Restricted STAT5 Activation Dictates Appropriate Thymic B versus T Cell Lineage Commitment1
http://www.100md.com
免疫学杂志 2005年第12期
Abstract
The molecular mechanisms regulating lymphocyte lineage commitment remain poorly characterized. To explore the role of the IL7R in this process, we generated transgenic mice that express a constitutively active form of STAT5 (STAT5b-CA), a key downstream IL7R effector, throughout lymphocyte development. STAT5b-CA mice exhibit a 40-fold increase in pro-B cells in the thymus. As documented by BrdU labeling studies, this increase is not due to enhanced B cell proliferation. Thymic pro-B cells in STAT5b-CA mice show a modest increase in cell survival (4-fold), which correlates with bcl-xL expression. However, bcl-xL transgenic mice do not show increases in thymic B cell numbers. Thus, STAT5-dependent bcl-xL up-regulation and enhanced B cell survival are not sufficient to drive the thymic B cell development observed in STAT5b-CA mice. Importantly, thymic pro-B cells in STAT5b-CA mice are derived from early T cell progenitors (ETPs), suggesting that STAT5 acts by altering ETP lineage commitment. Supporting this hypothesis, STAT5 binds to the pax5 promoter in ETPs from STAT5b-CA mice and induces pax5, a master regulator of B cell development. Conversely, STAT5b-CA mice exhibit a decrease in the DN1b subset of ETPs, demonstrating that STAT5 activation inhibits early T cell differentiation or lineage commitment. On the basis of these findings, we propose that the observed expression of the IL-7R on common lymphoid progenitors, but not ETPs, results in differential STAT5 signaling within these distinct progenitor populations and thus helps ensure appropriate development of B cells and T cells in the bone marrow and thymic environments, respectively.
Introduction
Lymphocyte development involves the differentiation of early lymphoid progenitors (ELPs)3 into lineage-restricted B cells and T cells (1). This process occurs in a stepwise manner, with ELPs first giving rise to two distinct progenitor subtypes: 1) the early T cell progenitor (ETP) in the thymus and 2) the common lymphoid progenitor (CLP) in the bone marrow (2, 3). The mechanisms by which ETPs and CLPs commit to the T or B cell lineage are largely unknown. However, recent results have identified the transcription factor Pax5 and the receptor Notch1 as master regulators of B and T cell lineage commitment, respectively (4, 5, 6, 7). Specifically, low levels of Notch1 expression can be detected on CD44+CD25– thymocytes (triple negative 1, TN1), a subpopulation that contains the earliest T cell progenitors. Notch1 expression levels increase as these progenitors differentiate into pro-T cells (TN2, CD44+CD25+), an event that correlates with commitment to the T cell lineage (8). Furthermore, Notch1 is required for T cell differentiation since T cell development is abrogated in Notch1-deficient mice; in fact, Notch1 deletion leads to intrathymic B cell development (9, 10, 11). Conversely, a constitutively active form of Notch1 results in T cell development in the bone marrow and essentially ablates B cell differentiation (12, 13). Thus, Notch1 signaling is both necessary and sufficient to commit early lymphoid progenitors to the T cell lineage.
In contrast to T cell development, pro-B cell development is regulated by a set of transcription factors called early B cell factor (EBF) and Pax5; mice lacking these factors show specific defects in pro-B cell differentiation (14). Commitment to the B cell lineage is completed at the CD19+ pro-B cell stage (15). Pax5 appears to play a particularly important role in this process, because pro-B cells from pax5–/– mice retain a great deal of lineage plasticity. For example, pax5–/– pro-B cells are able to differentiate into a variety of other cell types including T cells, myeloid cells, and erythrocytes (4). In contrast, forced expression of Pax5 in T cell progenitors alters T cell development and results in the generation of CD19-expressing thymocytes (16). These studies demonstrate that up-regulation of pax5 gene expression in early lymphocyte progenitor cells represses differentiation toward other cell fates and thereby promotes development toward the B cell fate (4, 17).
Although Notch1 and Pax5 play essential roles in governing lymphocyte development and lineage commitment, the molecular mechanisms that regulate expression of these factors are not well characterized. One intriguing possibility is that cytokine receptors may play an important role in this process. A key cytokine receptor required for both early B and T cell development is the IL7R (18). IL7R–/– mice show a profound block in thymocyte development at the pro-T cell stage (TN2 TN3), resulting in a 10- to 100-fold reduction in total thymocyte numbers (18). B cell differentiation is blocked at an even earlier stage, corresponding to the CLP to pro-B cell transition (Hardy fraction A) (18, 19). Interestingly, this correlates with expression of the IL7R, which can be observed on CLPs but not on ETPs (2), thereby illustrating a potentially important difference in B vs T cell lineage commitment.
To determine whether the IL7R plays a differential role in B vs T cell lineage commitment, we examined the molecular mechanisms by which this receptor regulates early lymphocyte development. We currently know that the IL7R activates a number of downstream signaling pathways, including those for PI3K, STAT3, STAT5, and possibly Ras (20, 21, 22). To identify the role that STAT5 plays in IL7R-dependent lymphocyte development, we generated transgenic mice expressing a constitutively active form of STAT5 (STAT5b-CA). Using these mice, we have recently shown that STAT5 activation is sufficient to drive IL7R-dependent B cell development (23). Specifically, crossing the STAT5b-CA transgene onto an IL7R-deficient background largely restores pro-B, immature B, and mature B cell populations. Thus, IL7R-dependent STAT5 activation plays a key role in directing B cell development.
More recently, we have examined the role of STAT5 activation in lymphocyte lineage commitment. We find that STAT5 activation leads to the surprising development of B cells in the thymus. Specifically, STAT5 activation results in the induction of pax5 and bcl-xL in ETPs and thereby redirects these cells down the B cell lineage. Based on our findings and the absence of IL7R expression on ETPs, we propose a model in which T vs B cell lineage choice is dictated by a competition between Notch1 and STAT5 activation.
Materials and Methods
]
Thymi from LMC and STAT5b-CA mice (4–12 wk old) were isolated and processed to examine DN1a–e thymocyte populations as described by Porritt et al. (30). Lineage-positive cells were removed via magnetic depletion on an AutoMACS using Abs against CD3, CD4, CD8, CD19, B220, TCR-, DX5, CD11b, Gr-1, and TER119. After purification, cells were stained with Abs against CD44 and CD25 to identify DN1 thymocytes, followed by CD24 (HSA) and c-kit to identify DN1a–e populations. Samples were analyzed by flow cytometry on an LSR II flow cytometer. Absolute cell numbers and p values were calculated for each cell population. These results are representative of 5 experiments (16 LMC and 13 STAT5b-CA mice).
Results
STAT5b-CA transgene expression
To examine the role that the STAT5 signaling pathway plays in lymphocyte development, we utilized transgenic mice expressing a constitutively active form of STAT5 (STAT5b-CA) (23)(Fig. 1, A and B). Expression of the STAT5b-CA transgene was verified using real-time RT-PCR. For these experiments, we purified ETP populations (TN1, TN2, and ETPs) via cell sorting from littermate control (LMC) and STAT5b-CA mice. As shown in Fig. 1, C and D, expression of the STAT5b-CA transgene was detected in TN1, TN2, and ETP cells from STAT5b-CA mice; no expression was detected in these populations from LMC mice. Relative expression levels were calculated following normalization to HPRT mRNA levels (Fig. 1, C and D). Although the presence of activated STAT5 can be monitored using Abs that recognize STAT5b phosphorylated on Y699, we were unable to obtain sufficient numbers of pro-T cells to carry out such immunoblotting experiments. However, we have previously documented that the STAT5b-CA transgene continues to be expressed in more mature T and B cell populations (24). In these mature lymphocyte populations, STAT5b-CA mRNA levels precisely paralleled phospho-STAT5 expression (and hence activated STAT5). Thus, STAT5b-CA expression strongly correlates with activation of the STAT5 signaling pathway.
STAT5 activation results in intrathymic B cell development
Initial characterization of STAT5b-CA mice revealed that there were increased numbers of TN T cells in the thymi of these mice; this increase in TNs correlated with age (Fig. 2A). However, overall thymic cellularity was not significantly different in these mice (LMC 124 x 106±16.6 x 106, STAT5b-CA 117 x 106±47.8 x 106). To more closely characterize this expanded TN population, we gated on TN cells and stained for B220 expression. Interestingly, STAT5b-CA TN cells exhibit a dramatic increase in the percentage of B220-expressing cells relative to LMC mice (27.7% vs 1.7%, respectively, Fig. 2B); this represents a 45-fold increase in thymic B220+ cells (Fig. 2E). However, since B220 is not uniquely expressed by B cells, we further characterized this population by staining with Abs against BP-1 and CD19, markers that are associated with developing pro-B cells. A significant subpopulation of TN cells from STAT5b-CA mice expressed both BP-1 and CD19 (Fig. 2C). In contrast, virtually no CD19+BP-1+ pro-B cells are found in the thymi of LMC mice. This difference corresponds to an 40-fold increase in thymic pro-B cells in STAT5b-CA mice (Fig. 2E). In addition, we found that both IgM+ and IgM– cells are present in the TN population from STAT5b-CA mice (Fig. 2D). In contrast, and consistent with previous reports, the few B220+ cells present in LMC mice are predominantly IgM+ B cells (Fig. 2D). These results demonstrate that B cell development is actively ongoing in the thymi of STAT5b-CA mice and suggest that STAT5 signaling alters B vs T cell lineage commitment in the thymus.
STAT5b-CA thymic B cells are derived from ETPs
To determine whether thymic B cells in STAT5b-CA mice are derived from ETPs, as opposed to CLPs from the bone marrow, we examined TCR DJ rearrangement patterns. These studies are important to rule out the possibility that thymic B cells in STAT5b-CA mice are 1) derived from CLPs that have aberrantly migrated to the thymus (unlikely but formally possible) or 2) derived from B cells that developed normally from CLPs in the bone marrow and then after maturation in the spleen migrated to the thymus. The TCR locus contains two distinct DJ regions: D1J1 and D2J2 (Fig. 3A). Previous studies have shown that ETPs exhibit TCR D1J1, but not D2J2 rearrangements (2). In contrast, bone marrow-derived B cells do not exhibit significant TCR D1J1 or D2J2 gene rearrangements (2). Therefore, we conducted semiquantitative PCR assays that allowed us to examine TCR D1J1 and D2J2 gene rearrangements in sorted B220+CD19+ B cell populations isolated from the thymi and bone marrow of LMC and STAT5b-CA mice. Fig. 3B shows the purity of lineage-depleted B220+CD19+ thymic populations before and after sorting; sorted populations were >99% pure. Genomic DNA was isolated from these cell populations and subjected to PCR amplification to assay for DJ gene rearrangements (27). As expected, there were virtually undetectable levels of TCR D1J1 and D2J2 gene rearrangements in bone marrow-derived B cells from LMC and STAT5b-CA mice (Fig. 3C, lanes 3 and 4). In contrast, robust TCR D1J1, but not D2J2 gene rearrangements, are seen in thymic B cells from STAT5b-CA mice (Fig. 3C, lane 2). As a positive control, whole thymocytes from LMC and STAT5b-CA mice were subjected to TCR D1J1 and D2J2 PCR assays. Importantly, both D1J1 and D2J2 gene rearrangements are observed in these populations as previously reported (2)(Fig. 3C, lanes 5 and 6). For a more accurate comparison, TCR D1J1 and TCR D2J2 rearrangements from STAT5b-CA thymic B cells were also compared with LMC ETPs. Our results show that thymic B cells in STAT5b-CA mice exhibit levels of TCR D1J1 rearrangements that are as high or higher than that found in ETPs, suggesting that virtually all thymic B cells in STAT5b-CA mice are derived from the ETP population (Fig. 3D). In addition, we also examined TCR V8 DJ2 gene rearrangements in these distinct cell populations. V8 DJ2 gene rearrangements were only observed in total thymocyte populations (Fig. 3E). The absence of D2J2 and V8 DJ gene rearrangements in STAT5b-CA thymic B cells documents that the presence of D1J1 gene rearrangements in these cells is not due to contamination from thymic T cells. Thus, these results demonstrate that thymic B cells in STAT5b-CA mice are derived from ETPs.
STAT5b-CA thymic B cells exhibit enhanced cell survival but not proliferation
During T cell development, ETPs have been shown to give rise predominantly to T cells in the thymus (2)(Fig. 4A). Although ETPs can be induced to give rise to B cells under artificial circumstances, this occurs, if at all, with very low efficiency in the thymus (2). To determine the mechanism responsible for directing B cell development in the thymi of the STAT5b-CA mice, we considered three alternative models. Enhanced B cell development in the thymus of STAT5b-CA mice could be due entirely to altered lineage commitment in ETPs favoring B cell differentiation (model I). Alternatively, our results could reflect increased proliferation or enhanced survival of a rare ETP-derived B cell population in the thymi of the STAT5b-CA mice (model II). Finally, our findings may result from a combination of both altered lineage commitment and increased B cell survival/proliferation (model III).
To examine proliferation of thymic B cells, BrdU assays were conducted using purified cell populations from the thymus and bone marrow of LMC and STAT5b-CA mice. As shown in Fig. 4B, TN2 pro-T cells from LMC and STAT5b-CA mice exhibit similar levels of proliferation. Thus, STAT5 activation does not alter pro-T cell expansion. Likewise, thymic B cells from STAT5b-CA mice do not proliferate more than LMC thymic pro-B cells. Thus, enhanced proliferation is not the mechanism responsible for the increase in intrathymic B cell numbers in STAT5b-CA mice. We next determined whether STAT5b-CA thymic B cells have a survival advantage. To examine this issue, thymocytes were depleted of lineage-positive cells (CD3, CD8, TCR-, DX5, CD11b, Gr-1, and TER119) and stained for surface IgM and B220. Annexin V and PI staining was then analyzed in the B220+IgM– pro- and pre-B cell populations. As is shown in Fig. 4C, the percentage of apoptotic cells (Annexin V+PI–) is decreased in thymic pro-B and pre-B cells from STAT5b-CA mice. Specifically, 1.9 ± 0.7% of freshly isolated thymic B cells from LMC mice were apoptotic; this contrasts with STAT5b-CA thymic B cells, of which only 0.5 ± 0.3% were apoptotic. This corresponds to a 3.8-fold increase (p = 0.005) in pro-B/pre-B cell survival in STAT5b-CA mice. In addition, placing isolated B cells in culture allows for apoptosis to continue without the removal of dead or dying cells by phagocytes. Therefore, we placed isolated thymic B cells in culture to re-examine cell survival. As seen in Fig. 4C, cell death increased following 6 h of culture, although once again STAT5b-CA B cells had a survival advantage.
Thymic pro-B cell numbers are increased 40-fold in STAT5b-CA mice relative to LMC. Thus, it appeared unlikely that the relatively modest increase in cell survival observed in STAT5b-CA mice accounted for the increase in pro-B cell numbers. However, to formally test this hypothesis, we made use of bcl-xL transgenic mice (Eμ-bcl-xL). In these mice, bcl-xL expression, which is driven by the Eμ enhancer, is observed throughout B cell development (25) and largely overlaps with the expression of our STAT5b-CA transgene in B cells. If increased STAT5-dependent bcl-xL expression was sufficient to account for the enhanced B cell development observed in STAT5b-CA mice, then bcl-xL transgenic mice should exhibit a similar phenotype. However, as shown in Fig. 4D, bcl-xL mice show no significant increase in either total thymic B220+ cells or in CD19+BP-1+ pro-B cells. Thus, STAT5b-dependent bcl-xL induction is clearly not sufficient to drive the enhanced B cell development observed in STAT5b-CA mice.
T cell progenitors in STAT5b-CA mice show up-regulation of pax5 and bcl-xL and bind to the pax5 gene promoter
To determine whether STAT5 activation in ETPs could perturb cell fate decisions, we examined whether pax5, a transcription factor associated with B cell lineage commitment, is expressed in T cell progenitor populations in STAT5b-CA mice. To assay for pax5 gene expression, TaqMan RT-PCR assays were conducted on sorted TN1 and TN2 cells from LMC and STAT5b-CA mice. As shown in Fig. 5A, pax5 expression is present in STAT5b-CA TN1 cells. Conversely, LMC TN1 cells or TN2 cells from both LMC and STAT5b-CA mice do not express pax5 (Fig. 5A). We verified that essentially equivalent levels of mRNA were analyzed using a TaqMan assay for HPRT (Fig. 5A, inset). Thus, STAT5 activation is sufficient to induce pax5 gene expression in TN1 cells.
Previous studies have demonstrated that TN1 cells are a heterogeneous population that include both ETPs (c-kithighIL7Rlow/–), which give rise to T cells, and non-ETPs (c-kitlow, IL7R+), which do not give rise to T cells (Fig. 5B) (2). To examine whether STAT5 activation changes genes involved in proliferation, survival, or B cell lineage commitment, we conducted real-time RT-PCR assays for cyclin D2, bcl-xL, and pax5 gene expression in ETPs. ETPs were sorted on a FACSAria and subsequently used in TaqMan RT-PCR assays as described in Materials and Methods. Importantly, pax5 is expressed in the STAT5b-CA ETP population (Fig. 5C, top panel), but was never detected in ETP populations from LMC mice. We also examined whether ETPs express different levels of genes involved in cell cycle induction or cell survival such as cyclin D2 or bcl-xL. Cyclin D2 levels are slightly higher in ETPs from LMC vs STAT5b-CA mice (Fig. 5C, bottom panel). This contrasts with bone marrow-derived STAT5b-CA pro-B cells which showed a 7-fold increase in cyclin D2 levels relative to LMC (23). However, similar to bone marrow-derived STAT5b-CA pro-B cells (23), there is a 7-fold increase in bcl-xL mRNA levels in STAT5b-CA ETPs relative to LMC ETPs (Fig. 5C, middle panel). These results are also in agreement with the increase in survival that we observed in STAT5b-CA thymic B cells (Fig. 4C). Thus, STAT5 activation in ETPs results in altered gene expression which promotes cell survival (bcl-xL) and differentiation (pax5) down the B cell lineage.
The pax5 RT-PCR results for the TN1 and ETP populations suggest that pro-T cell lineage commitment is altered via STAT5-dependent induction of pax5. If STAT5 is able to directly regulate pax5 transcription, we predict that STAT5 should exhibit binding to the pax5 gene promoter. Importantly, recent results by Hirokawa et al. (29) have identified a STAT5-binding motif in the pax5 gene promoter, and demonstrated that STAT5 binds to this promoter in a BAF pro-B cell line. On the basis of this work, we conducted ChIP assays on c-kithigh, TN1 and TN2 cells sorted from LMC, and STAT5b-CA mice. The presence of the STAT5-binding motif was detected by a PCR, followed by Southern blotting with an oligonucleotide that spans this motif. As shown in Fig. 5D, significant STAT5 binding to the pax5 promoter could be detected in STAT5b-CA, but not in LMC, TN1 cells. Importantly, this increased STAT5 binding does not simply reflect the increase in STAT5 expression levels in STAT5b-CA mice since it was not observed in TN2 cells from either LMC or STAT5b-CA mice (Fig. 5D). These results strongly suggest that STAT5 initiates B cell lineage commitment by directly binding to the pax5 gene promoter and thereby inducing pax5 transcription.
STAT5 activation inhibits early T cell development
Our data strongly suggest that STAT5 activation alters thymocyte lineage commitment by redirecting ETPs toward a B cell fate. One prediction of this hypothesis is that early T cell differentiation should be inhibited in STAT5b-CA mice as well. To examine this issue, we made use of recent findings by Porritt et al. (30) who have demonstrated that DN1 thymocytes can be broken down into five distinct subsets (DN1a–e) by the cell surface markers c-kit and heat stable antigen (HSA). Of these five subsets, DN1a–DN1c would be classified as ETPs (c-kithigh, IL7Rlow/–). Importantly, Porritt et al. (30) have demonstrated that DN1a cells give rise to the DN1b subset, and that these subsets appear to be the only progenitor populations capable of giving rise to larger numbers of mature T cells in the thymus. In contrast, the DN1c subset has both B and T cell developmental potential, although T cell development proceeds inefficiently. Based on these findings, we examined the relative distribution of DN1a–e populations in LMC and STAT5b-CA thymocytes, both by percentage and absolute cell number. For these studies, mature cell lineages were removed by magnetic depletion. The DN1 subsets are identified as lineage negative (CD3, CD4, CD8, B220, CD19, DX5, TCR-, CD11b, Gr-1, and TER119) CD44+CD25– (Fig. 6A). DN1 cells were then further subdivided using c-kit and HSA (CD24) into the previously described DN1a–DN1e subsets. The relative percentages of these cell populations for LMC and STAT5b-CA mice are shown in Fig. 6B; absolute cell numbers were calculated and are shown in Fig. 6C. STAT5b-CA mice exhibit a 3-fold decrease in both DN1a and DN1b ETPs (p = 0.016 and 0.001, respectively) and a corresponding increase in the DN1c subset (5.7x, p = 0.016). In contrast, STAT5b-CA mice show no significant changes in the DN1d and DN1e populations. These results document that the development of very early T cell progenitors (DN1a and DN1b) is inhibited upon STAT5 activation.
Discussion
The IL7R plays a key role in both B and T cell development. However, the molecular mechanisms underlying IL7R-dependent lymphocyte differentiation and lineage commitment remain poorly characterized. Interestingly, the IL7R is differentially expressed on ETPs (IL7R–) vs CLPs (IL7R+). Underscoring the potential significance of this differential expression pattern, a recent study has shown that distinct Ets family transcription factors, namely, GA binding protein and PU.1, regulate expression of the IL7R gene in T cells vs B cells, respectively (31). To explore the role of the IL7R on lymphocyte lineage commitment, we focused on the transcription factor STAT5, a downstream target of the IL7R, that has recently been shown to drive both B cell and T cell development (23, 32). Our current studies demonstrate that ectopic STAT5 activation in ETPs induces pax5 gene transcription and results in B cell development in the thymus. Thus, IL7R-dependent STAT5 signaling not only regulates B cell differentiation, but also plays an important role in reinforcing appropriate cell fate decisions during lymphocyte development.
Early lymphocyte development involves a binary cell fate decision made by ELPs as they differentiate into either ETPs, which give rise to T cells, or CLPs, which give rise to B cells (or NK cells). Based on the differential expression of the IL7R on ETPs vs CLPs, we hypothesized that IL7R-dependent STAT5 activation functions to regulate lymphocyte lineage commitment. Supporting this hypothesis, we observed a 40-fold increase in developing B cells in the thymi of STAT5b-CA vs LMC mice. Specifically, we detected pro-B cell, immature B cell, and mature B cell subsets in the thymi of STAT5b-CA mice (Fig. 2, C and D, and data not shown). In contrast, although LMC thymi have a very small population of B220+ cells, the majority of these cells are mature B cells characterized by IgM expression (Fig. 2D). This is consistent with previous reports documenting that thymic B cells predominantly exhibit a mature phenotype (33). Thus, these results demonstrate that STAT5b-CA mice exhibit B cell development in the thymus, an organ in which this typically does not occur.
We believe that the above findings reflect a role for STAT5 in altering ETP cell fate decisions and not simply enhanced proliferation or survival of rare B cell progenitors. Several pieces of evidence support this hypothesis. First, we have documented that the STAT5b-CA transgene is expressed in ETPs (Fig. 1) and can induce transcription of STAT5 target genes such as bcl-xL (Fig. 5C, middle panel) (34, 35). Second, we directly demonstrate that B cells in the thymus of STAT5b-CA mice are derived from ETPs by making use of TCRDJ gene rearrangement patterns. As has been reported for wild-type ETPs, STAT5b-CA thymic B cells exhibit D1J1, but not D2J2 gene rearrangements (2). In contrast, bone marrow-derived B cells from STAT5b-CA mice exhibit undetectable levels of TCR DJ gene rearrangements (Fig. 3C). These findings exclude the possibility that the thymic B cells observed in STAT5b-CA mice are derived from rare, infiltrating bone marrow CLPs that undergo expansion in the thymus; rather, they provide evidence that STAT5b-CA thymic B cells are derived from ETPs. Third, we demonstrate that increased intrathymic B cell numbers in STAT5b-CA mice are not due to enhanced proliferation of a rare B cell population (Fig. 4B). In fact, STAT5b-CA ETPs express slightly lower levels of cyclin D2 (Fig. 5C), and STAT5b-CA thymic B cells proliferate no better than those of LMC mice (Fig. 4B). Fourth, thymic STAT5b-CA B cells show a modest increase in cell survival (Fig. 4C), which correlates with increased bcl-xL expression in ETPs. However, this is clearly not sufficient to drive the enhanced B cell development observed in STAT5b-CA mice since forced expression of a bcl-xL transgene does not mimic the phenotype observed in STAT5b-CA mice (Fig. 4D). Fifth, the IL7R is expressed on pro-B cells but not ETPs. Hence, if the STAT5-dependent increase in B cell numbers is due to an effect on expansion/survival of pro-B cells rather than altered lineage commitment, then we would predict that IL7 transgenic mice would mimic the phenotype observed in STAT5b-CA mice. In fact, IL7 transgenic mice exhibit, at most, a modest 2-fold increase in total thymic B cells (36) and thus clearly do not recapitulate the dramatic 45-fold expansion of total B cells observed in STAT5b-CA mice. Taken together, these results eliminate the possibility that the pro-B cell expansion observed in STAT5b-CA mice is simply attributable to increased proliferation or enhanced survival of STAT5b-CA thymic B cells.
On the basis of our findings, we propose that STAT5b-CA acts to deregulate B vs T cell lineage commitment in ETPs. Specifically, we demonstrate that STAT5b-CA ETPs up-regulate pax5, a transcription factor essential for B cell lineage commitment (Fig. 5C, top panel). Previous studies have shown that forcing pax5 expression in ETPs blocks T cell differentiation (via inhibition of Notch) and results in the production of aberrant CD19-expressing thymocytes (16, 37). Thus, our results strongly suggest that STAT5 induces intrathymic B cell development via induction of pax5 in ETPs. Furthermore, we provide evidence that STAT5 plays a direct role in inducing pax5 transcription. Specifically, we observe that STAT5 binds to a portion of the pax5 promoter that has previously been shown to regulate pax5 expression (Fig. 5D) (29, 38). This finding strongly suggests that STAT5 influences lineage commitment by direct induction of pax5. Finally, our studies also demonstrate that STAT5 activation impairs early T cell development. Specifically, we observed a significant reduction in the DN1b subset of the ETP population (Fig. 6). Taken together, our results support a model in which STAT5 promotes thymic B cell development by both altering lineage commitment in ETPs, via pax5 induction, and by enhancing the survival of thymic pro-B cells, via bcl-xL induction (see Fig. 4A, model III).
Although STAT5b-CA mice show remarkable B cell development in the thymus, T cell development is not completely ablated. Hence, STAT5 activation by itself is clearly not sufficient to enforce total lineage diversion. Rather, our results suggest that STAT5 interacts with other factors to promote pax5 gene expression and B cell differentiation. Interestingly, it has been shown that there is a critical binding site for the transcription factor EBF in the pax5 gene promoter (38). Moreover, recent studies by Hirokawa et al. (29) illustrate that STAT5 binds to an overlapping sequence in the pax5 gene promoter. Importantly, this STAT5 binding site is suboptimal (TCCNNNGAA instead of TTCNNNGAA) and most likely requires that STAT5 interact with EBF to bind effectively and induce optimal pax5 gene transcription. Overexpression of STAT5 in EBF-expressing fibroblasts revealed a modest 1.5-fold enhancement of pax5 reporter constructs (29). Although these results suggested that STAT5 and EBF may cooperate to drive pax5 expression in fibroblasts, it was unclear whether the modest synergy observed in those cells would translate into a significant biological effect in B cells in vivo. Importantly, our results suggest that this synergy does indeed occur in vivo and that it has a significant impact on a real biological process, namely, lymphocyte lineage commitment.
Our results, taken together with previous findings, suggest the following model for regulation of B and T cell lineage commitment (Fig. 7). Several groups have demonstrated that E2A, a transcriptional regulator of EBF, and low levels of EBF itself, are expressed in ELPs and ETPs (1, 39). Differentiation of ELPs into CLPs results in IL7R up-regulation and allows for STAT5 activation. We propose that activated STAT5 synergizes with low levels of EBF to regulate pax5 transcription and thereby initiate B cell differentiation. Since Pax5 subsequently acts to induce high levels of EBF (40), this establishes a positive feedback loop that then maintains Pax5 expression and B cell differentiation in the absence of further STAT5 signaling. In contrast, differentiation of ELPs into ETPs does not result in IL7R up-regulation; rather, ETPs exhibit up-regulation of Notch1 expression. Based on the high level of conservation of Notch1 and EBF throughout evolution, we suggest that Notch1 represses EBF expression in ETPs, thereby preventing B cell differentiation. For example, Notch1 can repress signaling from EBF homologues in species such as Xenopus to influence neuronal development or in Drosophila to regulate myogenesis (41). In addition, transcription of E2A, an upstream activator of EBF, is inhibited by Notch signaling (12). More recently, work by Sun and colleagues (42) has demonstrated that Notch1 can target E2A for degradation, providing another mechanism by which Notch1 can influence EBF expression. Finally, upon differentiation of ETPs into TN2 pro-T cells, the IL7R is expressed. This allows IL7R-dependent signaling pathways such as PI3K and STAT5 to induce pro-T cell expansion and survival. However, the absence of EBF in these cells prevents STAT5-induced B cell differentiation. This would also explain why we observe STAT5 binding to the pax5 promoter in TN1 cells (which express EBF) but not in TN2 cells (which express higher levels of the STAT5b-CA transgene but do not express EBF). Furthermore, this provides an explanation for why STAT5 activation does not result in complete lineage conversion. Specifically, the STAT5b-CA transgene begins to be expressed in ETPs at about the same time that Notch1 is up-regulated and comes into contact with its ligand in the thymus. Thus, if Notch1 signaling precedes STAT5b-CA expression, this would result in EBF down-regulation and commitment to the T cell lineage; conversely, if STAT5b-CA expression precedes Notch1 signaling, this would result in pax5 induction and B cell lineage commitment. In this sense, early lymphoid progenitors exist in an unstable equilibrium that can be easily perturbed by environmental cues, resulting in either B or T cell commitment.
Seen in this light, STAT5 does not act as a master regulator of B cell differentiation, but rather as an amplifier that enhances the probability that lymphocyte progenitors will give rise to B lineage cells. Hence, in the bone marrow environment, where Notch ligands are not encountered by early B cell progenitors (43), IL7R-dependent STAT5 activation synergizes with EBF and leads to a very high probability of B cell differentiation. In contrast, in the thymus, where Notch ligands are abundantly expressed and STAT5 signaling is induced only following T cell lineage commitment, the probability of B cell development is drastically reduced. In conclusion, we propose that appropriate IL7R-induced STAT5 signaling reinforces lymphocyte cell fate decisions and ensures that B cell and T cell development are restricted to the bone marrow and thymic environments, respectively.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by grants to M.A.F. from the National Institutes of Health (AI05737), the Concern Foundation, the Minnesota Medical Foundation, a Pew Scholar Award, and a Cancer Research Investigator Award. C.A.G. and M.A.B. were partially supported by a gift from the estate of Eli and Dorothy Rosen and Bernard Collins. C.A.G. and M.A.B. are also supported by a National Institutes of Health Training Grant (2T32-AI07313).
2 Address correspondence and reprint requests to Dr. Michael A. Farrar, Center for Immunology, Cancer Center, Department of Laboratory Medicine and Pathology, University of Minnesota, 312 Church Street SE, BSBE Building, Minneapolis, MN 55455. E-mail address: farrar005{at}tc.umn.edu
3 Abbreviations used in this paper: ELP, early lymphoid progenitor; EBF, early B cell factor; ETP, early T cell progenitor; CLP, common lymphoid progenitor; TN, triple negative; ChIP, chromatin immunoprecipitation; PI, propidium iodide; HPRT, hypoxanthine phosphoribosyltransferase; GA binding protein, GABP; LMC, littermate control; HSA, heat stable antigen.
Received for publication December 21, 2004. Accepted for publication April 14, 2005.
References
Igarashi, H., S. C. Gregory, T. Yokota, N. Sakaguchi, P. W. Kincade. 2002. Transcription from the RAG1 locus marks the earliest lymphocyte progenitors in bone marrow. Immunity 17: 117-130.
Allman, D., A. Sambandam, S. Kim, J. P. Miller, A. Pagan, D. Well, A. Meraz, A. Bhandoola. 2003. Thymopoiesis independent of common lymphoid progenitors. Nat. Immunol. 4: 168-174.
Kondo, M., I. L. Weissman, K. Akashi. 1997. Identification of clonogenic common lymphoid progenitors in mouse bone marrow. Cell 91: 661-672.
Nutt, S. L., B. Heavey, A. G. Rolink, M. Busslinger. 1999. Commitment of the B-lymphoid lineage depends on the transcription factor Pax5. Nature 401: 556-562.
Rolink, A. G., S. L. Nutt, F. Melchers, M. Busslinger. 1999. Long-term in vivo reconstitution of T-cell development by Pax5-deficient B-cell progenitors. Nature 401: 603-606.
Izon, D. J., J. A. Punt, W. S. Pear. 2002. Deciphering the role of Notch signaling in lymphopoiesis. Curr. Opin. Immunol. 14: 192-199.
Radtke, F., A. Wilson, G. Stark, M. Bauer, J. van Meerwijk, H. R. MacDonald, M. Aguet. 1999. Deficient T cell fate specification in mice with an induced inactivation of Notch1. Immunity 10: 547-558.(Christine A. Goetz, Ian R)
The molecular mechanisms regulating lymphocyte lineage commitment remain poorly characterized. To explore the role of the IL7R in this process, we generated transgenic mice that express a constitutively active form of STAT5 (STAT5b-CA), a key downstream IL7R effector, throughout lymphocyte development. STAT5b-CA mice exhibit a 40-fold increase in pro-B cells in the thymus. As documented by BrdU labeling studies, this increase is not due to enhanced B cell proliferation. Thymic pro-B cells in STAT5b-CA mice show a modest increase in cell survival (4-fold), which correlates with bcl-xL expression. However, bcl-xL transgenic mice do not show increases in thymic B cell numbers. Thus, STAT5-dependent bcl-xL up-regulation and enhanced B cell survival are not sufficient to drive the thymic B cell development observed in STAT5b-CA mice. Importantly, thymic pro-B cells in STAT5b-CA mice are derived from early T cell progenitors (ETPs), suggesting that STAT5 acts by altering ETP lineage commitment. Supporting this hypothesis, STAT5 binds to the pax5 promoter in ETPs from STAT5b-CA mice and induces pax5, a master regulator of B cell development. Conversely, STAT5b-CA mice exhibit a decrease in the DN1b subset of ETPs, demonstrating that STAT5 activation inhibits early T cell differentiation or lineage commitment. On the basis of these findings, we propose that the observed expression of the IL-7R on common lymphoid progenitors, but not ETPs, results in differential STAT5 signaling within these distinct progenitor populations and thus helps ensure appropriate development of B cells and T cells in the bone marrow and thymic environments, respectively.
Introduction
Lymphocyte development involves the differentiation of early lymphoid progenitors (ELPs)3 into lineage-restricted B cells and T cells (1). This process occurs in a stepwise manner, with ELPs first giving rise to two distinct progenitor subtypes: 1) the early T cell progenitor (ETP) in the thymus and 2) the common lymphoid progenitor (CLP) in the bone marrow (2, 3). The mechanisms by which ETPs and CLPs commit to the T or B cell lineage are largely unknown. However, recent results have identified the transcription factor Pax5 and the receptor Notch1 as master regulators of B and T cell lineage commitment, respectively (4, 5, 6, 7). Specifically, low levels of Notch1 expression can be detected on CD44+CD25– thymocytes (triple negative 1, TN1), a subpopulation that contains the earliest T cell progenitors. Notch1 expression levels increase as these progenitors differentiate into pro-T cells (TN2, CD44+CD25+), an event that correlates with commitment to the T cell lineage (8). Furthermore, Notch1 is required for T cell differentiation since T cell development is abrogated in Notch1-deficient mice; in fact, Notch1 deletion leads to intrathymic B cell development (9, 10, 11). Conversely, a constitutively active form of Notch1 results in T cell development in the bone marrow and essentially ablates B cell differentiation (12, 13). Thus, Notch1 signaling is both necessary and sufficient to commit early lymphoid progenitors to the T cell lineage.
In contrast to T cell development, pro-B cell development is regulated by a set of transcription factors called early B cell factor (EBF) and Pax5; mice lacking these factors show specific defects in pro-B cell differentiation (14). Commitment to the B cell lineage is completed at the CD19+ pro-B cell stage (15). Pax5 appears to play a particularly important role in this process, because pro-B cells from pax5–/– mice retain a great deal of lineage plasticity. For example, pax5–/– pro-B cells are able to differentiate into a variety of other cell types including T cells, myeloid cells, and erythrocytes (4). In contrast, forced expression of Pax5 in T cell progenitors alters T cell development and results in the generation of CD19-expressing thymocytes (16). These studies demonstrate that up-regulation of pax5 gene expression in early lymphocyte progenitor cells represses differentiation toward other cell fates and thereby promotes development toward the B cell fate (4, 17).
Although Notch1 and Pax5 play essential roles in governing lymphocyte development and lineage commitment, the molecular mechanisms that regulate expression of these factors are not well characterized. One intriguing possibility is that cytokine receptors may play an important role in this process. A key cytokine receptor required for both early B and T cell development is the IL7R (18). IL7R–/– mice show a profound block in thymocyte development at the pro-T cell stage (TN2 TN3), resulting in a 10- to 100-fold reduction in total thymocyte numbers (18). B cell differentiation is blocked at an even earlier stage, corresponding to the CLP to pro-B cell transition (Hardy fraction A) (18, 19). Interestingly, this correlates with expression of the IL7R, which can be observed on CLPs but not on ETPs (2), thereby illustrating a potentially important difference in B vs T cell lineage commitment.
To determine whether the IL7R plays a differential role in B vs T cell lineage commitment, we examined the molecular mechanisms by which this receptor regulates early lymphocyte development. We currently know that the IL7R activates a number of downstream signaling pathways, including those for PI3K, STAT3, STAT5, and possibly Ras (20, 21, 22). To identify the role that STAT5 plays in IL7R-dependent lymphocyte development, we generated transgenic mice expressing a constitutively active form of STAT5 (STAT5b-CA). Using these mice, we have recently shown that STAT5 activation is sufficient to drive IL7R-dependent B cell development (23). Specifically, crossing the STAT5b-CA transgene onto an IL7R-deficient background largely restores pro-B, immature B, and mature B cell populations. Thus, IL7R-dependent STAT5 activation plays a key role in directing B cell development.
More recently, we have examined the role of STAT5 activation in lymphocyte lineage commitment. We find that STAT5 activation leads to the surprising development of B cells in the thymus. Specifically, STAT5 activation results in the induction of pax5 and bcl-xL in ETPs and thereby redirects these cells down the B cell lineage. Based on our findings and the absence of IL7R expression on ETPs, we propose a model in which T vs B cell lineage choice is dictated by a competition between Notch1 and STAT5 activation.
Materials and Methods
]
Thymi from LMC and STAT5b-CA mice (4–12 wk old) were isolated and processed to examine DN1a–e thymocyte populations as described by Porritt et al. (30). Lineage-positive cells were removed via magnetic depletion on an AutoMACS using Abs against CD3, CD4, CD8, CD19, B220, TCR-, DX5, CD11b, Gr-1, and TER119. After purification, cells were stained with Abs against CD44 and CD25 to identify DN1 thymocytes, followed by CD24 (HSA) and c-kit to identify DN1a–e populations. Samples were analyzed by flow cytometry on an LSR II flow cytometer. Absolute cell numbers and p values were calculated for each cell population. These results are representative of 5 experiments (16 LMC and 13 STAT5b-CA mice).
Results
STAT5b-CA transgene expression
To examine the role that the STAT5 signaling pathway plays in lymphocyte development, we utilized transgenic mice expressing a constitutively active form of STAT5 (STAT5b-CA) (23)(Fig. 1, A and B). Expression of the STAT5b-CA transgene was verified using real-time RT-PCR. For these experiments, we purified ETP populations (TN1, TN2, and ETPs) via cell sorting from littermate control (LMC) and STAT5b-CA mice. As shown in Fig. 1, C and D, expression of the STAT5b-CA transgene was detected in TN1, TN2, and ETP cells from STAT5b-CA mice; no expression was detected in these populations from LMC mice. Relative expression levels were calculated following normalization to HPRT mRNA levels (Fig. 1, C and D). Although the presence of activated STAT5 can be monitored using Abs that recognize STAT5b phosphorylated on Y699, we were unable to obtain sufficient numbers of pro-T cells to carry out such immunoblotting experiments. However, we have previously documented that the STAT5b-CA transgene continues to be expressed in more mature T and B cell populations (24). In these mature lymphocyte populations, STAT5b-CA mRNA levels precisely paralleled phospho-STAT5 expression (and hence activated STAT5). Thus, STAT5b-CA expression strongly correlates with activation of the STAT5 signaling pathway.
STAT5 activation results in intrathymic B cell development
Initial characterization of STAT5b-CA mice revealed that there were increased numbers of TN T cells in the thymi of these mice; this increase in TNs correlated with age (Fig. 2A). However, overall thymic cellularity was not significantly different in these mice (LMC 124 x 106±16.6 x 106, STAT5b-CA 117 x 106±47.8 x 106). To more closely characterize this expanded TN population, we gated on TN cells and stained for B220 expression. Interestingly, STAT5b-CA TN cells exhibit a dramatic increase in the percentage of B220-expressing cells relative to LMC mice (27.7% vs 1.7%, respectively, Fig. 2B); this represents a 45-fold increase in thymic B220+ cells (Fig. 2E). However, since B220 is not uniquely expressed by B cells, we further characterized this population by staining with Abs against BP-1 and CD19, markers that are associated with developing pro-B cells. A significant subpopulation of TN cells from STAT5b-CA mice expressed both BP-1 and CD19 (Fig. 2C). In contrast, virtually no CD19+BP-1+ pro-B cells are found in the thymi of LMC mice. This difference corresponds to an 40-fold increase in thymic pro-B cells in STAT5b-CA mice (Fig. 2E). In addition, we found that both IgM+ and IgM– cells are present in the TN population from STAT5b-CA mice (Fig. 2D). In contrast, and consistent with previous reports, the few B220+ cells present in LMC mice are predominantly IgM+ B cells (Fig. 2D). These results demonstrate that B cell development is actively ongoing in the thymi of STAT5b-CA mice and suggest that STAT5 signaling alters B vs T cell lineage commitment in the thymus.
STAT5b-CA thymic B cells are derived from ETPs
To determine whether thymic B cells in STAT5b-CA mice are derived from ETPs, as opposed to CLPs from the bone marrow, we examined TCR DJ rearrangement patterns. These studies are important to rule out the possibility that thymic B cells in STAT5b-CA mice are 1) derived from CLPs that have aberrantly migrated to the thymus (unlikely but formally possible) or 2) derived from B cells that developed normally from CLPs in the bone marrow and then after maturation in the spleen migrated to the thymus. The TCR locus contains two distinct DJ regions: D1J1 and D2J2 (Fig. 3A). Previous studies have shown that ETPs exhibit TCR D1J1, but not D2J2 rearrangements (2). In contrast, bone marrow-derived B cells do not exhibit significant TCR D1J1 or D2J2 gene rearrangements (2). Therefore, we conducted semiquantitative PCR assays that allowed us to examine TCR D1J1 and D2J2 gene rearrangements in sorted B220+CD19+ B cell populations isolated from the thymi and bone marrow of LMC and STAT5b-CA mice. Fig. 3B shows the purity of lineage-depleted B220+CD19+ thymic populations before and after sorting; sorted populations were >99% pure. Genomic DNA was isolated from these cell populations and subjected to PCR amplification to assay for DJ gene rearrangements (27). As expected, there were virtually undetectable levels of TCR D1J1 and D2J2 gene rearrangements in bone marrow-derived B cells from LMC and STAT5b-CA mice (Fig. 3C, lanes 3 and 4). In contrast, robust TCR D1J1, but not D2J2 gene rearrangements, are seen in thymic B cells from STAT5b-CA mice (Fig. 3C, lane 2). As a positive control, whole thymocytes from LMC and STAT5b-CA mice were subjected to TCR D1J1 and D2J2 PCR assays. Importantly, both D1J1 and D2J2 gene rearrangements are observed in these populations as previously reported (2)(Fig. 3C, lanes 5 and 6). For a more accurate comparison, TCR D1J1 and TCR D2J2 rearrangements from STAT5b-CA thymic B cells were also compared with LMC ETPs. Our results show that thymic B cells in STAT5b-CA mice exhibit levels of TCR D1J1 rearrangements that are as high or higher than that found in ETPs, suggesting that virtually all thymic B cells in STAT5b-CA mice are derived from the ETP population (Fig. 3D). In addition, we also examined TCR V8 DJ2 gene rearrangements in these distinct cell populations. V8 DJ2 gene rearrangements were only observed in total thymocyte populations (Fig. 3E). The absence of D2J2 and V8 DJ gene rearrangements in STAT5b-CA thymic B cells documents that the presence of D1J1 gene rearrangements in these cells is not due to contamination from thymic T cells. Thus, these results demonstrate that thymic B cells in STAT5b-CA mice are derived from ETPs.
STAT5b-CA thymic B cells exhibit enhanced cell survival but not proliferation
During T cell development, ETPs have been shown to give rise predominantly to T cells in the thymus (2)(Fig. 4A). Although ETPs can be induced to give rise to B cells under artificial circumstances, this occurs, if at all, with very low efficiency in the thymus (2). To determine the mechanism responsible for directing B cell development in the thymi of the STAT5b-CA mice, we considered three alternative models. Enhanced B cell development in the thymus of STAT5b-CA mice could be due entirely to altered lineage commitment in ETPs favoring B cell differentiation (model I). Alternatively, our results could reflect increased proliferation or enhanced survival of a rare ETP-derived B cell population in the thymi of the STAT5b-CA mice (model II). Finally, our findings may result from a combination of both altered lineage commitment and increased B cell survival/proliferation (model III).
To examine proliferation of thymic B cells, BrdU assays were conducted using purified cell populations from the thymus and bone marrow of LMC and STAT5b-CA mice. As shown in Fig. 4B, TN2 pro-T cells from LMC and STAT5b-CA mice exhibit similar levels of proliferation. Thus, STAT5 activation does not alter pro-T cell expansion. Likewise, thymic B cells from STAT5b-CA mice do not proliferate more than LMC thymic pro-B cells. Thus, enhanced proliferation is not the mechanism responsible for the increase in intrathymic B cell numbers in STAT5b-CA mice. We next determined whether STAT5b-CA thymic B cells have a survival advantage. To examine this issue, thymocytes were depleted of lineage-positive cells (CD3, CD8, TCR-, DX5, CD11b, Gr-1, and TER119) and stained for surface IgM and B220. Annexin V and PI staining was then analyzed in the B220+IgM– pro- and pre-B cell populations. As is shown in Fig. 4C, the percentage of apoptotic cells (Annexin V+PI–) is decreased in thymic pro-B and pre-B cells from STAT5b-CA mice. Specifically, 1.9 ± 0.7% of freshly isolated thymic B cells from LMC mice were apoptotic; this contrasts with STAT5b-CA thymic B cells, of which only 0.5 ± 0.3% were apoptotic. This corresponds to a 3.8-fold increase (p = 0.005) in pro-B/pre-B cell survival in STAT5b-CA mice. In addition, placing isolated B cells in culture allows for apoptosis to continue without the removal of dead or dying cells by phagocytes. Therefore, we placed isolated thymic B cells in culture to re-examine cell survival. As seen in Fig. 4C, cell death increased following 6 h of culture, although once again STAT5b-CA B cells had a survival advantage.
Thymic pro-B cell numbers are increased 40-fold in STAT5b-CA mice relative to LMC. Thus, it appeared unlikely that the relatively modest increase in cell survival observed in STAT5b-CA mice accounted for the increase in pro-B cell numbers. However, to formally test this hypothesis, we made use of bcl-xL transgenic mice (Eμ-bcl-xL). In these mice, bcl-xL expression, which is driven by the Eμ enhancer, is observed throughout B cell development (25) and largely overlaps with the expression of our STAT5b-CA transgene in B cells. If increased STAT5-dependent bcl-xL expression was sufficient to account for the enhanced B cell development observed in STAT5b-CA mice, then bcl-xL transgenic mice should exhibit a similar phenotype. However, as shown in Fig. 4D, bcl-xL mice show no significant increase in either total thymic B220+ cells or in CD19+BP-1+ pro-B cells. Thus, STAT5b-dependent bcl-xL induction is clearly not sufficient to drive the enhanced B cell development observed in STAT5b-CA mice.
T cell progenitors in STAT5b-CA mice show up-regulation of pax5 and bcl-xL and bind to the pax5 gene promoter
To determine whether STAT5 activation in ETPs could perturb cell fate decisions, we examined whether pax5, a transcription factor associated with B cell lineage commitment, is expressed in T cell progenitor populations in STAT5b-CA mice. To assay for pax5 gene expression, TaqMan RT-PCR assays were conducted on sorted TN1 and TN2 cells from LMC and STAT5b-CA mice. As shown in Fig. 5A, pax5 expression is present in STAT5b-CA TN1 cells. Conversely, LMC TN1 cells or TN2 cells from both LMC and STAT5b-CA mice do not express pax5 (Fig. 5A). We verified that essentially equivalent levels of mRNA were analyzed using a TaqMan assay for HPRT (Fig. 5A, inset). Thus, STAT5 activation is sufficient to induce pax5 gene expression in TN1 cells.
Previous studies have demonstrated that TN1 cells are a heterogeneous population that include both ETPs (c-kithighIL7Rlow/–), which give rise to T cells, and non-ETPs (c-kitlow, IL7R+), which do not give rise to T cells (Fig. 5B) (2). To examine whether STAT5 activation changes genes involved in proliferation, survival, or B cell lineage commitment, we conducted real-time RT-PCR assays for cyclin D2, bcl-xL, and pax5 gene expression in ETPs. ETPs were sorted on a FACSAria and subsequently used in TaqMan RT-PCR assays as described in Materials and Methods. Importantly, pax5 is expressed in the STAT5b-CA ETP population (Fig. 5C, top panel), but was never detected in ETP populations from LMC mice. We also examined whether ETPs express different levels of genes involved in cell cycle induction or cell survival such as cyclin D2 or bcl-xL. Cyclin D2 levels are slightly higher in ETPs from LMC vs STAT5b-CA mice (Fig. 5C, bottom panel). This contrasts with bone marrow-derived STAT5b-CA pro-B cells which showed a 7-fold increase in cyclin D2 levels relative to LMC (23). However, similar to bone marrow-derived STAT5b-CA pro-B cells (23), there is a 7-fold increase in bcl-xL mRNA levels in STAT5b-CA ETPs relative to LMC ETPs (Fig. 5C, middle panel). These results are also in agreement with the increase in survival that we observed in STAT5b-CA thymic B cells (Fig. 4C). Thus, STAT5 activation in ETPs results in altered gene expression which promotes cell survival (bcl-xL) and differentiation (pax5) down the B cell lineage.
The pax5 RT-PCR results for the TN1 and ETP populations suggest that pro-T cell lineage commitment is altered via STAT5-dependent induction of pax5. If STAT5 is able to directly regulate pax5 transcription, we predict that STAT5 should exhibit binding to the pax5 gene promoter. Importantly, recent results by Hirokawa et al. (29) have identified a STAT5-binding motif in the pax5 gene promoter, and demonstrated that STAT5 binds to this promoter in a BAF pro-B cell line. On the basis of this work, we conducted ChIP assays on c-kithigh, TN1 and TN2 cells sorted from LMC, and STAT5b-CA mice. The presence of the STAT5-binding motif was detected by a PCR, followed by Southern blotting with an oligonucleotide that spans this motif. As shown in Fig. 5D, significant STAT5 binding to the pax5 promoter could be detected in STAT5b-CA, but not in LMC, TN1 cells. Importantly, this increased STAT5 binding does not simply reflect the increase in STAT5 expression levels in STAT5b-CA mice since it was not observed in TN2 cells from either LMC or STAT5b-CA mice (Fig. 5D). These results strongly suggest that STAT5 initiates B cell lineage commitment by directly binding to the pax5 gene promoter and thereby inducing pax5 transcription.
STAT5 activation inhibits early T cell development
Our data strongly suggest that STAT5 activation alters thymocyte lineage commitment by redirecting ETPs toward a B cell fate. One prediction of this hypothesis is that early T cell differentiation should be inhibited in STAT5b-CA mice as well. To examine this issue, we made use of recent findings by Porritt et al. (30) who have demonstrated that DN1 thymocytes can be broken down into five distinct subsets (DN1a–e) by the cell surface markers c-kit and heat stable antigen (HSA). Of these five subsets, DN1a–DN1c would be classified as ETPs (c-kithigh, IL7Rlow/–). Importantly, Porritt et al. (30) have demonstrated that DN1a cells give rise to the DN1b subset, and that these subsets appear to be the only progenitor populations capable of giving rise to larger numbers of mature T cells in the thymus. In contrast, the DN1c subset has both B and T cell developmental potential, although T cell development proceeds inefficiently. Based on these findings, we examined the relative distribution of DN1a–e populations in LMC and STAT5b-CA thymocytes, both by percentage and absolute cell number. For these studies, mature cell lineages were removed by magnetic depletion. The DN1 subsets are identified as lineage negative (CD3, CD4, CD8, B220, CD19, DX5, TCR-, CD11b, Gr-1, and TER119) CD44+CD25– (Fig. 6A). DN1 cells were then further subdivided using c-kit and HSA (CD24) into the previously described DN1a–DN1e subsets. The relative percentages of these cell populations for LMC and STAT5b-CA mice are shown in Fig. 6B; absolute cell numbers were calculated and are shown in Fig. 6C. STAT5b-CA mice exhibit a 3-fold decrease in both DN1a and DN1b ETPs (p = 0.016 and 0.001, respectively) and a corresponding increase in the DN1c subset (5.7x, p = 0.016). In contrast, STAT5b-CA mice show no significant changes in the DN1d and DN1e populations. These results document that the development of very early T cell progenitors (DN1a and DN1b) is inhibited upon STAT5 activation.
Discussion
The IL7R plays a key role in both B and T cell development. However, the molecular mechanisms underlying IL7R-dependent lymphocyte differentiation and lineage commitment remain poorly characterized. Interestingly, the IL7R is differentially expressed on ETPs (IL7R–) vs CLPs (IL7R+). Underscoring the potential significance of this differential expression pattern, a recent study has shown that distinct Ets family transcription factors, namely, GA binding protein and PU.1, regulate expression of the IL7R gene in T cells vs B cells, respectively (31). To explore the role of the IL7R on lymphocyte lineage commitment, we focused on the transcription factor STAT5, a downstream target of the IL7R, that has recently been shown to drive both B cell and T cell development (23, 32). Our current studies demonstrate that ectopic STAT5 activation in ETPs induces pax5 gene transcription and results in B cell development in the thymus. Thus, IL7R-dependent STAT5 signaling not only regulates B cell differentiation, but also plays an important role in reinforcing appropriate cell fate decisions during lymphocyte development.
Early lymphocyte development involves a binary cell fate decision made by ELPs as they differentiate into either ETPs, which give rise to T cells, or CLPs, which give rise to B cells (or NK cells). Based on the differential expression of the IL7R on ETPs vs CLPs, we hypothesized that IL7R-dependent STAT5 activation functions to regulate lymphocyte lineage commitment. Supporting this hypothesis, we observed a 40-fold increase in developing B cells in the thymi of STAT5b-CA vs LMC mice. Specifically, we detected pro-B cell, immature B cell, and mature B cell subsets in the thymi of STAT5b-CA mice (Fig. 2, C and D, and data not shown). In contrast, although LMC thymi have a very small population of B220+ cells, the majority of these cells are mature B cells characterized by IgM expression (Fig. 2D). This is consistent with previous reports documenting that thymic B cells predominantly exhibit a mature phenotype (33). Thus, these results demonstrate that STAT5b-CA mice exhibit B cell development in the thymus, an organ in which this typically does not occur.
We believe that the above findings reflect a role for STAT5 in altering ETP cell fate decisions and not simply enhanced proliferation or survival of rare B cell progenitors. Several pieces of evidence support this hypothesis. First, we have documented that the STAT5b-CA transgene is expressed in ETPs (Fig. 1) and can induce transcription of STAT5 target genes such as bcl-xL (Fig. 5C, middle panel) (34, 35). Second, we directly demonstrate that B cells in the thymus of STAT5b-CA mice are derived from ETPs by making use of TCRDJ gene rearrangement patterns. As has been reported for wild-type ETPs, STAT5b-CA thymic B cells exhibit D1J1, but not D2J2 gene rearrangements (2). In contrast, bone marrow-derived B cells from STAT5b-CA mice exhibit undetectable levels of TCR DJ gene rearrangements (Fig. 3C). These findings exclude the possibility that the thymic B cells observed in STAT5b-CA mice are derived from rare, infiltrating bone marrow CLPs that undergo expansion in the thymus; rather, they provide evidence that STAT5b-CA thymic B cells are derived from ETPs. Third, we demonstrate that increased intrathymic B cell numbers in STAT5b-CA mice are not due to enhanced proliferation of a rare B cell population (Fig. 4B). In fact, STAT5b-CA ETPs express slightly lower levels of cyclin D2 (Fig. 5C), and STAT5b-CA thymic B cells proliferate no better than those of LMC mice (Fig. 4B). Fourth, thymic STAT5b-CA B cells show a modest increase in cell survival (Fig. 4C), which correlates with increased bcl-xL expression in ETPs. However, this is clearly not sufficient to drive the enhanced B cell development observed in STAT5b-CA mice since forced expression of a bcl-xL transgene does not mimic the phenotype observed in STAT5b-CA mice (Fig. 4D). Fifth, the IL7R is expressed on pro-B cells but not ETPs. Hence, if the STAT5-dependent increase in B cell numbers is due to an effect on expansion/survival of pro-B cells rather than altered lineage commitment, then we would predict that IL7 transgenic mice would mimic the phenotype observed in STAT5b-CA mice. In fact, IL7 transgenic mice exhibit, at most, a modest 2-fold increase in total thymic B cells (36) and thus clearly do not recapitulate the dramatic 45-fold expansion of total B cells observed in STAT5b-CA mice. Taken together, these results eliminate the possibility that the pro-B cell expansion observed in STAT5b-CA mice is simply attributable to increased proliferation or enhanced survival of STAT5b-CA thymic B cells.
On the basis of our findings, we propose that STAT5b-CA acts to deregulate B vs T cell lineage commitment in ETPs. Specifically, we demonstrate that STAT5b-CA ETPs up-regulate pax5, a transcription factor essential for B cell lineage commitment (Fig. 5C, top panel). Previous studies have shown that forcing pax5 expression in ETPs blocks T cell differentiation (via inhibition of Notch) and results in the production of aberrant CD19-expressing thymocytes (16, 37). Thus, our results strongly suggest that STAT5 induces intrathymic B cell development via induction of pax5 in ETPs. Furthermore, we provide evidence that STAT5 plays a direct role in inducing pax5 transcription. Specifically, we observe that STAT5 binds to a portion of the pax5 promoter that has previously been shown to regulate pax5 expression (Fig. 5D) (29, 38). This finding strongly suggests that STAT5 influences lineage commitment by direct induction of pax5. Finally, our studies also demonstrate that STAT5 activation impairs early T cell development. Specifically, we observed a significant reduction in the DN1b subset of the ETP population (Fig. 6). Taken together, our results support a model in which STAT5 promotes thymic B cell development by both altering lineage commitment in ETPs, via pax5 induction, and by enhancing the survival of thymic pro-B cells, via bcl-xL induction (see Fig. 4A, model III).
Although STAT5b-CA mice show remarkable B cell development in the thymus, T cell development is not completely ablated. Hence, STAT5 activation by itself is clearly not sufficient to enforce total lineage diversion. Rather, our results suggest that STAT5 interacts with other factors to promote pax5 gene expression and B cell differentiation. Interestingly, it has been shown that there is a critical binding site for the transcription factor EBF in the pax5 gene promoter (38). Moreover, recent studies by Hirokawa et al. (29) illustrate that STAT5 binds to an overlapping sequence in the pax5 gene promoter. Importantly, this STAT5 binding site is suboptimal (TCCNNNGAA instead of TTCNNNGAA) and most likely requires that STAT5 interact with EBF to bind effectively and induce optimal pax5 gene transcription. Overexpression of STAT5 in EBF-expressing fibroblasts revealed a modest 1.5-fold enhancement of pax5 reporter constructs (29). Although these results suggested that STAT5 and EBF may cooperate to drive pax5 expression in fibroblasts, it was unclear whether the modest synergy observed in those cells would translate into a significant biological effect in B cells in vivo. Importantly, our results suggest that this synergy does indeed occur in vivo and that it has a significant impact on a real biological process, namely, lymphocyte lineage commitment.
Our results, taken together with previous findings, suggest the following model for regulation of B and T cell lineage commitment (Fig. 7). Several groups have demonstrated that E2A, a transcriptional regulator of EBF, and low levels of EBF itself, are expressed in ELPs and ETPs (1, 39). Differentiation of ELPs into CLPs results in IL7R up-regulation and allows for STAT5 activation. We propose that activated STAT5 synergizes with low levels of EBF to regulate pax5 transcription and thereby initiate B cell differentiation. Since Pax5 subsequently acts to induce high levels of EBF (40), this establishes a positive feedback loop that then maintains Pax5 expression and B cell differentiation in the absence of further STAT5 signaling. In contrast, differentiation of ELPs into ETPs does not result in IL7R up-regulation; rather, ETPs exhibit up-regulation of Notch1 expression. Based on the high level of conservation of Notch1 and EBF throughout evolution, we suggest that Notch1 represses EBF expression in ETPs, thereby preventing B cell differentiation. For example, Notch1 can repress signaling from EBF homologues in species such as Xenopus to influence neuronal development or in Drosophila to regulate myogenesis (41). In addition, transcription of E2A, an upstream activator of EBF, is inhibited by Notch signaling (12). More recently, work by Sun and colleagues (42) has demonstrated that Notch1 can target E2A for degradation, providing another mechanism by which Notch1 can influence EBF expression. Finally, upon differentiation of ETPs into TN2 pro-T cells, the IL7R is expressed. This allows IL7R-dependent signaling pathways such as PI3K and STAT5 to induce pro-T cell expansion and survival. However, the absence of EBF in these cells prevents STAT5-induced B cell differentiation. This would also explain why we observe STAT5 binding to the pax5 promoter in TN1 cells (which express EBF) but not in TN2 cells (which express higher levels of the STAT5b-CA transgene but do not express EBF). Furthermore, this provides an explanation for why STAT5 activation does not result in complete lineage conversion. Specifically, the STAT5b-CA transgene begins to be expressed in ETPs at about the same time that Notch1 is up-regulated and comes into contact with its ligand in the thymus. Thus, if Notch1 signaling precedes STAT5b-CA expression, this would result in EBF down-regulation and commitment to the T cell lineage; conversely, if STAT5b-CA expression precedes Notch1 signaling, this would result in pax5 induction and B cell lineage commitment. In this sense, early lymphoid progenitors exist in an unstable equilibrium that can be easily perturbed by environmental cues, resulting in either B or T cell commitment.
Seen in this light, STAT5 does not act as a master regulator of B cell differentiation, but rather as an amplifier that enhances the probability that lymphocyte progenitors will give rise to B lineage cells. Hence, in the bone marrow environment, where Notch ligands are not encountered by early B cell progenitors (43), IL7R-dependent STAT5 activation synergizes with EBF and leads to a very high probability of B cell differentiation. In contrast, in the thymus, where Notch ligands are abundantly expressed and STAT5 signaling is induced only following T cell lineage commitment, the probability of B cell development is drastically reduced. In conclusion, we propose that appropriate IL7R-induced STAT5 signaling reinforces lymphocyte cell fate decisions and ensures that B cell and T cell development are restricted to the bone marrow and thymic environments, respectively.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by grants to M.A.F. from the National Institutes of Health (AI05737), the Concern Foundation, the Minnesota Medical Foundation, a Pew Scholar Award, and a Cancer Research Investigator Award. C.A.G. and M.A.B. were partially supported by a gift from the estate of Eli and Dorothy Rosen and Bernard Collins. C.A.G. and M.A.B. are also supported by a National Institutes of Health Training Grant (2T32-AI07313).
2 Address correspondence and reprint requests to Dr. Michael A. Farrar, Center for Immunology, Cancer Center, Department of Laboratory Medicine and Pathology, University of Minnesota, 312 Church Street SE, BSBE Building, Minneapolis, MN 55455. E-mail address: farrar005{at}tc.umn.edu
3 Abbreviations used in this paper: ELP, early lymphoid progenitor; EBF, early B cell factor; ETP, early T cell progenitor; CLP, common lymphoid progenitor; TN, triple negative; ChIP, chromatin immunoprecipitation; PI, propidium iodide; HPRT, hypoxanthine phosphoribosyltransferase; GA binding protein, GABP; LMC, littermate control; HSA, heat stable antigen.
Received for publication December 21, 2004. Accepted for publication April 14, 2005.
References
Igarashi, H., S. C. Gregory, T. Yokota, N. Sakaguchi, P. W. Kincade. 2002. Transcription from the RAG1 locus marks the earliest lymphocyte progenitors in bone marrow. Immunity 17: 117-130.
Allman, D., A. Sambandam, S. Kim, J. P. Miller, A. Pagan, D. Well, A. Meraz, A. Bhandoola. 2003. Thymopoiesis independent of common lymphoid progenitors. Nat. Immunol. 4: 168-174.
Kondo, M., I. L. Weissman, K. Akashi. 1997. Identification of clonogenic common lymphoid progenitors in mouse bone marrow. Cell 91: 661-672.
Nutt, S. L., B. Heavey, A. G. Rolink, M. Busslinger. 1999. Commitment of the B-lymphoid lineage depends on the transcription factor Pax5. Nature 401: 556-562.
Rolink, A. G., S. L. Nutt, F. Melchers, M. Busslinger. 1999. Long-term in vivo reconstitution of T-cell development by Pax5-deficient B-cell progenitors. Nature 401: 603-606.
Izon, D. J., J. A. Punt, W. S. Pear. 2002. Deciphering the role of Notch signaling in lymphopoiesis. Curr. Opin. Immunol. 14: 192-199.
Radtke, F., A. Wilson, G. Stark, M. Bauer, J. van Meerwijk, H. R. MacDonald, M. Aguet. 1999. Deficient T cell fate specification in mice with an induced inactivation of Notch1. Immunity 10: 547-558.(Christine A. Goetz, Ian R)