Differential Regulation of Mouse B Cell Development by Transforming Growth Factor β1

Transforming growth factor β (TGFβ) can inhibit the in vitro proliferation, survival and differentiation of B cell progenitors, mature B lymphocytes and plasma cells. Here we demonstrate unexpected, age-dependent reductions in the bone marrow (BM) B cell progenitors and immature B cells in TGFβ1-/- mice. To evaluate TGFβ responsiveness during normal B lineage development, cells were cultured in interleukin 7 (IL7)±TGFβ. Picomolar doses of TGFβ1 reduced pro-B cell recoveries at every timepoint. By contrast, the pre-B cells were initially reduced in number, but subsequently increased compared to IL7 alone, resulting in a 4-fold increase in the growth rate for the pre-B cell population. Analysis of purified BM sub-populations indicated that pro-B cells and the earliest BP1- pre-B cells were sensitive to the inhibitory effects of TGFβ1. However, the large BP1+ pre-B cells, although initially reduced, were increased in number at days 5 and 7 of culture. These results indicate that TGFβ1 is important for normal B cell development in vivo, and that B cell progenitors are differentially affected by the cytokine according to their stage of differentiation.

Limitations to studying TGFb effects in vivo are imposed by its importance during embryogenesis. Gene-targeting of each isoform (TGFb1, 2 or 3) as well as each of the two receptor subunits (TbRI and TbRII) results in lethality at or prior to birth Kaartinen et al., 1995;Martin et al., 1995;Bonyadi et al., 1997;Sanford et al., 1997). The earliest lethality is seen in TbRI 2/2 , TbRII 2/2 (Larsson et al., 2001), and in , 50% of TGFb1 2/2 embryos (Shull et al., 1992;, which expire at , 10.5 days post-coitus, due to aberrantly developed yolk sac vasculature and anemia. The embryonic anemia in vivo is likely a secondary result of inadequate vascularization, since endothelia from the TGFb1 mutant embryos fail to differentiate in culture , whereas in vitro development of yolk sac-derived TbRI 2/2 hematopoietic progenitors into various blood cell lineages is similar to controls (Larsson et al., 2001).
Once a hematopoietic progenitor enters the B lineage pathway, it progresses through a number of developmental stages defined by expression of cell surface differentiation antigens (Hardy et al., 1991;Rolink et al., 1999), cell cycle status (Osmond, 1991;Itoh et al., 1996), antibody variable region gene rearrangements (Hardy et al., 1991;Hardy, 1992;Li et al., 1993;Papavasiliou et al., 1997;Rolink et al., 1999), responsiveness to and requirements for interleukin 7 (IL7) receptor signaling (Peschon et al., 1994;Candeias et al., 1997;Marshall et al., 1998) and interaction with the bone marrow (BM) stroma (King et al., 1988;Gimble et al., 1989;Dittel et al., 1993;Dittel and LeBien, 1995;Borghesi et al., 1997). The IL7 receptor is indispensable for mouse B cell development during the V-to-DJ heavy chain (HC) variable region gene rearrangement process (Corcoran et al., 1998). Acquisition of mHC and formation of the pre-B cell receptor are associated with a decreased IL7 dose-response threshold (Marshall et al., 1998). The resulting increases in IL7 sensitivity may be responsible for the large size and mitotic status of early/intermediate pre-B cells. Late pre-B cells exit the cell cycle and undergo light chain V -J rearrangement in preparation for full antibody assembly and surface expression on the more differentiated B cell (Meffre et al., 2000;Melchers et al., 2000).
The effects of exogenous TGFb have been examined in cultured B lineage cells representative of almost every developmental stage, and are usually inhibitory. Early studies showed that TGFb inhibits the proliferative response of BM B cell progenitors (BCP) to IL7, and that it can inhibit k LC acquisition in a differentiating B lineage cell line (Lee et al., 1987;Kincade et al., 1989). Similar observations have been made for k light chains in human fetal BM cultures (Rehmann and LeBien, 1994). However, these studies did not distinguish the effects of TGFb on pro-B versus pre-B cells within one system. Induction of the transcriptional regulator Id3 by TGFb, together with inhibition of cell cycling and Rag1 mRNA expression has also been demonstrated (Kee et al., 2001). TGFb effects at later mature B and plasma cells stages are almost exclusively negative with the exception of inducing IgA isotype switching (Kim and Kagnoff, 1990;Lebman et al., 1990;Shockett and Stavnezer, 1991). These studies indicate that TGFb can inhibit the in vitro survival, proliferation and differentiation of antibody-producing B cells at all stages of development.
An inhibitory role for TGFb in the immune system is supported by the phenotype of juvenile TGFb1 2/2 mice (Christ et al., 1994). Infiltrates of plasma cells are found in secondary lymphoid organs and also in non-lymphoid tissues where they accompany inflammatory infiltrates of other hematopoietic cells (Christ et al., 1994;Kulkarni et al., 1995;van Ginkel et al., 1999). The mice also have increased levels of anti-nuclear and anti-collagen serum antibodies (Dang et al., 1995;Yaswen et al., 1996) and hyperproliferation in the splenic B cell follicles (Christ et al., 1994).
These observations, together with the described inhibitory effects of TGFb on in vitro B cell development, predicted the expansion of B cell progenitors in the absence of TGFb in vivo. We found instead an age-related deficiency in B cell development in TGFb1 2/2 mice. The complication of the co-existing inflammatory disease in these mice lead us to re-examine the in vitro effects of TGFb1 on defined sub-populations of normal BM B lineage cells. The results of this combined approach indicate that deficiencies in the earliest B cell progenitors in the TGFb1 2/2 mice are likely to be due to secondary effects of the phenotype, since pro-B cell growth is inhibited by TGFb1 in vitro. By contrast, TGFb1 increases the recovery of the large pre-B cells. Collectively, these observations demonstrate that TGFb1 is required for normal B cell development in vivo, and indicate differential sensitivity of B cell progenitors to TGFb according to their stage of differentiation.

Cell Culture
Erythrocyte-depleted BM from 4-to 5-week-old female C57BL/6 mice was purified by centrifugation over a Lympholyte M gradient (Cedar Lane, Hornby, Ont., Canada). B220 þ cells were isolated by positive selection with magnetic beads (Miltenyi, Auburn, CA); sorting efficiency was assessed by flow cytometry to be 85^8%. FACS w -sorted cells were purified with a MoFlo flow cytometer (Cytomation, Fort Collins, CO) using anti-CD19 (clone 6D5 from Southern Biotechnology Associates, Birmingham, AL), anti-BP1 and anti-IgM (as above). 5 £ 10 4 sorted cells/ml were plated in complete IMDM (5% FCS, 5 £ 10 25 M 2ME, 1% each L-glutamine, penicillin/streptomycin, non-essential amino acids) and treated with 10 ng/ml recombinant mouse IL7 (PeproTech, Rocky Hill, NJ)^recombinant human TGFb1 (R and D Systems, Minneapolis, MN). Two doses of 0.04 ng/ml (1.6 pM) or 1 ng/ml (40 pM) were compared in all of the experiments shown here because dose-response experiments showed distinct read-outs at these two concentrations. Harvested cells were counted by Trypan Blue exclusion and surface-stained for B220 (as above) and sIgM [goat anti-mouse IgM cy5 from Jackson Laboratories (Bar Harbor, ME) or MB86 Alexa647 from John Kearney (Birmingham, AL)], or control antibodies. These cells were fixed in 1% paraformaldehyde, permeablized with Tween 20 and stained intracellularly with a goat antimouse m F antibody. For some experiments, the CytoFix/ Cytoperm kit from Pharmingen was used according to the manufacturer's instructions. The pre-B cell growth rate was calculated as (pre-B cell number recovered on day 7 2 pre-B cell number recovered on day 3)/4 days ¼ cells/day.

Age-dependent BM B Lineage Cell Reductions in TGFb1 2/2 Mice
Flow cytometry was used to examine the proportions (Fig. 1A,B) and absolute numbers ( Fig. 1C) of B220 þ B lineage cells in the BM of neonatal (1.5-week-old) and juvenile (3.5-week-old) TGFb1 2/2 mice. The 1.5 weekold TGFb1 2/2 mice were comparable to the TGFb1 þ/þ littermate (LM) controls at the early, sIgM 2 and later, sIgM þ stages. By contrast, 3.5-week-old mice showed a significant 2.6-fold reduction in the percentage of B220 þ sIgM 2 cells, corresponding to a significant 4.6fold reduction in absolute cell number. Absolute numbers and percentages of B220 þ sIgM þ B cells were also reduced, although not consistently.
The cell surface marker system described by Hardy et al. (1991) was used to further define the B lineage developmental stages affected by the TGFb1 deficiency, and the results were calculated both as a percentage of total BM ( Fig. 2A,B) and as absolute cell numbers (Fig. 2C). In TGFb1 2/2 mice examined at 3.5 weeks of age, the percentage of cells in Fraction B (B220 þ CD43 þ HSA þ BP1 2 ), including pro-and pre-B cells (Wasserman et al., 1997), was not significantly changed, although the absolute numbers of these cells were 2.6-fold reduced. The percentage and absolute number of  were significantly reduced by 3.0-and 5.0fold, respectively. The late pre-B cells in Fraction D (B220 þ CD43 2 IgM 2 IgD 2 ) were proportionally reduced by 4.6-fold and in absolute cell number by 8.2fold. The subsequent immature B cells, Fraction E (B220 þ CD43 2 IgM þ IgD 2 ) were proportionally reduced by 5.7-fold and in absolute number by 9.2-fold. Mature B cells (B220 þ CD43 2 IgM þ IgD þ , Fraction F) in the BM were variably increased or decreased in proportion and in absolute number (Fig. 2C) as was observed in the periphery [not shown and Christ et al. (1994)]. It should be noted that Fraction A (B220 þ CD43 þ HSA 2 BP1 2 ) is not consistently affected in the TGFb1 2/2 mice ( Fig. 2A and not shown); however, not all the cells in this population are progenitors of the B lineage .
In vivo, the reductions seen in Fraction B were statistically significant when calculated as absolute numbers of cells recovered (2.6-fold), but not as a percentage of total BM cells. This might suggest that Fraction B itself is unchanged, and that the smaller size of the TGFb1 2/2 mice (Shull et al., 1992;Boivin et al., 1995;Kulkarni et al., 1995), and therefore smaller bone cavity, is the cause of this reduction in cell number. However, normalizing the cell number to body weight of each mouse still results in a significant reduction in Fraction B, although the degree of reduction is less, at 1.8fold (þ /þ , 1:46 £ 10 4^4 :98 £ 10 3 cells=g; 2 /2 , 8:08 £ 10 3^2 :27 £ 10 3 cells=g; p ¼ 0:048). The reduction in Fraction C, however, is significant regardless of how the data are calculated: as a proportion of total BM (3.0-fold), as an absolute cell number (5.0-fold), and also as a normalized cell number [3.6-fold (þ /þ , 2:90 £ 10 36 32 cells=g; 2 /2 , 810^463 cells=g; p ¼ 0:00091)].
A summary of a more extensive analysis of mice at different ages is shown in Table I as the frequency of TGFb1 2/2 mice with reductions of $ 2-fold in the absolute numbers of BM B lineage cells. These mice are not included in Fig. 2 because a different number of bones per mouse were used for the analysis. Although there is some variability, these results confirm the age-dependent decrease in BCP in the TGFb1 2/2 mice. The frequencies of mature B cells are reduced in slightly more than half of the mice, which may be due to variable experiences of these recirculating cells in the periphery. Phenotypic variability in TGFb1 2/2 mice has been noted in other contexts, for example, inflammation in organs other than heart and lung (Shull et al., 1992;Boivin et al., 1995;Kulkarni et al., 1995).
Activated T cells are responsible for much of the characteristic inflammatory phenotype of the TGFb1 2/2 mouse Kobayashi et al., 1999). To ask whether the BCP reductions in the TGFb1 2/2 mice were associated with an altered BM microenvironment, Thy1.2 and Mac1 were used as markers for BM T and myeloid cells, respectively. Although the Thy1.2 þ population was proportionally increased in all 3.5-weekold mice examined (Fig. 3A), this did not correlate with increased cell numbers in most mice (Fig. 3B). The proportions of Mac1 þ cells were increased in half of the mice examined (Fig. 3C); however, once again, this generally did not correspond to an increase in the absolute numbers of myeloid lineage cells (Fig. 3D). One explanation for the observed discrepancy between the percentages and absolute cell numbers may again be the smaller size of the TGFb1 2/2 mice (Shull et al., 1992;Boivin et al., 1995;Kulkarni et al., 1995). However, when the data were normalized to body weight, a similar pattern was observed for Thy1.2 (þ /þ , 1:8 £ 10 4^1 :1 £ 10 4 cells=g; 2 /2 , 4:4 £ 10 4^2 :6 £ 10 4 cells=g), and for Mac1 (þ /þ , 7:5 £ 10 4^1 :1 £ 10 4 cells=g; 2 /2 , 11:1 £ 10 4^4 :8 £ 10 4 cells=g). These findings show that there is an altered cellular composition of the BM in the TGFb1 2/2 mice, including changes in the proportions of  (Hardy et al., 1991;Li et al., 1993;Li et al., 1996). myeloid and T lineage cells. However, the absolute numbers of these cells are not consistently increased in all of the mice that had an equally severe reduction in BCP. Therefore, a global disruption of the BM microenvironment seems unlikely, although our analysis does not exclude possible inhibitory effects of inflammatory/ myelopoietic foci on the B lineage cells in TGFb1 2/2 BM.

Exogenous TGFb1 Effects upon Normal Pro-and Pre-B Cells In Vitro
The complex pathology of the TGFb1 2/2 mice and the lack of correlation between the frequency of Thy1.2 þ and Mac1 þ BM cells and the BCP deficiency lead us to re-examine the effects of TGFb1 on B cell development in vitro. We asked whether TGFb1 might be beneficial for B cell development as suggested by the phenotype of the TGFb1 2/2 mice. B220 þ BM B lineage cells from normal (C57BL/6) mice were treated with TGFb1 in the presence of IL7, a cytokine that stimulates BCP proliferation prior to the late pre-B cell stage (Hardy et al., 1991;Marshall et al., 1998). Stromal cells were not included in our system due to their ability to produce and respond to TGFb (Dittel et al., 1993;Dittel and LeBien, 1995;Robledo et al., 1998;Olsen et al., 2001). Intracellular (i.c.) mHC expression in B220 þ sIgM 2 BCP was used to identify pre-B cells recovered from cultures of IL7-stimulated B220 þ BM cells (Fig. 4A). At day 3 in the control sample of IL7 alone (first column), there was a predominance of pre-B cells; however, the pro-B cell-enriched population (i.c. mHC 2 BCP) predominated by day 7. It should be noted that a majority of the pre-B cells in the starting population are late pre-B cells, which do not proliferate in response to IL7 (Hardy et al., 1991;Marshall et al., 1998). During the course of a weeklong culture, these cells should either expire or mature to become sIgM þ B cells and thus be excluded from the analysis. Meanwhile, IL7-responsive pro-B cells accumulate and predominate in the cultures by day 7.
Addition at day 0 of either 0.04 or 1 ng/ml TGFb1 to the IL7 cultures resulted in little change at day 3 in the percentage of pre-B cells compared to IL7 alone (Fig. 4A). However, both treatments resulted in a consistent reduction in the total viable cell numbers recovered (not shown) corresponding to a 2-fold reduction in both proand pre-B cell numbers at day 3 of culture [ Fig. 4B (Bottom) and C, respectively].
At the later timepoints of 5 and 7 days, TGFb1 treatment resulted in a small increase in the proportion of pre-B cells in comparison to IL7 alone (Fig. 4A). This was partially due to reductions in pro-B cell numbers in the presence of TGFb1 (Fig. 4B). In contrast to reductions in pre-B cell numbers seen at day 3 of culture, treatment with 0.04 ng/ml TGFb1 resulted in a modest increase in the numbers of pre-B cells recovered at 5 and 7 days. By contrast, treatment with 1 ng/ml TGFb1 showed minor pre-B cell reductions at later timepoints (Fig. 4C).
The initial reductions in pre-B cells indicate that some cells are likely to be sensitive to the inhibitory effects of TGFb1. However, the remaining pre-B cell population either has, or acquires a very high proliferative capacity, indicated by a significant 4-fold increase in the rate of growth between 3 and 7 days (Fig. 4D).
To identify the populations responsible for these distinct outcomes, we initiated similar experiments using sortpurified BM. The BP1 cell surface marker was used to subdivide normal BM into two sub-populations of CD19 þ sIgM 2 BCP: (1) BP1 2 cells, , 60% of which are i.c. mHC þ and (2) large BP1 þ cells, , 90-99% i.c. mHC þ (not shown). Between days 3 and 7 in culture with IL7 alone, the number of BP1 2 -derived pro-and pre-B cells increased (Fig. 5A,B, respectively); in each case and at all time points, 0.04 and 1 ng/ml exogenous TGFb1 resulted in reduced cell recovery. TGFb1-mediated reductions in pro-B cell recoveries were also confirmed using CD19 þ Rag1 2/2 BM as a source of pro-B cells (not shown). In contrast to the cell growth observed with IL7 alone in cultures of BP1 2 BCP, the large BP1 þ pre-B cells decreased in number between days 3 and 7 (Fig. 5C). In this context, treatment with either dose of TGFb1 resulted in an initial decrease in pre-B cell recoveries, but subsequently, low-dose TGFb1 treatment resulted in a net increase at days 5 and 7 of culture (2.1^0.7 and 3.0^0.6-fold increase, respectively over IL7 alone; n ¼ 3 experiments). Figure 5D summarizes the effects of low-dose TGFb1 treatment where the diameter of each circle relative to the control (1.0) represents the average fold change in cell number. As seen on day 3, each stage analyzed contains cells that are sensitive to the inhibitory effects of TGFb1. TGFb1-mediated reductions continue over time for BP1 2 BCP-derived pro-and pre-B cells, whereas the BP1 þ fraction contains cells that are positively affected by exogenous TGFb1 treatment at a low-dose (0.04 ng/ml). At a later timepoint in the culture, IgM þ B cells accumulate as well ( Fig. 5D and data not shown).

DISCUSSION
We have demonstrated an age-dependent reduction in BM B lineage cells in TGFb1 2/2 mice. The deficiency is apparent as early as Hardy's Fraction B, containing pro-B cells and extends through the immature B cell stage. The mice also have variable increases in the proportions and numbers of BM Thy1.2 þ and Mac1 þ cells, but these do not consistently correlate with the reduction in B lineage cells. Subsequent studies of the in vitro effects of exogenous TGFb1 on BM B lineage cells from normal mice showed reductions in pro-B cell recoveries as early as day 3 and continuing through day 7 of incubation at pg/ml doses of TGFb1. Although the same cultures showed an initial decrease in pre-B cell numbers, this was followed by increases at days 5 and 7 translating into a 4-fold increase in the rate of growth for the pre-B cell population. By sort-purification of the starting BCP subpopulations, the large BP1 þ pre-B cell fraction was identified as containing the cells that are increased in response to TGFb1.
Reductions in BM-BCP in TGFb1 2/2 mice were unanticipated because TGFb had been previously shown to have inhibitory effects upon the B lineage in vitro (Lee et al., 1987;Kincade et al., 1989;Lee et al., 1989;Rehmann and LeBien, 1994;Kee et al., 2001). The reductions in vivo may thus be due to an unrecognized necessity for a TGFb receptor signal directly upon B lineage cells. Alternatively, effects secondary to the TGFb1 deficiency, e.g. soluble factors produced by infiltrating inflammatory cells, circulating prostaglandins or sex hormones Wang et al., 1995;Kincade et al., 2000), which could be directly regulated by TGFb1 or induced in response to inflammatory stress may be responsible. Alterations in cellular adhesion may also contribute to dysregulated B lymphopoiesis in the BM (Dittel et al., 1993;Dittel and LeBien, 1995).
B cell development is apparently normal in very young (1.5 week) TGFb1 2/2 mice and deteriorates thereafter. This age dependency may be due to maternal TGFb1, acquired in utero or during nursing, compensating early in life for the lack of de novo production (Letterio et al., 1994) and delaying the onset of defects in B lymphopoiesis. Alternatively, BCP derived from older mice may have a differential sensitivity to TGFb1 in comparison to those generated earlier in life, as differences have been observed for BCP at different stages of ontogeny (Kearney et al., 1997;Hardy et al., 2000;Igarashi et al., 2001). In either case, the mature B cells found in the BM and periphery of 3-to 5-week old TGFb1 2/2 mice are likely generated at an earlier age when B cell development is unaffected by the TGFb1 mutation.
The analysis of T and myeloid cells in individual TGFb1 2/2 mice indicates variable alterations in the cellular composition of the BM; however, there was no consistent correlation with BCP reductions. Moreover, when TGFb1 2/2 mice are rendered deficient in CD8 þ T cells, by backcrossing to a b2-microglobulin (MHC Class I) deficient genetic background, which prevents the T cell-mediated inflammatory disease, total B220 þ BM B lineage cells are still reduced (Kobayashi et al., 1999). This deficiency in B lineage cells, as in our studies, is specific to the BM, and is not observed in the spleen. We have attempted to address the role of the T cell-mediated inflammatory response in the BCP deficiency by breeding TGFb1 2/2 and TCRa 2/2 mice. However, no doubly homozygous mutant offspring were obtained.
Another approach has been to examine B cell development in a model where only B lineage cells are unresponsive to TGFb. In mice conditionally gene-targeted for the TbRII subunit of the TGFb receptor specifically in B cells, the early B lineage cells in the BM were reportedly unchanged in comparison to the controls (Cazac and Roës, 2000). However, the CD19cre deletion system used in these studies is less efficient in late pre-B cells (75 -80%) than in splenic B cells (90 -95%) (Rickert et al., 1997). It is thus unclear whether and at what efficiency TbRII deletion occurs at earlier stages of B cell development, since TGFb responsiveness was not examined in the BM B lineage cells of these mice. However, in the periphery, where B lineage cells had defective TGFb receptor signaling, populations of mature B lymphocytes were increased, as were serum levels of anti-nuclear antibodies indicating a direct inhibitory role for TGFb on B lineage cells at later stages (Cazac and Roës, 2000).
If BCP deficiencies in TGFb1 2/2 mice are due to a requirement for a direct TGFb receptor signal on these cells, we reasoned that using a defined culture system with rIL7 and purified BM B lineage cells from normal mice, we could identify sub-populations that benefit from exposure to exogenous TGFb1. In IL7-containing cultures, pro-B cell recoveries were consistently decreased by low-dose TGFb1 treatment. An often-overlooked population of i.c. mHC þ pre-B cells, included in Hardy's Fraction B, is also reduced by TGFb1 in culture. By contrast, the numbers of pre-B cells derived from large BP1 þ BCP were increased. Notably, the more severe in vivo BCP reductions in TGFb1 2/2 mice begin in Fraction C, which corresponds to the large BP1 þ pre-B cells in our cultures.
The effects of TGFb on cell cycle and apoptosis may provide an alternative explanation for our in vitro data. Pro-B cells proliferating in response to IL-7 eventually mature into i.c. mHC þ pre-B cells and then exit cell cycle. At low doses, TGFb may inhibit proliferation of the pro-B cells, which then may complete IgH chain gene rearrangement and express mHC. This outcome would result in the observed decrease in pro-B cells and increase in pre-B cells. Higher doses of TGFb may induce pro-B cell apoptosis, thus accounting for the decrease in both pro-and pre-B cells. However, this interpretation does not account for the net increase in pre-B cells with low dose TGF-b in cultures initiated with large BP1 þ cells, 90-99% of which already express mHC.
In conclusion, TGFb1 influences B cell development in multiple ways including directly inhibiting pro-B cell/early pre-B cell populations. The BP1 þ pre-B cells are unique among B lineage cells in their positive response to exogenous TGFb1. It is currently unclear why the large pre-B cells respond to TGFb in this way. It has been suggested that the stability of the preBCR, which varies depending upon the mHC variable region, may regulate cellular viability and proliferation (Melchers et al., 2000). A pre-B cell with a well-fitting preBCR would receive proliferative/viability signals via the receptor itself, and consequently the cell could be changed in other ways to give it the greatest advantage over cells with a poorly fitting preBCR. Differential responsiveness to TGFb may be one such phenotypic change and would be advantageous for expanding the numbers of these positively selected pre-B cells.