IL-7 functionally segregates the pro-B cell stage by regulating transcription of recombination mediators across cell cycle

Abstract

Antigen receptor diversity involves the introduction of DNA double stranded breaks during lymphocyte development. To ensure fidelity, cleavage is confined to the G0/G1 phase of cell cycle. One established mechanism of regulation is through periodic degradation of the RAG2 recombinase protein. However, there are additional levels of protection. Here we show that cyclical changes in the IL-7R signaling pathway functionally segregate pro-B cells according to cell cycle status. In consequence, the level of a downstream effector of IL-7 signaling, phospho-STAT5, is inversely correlated with cell cycle expression of Rag, a key gene involved in recombination. Higher levels of phopho-STAT5 in S/G2 correlate with decreased Rag expression and Rag relocalization to pericentromeric heterochromatin (PCH). These cyclical changes in transcription and locus repositioning are ablated upon transformation with v-Abl, which renders STAT5 constitutively active across the cell cycle. We propose that this activity of the IL-7R/STAT5 pathway plays a critical protective role in development, complementing regulation of RAG2 at the protein level, to ensure that recombination does not occur during replication. Our data, suggesting that pro-B cells are not a single homogeneous population explain inconsistencies in the role of IL-7 signaling in regulating Igh recombination.

Introduction

Developmental progression involves a delicate balance between differentiation, survival and proliferation. The juxtaposition of the differentiation and proliferation programs is particularly important during lymphocyte development as early developmental stages are punctuated by V(D)J recombination events, which enable diversification of antigen receptors in B and T lineage cells (1). Because recombination involves the repeated cutting and joining of widely separated gene segments the process must be tightly regulated to ensure that mistakes do not occur. Indeed deregulation of recombination can have serious consequences resulting in aberrant chromosomal rearrangements that give rise to leukemia and lymphoma.

The recombination activating gene proteins (RAG1 and RAG2) play an essential part in recombination, guiding the process from the cleavage phase through to synapsis and repair. First, RAG proteins recognize and cleave conserved recombination signal sequence (RSS) elements that flank individual antigen receptor gene segments (2). Next, RAG stabilizes the post-cleavage complex to ensure that the four cleaved ends are held in place, which is critical for their proper repair by classical nonhomologous end joining (NHEJ) (3). This repair pathway predominates during the G1 early S phase of cell cycle (4) and is important for maintaining genome stability (5-7). RAG cleavage is also directed to occur during the G1/G0 phase of the cell cycle, which couples cleavage and repair to ensure that replication does not occur over unrepaired breaks (8).

It has been known for some time that RAG2 is regulated across cell cycle by a mechanism that involves phosphorylation dependent protein degradation during the G1 to S transition (9-11). The importance of this restriction was recently shown in mice harboring a targeted threonine to alanine mutation at position 490 on RAG2, which prevents degradation at the appropriate stage of cell cycle. When RAG2^T490A/T490A is expressed on a p53 deficient background it gives rise to lymphoid tumors that contain translocations involving antigen receptor genes (12). Somewhat surprisingly though, death is not accelerated in these animals compared to p53 deficient mice. Furthermore, the lymphoid tumors are phenotypically more mature, with a broader spectrum of translocations, compared to lymphomas isolated from Core RAG2, p53 double deficient mice (13). This implies additional mechanisms of protection at the early stages of lymphocyte development.

Environmental signals govern lineage and stage specific gene programs, including the processes of proliferation and recombination. In this context, the IL-7 signaling pathway serves a pivotal role. Not only is it required for survival and proliferation of early B and T cell progenitors (14), but it is also involved in the negative regulation of Rag expression (15, 16). Further, phospho-STAT5, a downstream signaling component of the IL-7 signaling pathway, has been shown to enhance accessibility of the immunoglobulin heavy chain (Igh) locus for rearrangement (17, 18), a stage at which IL-7 is purported to inhibit Rag expression (15). Thus, it remains unclear how IL-7 could coordinate these disparate activities at the pro-B cell stage during Igh recombination.

Here we show that pro-B cells are in fact, a heterogeneous population that can be subdivided on the basis of IL-7R expression and levels of phospho-STAT5. Expression of IL-7R/phospho-STAT5 is found predominantly in the actively dividing population. As a consequence Rag1 has a different transcriptional profile within B cell subsets. Repositioning of the Rag locus to repressive pericentromeric heterochromatin (PCH) occurs preferentially within cells at the G2 phase and correlates with increased phospho-STAT5 levels. Our data reconcile the role of IL-7 in positively regulating Igh accessibility and negatively regulating Rag expression in pro-B cells. Importantly, we reveal an additional mechanism to enforce segregated recombination and proliferation in a developmental context.

Materials and Methods

Mice

CaStat5 and Erag mice have been previously described (19) (20). Wild-type mice were littermate controls and/or when wild-type were used alone were C57Bl/6 mice (Jackson labs). Mice were housed in specific pathogen-free conditions and were maintained and used in accordance with the Institutional Animal Care and Use Committee guidelines.

Cells and Culture Conditions

Short-term bone marrow cells were established either by harvesting total bone marrow or isolated through positive selection using CD19 microbeads (Miltenyi Biotec Inc.) and placed at a concentration of 1 × 10⁶ to 2 × 10⁶ cells/ml in T-25 flasks in 8 ml in Optimem media supplemented with 5% FCS and 5ng/ml of IL-7. One-half of the media was replaced every 3 to 4 days for a total of 6-10 days, with fresh media being added 1 day prior to analysis. This culture system makes use of endogenous stromal cells present within the bone marrow and produces a pro-B cell population that is over 90% pure as measured by CD19⁺/CD25^- surface expression. For IL-7 dilution experiments, cultures were established as above for 6-7 days, counted, washed and replated for 36-40hrs at 2 × 10⁶ cells/ml in 6 well dishes without stroma with either 10, 5, 2.5, 1ng/ml or no IL-7 and then directly analyzed. When indicated cells were sorted prior to analysis on CD19⁺/CD25^-/IgM^- to ensure equivalent purity of pro-B cell populations were assessed.

The v-Abl lines were cultured without stromal cells or cytokine, in RPMI supplemented with 10%FBS.

The gating strategy of sorted cells is as follows: pro-B cells are CD19⁺/c-kit⁺/CD25^-/IgM^-, pre-B cells are CD19⁺/c-kit^-/CD25⁺/IgM^- and DP cells are Thy1.2⁺/CD19^-/CD4⁺/CD8⁺. Experiments were done directly after sorting. Fetal liver pro-B cells were analyzed directly after sorting and were directly compared to bone marrow pro-B that were similarly sorted and directly analyzed.

RT-PCR

Total RNA was isolated with TRIzol (Invitrogen) and cDNA was made using SuperScript II reverse transcriptase (Invitrogen). Quantitative PCR was performed in triplicate with a SYBR green kit (Stratagene) using gene specific primers described previously (16). All samples are normalized against β₂ microglobulin.

Three-dimensional DNA FISH

Cells were washed three times in PBS and then fixed onto poly-L-lysine-coated slides for 3D DNA-FISH analysis as described in detail (21). The Rag locus (containing both Rag1 and Rag2 genes) was detected using the BAC RP23-313G3. The Igh locus was detected using the two BACs CT7-34H6 and CT7-526A21, mapping the 3’ constant and the 5’ variable regions respectively (22). BAC probes were directly labeled by nick translation with dUTP-A594 or dUTP-A488 (Invitrogen). The γ-satellite probe was prepared from a plasmid containing eight copies of the γ-satellite repeat sequence (23) and was directly labeled with dUTP-Cy5 or dUTPA488 (GE Healthcare).

Immuno-DNA FISH

Combined detection of H3S10ph and Rag or Igh loci was carried out on cells adhered to poly-L lysine coated coverslips as previously described (18). Briefly, cells were fixed with 2% paraformaldehyde / PBS for 10 minutes and permeabilized for 5 minutes with 0.4% Triton / PBS on ice. After 30 minutes blocking in 2.5% BSA, 10% normal goat serum and 0.1% Tween-20 / PBS, H3S10ph staining was carried out using an antibody against phosphorylated serine-10 of H3 (Millipore) diluted at 1:400 in blocking solution for one hour at room temperature. Cells were rinsed 3 times in 0.2% BSA, 0.1% Tween-20 / PBS and incubated for one hour with goat-anti-rabbit IgG Alexa 488 or 594 or 633 (Invitrogen). After 3 rinses in 0.1% Tween-20 / PBS, cells were post fixed in 3% paraformaldehyde / PBS for 10 minutes, permeabilized in 0.7% Triton-X-100 in 0.1M HCl for 15 minutes on ice and incubated in 0.1 mg/ml RNase A for 30 minutes at 37°C. Cells were then denatured with 1.9 M HCl for 30 minutes at room temperature and rinsed with cold PBS. DNA probes were denatured for 5 minutes at 95°C, pre-annealed for 45 minutes at 37°C and applied to coverslips which were sealed onto slides with rubber cement and incubated overnight at 37°C. Cells were then rinsed 3 times 30 minutes with 2xSSC at 37°C, 2X SSC and 1xSSC at room temperature. Cells were mounted in ProLong Gold (Invitrogen) containing DAPI to counterstain total DNA.

Confocal microscopy and analysis

Cells were analyzed by confocal microscopy on a Leica Sp5 AOBS (Acoustica Optical Beam Splitter) system. Optical sections separated by 0.3 μm were collected, and only cells with signals from both alleles were analyzed using ImageJ software. Alleles were defined as associated with pericentromeric heterochromatin if BAC probe signals were overlapping or juxtaposed to the γ-satellite signal. Individual alleles from the same cell are shown in different confocal sections. Samples sizes were 100 cells minimum per experiment (see supplementary tables for exact numbers) and experiments were repeated at least 2-3 times.

Flow Cytometry

Surface staining, intracellular staining and cell cycle staining using Hoechst have all been described previously (16, 24). Antibodies specific for murine CD19 (1D3), IL-7R alpha-chain (A7R34), c-kit (2B8), CD25 (PC61), IgM(II/41), and pY695 STAT5 were purchased from BD Pharmingen. Data were collected with the LSR II, and were analyzed with FCS Express (De Novo Software).

ChIP-seq analysis

We obtained raw ChIP-seq data sets of H3K27me3, STAT5, and total input from recent experiments performed by the Clark laboratory (25). We aligned ChIP-seq reads with Bowtie 0.12.7 software to the mm9 mouse genome data, using the following command line option --best --all -m1 -n2 (26). To identify regions of increased sequence tag density obtained after enrichment by ChIP with specific antibodies relative to the measured background along the genome (input chromatin), we used the Qeseq algorithm (27). We analyzed the H3K27me3 using –s 150 setting, reflecting the experimental fragment size. For STAT5, the setting was –s 250. We visualized the obtained peaks and the enrichment scores of the reads located within the peaks using Integrated Genome Browser (http://bioviz.org/igb/).

Statistical analysis

The two-tailed Fisher's exact test was used to analyze the significance of association with PCH and association of γ-satellite.

Standard deviation was used to create errors bars for transcription analysis. In some cases single representative experiments are shown in order to compare multiple experimental parameters within a single experiment. Replicate experimental data is provided within supplementary tables to show reproducibility.

The statistical tests described above were applied to combined data from repeated experiments. Data for individual experiments is displayed in supplementary tables to show the low level of variation between the repeats.

Results

Differential Rag1 transcription occurs within pro-B cells at different phases of the cell cycle as a result of fluctuations in IL-7 responsiveness

To explain the seemingly contradictory roles of IL-7 in regulating V(D)J recombination within pro-B cells we considered the possibility that pro-B cells are in fact a heterogeneous population in terms of their responsiveness to IL-7 signaling. To test this hypothesis we assessed surface IL-7R levels within pro-B cells at different phases of the cell cycle (Supplementary Figure A). Our analyses indicate that a higher proportion of S/G2/M cells expressed IL-7R as compared to G0/G1 pro-B cells (Figure 1A). These data demonstrate that pro-B cells are a heterogeneous population and differentially express the IL-7R on their surface in a fashion that correlates with cell cycle status.

Figure 1. Differential Rag1 transcription occurs within pro-B cells at different phases of the cell cycle as a result of fluctuations in IL-7 responsiveness.

(A) Wild-type bone marrow pro-B cells were analyzed for IL-7R expression and cell cycle stage using flow cytometry. CD19⁺/IgM^-/c-kit⁺ pro-B cells were divided into G0/G1 and S/G2/M by DAPI staining, and back gated for either IL-7R (blue) or an isotype (IG) control (red). (B) Wild-type bone marrow pro-B cells cultured short-term with 5ng/ml of IL-7 in vitro were analyzed for phospho-STAT5 levels and cell cycle stage using flow cytometry. Cells were divided into G0/G1 and S/G2/M by DAPI staining, and back gated for either phospho-STAT5 (blue) or an isotype (IG) control (red). (C) Phospho-STAT5 levels were analyzed across cell cycle stages within v-Abl transformed pro-B cells as described in (B).

(D), (E) Graphs showing the level of Rag1 and Iμ transcripts assessed by Q-PCR (lower panels) in sorted G0/G1 and S/G2/M bone marrow pro-B cells cultured short-term in the presence of IL-7 (5ng/ml) (D) or v-Abl pro-B cells (E) (upper panels). Transcripts were normalized against β₂ microglobulin and the G0/G1 population set at 1. Each data set is representative of 3-4 experiments.

Signaling components downstream of the IL-7R pathway must also be differentially regulated during cell cycle for there to be functional consequences. Activation of the STAT pathway begins after IL-7R engagement, which allows activation of the STAT protein via phosphorylation, dimerization and subsequent translocation to the nucleus where transcription of target genes can be directly controlled. Thus, we analyzed levels of activated STAT5 (phospho-STAT5) during G0/G1 and S/G2/M in pro-B cells. For these experiments we used ex-vivo derived wild-type pro-B cells, cultured short-term in the presence of 5ng/ml of recombinant IL-7. The use of cultured cells was necessary because of technical limitations of the assay including, (i) the inability to use multi-parameter surface and DAPI staining in conjunction with the methanol fixation step that is required for intracellular staining with the phospho-STAT5 antibody, and (ii) the transient nature of the downstream effects of IL-7 signaling: phospho-STAT5 levels fall off rapidly after removal of cells from their IL-7 containing environment and are therefore lost during cell sorting. In addition, the use of cultured cells allowed us to obtain adequate cell numbers to analyze gene expression in concurrent experiments (see below). Intracellular staining was performed on cultured ex-vivo derived pro-B cells using an antibody to phospho-STAT5 or an isotype control. As shown in Figure 1B, phospho-STAT5 was present in all S/G2/M cells and in only a proportion of G0/G1 cells. This pattern mirrors the pattern of IL-7R expression (Figure 1A) indicating that both these signaling components are dynamically regulated within pro-B cells.

We next asked whether differential levels of IL-7/phospho STAT5 in G0/G1 and S/G2/M impact Rag expression in pro-B cells. Ex-vivo derived wild-type pro-B cells were stained with Hoechst and sorted according to cell cycle status prior to analyzing levels of Rag1 transcription. As shown in Figure 1D in G0/G1 cells Rag1 transcripts were detected at a level that was 8-11 fold higher compared to S/G2/M, whereas other transcripts such as Iμ did not vary substantially in the same cells. These data demonstrate that Rag transcription is specifically decreased in S/G2/M where IL-7R/phospho-STAT5 is present in the majority of cells. This result is consistent with previous analyses showing that IL-7 signaling represses Rag transcription (15, 16). In addition, these data are also consistent with V(D)J recombination occurring in the G1 phase of the cell cycle (8) and highlight a second level of control for restricting Rag expression to the G0/G1 compartment.

The above data links IL-7R/phosho-STAT5 expression with cell cycle status and Rag transcription. However it remains possible that differential Rag transcription is a consequence of a change in cell cycle status rather than a consequence of IL-7 signaling. To differentiate between these possibilities we analyzed Rag1 transcription during cell cycle in v-Abl transformed pro-B cells. Transformation with v-Abl dissociates proliferation from exogenous IL-7 resulting in constitutive activation of downstream signaling components (28). Indeed, our data indicate that in contrast to ex-vivo derived cells STAT5 remains phosphorylated in v-Abl transformed cells during G0/G1 (Figure 1C) and Rag1 transcription is not altered to the same extent across the cell cycle relative to untransformed cells (in v-Abl transformed cells there is a less than 2 fold difference between G0/G1 and S/G2/M compared to an 8-11 fold difference in Rag1 transcription in untransformed cells) (Figure 1E). Thus we conclude that differential Rag transcription is not simply a consequence of cell cycle status.

The Rag locus is repositioned to pericentromeric heterochromatin in G2 cells

Association of genes with pericentromeric heterochromatin (PCH) correlates with gene silencing (29). Furthermore, repositioning of select loci within the nucleus, including Rag, has been shown to occur in cycling but not quiescent cells (30). We considered the possibility that a similar mechanism may function to silence Rag expression within a population of actively dividing cells. To address this we performed 3-D immuno-FISH using an antibody to the phosphorylated form of serine 10 on histone H3 (H3S10ph) in combination with a BAC probe that spans both the Rag1 and Rag2 loci (RP23-313G3) and a γ-satellite probe which hybridizes to PCH (Figure 2A). H3S10 phosphorylation begins in early G2 at the chromocentres, overlapping with PCH, and continually spreads along the chromosome until prophase (31). By this method, we were able to determine whether the Rag locus was preferentially associated with repressive PCH during G2, when cells have committed to the process of cell division. Alleles were considered to be associated with PCH if signals overlapped or were directly adjacent to γ-satellite. Our analysis of wild-type pro-B cells, cultured short-term in recombinant IL-7, indicate that the Rag locus was preferentially associated with PCH in G2 cells (Figure 2 and Supplementary Table 1). This trend is not common to other loci that are regulated by IL-7: as shown in Figure 2B, association of Igh with PCH did not change between cell cycle phases. Moreover, relocation of the Rag locus to PCH during G2 is not simply due to a progression through cell cycle, as repositioning did not occur during G2 in v-Abl transformed cells. These data demonstrate that the Rag locus is dynamically repositioned to PCH during G2 in IL-7 responsive B cells, which correlates with reduced Rag transcription.

Figure 2. The Rag locus is repositioned to pericentromeric heterochromatin in G2 cells.

(A) 3-D immuno-DNA FISH was performed on wild-type pro-B cells (cultured short-term in the presence of IL-7) and v-Abl transformed pro-B cell lines. Cells were separated into G2 or G0/G1/S on the basis of their immuno-staining pattern for H3S10ph (green). DNA FISH probes used were γ-satellite (white) in conjunction with either Rag (red) or Igh (not shown). Both alleles were scored for association with repressive pericentromeric heterochromatin (PCH) as determined by overlapping or juxtaposition of the locus specific and γ-satellite signals. Mitotic cells were excluded. A representative example of a G0/G1/S and G2 cell is shown. The positions of the two Rag alleles are shown in separate confocal sections. Scale = 1 μm. (B) Graph showing the percentage of association of at least one Rag or Igh allele with PCH in the total population and in G2 or G0/G1/S cells from both short-term wild-type pro-B cell cultures or v-Abl B cell lines. Over 250 cells of each cell type was counted. Data shown is from one representative experiment, with experiments repeated 2-3 times. Data from two individual experiments is shown in Supplementary Table 1 to show the low variability between experiments.

Phosho-STAT5 levels inversely correlate with Rag expression

During ontogeny, B cell development begins in the fetal liver and subsequently moves to the bone marrow. Interestingly, there is a differential dependence on IL-7 signaling within these two anatomical locations, such that in the absence of the IL-7R, B cell development is ablated in the adult bone marrow, but only partially blocked in fetal liver (32, 33). Thus, we asked whether differential dependence on the IL-7 signaling pathway within these two physiological settings manifested itself in differential levels of phospho-STAT5 in bone marrow and fetal liver derived pro-B cells. As shown in Figure 3A, phospho-STAT5 is present at lower levels in ex-vivo sorted fetal liver pro-B cells compared to their bone marrow counterparts. Again, we compared the phospho-STAT5 signals with an isotype control antibody because (as shown here) the background signal can change as a result of cell size. FACs analyses indicate that fetal liver pro-B cells also have fewer cells in S/G2/M compared to their bone-marrow counterparts (Supplementary Figure Bi). Importantly, reduced levels of phospho-STAT5 in fetal liver pro-B (compared to bone marrow pro-B cells) correlated with increased Rag expression (Figure 3A). Thus reduced levels of phospho-STAT5, impacts on the cell cycle profile and levels of Rag expression in the two environments.

Figure 3. Phosho-STAT5 levels inversely correlate with Rag expression.

(A) Intracellular staining and FACS analysis of phospho-STAT5 levels in pro-B cells (CD19⁺/c-kit⁺) derived from bone marrow and E16 fetal liver. Phospho-STAT5 (blue) staining and the IG control (red) are shown for each cell type (left panels). Q-PCR analysis of the level of Rag1 transcripts in bone marrow and fetal liver derived pro-B cells is shown (right panel). Transcripts were normalized against β₂ microglobulin and the bone marrow pro-B cell population set at 1. (B) Comparison of phospho-STAT5 levels in wild-type and caStat5 pro-B cell (CD19⁺/CD25^-) short-term cultures (left panels). Corresponding Rag1 transcriptional analysis is also shown (right panel). (C) Comparison of phospho-STAT5 levels across cell cycle stages within wild-type (top) or caStat5 (bottom) pro-B cell cultures as described in Figure 1B (left panels). (D) Graph showing the percentage of association of the Rag locus with PCH as determined by 3-D DNA FISH in sorted wild-type or caStat5 pro-B short-term cultures. A γ-satellite DNA probe was used in conjunction with a probe that hybridizes to the Rag locus. 200 cells of each genotype were scored. Experimental variability is shown in Supplementary Table 2. Data is representative of 3 experiments. Data from two individual experiments is shown in Supplementary Table 2 to show the low variability between experiments.

To extend these observations we next asked whether reciprocal alterations in phospho-STAT5 levels also influence Rag expression. For this we analyzed Rag expression and cell cycle status in pro-B cells isolated from mice that express a constitutively active form of STAT5b (caStat5). In these mice, although transgenic STAT5b can be activated independent of IL-7 signals, it becomes hyper-phosphorylated upon IL-7 stimulation and decays at a slower rate (24). STAT5 is phosphorylated in both wild-type and caStat5 pro-B cells that have been cultured short term in the presence of IL-7. In the overall population, phospho-STAT5 levels are slightly increased in transgenic cells (Figure 3B), but importantly, in contrast to their wild-type counterparts, phopho-STAT5 levels were not reduced in G0/G1 caStat5 pro-B cells (Figure 3C). However, increased levels of phospho-STAT5 did not correlate with an increased percentage of caStat5 S/G2/M cells (Supplementary Figure Bii). To ensure equivalent developmental status we sorted CD19⁺CD25^- pro-B cell from the two cultures before analyzing Rag expression. Consistent with our previous observations we found that in caStat5 pro-B cells Rag transcripts were reproducibly decreased compared to wild-type controls (Figure 3B).

To determine whether STAT5 was responsible for Rag locus repositioning to PCH, 3-D DNA FISH was performed on wild type and caStat5 sorted pro-B cells using probes that hybridize to Rag and γ-satellite. As shown in Figure 3D, we found an increased frequency of association of Rag with PCH in caStat5 pro-B cells compared to wild-type counterparts (data from 2 independent experiments is shown in Supplementary Table 2A). In sum these data demonstrate that phosho-STAT5 plays a role in repositioning the Rag locus to a repressive compartment of the nucleus and downregulating its expression in a manner that relates to cell cycle status.

STAT5 binds downstream of the Rag1 locus in pro-B cells

The transcriptional regulation of the Rag genes is complex. The two genes (Rag1 and Rag2) are closely linked, convergently transcribed and share multiple regulatory elements within a 110 kb region, including several lineage specific enhancers (19, 34, 35). To date Erag is the only defined B cell regulatory element involved in the transcriptional regulation of the Rag genes (2, 19). To test whether Erag is responsible for Rag locus positioning relative to PCH within the nucleus we analyzed Rag locus recruitment to PCH in Erag null mice relative to their wild-type counterparts. Deletion of Erag has been previously been shown to result in reduced B lineage specific transcription of Rag (Figure 4A). Intriguingly, despite reduced transcription within Erag null pro-B cells, we observed no difference in Rag locus association with PCH (Figure 4B) and Supplementary Table 3). These data allow us to draw two important conclusions. First, association of the Rag locus with PCH is not always linked to transcriptional repression because deletion of Erag does not affect nuclear localization. Secondly, an alternate STAT5 dependent regulatory element is involved in Rag locus repositioning.

Figure 4. STAT5 binds downstream of the Rag1 locus in pro-B cells.

(A) Q-PCR analysis of the level of Rag1 transcripts in wild-type and Erag^-/- pro-B cells (B) Graph showing percentage association of the Rag locus with PCH in wild-type and Erag^-/- pro-B cells. Over 180 cells of each genotype were counted for the pair. Data from two individual experiments is shown in Supplementary Table 3 to show the low variability between experiments. (B) ChIP-seq enrichment and peaks of STAT5 and H3K27me3 at the Rag1 and Rag2 loci and surrounding genomic region encompassing known regulatory elements Erag and the ASE as well as the newly identified STAT5/H3K27me3 enriched binding site. Genomic location of these elements is indicated.

To identify putative repressive STAT5 binding sites within the 110 kb region of the genome containing the Rag genes and their regulators, we analyzed ChIP–seq data sets of H3K27me3 and STAT5 that we obtained from a recent analysis of these factors (25). These ChIP-seq experiments were performed by the Clark lab using Rag2^-/- pro-B cells cultured in high IL-7 (10ng/ml). Using these data sets we identified several small peaks of STAT5 within the 110 kb Rag region that encompasses all the known regulatory elements (Figure 4B). One of these peaks is located on Erag. However, only one site, located approximately 6Kb downstream from the Rag1 locus on Chromosome 2 at 101495,000, was enriched for both STAT5 and the repressive histone modification H3K27me3 (Figure 4B). (It should be noted that we cannot rule out that other H3K27me3 enriched STAT5 binding sites can be found in these cells within the Rag2 locus because at least part of this locus is deleted in these cells.) The site that we have identified downstream of the Rag1 locus is a previously unidentified site which we have named PRCC (putative regulator of RAG expression through cell cycle). To validate the role of this element in downregulating expression of Rag1 in G2/M in pro-B cells targeted deletion of this region would be required.

IL-7 concentration influences phospho-STAT5 levels across cell cycle

Our data suggests a role for the dynamic regulation of phospho-STAT5 levels in controlling the expression of Rag1 in a manner that relates to cell cycle status. However it remains unclear how IL-7R/phospho-STAT5 levels are controlled within the pro-B cell stage. IL-7R and phospho-STAT5 are only ablated in a proportion of wild-type pro-B cells in G0/G1 (Figure 1A and B), indicating that levels do not change during each replication cycle. Additionally, constitutive activation of phospho-STAT5 (either via Abl transformation or the presence of a constitutively active transgene) negates alterations in phospho-STAT5 levels relative to cell cycle status. We next asked whether increasing IL-7 concentrations could “mimic” constitutive activation of the pathway, effectively reducing the proportion of cells that downregulate phospho-STAT5 levels in G0/G1. For this analysis we cultured pro-B cells in increasing concentrations of IL-7 and analyzed phospho-STAT5 levels by intracellular staining (Figure 5A). In previous experiments, pro-B cell cultures were established on bone marrow stroma (which secrete IL-7) using media containing 5ng/ml of recombinant IL-7. During a 6-10 day culture half of the media was replaced a total of 2 times. Under these conditions, it is unclear what the actual IL-7 concentration is at any particular time point of the culture. We therefore revised these conditions to more accurately control IL-7 levels. Pro-B cell cultures were established during 6-7 days as described above. We then removed the pro-B cells, washed them and replated them in the absence of stroma in 10, 5, 2.5, 1 ng/ml or no IL-7 for ~36 hrs and directly analyzed phospho-STAT5 levels within the G0/G1 or S/G2/M compartments. Strikingly, the proportion of cells that contain phospho-STAT5 within the G0/G1 population remains consistent within cells cultured in the highest concentrations of IL-7. However, this proportion is visibly altered between the 2.5 and 1 ng/ml culture conditions (Figure 5A). We note that the percentage of cells within S/G2/M did not change significantly until IL-7 was completely withdrawn from the cultures. This data again separates phospho-STAT5 levels and cell cycle status. We conclude that the regulation of phospho-STAT5 levels in G0/G1 is influenced by the amount of available IL-7. However additional mechanisms must also exist to downregulate phospho-STAT5 when IL-7 is present in excess. Though Rag transcription is reduced at low concentrations of IL-7 within G0/G1 cells this may be an indirect effect as IL-7 withdrawal induces progression to the pre-B cell stage.

Figure 5. IL-7 concentration influences phospho-STAT5 levels across cell cycle.

(A) Intracellular FACS analysis of phospho-STAT5 levels across the cell cycle within pro-B cells cultured in decreasing amounts of IL-7. Phospho-STAT5 is shown in blue and an isotype (IG) control is shown in red. Cell cycle profiles within each culture condition assessed by DAPI staining are shown within histograms on the far right. (B) Model shows how IL-7 segregates proliferation and recombination in pro-B cells via cell cycle mediated control of Rag transcription.

Discussion

Genetic studies have shown that it is critical to restrict recombination to the G0/G1 phase of the cell cycle in order to prohibit translocations. Periodic destruction of RAG2 protein during S phase entry is one mechanism by which this is accomplished. As with most important biological processes multiple layers of protection are required. Here we elucidate an additional regulatory constraint that segregates proliferation from recombination during early B cell development. We find that IL-7R is expressed within a larger proportion of S/G2/M than G0/G1 pro-B cells, and that S/G2/M cells have reduced Rag expression as a consequence of the action of the downstream IL-7 signaling effector molecule, STAT5. These data are consistent with a known role for IL-7 in promoting proliferation, as well as inhibiting Rag transcription.

The early development of B and T cells share many parallels; each proceed through windows of locus specific recombination flanked by bursts of proliferation, thus developmental stages between the two lineages are often compared. Double negative T cells are considered to be the counterpart for pro-B cells. Yet double negative T cells are subdivided into four distinct compartments with different cell cycle profiles. It is thus not unexpected that pro-B cells can also be subdivided into different compartments with different cell cycle profiles. A role for IL-7 in this subdivision is supported by studies showing that constitutive expression of IL-7R within hematopoietic progenitor cells blocks B cell development at the pre-pro-B cell stage (39). A similar developmental block is seen within mice that express c-myc and bcl-2 constitutively (40), a state in which cell cycle exit or pausing in G0/G1 would be unlikely. In the context of the data presented here, these phenotypes could in part be explained by an inability of those cells to exit cell cycle and up-regulate Rag expression for recombination. Because D-J rearrangement is initiated within the Igh locus at the pre-pro-B cell stage it is tempting to speculate that IL-7 signals are transiently down-regulated at each stage in which recombination occurs. In fact an IL-7R negative or low population has been found within pre-pro-B cells (41). However it is unlikely that the block seen in cells expressing a constitutive IL-7R would be solely due to differential Rag expression as Rag deficient mice are able to proceed to the pro-B cell stage. Instead it is more likely that dynamic regulation of the IL-7 signaling pathway is required for regulating additional genes outside of the recombination pathway. Indeed we provide evidence that differential expression of STAT5 targets is not limited to Rag.

Interestingly we found that IL-7 levels regulate the proportion of cells with phospho-STAT5 in G0/G1. An intriguing possibility is that dividing cells in the bone marrow move back and forth between IL-7 expressing and non-IL-7 expressing stroma, resulting in a transient attenuation of IL-7R signaling. Indeed, IL-7R/phospho-STAT5 levels do not change each time a cell passes through G0/G1, as evidenced by our data showing a bimodal expression within G0/G1 cells. This suggests that only a subset of daughter cells downregulate IL-7R/phospho-STAT5, a result that could be attributed to either limiting concentrations of IL-7 or asymmetric cell division. One idea is that these two possibilities are linked. In this model the process of cell division causes one daughter cell to be positioned further away from IL-7 stroma cells in a manner that downregulates the IL-7 receptor signaling pathway. These possibilities need to be further investigated.

Taking into account that IL-7 signals are generally thought to promote Igh locus accessibility for recombination (36, 42-45), we propose a model for an ordered subdivision of the pro-B cell compartment based on IL-7R expression (Figure 5B). In this model the (first) actively cycling, IL-7R positive pro-B subpopulation prepares the Igh locus for recombination by enhancing accessibility. In the subsequent G0/G1 IL-7R negative population, Rag is upregulated to recombine the accessible Igh locus. Cells could transit between these states until recombination is complete. The exact relationship between the IL-7R/phospho-STAT5 low/negative population and IL-7R/phospho-STAT5 high population remains to be determined. While it is possible that IL-7R/phospho-STAT5 is regulated dynamically during the cell cycle in individual cells, the bimodal expression of IL7-R/phospho-STAT5 within G0/G1 is consistent with the existence of a population of IL-7R/phospho-STAT5 low/negative daughter cells that truly exits the cell cycle (at least temporarily) to allow recombination to occur.

In the B cell lineage, IL-7R/STAT5 is known to have a role in promoting the sequential ordering of recombination of the immunoglobulin loci (16, 36, 46). Now, we find that IL-7R/STAT5 is also involved in segregating recombination and cell division by differentially regulating the transcription of key gene targets involved in the recombination process in a manner that is linked to cell cycle progression. Unfortunately, since current antibodies do not allow detection of endogenous RAG1 within ex-vivo derived wild-type pro-B cells (47) or in ex-vivo cultured cells we were not able to confirm these findings at the protein level. Instead we focused on transcriptional regulation via STAT5 and regulatory mechanisms associated with transcriptional control within the nucleus.

STAT5 can act as both an activator and a repressor of transcription (25). Recent studies indicate that H3K27me3 is enriched at sites where STAT5 acts as a repressor. Interestingly through analysis of the Clark lab ChIP-seq data sets for STAT5 and H3K27me3, we uncovered a putative new regulatory element (PRCC) that binds both factors at a site downstream of the Rag1 locus in pro-B cells. Though this element lies outside what is considered to be the main promoter region, it is certainly possible that it functions to control promoter activity, which in turn affects transcription levels. We note that the Rag1 promoter is positively regulated by E2A and the IL-7R/STAT5 pathway has previously been shown to prohibit E2A function and binding at the Igκ locus (46, 48). It will be interesting to determine whether an important element of STAT5 repressive function is the antagonism of the E2A protein. Further genetic targeting studies will be required to determine whether this element can regulate cell cycle expression of Rag1 through the repositioning of the latter to PCH.

It is clear that not all of the targets of STAT5 are regulated similarly across the cell cycle, as indicated by differences in the behavior of the Igh and Rag loci. We have previously shown that STAT5 can maintain the accessibility of Igh by keeping it euchromatic (18), while STAT5 regulates Rag in an opposite manner. The cell cycle regulation of Rag by STAT5 occurs at the transcriptional level and correlates with the repositioning of the Rag locus to PCH. As with all studies that link gene silencing with PCH repositioning there is no clear indication as to whether a change in nuclear localization is a cause or a consequence of repression. How STAT5 functions in this context remains to be determined.

IL-7R/STAT5 has been shown to regulate the expression of genes involved in survival, proliferation and lineage specification (36, 39, 49). Importantly, STAT5 is rendered constitutively active and independent of extrinsic signals in numerous cancers (50). It is possible that some STAT5 targets outside of the recombination pathway are also differentially regulated within pro-B cells in a manner that relates to cell cycle status. In this context, changes in gene dosage across cell cycle could ultimately promote oncogenesis in many different settings as indicated by recent findings which show that alterations in the levels of transcription factors cooperate with STAT5 to initiate ALL (38).