,63  (see “Materials and methods” for details on this and other procedures). In agreement with the reported photostability, fibrin gels stored in the dark at 37°C in PBS with sodium azide lost just 8% fluorescence signal in 14 days, when compared with control gels stored in the dark at −80°C. We first implanted fluorescent fibrin gels into the subcutis of mice to determine the fate of fibrin placed in extravascular space. Mice were killed 1.25, 4, 7, and 15 days later, the gels were extracted, and their weight and total fluorescence was compared with control gels stored in the dark at 37°C in PBS with sodium azide for the identical time period (Figure 1A). The weight of the implanted fibrin gels decreased rapidly to 9% of their original weight within 30 hours. In contrast, the fluorescence of implanted gels displayed a slower reduction with time, with a half-life of ∼9 days.

To determine the cellular contribution to extravascular fibrin degradation in vivo, we implanted fluorescently labeled fibrin together with Alexa Fluor 647–labeled 10 kDa dextran, which accumulates in lysosomes within 2 hours of administration.^62,64,65  Furthermore, we systemically injected fluorescent Hoechst 33342 dye prior to imaging to visualize cell nuclei.⁶²  Enumeration of cell density in the fibrin implantation zone, by analysis of confocal z-stacks, revealed a large increase in cellularity, when compared with control mice or mice undergoing sham surgery, showing that fibrin implantation causes cell recruitment (Figure 1B). Inspection of confocal z-stacks 2, 6, 12, and 24 hours after fibrin implantation showed progressive accumulation of cells within the fibrin implantation zone (Figure 1C-F). Importantly, most cells within the implantation zone displayed fibrin-containing intracellular vesicles. Many of these fibrin-containing vesicles colocalized with the Alexa Fluor 647 dextran lysosomal marker (Figure 1G), suggesting that endocytosed extravascular fibrin is targeted to the lysosome. To provide independent support for an endocytic pathway for extravascular fibrin clearance, we implanted fluorescent fibrin into mice that express plasma membrane-localized tdTomato fluorescent protein.⁶⁶  Inspection of confocal z-stacks from fibrin implantation zones of these mice indeed revealed numerous fibrin-containing vesicles completely circumscribed by plasma membrane (Figure 1H). Taken together, these data show that degradation of extravascular fibrin involves endocytosis and lysosomal targeting.

We set out to identify the cell types engaged in endocytic degradation of extravascular fibrin. For this purpose, we used a combination of whole-mount staining with antibodies against cell-type–specific markers and transgenic mice expressing cell-type–specific fluorescent tags. Interestingly, 54.7% of all cells observed within the fibrin implantation zone at 24 hours postimplantation expressed the pan-macrophage marker F4/80. Furthermore, 58.3% of these cells endocytosed fibrin, and 70.2% of all fibrin-endocytosing cells were F4/80 positive, suggesting a key role of macrophages in mediating extravascular fibrin internalization (Figure 2A,B,G,H). In contrast, whole-mount staining using the neutrophil marker NIMP.R14 revealed that neutrophils constituted only 11.4% of cells in the fibrin implantation zone and just 4.9% of fibrin-uptaking cells, suggesting a minor contribution of neutrophils to endocytic fibrin degradation (Figure 2C,D,G,H). To determine if nonleukocyte cell populations contribute to endocytic fibrin turnover, we implanted fibrin gels into mice hemizygous for a Col1a1-GFP transgene (Col1a1-GFP^+/0 mice) expressed in dermal fibroblasts⁶⁷  and quantified the abundance of GFP-positive cells in the implantation zone and their uptake of fibrin. This revealed that dermal fibroblasts engage in extravascular fibrin turnover to a minor extent, constituting 6.4% of all cells and 9.7% of all fibrin-uptaking cells (Figure 2E-H). A total of 27.5% of all cells (Figure 2G) and 15.3% of fibrin-internalizing cells (Figure 2H) in fibrin implantation zones did not score as macrophages, neutrophils, or fibroblasts, as defined by expression of either of the 3 markers. These cells may be macrophages, neutrophils, or fibroblasts expressing insufficient levels of, respectively, F4/80, NIMP.R14, or the Col1a1-GFP transgene to be scored as positive or may represent an unidentified cell population. This analysis reveals a dominant role of macrophages in endocytic fibrin uptake, with minor contributions from neutrophils and fibroblasts.

Tissue-infiltrating macrophages arise from distinct populations of circulating CCR2-positive, CX3CR1-negative (or CCR2-positive, CX3CR1-low expressing) and CCR2-negative, CX3CR1-positive monocytes.⁶⁸  To determine if specific subpopulations of macrophages engaged in endocytic fibrin degradation, we implanted fibrin into mice that carried 1 wild-type and 1 RFP-tagged Ccr2 allele (CCR2^+/RFP mice) and quantified cellular fibrin uptake after 24 hours. CCR2-positive macrophages constituted 38.6% of all cells and 54.4% of all fibrin-uptaking cells within the implantation zone (Figure 3A-B). Quantification of cellular fibrin uptake after implantation of fluorescent fibrin gels into mice that carry 1 wild-type and 1 GFP-tagged Cx3cr1 allele (Cx3cr1^+/GFP mice) showed that CX3CR1-positive macrophages constituted 32.2% of all cells and 40.9% of all fibrin-uptaking cells (Figure 3D-E). We next implanted fibrin into mice that carried 1 wild-type and 1 GFP-tagged Cx3cr1 allele as well as 1 RFP-tagged Ccr2 allele (Cx3cr1^+/GFP;CCR2^+/RFP mice). Interestingly, nearly all CX3CR1-positive cells and nearly all fibrin-endocytosing CX3CR1-positive cells were also positive for CCR2, with CX3CR1 single-positive cells constituting just 1.36% of all cells and 1.40% of all fibrin-uptaking cells (Figure 3G-H). These macrophage populations, in part, accumulated in response to fibrin implantation, with the density of CCR2-positive cells in fibrin-implanted mice being, respectively, 25.4- and 5.4-fold higher than control and sham-operated mice, and the density of CX3CR1-positive cells being, respectively, 34.5- and 1.75-fold higher than control and sham-operated mice (Figure 3C,F).

Fibrin-uptaking cells were morphologically different from collagen-degrading macrophages⁶²  by being smaller with fewer and smaller lysosomes. This indicated that fibrin- and collagen-endocytosing macrophages constitute distinct macrophage subpopulations. Consistent with this, analysis of injection fields from fibrillar collagen-injected CCR2^+/RFP mice showed that collagen endocytosing cells and CCR2-positive cells were distinct cell populations (Figure 3I).

We next isolated CD45-positive cells from fibrin implantation zones and performed flow cytometry analysis after staining with antibodies against the proliferation marker Ki-67. This analysis showed that fibrin-internalizing and fibrin noninternalizing CD45-positive cells displayed low proliferation rates (Figure 4A-E). Of potential interest, a 4.5-fold reduction in Ki-67–positive cells was observed in fibrin-internalizing cells compared with noninternalizing cells, tentatively suggesting that fibrin uptake may modulate cell proliferation. However, CCR2 fibrin-positive and fibrin-negative cells could not be compared directly, due to loss of RFP signal by cell permeabilization.

To analyze the expression of plasminogen activator, we performed real-time PCR of populations of CD45, CCR2 double-positive fibrin-internalizing and CD45-positive fibrin-noninternalizing cells from fibrin implantation zones isolated by fluorescence-activated cell sorting (CD45, CCR2 double-positive cells were not obtained in sufficient numbers for analysis). Both fibrin-internalizing and fibrin-noninternalizing cells expressed uPA and uPAR messenger RNA (mRNA) at comparable levels (Figure 4F). tPA mRNA expression was below the detection limit (data not shown).

To determine the effect of the selective depletion of CCR2 macrophages on endocytic fibrin uptake, we used a transgenic mouse strain expressing the human diphtheria toxin receptor controlled by the Ccr2 gene promoter (CCR2-depleter^+/0 mice⁶⁹ ). This mouse was crossed to CCR2^+/RFP mice to generate CCR2^+/RFP;CCR2-depleter^+/0 bitransgenic mice and CCR2^+/RFP littermates. These mice were pretreated with diphtheria toxin and implanted with fibrin, and the fibrin implantation zones were analyzed 24 hours later. As expected, a dramatic reduction of CCR2-positive cells was observed in diphtheria toxin–treated CCR2^+/RFP;CCR2-depleter^+/0 bitransgenic mice, when compared with diphtheria toxin–treated single-transgenic CCR2^+/RFP littermates (Figure 5A, examples in Figure 5C-D) and caused a 43% reduction in the total number of fibrin-endocytosing cells (Figure 5B). Fibrin uptake in CCR2-negative cells, characterized in Figure 3, was not affected by loss of CCR2-positive cells. The discrepancy in the abundance of these cells (Figure 3A and Figure 5B) may represent experimental variation and/or difference in method of quantitation (fraction of cells vs cells per volume unit).

Myeloid cells bind purified fibrin via an interaction between the αMβ2 (Mac-1, CD11b/CD18) integrin and an integrin-binding RLTIGE (human)/RLSIGE (mouse) motif within the fibrinogen γ chain.^3,70,71  Furthermore, soluble fibrin monomer is endocytosed by cultured myeloid cells in an αMβ2-dependent manner.⁵⁸  Indeed, flow cytometry analysis showed that nearly all CCR2-positive, fibrin-internalizing cells expressed αMβ2 (Figure 6A-D), making αMβ2 a strong fibrin endocytosis candidate receptor. To determine if αMβ2 integrin is critical for endocytic fibrin uptake, we implanted fibrin into αM-deficient (Itgam^−/−) mice and wild-type littermates. Loss of αMβ2 caused a small, nonsignificant, increase of cells in the fibrin implantation zone (Figure 6M). Unexpectedly, however, loss of αMβ2 integrin did not affect endocytic fibrin uptake, as assessed by the fraction of cells with endocytosed fibrin, indicating that fibrin endocytosis can take place in the absence of this integrin (Figure 6Q). One possible explanation for this unanticipated result was that other β2 integrins engaged the RLSIGE fibrinogen γ-chain motif during fibrin endocytosis. To investigate this, we employed fibrinogen from Fgg^390-396A knockin mice, in which the fibrinogen RLSIGE γ-chain motif is replaced by alanines. This mutant fibrinogen fails to support myeloid cell adhesion in vitro.³  Fibrinogen was purified from Fgg^390-396A mice and wild-type littermates, fluorescently labeled, and implanted into mice. Total cell numbers in fibrin implantation zones were similar in wild-type and mutant fibrin-implanted mice (Figure 6N). Surprisingly, analysis of implantation fields 24 hours after fibrin implantation showed no difference in the fraction of cells endocytosing mutant and wild-type fibrin (Figure 6R). The lack of evidence for involvement of a β2 integrin–dependent mechanism of cellular uptake led us to examine the contribution of ICAM-1, a non–integrin-type fibrin(ogen) receptor expressed on leukocytes.⁷²  Indeed, most fibrin-internalizing CCR2-positive cells expressed ICAM-1 (Figure 6E-H), and total cell accumulation in fibrin implantation zones was unaffected by ICAM-1 deficiency (Figure 6O). Thus, we implanted fibrin gels into ICAM-1–deficient (Icam1^−/−) mice and wild-type littermates and determined the fraction of cells internalizing fibrin after 24 hours (Figure 6S). Again, no effect of ICAM-1 deficiency was observed on fibrin endocytosis.

Mannose receptor (MR) is an endocytosis receptor responsible for macrophage-mediated collagen endocytosis,^62,73-76  and we found it to be is expressed in a subset of CCR2-positive fibrin-internalizing cells (Figure 6I-L). To examine if the mannose receptor functions as a fibrin endocytosis receptor, for example via a carbohydrate-mediated interaction, we implanted fibrin into MR-deficient (Mrc1^−/−) mice and wild-type littermates and analyzed cell accumulation (Figure 6P) and fibrin endocytosis (Figure 6T). This analysis revealed no effect of loss of mannose receptor on fibrin endocytosis. Overall, this analysis shows that fibrin endocytosis can take place in mice with individual deficiencies in αMβ2, ICAM-1, and MR.

We next investigated the involvement of extracellular proteases in endocytic fibrin uptake and, thus, the possible mechanistic coupling of extracellular and intracellular fibrin degradation pathways. Plasmin is the principal fibrinolytic protease and was recently shown to directly stimulate macrophage phagocytosis.⁷⁷  Therefore, we first implanted fibrin gels into plasminogen-deficient (Plg^−/−) and wild-type littermates and quantified fibrin uptake. Interestingly, although loss of plasminogen did not significantly affect the abundance of cells within the implantation zone (Figure 7A, examples in Figure 7D-E), it caused a dramatic reduction in the fraction of cells that internalized fibrin at 24 hours (Figure 7B, examples in Figure 7D-E), demonstrating a pivotal role of plasminogen in enabling the cellular uptake of extravascular fibrin. Consistent with this loss of cellular fibrin uptake, fluorescent fibrin gels implanted into plasminogen-deficient mice remained essentially intact when followed for up to 14 days, and gels implanted into plasminogen heterozygous-deficient (Plg^+/−) mice displayed reduced degradation when compared with littermate wild-type mice (Figure 7C). Even at 14 days after fibrin implantation, very few cells internalized fibrin in plasminogen-deficient mice (Figure 7F-G), showing that plasminogen deficiency likely is poorly compensated by alternative leukocyte fibrinolytic pathways⁷⁸  in the context of endocytic fibrin uptake.

To determine the role of plasminogen activators in endocytic fibrin uptake, we next implanted fibrin gels into uPA-deficient (Plau^−/−) mice, uPAR-deficient (Plaur^−/−) mice, and tPA-deficient (Plat^−/−) mice and their respective wild-type littermates. In contrast to the loss of plasminogen, individual loss of uPA or uPAR did not affect fibrin endocytosis, and loss of tPA caused only a small, nonsignificant reduction in endocytosis (Figure 7H-J).

The lack of effect of individual loss of uPA or tPA on fibrin endocytosis indicated that plasminogen activation was dispensable for plasminogen-dependent fibrin endocytosis, or, alternatively, that uPA and tPA activated plasminogen in a functionally redundant manner in this context. To discriminate between the 2 possibilities, we implanted fibrin gels into knockin mice homozygous for point mutations in the endogenous Plg gene that substitute the active site serine within the catalytic His-Ser-Asp triad of plasmin with alanine (Plg^S743A/S743A mice) and their wild-type littermates. Plg^S743A/S743A mice express normal levels of the mutant plasminogen, which can undergo activation site cleavage but is enzymatically inactive.⁷⁹  Like wholesale loss of plasminogen, loss of plasmin enzymatic activity dramatically reduced fibrin endocytosis (Figure 7K), demonstrating that plasmin-mediated fibrin cleavage is a prerequisite for cellular fibrin uptake and lysosomal degradation in vivo. Additional evidence for overlapping functions of uPA and tPA in activating plasminogen in this process was revealed by the much lower number of fibrin-endocytosing cells in fibrin-implanted uPA and tPA double-deficient (Plau^−/−;Plat^−/−) mice, as compared with tPA single-deficient Plat^−/− littermates (Figure 7L). Taken together, the above studies reveal a novel mechanistic coupling of extracellular and endocytic fibrin degradation pathways (Figure 7M).

We used intravital microscopy to investigate how fibrin is removed from extravascular space and found that removal of fibrin deposited in the mouse dermis involves endocytic uptake and lysosomal routing of the fibrin. We proceeded to characterize this endocytic fibrin degradation pathway at the cellular and molecular level and elucidated the following characteristics:

• a. Fibrin endocytosis is executed primarily by CCR2 single-positive and CCR2, CX3CR1 double-positive macrophages that are recruited or emerge in response to fibrin deposition. Minor contributions from CX3CR1-positive, CCR2-negative macrophages, neutrophils, and fibroblasts were also identified.

• b. Fibrin endocytosis requires plasminogen and plasminogen activation, likely to fragment fibrin to make it amenable for endocytic uptake, to expose cellular binding sites on the partially digested fibrin, or to stimulate general phagocytic activity.⁷⁷ 

• c. uPA and tPA have redundant functions in the activation of plasminogen in the context of fibrin endocytosis.

• d. Fibrin endocytosis can take place in the absence of the fibrin receptors αMβ2 and ICAM-1 the multiligand macrophage endocytosis receptor MR, and the principal myeloid integrin engagement site on fibrin.

The aggressive cellular uptake and lysosomal targeting of extravascular fibrin observed here at first glance is surprising, in light of the well-established pathways for extracellular plasminogen activation in the context of fibrinolysis. Our findings, however, support the emerging notion that extracellular and intracellular matrix degradation pathways are functionally linked in a sequential process orchestrated by specialized cells through the coordinated expression of specific cell-surface–located matrix-degrading enzymes and extracellular matrix receptors.^80-82  In this scenario, a limited number of proteolytic enzymes, endowed with the capacity to act on nondenatured cross-linked matrices (eg, plasmin and collagenase-type matrix metalloproteinases), execute the initial fragmentation of extracellular matrices, whereas bulk degradation takes place in the lysosome following the uptake of large extracellular matrix fragments.

The identification of macrophages as the principal cell type mediating intracellular fibrin degradation is not unexpected in light of the well-established role of macrophages as professional phagocytes and their high-level expression of plasminogen activators and receptors for plasmin(ogen) and plasminogen activator.^83-85  Importantly, however, the population of macrophages engaged in endocytic fibrin turnover in the dermis is different from the population of macrophages engaged in endocytic collagen turnover in the same tissue. Thus, whereas CCR2-positive inflammatory macrophages are primarily engaged in fibrin endocytosis, interstitial collagen is predominantly degraded through endocytosis by CCR2-negative macrophages with a noninflammatory, tissue-remodeling M2-polarization (this study and Madsen et al⁶² ). Macrophages display plasticity in terms of polarization,⁸⁶  and it remains to be established if this different phenotype of fibrin- and crollagen-degrading macrophages is induced in response to the engagement of the 2 matrices, or if fibrin and collagen deposition attracts/retains fundamentally different macrophage subpopulations. Whichever is the case, this finding makes excellent sense from a wound-healing perspective. Fibrin is the first matrix to be deposited in response to tissue injury, and, as the healing process progresses, this provisional fibrin matrix is replaced by a collagenous scar. The capacity of fibrin to attract or induce inflammatory macrophages capable of orchestrating a robust antibacterial response may be critical to the initial phase of the healing process, and the ability of collagen to attract noninflammatory M2-polarized wound-healing macrophages may be essential for scar resolution.

An unexpected finding in this study was the capacity of fibrin internalization to take place in the absence of the principal fibrin-binding myeloid cell integrin αMβ2 or the integrity of the principal integrin-binding site on fibrin. Although the former could be explained by compensation by other myeloid cell β2 integrins, removal of the integrin-binding RLSIGE motif on the fibrin γ chain completely abrogates macrophage adhesion to fibrin ex vivo.³  The fibrin endocytosis pathway identified here thus appears to be fundamentally different from a previously described endocytosis pathway for soluble fibrin on cultured U-937 monocytoid cells,⁵⁸  which was described as being plasmin independent but dependent of αMβ2-mediated fibrin binding. An early study, in which endocytosis of soluble fibrin monomer by cultured peritoneal rabbit macrophages was followed by electron microscopy, provided evidence for a specific fibrin endocytosis receptor binding the N terminus of the α chain.⁵⁷  The proposition of a specific interaction with the α chain was based on the ability of a Gly-Pro-Arg peptide, mimicking the N terminus of the α chain, to block fibrin monomer endocytosis. However, we were unable to observe an effect of this peptide on the endocytosis of radiolabeled fibrin by cultured J774A.1 macrophage-like cells (unpublished data). Furthermore, overall fibrin uptake by these cells was very low, and the absence of an in vitro assay to mechanistically dissect the novel fibrin endocytotic pathway is a significant limitation of our study.

The redundancy of uPA and tPA in mediating plasminogen activation required for cellular uptake and lysosomal degradation of fibrin is well aligned with the observed fibrin-associated phenotypes of mice with single and combined plasminogen activator deficiency. Thus, deficiency in either uPA or tPA causes modest or no extravascular fibrin deposition and only small wound-healing defects, whereas combined uPA and tPA deficiency causes extensive extravascular fibrin deposition and severely compromised wound healing.^51,53  It is also consistent with the robust expression of receptors for both uPA and tPA on macrophages.^85,87  It should be noted, however, that our study does not exclude that alternative plasminogen activator–independent fibrinolytic pathways^78,88  contribute to endocytic fibrin uptake.

In summary, by using intravital microscopy, we have identified an intracellular fibrin degradation pathway that is engaged in extravascular fibrinolysis and is mechanistically distinct from previously described fibrinolytic pathways and from other extracellular matrix–degrading pathways.

The authors thank Dr Katerina Akassoglou for discussions stimulating the initiation of this work and Dr Eric G. Pamer for providing transgenic mice. We thank Drs Silvio Gutkind and Mary Jo Danton for critically reviewing this manuscript.

This research was supported by the Intramural Research Program of the National Institutes of Health, National Institute of Dental and Craniofacial Research, by the National Heart, Lung, and Blood Institute (R01 HL55374) and National Institute of Neurological Disorders and Stroke (R01 NS079639) (D.A.L.), National Heart, Lung, and Blood Institute (HL013423) (F.J.C.), and National Institute of Arthritis and Musculoskeletal and Skin Diseases (R01 AR056990) (M.J.F.). H.J.J. was supported by a Fellowship from the Danish Cancer Society.

Contribution: M.P.M, D.H.M., H.J.J., D.E.S., R.S., K.H., and R.W. designed and performed the research; D.A.L. and F.J.C. bred and provided transgenic mice; M.J.F. purified and provided mutant mouse fibrinogen; T.H.B. and D.H.M. designed research, analyzed data, and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Thomas H. Bugge, National Institute of Dental and Craniofacial Research, National Institutes of Health, 30 Convent Dr, Room 320, Bethesda, MD 20892; e-mail: thomas.bugge@nih.gov.