The cellular microenvironments in which dendritic cells (DCs) develop are not known. DCs are commonly expanded from CD34+ bone marrow precursors or blood monocytes using a cocktail of growth factors including GM-CSF. However, cytokine-supported cultures are not suitable for studying the intermediate stages of DC development, since progenitors are quickly driven to become mature DCs that undergo limited proliferation and survive for only a short period of time. This lab has developed a long-term culture (LTC) system from spleen which readily generates a high yield of DCs. Hematopoietic cells develop under more normal physiological conditions than in cultures supplemented with cytokines. A spleen stromal cell monolayer supports stem cell maintenance, renewal, and the specific differentiation of only DCs and no other hematopoietic cells. Cultures maintain continuous production of a small population of small-sized progenitors and a large population of fully developed DCs. Cell–cell interaction between stromal cells and progenitor cells is critical for DC differentiation. The progenitors maintained in LTC appear to be quite distinct from bone marrow–derived DC progenitors that respond to GM-CSF. The majority of cells produced in LTC are large-sized cells with a phenotype reflecting myeloid-like DC precursors or immature DCs. These cells are highly endocytotic and weakly immunostimulatory for T cells. This model system predicts in situ production of DCs in spleen from endogenous progenitors, as well as a central role for spleen in DC hematopoiesis.
The study of the dendritic cell (DC) lineage and of the signals and factors regulating cell development and function is an area of intense interest to immunologists. DCs are highly endocytotic and the most efficient antigen-presenting cells for the immune system. They are therefore the ultimate controllers of the immune response and an extremely important target in the development of strategies for immunotherapy. Multiple DC subsets have now been defined in many organs on the basis of cell surface marker expression and function [reviewed in 1, 2]. DC subsets can also interact with a range of immune cells, presenting antigen to naïve or memory T cells and inducing activation of natural killer cells . They also present unprocessed antigen to B cells and modulate B-cell responses . Immature DCs in peripheral tissues are thought to be the predominant DC population in the immune steady-state. They are highly endocytotic but not activated in terms of immunostimulatory capacity. Full maturation or activation depends on exposure to pathogens, inflammatory cytokines, or necrotic cells . Only activated DCs can induce a T-cell response, and different subsets of DCs appear to be responsible for induction of either tolerance or immunity . At this stage, it is unclear whether functional division among DC subsets reflects cell specialization during development from progenitors or functional plasticity of fully differentiated DCs. The answer to this question depends on direct analysis of the development of DCs from progenitors. While there are several ways to study DC development, the procedures involved are only possible in animal models. This review considers current approaches to the study of lineage commitment in DCs. Reference is made primarily to the murine DC lineage(s), although many similarities are known to exist between human and murine DCs .
Difficulties Associated with the Study of Dendritic Cell Development
DCs are scattered in most organs of the body, including lung, gut, skin, and lymphoid tissues. Unlike other leucocytes, these cells are not present in high numbers in blood. They are widely distributed in most peripheral tissues, but their numbers are so small as to make isolation an extremely difficult task. There are very few markers specific for the DC lineage, although the CD11c marker in mouse has been particularly useful . Cells are identified on the basis of combined expression of a number of markers common to other leucocytes. Langerhans cells in skin represent a distinct lineage of DCs and are identifiable by expression of Langerin . It is well accepted that DCs in peripheral tissues, including Langerhans cells in skin, are immature DCs that take up antigen and mature as they traffic to lymph nodes where they encounter T cells. Lymphoid tissues also contain endogenous populations of immature DCs that are distinguishable from antigen-carrying DCs entering from peripheral tissues . It is not yet known whether endogenous immature DCs in lymphoid tissues differ from immature DCs in peripheral tissue sites. Cells in both sites could have an important role in the maintenance of peripheral tolerance, but it is not known exactly how cells differentiate to acquire those distinctly different functional properties. While endogenous spleen DCs are thought to be blood-derived, at this stage it is unclear whether they undergo differentiation within the spleen from endogenous progenitors or arrive in a more mature state, perhaps as blood-derived DC precursors [10,11].
Many labs have developed procedures for fractionation of DC subsets of increasingly higher purity. Methods for cell isolation depend on known cell properties or known cell surface marker expression and are fraught with problems of cell contamination . Another problem is that once cells are isolated from tissue and exposed to in vitro culture, they immediately become activated, which can lead to alterations in functional capacity and marker expression . For experimentation, isolated cell fractions enriched for DC precursors are commonly expanded in vitro by culture in medium containing a cocktail of growth factors, usually containing granulocyte-macrophage colony-stimulating factor (GM-CSF), stem cell factor (SCF), and tumor necrosis factor alpha (TNF-α ). However, cytokine-supported cultures yield a heterogeneous mixture of cells, including DCs and in some circumstances monocytes and macrophages. Most cytokine-supplemented cultures are not suitable for studying the intermediate stages of DC development, since progenitors are quickly driven to become mature DCs, which undergo limited proliferation and survive for only a short period of time. These in vitro culture conditions do not maintain progenitor cell populations and provide little opportunity to study intermediate stages in development. The pathways that connect intermediate stages in DC development with progenitors and mature cells remain difficult to analyze by studying the population of cells resulting from tissue fractionation or in vitro culture.
Ways to Study Dendritic Cell Hematopoiesis
The study of DC development from progenitors is inherently difficult because progenitors are present in tissues in very low numbers, and specific markers for progenitors have not yet been defined. Characterization of DC progeny is all the more difficult because it is dependent on the production of identifiable progeny DCs. Progenitor or stem cells are also intimately tied to a niche or microenvironment that provides a complex arrangement of soluble factors, as well as cell–cell and cell–matrix interactions that are essential for maintenance of stem cells and their eventual differentiation [13,14]. The microenvironments in which DCs develop are, so far, undefined. It is likely that a number of signaling events apart from cytokines combine to drive progenitor cells to differentiate into DCs. While growth factors like Flt3Land GM-CSF can support DC development both in vivo and in vitro, these cytokines do not have an essential or even a specific role in lineage commitment [15,16]. The study of progenitors requires a suitable model system in which to study their differentiation into functional DCs. There are three main approaches to the study of hematopoiesis: subset analysis, lineage analysis, and production of cells in long-term cultures. A more thorough understanding of developmental pathways could stem from the application of all three approaches.
Subset analysis involves identification of cells on the basis of differential expression of multiple cell surface markers and predictions about their lineage relationship. This is usually the first approach, and it uses isolated cells, flow cytometry, and any available antibodies. Multiple DC subsets have been identified in murine lymphoid tissues on the basis of expression of markers including CD11c, CD11b, CD205, CD80, CD86, B220, MHC-II (major histocompatibility complex-2), CD40, CD8α, and CD4. Currently there are three known subsets in spleen , two in thymus , and five in lymph node . More recently, the distinct lineage of murine plasmacytoid DCs was identified in multiple organs on the basis of B220 expression . DC subsets in murine spleen are located in two distinct locations: some are located in the marginal zone, and others, notable by expression of CD205 and CD8α, are located in T-cell areas . Marker combinations have been identified which can distinguish Langerhans cells, thymic lymphoid DCs, plasmacytoid DCs, and the distinct populations of myeloid CD8a+ and CD8a−DCs. Definition of the functional capacity of the many known DC subsets lags behind phenotypic definition. Furthermore, the developmental origin of each of the different subsets is one of the most controversial areas of DC biology. To date, very little is known about the lineage relationship among these DC subsets.
Lineage analysis involves ex vivo isolation of a cell subset containing progenitor cells and transfer of these into a syngeneic host that may have been sublethally irradiated to destroy host hematopoietic cells. This can be done effectively if donor and host express different allelic markers delineating the different cells. Progeny cells are identified by antibody staining to detect cell surface markers that indicate the lineage of donor-derived cells. The drawback with these studies is that information obtained is limited by knowledge of the starting cell population. It is difficult to clearly delineate lineage when the progenitor population contains more than one progenitor type. The most definitive experiments involve isolation of a pure starting population of otherwise rare cells. Another requirement is that a sufficient number of progeny cells is produced so that it is possible to detect them with certainty among the cells of the host. Using the lineage approach, both DCs and T cells were shown to develop in thymus from thymic CD4lo lymphoid precursors following intrathymic injection of cells into syngeneic host mice . This finding led to the hypothesis for a separate lymphoid-like lineage of DCs in thymus. Recently this approach was used to demonstrate plasticity in DC development. DCs of different phenotype, previously thought to be representative of distinct myeloid and lymphoid lineage subtypes, could be derived from both the common myeloid and the common lymphoid progenitor subsets in bone marrow following transplantation into host mice . This study predicted the existence of a common DC precursor derived from both myeloid and lymphoid progenitors and questioned hypotheses about the existence of separate myeloid and lymphoid lineages of DCs.
The third approach to cell development is the study of hematopoiesis in long-term stroma-dependent cultures. In these cultures, hematopoiesis is dependent on the establishment of a layer of stromal cells that support stem cell survival, self-renewal, and differentiation . Hematopoiesis was first achieved in vitro in long-term bone marrow cultures in which granulopoiesis predominated [23,24]. Whitlock et al.  also developed a long-term culture (LTC) system that supported production of cells representing early stages in B-cell development. Similarly, thymic organ cultures have been used to study the development of T cells in contact with the thymic stromal cell environment . In each of these culture systems, early progenitors are maintained in a complex cellular environment conducive to lineage commitment. The main advantage of these cultures is that the microenvironment supports self-renewal and maintenance of a population of progenitor cells which would otherwise be impossible to isolate. Long-term cultures mimic the normal developmental process within tissue niches. The niche established in a long-term culture supports a variety of interactions between stem cells, stromal cells, extracellular matrix, and growth and differentiation factors . Cells develop under more normal physiological conditions than in cultures supplemented with cytokines alone. This approach has limitations in that it represents in vitro development and any findings need to subsequently be tested in vivo. However, cell development and function can be observed in vitro, and cells can be isolated for study at various time points.
This lab has developed a long-term culture system that produces DCs. While some success has been obtained in establishing cultures from cells derived from a number of lymphoid sites, by far the most productive long-term cultures have been those derived from spleen [28,29]. Areadily identifiable common class of immature DCs has been found to develop in spleen long-term cultures [28–32]. The ease with which these cultures are established and their consistent production of DCs of a common phenotype suggest that an equivalent normal process of DC hematopoiesis may occur in spleen. Our prediction is that spleen contains DC progenitors that develop into DCs in situ and are supported by stromal cells and other factors. An endogenous population of spleen immature DCs could develop from progenitor cells resident in spleen. This hypothesis is under further test.
The preparation of long-term cultures involves culturing a suspension of cells prepared from dissociated lymphoid tissue of young mice. This ensures maintenance of all stromal and hematopoietic cell types present in the original tissue. Lymphoid tissues from many different strains of mice have been used successfully to generate long-term cultures. However, spleen cultures are unique in their capacity to continuously support the production of DCs . Within a week of culture, stromal cells are adhered to the tissue culture flask and begin to slowly divide to produce a confluent cell layer. The stroma comprises a mixture of cell types that is characteristic of the starting tissue type . For example, thymic stroma is quite distinct from spleen stroma in that it contains epithelial cells. Spleen stroma, however, comprises a mixture of fibroblasts and endothelial cells with some adherent DCs and macrophages detectable in early cultures. The presence of endothelial cells has been confirmed by staining cells with antibody specific to Factor VIII–related antigen . Spleen long-term cultures were the easiest to produce and became immediately of interest when it was discovered that they yielded a population of cells resembling immature DCs. Spleen stromal layers also support secondary LTC and the outgrowth of allogeneic DCs from overlaid bone marrow . Most of our work has therefore concentrated on the characterization of spleen long-term cultures. Cells reflecting various stages in development can be identified in these cultures (Fig. 1). Aphotomicrograph showing the cells in an early LTC is shown in Figure 2, and adherent large endothelial cells and fibroblasts are evident (Fig. 2A, B). Foci of dividing hematopoietic cells are attached to the stromal cells, and scattered hematopoietic cells are loosely attached before they are released into the supernatant. In Figure 2C, nonadherent cells released into supernatant have the morphological characteristics of DCs with long membrane extensions, or pseudopodia.
Cells are cultured in medium without the addition of cytokines commonly used to expand DCs from isolated precursor populations. Supplemented Dulbecco's Minimal Essential Medium containing 10% endotoxin-free fetal calf serum is used routinely . Within 3 weeks of establishment of spleen cultures, stroma becomes confluent and most mature lymphoid cells and erythrocytes have died off. Frequent medium change is needed between 2 and 3 weeks of culture to dilute out debris from dying cells . Foci of small round dividing cells then become detectable above the stroma (Fig. 2A, B) . Over time, larger round cells appear above the foci and are then released into the culture medium (Fig. 2C). These show the characteristic small-membrane extensions of developing DCs. Large, granular DCs with long pseudopodia are also detectable in the medium. The establishment of productive cultures requires selection of cultures with an appropriate mixture of both fibroblastic and endothelial cells from an early stage. The relative contribution of these two cell types to DC hematopoiesis is not yet completely understood but is under further investigation. The characteristics of cloned stromal cell lines are under investigation for functional capacity to support DC differentiation, and gene profiling is being used to confirm cell types. A productive 30-ml long-term culture will yield 0.5–1.0 × 106 nonadherent DCs every 48 hours. A homogeneous population of cells is produced in long-term cultures, and no further enrichment of cells is needed to obtain DCs.
For experimentation, culture flasks are gently rocked and nonadherent cells are collected by pipetting off supernatant without disturbing the stroma. The longevity of the culture is maintained by passaging both stromal cells and nonadherent cells into new tissue culture flasks every 2–3 months. A small section of stroma is scraped from the donor flask. Medium containing stromal and nonadherent cells is transferred to a fresh flask with fresh medium. Subcultures then take 3–4 weeks to form a stromal layer and to become productive. DCs can be collected from a single long-term culture line for experimental use over a period of years if cultures are maintained by passage . Passage of cultures is necessary to prevent fibroblast overgrowth in the culture and to maintain progenitors that are essential for continuity of long-term cultures. Nonadherent cells alone seeded into a flask of medium will not survive beyond about 7 days [28, 29, 32]. Passage of the stromal cell layer is essential for maintenance of the culture.
One can argue that stroma-supported hematopoiesis in vitro simulates conditions much closer to normal physiological conditions than do cultures supplemented with high concentrations of growth factors. Long-term spleen cultures are unique in terms of DC production because they rely entirely on endogenously produced growth factors. While soluble factors are known to be important in this system , GM-CSF and TNF-α, commonly used for in vitro generation of DCs from isolated precursors, are not produced in LTC [28,32]. Consistent with this is evidence that DCs that produce long-term cultures can be derived from GM-CSF−/−mice . Our hypothesis is that steady-state or homeostatic DC development in spleen LTC occurs in the absence of GM-CSF. Addition of GM-CSF does not increase the production of DC from progenitors, although it has been shown to amplify the number of large DCs in culture . This finding is consistent with a role for GM-CSF in an alternative pathway for development. In fact, GM-CSF could be an important factor for DC development under inflammatory conditions. The important role for inflammatory factors like GM-CSF, SCF, and TNF-α in DC development is supported by the many studies that effectively use the factors to amplify DC numbers for immunotherapy. However, this should not be taken to mean that cytokine-induced cultures are the only or most effective way to produce DCs for clinical or immunotherapeutic use .
Characteristics of Dendritic Cells Produced in Long-Term Spleen Cultures
Nonadherent cells produced in long-term cultures have been characterized over many years as DCs on the basis of multiple parameters, including morphology, cell surface phenotype, and antigen-presenting capacity [28–30]. The size distribution of nonadherent cells collected from long-term cultures has also been measured using flow cytometry to record light scatter properties of cells. LTC produces two clear subpopulations of “small” and “large” cells, as defined by Forward and Side scatter profiles . These have been referred to as small LTC-DC and large LTC-DC (Fig. 3). While the division into subsets on the basis of size appears arbitrary, it represents the best division to delineate a phenotypically homogeneous population of fully differentiated large-sized DCs. The small-cell subset is not so homogeneous and reflects less-differentiated cells with only some cells expressing the markers of DCs .
The cell surface phenotype of small and large cells produced in long-term cultures has been analyzed using flow cytometry and antibodies specific for a range of surface markers. The collective data from many different long-term lines maintained over a 4-year period are shown in Table 1. These data clearly delineate the phenotype of the small and large cells produced in long-term culture and also demonstrate the consistency in the percentage of positive cells and mean fluorescence achieved over many repeat experiments. The cell population produced in long-term cultures has remained remarkably constant in terms of cell size and surface marker expression over many years and over many lines established from different strains of mice .
Table Table 1.. Marker expression on small and large subsets of cells produced in long-term cultures
aSmall and large LTC-DCs were delineated by flow cytometry using postacquisitional gating.
bThe percentage of positive-staining cells was calculated by subtracting background from specific antibody staining. Background was obtained by replacement of primary antibody with medium or isotype control antibody.
cShift in median fluorescence intensity of positive cells relative to background (signal-to-noise ratio). This was calculated to allow for variation in voltage amplification used over the course of many experiments.
dData represent mean ± standard error across multiple experiments.
The majority of cells produced in long-term cultures are large-sized cells with a clear DC phenotype of CD11c+CD11b+CD80+CD86+MHC-II−/lo (Table 1; Fig. 3). Absence of CD40, B200, and CD8α–reflects precursor or immature myeloid-like DCs (Table 1) [33, 36, 37]. Numerous tests have shown these cells to be highly endocytotic and immunostimulatory for T cells [31, 33, 37]. Small DCs represent a minor subset of cells expressing variable levels of CD117 (c-kit), CD80, CD11c, and CD11b. They are only weakly endocytotic and do not stimulate T cells . A minor (1%–3%) subset of large cells produced in LTC is MHC-II+ and is thought to represent DCs activated in culture . Large LTC-DCs collected from stromal cultures can be weakly activated if they are cultured directly on plastic surfaces. This leads to upregulation of CD86 and MHC-I, although not of MHC-II . These cells also respond weakly to activation with known DC-activating agents like lipopolysaccharide (LPS), TNF-α, and CD40 ligand. While these factors typically induce upregulation of CD86, MHC-I, CD44, and MHC-II in bone marrow–derived or monocyte-derived DCs, LTC-DCs respond to these particular stimuli by upregulation of MHC-I and CD86 but not MHC-II or CD80 [37–39]. The significance of this response is under further investigation. One hypothesis is that DCs in different sites may have different functional capacity with respect to activation consistent with exposure to different types of pathogens or activators. This would be consistent with site-specific immune responses and compartmentalization of the immune system to best meet the invasion of a range of different pathogens.
In relation to known DC subsets, large LTC-DCs represent fully developed myeloid DCs that are unusual in that they do not express MHC-II and CD40. An in vivo equivalent cell type can be characterized in mouse spleen (unpublished data). In terms of cell surface phenotype, these cells resemble recently described murine CD11c+CD11b+MHC-II−CD40−blood DC precursors (Fig. 3) [10, 11, 40]. A similar subset of DCs in spleen has also been described, which can activate marginal zone B cells to respond to bloodborne bacterial antigen . DCs produced in long-term cultures are phenotypically and functionally distinct from the “gold standard” CD11c+CD11b+MHC-II+ DCs generated in cytokine-supported cultures. The commonly described DC subsets isolated by cell enrichment from the spleen and from lymph nodes are also different in that they have high MHC-II expression. Indeed, many of the procedures used to isolate DCs would eliminate subsets of cells resembling LTC-DCs on the basis of no MHC-II expression. However, it is difficult to determine from combined studies whether MHC-II levels are low to medium or high on spleen DC subsets, and, in fact, contradictory reports exist [9,42]. DCs produced in the LTC system described here also differ from other cytokine-dependent DC-producing lines, like the D1 line described by Rescigno et al. . D1 cells are myeloid DCs that express high levels of immunostimulatory markers, including MHC-II. These levels also increase after activation of cells with LPS and other bacterial activators . Low expression of MHC-II in LTC-DC could be a result of in vitro culture conditions, however. LTC-DCs have been shown to synthesize the MHC-II invariant chain by reverse transcription polymerase chain reaction (RT-PCR) (unpublished data). It is possible that with their extremely high endocytotic capacity, cells lose surface MHC-II due to rapid turnover of the plasma membrane. DCs with a high rate of endocytosis have been shown to reabsorb and degrade MHC-II/peptide complexes faster than weakly endocytotic DCs .
Spleen also contains subsets of DCs that resemble LTC-DCs in terms of their absence of the CD8α and CD205 markers . Low to nondetectable expression of CD205 in LTC-DCs is indicative of DCs residing outside the T-cell area of spleen [20,47], suggesting that LTC-DCs may represent a marginal zone population of DCs. CD8α was originally thought to be a stable marker of spleen DC subsets, but it is now thought to fluctuate with activation and differentiation of cells. CD8α–DCs can develop into CD8α + DCs  in a process that also involves upregulation of CD205 and is thought to be associated with maturation and movement of DCs into the T-cell areas. Alternatively, there is also evidence for a CD8α + blood precursor of spleen CD8α +DCs, which suggests that separate lineages of CD8α +cells may exist .
Large DCs produced in long-term cultures also express high levels of the FcγII/III receptor and are highly endocytotic [34 and unpublished data]. This is consistent with immature DCs that have a high capacity to endocytose antigen in preparation for processing and presentation to Tcells . The capacity of cells to present antigen leading to T-cell activation has been demonstrated in vitro using both the conalbumin-specific D10.G4.1 Th cell clone  and hen egg lysosome (HEL)–specific T cells isolated from T-cell receptor 3A9 (TCR)-transgenic (Tg) mice [37,38]. Since CD4+T cells derived from 3A9 TCR-Tg mice are naïve, LTC-DCs have a distinct antigen-presenting capacity for unprimed T cells, which is a defining property of DCs. LTC-DCs pulsed with tumor cell membranes can induce a specific antitumor CD8+ cytotoxic T-cell response following adoptive transfer into mice . This response has also been shown to be protective and to reduce mortality among tumor-bearing mice.
DCs derived from most long-term cultures have no or weak capacity to stimulate allogeneic T cells in a mixed lymphocyte reaction [31 and unpublished data]. It is possible that incapacity to stimulate T cells in mixed lymphocyte reaction (MLR) is related to low expression of immunostimulatory molecules on cells. While LTC-DCs have high expression of CD80, CD86, and MHC-I, they have low or no expression of other immunostimulatory markers such as MHC-II, CD40, and CD40L. Some DC subsets have been shown to induce tolerance rather than immunity . It is not yet known whether the limited CD4+ T-cell stimulation by LTC-DCs is immunogenic or whether it leads to abortive T-cell proliferation and apoptosis of cells, but this is under further investigation. However, this inability to stimulate an MLR could be due to the immature nature of LTC-DCs and a lack of MHC-II expression on most cells. There could also be differences in the activation threshold of T-cell clones or TCR-Tg T cells, compared with the heterogeneously responding T-cell population used in MLR. Recent studies have also demonstrated that the response in T cells generated by DCs in an allogeneic MLR may depend on the presence of syngeneic DCs that present allogeneic antigens to the responding leucocyte population . Another consideration is that LTC-DCs have a distinct functional capacity for stimulation of B cells or natural killer cells, in line with their location in spleen and exposure to bloodborne pathogens .
In summary, DCs produced in long-term spleen cultures represent immature or precursor CD8α–CD205−/lo DCs. These could be representative of marginal zone type DCs that show some weak expression of CD205. Long-term cultures could provide an environment that promotes DC differentiation to the immature DC state but limits development of cells whose end result might well have been to reside in T-cell areas. Alternatively, low expression of CD205 could place these cells on the periphery of T-cell areas. As early immature or precursor DCs lacking expression of MHC-II and CD40, they could also represent a DC subclass with distinct functional capacity for mediating early natural killer or B-cell immune responses. These hypotheses are under further test.
Cultured Stromal Cells Support Dendritic Cell Development from Progenitors
Very little is known about stromal niches that support DC development. In long-term culture, spleen stroma provides both cell–cell contact and growth factors that support DC development . Assays for common growth factors released into the supernatant of long-term cultures have detected only interleukin-6 (IL-6) and no factors like GM-CSF and TNF-α , which are known to support the production of DCs under different conditions. However, spleen stroma has the specific capacity to support only DC development, and no hematopoietic cell types other than DCs are detectable in long-term cultures [28–30]. The unique importance of spleen stroma for DC development is evident, since it not only supports DC production from spleen progenitors but also from bone marrow cells overlaid onto stroma . The developmental potential of cells produced in long-term cultures has also been analyzed in colony assays in the presence of various known hematopoietic growth factors. Again, colony assays supported development of only DCs and no other lymphomyeloid cells . A small number of stromal cells was found necessary for production of large DC colonies, while conditioned medium from long-term spleen cultures supported proliferation of many small clusters of DCs.
Cell–cell contact between small progenitor cells produced in long-term cultures and the stromal layer is essential for DC development. Production of DCs ceases when small cells are overlaid in a transwell above stroma, which prevents cell contact [33,34]. Direct contact with stroma is necessary for the proliferation and differentiation of small progenitor cells into DCs. In contrast, large LTC-DCs are dividing cells that do not need contact with the stromal layer for survival , although the stroma does provide soluble factors that prolong the lifespan of large cells . When isolated small cells from long-term cultures are cocultured with stromal cells, the small-cell population does not self-renew. Instead, small cells differentiate into a population of large-sized DCs within about 15 days , and the small-cell population disappears . While the nonadherent small-cell population produced in long-term cultures does contain DC progenitors, these are not capable of self-renewal. The prediction is that self-renewing progenitor cells are either adhered to or resident within the stromal cell monolayer  and not present in the nonadherent cell population collected at medium change. Recently the HJC cell line was isolated from a long-term culture that began to produce large DCs at a higher rate. The naturally transformed cell line was obtained after many passages to grow large cells independently of stroma. This DC cell line retains all the phenotypic and functional characteristics of immature DCs produced in spleen long-term cultures (unpublished data). It is under further investigation for functional capacity.
Early experiments involved stromal layers from long-term cultures that had been washed free of hematopoietic cells and irradiated to prevent cell proliferation . Subsequent experiments have used the spleen stromal cell population, ST-X3. This mixed cell line was isolated from a long-term culture that had been selectively washed free of hematopoietic cells to derive a line which had lost hematopoietic progenitor cells and no longer produced DCs [33,34]. ST-X3 is a mixed cell population of endothelial and fibroblastic cells and supports the outgrowth of DCs from unfractionated bone marrow and spleen cells within 4 weeks of coculture [29,30]. These are termed secondary long-term cultures. Many cloned lines derived from ST-X3 are now under analysis for functional capacity to support DC development. Gene profiling using Affymetrix (Santa Clara, CA) microarrays has been used to detect gene expression in ST-X3 spleen stroma. The differential expression of genes between ST-X3 and an unrelated lymph node stromal population 2RL22 will be used to delineate genes of importance for DC development. 2RL22 is a mixed stromal cell line derived from a long-term culture established from lymph nodes and does not support DC development . Gene profiling will subsequently be used to identify the cell type of cloned stromal lines that do support DC differentiation.
Gene Expression during Dendritic Cell Development in Long-Term Cultures
The spleen long-term cultures described here are ideal for the study of differential gene expression, since progenitors and their progeny are maintained within the same culture environment. Subtracted cDNA libraries have been generated; these contain sequences that are differentially expressed in either “small minus large” or “large minus small” subsets of LTC-DCs . Differential screening was used to select clones expressed in either the small or the large LTC-DC populations for sequencing and identification. Known genes isolated from subtracted libraries relate to different stages in DC development and support previous findings regarding the function of the small and large cell subsets (Fig. 3) . This lab has had very good success with this procedure, with no overlapping clones detected in the two cell subsets. Large LTC-DCs express a number of immunologically important genes, including CD86, CCR1, osteopontin, and lysozyme. Small LTC-DCs resemble progenitors expressing genes relating to organization of the cytoskeleton and regulation of antigen processing. Novel transcripts have been isolated from both small and large LTC-DC–subtracted libraries that could encode novel proteins important in DC development.
The individual small- and large-cell subsets are also highly suitable for gene profiling using Affymetrix microarrays. Multiple cell sorts have been performed to isolate subsets of small and large cells from long-term cultures and RNA isolated for probe preparation. Procedures have been developed to prepare labels from small quantities of RNA isolated from the rare small-cell subset. This type of analysis should identify genes that are differentially expressed between progenitors and DCs generated in spleen long-term cultures and should help to identify factors and genes important in early DC development.
Conclusion: A Model for Dendritic Cell Development in Spleen
The replicative and developmental potential of cells produced in long-term spleen cultures has been investigated as a model for production of DCs. The model predicts the in situ production of DCs in spleen from endogenous progenitors and identifies a central role for spleen in DC hematopoiesis. While other models are possible, they depend on a continuous influx of DC precursors into spleen from blood. Our preliminary evidence, however, favors the existence of a self-renewing DC progenitor within spleen. The identity of a resident spleen DC progenitor is not yet clear. An endogenous population of immature DCs in spleen has recently been distinguished from DCs entering spleen from peripheral sites . Amodel for DC development is shown in Figure 4, which incorporates the four distinct hematopoietic cell types identified in long-term cultures: a self-renewing stem cell, the small-cell subset of DC progenitors, a large population of immature DCs, and a small population of mature or activated MHC-II+ DCs.
Most research to date on spleen DCs has focused on the mature MHC-II+ cells and the concept that maturing DCs migrate into spleen from the periphery after encounter with antigen and activation by inflammatory factors. This study focuses instead on early DC development and emphasizes the need to use in vitro methods of hematopoiesis in order to study the differentiative process within lymphoid tissues. The essential role for stromal cells is identified, and the potential exists using cloned stromal lines to identify cell-adhesion molecules, soluble factors, and signaling cascades that are essential to early DC development. DCs produced in long-term cultures are clearly immature or partially mature DCs and are quite distinct from DCs in peripheral tissue sites, which express MHC-II and CD40 and which migrate to lymph nodes to present antigen to T cells. A resident immature spleen DC subset may be expected to express a different range of genes and receptors to peripheral DCs located in tissues. Development of this distinct subset may depend on the presence of a unique niche environment in spleen-containing stromal cells that support stem cell maintenance, self-renewal, and DC differentiation. Several different stromal cell types may contribute to the niche, and a unique set of signaling pathways may operate to maintain cell production.
One model proposes the presence of a distinct, endogenous DC subset in spleen that functions specifically to control bloodborne antigen. This immature DC subset would be generated from self-renewing progenitors maintained within the spleen environment. In the steady-state these cells could function to maintain tolerance to bloodborne self-antigens. They could also have a distinct role in initiating early-phase responses such as the T-independent antibody response or the natural killer cell response. Upon activation with environmental antigens, these DCs could mature, migrate into T-cell areas, and become immunostimulatory for T cells. The HJC cell line representing large LTC-DCs has recently been isolated as a continuous stroma-independent cell line that maintains all phenotypic characteristics and known functions of large LTC-DCs. This cell line should be invaluable in future studies to investigate the functional capacity of spleen-derived DCs.
LTC provides a means to generate DCs that are immature. Most other procedures for DC culture involve procedures and cytokines that lead to activation of DCs. Activation heralds the end of DC differentiation, and so only limited cell replication is possible. While questions still remain unanswered about the immunogenic capacity of DCs produced in LTC, this system has advantages over other procedures for generating DCs in that cell production is continuous and long term. In the clinical setting one could envisage establishment of secondary LTC involving DC progenitors isolated from blood or bone marrow cocultured with an established allogeneic stromal cell line. DCs produced continuously in this system could be used over time for intermittent immunotherapy of tumor patients. In the least, DCs could be exposed to tumor membranes and then adoptively transferred for induction of a protective cytotoxic T-cell response as achieved previously in a mouse tumor model .
This work was supported by grants to H.O. from the National Health and Medical Research Council of Australia, the Australian Research Council, the Clive & Vera Ramaciotti Foundation, and the Australian National University Faculties Research Grant Scheme. H.W.and B.Q.were supported by Australian Postgraduate Awards. J.A.was supported by an ANU Graduate Scholarship. G.D. was supported by a scholarship from the Fonds Nature et Technologies-Fonds de laRecherche en Santé du Québec.