uPA/uPAR System Is Active in Immature Dendritic Cells Derived from CD14⁺CD34⁺ Precursors and Is Down-Regulated upon Maturation¹

Dendritic cells (DC)³ are professional APCs strategically located in the sites of Ag entry, either in epithelial microenvironment or in lymphoid organs (1, 2). They are thought to derive from precursors which undergo multiple distinct pathways of differentiation depending on the cytokine(s) released in the peripheral tissues where they localize (3, 4). At least two pathways of DC differentiation have been described: thymic DC appear to originate from a hematopoietic progenitor with lymphoid but not myeloid potential, whereas DC related to Langerhans cells can be generated in vitro from myeloid precursors using GM-CSF (5, 6, 7, 8). Human myeloid progenitors cultured with GM-CSF plus TNF-α give rise to DC with phenotypic features of Langerhans cells and a monocyte-derived type of DC which induces the production of IgM by activated B cells (5, 9).

Circulating monocytes or bone marrow precursors extravasate to enter tissues, such as tonsils, Peyer’s patches, or epidermis, where they differentiate into macrophages or DC upon inflammatory or immunological stimuli (10, 11). Once DC have encountered the Ag, they undergo full maturation and recirculate through the afferent lymph stream to the T cell-rich areas of the regional draining lymph nodes where Ag presentation takes place (12). Thus, both immature and mature DC should be able to modify their transendothelial migratory ability and matrix invasion properties, depending on the stage of differentiation, the microenvironment, and the presence and capture of the Ag. To accomplish this purpose, DC precursors follow the well-known step-wise model of rolling mediated by selectins (13) and cell adhesion involving integrins (14, 15) and migration (13, 16). The last step is also dependent on the ability of migrating cells to degrade and subsequently invade the subendothelial extracellular matrix (ECM) (13, 17). It is believed that the urokinase plasminogen activator (uPA)/uPAR system plays a critical role in favoring the directional migration of leukocytes from the bloodstream to peripheral tissues, driving leukocytes through subendothelial ECM (18, 19).

We have recently described a subset of CD14⁺CD34⁺ precursors capable of transendothelial migration and differentiation into immunostimulatory DC (20). In this paper we show that immature DC derived from such precursors use the uPA/uPAR system to invade the subendothelial matrix and reverse transmigrate across endothelial monolayers. They also respond to chemotactic agents, such as macrophage-inflammatory protein (MIP)-1α. Upon maturation induced by TNF-α, CD14⁺CD34⁺-derived DC become potent APCs, shift their sensitivity to chemokines from MIP-1α to MIP-3β, and significantly reduce their ability to migrate through ECM. These changes are accompanied by a decreased usage of the uPA/uPAR system.

CD34⁺ and CD34⁻ cell subsets were fractionated from peripheral blood CD14⁺ cells using immunomagnetic beads (Dynal, Milan, Italy) as described (20, 21). Cells were cultured in RPMI 1640 medium supplemented with 2 mM l-glutamine, 100 IU/ml penicillin and 100 μg/ml streptomycin (Biochrom, Berlin, Germany), 10% heat-inactivated FCS (PAA Labour, Linz, Austria), and 40 ng/ml recombinant GM-CSF (Shering-Plough, Milan, Italy) (20) for 5–7 days to generate DC. In some experiments, TNF-α (100 ng/ml, Genzyme, Boston, MA) was added to DC cultures during the last 48 h to obtain mature DC. Media were endotoxin-free as shown by the Limulus lysate colorimetric assay (PBI, Milan, Italy).

A total of 10⁵ cells/sample were stained with the various mAbs and then by the anti-isotype-specific FITC- or PE-conjugated goat anti-mouse Igs (GAM) (Zymed Laboratories, San Francisco, CA). The anti-uPAR R2 and R3 mAbs were kindly provided by E. Rønne and G. Høyer-Hanse (K. Danø Finsen Laboratory, Copenhagen, Denmark), whereas the anti-HLA-DR (class II) mAb was a gift from R. Accolla (ABC, Genoa, Italy). The anti-CD80/B7.1 and the anti-CD86/B7.2 mAbs were purchased from Becton Dickinson (Mountain View, CA) and the anti-CD54/ICAM-1 mAb was purchased from Immunotech (Luminy, France). Samples were run on a FACStar^Plus that was equipped with an argon-ion laser (Becton Dickinson) and gated to exclude nonviable cells. At least 5000 events/sample were analyzed. Results are expressed as Log green fluorescence intensity (arbitrary units, a.u.) vs number of cells. Cells to be examined by confocal microscopy were grown on glass coverslips, fixed in 2% paraformaldehyde, and stained with the indicated mAbs and then with FITC-GAM or tetramethylrhodamine isothiocyanate/GAM (TRITC-GAM; Zymed). Coverslips were then mounted in 50% glycerol-PBS. Microscopic analysis was conducted in a Bio-Rad MRC 1000 confocal scanning microscope (Bio-Rad Laboratories, Milan, Italy) and fluorescence images were recorded on Kodak T-Max 100 film using a Focus Imagecorder Plus (Focus Graphics, Foster City, CA) (22).

uPAR content was also measured in the supernatants of DC obtained from GM-CSF cultures (immature DC), exposed to TNF-α for the last 48 h (mature DC), or after digestion with phosphoinositide-specific phospholipase C (PI-PLC) (2 U/ml; Sigma, St. Louis, MO) for 2 h at 37°C with a sandwich ELISA as described (23).

Transmigration of radiolabeled (⁵¹Cr; NEN, Boston, MA) CD14⁺CD34⁺ cells or DC through HUVEC, isolated and cultured as described (22) and used within four passages, or Matrigel (Collaborative Research Biomedical Products, Bedford, MA) was performed with the Transwell cell culture chambers (polycarbonate filters, 5 μm pore size; Costar, Cambridge, MA) in RPMI 1640 containing 1% Nutridoma (Boehringer Mannheim, Mannheim, Germany) as described (19, 20). When indicated, 100 ng/ml of the chemokines RANTES, MIP-1α, or MIP-3β (kind gift of P. Allavena, Istituto Farmacologico M. Negri, Milan, Italy) (24), 100 μg/ml amiloride (Sigma), 10 μg/ml aprotinine (Bayer Trasylol, Zurich, Switzerland), or 0.1 mM plasmin-free plasminogen (Sigma) were added at the beginning of the transmigration assay (19). In other experiments, DC were incubated with the anti-CD18/β₂ integrin mAb TS1.18 (American Type Culture Collection, Manassas, VA) or with blocking or nonblocking anti-uPAR Abs (R3 or R2, respectively; 5 μg/ml). Alternatively, DC were treated with 100 ng/ml TNF-α during the transmigration assay or exposed to TNF-α for 4, 12, or 24 h before transmigration (19). After 8 h of incubation (kinetics experiments show that after 8 h migration reached a plateau with a random migration of 10% or less (Table I)), migrated cells were recovered from the lower compartment and lysed with 100 mM Tris buffer containing 0.1% Triton X-100. The radioactivity of the samples was measured in a gamma-counter (Packard, Sterling, VA). Reverse transmigration was performed following the method described by Randolph et al. (24) partially modified according to D’Amico et al. (25) in the absence or presence of the R2 or R3 mAbs. Briefly, ECM was prepared by growing a monolayer of HUVEC on an upper polycarbonate filter (5-μm pores). After 5 days the monolayer was stripped by a 30-s treatment with a 20-mmol/L NH₄OH solution + 0.5% Triton X-100 (Sigma). The lower polycarbonate filter was placed upside down, coated by a monolayer of the same line of HUVEC, and mounted in a Boyden chamber. ⁵¹Cr-labeled DC were seeded on the upper compartment and the chamber was placed in an incubator at 37°C. After 4 h the migrated cells were recovered from the lower compartment, lysed, and counted in a gamma-counter. The relative percentage was calculated by comparing this value to that obtained by the lysis and the counting of the total original input. Statistical analysis was performed using the Student t test.

DC cultured with GM-CSF alone for 7 days or with TNF-α during the last 48 h were used as stimulators for allogenic or tetanus-toxoid- (TT; gift of S. Burastero, HSR-Dibit, Milan, Italy) specific T lymphocytes as previouly reported (20). About 5 × 10⁴ allogeneic or TT-specific autologous T cells were added to irradiated (4000 rad) DC (from 10³ to 2 × 10⁴) in 96-well round-bottom microwell plates. For TT presentation, DC were loaded with the Ag for 24 h before the addition of specific T cells (20). After 3 days for the allogeneic reaction (MLR) or after 48 h for TT-specific response, cells were pulsed with 1 μCi of [³H]TdR (NEN-DuPont) per well for the last 18 h of culture, harvested, and counted in a beta-counter (Packard). Tests were conducted in triplicate and results are expressed as mean cpm ± S.D. [³H]TdR uptake by stimulatory DC or responder T cells alone was <400 cpm (data not shown).

As DC derived from CD14⁺CD34⁺ precursors upon culture with GM-CSF are capable of transendothelial migration (20), we addressed the question of whether they can also actively migrate through the subendothelial matrix. To this purpose, transmigration was performed in a two-chamber Transwell system after coating of the porous membrane with Matrigel, as an equivalent for ECM, in the absence of serum. Fig. 1 shows that a sizeable fraction of immature DC (cultured in GM-CSF for 5 days) can efficiently invade Matrigel in the absence of added chemokines, suggesting an alternative mechanism capable of driving transmigration. A similar effect has been described for the uPA/uPAR system in leukocytes, in which uPA/uPAR interaction activates the receptor which acts as a chemoattractant (18). Interestingly, we found that migration through Matrigel was significantly impaired when immature DC (Fig. 1) were pretreated with the anti-uPAR R3 mAb directed against the uPA binding site of uPAR, but it was not impaired with the nonblocking R2 mAb. A similar trend was observed using CD14⁺CD34⁺ DC precursors (data not shown).

The inhibiting effect of R3 mAb was not detectable when fibronectin was used as a substrate (percentage of migrating DC was 35 ± 3 in the absence and 30 ± 4 in the presence of R3 mAb; not shown in tables or figures), in keeping with others (24). To further define the involvement of uPA/uPAR, Matrigel invasion assay was performed upon addition of plasminogen, the substrate of the uPA/uPAR system, in the presence of aprotinin, a plasmin inhibitor, or plus amiloride, a urokinase inhibitor (19). Plasminogen increased the fraction of invading immature DC (Fig. 1). It is worth noting that R3 mAb, amiloride, and aprotinin, at variance with R2 mAb, strongly inhibited migration of immature DC, further supporting that these cells can invade ECM using, among others, the uPA/uPAR system.

We found that immature DC can penetrate endothelial cell monolayers using the uPA/uPAR system, as shown by the fact that the blocking R3 mAb can reduce their transmigration through HUVEC monolayers (Fig. 2,A). However, the involvement of such a system is less evident than using Matrigel, possibly because in vitro-cultured endothelial cells usually produce low amounts of subendothelial matrix, thus masking the real contribution of the uPA/uPAR system in a direct transmigration assay. Moreover, a wide number of adhesion molecules, both for endothelial cells and matrix ligands, are operating during transmigration. Because there is evidence that chemotaxis stimulated by uPAR is mediated, at least in part, by β₂ integrin activation (26, 27, 28), the relative role of CD18/β₂ integrin and uPA/uPAR in transendothelial migration by DC has been investigated. As depicted in Fig. 2,A, while both R3 and an anti-CD18 mAb can decrease the percentage of migrating DC, no additive effect is observed using the two mAbs. This might be due to the fact that our DC population, as we reported for their precursors (20), preferentially uses other adhesion systems involved in transmigration rather than CD18/ICAM-1. Because immature DC have been reported to migrate across endothelial monolayers, also in the basal-apical direction (24, 25), we investigated the role of uPA/uPAR in a reverse transmigration assay. Fig. 2,B shows that immature DC can efficiently reverse transmigrate across HUVEC. Interestingly, the blocking anti-uPAR R3 mAb, at variance with the nonblocking R2 mAb, could significantly inhibit reverse transmigration (Fig. 2 B), suggesting that the uPA/uPAR system contributes to the motility and recirculating potential of immature DC.

Recently, it has been reported that immature DC respond to selected CC chemokines, including RANTES and MIP-1α, enhancing their chemotaxis and transendothelial migration (29, 30, 31). Thus, we addressed the questions of whether and how the use of the uPA/uPAR system by immature DC is regulated when matrix invasion is tested in the presence of RANTES or MIP-1α. As shown in Fig. 3,A, immature DC derived from CD14⁺CD34⁺ precursors are responsive to MIP-1α, enhancing their transmigration through Matrigel by 50%. A lesser increase in transmigration (10%) was found in response to RANTES (Fig. 3,A). Interestingly, MIP-1α was also capable of preventing the decreased matrix invasion due to the inhibitory effect of the anti-uPAR R3 mAb; however, the inhibitory effect of the Ab was not fully reverted by the MIP-1α (Fig. 3,A), suggesting that the chemokine stimulates DC migration via alternative/additional pathways not influenced by uPAR and therefore not blocked by R3 mAb. In the presence of plasminogen, whereby the uPA/uPAR system is enhanced, the effect of MIP-1α was much less evident, and that of RANTES was absent (Fig. 3 B).

It is known that upon maturation DC lose their response to chemokines and change their migratory pattern (30, 31). Thus, we studied the effects of TNF-α, which contributes to DC differentiation (4, 5, 9, 21), on DC derived from CD14⁺CD34⁺ precursors. First, we found that these cells undergo maturation upon exposure to TNF-α for 48 h after 5 days of culture in GM-CSF; indeed, they up-regulate costimulatory molecules, such as CD86/B7.2, HLA-DR, and to a lesser extent CD80/B7.1 and CD54/ICAM-1 (Fig. 4,A). Expression of other surface molecules, such as β₁, β₂, or β₃ integrins, was not significantly altered upon maturation (data not shown). Because these phenotypic changes represent a hallmark of differentiation toward fully mature DC (4, 11, 12), we analyzed the immunostimulatory capacity of DC derived from GM-CSF cell culture (5 days) with or without TNF-α for an additional 48 h. To this aim, the two DC populations were used as stimulators to TT-specific T cell lines. As shown in Fig. 4,B, TNF-α-treated DC pulsed with TT increased their capacity for stimulating the proliferation of TT-specific T lymphocytes. As shown in Fig. 4,C, this phenomenon was even more evident in allogenic stimulation (MLR). Thus, DC derived from CD14⁺CD34⁺ precursors respond to maturative stimuli and potentiate their APC function, a response similar to those of other DC populations. Interestingly, upon maturation induced by TNF-α, DC significantly reduced their capacity for invading ECM in vitro, both in the absence and in the presence of plasminogen (compare Fig. 5,A with Fig. 1; p < 0.05). Migration was decreased by DC treatment with amiloride and aprotinin (although to a lesser extent than in immature DC; compare Fig. 5,A to Fig. 1), but not with the R3 anti-uPAR Ab (Fig. 5 A). These data suggest that the uPA/uPAR system is less active in mature DC than in immature DC.

Because it has been reported that the sensitivity of DC shifts from MIP-1α to MIP-3β upon maturation (29, 30, 31), we addressed the question of whether this chemokine could stimulate migration of mature DC through Matrigel. Fig. 5,B shows that mature DC derived from CD34⁺CD14⁻ precursors respond to MIP-3β and enhance their capability for invading ECM. Moreover, R3 mAb does not interfere with MIP-3β-driven migration of mature DC through Matrigel (Fig. 5,B). However, it has to be noted that uPAR is less functional in mature DC (see above and Fig. 5 A).

In another series of experiments, migration through Matrigel was tested using immature DC treated with TNF-α for less than 2 days. A short exposure (4 h) of DC to TNF-α enhanced Matrigel invasion (at variance with treatment for 12 h or 24 h) without significantly affecting uPAR expression (data not shown). The presence of the cytokine during migration assay did not alter the degree or rate of transmigration (Fig. 6). This suggests that DC can move in response to TNF-α when immature and then become less sensitive and decrease their motility upon exposure to maturative stimuli.

We asked whether the decreased usage of the uPA/uPAR system in matrix invasion was due, at least in part, to changes in uPAR expression and/or surface distribution in mature vs immature DC. Fig. 7A shows that uPAR is detectable by indirect immunofluorescence on DC derived from 7-day cultures with GM-CSF; treatment with TNF-α for the last 48 h decreased its intensity of expression (Fig. 7B). Apparently, this was not due to the shedding of uPAR; indeed, the amount of the receptor detectable by ELISA in the supernatants of DC cultured in GM-CSF did not change after exposure to TNF-α (Fig. 7,C). As expected, treatment of immature or mature DC with PI-PLC, an enzyme that cleaves phosphoinositide bridges whereby uPAR is linked to the cell surface (32), led to the release of detectable amounts of uPAR (Fig. 7 C).

A possible explanation for the different usage of uPAR in immature vs mature DC might be represented by a different surface distribution of this receptor. Confocal microscopy performed on DC, before and after exposure to TNF-α, showed that in GM-CSF-cultured DC uPAR is brightly expressed and that it is enriched on filopodia (Fig. 8,A), cytoplasmic structures involved in cell migration (2, 17). However, TNF-α-treated DC uPAR staining is dull and detectable along the plasma membrane only (Fig. 8,B). Conversely, MHC class II molecules are still detectable on filopodia of mature DC (compare Fig. 8, C and D). Altogether, these data suggest that DC down-regulate the uPA/uPAR system and become less motile upon exposure to maturative stimuli such as TNF-α.

In this paper we provide evidence that the uPA/uPAR system plays an important role in the dynamic interactions between DC and ECM. That uPA/uPAR is involved in driving leukocyte chemotaxis and matrix degradation is well substantiated (19, 33, 34, 35), although other enzymatic systems can be operating, as reported for T lymphocytes (36, 37). In this paper we show that the uPA/uPAR system is active both in CD34⁺CD14⁺ circulating precursors and in immature DC derived from these cells, contributing to their motility through ECM. Upon maturation, uPAR is down-regulated and less functional on DC, thus slowing down matrix invasion.

DC precursors migrate from bone marrow toward peripheral tissues using multiple adhesion systems that allow them to stick and detach to vascular endothelium and subendothelial matrix to reach their final destination (11, 12). From this viewpoint uPA/uPAR might represent a useful tool for shifting from transendothelial to subendothelial migration; indeed, upon binding to uPAR, pro-uPA is changed into its active form that converts plasminogen into plasmin which, in turn, can degrade fibrin and other proteins, thus driving cell migration through the extracellular milieu (18, 33, 34, 35). The activation of this enzymatic system may contribute to completing the cell polarization that begins when DC migrate between endothelial cells, maintaining a directional movement through subendothelial matrix toward the final destination. Support for this hypothesis comes from our observation that DC can use the uPA/uPAR system to migrate across HUVEC monolayers. Interestingly, we found that uPA/uPAR is involved in the reverse transmigration of immature DC as well, thus possibly contributing to their recirculating potential. The finding that immature DC respond to MIP-1α, enhancing their ability to invade ECM, supports the idea that DC are selectively driven at the site of inflammation, possibly where the Ag is present at high concentrations, which is in agreement with other reports (29, 30, 31). Interestingly, MIP-1α was also capable of preventing the decreased matrix invasion observed by blocking uPAR, suggesting that the uPA/uPAR system and MIP-1α cooperate in driving immature DC migration through the subendothelial matrix.

When DC undergo maturation, they usually down-regulate most of surface chemokine receptors, shifting their sensitivity toward other chemokines such as MIP-3β (30, 31). In keeping with this, we found that mature DC derived from CD34⁺CD14⁺ precursors do not respond to MIP-1α, acquiring the ability to migrate in response to MIP-3β. Moreover, uPAR is down-regulated at the cell surface and the uPA/uPAR sytem is less active in mature DC. From this viewpoint, our data are in line with the hypothesis that leukocytes can respond sequentially to chemokines in a “multistep navigation” mode (38).

However, it is largely accepted that, after capture of the Ag, mature DC also travel to the lymphoid tissues where they encounter T lymphocytes (2, 10, 12). DC are also detectable in the afferent lymph and in the absence of pathogens or tissue injuries (12). Thus, it is conceivable that immature DC can move actively in the inflamed tissue until they reach the site of pathogen entry, where they undergo complete maturation. In this timeframe, their motility might be modulated by microenvironmental stimuli, such as chemokines and/or cytokines, released during inflammation. Such factors include TNF-α, which is involved in DC maturation.

TNF-α has been reported to promote DC migration in mice (39, 40). However, it should be noted that when these phenomena are evaluated in vivo, the overall effect of the cytokine on both migrating cells and neighboring tissues, including endothelium, are considered. The majority of the reported effects of TNF-α on cell migration are referred to its action on endothelial cells; indeed, this cytokine can increase vascular permeability, induce proinflammatory cytokine production by endothelial cells themselves, and regulate the expression of adhesion and junctional molecules (22, 41, 42, 43). We found that TNF-α up-regulates costimulatory molecules on immature DC and enhances Ag presentation; in turn, it down-regulates the expression and function of uPA/uPAR, thus contributing to the decreased motility of DC during maturation, in line with data from the literature (30, 31). However, when used as a short-term challenge, TNF-α can potentiate matrix invasion. It is tempting to speculate that immature DC actively respond to different chemotactic stimuli until they undergo maturation and then become less motile and remain in the tissue as long as the pathogen is present to complete Ag capture. This is in line with reported data showing that mature DC are less motile than immature DC (30, 31).

After capturing the Ag, mature DC should reacquire the ability of traverse the endothelium in the abluminal-luminal direction to reach lymph nodes and initiate T cell-dependent immune responses. Evidence has been reported for the so-called reverse transmigration by either immature or mature DC (25, 29, 30). During this process mature DC use adhesion/migratory pathways other than those used by immature DC and their precursors (29, 30). This might explain our observation that the CD18/β₂ integrin is poorly functional in immature DC and does not seem to cooperate with the uPA/uPAR system in driving transendothelial migration. Along this line we found that while transendothelial migration of CD14⁺CD34⁺ precursors seems to be preferentially driven by a CD31 aptotactic gradient (20), mature DC shift to VCAM-1-driven transmigration (data not shown), possibly favoring the reverse transmigration. The down-regulation of the uPA/uPAR system might be useful to enhance this process, thus allowing the detachment from matrix and permitting mature DC recirculation.

We thank C. Rugarli and F. Blasi for support, A. Rubartelli for critical reading of the manuscript, and G. Consogno for skillful technical assistance. We are also grateful to R. Accolla (Advanced Biotechnology Center, Genoa, Italy), P. Allavena (Istituto Framacologico M. Negri, Milan, Italy), and E. Rønne and G. Høyer-Hanse (K. Danø Finsen Laboratory, Copenhagen, Denmark) for sharing their reagents.