|
|
||||||||
CONCISE REVIEW |
a Laboratory of Cellular Physiology and Immunology, The Rockefeller University, New York, New York, USA;
b Allogeneic Bone Marrow Transplantation and
c Clinical Immunology Services, Division of Hematologic Oncology, Dept. of Medicine, Memorial Sloan-Kettering Cancer Center, Cornell University Medical College, New York, New York, USA
Key Words. Dendritic cells • Langerhans cells • Hematopoiesis • Myeloid • Cytokines • Antigen-presenting cells • Accessory cells • Cell-mediated immunity
Dr. James W. Young, Box 176, The Rockefeller University, 1230 York Avenue, New York, NY 10021-6399, USA.
| Abstract |
|---|
|
|
|---|
This review outlines some of the ways in which DC are distinguished from two other myeloid lineages, macrophages and granulocytes. Recent data regarding DC development from class II MHC-negative precursors in the mouse, as well as unselected and selected CD34+ progenitors in human bone marrow and peripheral and cord blood, are reviewed. Additional pathways via post-colony-forming units, intermediate cell types have also become evident in suspension cultures where the cytokine milieu can alter terminal differentiation. The availability of larger numbers of DC is opening new avenues for immune therapy that use this physiologic adjuvant.
| Introduction |
|---|
|
|
|---|
Some of the distinctive features of DC will be reviewed first, but the main emphasis will be on the hematopoietic pathways whereby DC develop from human CD34+ progenitor populations or class II MHC-negative precursors in the mouse. The identification of active cytokines has stimulated these recent studies of DC development. Results of these investigations are attracting considerable attention because the availability of larger numbers of these physiologic adjuvants has opened new avenues for immune therapy.
| Dendritic Cells: A Distinct Pathway for Myeloid Differentiation |
|---|
|
|
|---|
|
Soon after DC were identified, bone marrow transplant chimeras were generated to document their marrow origin [14]. DC proved to be myeloid rather than lymphoid in nature. DC lack lymphocyte antigen receptors and do not rearrange either Ig or T cell receptor genes [5]. DC are CD33 positive [6], as are macrophages and granulocytes, and like these other cell types, they are responsive to GM-CSF. However, DC are not known to respond to the more lineage-restricted macrophage and granulocyte colony-stimulating factors (M-CSF, G-CSF). A "DC-CSF" has yet to be identified.
The distinction of mature DC from mononuclear and polymorphonuclear phagocytes is made on the basis of their unusual morphology and pattern of motility, the abundance of class II MHC molecules and the lack of phagocytic activity and typical phagocyte markers (Fig. 2
). The CD83 molecule has been recently identified as a positive marker for DC [7], although it is also weakly expressed on some activated lymphocytes [8].
|
R) or C3 and its fragments. Low levels of the CD32 Fc
R and the CD11b C3biR can be found, particularly when the DC are not fully mature. Recent data indicate that some of the allostimulatory cells in marrow can express CD14 [9], and Szabolcs et al. have identified a CD14+HLA-DR+ bipotential intermediate cell in the differentiation pathway of marrow-derived human DC [10]. However, terminally differentiated DC lack CD14, a receptor that is more abundant on macrophages and mediates the response of phagocytes to lipopolysaccharide (LPS) and LPS binding protein [11]. Mature CD14 DC are nonadherent to glass or plastic. DC also lack two characteristic antimicrobial enzymes of phagocytes, myeloperoxidase and lysozyme. Nonspecific esterases are also weak or absent, although again, these activities can be expressed in immature DC progenitors [12, 13]. In essence, the pathway of DC differentiation sacrifices many of the scavenging and antimicrobial properties of phagocytes for the very high levels of MHC products and accessory molecules that underlie efficient presentation of antigens to quiescent T cells. | Distribution and Nomenclature |
|---|
|
|
|---|
|
In the thymus, DC are confined primarily to the medulla where they are responsible for self-tolerance. Thymic DC process self-antigens and mediate the deletion of self-reactive T cells [27, 28]. In the thymic medulla, and in the T cell areas of peripheral lymphoid organs, the DC have been called interdigitating cells.
Despite their many names, all the different mature members of the DC family can exhibit the features diagramed in Fig. 1
. In contrast is a different type of DC found in the B cell areas or follicles of lymphoid organs [29]. These follicular dendritic cells (FDC) are thought to be stromal rather than myeloid cells. FDC express neither the common leukocyte antigen CD45 nor the leukocyte ß2 integrins. FDC function to present native antigens as immune complexes to B cells, rather than processed antigens as MHC-peptide complexes to T cells.
| GM-CSF Mediates the Maturation of Epidermal Langerhans Cells into Typical DC |
|---|
|
|
|---|
Langerhans cells had some features in situ that were atypical of the DC that had been earlier characterized in lymphoid organs, lymph and blood. For example, freshly isolated Langerhans cells had nonspecific esterase and ATPase, were reactive with the F4/80 antimouse macrophage monoclonal antibody (mAb) and expressed the Fc
R for immune complexes. However, when Langerhans cells were cultured in vitro, all of these features were lost [30]. Concomitantly, the Langerhans cells increased greatly in size, extended many motile lamellipodia and escalated the amount of class II MHC molecules on the cell surface.
The T cell stimulatory properties of Langerhans cells also changed dramatically in culture. A cardinal manifestation of DC function is that very low numbers induce a strong primary response by allogeneic T cells in a mixed leukocyte reaction (MLR). MLRs are classically evaluated using whole blood leukocytes or bulk mononuclear cells at stimulator:responder ratios of 1:1. When DC are the stimulators for allogeneic T cells, stimulator:responder ratios of
1:100 induce even stronger responses than bulk leukocyte stimulators at ratios of
1:1. When Langerhans cells were cultured before addition to primary allogeneic T cells in MLRs, their stimulatory capacity increased at least 10- to 30-fold. This maturation was accompanied by a much greater capacity to bind T cells [31], and an associated upregulation of accessory molecules like CD40, ICAM-1/CD54, LFA-3/CD58, B7-1/CD80 and B7-2/CD86 [12, 32, 33].
When fresh and cultured Langerhans cells were tested for their capacity to present native protein antigen to primed T cells, activity opposite to that of allostimulation in the MLR was noted. In this case freshly isolated cells were capable of processing intact protein antigens for presentation, while cultured cells were very weak [34, 35].
The maturation of Langerhans cells did not occur if the cells were depleted of keratinocytes during purification before in vitro culture. It became evident that the keratinocytes were a paracrine source of cytokines, in particular GM-CSF. Recombinant GM-CSF, when added to immature DC, maintained their viability and maturation into potent immunostimulatory cells [36, 37]. Interleukin 1 (IL-1) could further enhance maturation when used in combination with GM-CSF [37].
M-CSF and G-CSF did not mediate Langerhans cell maturation in vitro [36]. Subsequently, it became evident that DC have few if any M-CSF receptors [5, 10] but have abundant GM-CSF receptors [5, 38]. There is some interest in the idea of a corresponding DC-CSF, but there is no direct evidence for this. As will be discussed below, tumor necrosis factor
(TNF-
), CD40 ligand and possibly other related cytokines can also promote the development and function of DC [39, 40].
A comparable maturation of Langerhans cells almost certainly takes place in vivo under several circumstances. For example, when skin is transplanted or explanted, the Langerhans cells leave the skin to enter the afferent lymph [41, 42]. Simultaneously, expression of MHC products rises dramatically, as does the expression of B7 costimulatory molecules [43]. When skin is placed into organ culture, Langerhans cells also migrate from the skin into the medium [44, 45]. The emigrated cells are rich in adhesion and costimulatory molecules that are not evident in the epidermis prior to migration, e.g., B7-2, ICAM-1, LFA-3 and CD40. This has led to the concept that Langerhans cells would first capture and process antigens while actively synthesizing MHC products. Large amounts of MHC-peptide complexes would then be formed during maturation into typical stimulatory DC [4648].
| Immature, Cytokine-Responsive DC are Present in Other Tissues Including Human Blood |
|---|
|
|
|---|
IL-4 is another cytokine that has been extensively evaluated because of its known suppression of monocyte development even at the clonogenic level [52]. Investigators have reported the development of candidate DC from unselected precursors in human peripheral blood with the support of GM-CSF and IL-4 [39, 53]. Under these cytokine conditions large numbers of cells developed which expressed high levels of class II MHC and costimulator molecules and exhibited stimulatory activity in the MLR. In combination with GM-CSF, IL-4 also supported presentation of soluble antigen by immature DC, a process further enhanced by immune complexes that had been endocytosed by FcR-mediated capture [39]. Even brief exposure to TNF-
resulted in diminished antigen capture and presentation by this mechanism. In contrast, the addition of TNF-
to GM-CSF led to upregulation of class II MHC molecules, accessory ligands, and T cell stimulatory capacity in the MLR.
Other cell types can express the currently known costimulatory and adhesion molecules like B7-1/CD80, B7-2/CD86, ICAM-1/CD54, ICAM-3/CD50, LFA-3/CD58 and so on. What appear to be distinctive about DC are the high levels that can be expressed and their regulation. Marked upregulation of these accessory ligands results from in vitro culture and maturation of resident DC from a solid organ [33] or peripheral blood [6, 23, 54], or antigen-specific T cell binding to stimulatory DC [54, 55]. Other cells, like macrophages or B cells, require additional stimuli such as interferon
(IFN-
) [56], LPS [33] or anti-Ig [33].
The mature DC is therefore a motile, nonadherent leukocyte of the myeloid lineage, with many sheet-like veils or cytoplasmic processes. Maturation provides DC with potent immunostimulatory properties due to the abundant amounts of MHC products and accessory molecules expressed during terminal differentiation. Mature DC typically are short-lived in vitro (one to five days) and are not known to revert back to a less mature state. Viability can be prolonged with the cytokines GM-CSF and TNF-
[36, 37, 57, 58], and by stimulation of CD40 using gp39/CD40L [40]. Because the latter is expressed on activated T cells, T cell contact and stimulation may be a critical means of maintaining DC viability and stimulatory function during the onset of an immune response.
| Proliferating Progenitors of DC in the Mouse |
|---|
|
|
|---|
Upon addition of GM-CSF, class II MHC-negative precursors gave rise to three myeloid cell types, i.e., macrophages, granulocytes and DC. The marrow from two femurs produced about 5 x 106 DC in seven to eight days of culture. In these cultures, granulocytes were the first to develop, but these were nonadherent cells that could be rinsed away. Thereafter, especially between days 4 and 6 of culture, DC developed in aggregates that were loosely adherent to stromal cells alongside typical adherent monocytes. Pulse chase experiments with [3H]TdR confirmed that the aggregates contained proliferating cells from which mature progeny DC were released.
The DC progeny exhibited typical motile sheet-like processes, or veils. MHC class II molecules were abundant, and a characteristic repertoire of other surface markers was displayed. The DC progeny proved to be potent stimulators of resting T cells, and in vivo the cells could migrate and home to the T cell areas of draining lymph nodes. Other hematopoietic growth factors like G-CSF, M-CSF and IL-3 did not support the growth of DC. Furthermore, despite the parallel development of granulocytes and macrophages in these cultures, the commitment to the DC pathway of differentiation was fixed and could not be reversed by subsequent exposure to M-CSF.
The development of murine DC was also followed in clonogenic assays in methylcellulose, beginning with class II MHC-negative precursors from bone marrow [61]. The colonies were manually harvested, and enriched populations of the respective myeloid cell types prepared. The DC represented only about 1% of the progeny but were typical in their large size and irregular shape, abundance of class II MHC molecules, and absence of certain macrophage and granulocyte markers. The DC were separated from the macrophages and granulocytes by plastic adherence methods. The DC were potent MLR stimulators, in contrast to the macrophage progeny. These colony assays demonstrated that murine granulocytes, monocytes and DC share a common, GM-CSF responsive progenitor at some point in their development. The proportion of each type of myeloid progeny, e.g., 0.5 to 1.5% for DC, reflected its relative frequency in most sites where myeloid cells circulate or reside in vivo.
During the development of DC in cytokine-driven suspension cultures of mouse marrow, the cells can display some phagocytic activity, albeit less so than macrophages in the adherent monolayer [62]. With Calmette-Guerin bacillus (BCG) as the particulate antigen for phagocytosis, it was shown that the BCG-bearing DC were stimulatory of T cells both in vitro and in vivo, to a degree not matched by similar exposure of mature DC to BCG. Antigen uptake and processing by immature progenitors should permit the presentation of peptides tailored to newly synthesized self-MHC molecules, concomitant with increasing numbers of DC from proliferating precursors, all under the aegis of GM-CSF.
Thomson et al. have reported the generation of DC from mouse liver [63]. The liver has hematopoietic potential, and it is known to be one of the most tolerogenic solid organs when transplanted. GM-CSF alone was unable to generate mature functional DC from bulk hepatic nonparenchymal cells. Instead, additional exposure to type I collagen was required. Upon such stimulation the DC had much higher levels of MHC class II, were better stimulators of T cells in the MLR and could home in vivo to the T cell areas of lymphoid organs. Thomson et al. hypothesized that the generation of immature DC might impart tolerogenic activity to the newly transplanted liver.
The Combination of GM-CSF and TNF- Stimulates Human CD34+ Progenitors to Develop into DC and Other Myeloid Cell Types in Suspension Cultures
|
|---|
|
|
|---|
[64]. As many as 40% of these cells met some of the criteria for identification as DC. However, the cells were not potent stimulators of T cell responses.
Caux et al. detailed the generation of DC from cord blood CD34+ precursors in the presence of GM-CSF and TNF-
[65]. Candidate DC, identified by a stellate cell shape, potent MLR stimulation, and an array of typical surface markers, constituted at least 20%-50% of the progeny. Electron microscopy showed that at least 20% of the progeny contained Birbeck granules, a selective marker for Langerhans cells. One of the important criteria for the production of DC was the demonstration of potent stimulatory function for quiescent allogeneic CD4+ T cells. This was enriched in the fraction of cells that expressed CD1a+ and depleted among cells lacking CD1a. Mechanisms proposed to explain the stimulatory effects of TNF-
on DC development include the recruitment of primitive, transferrin-receptor (TfR) negative progenitors to become TfR positive and responsive to other growth factors like IL-3 and GM-CSF [66], as well as the downmodulation of CSF receptors [6769].
Szabolcs et al. also described the generation of DC in liquid culture, but used CD34+ precursors from normal human bone marrow [70]. These authors stressed the fact that the combination of GM-CSF and TNF-
led to the development of granulocytes, macrophages and DC alongside each other. Approximately 50% of the cells were class II MHC positive, and these included typical adherent CD14+ macrophages and nonadherent CD14 DC in a ratio of about 4:1. This constituted an enormous expansion of DC, i.e., absolute yields from the CD34+ cells in
5 ml of a cellular marrow specimen approximated the usual yield of ~106 DC from an entire unit of peripheral blood. Nevertheless, their proportion relative to other myeloid cells seemed smaller than that reported from cord blood, and the incidence of CD14+ cells seemed greater from marrow [65].
After two weeks' growth and maturation in GM-CSF and TNF-
, these cells were cytofluorographically sorted according to the presence or absence of CD14 among the HLA-DR+ progeny. The cells in the CD14-HLA-DR++/+++ fraction were nonadherent, motile, and had the cytologic and phenotypic features of DC. In the critical test of potent immunostimulatory function for a primary T cell response, these candidate DC exceeded the stimulatory capacity of the sorted CD14+HLA-DR+ macrophages by at least 1.5 to 2 logs.
c-kit ligand increased expansion 10- to 15-fold over that achieved by GM-CSF and TNF-
alone, but did not directly influence DC differentiation. Neither cytologic, phenotypic, nor stimulatory properties were altered by the addition of c-kit ligand to GM-CSF and TNF-
. The value of a cytokine like c-kit ligand became more apparent in clonogenic assays (see below and [13]) where primary cloning efficiency as well as expansion of progenitor populations were both enhanced.
These results with human CD34+ progenitors should serve to stimulate further research in the murine system where only GM-CSF is applied to obtain sizable numbers of DC. For example, it is possible that endogenous TNF is being generated in the murine cultures, and that signaling via the TNF receptor or a homologue like CD40 will be an important common feature of DC growth and maturation. In fact, soluble gp39/CD40L could substitute and may even optimize many of the effects attributed to TNF-
in the human system [39]. Caux et al. [40] have demonstrated the high expression of CD40 by human DC generated by GM-CSF and TNF-
from CD34+ cord blood progenitors. CD40L:CD40 interactions maintained DC viability and augmented accessory molecule expression and secretion of limited cytokines. They have proposed that insofar as DC-activated T cells upregulate CD40L, the CD40L:CD40 interaction likely plays an important physiologic role in DC-initiated T cell immune responses.
| A Distinct CD34+ Progenitor for DC in Colony Forming Assays |
|---|
|
|
|---|
under serum-free conditions [72]. Colonies generated from unselected bulk bone marrow mononuclear cells yielded rare candidate colonies of DC only. However, colony identification was based solely on morphology and CD1a expression, and cloning efficiency was <0.05% from this heterogeneous starting population. Because cytokines like GM-CSF and IL-4 can induce expression of CD1 epitopes on monocytes [73, 74], CD1 is not specific to dendritic or Langerhans cells under cytokine-driven conditions. Limited cell yields did not permit further phenotypic or functional characterization.
By combining GM-CSF and TNF-
in serum-containing conditions for the culture of bone marrow CD34+ progenitors, CFU-DC that gave rise to pure DC colonies were readily identified [13]. The addition of TNF-
to GM-CSF led to the appearance of these pure DC colonies, in addition to granulocyte/macrophage (GM) colonies. The progenitors for each colony type were therefore distinct.
c-kit ligand proved synergistic with respect to the effects of GM-CSF and TNF-
on DC, just as it has been with other cytokines that affect other cell types. c-kit ligand afforded an approximately twofold increase in primary colony growth, but expanded CFU-DC almost 100-fold over 14 days as documented in secondary clonogenic assays. As in the bulk suspension cultures, however, c-kit ligand did not directly affect DC differentiation.
The cells recovered from CFU-DC-derived colonies were typical DC. They were nonadherent and motile when transferred to liquid culture, with striking cytoplasmic veils or dendrites. They were CD14, intensely HLA-DR+++, and exerted MLR stimulatory activity comparable to that effected by mature blood DC. Other phenotypic and cytochemical traits did not prove sufficiently discriminatory from macrophages. CFU-GM gave rise to colonies alongside DC colonies in the presence of GM-CSF and TNF-
. In contrast to GM colonies grown in GM-CSF only, however, the addition of TNF-
led to the development of ~1% cells in the GM colonies with the same morphologic, phenotypic, and functional features of the CFU-DC progeny.
GM colonies that include ~1% DC, when grown in conditions that also support CFU-DC expansion, are consistent with the murine data reported by Inaba et al. [61]. This trace representation of DC in mixed myeloid colonies is comparable to their proportion among myeloid cells in most resident sites in vivo. The purely DC-committed pathway, however, may in the steady-state, contribute DC to sites where they are the principal myeloid cell type, e.g., epidermis, afferent lymph, and T cell areas of lymphoid organs.
| Intermediates in the Development of Human DC from CD34+ Precursors |
|---|
|
|
|---|
, a nonclonogenic, CD14+HLA-DR+ cell could still be identified (Fig. 2
resulted in the terminal differentiation of DC. Reculture of the same cells, with or without M-CSF, generated macrophages. This CD14+HLA-DR+ intermediate cell therefore had bipotential differentiation capacity dependent on the cytokine milieu. Terminal differentiation could not be reversed, and cells of the same CD14+HLA-DR+ phenotype isolated beyond six to seven days of culture were no longer bipotential. These investigators concluded that DC differentiation via the CD14+HLA-DR+ intermediate is ordinarily not seen due to the suppressive effect of macrophages that develop alongside in bulk culture. This may explain the scant numbers of candidate DC in GM colonies that develop under cytokine conditions which are otherwise supportive of DC growth. The CD14HLA-DR+++ DC that are already well-established by day 6 may correspond to the progeny of marrow-derived CFU-DC described above.
Caux et al. have reported two intermediates that arise by day 5 from CD34+ precursors in cord blood cultured with GM-CSF and TNF-
[75]. One is CD1a+CD14 and the other is CD1aCD14+. GM-CSF, but not TNF-
, is critical to their subsequent differentiation. CD1a+CD14 intermediates give rise to cells of the same phenotype, but which develop Birbeck granules and are comparable to dendritic/Langerhans cells. The CD1aCD14+ intermediates also give rise to CD1a+CD14 progeny, but they lack Birbeck granules and are comparable to dermal DC.
The characterization and application of intermediate or immature DC precursors, which no longer have colony-forming capacity, have already proven useful in two settings. One is the demonstration of phagocytosis by immature DC for particulate antigen by Inaba et al. [62], and the other is the capture and presentation of soluble antigen [39]. Functional counterparts to maturation in vitro and in vivo have precedent in the more extensive studies of Langerhans cells and class II MHC-restricted T cell responses [12, 34]. The definition and functional characterization of progenitors, intermediates, and terminally differentiated DC should facilitate their application to the manipulation of T cell immune responses.
| Discussion |
|---|
|
|
|---|
. This pathway may generate DC primarily at body surfaces, especially the skin and mucosal epithelium. A different CD34+ progenitor gives rise in the presence of GM-CSF to mixtures of granulocytes and macrophages, and trace numbers of DC when TNF-
is added. Within this pathway, it may be possible to skew differentiation along either a DC or a macrophage pathway by cytokines, e.g., for DC, a combination of GM-CSF and TNF-
, gp39/CD40L, or IL-4, and for macrophages, M-CSF. Yet additional pathways have been reported, i.e., the generation of both thymic DC and T cells [76], or marrow-derived DC, T and B lymphocytes, and natural killer cells [77], each set of progeny from its own common precursor. While there may be many routes to DC development, it has not been possible to interconvert the different, mature myeloid progeny, i.e., DC, macrophages and granulocytes. Furthermore, fully mature DC coordinately express a large number of properties that are very different from phagocytes. Phagocytes handle antigen primarily for clearance rather than for presentation to resting T cells. Additional differences include the complete or near lack of important receptors for LPS and immune complexes (CD14, CD16, CD32, CD64, CD35) on DC, as well as several phagocytic enzymes (lysozyme, myeloperoxidase, nonspecific and specific esterases). DC are also nonadherent and motile, unlike macrophages on tissue culture surfaces. In many tissues, if one selects for cells that exhibit the unusual shape and motility of mature DC, one will simultaneously secure cells that stably express high levels of MHC products and accessory molecules and have potent immunostimulatory activity in vitro and in vivo. It is the existence of such specialized APC with potent accessory properties that is currently intensifying the efforts to understand DC in molecular terms and to use these cells as adjuvants in humans.
| Acknowledgments |
|---|
| References |
|---|
|
|
|---|
. J Exp Med 1995;182:11111120.
. J Exp Med 1994;179:11091118.
stimulated monocytes: A potential mechanism for amplification of T cell activation through the CD28 pathway. Cell Immunol 1991;137:429437.[Medline]
cooperate in the generation of dendritic Langerhans cells. Nature 1992;360:258261.[Medline]
cooperates with interleukin 3 in the recruitment of a primitive subset of human CD34+ progenitors. J Exp Med 1993;177:18151820.
directly and indirectly regulates hematopoietic progenitor cell proliferation: Role of colony-stimulating factor receptor modulation. J Exp Med 1992;175:17591772.
. J Immunol 1991;146:26482653.[Abstract]
. J Immunol 1995;154:58515861.[Abstract]
| |||||||||||||||||||||||||||||||