Differentiation of Monocytes into Dendritic Cells in a Model of Transendothelial Trafficking

See allHide authors and affiliations

Science  16 Oct 1998:
Vol. 282, Issue 5388, pp. 480-483
DOI: 10.1126/science.282.5388.480


Essential to the dendritic cell system of antigen-presenting cells are the veiled dendritic cells that traverse afferent lymph to enter lymph nodes, where they initiate immune responses. The origin of veiled cells, which were discovered 20 years ago, is unclear. Monocytes cultured with endothelium differentiated into dendritic cells within 2 days, particularly after phagocytosing particles in subendothelial collagen. These nascent dendritic cells migrated across the endothelium in the ablumenal-to-lumenal direction, as would occur during entry into lymphatics. Monocytes that remained in the subendothelial matrix became macrophages. Therefore, monocytes have two potential fates associated with distinct patterns of migration.

One of the important features of dendritic cells (DCs) is their capacity to migrate from peripheral tissues to lymphoid organs and initiate immunity. DCs gain access to the spleen from the bloodstream and enter lymph nodes by migration through afferent lymphatic vessels (1). Bearing soluble proteins (2) and particulates (3) that they acquire before entry into lymph nodes, DCs localize to the T cell areas where they are ideally positioned to select and activate clones of antigen-reactive T lymphocytes. When afferent lymphatic conduits are severed, immunity to peripherally administered antigen does not develop (4). Lymph DCs may also induce peripheral tolerance to self-antigens acquired during the capture of apoptotic cells (5, 6). After interacting with T cells, DCs probably die within the lymph nodes, because they do not enter efferent lymph (1).

The origin of lymph-borne DCs, often called veiled cells because of their many sheetlike cellular processes, is unclear. Large-scale migration of DCs from immunologically perturbed tissues can be attributed in part to the mobilization of interstitial and epidermal DCs (7, 8). Yet even in the absence of overt stimuli, numerous DCs traffic through lymph (9) at an estimated rate of 104 cells per hour (10,11), suggesting that veiled DCs are constitutively generated from a precursor population. When rats are pulsed with [3H]thymidine, it is evident that veiled cells, which are not in cell cycle, arise from a precursor that underwent cell division 1 to 4 days (mode = 3 days) previously (11).

Monocytes continuously emigrate from the blood into peripheral tissues with a half-life in the blood of about 1 day in mice (12). Nondividing monocytes can differentiate into DCs in culture. However, this conversion requires treatment of monocytes with combinations of cytokines, typically interleukin-4 (IL-4) and granulocyte-macrophage colony-stimulating factor, and takes 7 to 11 days in culture (13, 14). We now describe a culture system in which monocytes differentiate into DCs in only 2 days as they traverse endothelial monolayers in the basal-to-apical direction (reverse transmigration), a migratory process that is reminiscent of the movement of cells from tissues into lymphatic vessels. Monocytes that do not reverse transmigrate become macrophages, exhibiting a phenotype entirely different from DCs.

In this culture system, freshly isolated blood mononuclear cells were incubated with unstimulated monolayers of human umbilical vein endothelial cells grown on an endotoxin-free collagenous matrix (15). Within 1 hour, monocytes bound to and diapedesed into the subendothelial collagen, by use of β2 integrins (CD18) (15) and CD31 (16), respectively. Because few lymphocytes transmigrated under these conditions, subendothelial cells were greatly enriched for CD14+cells (about 90%). During 2 days of further culture, half of the subendothelial monocytes subsequently reverse transmigrated across the overlying, intact endothelium (17,18). Reverse transmigration does not require CD18 or CD31 but uses functional P-glycoprotein (MDR-1), a transmembrane transporter of the adenosine triphosphatase–binding cassette superfamily (18). Similarly, the migration of DCs from skin through authentic lymphatic channels does not require CD18 or CD31 but is blocked by antagonists of P-glycoprotein (18).

Using this model (19), we investigated whether monocyte-derived, reverse-transmigrated cells could develop into DCs by monitoring the phenotype of the cells at different stages in culture. The cells that initially transmigrated into the collagen (lumenal-to-ablumenal transendothelial migration) after a 1.5-hour incubation of endothelium with bulk peripheral blood mononuclear cells (PBMCs) were typical monocytes. These cells expressed CD14 and the Fcγ receptors CD32 and CD64 (Fig. 1A). Human leukocyte antigen–DR (HLA-DR) was present but at relatively low levels. The T cell costimulatory molecules CD80 and CD86 were absent or very low, respectively, and the DC maturation marker CD83 (20) was undetectable. A few CD3+ T cells were also retained in the cultures (about 10% of total); CD3CD14CD19HLA-DR+nonmonocytic DC precursors (21) were present at 2 to 5% relative to CD14+ monocytes.

Figure 1

Flow cytometric evaluation and immunohistochemistry of monocyte-derived cells before and after culture with endothelial cell–collagen constructs. Monocytes were incubated with endothelium grown on type I collagen gels. (A) Cells were collected from cultures that received no exogenous stimulus (unfilled profiles) or from cultures in which one to five zymosan particles per migrated monocyte were incorporated into the subendothelial collagen (filled profiles). Flow cytometric profiles (30) show monocytes (left) collected immediately after their transmigration into the collagen (profiles from unstimulated and zymosan-containing cultures overlap), reverse-transmigrated cells (middle) collected after 48 hours, and monocyte-derived cells remaining in the subendothelial matrix (right) at 48 hours. Tick marks on thex axis designate logarithmic increments in fluorescence intensity. (B to E) Immunostaining (31) for p55 (dark cytoplasmic staining indicates positive staining) is shown for reverse-transmigrated (B) and subendothelial (C) cells derived from cultures containing zymosan. Arrows illustrate cells with ingested zymosan particles. Digests to obtain cells residing in the collagen contain some clusters of endothelial cells (asterisk). Lysosomal markers were examined with cells derived from IL-1–stimulated cultures. The presence of DC-LAMP in perinuclear lysosomes (arrowhead) was observed in reverse-transmigrated cells (D). Subendothelial monocyte-derived cells expressed another lysosomal marker, CD68, throughout the cytoplasm (E).

Reverse-transmigrated cells were collected after 48 hours (22), when most cells that will reverse transmigrate have done so (17, 18). The monocyte-derived cells that did not reverse transmigrate but remained in the collagen (23) were also analyzed. In the absence of additional stimuli, most reverse-transmigrated cells and subendothelial mononuclear phagocytes had a similar phenotype (Fig. 1A). CD32 was maintained at levels similar to monocytes, but CD14 and CD64 typically decreased by 50%. CD80 and CD86 remained low. Other T cell stimulatory molecules, including HLA-DR (Fig. 1A) and CD54 (24), increased by an order of magnitude.

Incorporation of phagocytic particles, including latex microspheres and zymosan (yeast cell walls), in the collagen gel had a striking differential effect on the cells that reverse transmigrated compared with those that remained in the subendothelial matrix. About half of reverse-transmigrated cells displayed characteristics of mature DCs (Fig. 1A and Table 1). CD14, CD32, and CD64 (Fig. 1A) were lost, as was the integrin CD11b (24). T cell costimulators CD80, CD86, and HLA-DR were all up-regulated, and there was de novo expression of the DC maturation marker CD83. CD1a, a marker expressed by epidermal DCs but not other DC subsets, was absent. In comparison, cells that remained beneath the endothelium in the presence of phagocytic particles retained CD14 and Fc receptors and either reduced or failed to up-regulate CD80, CD83, CD86, and HLA-DR (Fig. 1).

Table 1

Total cells recovered and extent of DC maturation in reverse-transmigrated (RT) and subendothelial (SE) monocyte-derived populations.

View this table:

When cytospin preparations of the cells were examined for intracellular markers, many reverse-transmigrated cells had the features of DCs. These cells were large, with extended dendritic processes. They were positive for the DC marker actin bundling protein p55 (25) (Fig. 1B). In contrast, the subendothelial cells had the appearance of typical phagocytes, lacking p55 (Fig. 1C). Reverse-transmigrated cells were also positive for the newly identified DC-specific lysosomal marker DC-LAMP (26) (Fig. 1D), whereas subendothelial cells were strongly positive throughout the cytoplasm for a different lysosomal marker, CD68, expressed by macrophages (Fig. 1E) (27). Both reverse-transmigrated and subendothelial cells phagocytosed the zymosan when it was present in the collagen (Fig. 1, B and C).

After 48 hours of culture, total cell yields were similar in all conditions tested, so that only the extent to which reverse-transmigrated cells had become mature DCs varied depending on the stimulus (Table 1). The number of nonmonocytic CD14 large mononuclear cells retained in the cultures was too low to account for the abundance of DCs recovered, strongly implying that the majority of DCs were derived from CD14+monocytes. Irradiation of monocytes before they were added to endothelial cultures produced the same number of DCs by all criteria as cultures that received unirradiated monocytes [14% mature DCs compared with 13% DCs, respectively, from lipopolysaccharide (LPS)–stimulated cultures], indicating that proliferation of precursors is not required for the development of DCs. Sorting monocytes to purity by selecting CD64+ cells (28), of which more than 97% were also CD14+, resulted in 29% mature DCs when CD64+ cells were added to endothelial cultures as a purified population. By comparison, when we recombined sorted CD64+ cells with CD64 blood DC precursors to match representation of the latter to their abundance in the original nonsorted population (28), 30% of reverse-transmigrated cells matured to DCs. Because DCs developed in the absence of CD64 cells to a similar extent as in their presence, these data indicate that reverse-transmigrated DCs originated predominantly from CD14+CD64+monocytes.

Reverse-transmigrated cells from unstimulated endothelial cell cultures were studied further to examine whether they would differentiate to DCs. Removed from the endothelium after 48 hours, these cells were subsequently cultured in medium lacking additional stimuli or in medium containing LPS or five zymosan particles per cell. After 3 days of incubation in these conditions, 16, 24, and 81% were mature DCs, respectively (Fig. 2, B to D). Thus, although only 6% of reverse-transmigrated cells from unstimulated cultures resembled mature DCs after collection from endothelial cultures (Fig. 2A), these data indicate that such cells can mature to DCs upon further stimulation. However, cells that did not mature fully to DCs could also revert to macrophages when continued in culture on plastic dishes, as previously documented for immature DCs (14). Fully mature CD83+ DCs did not revert to a macrophage phenotype.

Figure 2

Maturation of unstimulated reverse-transmigrated cells to DCs after collection from endothelium and analysis of the requirement for endothelium. Maturation was monitored by expression of CD83 and high levels of HLA-DR. Quadrants are marked on the basis of the amount of fluorescence intensity observed in samples stained with control mAbs. Cells were analyzed just after collection from unstimulated endothelial cultures at 48 hours (A) or after 3 days of further culture in Teflon beakers containing 20% heat-inactivated human serum and Medium 199 with no supplements (B), a supplement of LPS (10 ng/ml) (C), or five zymosan particles per cell (D). Other monocytes were never added to endothelial cell cultures but instead were incubated in Teflon beakers in the same medium containing LPS (10 ng/ml) (E) or were incubated with fibronectin-coated collagen gels impregnated with zymosan particles but lacking an endothelial monolayer (F). Data are representative of three experiments.

Exposure to the endothelium was essential for this rapid differentiation of monocytes to DCs. When monocytes were incubated in the presence of LPS or zymosan particles but were not added to endothelial-collagen cultures (Fig. 2E) or were incubated with the collagen matrix containing zymosan particles but lacking an endothelial monolayer (Fig. 2F), they expressed low HLA-DR and had little to no detectable expression of CD83 or p55.

To verify that the reverse-transmigrated cells had acquired the functional features of DCs, we carried out mixed lymphocyte reactions. The reverse-transmigrated cells were potent stimulators of allogeneic T cell proliferation and much more efficacious than their subendothelial counterparts (Fig. 3). The stimulatory capacity of reverse-transmigrated cells was increased upon addition of a macrophage-conditioned medium (MCM) (Fig. 3A) that has been described as a potent maturation cocktail for DCs (14) or by treatment with LPS (Fig. 3B) or IL-1. Reverse-transmigrated cells derived from monocytes that were irradiated before they were added to endothelial cultures promoted proliferation of T cells as well as their unirradiated counterparts, underscoring the conclusion that DCs from these cultures derived from a pool of nonproliferating mononuclear cells (Fig. 3B). In agreement with the potentiating effects of phagocytic stimuli on surface markers of DC maturation in the vessel wall cultures (Table 1), reverse-transmigrated cells derived from cultures containing latex (24) or zymosan (Fig. 3A) were the most potent stimulators of T cell proliferation.

Figure 3

Capacity of reverse-transmigrated and subendothelial cells to stimulate allogeneic T cell proliferation. After monocytes entered collagen gels (1.5 hours), cultures were treated with LPS (1 ng/ml) and 30% MCM or incubated in medium without additional additives. Zymosan particles were incorporated into the collagen of some cultures. Monocyte-derived cells taken from reverse-transmigrated (filled symbols) or subendothelial populations (open symbols) were collected from endothelial cell cultures at 48 hours of incubation and irradiated with 30 grays before being combined with allogeneic T cells (A) (32). Alternatively, monocytes were irradiated just after their isolation and subsequently added to endothelium (B) (preirradiated samples) for a 48-hour incubation, and then reverse-transmigrated and subendothelial populations were collected. For all conditions, monocyte-derived cells were mixed with T cells at indicated ratios and cultured for 5 days and then finally in [3H]thymidine (4 μCi/ml) for 18 hours. Counts per minute for T cells alone was <300. Standard errors were not greater than 10% of the mean value. (A) and (B) show two independent analyses that are representative of five experiments.

Consistent with our observations, previous studies have observed that the composition of large mononuclear cells in lymph (excluding lymphocytes) is heterogeneous with respect to the extent to which they have matured to DCs. Some of these cells exhibit features characteristic of monocytes (7, 9, 29), including residual phagocytic activity. A few studies have noted that lymph DCs have phagolysosomes containing cellular debris (7, 11) and other particles (3). Thus, phagocytic stimuli, possibly including the ingestion of apoptotic cells (6), may well be a maturation stimulus for DCs in vivo. However, development of monocytes to fully differentiated DCs may not always predominate in vivo. In one study of lymph collected from superficial leg lymphatics of humans (29), cells that resembled monocytes were three to four times more abundant than those that resembled mature DCs. These monocyte-like cells may be analogous to reverse-transmigrated cells from our unstimulated cultures.

Our studies indicate that, in the presence of endothelial cells grown on an extracellular matrix, the differentiation of monocytes diverges along two distinct pathways toward DCs or macrophages. DCs arise from monocytes that migrate across endothelium in an ablumenal-to-lumenal direction, whereas macrophages develop from monocytes that remain in the subendothelial matrix. We suggest that endothelial tissues initiate differentiation of monocytes to DCs and that full differentiation requires an additional stimulus such as the uptake of particles or exposure to microorganisms.

  • * To whom correspondence should be addressed. E-mail: GJRandol{at}


Stay Connected to Science

Navigate This Article