|Home | About | Journals | Submit | Contact Us | Français|
R.S.W. and B.L.E. contributed equally to this work.
Changes in cell surface markers and patterns of gene expression are commonly used to construct sequences of events in hematopoiesis. However, the order may not be as rigid as once thought and it is unclear which changes represent the best milestones of differentiation. We developed a fate mapping model where cells with a history of RAG-1 expression are permanently marked by red fluorescence. This approach is valuable for appreciating lymphoid lineage relationships without need for irradiation and transplantation. Hematopoietic stem cells (HSC), as well as myeloid and dendritic cell (DC) progenitors were unlabeled. Also as expected, most previously identified RAG-1+ early lymphoid progenitors (RAG-1+ ELP) in bone marrow and all lymphoid affiliated cells were marked. Of particular interest, there was heterogeneity among canonical common lymphoid progenitors (CLP) in bone marrow. Labeled CLP expressed slightly higher levels of IL-7Rα, displayed somewhat less c-kit and generated CD19+ lymphocytes faster than the unlabeled CLP. Furthermore, CLP with a history of RAG-1 expression were much less likely to generate dendritic and NK cells. The RAG-1 marked CLP were lineage stable even when exposed to LPS, while unlabeled CLP were re-directed to become dendritic cells in response to this TLR4 ligand. These findings indicate that essential events in B lymphopoiesis are not tightly synchronized. Some progenitors with increased probability of becoming lymphocytes express RAG-1 while still part of the lineage marker negative Sca-1+ c-KitHi (LSK) fraction. Other progenitors first activate this locus after cKit levels have diminished and cell surface IL-7 receptors are detectable.
Many studies have investigated precursor-product relationships between hematopoietic cells, but unifying models have proven difficult to construct. In part, this is because of the absence of standardized nomenclature, but recent studies suggest there is also substantial flexibility in the pathways by which various cell types are made. In fact, hematopoietic processes in normal animals may be dramatically altered during disease. For example, we found that stem and progenitor cells express functional Toll-like receptors (TLR), and encounter with pathogen products such as lipopolysaccharide (LPS) elicits a series of responses (1, 2). Stem cells are driven into cycle, myeloid progenitors become less dependent on cytokines and common lymphoid progenitors (CLP) are directed to become dendritic cells. This might represent a protective mechanism whereby innate effector cells are rapidly replenished at the expense of lymphopoiesis in response to pathogen products. However, that remains to be shown, and it is equally possible that pathogen products are harmful to marrow functions.
Transplantation has been a powerful tool for determining differentiation potential of hematopoietic cells, but the normal equilibrium is disturbed, microbes can be released into the blood stream and homing patterns are altered when conditioning irradiation is used (3, 4). Furthermore, there has been little way to monitor pathway shifts in real time and no means to establish the genealogy of mature cells in the immune system. For example, dendritic cells can derive from lymphoid and non-lymphoid progenitors but the extent of utilization of alternate pathways in steady state conditions has not been explored (5-7). Consequently, a non-invasive fate mapping model that can be used with intact animals would have substantial utility. RAG expression, immunoglobulin locus accessibility and V(D)J recombination are sequential hallmarks of lymphoid lineage progression(8). Experimental models such as RAG-1/GFP knock-in mice or recombination substrate transgenics have exploited these events but leave gaps in our knowledge (9, 10). For example, it remains unclear whether the early expression of RAG-1 in uncommitted progenitors always reflects a restriction in developmental potential or, instead, represents promiscuous lineage priming without constraining developmental outcome.
We now report that RAG-1/Cre X Rosa26 tandem dimer red fluorescent protein (tdRFP) mice can be used to address these issues, as cells with a history of RAG-1 expression can be readily identified. They provide a valuable index of lymphoid relatedness in a variety of cell types. CLP in bone marrow were determined to be heterogeneous with respect to current or previous RAG-1 expression, and the unmarked CLP represented good sources for dendritic and NK cells. These differentiation options were greatly diminished in the labeled CLP, a subset that was heavily biased to become B lymphocytes, while LPS preferentially redirected the less mature of the two types of CLP to become dendritic cells. The findings reveal patterns of immune system replenishment in healthy animals and illustrate how alternate differentiation pathways can be utilized when environmental circumstances change.
RAG-1/Red mice were created by crossing RAG-1/Cre mice (129 background) to animals with tandem-dimer red fluorescent protein (tdRFP) knocked into the Rosa26 locus (C57BL/6, B6 background) (11, 12). Wild-type 129/B6Rosa26 F1 mice were used as negative controls. All animals were bred and maintained in the Laboratory Animal Resource Center at the Oklahoma Medical Research Foundation (Oklahoma City, OK). All experimental procedures were conducted under approved IACUC protocols.
Wild-type and RAG-1/Red mice were injected with 100 μg LPS (O55:B5, Sigma, St. Louis, MO) in 200 μL PBS. Isolation of progenitors and analysis of hematopoietic tissues by flow cytometry was performed one week later.
Cell isolations were performed in HBSS with 5% FCS. Marrow cells were harvested from the femurs, and erythrocytes were lysed by briefly resuspending in NH4Cl− hypotonic solution. Isolation of progenitor populations for culture was done as follows: BM cells were enriched with negative selection by labeling marrow with Ly6G+C/Gr-1 (RB6-8C5), CD11b/Mac-1, TER-119, CD3 (17A2), CD8 (53-6.7), CD19 (1D3), B220 (14.8), and then immuno-magnetically depleted with the BioMag goat anti–rat IgG system (Qiagen, Valencia, CA). All cells were treated with Fc-receptor block (2.4G2) prior to fluorescent staining and FACS sorting. After staining marrow with biotin-anti-lineage markers APC-Cy7 anti-cKit (2B8), PE-Cy5 anti-Flk2/Flt3(A2F10), Pacific Blue anti-Sca-1 (D7, eBioscience, San Diego, CA), (A7R34, eBioscience), PE-Cy7 anti-CD150(TC15-12F12.2) and PerCp Cy5.5-anti-IL-7Rα (Biolegend), biotin-anti-CD27 (LG.3A10), PerCP-Cy5.5 anti-CD34 (HM34, Biolegend), and secondary streptavidin PE-Texas Red; (Caltag Laboratories, Burlingame, CA), Lin− populations were sorted using either a MoFlo (DakoCytomation, Ft. Collins, CO) or FACSAria cytometer (BD Biosciences, San Diego, CA) into specific populations. CLP were sorted as Lin−IL-7Rα+cKitLoSca-1+, and subdivided with respect to tdRFP labeling. Myeloid progenitors were sorted as Lin− Sca-1−ckitHi and further resolved into CMP (CD34+FcγRII/IIIlo/-) and GMP (CD34+FcγRII/III+).
Dead cells were excluded in FACS by propidium iodide staining (Molecular Probes, Eugene, OR). Subset purification was achieved by double sorting, and confirmed by post-sort analysis. All antibodies came from BD Pharmingen, unless otherwise stated. Isotype control stains were used for gating. Flow cytometry analyses were performed on a BD LSRII (BD Biosciences, San Jose, CA), and FlowJo software (Treestar, San Carlos, CA) was used for analysis.
The mRNAs were isolated from sorted cells by using MicroPoly(A) Purist (Ambion, Austin, TX). The cDNAs were then prepared from DNaseI - treated mRNA by using oligo-dT and Moloney murine leukemia virus reverse transcriptase (Invitrogen). Reactions were quantified with the fluorescent TaqMan® technology. TaqMan primers and probes specific for murine rag1 and GAPDH were used in the ABI7500 sequence detection system (Applied Biosystems, Germany) using TaqMan Universal PCR Master Mix (Applied Biosystems). Reactions were run at an annealing temperature of 60°C with 40 cycles. Each sample was measured in triplicate and comparative Ct method was used for relative quantification of gene expression.
Details of our culture procedures have been published (13). Briefly, sorted cells were cultured in 96-well plates (Corning, Inc.) with X-VIVO15 medium (Biowhittaker, Walkersville, MD) containing 1% detoxified bovine serum albumin (Stem Cell Technologies, Vancouver, Canada), 5 × 10−5 M 2-mercaptoethanol, 2 mM L-glutamine, 100 U/ml penicillin, and 100 mg/ml streptomycin. Culture medium was enriched with 100 ng/mL FL, 20 ng/mL SCF, 1 ng/mL IL-7, and 50 ng/mL IL-15 in combinations as indicated. DC-lineage promoting conditions included SCF- and FL-enriched medium without IL-7. Two-step NK-lineage culture conditions have been described previously(14). Briefly, cultures were initiated with CLP in medium with SCF, FL, and IL-7. Cells were washed at day 5, and IL-7 was replaced by IL-15 for the remaining 5 days. LPS (O55:B5, Sigma, St. Louis, MO) was added at a concentration of 10 μg/mL for the initial 48 hours of culture, before washing with plain medium and re-incubating with cytokine-enriched medium for the designated culture period. Incubation was maintained at 37°C in a 5% CO2 humidified atmosphere. Cells were fed by replacing half volume with fresh media and cytokines every three days. Harvested cells were stained with mAbs to lineage markers. Total cell yields were determined with Countbrite beads (Invitrogen, Carlsbad, CA) according to manufacturer protocol.
The Prism V3.02 software (GraphPad, San Diego, CA) was used for statistical analysis. Intergroup comparisons were performed with two-tailed, unpaired t-tests. P values were considered significant if less than 0.05.
Receptor gene recombination is essential for lymphopoiesis and reporter mice that reflect that process have been extremely valuable (9, 10). We constructed a new model by crossing previously generated RAG-1/Cre knock-in mice (11) with Rosa26tdRFP reporter mice (12) and used it to detect cells that either express RAG-1 or derive from RAG-1+ progenitors. Patterns of expression will be detailed elsewhere, but As expected, greater than 95% of mature B and T lymphocytes in all tissues but no more than 3% of granulocytes (CD19− CD3− CD11b+Gr-1+) in the marrow or peripheral blood were labeled in the RAG-1/Red mice (Figure 1). As shown below, the marking is stable because cells generated from marked progenitors continued to fluoresce. Lin−c-Kit+ cells were sorted according to red fluorescence and assessed for RAG-1 transcripts. These were 81 fold increased in RAG-1 mRNA relative to the non-fluorescent cohort as determined by rq-PCR.
Hematopoietic stem cells (HSC) rigorously defined as Lin−CD48-cKitHISca1+CD150+ were unmarked (less than one out of ~600 LT-HSC) in RAG-1/Red mice (Figure 2A). There was also no labeling among non-lymphoid progenitors (MEP, CMP and GMP) in the Lin−Sca-1−cKitHi fraction of bone marrow (Figure 2B). The HSC containing Lin− Sca-1+ c-KitHI (LSK) fraction contains primitive cells that express some lymphoid genes (9, 15-17). Our laboratory showed that CD27+ Flt3+ LSK early lymphoid progenitors (ELP), are potent with respect to lymphopoiesis and have greatly reduced, but detectable myeloid potential (9, 18). ELP are heterogeneous, expressing different combinations of terminal deoxynucleotidyl transferase (TdT), a human μ transgene or RAG-1 as revealed in a RAG-1/GFP knock-in model (Figure 2C). Consequently, we expected that permanently marked RAG-1/Red ELP would be a subset of the RAG-1/GFP+ cells. That was shown to be the case by crossing RAG-1/Red mice with RAG-1/GFP knock-in mice (Figure 2D). Approximately 90% of the RAG-1/GFP+ cells also fluoresced red. Furthermore, all Lin−cKitLoSca-1LoCD27+ pro-lymphocytes previously identified in RAG-1/GFP knock-in mice uniformly had the tdRFP marker (Figure 2E).
Down-regulation of cKit in bone marrow has long been known to correlate with progression to the B and NK lineages (9, 14, 18, 19). For example, Lin−cKitLo cells give rise to CD19+ lymphocytes much more quickly than cKitHi cells in culture or transplant and have less thymus colonizing potential (9, 20-22). Furthermore, this category includes Lin−IL-7Rα+cKitLoSca-1+ common lymphoid progenitors (CLP). We found that approximately half of these CLP had a history of RAG-1 expression (Figure 3A). This is similar to the percentage of CLP that are marked in RAG-1/GFP mice (9, 21). It has recently been shown that CLP can be sub-divided on the basis of Flt3 expression (23), but RAG-1/Red-marked CLP were heterogeneous with respect to that marker (Figure 3B). Additionally, Allman and Hardy have used CD93/AA4.1 as a defining feature of CLP (22, 24). A majority (~80%) of both marked and unmarked cells in RAG-1/Red mice were AA4.1+ (Figure 3B). Interestingly, the tdRFP+ CLP had lower densities of cKit and higher levels of IL-7Rα than their tdRFP− cohorts (Figure 3C).
These findings indicated that the new reporter mice are a powerful tool for identifying cells that derive from RAG-1 expressing progenitors. Marking begins among previously identified ELP and overlaps completely with Lin−cKitLoSca-1LoCD27+ prolymphocytes (9). However, only some of the extensively studied CLP category have a history of RAG-1 expression. Subsequent experiments were designed to explore developmental relationships between these two CLP subsets and compare their differentiation potentials.
The slightly reduced density of cKit and higher amounts of IL-7Rα seen on marked CLP in RAG-1/Red mice could indicate that they have progressed to a developmental status beyond the unmarked cohort. To investigate this possibility, we sorted the two subsets to high purity and tested them in defined, serum-free, stromal cell-free cultures. Most red fluorescent CLP up-regulated B220/CD45R in just 24 hours, while the two markers were not synchronously acquired by the non-red CLP cohort (Figure 4A). That is, some cells up-regulated B220 alone and others became red fluorescence without B220. Possible relationships between the subsets were also explored with time-course experiments. A comparison of CLP subsets in IL-7 containing cultures assessed their relative maturity. Total CLP from wild type mice and the red-labeled CLP from RAG-1/Red mice generated CD19+ lymphocytes in as little as five days (Figure 4B and 4C). The unlabeled cohort also generated B lineage lymphocytes, but only after longer intervals of culture. One week limiting dilution experiments revealed that the red fluorescent CLP were enriched for B lineage progenitors relative to the non-fluorescent cohort. That is, ~1:6 of the marked CLP had potential for generating B220+CD19+ cells as compared to ~1:14 of the unlabeled progenitors (data not shown, r2=0.923).
The tdRFP− CLP consistently produced more CD19− non-lymphoid cells than did the tdRFP+ CLP. This suggested that the unlabeled subset might have broader differentiation potential than its tdRFP+ counterpart. To address this point, our analyses were modified to assess production of additional lineages (Figure 5A-D). Specifically, omission of IL-7 prevents preferential expansion of lymphoid cells while staining with CD11b, and Ly6C allows rigorous discrimination of plasmacytoid dendritic cells (pDC) and myeloid cDC. In this model, red-labeled CLP expanded poorly and generated few pDC or cDC. In striking contrast, total CLP from wild type controls and cells with no history of RAG-1 expression produced CD19−B220+CD11c+Ly6C+ pDC (Figure 5A and B) as well as CD19−B220−CD11c+CD11b+ cDC (Figure 5C and D). When unseparated CLP from RAG-1/Red mice were used to produce pDC in culture, 18% of the newly generated pDC had a history of RAG-1 expression (data not shown). Addition of IL-15 with the removal of IL-7 on day 5 of culture promotes NK cell production from CLP12. With this experimental design, unlabeled CLP were much better NK cell progenitors than those with a history of RAG-1 locus activation (Figure 4E and F). Also included in this analysis were recently described Lin−cKit Lo Flt3+ CD115+ proDC/CDCP as discussed below (Figure 5, right panels).
Thus, there is no rigid sequence whereby Lin−IL-7Rα+cKitLo lymphoid progenitors must acquire B220 before becoming labeled by RAG-1 expression. The non-red CLP subset contains fewer B lineage progenitors than the fluorescent cohort and has most of the DC and NK potential. Although progenitors with clear evidence of RAG-1 locus activation were not totally lineage restricted, their differentiation options were much more limited.
Exposure of lymphoid progenitors to TLR ligands alters their differentiation (1, 2). The RAG-1/Red model made it possible to re-examine that issue and determine if the two subsets of CLP are equally directed to non-lymphoid fates in response to pathogen products. Interestingly, the tdRFP+ CLP were lineage stable, and the vast majority of their progeny were CD19+CD45R/B220+ B-lineage cells even after brief exposure (48hr) to the TLR4 ligand LPS (Figure 6A). In fact, yields of B lineage cells from this subset were reduced by only 10% (Figure 6B). It is noteworthy that the normally limited potential these progenitors have for cDC production was slightly increased by this TLR ligand (Figure 6C and D). Thus RAG-1 expression correlates with impaired ability to be directed to a non-B cell fate.
Results with the less mature tdRFP− subset of CLP were completely different. LPS caused a ~90% reduction in yields of CD19+ cells in cultures initiated with those progenitors (Figure 6A and B), while there was preferential generation of CD19−B220−CD11c+CD11b+ cDC (Figure 6C and D). Approximately one third of DC emerging in cultures initiated with these non-labeled CLP expressed tdRFP, and thus had a history of RAG-1 expression (data not shown). The unlabeled lymphoid progenitors made twice as many DC when exposed to this TLR ligand (Figure 6C and D).
The Lin−cKitLo fraction also includes potent Flt3+ CD115+ progenitors of dendritic cells, but the possibility that they are developmentally related to CLP had not been explored (25-27). That is, CLP directed to produce dendritic cells might transit a Flt3+ CD115+ stage. However, none of these proDC/CDCP were labeled in RAG-1Red mice, and CLP had no potential for generating them in culture (data not shown). Furthermore, ProDC/CDCP could not make B lineage cells even when exposed to LPS (Figure 5A and B). ProDC/CDCP taken from RAG-1/Red marrow generated DC in our cultures, but none of them were marked by red fluorescence regardless of whether LPS was present (Figure 6C, 6D and Figure 7A). As yet another dissimilarity between CLP and ProDC/CDCP, the latter were estrogen resistant (Figure 7B).
DC subsets that have been previously described and studied in relationship to lymphopoiesis were then examined to determine if any were derived from RAG-1 expressing progenitors (Table 1). Approximately 3-5% of total cDC in the spleens of RAG-1/Red mice were labeled (Table 1). We previously reported that a substantial subset of pDC constitutively express RAG-1, even in peripheral tissues such as the spleen and lymph nodes(28). The same cells were readily identified in RAG-1/Red mice. A tolerogenic B220HiCD11cLoMHC-IILoCCR9+ subset of pDC has been recently described (29), and approximately one third of them had a history of RAG-1 expression lymphoid affiliation in our model. In the steady state, RAG-1 expressing progenitors contributed to smaller populations of conventional DC subdivided with respect to CD8 and CD205 (Table 1 and Figure 8A). Mice were then given single injections of LPS and examined one week later. Numbers of red labeled cDC doubled in the marrow and thymus of treated mice, while there were negligible changes in the spleen at that interval (Table1 and Figure 8A). This TLR ligand does not effectively induce formation of pDC (2). Of special interest, percentages of CD8+ cDC that were tdRFP labeled did not change. CLP subsets were also isolated from marrows of the same treated mice and placed in defined cultures for an additional week. Non-labeled CLP from the treated mice made many cDC and had greatly reduced potential to generate CD19+ lymphocytes (Figure 8B and C). In contrast, tdRFP+ CLP taken from the treated animals were still good B lineage progenitors.
These findings indicate that lymphoid progenitors become more lineage committed upon RAG-1 locus activation. This is not a complete transition, as small numbers of DC were induced from tdRFP+ CLP under the influence of LPS. Also, numbers of dendritic cells with the red marker increased when tdRFP− CLP were exposed to LPS. Lymphoid related progenitors made an increased contribution to some DC subsets when animals were treated with LPS. The culture results suggest it is unlikely to involve transition of CLP to proDC/CDCP. Thus, this TLR ligand can selectively influence the more primitive, non-labeled CLP, causing a dramatic change in their fates.
We developed a model that allows RAG-1 expressing progenitors to be tracked through all of their progeny and used it here to monitor progression in the B lymphocyte lineage. Like other milestones of differentiation, the RAG-1 locus is not synchronously and uniformly activated in primitive bone marrow cells. For example, some progenitors with increased probability of becoming lymphocytes express RAG-1 while still part of the LSK fraction. Other progenitors first activate this locus after cKit levels have diminished and cell surface IL-7 receptors are detectable. Some, but not all, expressed B220 prior to becoming labeled. Consistent with other reports, cells in the CLP fraction are heterogenous and continuously segregating with respect to differentiation potential. Their probability of generating non-lymphoid cells markedly diminishes, but is not entirely extinguished when RAG-1 is expressed. That is, RAG-1 negative CLP are more easily directed to become dendritic cells under the influence of a pathogen product. Thus, a picture is emerging where multiple steps in B lymphopoiesis are essential, but need not take place in tight synchrony with each other. In fact, a particular event may frequently occur out of the typical expected sequence. This perspective should influence nomenclature associated with “stages” in differentiation diagrams and considerations of the underlying patterns of transcription factor expression.
A substantial amount of information is available about subsets of multipotent progenitors that replenish the immune system. Our attention was first drawn to lymphopoietic cells in the Flt3+ CD27+ LSK category because of their unique sensitivity to steroid hormones (18). Since some of these cells expressed a human P transgene and/or TdT, we designated them early lymphoid progenitors (ELP). RAG-1/GFP knock-in mice made it possible to isolate some ELP in viable form and learn that that they are potent sources of B, T and NK cells (9). Jacobsen and colleagues refer to a slightly larger population as lymphoid primed multipotent progenitors (LMPP) and have shown that Flt3 upregulation corresponds to loss of megakaryocytic and erythroid potential (30). Formation of LMPP/ELP is dependent on Ikaros and the subset can be identified with Ikaros based reporter mice (17). Additional resolution and tracking of short term repopulating HSC to ELP was made possible by staining for the VCAM-1 adhesion molecule (16). RAG-1+ ELP have a much lower cloning efficiency in Methocel cultures with myeloid supporting growth factors than the otherwise similar RAG-1− cohort (9, 31, 32). This suggests that they have initiated the process of lymphopoiesis, but not completely lost other differentiation options. ELP defined by red fluorescence in our new model are included in previously identified in RAG-1/GFP+ ELP and were not considered further.
Down-regulation of the c-Kit receptor for stem cell factor and display of the IL-7 receptor are important indicators for progression in lymphoid lineages. For example, cKitLoIL-7Rα+ CLP generate lymphocytes more rapidly than c-KitHiIL-7Rα− MPP or ELP (9, 21, 33, 34). However, it is striking that RAG-1 expression is not homogenous in either category. Stochastic aspects of lymphopoiesis have been appreciated before. For example, ELP may express multiple combinations of lymphoid related genes and proteins. One recent study concluded that lymphoid related genes can even be expressed at low-levels in HSC and a newly defined Lin− c-Kit+ CD150− Sca-1− CD41− FcgRII/III− pre-GM subset (17, 35). Some randomness is also associated with later events in lymphopoiesis. While CD19 is usually acquired after CD45R/B220, the reverse occurs with specialized B1 restricted progenitors (36, 37).
The originally defined Lin−IL-7Rα+cKitLoSca-1+ CLP subset of bone marrow probably includes all progenitors destined to produce B lymphocytes as well as some of the cells in other immune lineages (8, 14, 19). It remains contentious whether CLP are a principal source of T cells and if they retain potential for generating granulocytes or macrophages (21, 38, 39). Those issues were not addressed in this study. Rather, we focused instead on relationships between CLP and NK cells, pDC or cDC (5, 6, 40, 41). We found that the approximately half of CLP that were marked with red fluorescence had slightly less cKit along with more IL-7Rα. These differences suggested that the labeled cells were more differentiated, and indeed they required less time in culture to generate CD19+ lymphocytes. Importantly, red labeled CLP were much less likely to generate NK and DC than the unlabeled cohort. While expression of RAG-1 is clearly an important milestone, the unmarked CLP could generate labeled cells in all of these lineages. Therefore, the threshold of RAG-1 expression required for marking can be reached before or during progression in those pathways. This heterogeneity of CLP is consistent with a prior study, where individual cells were found to have B+NK, NK or B potential (14). Furthermore, CLPs have been re-defined and sub-divided by several other groups. The Flt3 cytokine receptor is required for efficient dendritic cell formation, and is useful for sub-dividing CLP into populations with more or less T lineage potential (23, 42). Since Flt3 is highest on ELP and appears to decline with B lineage progression, we expected to find a correlation with marking in RAG-1/Red mice. However, most unlabeled CLP were Flt3+, and red labeled CLP were heterogeneous with respect to that receptor. In contrast, most RAG-1 marked cells had the CD93/AA4.1 antigen used by other laboratories to define lymphoid progenitors (22, 24). Mansson and colleagues found that 55% of single CLP had RAG-1 transcripts detectable in a multiplex PCR assay and used a λ5 based reporter to isolate a small population of B lineage restricted progenitors (43). These studies suggest that no CLP subsets are homogeneous in all respects. Moreover, it may be that all B lymphocytes are not produced via the same sequence of events.
RAG-1, RAG-2 and other components of the receptor gene recombination system can be expressed in cells where their function is not obvious. It remains possible that this represents low level, unimportant gene activation (44). On the other hand, bone marrow progenitors isolated on the basis of present or past RAG-1 synthesis exhibit strong lymphoid bias (9). Thus, RAG-1 is an index of lymphoid affiliation that may be useful for distinguishing alternate differentiation pathways. Plasmacytoid dendritic cells (pDC) represent one example where this appears to be the case. For example, we previously resolved two pDC subsets according to green fluorescence in RAG-1/GFP knock-in mice (28). Both RAG-1+ and RAG-1− pDC were present in marrow, spleen and lymph nodes. The RAG-1+ pDC were closely related to the B lymphocyte lineage, as indicated by their expression of EBF and Pax-5 transcription factors. Similarly, we have now found that pDC subsets can be generated in culture from CLP under the influence of a pathogen product (see below). Several observations suggest that the genealogy of pDC may correspond with special functions. For example, cytokine production patterns differed between RAG-1+ and RAG-1− subsets (28). Also, pDC produced by intrasplenic transplantation from myeloid progenitors had greater ability to activate T cells and produce cytokines than those generated from CLP (40). A tolerogenic category of pDC was recently discovered based on expression of the CCR9 chemokine receptor and low MHC-Class II (29). However, cells with those characteristics were heterogeneous with respect to RAG-1 expression history in RAG-1/Red mice. It will be interesting to learn if functionally specialized pDC subsets described in humans have distinct origins (45). In addition to pDC, some NK cells were previously shown to come from RAG-1 expressing progenitors and to undergo D-J rearrangements (10, 40, 46). These were also marked in our RAG-1/Red model and were generated from CLP in culture. The absence of NK cell labeling in RAG-1/GFP knock-in mice indicates RAG-1 is only transiently made by their progenitors, and again demonstrates the utility of RAG-1/Red mice for fate mapping.
We and others have described lymphopoiesis as a continuous, progressive process, where functionally related effector cells may arise via alternate differentiation pathways (47, 48). In addition, early steps in differentiation may be reversible (49). Given that flexibility, fate mapping systems such as the one used here are needed to appreciate the origins of cells and explore the importance of environmental cues for their formation. Proximity of progenitors to growth and differentiation factors may determine genealogy in bone marrow of healthy animals, while Inflammatory cytokines and pathogen products become important during infections (1, 2, 50). Precursor frequencies for B lymphopoiesis dramatically decline when CLP are exposed to TLR ligands in culture (1, 2). At the same time, numbers of cells capable of becoming DC increased. We now report that LPS selects among CLP subsets, preferentially directing the least differentiated ones to become dendritic cells. This was not absolute however, and numbers of DC increased when red labeled CLP were exposed to LPS. Additionally, some unlabeled CLP gave rise to labeled DC in LPS containing cultures. In any case, RAG-1+ lymphoid progenitors have no developmental relationship to DC restricted proDC/CDCP and do not pass through that compartment when directed to become DC. We conclude that commitment to immune cell lineages is a gradual and asynchronous process that can change in response to disease.
We thank Dr. Linda F. Thompson for advice on the manuscript, Jacob Bass and Dr. Diana Hamilton for cell sorting, and Shelli Wasson for editorial assistance. P.W.K. holds the William H. and Rita Bell Endowed Chair in Biomedical Research.
1This work was supported by grants AI020069, AI058162, AI069024 (P.W.K.) and F30AG031646 (B.L.E.) from the National Institutes of Health.
The authors have no financial conflict of interest.