Two types of dendritic cells (DC) are circulating in human blood and can be identified by their differential expression of the myeloid Ag CD11c. In this study, we show that CD11c− peripheral blood (PB)-DC correspond to plasmacytoid DC of lymphoid tissue not only by their surface Ag expression profile but, more impressively, by their peculiar ultramorphology. We also demonstrate that CD11c− and CD11c+ DC differ in the quality of their response to and in their requirement for certain cytokines. Freshly isolated CD11c− cells depend on IL-3 for survival and use autocrine or exogenous TNF-α as maturation signal, leading to the appearance of a highly dendritic phenotype, the up-regulation and redistribution of MHC class II from lysosomal compartments to the plasma membrane, the increased expression of costimulatory molecules, and the switch from a high Ag-processing to a low Ag-processing/potent accessory cell mode. Surprisingly, IL-4 efficiently killed freshly isolated CD11c− PB-DC, but did not impair the viability of CD11c+ PB-DC and, together with GM-CSF, induced maturation of these cells. A direct functional comparison revealed that neo-Ag-modified and subsequently matured CD11c− but to a lesser extent CD11c+ DC were able to prime naive Ag-specific CD4+ T cells. Our findings show that two diverse DC types respond to certain T cell-derived cytokines in a differential manner and, thus, suggest that suppression or activation of functionally diverse DC types may be a novel mechanism for the regulation of the quantity and quality of immune responses.
Recent studies on dendritic cell (DC)3 subpopulations present in human lymph nodes and tonsils have revealed that in addition to interdigitating cells of T-dependent regions (1), there exist two previously ignored DC species that display differences with regard to their Ag expression profile, ultrastructural appearance, and anatomical distribution (2, 3). One subtype that localizes to T cell-rich areas of lymphoid tissues is a HLA-DR+CD4+CD45RA+ but CD11c−CD1a− cell exhibiting a plasmacytoid phenotype (2). It apparently corresponds to a cell type that previously has been designated erroneously plasmacytoid T cell or plasmacytoid monocyte due to its peculiar morphology, and its anti-CD4 and anti-CD68 immunoreactivity (4). This obscure cell occurs in sterile, fetal lymph nodes and, thus, appears to enter T cell areas in the virtual absence of inflammation-related signals (5). Unlike other DC types, tonsilar CD11c− DC express high levels of IL-3Rα and depend on the presence of IL-3 rather than of GM-CSF for survival and the acquisition of allostimulatory potential (2). The second DC type is found in the dark and light zones of germinal centers, is a HLA-DR+CD4+CD45RO+CD11c+ cell lacking CD1a expression, and, by ultrastructure, displays the cardinal features of DC present in many other tissues (3).
In the search of putative precursors of tissue-bound DC, we and others detected and isolated a lineage Ag (i.e., CD3,CD19,CD14/CD11b,CD16/56,CD34) negative (lin−), but MHC class II+CD4+ PBMC population, which is now collectively referred to as PB-DC (6, 7, 8, 9, 10, 11). Their belonging to the family of dendritic APC was demonstrated by the observation that these cells, upon in vitro culture in medium or in the presence of a complex mix of monocyte-derived factors, acquire a dendritic morphology and pronounced allostimulatory capacity (6, 7, 10, 11). Extensive immunophenotypic analyses revealed that this MHC class II+CD4+ PB-DC population is not homogeneous, but contains CD11c−CD45RA+ and CD11c+CD45RO+ DC subtypes (6, 7, 12). Freshly isolated CD11c+ DC exhibit high levels of MHC gene products and significant amounts of costimulatory molecules and are, or spontaneously become, potent allostimulatory APC (7). In contrast, CD11c− PB-DC were found to depend on preculture in monocyte-conditioned or IL-3/GM-CSF-supplemented medium to acquire DC morphology and allostimulatory function (5, 7, 12). Concerning the interrelationship of CD11c− and CD11c+ PB-DC, it has been postulated that these populations represent different maturational stages of the same cell type possibly on the way to and from nonlymphoid organs, respectively (6, 7). This assumption finds support by the observation that DC freshly isolated from PB by immunodepletion are mainly CD11c−CD45RA+CD45RO−, while those isolated from cultured PBMC are largely CD11c+CD45RA−CD45RO+ (6). On the other hand, it is also conceivable that the two PB-DC populations represent rather unrelated DC types and, moreover, constitute the circulating precursors of the CD11c− and CD11c+ DC located in T cell- and B cell-rich areas, respectively. In support of this concept, recent findings demonstrate that both CD11c− PB-DC and CD11c− plasmacytoid DC of lymphatic tissues abundantly display IL-3Rα (5). However, electron microscopy studies of crude unseparated PB-DC have failed to date to demonstrate substantial ultrastructural heterogeneity among this cell population and, thus, to identify circulating counterparts of lymphatic tissue-bound plasmacytoid DC (13, 14).
To elucidate the interrelationship of CD11c− and CD11c+ PB-DC and to investigate their putative similarities and dissimilarities with lymphatic tissue-derived DC, we comparatively analyzed their immunophenotype and ultrastructural appearance, and investigated their growth/differentiation factor requirements to undergo maturation and to acquire a terminally differentiated phenotype. Moreover, we asked whether the two DC types may differ in the regulation of Ag uptake/processing functions as well as in their capabilities to prime naive T cells for epitopes derived from exogenous protein Ags.
Materials and Methods
Abs and reagents
Purified mAbs (mAb clones) used in this study were: anti-CD3 (UCHT-1), anti-CD8 (B9.11), anti-CD11b/CR3 (BEARI), anti-CD16/FcγRIII (3G8), anti-CD19 (J4.119), anti-CD34 (581), and anti-CD56 (C218; all from Immunotech, Marseille, France); anti-CD40 (G28-5; Ref. 15); anti-CD4 (SK3) and anti-HLA-DR (L243; both from Becton Dickinson Immunocytometry Systems, San Jose, CA); anti-CD45RO (UCHL-1; Dako, Glostrup, Denmark); anti-CD115/M-CSFR (12-2D6; Zymed Immunochemicals, South San Francisco, CA); and anti-CD120a/TNFR type 1 (H398) and anti-CD120b/TNFR type 2 (M1; both from Serotec, Oxford, U.K.). Neutralizing anti-human TNF-α mAbs (clone 1825.21) were from R&D Systems (Minneapolis, MN). Fluorochrome- or biotin-conjugated mAbs (mAb clones) included: anti-CD1a (OKT6) FITC (Ortho Diagnostics, Raritan, NJ); anti-HLA-DR (L243) PerCP, anti-CD11c (Leu-M5) PE, anti-CD13 (Leu-M7) PE, anti-CD33 (Leu-M9) PE, anti-CD45RO (UCHL-1) PE, and anti-CD45RA (Leu-18) FITC (all from Becton Dickinson); anti-CD4 (MT310) PE-Cy5 (Dako); anti-CD40 (B-B20) PE and anti-CD11c (MA460) FITC (from Serotec); anti-CD1a (HI149) FITC, anti-CD80/B7-1 (BB1) FITC, anti-CD86/B7-2 (IT2.2) PE, anti-CD107b/LAMP-2 (H4B4) FITC, anti-CDw116/GM-CSFRα (M5D12) biotin, anti-CDw123/IL-3Rα (7G3) PE, and anti-CD131/common β-chain of IL-3R, IL-5R, and GM-CSFR (3D7) biotin (all from PharMingen, San Diego, CA); and anti-CD83 (HB15a) PE and anti-CD124/IL-4Rα (S4-56C9) PE (from Immunotech). Streptavidin-PE (SA-PE) was from Becton Dickinson. FITC-labeled goat anti-mouse IgG and rabbit anti-rat IgG were from Sigma (St. Louis, MO). Tetramethylrhodamine isothiocyanate (TRITC)-labeled goat F(ab′)2 anti-mouse IgG was purchased from Jackson ImmunoResearch (West Grove, PA). Control mAbs were FITC-, PE-, or PerCP-labeled mouse IgG1 or IgG2a (Becton Dickinson). Keyhole limpet hemocyanin (KLH) and tetanus toxoid (TT) were obtained from Sigma and Calbiochem (La Jolla, CA), respectively. Recombinant human cytokines used in this study included rhGM-CSF (final concentration: 1000 U/ml), rhIL-4 (800 U/ml; both kindly provided by Novartis AG, Basel, Switzerland), rhIL-3 (100 U/ml), and rhTNF-α (100 U/ml; both from Genzyme, Cambridge, MA). Penicillin/streptomycin, l-glutamine, and RPMI 1640 culture medium were from Life Technologies (Paisley, Scotland). X-Vivo 15 was purchased from BioWhittaker (Walkersville, MD). Human AB serum was from PAA Laboratories (Linz, Austria).
PBMC from healthy donors were prepared as previously described (8). Briefly, leukapheresis products were diluted 1/5 in PBS and sedimented over Ficoll-Hypaque (Pharmacia, Uppsala, Sweden) at 400 × g for 30 min at room temperature. This procedure routinely allowed the recovery of 1–2 × 109 PBMC, which, thereafter, were subjected to counter current elutriation centrifugation using a J2-MC centrifuge (Beckman, Palo Alto, CA) equipped with an elutriation device (JE-6B elutriation system; Beckman). A total of 1–2 × 109 PBMC was resuspended in 175 ml PBS, and the elutriation process was initiated at a counterflow rate of 9.6 ml/min and a rotor speed of 1960 rpm. Cellular debris and small lymphocytes were depleted by stepwise (0.2 ml/min) increasing the flow rate to 10.5 ml/min. PB-DC-enriched elutriation fractions were reproducibly obtained at flow rates between 10.5 and 11.8 ml/min. DC enrichment in individual elutriation fractions was monitored by analyzing the frequency of HLA-DR+CD4+lin−PBMC and/or the cellular light scatter profiles using a FACScan (Becton Dickinson). At flow rates of ≥11.8 ml/min, CD14+ monocytes rather than HLA-DR+CD4+lin−PBMC were recovered. The PB-DC-containing elutriation fractions were pooled and the resulting cell population was depleted of residual T, B, NK, and hemopoietic stem cells as well as of monocytes and basophils by anti-CD3/CD11b/CD16/CD19/CD34/CD56 immunolabeling, followed by anti-mouse IgG immunomagnetic depletion (MACS; Miltenyi Biotec, Bergisch-Gladbach, Germany). Ninety to 95% of the remaining nonbound cell fraction (3–10 × 106 PBMC) fulfilled the immunophenotypic criteria of PB-DC (8). In most experiments, this DC-enriched cell population was further reacted with anti-CD11c FITC, anti-CD13 PE, and anti-CD4 Cy-5 mAbs and sorted into CD11c−CD13−CD4high and CD11c+CD13+CD4+ subpopulations (further on referred to as CD11c− DC and CD11c+ DC, respectively) using a FACStarPlus flow cytometer (Becton Dickinson). Typically, this procedure yielded CD11c− (1–3 × 106) and CD11c+ (0.2–0.5 × 106) DC populations of >99% purity. Syngeneic and allogeneic CD4+ T cells were purified by anti-CD8/CD16/CD19/CD56/HLA-DR-based immunomagnetic depletion of lymphocyte-enriched elutriation fractions. CD8+ T cells and naive CD4+ T cells were obtained by exchanging the anti-CD8 for an anti-CD4 mAb and by including an anti-CD45RO mAb in the mAb depletion mixture, respectively.
Flow cytometry analysis
For three-color immunolabeling, 5 × 104 cells were washed twice in ice-cold PBS, resuspended in 30 μl PBS containing the mAb mixture (5 μg/ml each), and incubated on ice for 30 min. The cells were then washed twice and resuspended in 50 μl of PBS. At least 10,000 cells were analyzed on a FACScan (Becton Dickinson). For the detection of M-CSF receptor (CD115) and TNFR type 1 (CD120a) and 2 (CD120b) expression, cells were first reacted for 30 min at 4°C with appropriate mouse or rat mAbs and then incubated with FITC-labeled mouse anti-rat IgG (anti-CD115, anti-CD120b) or goat anti-mouse IgG (anti-CD120a). CDw116 and CDw131 were detected with biotinylated mAbs and streptavidin-PE (SA-PE). In the case of anti-CD120a immunostaining, free goat anti-mouse IgG binding sites were saturated by excess concentrations of mouse IgG before exposing the cells to PE- and PerCP-labeled mAbs.
Purified PB-DC subsets were seeded at a density of 0.5–1 × 106/ml in 96-well flat-bottom microtiter plates (Costar, Cambridge, MA) in X-Vivo 15 or RPMI 1640 medium supplemented with 10% human AB serum, 2.5 mM l-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin, and cultured in the presence of different stimuli for 3 days. Where indicated, cell cultures were supplemented with endotoxin-free neutralizing anti-TNF-α mAbs in a final concentration of 10 μg/ml. Numbers of viable cells were assessed by counting of trypan blue-negative cells using a hemocytometer. To measure cellular proliferation, cells were pulsed for the last 18 h with 0.5 μCi [3H]thymidine (Amersham Life Science, Buckinghamshire, U.K.) per well, and incorporation of the radionucleotide was measured by β-scintillation spectroscopy (Betaplate; Wallac, Turku, Finland).
DC were spun onto l-lysine-coated slides, fixed in methanol (Merck, Darmstadt, Germany) for 15 min at −20°C, quenched with 1% normal horse serum (Vector Laboratories, Burlingame, CA), and exposed to 2 μg/ml of either the IgG1 control mAb MOPC-21 (Sigma) or an anti-HLA-DR mAb clone TAL.1B5; Dako). After three washings in PBS, cell-bound mAbs were conjugated with biotinylated horse anti-mouse IgG (Vector Laboratories) and visualized with avidin/biotinylated peroxidase complexes (ABC Elite Vectastain Kit; Vector Laboratories) and 3-amino-9-ethylcarbazole (Sigma). The samples were counterstained with Harris’ hematoxylin (Merck) and finally embedded in Glycergel (Dako).
Confocal laser-scanning microscopy
A total of 1 × 104 freshly isolated or cultured PB-DC was mounted onto individual fields of adhesion slides (Bio-Rad, Richmond, CA), fixed in 4% paraformaldehyde (Fluka Chemie AG, Buchs, Switzerland), quenched with 50 mM NH4Cl/PBS, and permeabilized in 0.1% saponin (Sigma)/PBS. Slide-bound cells were then reacted with anti-HLA-DR (L243; Becton Dickinson) or isotype-matched control mAbs (2 μg/ml each) for 30 min at 4°C, washed, and exposed to TRITC-labeled goat F(ab′)2 anti-mouse IgG (5 μg/ml; Jackson ImmunoResearch). After two washings, cells were preincubated in 10% mouse serum/PBS before their exposure to FITC-labeled anti-CD107b/LAMP-2 or isotype-matched control mAbs (5 μg/ml each). All Ab dilutions and washing buffers were supplemented with 0.1% saponin and 1% BSA. After rinsing and embedding in Fluoprep medium (BioMerieux, Marcy L’Etoile, France), cells were examined using a confocal laser-scanning microscope system (LSM 410; Zeiss, Oberkochen, Germany) equipped with lasers emitting light at 488 and 568 nm for the excitation of FITC and TRITC, respectively.
Transmission electron microscopy
Sample preparation for transmission electron microscopy was performed as described previously (16). Briefly, 1 × 106 freshly isolated or cultured DC were washed twice in PBS, resuspended in 0.1 M cacodylate/PBS, and fixed for 30 min in 2.5% glutaraldehyde/2% paraformaldehyde (both from Polaron, Watford, U.K.)/0.1 M cacodylate, pH 7.2. Thereafter, cells were postfixed in 4% OsO4 for 1 h at room temperature, dehydrated in a graded series of ethanol dilutions, and infiltrated with and polymerized in Epon 812 (Fluka). Microtome sections (100 nm) were stained with uranyl-actetate and lead citrate (both from Merck) and examined by transmission electron microscopy (1200 EX; Jeol, Tokyo, Japan).
T cell proliferation assays
Graded numbers of freshly isolated or cultured PB-DC were dispensed in individual wells of 96-well round-bottom microtiter plates (Costar) and cocultured with 105 allogeneic or syngeneic CD4+ or CD8+ T cells in X-Vivo 15 or RPMI medium/10% human AB serum for 5 days. During the last 18 h, cultures were pulsed with 0.5 μCi [3H]thymidine (Amersham), and incorporation of the radionucleotide was measured by β-scintillation spectroscopy (Wallac). Results are expressed as mean cpm of triplicate or duplicate cultures. Background counts of controls (responder T cells or stimulator cells alone) were always <500 cpm.
Ag-specific T cell proliferation.
CD11c− and CD11c+ PB-DC were incubated overnight with 25 μg/ml KLH or 10 Lf/ml TT either immediately after beginning or at the indicated time points of a 3- to 4-day culture period. After harvesting, DC were washed twice and graded cell numbers were dispensed into individual wells of 96-well round-bottom microtiter plates and cocultured with syngeneic CD4+ T cells for 7 days. During the last 18 h, cultures were pulsed with 0.5 μCi [3H]thymidine and processed as described for the MLR experiments. Background counts of controls (DC, DC plus Ag, T cells alone, T cells plus Ag) were always <500 cpm.
Circulating CD4+CD11c− but not CD11c+ lineage Ag-negative (lin−) PBMC exhibit a plasmacytoid morphology
To better characterize the (immuno)phenotypic and functional heterogeneity among circulating human DC, we developed a strategy to highly enrich these cells by Ficoll-Hypaque centrifugation, counter current elutriation, and immunomagnetic depletion of non-DC (Fig. 1,A). The resulting cell fraction almost exclusively contained HLA-DR+ PBMC (Fig. 1,B) and was virtually devoid of lin+ cells (<5%, data not shown). A fraction of the HLA-DR+ PBMC, particularly those expressing the highest levels of MHC class II, were CD11c+ (24 ± 5% of HLA-DR+ cells, Fig. 1,B). The remaining CD11c− cells moderately though homogeneously expressed MHC class II. As shown in Fig. 2,A, the HLA-DR+CD11c+ and the CD11c− cells are diverse DC types with regard to myeloid and CD45 isoform expression. CD13 and CD33 as well as the RO isoform of CD45 were selectively expressed by CD11c+ DC. In contrast, CD11c− but not CD11c+ DC expressed CD45RA. Although both exhibited homogeneous anti-CD4 immunoreactivity, neither of the subsets contained CD1a+ cells. This immunophenotypic information was used to purify these DC populations by three-color flow cytometry and cell sorting. As shown in Fig. 1, C and D, enriched PB-DC were stained with fluorochrome-labeled anti-CD11c, anti-CD13, and anti-CD4 mAb, and CD11c−CD13−CD4+ (designated CD11c− DC) and CD11c+CD13+CD4+ (CD11c+ DC) subpopulations were sorted by flow cytometry and subjected to transmission electron microscopy (EM). The ultrastructural analysis of the sorted subsets revealed that the CD11c− DC are round to oval cells, exhibit an oval or kidney-shaped nucleus, and display a plain, unruffled plasma membrane. As their most prominent feature, they exhibit abundant endoplasmic reticulum (ER) membranes (Fig. 3,A). Thus, both by immunophenotype and ultramorphology, these cells resemble the CD11c− plasmacytoid DC that reside in T cell-rich areas of tonsils and lymph nodes (2). Compared with CD11c− DC, CD11c+ DC have a larger cell diameter and show more pronounced protrusions and microvillous projections of their plasma membrane (Fig. 3 B). In addition, most of these cells have multilobulated nuclei and lack the prominent ER of their CD11c− counterparts. The cytoplasm of both PB-DC types contains abundant vacuolar, membrane-bound organelles filled with electron-lucent material.
CD11c− and CD11c+ DC differ in their cytokine receptor expression profile, growth factor requirement for survival, and biologic behavior upon growth factor signaling
To better understand the nature of the stimuli putatively required for the survival and maturation of these distinctive DC populations, we determined the cell surface expression of lymphokine receptors by FACS. CDw131, the common β-chain and signal-transducing unit of the IL-3, IL-5, and GM-CSF receptor, was found on both PB-DC subsets, more pronouncedly though on CD11c− cells (Fig. 4). From the various CDw131-pairing α-chains, CD11c− PB-DC strongly expressed IL-3Rα and were moderately GM-CSFRα+ (Fig. 4). In contrast, CD11c+ DC bear high levels of GM-CSFRα and low amounts of IL-3Rα on their surface (Fig. 4). Both subsets were found to be devoid of anti-IL-5Rα immunoreactivity (data not shown). Whereas both DC types express similar amounts of p55 TNFR type 1, CD11c+ DC display higher numbers of p75 TNFR type 2. By immunostaining, both cell types were found to be negative for M-CSFR and to express IL-4Rα at levels too low to be discerned by FACS (Fig. 4). In RT-PCR experiments not shown, we could detect IL-4Rα-encoding mRNA in both the CD11c− and CD11c+ DC, while M-CSFR transcripts could be amplified from monocyte-derived cDNAs, but not from cDNAs prepared from CD11c− or CD11c+ DC.
Next, we investigated whether the cytokine receptor expression profiles of the two PB-DC subsets correlate with the responsiveness to the respective growth factors in terms of DC survival and/or proliferation. Baseline evaluations revealed that the two subsets strikingly differ in their ability to survive in non-cytokine-conditioned culture medium. Although 20–30% of seeded CD11c+ DC could be recovered after a 3-day culture period even in non-cytokine-conditioned FCS- or human AB serum-containing medium, these culture conditions did not support significant survival of CD11c− cells (Fig. 5, A and B). Most strikingly, IL-3 as well as GM-CSF rescued these cells from spontaneous cell death. Interestingly, in the presence of IL-3, and to a much lesser extent in the presence of GM-CSF, CD11c− cells incorporated [3H]thymidine, indicating that these cells, upon receipt of adequate growth factor signals, can undergo at least limited proliferation (Fig. 5,A). This phenomenon occurred independent of whether or not the cultures were additionally supplemented with neutralizing anti-TNF-α mAbs or stimuli that have been reported to promote DC survival, i.e., rTNF-α and/or stimulatory anti-CD40 mAbs (Fig. 5,A). In the absence of IL-3, the latter agents could not rescue the CD11c− cells from spontaneous cell death (not shown). Fig. 5, A and C, also shows that IL-4, which is of critical importance for the development of DC from monocyte precursors (17), has a detrimental effect on CD11c− DC. When IL-4 was added at the start of the culture, it could entirely abolish IL-3-mediated proliferation and survival of CD11c− DC (Fig. 5, A and C), an effect that could not be reverted by CD40 signaling (Fig. 5,A). This death-promoting potency of IL-4 appears to be DC subtype restricted because a similar effect on CD11c+ DC was not observed (Fig. 5,B), and to depend on the maturational stage of CD11c− DC because these cells progressively lost their IL-4 sensitivity during culture in IL-3-supplemented medium (Fig. 5,C). In contrast to the situation seen with CD11c− cells, neither IL-3 nor GM-CSF could induce CD11c+ DC to proliferate. Nevertheless, culture of CD11c+ DC in the presence of GM-CSF, either alone or in combination with IL-4 and/or anti-CD40 mAbs, regularly resulted in higher recovery rates of viable cells than observed in cultures with non-cytokine-conditioned or IL-3-supplemented medium (Fig. 5 B).
Two phenotypically different mature DC types emerge from circulating CD11c− DC and CD11c+ DC upon cytokine signaling
EM and immunohistochemical analyses revealed that IL-3/TNF-α-stimulated CD11c− DC, as compared with their freshly isolated counterparts (Fig. 3,A), display an increased overall cell size and develop characteristic dendritic plasma membrane protrusions (Fig. 3, G and H). Moreover, IL-3/TNF-α stimulation resulted in an extensive remodeling of their cytoarchitecture, best evidenced by a striking reduction in the abundance of ER membranes and, thus, in a loss of their initial plasmacytoid appearance (Fig. 3,G). Although neutralizing anti-TNF-α mAbs did not counteract the survival-promoting effect of IL-3 on CD11c− DC (Fig. 5,A), TNF-α neutralization inhibited the loss of the plasmacytoid appearance and the evolution of a typical DC morphology (Fig. 3, D and E). Remarkably, the cytoplasm of IL-3-stimulated CD11c− DC, cultured in the presence of neutralizing anti-TNF-α mAbs, contained giant multivesicular compartments (Fig. 3,D) and, in rare cases, nuclei of apoptotic cells (Fig. 3,F). Similar to IL-3/TNF-α-stimulated CD11c− DC, cytokine (GM-CSF/IL-4)-stimulated CD11c+ DC showed an increase in their overall cell size and exhibited numerous dendritic plasma membrane projections (Fig. 3,C). In contrast to their CD11c− counterparts, most cytokine-stimulated CD11c+ DC contained membrane-bound, macropinosome-like vesicles of ∼1 μm diameter with moderately electron-dense proteinacious content. Culture of CD11c− DC in the presence of IL-3/TNF-α resulted in the de novo expression of moderate amounts of the myeloid Ags CD11c, CD13, and CD33 (Fig. 2,B). Cytokine-stimulated CD11c+ DC continued to express the myeloid Ags CD13 and CD33 and, in contrast to CD11c− DC, could be induced to express CD1a. Importantly, both cell types retained the mutually exclusive CD45 isoform expression pattern during in vitro maturation (Fig. 2, A and B). Thus, freshly isolated CD11c− DC and their in vitro matured progeny are CD45RO−RA+, while freshly isolated CD11c+ DC and their progeny display a CD45RO+RA− phenotype.
To determine the influence of cytokines that regulate DC survival and/or proliferation on the occurrence of maturation-related events, immunocytometry and histochemistry experiments were performed. Culture of CD11c− PB-DC in IL-3 alone and, more pronouncedly, in IL-3/TNF-α-conditioned medium, resulted in the vigorous and homogeneous up-regulation of MHC class II Ags as well as of the costimulatory molecules CD40, CD80, and CD86, and led to the neo-expression of CD83 at moderate levels (Fig. 6,A). In keeping with our ultrastructural observations (Fig. 3), anti-TNF-α mAbs strikingly reduced the potency of IL-3 to promote maturation of CD11c− DC (Fig. 6,A). One must therefore assume that IL-3 induces CD11c− DC to produce TNF-α and to use this cytokine as an autocrine factor promoting maturation rather than survival. Although IL-3 was highly effective for the induction of maturation-related events in CD11c− DC, GM-CSF was a more potent stimulus than IL-3 for the induction/up-regulation of costimulatory molecule expression by CD11c+ DC (Fig. 6,B). Contrasting the findings with CD11c− DC, the neutralization of TNF-α did not affect the level of HLA-DR or costimulatory molecule expression by cultured CD11c+ DC (data not shown). Another fundamental difference between the two PB-DC subsets is their fate upon encounter of IL-4. This cytokine induces the rapid demise of CD11c− DC but, on the other hand, enhances MHC class II and CD83 expression in GM-CSF-stimulated CD11c+ DC (Fig. 6 B).
The cytokine-induced maturation of both PB-DC subsets is also evidenced by changes in their MHC class II expression and distribution. It has been shown previously that MHC class II Ags, while residing mainly in endo-/lysosomal compartments in immature DC, almost exclusively occur on the cell surface in mature DC (18, 19). When we subjected freshly isolated, fixed, and permeabilized CD11c− and CD11c+ DC to anti-HLA-DR/anti-LAMP-2 double labeling, we detected MHC class II preferentially in cytoplasmic LAMP-2+ organelles (Fig. 7, C and I). CD11c+ DC consistently displayed higher levels of HLA-DR than CD11c− DC. After culture in the presence of stimuli found to induce DC morphology and CD83 expression, CD11c− and, somewhat less so, CD11c+ DC not only displayed highly increased total cellular amounts of MHC class II molecules but, more importantly, mainly expressed these moieties on the cell surface (Fig. 7, D, F, J, and L). As a result, LAMP-2+ lysosomal organelles in CD11c− DC were entirely devoid of HLA-DR, whereas some residual MHC class II could be detected in lysosomes of in vitro matured CD11c+ DC (Fig. 7, F and L). In CD11c+ cells, cytokine treatment also resulted in the accumulation of the initially multiple and peripherally located LAMP-2-containing organelles in a perinuclear position (Fig. 7, H and K).
Freshly isolated CD11c+ DC, but not CD11c− DC, are potent stimulators of allogeneic CD4+ and CD8+ T cell proliferation
To determine the accessory cell (ACC) potential of PB-DC subsets, we assessed their capacity to induce proliferation of allogeneic naive CD4+ and CD8+ T cells. Although freshly isolated CD11c− DC were incapable of inducing proliferative MLR responses (Fig. 8,A), cells that had been exposed to IL-3 and TNF-α for 3 days before the initiation of the MLR culture were potent stimulators of naive CD4+ and CD8+ T cells (Fig. 8, B and C). When CD11c− DC were stimulated with IL-3 in the presence of neutralizing anti-TNF-α mAbs instead of TNF-α, their allostimulatory potential for CD4+ T cells and, to a lesser extent, for CD8+ T cells was strikingly reduced (Fig. 8, B and C). Freshly isolated CD11c+ DC elicited a vigorous proliferative response of allogeneic naive CD4+ as well as CD8+ T cells (Fig. 8,A) and did not increase their immunostimulatory capacity upon preculture in GM-CSF/IL-4 (±anti-CD40 mAb)-supplemented medium (Fig. 8 D). Thus, it appears that, in the presence of allogeneic T cells, freshly isolated CD11c+ DC, but not CD11c− DC, can convert into potent APC, and that CD11c− DC require cooperative IL-3 and TNF-α receptor triggering to acquire competent ACC functionality.
During in vitro culture, CD11c+ PB-DC continuously down-regulate their Ag-processing capacity, while CD11c− DC enter a temporary state of maximized Ag processing before terminal maturation
To investigate whether the two PB-DC subtypes differ in their capacity to process exogenous Ag, sorted CD11c− and CD11c+ DC were pulsed with full protein Ag (TT) immediately after their purification and at various time points during a 3-day culture period in cytokine-conditioned medium. Thereafter, TT-modified DC and, for control purposes, non-Ag-exposed DC were assayed for their ability to activate autologous CD4+ T cells from TT-sensitized donors. As shown in Fig. 9, A and B, CD11c+ DC most efficiently induced TT-specific T cell proliferation when pulsed immediately after their purification and, thereafter, progressively lost their ability to generate and display exogenous Ag-derived peptides. In contrast, CD11c− DC stimulated TT-specific autologous T cells most potently when they were exposed to IL-3/TNF-α for 2 days before Ag pulsing (Fig. 9, A and B). In contrast, Ag pulsing between day 3 and 4 resulted in much less pronounced or, even, no TT-specific T cell response (Fig. 9, A and B, and data not shown). Particularly at low DC:T cell ratios, the cytokine-stimulated progeny of CD11c− DC, rather than that of CD11c+ DC, appeared to be more potent stimulators of full protein Ag-specific T cell proliferation (Fig. 9 B).
Neo-Ag-modified and subsequently matured CD11c− DC are able to prime naive Ag-specific CD4+ T cells
To see whether the two PB-DC types also differ in their ability to generate T cell responses against neo-Ag, CD11c− as well as CD11c+ DC were pulsed with KLH immediately after isolation and at different time points during a 4-day culture period and were then incubated with naive autologous CD4+ T cells. These experiments revealed that CD11c− DC are far superior to CD11c+ DC in their capacity to stimulate KLH-dependent T cell proliferation (Fig. 9,C). In analogy to the results obtained with TT, KLH-dependent T cell proliferation was optimal when CD11c− DC were Ag modified on day 2 and did not occur when Ag pulsing was delayed until day 4 (Fig. 9,C). While poor in stimulating primary Ag-dependent T cell responses, in vitro matured CD11c+ DC were more potent stimulators of autologous MLR than their CD11c− counterparts (Fig. 9 D). Although the high autostimulatory potency of CD11c+ DC may limit the detectability of weak KLH-dependent responses, it appears that the two DC subsets differ remarkably in their abilities to present self or exogenous Ags in quantities sufficient for T cell priming.
In this study, we show that CD11c− and CD11c+ PB-DC exhibit fundamental differences in their immunophenotype, ultrastructure, cytokine responsiveness, and growth factor requirements to undergo functional maturation as well as in their capability to prime naive T cells for exogenous Ag. These findings argue for the existence of two developmentally and functionally diverse immature DC subsets that are circulating in the blood and, upon extravasation and receipt of appropriate signals, can give rise to two types of mature DC, presumably serving specialized immune functions in vivo.
FACS-purified PB-DC subsets exhibited striking differences in their ability to survive spontaneously. Although the plasmacytoid CD11c− PB-DC underwent cell death within a few hours of in vitro culture, a substantial proportion of seeded CD11c+ PB-DC could be recovered even following a 3- to 4-day culture period. In keeping with previous observations (5, 12), CD11c− PB-DC could be rescued from cell death by exogenous IL-3 and displayed limited, though reproducible, thymidine incorporation. In accordance with their cytokine response profile, freshly isolated CD11c− PB-DC abundantly expressed IL-3R, moderate levels of GM-CSFR, and not at all displayed IL-5R. The maturation-promoting effect of IL-3, on the other hand, is an indirect rather than a direct consequence of IL-3R triggering. Autocrine or exogenous TNF-α, while not directly supporting the survival of CD11c− PB-DC, is critical for 1) the conversion from the plasmacytoid to a DC phenotype; 2) the up-regulation of MHC class II products, costimulatory molecules, and CD83 expression; and 3) the acquisition of a potent ACC function. The assumption that CD11c− PB-DC can use an autocrine, maturation-promoting TNF-α loop is further supported by our observations that 1) TNF-α-encoding mRNA could be detected in IL-3-stimulated, but not in freshly isolated CD11c− PB-DC (data not shown); 2) as seen in preliminary experiments, IL-3-stimulated CD11c− PB-DC secreted significant amounts of TNF-α during a 2-day culture period (290 pg/106 cells, mean of two experiments); and 3) neutralization of endogenously produced TNF-α prevents maturation, but does not impair the IL-3-induced survival/proliferation of these cells. Thus, it appears that exogenous IL-3 and autocrine TNF-α regulate discrete cell functions, i.e., survival vs maturation, during the transition from the circulating to an advanced maturational stage of CD11c− PB-DC.
By immunomorphologic and functional criteria, we have now identified signals that allow CD11c− and CD11c+ PB-DC to arrive at defined maturational stages. As shown in this study, freshly isolated and, to an even greater extent, short-term IL-3/TNF-α-stimulated CD11c− PB-DC display a potent Ag uptake/processing function. Furthermore, CD11c− PB-DC redistributed MHC class II Ags from lysosomal, LAMP-2+ compartments toward their dendrite-bearing plasma membrane upon stimulation with IL-3/TNF-α. Upon continued stimulation with IL-3/TNF-α, these cells gradually reduce and finally lose this ability, but greatly increase their ACC potency. Together, these findings support the notion that IL-3 and TNF-α are indeed critical and sufficient stimuli for the terminal maturation of circulating CD11c− DC. In contrast to the observations with CD11c− DC, in vitro cytokine stimulation of CD11c+ DC never resulted in an increased Ag uptake/processing function. These cells most efficiently processed and presented Ag when pulsed in their freshly isolated state and, then, progressively lost this cellular function during the subsequent culture period.
Surprisingly, CD11c− but not CD11c+ PB-DC elicited significant T cell responses against the primary Ags KLH and OVA (Fig. 9,C and data not shown). As expected from the recall Ag-pulsing experiments, CD11c− DC antigenically modified 2 days after the initiation of IL-3/TNF-α stimulation displayed a superior ability to elicit primary Ag-specific T cell responses than cells that were pulsed in their freshly isolated or 4-day cultured state. The poor capacity of CD11c+ PB-DC to induce primary Ag-specific T cell responses can be explained neither by an inability of these cells to take up/process full protein Ags nor by their limited ACC function. In fact, MHC class II and costimulatory molecule expression of cultured CD11c+ DC was always more pronounced than that of in vitro activated CD11c− DC isolated from the same donor (Fig. 6). Moreover, freshly isolated as well as cytokine-stimulated CD11c+ PB-DC were even far more potent APC in the autologous MLR than were their CD11c− counterparts (Fig. 9 D). Thus, it appears that factors other than the MHC class II and B7 surface expression density determine the potent primary Ag presentation function of DC derived from CD11c− precursors. Such conditions may include the abilities of these cells 1) to contact large numbers of Ag-specific naive T cells due to a pronounced migratory capacity, 2) to display long- rather than short-lived MHC-peptide complexes along with potent costimulatory signals other than B7, and, perhaps as a result, 3) to efficiently conjugate naive Ag-specific T cells for time periods that are critical for priming.
Concerning the interrelationship of CD11c− and CD11c+ PB-DC, we have evidence that the two DC types belong to separate pathways of DC development. This is best evidenced by the observation that CD11c− and CD11c+ PB-DC maintain their divergent CD45 isoform expression profile during the whole process of maturation. Moreover, it appears that CD11c+ PB-DC are of myeloid derivation, while the lineage commitment of CD11c− DC is less clear. It has been observed that circulating CD11c− DC express high levels of the myeloid lineage-restricted Ag CD68 in their cytoplasm (12), display CD13, CD33, and even CD11c upon cytokine-mediated maturation (this study), and, apparently, can develop from granulomonocytic progenitors in vitro (5). Although all these arguments support a myeloid origin of these cells, the possibility that they derive from early lymphoid precursors cannot be definitively excluded. CD11c− PB-DC resemble the plasmacytoid DC previously identified in T cell-rich areas of tonsils and lymph nodes (2) not only by differentiation Ag (common lineage/myeloid Ag−, CD4+), cytokine receptor (IL-3Rα+), and CD45 isoform (RA+RO−) expression, but, as also demonstrated in this study, by ultramorphology (Fig. 3). Selective tissue-homing properties of CD11c+ PB-DC are less obvious, and it is conceivable that these cells give rise to different types of tissue (e.g., dermal) DC. However, it is noteworthy that BM-derived DC in germinal centers display an (immuno)phenotype (i.e., MHC class I/II+, myeloid Ag+, CD4+, CD1a−, CD45RA−, but CD45RO+) that is indistinguishable from that of circulating CD11c+ DC (3).
In this study, we show that CD11c− PB-DC acquire all the cardinal features of mature DC upon cytokine stimulation in vitro. Currently, no experimental information exists if and where ontogenetically related mature DC reside in lymphatic tissue. An easy identification of these cells may be hampered by the fact that activated CD11c− DC can express myeloid Ags and, thus, resemble other myeloid DC. The finding that the mature form of CD11c− PB-DC, but not of CD11c+ PB-DC and other myeloid DC, is a CD45RA+RO− cell, may now allow the identification of its counterpart in situ. This mature CD11c− cell-derived DC could be the enigmatic macrophage inflammatory protein-3β/EBV-induced molecule 1 ligand chemokine-producing dendritically shaped cell of lymphoid tissues that occurs exquisitely in T cell-rich areas (20). This assumption is supported by our recent findings that IL-3/TNF-α-stimulated CD11c− PB-DC, but not in vitro matured epidermal Langerhans cells, express MIP-3β mRNA (manuscript in preparation) that displays chemotactic activity for CCR7-bearing T cells and mature DC.
Our demonstration that DC matured from CD11c− precursors are potent inducers of primary Ag-specific T cell responses together with most recent observations that these cells, unlike other mature DC, are incapable of producing IL-12 p35 and p40 (Ref. 21 ; N. Kohrgruber and D. Maurer, unpublished data), suggests that they may be involved in the elicitation of Th2 rather than Th1 responses. Our data (this study) and most recent observations by others (21) demonstrate that CD11c− DC, in their immature state, are susceptible to IL-4-mediated cell death. Our further observation that CD11c− DC upon IL-3 stimulation progressively lose this IL-4 sensitivity indicates that the timing of IL-3 and IL-4 signals received by the CD11c− DC could decisively influence the quality of T cell responses. The effect of IL-4 on CD11c− DC is unique because for most types of DC/DC precursors IL-4 acts as a cytokine with differentiation-inducing and/or antiapoptotic properties (17, 22, 23). Although it is still enigmatic where CD11c− DC acquire Ag and which Ags they display in vivo, our data suggest that these cells in their various states of maturation may well be critically involved in the elicitation of specific T cell responses and, perhaps also, in the regulation of T cell-dependent B cell responses in lymphoid tissues.
We thank Dr. Sybille Wichlas for helpful technical advice.
This work was supported in part by grants from the Ministry of Science and Transportation and from Novartis (Basel, Switzerland).
Abbreviations used in this paper: DC, dendritic cell; ACC, accessory cell; EM, electron microscopy; ER, endoplasmic reticulum; KLH, keyhole limpet hemocyanin; LAMP, lysosome-associated membrane protein; PB-DC, peripheral blood DC; PerCP, peridine chlorophyll protein; rh, recombinant human; TRITC, tetramethylrhodamine isothiocyanate; TT, tetanus toxoid.