Monitoring of Blood Vessels and Tissues by a Population of Monocytes with Patrolling Behavior

See allHide authors and affiliations

Science  03 Aug 2007:
Vol. 317, Issue 5838, pp. 666-670
DOI: 10.1126/science.1142883


The cellular immune response to tissue damage and infection requires the recruitment of blood leukocytes. This process is mediated through a classical multistep mechanism, which involves transient rolling on the endothelium and recognition of inflammation followed by extravasation. We have shown, by direct examination of blood monocyte functions in vivo, that a subset of monocytes patrols healthy tissues through long-range crawling on the resting endothelium. This patrolling behavior depended on the integrin LFA-1 and the chemokine receptor CX3CR1 and was required for rapid tissue invasion at the site of an infection by this “resident” monocyte population, which initiated an early immune response and differentiated into macrophages.

Mammalian monocytes consist of two main subsets of immune cells (1, 2), which arise from a common hematopoietic progenitor, the macrophage and dendritic cell (DC) precursor (MDP, or monoblast) that also gives rise to conventional resident spleen DCs (cDCs) and several tissue macrophage subsets (3, 4). So-called “inflammatory” monocytes express the cell surface protein Ly6c (Gr1+), the chemokine receptor CCR2, and the adhesion molecule L-selectin and are selectively recruited to inflamed tissues and lymph nodes (1, 5). They are able to differentiate into inflammatory DC (1, 6, 7) and can replenish resident cell compartments in the skin, digestive tract, and lung (3, 8). The second subset of monocytes has been termed “resident” in mice (1, 2) because they were found in both resting and inflamed tissues, although their functions are still unknown. This subset is defined by a smaller size, high expression of the chemokine receptor CX3CR1 and LFA-1 integrin, and by the lack of expression of Ly6c (Gr1), CCR2, and L-selectin (1, 2). Two monocyte subsets can also be identified in humans (9), with CD14+ CD16 monocytes resembling mouse inflammatory monocytes and CD14low CD16+ monocytes sharing a phenotype similar to that of mouse resident monocytes (1). Resident and inflammatory monocytes thus appear to be defined by distinct sets of adhesion molecules and chemokine receptors, which suggests different modes of tissue trafficking.

To explore this possibility further, we developed a strategy to study the behavior and functions of blood monocytes, in real time, under steady state or inflammatory conditions. Intravital confocal microscopy imaging was undertaken in vivo in a way that allowed us to observe cells within capillaries and postcapillary vessels in the dermis (Fig. 1A) and in larger veins and arteries, that is, mesenteric vessels (Fig. 1B) (10). Cx3cr1gfp/+ mice, which express green fluorescent protein (GFP) in monocytes [but also in natural killer (NK) cells and some T cells] and Rag2–/–, γc–/–, Cx3cr1gfp/+ mice, in which monocytes are the only blood cells expressing GFP, were used as reporters (Fig. 1, C and D) (10).

Fig. 1.

Intravital imaging of mouse monocytes. (A) CX3CR1-expressing cells express gfp in reporter mice, and dermal blood vessels are labeled in red rhodamine-conjugated dextran. (B) The branches of the mesenteric vein (arrowhead) and mesenteric artery (arrow) are surgically exposed. (C) Flow cytometry analysis of blood leukocytes from Cx3cr1gfp/+ mice, Rag2–/– Cx3cr1gfp/+ mice Rag2–/–, γc–/– Cx3cr1gfp/+ mice, and Rag2–/–, γc–/–, Cx3cr1gfp/+ mice reconstituted with wild-type lymphoid cells. (D) Gr1+ (gfplow) and Gr1 (gfphigh) monocytes were sorted by flow cytometry and examined by confocal microscopy. Based on gfp intensity, Gr1 (gfphigh) monocytes sorted by fluorescence-activated cell sorting (FACS) are easily distinguishable by confocal microscopy from FACS-sorted Gr1+ (gfplow) monocytes labeled with a far-red cell tracker (shown in blue in the middle panel).

Intravital microscopy observation of tissues in the steady state revealed that monocytes within most blood vessels in the dermis (Fig. 2, A and B, and movies S1 and S2) and in the branches of the mesenteric vein and the mesenteric artery (Fig. 2C, fig. S1, and movies S3 and S4) exhibited a constitutive “crawling” type motility. In contrast, rolling gfp+ monocytes were observed only transiently after surgery in the mesenteric veins (fig. S1 and movies S3 and S4) but not in arteries (movie S4). Analysis of gfp signal intensity per pixel indicated that crawling monocytes belonged predominantly to the gfphigh subset (CX3CR1high Gr1) as compared with monocytes that perform rolling in the same vessels (Fig. 2D and fig. S2). These finding suggested either that gfphigh monocytes were crawling inside blood vessels onto endothelial cells, despite the blood flow, or that they were located outside the lumen of the blood vessel. When the green gfp signal was omitted, crawling gfphigh monocytes appeared as dark spots on confocal sections of the vessel labeled in red with the fluorescent dextran, indicating that monocytes are located inside the blood vessels (Fig. 2E and movie S5). Crawling polymorphonuclear neutrophils (PMN) or lymphocytes would appear as dark crawling cells in the blood of mice; however, gfp crawling cells were not observed in the steady state (movie S1), suggesting that these cells did not behave in the same way and that gfp+ monocytes were the majority of cells that crawl on the endothelium in the steady state.

Fig. 2.

Gfphigh Gr1 monocytes crawl inside blood vessels. (A) Dermal blood vessels in Rag2–/–, γc–/– Cx3cr1gfp/+ reporter mice. Monocytes appear as green dots (arrows). (B and C) Green signal from images of time-lapse series are summed to display the path of labeled cells in the dermis (B) and mesenteric vessels (C). In (C), the open arrow indicates a branch of the mesenteric vein, and the closed arrow indicates a branch of the mesenteric artery. (D) gfp maximal pixel intensity was calculated as indicated in fig. S2. (E) Confocal section (6 μm) examined with red (intravenous dextran) and green (gfp+ monocytes) signals, or for the red signal only (shown in gray, right panel). Cells that are located within the vessels appear as a dark signal (arrow). (F) Overlay of monocyte tracks from two representative vessels, plotted after aligning their starting positions. (G to I) Path (G), kymograph (H), and instantaneous velocity (I) of a cell from movie S1. (J) Monocyte paths were classified as loops, hairpin, waves, mixed pattern, and short path (<40 μm) (fig. S3). (K) Average velocity was calculated as indicated in (10). (L) Confinement ratio of gfphigh monocytes. Scatter plot represents the ratio of the distance to origin of tracked cells versus their path length.

Although monocytes were located inside the vessels, overlay of individual cell tracks plotted after aligning their starting positions indicated that the direction of their crawling movement was not dependent on the blood flow (Fig. 2F). Extravasation was rarely observed in the steady state. The path of individual cells indicated that monocytes in blood vessels appeared to describe loops (25%), hairpin (17%), waves (38%), mixed pattern (9%), and short path (<40 μm, 11%) (Fig. 2, G to J, and fig. S3). The average instantaneous velocity of crawling monocytes was 4 to 20 μm/min (average 12 μm/min) (Fig. 2K and fig. S4). The distance traveled by individual cells was, on average, only half their path length, indicating that the cells have a high confinement ratio (Fig. 2L). The result of these crawling movements was that in small postcapillary venules, after an hour, monocytes appeared to have extensively monitored the endothelium of a given vessel (Fig. 2B and movie S2), leading us to describe their behavior as patrolling.

The velocity profile of flow in blood vessels is generally parabolic across the cross section of the vessel, and calculated values decrease to zero at the blood vessel wall (11). However, crawling against the midstream blood flow is counterintuitive, and this suggested that monocytes were closely adherent to the luminal side of the endothelium. We therefore explored the molecular basis of monocyte patrolling. Lymphocytes and PMNs have been shown to roll at speeds of ∼40 μm/sec under conditions of flow at sites of inflammation along the endothelium of postcapillary venules (12, 13). This initial contact allows endothelial membrane-bound chemokines to activate leukocyte integrins through Gαi-linked chemokine receptor (14). Changes in integrin affinity then allow the rolling leukocytes to stick firmly, occasionally crawl onto the endothelial cell toward the closest intracellular cell junction (15, 16), and diapedese across the endothelium. Patrolling was slower than rolling by a factor of 100 to 1000 and was observed in the steady state, long-range, and independent of the direction of midstream blood flow and of extravasation. However, it appears to involve a “firm binding” to endothelium, and therefore we reasoned that integrin may be involved. Gfphigh (Gr1) monocytes express the β2 integrins LFA-1 (CD11a/CD18, αLβ2) and Mac-1 (CD11b/CD18, αMβ2) (1). Intravenous injection of blocking antibodies to either CD11a or CD18, but not of antibodies to CD11b or control immunoglobulin (Ig), resulted in the rapid, complete, and prolonged release of monocytes from the endothelial wall (Fig. 3, A and B, and movie S6), indicating that LFA-1 is required for crawling. Gfphigh monocytes also express high levels of the chemokine receptor CX3CR1 (1), whereas its ligand, Fractalkine, is a transmembrane molecule expressed on endothelial cells (1719). CX3CR1-fractalkine interaction in vitro has been described to mediate adhesion between monocytes and endothelial cells through activation of integrins (20) and through an intrinsic adhesion function (18, 20, 21). The number of crawling monocytes was reduced by two-thirds in CX3CR1-deficient mice, and the average path length of crawling cells was reduced by one-half, resulting in a six-fold decrease in patrolling (Fig. 3, C and D, and movie S7), whereas the velocity of the remaining crawling monocytes and the numbers of circulating blood gfphigh monocytes were similar in Cx3cr1–/– and Cx3cr1+/– mice (Fig. 3, E and F, and fig. S5). It is notable that blocking antibodies to LFA-1 can detach crawling monocytes in Cx3cr1+/– mice, which suggests that CX3CR1-dependent crawling is largely mediated through LFA-1 in vivo.

Fig. 3.

Monocyte crawling is mediated in vivo through the integrin LFA-1 and the chemokine receptor CX3CR1. (A) Kymograph of a time-lapse series before and after intravenous injection of CD11a-blocking antibody (4 mg per kg of weight). (B) Percentage of crawling cells was determined as a ratio between the numbers of cells crawling before and 10 min after injection of control Ig or blocking antibody directed against CD11b, CD18, and CD11a. Experiments were repeated at least three times for each antibody. (C) Maximum projection of time-lapse series obtained from intravital microscopy experiments in postcapillary venules of the ear in steady-state conditions. (D) Histograms represent number of crawling cells per hour, mean path length of these crawling cells, and distance to origin, either in Rag2–/–, γc–/– Cx3cr1gfp/+ or Rag2–/–, γc–/– Cx3cr1gfp/gfp mice in the steady state. (E) Histograms represent number of Gr1 monocytes in the blood of 10-week-old Rag2–/–, γc–/– Cx3cr1gfp/+ and Rag2–/–, γc–/– Cx3cr1gfp/gfp mice. (F) Representation of the crawling velocity of Gr1 monocytes from Rag2–/–, γc–/– Cx3cr1gfp/+ (solid line) and Rag2–/–, γc–/– Cx3cr1gfp/gfp (dashed line) mice.

The patrolling monocytes are ideally located to provide immune surveillance of endothelial cells and surrounding tissues. In response to tissue damage, gfphigh monocytes extravasated rapidly within 1 hour and invaded the surrounding tissues after exposure to irritants (Fig. 4, A and B), aseptic wounding (Fig. 4C), and peritoneal infection with Listeria monocytogenes (Fig. 4D). To study in depth the kinetics, phenotype, and functions of extravasated monocytes, we used the L. monocytogenes peritoneal infection model, because extravasated cells can easily be recovered by peritoneal lavage. In this model, extravasation of Gr1 gfphigh monocytes peaked at 2 hours after infection, at a time when PMN is only beginning to enter the peritoneum and several hours before the extravasation of conventional Gr1+ gfplow monocytes is observed, and was significantly delayed in patrolling-deficient Cx3cr1–/– mice (Fig. 4D and table S1). Therefore, patrolling was associated with, and required for, early extravasation and tissue invasion by Gr1 gfphigh monocytes. Extravasated Gr1 gfphigh monocytes were responsible for an early inflammatory response (Fig. 4, E and F, and fig. S6). At 1 and 2 hours after infection, Gr1 gfphigh monocytes were the only producers of tumor necrosis factor-–α (TNFα), a cytokine central to inflammation, as detected by intracellular flow cytometry and polymerase chain reaction (PCR) (Fig. 4F). Genes coding for interleukin-1 (IL-1), lysozyme, defensins, complement, pattern recognition receptors such as TLRs, scavenger receptors, and IgFc receptors, and chemokines involved in the recruitment and activation of other effector cells were also up-regulated (Fig. 4E and fig. S6).

Fig. 4.

Rapid tissue invasion by patrolling Gfphigh Gr1 monocytes after tissue damage or infection. (A to C) Crawling gfphigh monocytes in the mouse dermis are tracked for 1 hour after application of phosphate-buffered saline (PBS) (A) or 1:1 dibutylphtalate acetone (B), or after aseptic damage with a microinjection pipette loaded with a far-red dye (shown in blue) (C). (D) Number of extravasated cells recovered from the peritoneal cavity of mice infected with L. monocytogenes. (E) Analysis of gene expression on purified GFPhigh monocytes (fig. S6) (10). (F) GFPhigh monocytes recovered from the peritoneum of infected mice are the main producers of TNFα in vivo at 2 hours after L. monocytogenes infection. (Data are mean ± SD; n = 3 in each group; representative experiment out of four). (G) Extravasated Gr1 gfphigh monocytes initiated a macrophage differentiation program, at the expense of DC differentiation, whereas Gr1+ gfplow monocytes initiated a DC differentiation program. Regulation of Mafb, cMaf, Pu.1, and RelB genes was analyzed as described in figs. S6 and S7.

Notably, the production of TNF and IL-1 was transient and turned off at 8 hours (Fig. 4F), whereas extravasated Gr1 gfphigh monocytes turned on—at 2 and 8 hours—genes involved in tissue remodeling (22), including arginase1, Fizz1, Mgl2, and mannose receptor (MR) (Fig. 4E and fig. S9).

Indeed, the study of the balance of transcription factors that specifies the alternative macrophage or dendritic cell fate of monocytes (23) indicated that extravasated Gr1 gfphigh monocytes initiated a typical macrophage differentiation program, characterized by up-regulation of cMaf and MafB but not RelB and Pu.1 (sfpi1) (Fig. 4G and figs. S6 and S7). In contrast, the conventional Gr1+ gfplow monocytes initiated a DC differentiation program, as described previously (1, 6), by up-regulating RelB and Pu.1 but not cMaf and MafB (Fig. 4G and fig. S6). Analysis of the expression of a larger panel of genes differentially regulated in monocyte-derived macrophages and DC supported this conclusion (fig. S8).

These findings demonstrate a new mechanism of leukocyte crawling on endothelial cells and a new role for LFA-1. The present data also assign a function to gfphigh Gr1 monocytes. Patrolling of blood vessels by these resident monocytes allow rapid tissue invasion by monocytes in case of damage and infection, followed by the initiation of an innate immune response and their differentiation into macrophages. This is in contrast to the role of Gr1+ monocytes, which reach the inflammatory site later and give rise to inflammatory DCs (1, 6). This reveals an unsuspected dichotomy of the differentiation potential and functions of blood monocyte subsets during infection. The extravasation of patrolling Gr1 gfphigh monocytes is likely to be dependent on a yet-unidentified signal(s) from damaged tissue and/or endothelium. Interestingly, the existence of a pool of “marginated” monocytes expressing CD16+ has been proposed in humans (24) and may correspond, at least in part, to the patrolling behavior that we describe here, suggesting a similar in vivo function of mouse resident CX3CR1high (gfphigh) Gr1low monocytes and human CX3CR1high CD16+ CD14low monocytes. Monocytes are abundant in arthritis and atherosclerotic lesions (25, 26), and CX3CR1, TNF-α, and LFA-1 have been implicated in the pathogenesis of these inflammatory diseases (27, 28); thus, patrolling monocytes may contribute to their pathogenesis and may represent a target for treatment.

Supporting Online Material

Materials and Methods

Figs. S1 to S8

Table S1


Movies S1 to S7

References and Notes

View Abstract

Navigate This Article