CD14⁺CD34⁺ Peripheral Blood Mononuclear Cells Migrate Across Endothelium and Give Rise to Immunostimulatory Dendritic Cells¹

Monocytes (Mo)³ derived from bone marrow precursors circulate in the blood and eventually traverse vascular endothelial lining to enter tissues, where they differentiate into macrophages or dendritic cells (DC) after inflammatory or immunologic stimuli (1, 2). Transmigration of mature Mo, expressing CD14, is a multistep process that follows the well-known stepwise model of rolling, mediated by selectins (3), cell adhesion, involving integrins (4, 5), and migration (3, 6). The last step is likely to depend on several junctional and nonjunctional molecules, expressed both by leukocytes and endothelial cells; among them, the platelet endothelial cell adhesion molecule-1 (PECAM1/CD31), a member of the Ig family able to support both homophilic and heterophilic adhesion, has been reported to play an important role (7, 8, 9).

Transendothelial migration of hemopoietic progenitor cells occurs during mobilization from bone marrow in response to cytokines and during homing of circulating progenitors (10). CD34⁺ leukocyte precursors express several adhesion molecules, such as selectin and integrins, which are known to mediate the different steps of extravasation (3, 11, 12). The interaction between the CD34 molecule itself and ligands of the selectin family is thought to be important for the initial binding of hemopoietic progenitors to endothelial cells (13). Likewise, members of the integrin family, such as the α₄β₁ and the α_Lβ₂ integrins, proved to be important for progenitor cell trafficking both in vitro and in vivo (14, 15). Recently, PECAM-1/CD31 has been reported to enhance the adhesivity of α₄β₁ integrin expressed by hemopoietic progenitor cells (16).

Mobilization of CD34⁺ cells from bone marrow is a rare event in the adult life, and it is conceivably due to changes in the expression, or affinity to their ligand, of different adhesion molecules during differentiation (17, 18, 19). Indeed, the adhesion molecules expressed on those progenitor cells that are committed to leave the bone marrow, should acquire a higher affinity for counterreceptors expressed by endothelial cells than by stromal cells and extracellular matrix (20, 21, 22). Likewise, the ability to localize to peripheral tissues depends, in mature leukocytes, on a sequence of molecular events that are tightly connected to each other (3, 4, 5, 6). These events can also be regulated by local conditions, such as cytokine production during inflammation, Ag sensitization in the immune response, and secretion of chemokines (23, 24).

We show that a subset of CD14⁺ PBMC, coexpressing the CD34 precursor marker, is able to migrate across endothelial cells and differentiate into immunostimulatory DC. Migration is apparently mediated by the PECAM-1 molecule through a haptotactic gradient. Once differentiated, DC can use both PECAM-1 and vascular cell adhesion molecule-1 (VCAM-1) for transmigration.

The purified, or the FITC- or phycoerythrin (PE)-conjugated, anti-CD14, anti-CD80 (B7.1), anti-CD86 (B7.2), and the anti-surface Igs (sIg) mAbs were purchased from Becton Dickinson (Sunnyvale, CA); the anti-ICAM-1 (CD54; 84H10), the anti-CD83 (HB15a), and the PE-conjugated anti-CD34 (HPCA-1) mAb were purchased from Immunotech (Luminy, Marseille, France); the anti-CD106/VCAM-1 (BBA6) mAb were purchased from British Biotechnology (Oxford, U.K.). The anti-HLA-DR mAb (D1.12) was kindly provided by R. Accolla (ABC, Genoa, Italy), the anti-α₄ integrin (CD49d) mAb (HP2/1) was provided by F. Sanchez-Madrid (Hospital de la Princesa, Madrid, Spain), the anti-CD36 mAb (NL07) was a kind gift of M. Alessio (Dibit, Milan, Italy), and the anti-human mannose receptor (pan1) was a gift of A. Mantovani (Istituto Farmacologico M. Negri, Milan, Italy). The anti-CD11a/α_L integrin mAb 70H12 and the anti-CD31/PECAM-1 mAb M89D3 were obtained in our laboratory (25, 26), and the anti-CD1a (OKT6) mAb was obtained from the American Type Culture Collection (Rockville, MD). These mAbs were purified from ascites fluids by affinity chromatography. All the purified Abs were used at concentrations of 5 μg/ml in functional studies and at 1 μg/ml in immunofluorescence. When indicated, to avoid any unspecific effect due to FcγR binding, mAbs were pepsin digested. F(ab′)₂ fragments were prepared according to the method of Parham (27). PE-conjugated goat anti-isotype mouse Igs (GAM) was from Zymed Laboratories Inc. (South San Francisco, CA).

Peripheral blood CD14⁺ cells were isolated from healthy donors (buffy coats, kindly provided by the Blood Transfusion Department of our institute), after density gradient centrifugation, according to the method of Sallustio and Lanzavecchia (28). CD34⁺ and CD34⁻ cell subsets were fractionated using immunomagnetic beads (Dynal, Milan, Italy) according to the manufacturer’s procedure. Purified T cells were obtained after two rounds of plastic adherence followed by immunodepletion of CD14⁺ and HLA-DR⁺ cells. When indicated, cells were cultured in RPMI 1640 medium supplemented with 2 mM l-glutamine, 100 IU/ml penicillin, 100 μg/ml streptomycin (Biochrom, Berlin, Germany), 10% heat-inactivated FCS (PAA Labour, Linz, Austria), and 20 ng/ml recombinant granulocyte-macrophage CSF (GM-CSF) as described (29). Media were endotoxin free as shown by the Limulus lysate colorimetric assay (BioWhittaker, Verviers, Belgium).

Cells were fixed for transmission electron microscopy (EM) in 2.5% glutaraldehyde for 20 min at 4°C and washed three times in phosphate buffer. Postfixation was performed in 1% osmium tetroxide. Samples were then dehydrated in ethyl alcohol and propylene oxide and embedded in Epon-Araldite resin (Fluka (Sigma-Aldrich), Milan, Italy). Thin sections (80 nm) were then stained with uranyl acetate and lead citrate and analyzed under a Zeiss CEM 902 electron microscope.

PECAM-1- and ICAM-1-stable transfectants were obtained in NIH/3T3 murine fibroblasts as described (9). CD31/PECAM-1 was subcloned into pc/DNAI/Neo expression vector (Invitrogen, San Diego, CA) at the XhoI (5′) NsiI (3′) sites from the original pGEM7 vector (CD31/PECAM-1/pGEM7 kindly provided by Peter Newmann, Blood Center of Southeastern Wisconsin, Milwaukee, WI). CD54/ICAM-1 was subcloned into pcDNAI/Neo at the XhoI site from the original pRc/CMV vector obtained from T. Springer (Harvard University, Boston, MA). Transfection was performed by calcium phosphate-DNA coprecipitation, and stable transfectants were selected by addition of the neomycin analogue G418 to a final concentration of 0.8 mg/ml. NIH/3T3 cells stably expressing a truncated form of CD31/PECAM-1 lacking the cytoplasmic tail (CD31/PECAM-1-Δcyto) were obtained as described (9). The VCAM-1 transfectants in Chinese hamster ovary (CHO) cells were kindly provided by A. Dobrina (Department of Physiology and Pathology, University of Trieste, Trieste, Italy). Mock NIH/3T3 and CHO cells, transfected with the vector alone, were used as controls.

For immunofluorescence, 10⁵ cells/sample were fixed with 3% paraformaldehyde in PBS (30 min of incubation) and stained with the various mAbs followed by the anti-isotype-specific PE-conjugated GAM. Double staining was performed using the FITC- or PE-conjugated anti-CD14 and the PE-conjugated anti-CD34 mAbs or the FITC-anti-sIg antiserum. After washing, samples were run on a FACStar^Plus equipped with an argon ion laser (Becton Dickinson, Mountain View, CA), gated to exclude cell debris and nonviable cells. At least 5000 events/sample were analyzed. Results are expressed as mean red fluorescence intensity vs mean green fluorescence intensity (arbitrary units, a.u.) or mean log fluorescence intensity (a.u.).

Ig heavy chain gene rearrangement was analyzed by PCR in CD34⁺ and CD34⁻ cell subsets and in cell populations derived from the above mentioned subsets after culture with GM-CSF. Briefly, cell were digested overnight at 37°C in 100 μl of lysis buffer containing proteinase K (Sigma Chemical Co., St. Louis, MO; 200 μg/ml); after 10 min of heating at 95°C, 3 μl of each sample were used as template in 50 μl of a PCR mixture containing 50 mM KCl, 10 mM Tris-HCl (pH 8.3), 50 pmol of each primer, 200 μM concentrations of each dNTP, 1.25 U of Taq polymerase, and 1.5 mM MgCl₂. A seminested reaction was applied for amplification of the Ig heavy chain gene, using the framework 3 V-region primer in conjunction with nested primers directed to the J regions (LJH and VLJH). Twenty microliters of the reaction products were analyzed by electrophoresis on a 3% agarose gel.

HUVEC were isolated and cultured as described (30). Cells were used within four passages. Adhesion or transmigration of radiolabeled (⁵¹Cr, DuPont NEN, Boston, MA), unsorted CD14⁺ or CD14⁺CD34⁺, and CD14⁺CD34⁻ sorted cells, through HUVEC, ICAM-1, VCAM-1, PECAM1, or PECAM-1-Δcyto transfectants, was performed with Transwell cell culture chambers (polycarbonate filters, 3-μm pore size, Costar, Cambridge, MA) as described (9). In some samples, cells were preincubated for 15 min at 4°C with saturating amounts (5 μg/ml) of the anti-CD31 (M89D3), anti-CD11a (70H12), or anti-CD49d (HP/1) F(ab′)₂ and washed twice, before the onset of the transmigration assay. At different time points (4 or 12 h), 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). Results are expressed as percent migrating cells, calculated as described (9), i.e., (cpm of lysates from migrated cells/cpm of lysates from cells of the total input) × 100. In some experiments, both migrated and nonmigrated cells were recovered and cultured for 3, 7, or 14 days in the presence of GM-CSF and tested in the transmigration assay.

To test the immunostimulatory potential of migrated and nonmigrated cells cultured with GM-CSF, after 7 days of culture cell derived from either CD14⁺CD34⁺ or CD14⁺CD34⁻, cell precursors were used as stimulators for allogeneic or tetanus toxoid (TT, kindly provided by S. Burastero, HSR-Dibit, Milan, Italy)-specific T lymphocytes at different T:DC ratios. TT-specific T cell lines were obtained by stimulation of syngeneic PBMC with TT (10 μg/ml), Percoll gradient separation of T cell blasts after 7 days, culture in T cell growth factor (Lymphocult, Biotest Diagnostics Inc., Dreieich, Germany), and weekly restimulation with TT. About 10⁵ allogeneic or 2 × 10⁴ TT-specific syngeneic T cells, washed, and maintained without Lymphocult for 48 h, were added to irradiated (4000 rad) DC (from 2 × 10⁴ to 10³) in 96-well round bottom microwell plates in RPM1 1640 culture medium supplemented with 10% heat-inactivated FCS and antibiotics as above. For TT presentation, DC were loaded with TT for 12 h before addition of specific T cells. After 3 days, for the allogeneic reaction (MLR), or 48 h for TT-specific response, cells were pulsed with 1 μCi of [³H]TdR (DuPont NEN) 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 ± SD [³H]TdR uptake by stimulatory DC or responder allogenic T cells alone was <100 cpm; [³H]TdR uptake by TT-specific T cell lines alone was <1000 cpm (not shown).

CD14⁺ cells collected from peripheral blood of healthy donors usually represent mature Mo; however, we found that a small proportion of these cells, ranging from 5 to 10%, coexpress the CD34 molecule (Fig. 1,A shows a representative experiment of 10), which is a marker of hemopoietic progenitors (31). CD34 expression in this cell population was also confirmed by Western blot (not shown). Since CD34 is usually lost during leukocyte maturation, the CD14⁺CD34⁺ cell population might represent an intermediate stage of differentiation toward mature Mo. This subset of double positive cells can transmigrate across endothelial cell monolayers: indeed, when purified CD14⁺ cells were used in a transmigration assay, almost all the cells (95%) recovered from the lower chamber after 12 h were CD34⁺ (Fig. 1,B). This transendothelial migration apparently occurs in the absence of chemokines, because we did not find detectable levels of monocyte chemoattractant protein 1 (MCP1), MCP2, or IL-8 in the supernatants of HUVEC after 3 days of culture (not shown). The migration ability of sorted CD14⁺CD34⁺ cells was compared with that of CD14⁺CD34⁻ cells. As shown in Figure 1,C, only the former subset transmigrated without the addition of exogenous chemokines, whereas the latter could migrate only after addition of 50 nM MCP1 (not shown), in keeping with reported data (32). Along this line, Table I shows that CD14⁺CD34⁻ cells displayed a higher tendency to adhere to HUVEC, compared with CD14⁺CD34⁺ cells, and this adhesion is blocked by anti-CD11a mAbs.

Both peripheral blood and bone marrow CD34⁺ precursors express β₁ or β₂ integrins and PECAM-1 (15, 16). We found (Fig. 2 shows a representative experiment) that the CD14⁺CD34⁺ cell subset expresses higher levels of PECAM-1 (mean fluorescence intensity (MFI) ± SD from 10 independent experiments, 110 ± 4 a.u.) than the CD14⁺CD34⁻ counterpart (MFI 28 ± 3 a.u.). Likewise, the CD34⁺ cell population was brightly positive for the α₄ integrin (MFI 80 ± 5 a.u.), whereas the CD34⁻ cells were weakly stained with the anti-α₄ integrin mAb (MFI 25 ± 3 a.u.). Conversely, the α_L integrin was highly expressed on CD14⁺CD34⁻ cells (MFI 66 ± 7 a.u.), but not on the CD14⁺CD34⁺ cell subset (MFI 22 ± 5 a.u.).

To investigate whether the higher expression of PECAM-1 and α₄ integrin by CD14⁺CD34⁺ cells was responsible for transendothelial migration, transmigration assays were performed using purified CD14⁺CD34⁺ cells and NIH/3T3 cell monolayers stably expressing PECAM-1, VCAM-1, or ICAM-1. Both PECAM-1 (Fig. 3,A) and, to a lesser extent, VCAM-1 (Fig. 3,B) transfectants proved to support CD14⁺CD34⁺ cell migration, at variance with ICAM-1⁺ cell monolayers (Fig. 3,C). Migrated cells were detectable in the lower Transwell compartment after 4 h (not shown), reaching the maximum after 12 h (Fig. 3), in particular when transmigration was evaluated across PECAM-1-transfected cell monolayers (Fig. 3,A). Pretreatment of migrating cells with the F(ab′)₂ of mAbs directed against either PECAM-1 (Fig. 3,A) or α₄ integrin (Fig. 3 B) prevented cell migration. Likewise, preincubation of PECAM1⁺ transfectants with the F(ab′)₂ of M89D3 mAb abrogated cell migration (not shown), suggesting that a CD31/PECAM-1-mediated homophilic adhesion was operating. This was also supported by the observation that the αvβ3 integrin, another reported ligand for PECAM-1 (33), was virtually absent from both CD14⁺CD34⁺ cells and PECAM-1⁺ transfectants (not shown). Background migration across mock-transfected NIH/3T3 or CHO cells was always <10% (9).

We have recently reported that, unlike ICAM-1, PECAM-1 can drive transmigration of a subset of lymphocytes by creating a haptotactic gradient because of preferential intercellular localization of the molecule in endothelial and transfected fibroblast monolayers (8, 9). To investigate whether a similar mechanism is involved in CD14⁺CD34⁺ cell migration, we analyzed their capacity of migrating across monolayers of NIH/3T3 cells transfected with the PECAM-1-Δcyto deletion mutant. This mutant lacks the cytoplasmic tail and displays a reduced tendency to localize at the intercellular junctions, thus being homogeneously distributed along the plasma membrane (9). Transmigration was reduced by 80% using monolayers of NIH/3T3 cells transfected with the PECAM-1-Δcyto mutant (Fig. 3D), indicating that a PECAM-1 haptotactic gradient can drive transmigration of CD14⁺CD34⁺ cells.

Migrated CD14⁺CD34⁺ cells (Fig. 4, a and b) expressed the CD80 Ag (Fig. 4,c), which is a DC hallmark (34), beside CD86 (not shown), and very low levels of CD36 (Fig. 4,d), a marker of mature Mo or macrophages (35). This would suggest that CD14⁺CD34⁺ cells might represent a subset of precursors capable of differentiating into DC, rather than Mo. Indeed, after 7 days of culture in the presence of GM-CSF, migrated cells lost the CD14 (Figs. 4 and 4f), the CD34 (Fig. 4,g), and the CD36 (Fig. 4,i) molecules and up-regulated HLA-DR (Fig. 4,l), remaining positive for both CD80 (Fig. 4,h) and CD86 (not shown) costimulatory (34) molecules; furthermore, they were not stained with dyes for the nonspecific esterases (not shown). On the contrary, nonmigrating CD14⁺CD34⁻CD36⁺ cells (Fig. 4, m–q), cultured in GM-CSF, gave rise to CD14⁺CD34⁻CD80⁻CD36⁺ cells (Fig. 4, r–v), positive for the nonspecific esterase staining (not shown), which conceivably represent macrophages. However, activated B cells have been reported to share some phenotypic features of monocyte/macrophages (36); indeed, we found that about 2% of purified peripheral blood CD14⁺ leukocytes coexpressed sIg (Fig. 5,A), at variance with CD14⁺CD34⁺ migrated cells, which were sIg negative (not shown). Thus, we analyzed for Ig heavy chain gene rearrangement both CD14⁺CD34⁺ and CD14⁺CD34⁻ subsets, before and after culture with GM-CSF. The electrophoresis of PCR products revealed the presence of a smear, suggestive of polyclonal rearrangement, in the CD14⁺CD34⁻ population (Fig. 5,B, lane 2), while samples run in the remaining lanes, from the same subset after culture in GM-CSF (lane 4) or from CD14⁺CD34⁺ cells, before (lane 3) and after culture (lane 5), were essentially devoid of amplified DNA fragments. That cells derived from migrating CD14⁺CD34⁺ leukocytes were actually DC was confirmed by electron microscopy. Figure 6,A shows that CD14⁺CD34⁺ cells have few cytoplasmic projections and high nucleo-cytoplasmic ratio, are roundish, and are larger than CD14⁺CD34⁻ leukocytes. Nuclei are smooth-countered with finely dispersed chromatin and usually one small nucleolus is usually present; the cytoplasmic organelles are mainly represented by mitochondria, rare rough reticulum cisternae, and small Golgi complexes (Fig. 6,A). CD14⁺CD34⁻ cells show, in turn, slender cytoplasmic projections, irregular nuclei with marginated chromatin, and abundant cytoplasm containing some electron-dense lysosome-like granules (Fig. 6,B). Cells derived from CD14⁺CD34⁺ precursors display many dendritic-like cytoplasmic protrusions (filopodia) and a cytoplasm poor of granules and lysosomes (Fig. 6,C), thus being bona fide DC, whereas cells derived from CD14⁺CD34⁻ leukocytes have ultrastructural characteristics consistent with monocyte/macrophage lineage such as short and thick membrane projections and a cytoplasm filled with dense lysosome-like granules (Fig. 6,D). We failed to demonstrate Birbeck’s granules (usually found in Langerhans cells) in our DC; however, unlike cells from CD14⁺CD34⁻ leukocytes (not shown), DC acquired CD1a (Fig. 7,d) in ∼50% of the donors examined, and CD83, which is a DC marker (38), in all of the 10 donors analyzed (Fig. 7,e) and expressed low levels of mannose receptor (Fig. 7 f), as reported in mature DC (39).

Because activation of naive T lymphocytes in response to allogenic Ags is one of the functional property of DC (39, 40), the two cell populations derived from migrating CD14⁺CD34⁺ or nonmigrated CD14⁺CD34⁻ precursors were tested for their ability to induce allogenic or TT-specific T cell proliferation. As shown in Figure 8, DC obtained from CD14⁺CD34⁺ migrating precursors could stimulate both allogenic (Fig. 8,A) and syngenic TT-specific T cell proliferation (Fig. 8,B). On the contrary, CD14⁺ cells derived from nonmigrated peripheral blood CD14⁺CD34⁻ Mo were poor stimulators in MLR (Fig. 8,A), and they were less efficient than DC in the stimulation of TT-specific T cell lines (Fig. 8 B).

One of the predictable characteristics of DC with potential Ag presenting capacity is the ability to become veiled cells and recirculate (2, 40). Thus, we addressed the question of whether DC derived from CD14⁺CD34⁺ cells displayed any migratory activity. To this aim, transmigration assays were performed using both HUVEC, and the various transfectants expressing either ICAM-1, VCAM-1, or PECAM-1. We found that DC obtained from CD14⁺CD34⁺ precursors displayed a higher migratory capacity through HUVEC monolayers than cells derived from the CD14⁺CD34⁻ counterpart (Fig. 9). This was confirmed by the finding that DC obtained from CD14⁺CD34⁺ precursors could efficiently migrate across PECAM-1 and VCAM-1 transfectants as well, at variance with cells from CD14⁺CD34⁻ Mo (Fig. 9). Moreover, GM-CSF-cultured CD14⁺CD34⁻ cells could adhere to ICAM-1-transfected cell monolayers more efficiently (35 ± 4%) than the CD14⁺CD34⁺-derived DC (10 ± 3%), whereas adhesion of CD34⁻ or CD34⁺-derived populations to PECAM-1 or VCAM-1 transfectants was superimposable (not shown). These data support the hypothesis that DC derived from CD14⁺CD34⁺ precursors show peculiar migratory properties.

Herein we describe a subset of CD14⁺ circulating leukocytes, coexpressing the CD34 progenitor marker, which are able to migrate across endothelial monolayers, unlike the CD14⁺CD34⁻ counterpart. Moreover, CD14⁺CD34⁺ cells differentiate into immunostimulatory DC on culture with GM-CSF.

This population possibly represents an intermediate stage of differentiation from bone marrow precursors to mature Mo or DC. Although DC maturation has been described as a single pathway, the heterogeneity of DC within peripheral blood points to the existence of two separate pathways in DC development, one committed directly from an early precursor and the other from a CD14⁺ monocyte-like stage (41). Our data would indicate that such a bipotent precursor is present in peripheral blood, in keeping with a recent interpretation of DC origin and maturation (41, 42). Unlike mature CD14⁺ leukocytes, the CD14⁺CD34⁺ subset shows high expression of PECAM-1 and α₄ integrin, being dull positive for all the other integrins. Both β₁ and β₂ integrins are expressed in lower levels by circulating CD34⁺ cells, compared with noncirculating bone marrow CD34⁺ progenitors (43). It is conceivable that the preferential expression of certain molecules accounts for the different adhesive/migratory behavior of the two cell populations. Indeed, CD14⁺CD34⁺ precursors use both PECAM1-dependent homophilic interaction and an α₄-VCAM-1 adhesion system for transmigration, at variance with CD14⁺CD34⁻ cells which adhere, rather than migrate, through the α_L-ICAM-1 receptor-ligand pair.

These findings are in agreement with data from other authors showing that human CD34⁺ progenitor cells use both PECAM-1 and VCAM-1 adhesion systems (16). However, although they describe an indirect involvement of PECAM-1 in cell adhesion, as PECAM-1 engagement leads to up-regulation of integrin affinity for their ligands, we found that PECAM-1 can drive directly the migration of CD34⁺ precursors through a homophilic haptotactic gradient. It is tempting to speculate that in addition to modulation of integrin avidity state, PECAM-1 can regulate the adhesive and migratory properties of CD34⁺ precursors by two additional mechanisms: 1) up- or down-regulation of PECAM-1 surface expression during leukocyte maturation; 2) modulation of PECAM-1 surface distribution on endothelial cells, thus favoring either migration (haptotactic gradient) or adhesion (homogeneous distribution).

We have reported that PECAM-1 localization at the intercellular junctions in endothelial cells depends on the connection with cytoskeletal proteins and is tightly regulated by PECAM-1 phosphorylation (30). The balance between phosphorylated and dephosphorylated endothelial PECAM-1, can be perturbed by TNF-α, through the activation of cellular kinases, both in vitro and in vivo (30), thus implying that tissue microenvironment can influence both PECAM1 distribution and function. Nevertheless, the finding that PECAM1-driven transmigration of CD14⁺CD34⁺ cells is apparently chemokine-independent would favor the hypothesis that this population is committed to localize to peripheral tissues, whatever the microenvironment, to complete their maturation and exert their function. This is further supported by the observation that both migrated and sorted CD14⁺CD34⁺ cells can differentiate into immunostimulatory DC during culture with GM-CSF, which is known to enhance DC maturation (36, 44, 45).

DC precursors originating from the bone marrow migrate to peripheral tissues and primary lymphoid organs, where they become professional APC (46). After processing, Ag-carried DC migrate to the lymph nodes for the induction of specific T lymphocytes (47). Accordingly, we found that DC derived from migrating CD14⁺CD34⁺ precursors retain the ability to migrate using both PECAM-1 and α₄-VCAM-1 adhesion systems in the absence of detectable chemotactic stimuli. DC with an abundance of adhesion and accessory molecules, able to stimulate quiescent T cells in MLR and to home to the T cell areas of lymphoid tissues, have been described (48) in keeping with our results. Moreover, a regulated expression and membrane distribution of PECAM-1, in the bone marrow or in peripheral lymphoid organs, might contribute to the progression of DC maturation and recirculation.

We thank A. Rubartelli and A. Manfredi for critical reading of the manuscript and E. Dal Cin and G. Consogno for skilful technical assistance. We are also grateful to R. Accolla (Advanced Biotechnology Center, Genoa, Italy), F. Sanchez-Madrid (Hospital de la Princesa, Madrid, Spain), P. Newman (Blood Center of Southeastern Wisconsin, Milwaukee, WI), T. Springer (Harvard University, Boston, MA), A. Dobrina (Department of Physiology and Pathology, University of Trieste, Trieste, Italy), A. Mantovani (Istituto Farmacologico M. Negri, Milan, Italy), and S. Burastero and M. Alessio (HSR-Dibit, Milan, Italy) for sharing their reagents.