© 2006 Nature Publishing Group http://www.nature.com/natureimmunology
ARTICLES
Regulation of T lymphopoiesis by Notch1 and Lunatic fringe–mediated competition for intrathymic niches Ioana Visan, Joanne B Tan, Julie S Yuan, James A Harper, Ute Koch & Cynthia J Guidos Notch1 activation regulates T lineage commitment and early T cell development. Fringe glycosyltransferases alter the sensitivity of Notch receptors to Delta-like versus Jagged Notch ligands, but their functions in T lymphopoiesis have not been defined. Here we show that developmental stage–specific expression of the glycosyltransferase lunatic fringe (Lfng) is required for coordination of the access of T cell progenitors to intrathymic niches that support Notch1-dependent phases of T cell development. Lfng-null progenitors generated few thymocytes in competitive assays, whereas Lfng overexpression converted thymocytes into ‘supercompetitors’ with enhanced binding of Delta-like ligands and blocked T lymphopoiesis from normal progenitors. We suggest that the ability of Lfng and Notch1 to control progenitor competition for limiting cortical niches is an important mechanism for the homeostatic regulation of thymus size.
Notch1 receptor signaling is required at many early stages of intrathymic T cell development1,2. When thymus-seeding progenitors (TSPs) are unable to activate Notch1, they generate B cells rather than T cells in the thymus3–7, demonstrating that Notch1 activation regulates the T lineage–versus–B lineage ‘decision’. Intrathymic Notch1 activation is required for suppression of the B cell potential of TSPs3,8. Notch1 activation is also necessary for TSPs to produce early T cell progenitors8,9 that are CD4–CD8– double-negative (DN) and CD44hiCD117hiCD25– and regulates their further maturation into CD44hiCD117hiCD25+ DN2 and CD44loCD117loCD25+ DN3 thymocytes8,10,11. Progression beyond the DN3 stage also depends on Notch1 activation10,12, which acts in synergy with pre–T cell receptor (TCR) signals to control survival and proliferation during the DN3–to–CD4+CD8+ double-positive (DP) transition13,14. Finally, Notch signaling can regulate mature T cell activation and differentiation15–19. However, excessive Notch activation causes extrathymic and intrathymic accumulation of developmentally arrested tumorigenic DP thymocytes20–22. Thus, Notch signaling is continuously required for the induction of T lineage commitment, specification and survival during the DN stages but must be attenuated in DP thymocytes to allow their further maturation and to prevent their oncogenic transformation. Those observations indicate stage-specific regulation of intrathymic Notch activation, but the mechanisms that confine Notch1 activation to the DN stage of intrathymic T cell development have not yet been elucidated. Blood-borne TSPs enter the postnatal thymus through postcapillary venules near the corticomedullary junction (CMJ)23. This region is thought to provide a limiting number of specific microenvironmental niches in which TSPs undergo T lineage specification and commitment24,25. Thus, CMJ niches probably contain Notch
ligands, but that has not been demonstrated functionally. Progression through the DN2 and DN3 stages occurs as progenitors undergo transcortical migration from the CMJ to the outer subcapsular zone of the thymus26,27. That region is thought to provide a second specialized niche in which DN3 thymocytes undergo b-selection to generate a large pool of DP thymocytes expressing functional TCRb chains28. Notably, Notch1+/– T cell progenitors generate few DP thymocytes in mixed chimeras because they are ‘out-competed’ by Notch1+/+ progenitors at each DN stage of intrathymic T cell development8,29. Such haploinsufficiency suggests that expression of Notch ligands is functionally limiting in cortical niches that support maturation of early T cell progenitors, but that idea has not been tested experimentally. There are two structurally distinct families of Notch ligands, Deltalike (Dll) and Jagged (Jag), and PCR studies have shown that several members of each family are expressed by thymic epithelium30. However, Jag1 is less efficient than Dll1 or Dll4 in inhibiting B cell development and promoting T cell development from hematopoietic progenitors in vitro31,32. The differential sensitivity of T cell progenitors to Dll versus Jag ligands may reflect modulation of Notch1 activation by Fringe proteins, which enhance Notch activation by Dll ligands but inhibit activation by Serrate or Jag-type ligands33. Fringe proteins are glycosyltransferases that add N-acetylglucosamine moieties to O-linked fucose to the extracellular domains of Notch receptors. Although Fringe proteins can be secreted, they must be localized to the Golgi to exert their normal biological function34, explaining the cell-autonomous activity described in genetic studies35,36. In contrast to core components of the Notch signaling pathway such as RBPJk37,38, Fringe proteins are not essential for Notch signaling. Instead, they coordinate the timing and localization of Notch activation during several developmental processes33.
Program in Developmental Biology, Hospital for Sick Children Research Institute, and Department of Immunology, Faculty of Medicine, University of Toronto, Toronto, Ontario M5G 1L7, Canada. Correspondence should be addressed to C.J.G. (
[email protected]). Received 12 January; accepted 4 April; published online 14 May 2006; doi:10.1038/ni1345
634
VOLUME 7
NUMBER 6
JUNE 2006
NATURE IMMUNOLOGY
ARTICLES
0
10–1
10–2
b
Normalized Lfng mRNA
ND DN2
DN3
DN4
DP
CD4 SP
CD8 SP
1
10
0
10
ND
–1
10
–/–
Rag2 : Lfng :
Tg– Tg+ Endogenous
Tg– Tg+ Transgenic
ND
ND
Tg+ B6 DP: Tg– Lfng : Endogenous
ND Tg+ Tg– Transgenic
Lunatic fringe (Lfng) is the most well studied vertebrate Fringe protein; in vertebrates it is a crucial component of the somite segmentation ‘clock’ during embryogenesis39. Notch-Delta signaling induces and then represses Lfng expression in the presomitic mesoderm, leading to the formation of a new somite every 2 h. Both homozygous deletion and overexpression of Lfng in the presomitic mesoderm cause segmentation defects, albeit less severe than those seen in Notch1–/– mice40,41. Thus, either loss or overexpression of Lfng interferes with Notch1 activation during somitogenesis, demonstrating that precisely regulated Lfng expression is needed to temporally and spatially coordinate Notch1-dependent patterning and segmentation during somitogenesis. Cortical thymocytes overexpressing Lfng inhibit many TSPs from activating Notch1, allowing them to adopt the B cell fate rather than the T cell fate intrathymically3,8. Unexpectedly, that study3 suggested that Lfng can regulate Notch1 activation nonautonomously. Here we demonstrate that Lfng transgene–positive (Tg+) DP thymocytes dominantly blocked TSP access to CMJ niches, explaining why transgenic Lfng seems to act non–cell autonomously to influence the intrathymic T lineage–versus–B lineage ‘decision’3. Furthermore, Lfng-Notch1 interactions controlled access of DN3 thymocytes to cortical niches that support DP thymocyte production, providing a mechanism for
homeostatic regulation of thymus size. Our data suggest that Dll ligands are functionally limiting in the postnatal thymus and demonstrate that the abundance and developmental stage of Lfng expression must be tightly controlled to ensure that Notch1-dependent DN thymocytes, rather than the more abundant and Notch1independent DP population, interact efficiently with Dll ligands in the thymic cortex. RESULTS Lfng function in early T cell development The ability of cortical thymocytes overexpressing Lfng to inhibit Notch1 activation in TSPs and alter the intrathymic T lineage– versus–B lineage ‘decision’3,8 indicates endogenous Lfng is important in thymopoiesis. In accordance with that idea, we noted stage-specific regulation of endogenous Lfng mRNA during T cell development (Fig. 1a). All DN thymocyte subsets and mature CD4 and CD8 singlepositive thymocytes had high expression of Lfng mRNA, but it was undetectable in DP thymocytes. Lfng mRNA and Lfng protein are highly unstable, so the abundance of Lfng mRNA reflects ongoing transcription and protein expression39. As Lfng is also expressed in
a
1.0 0.8 0.6 0.4 0.2 0.0
NATURE IMMUNOLOGY
VOLUME 7
NUMBER 6
JUNE 2006
Lfng FL:
b
WT
Single chimeras 120 DP thymocytes (×106)
Figure 2 Lfng–/– progenitors fail to efficiently reconstitute the thymus in mixed chimeras. Lin– fetal liver cells (FL) from Lfng–/– mice (–/–; n ¼ 3) or wild-type (Lfng+/+ and Lfng+/–) mice (WT; n ¼ 3) were injected into lethally irradiated hosts, followed by analysis 10 weeks later. Mixed chimeras were generated with mixtures of equal numbers of B6 Lin– bone marrow (BM) and Lin– fetal liver cells from Lfng–/– (n ¼ 6) or wild-type (n ¼ 5) fetuses, followed by analysis 5 weeks later. (a) Abundance of fetal liver cell progeny in the thymus versus bone marrow of mixed chimeras. The percentage of bone marrow cells and thymocytes derived from the fetal liver cell versus bone marrow donor was assessed by staining with CD45.1 and CD45.2 as described8. Each bar presents the percentage of fetal liver cell progeny in the thymus divided by the percentage of fetal liver cell progeny in the bone marrow from an individual chimera. Mean ratio (± s.e.m.) for fetal liver cell chimeras: 0.85 ± 0.06 (wild-type) and 0.09 ± 0.04 (Lfng–/–). (b) Abundance of fetal liver donor-derived thymic DP and B220+ cells in individual single versus mixed chimeras. Numbers of donor-derived DP thymocytes and B220+ cells were determined as described8. Small horizontal bars, mean values. Mean values (± s.e.m.) for fetal liver cell progeny in single versus mixed chimeras are summarized in Supplementary Table 1. Similar results were obtained in three independent experiments for mixed chimeras and two independent experiments for single chimeras.
–/– Mixed chimeras FL
P < 0.001
P < 0.001
100
BM
80 60 40 20 0 101
B220+ cells (×106)
© 2006 Nature Publishing Group http://www.nature.com/natureimmunology
10–3 Subsets : ETP
Figure 1 Expression of endogenous and transgenic Lfng mRNA in thymocyte subsets. (a) Quantitative RT-PCR8 of endogenous Lfng and Cd45 transcripts in B6 thymocyte subsets (horizontal axis). ETP, early T cell progenitor. (b) Quantitative RT-PCR of endogenous and transgenic Lfng and Cd45 transcripts in thymocytes from Tg– mice (filled bar) or Tg+ mice (open bars). Left, total DN thymocytes from Rag2–/– mice; right, sorted DP thymocytes from B6 mice. The number of endogenous or transgenic Lfng cDNA templates was divided by the number of Cd45 cDNA templates to yield normalized Lfng expression values; data are mean normalized values (± s.d.) of triplicate measurements and are representative of two to three independent experiments (sorts) for each subset. ND, not detectable (o10 templates/sample).
Ratio of FL donor cells in thymus/BM
10 Normalized Lfng mRNA
a
100 10–1 10–2 10–3
Lfng FL:
WT
–/–
WT
–/–
635
ARTICLES DN3
DN4
DP
100 80 WT
60
9.4
6
13.5
77.5
43.5
57.5
40
0 100 80
+
Tg
60 40 20 0 Fc protein
hematopoietic stem cells (HSCs)42, those data suggest that Lfng may modulate Notch-ligand interactions in prethymic T cell progenitors as well as intrathymically in DN thymocytes, but not in DP thymocytes. As expected from the developmental stage specificity of the Lck proximal promoter43, there was high expression of transgenic Lfng mRNA in DN thymocytes from Tg+ recombination-activating gene 2–deficient (Rag2–/–) mice (eight- to tenfold higher than that of endogenous Lfng), and this was even higher in DP thymocytes from Tg+ C57BL/6 (B6) mice (Fig. 1b). Thus, Lfng was overexpressed (relative to endogenous Lfng) in Tg+ DN thymocytes and was ectopically expressed or ‘misexpressed’ in Tg+ DP thymocytes. To determine the requirement for Lfng in early T cell development, we evaluated thymocyte production from Lfng–/– hematopoietic progenitors. As Lfng–/– mice survive poorly after birth because of vertebral and rib cage malformations44, we obtained lineage-negative (Lin–) fetal liver cells from Lfng–/– or wild-type (Lfng+/– and Lfng+/+) B6 CD45.2 mice at embryonic days 13.5–15.5 (E13.5–E15.5). We assessed their ability to make various hematopoietic lineages by intravenous adoptive transfer into lethally irradiated B6 CD45.1 hosts. In those single chimeras, Lfng–/– and wild-type fetal liver cells generated similar numbers of B cell progeny and myeloid progeny in the bone marrow (data not shown). In addition, late fetal, neonatal
and adult Lfng–/– mice had normal numbers of all hematopoietic lineages, including DP thymocytes and peripheral T cells (data not shown). Thus, Lfng was not required for short-term self-renewal or the differentiation of HSCs and progenitors into the B lineage and myeloid lineage. Moreover, the CD4-versus-CD8 distribution of donor-derived thymocytes was similar in chimeras reconstituted with Lfng–/– or wild-type fetal liver cells, indicating that Lfng was not absolutely required for intrathymic T cell development (data not shown). However, Lfng–/– fetal liver cell progenitors generated half the number of DP thymocytes that wild-type fetal liver cells generated (average numbers, 41.6 106 versus 75.9 106; Fig. 2). That subtle but statistically significant defect in generating a normal sized pool of DP thymocytes (P o 0.01) prompted us to examine T cell production from Lfng–/– fetal liver cell progenitors in competitive repopulation experiments, as Notch1+/– progenitors are also profoundly defective in generating DP thymocytes, but only in competitive conditions8,29. We injected mixtures of equal numbers of Lin– fetal liver cell progenitors from Lfng–/– B6 CD45.2 mice and wild-type competitor Lin– bone marrow cells from B6 CD45.1;CD45.2 heterozygous mice into lethally irradiated B6 CD45.1 hosts. We also generated control chimeras from mixtures of equal numbers of Lin– wild-type fetal liver cells and wild-type bone marrow progenitors. In all cases, the host contribution was less than 2%. Fetal liver cell progeny (Lfng–/– or wild-type) always ‘out-competed’ wild-type bone marrow progeny in colonizing the bone marrow, probably because of a
Figure 4 Transgenic Lfng converts T cell P < 0.001 100 120 Tg+ precursors into ‘supercompetitors’. (a) Tg+ – 100 Tg 80 progenitors block the generation of DP 80 Host thymocytes from Tg– progenitors in mixed 60 P < 0.0001 60 thymic chimeras. Lin– bone marrow progenitors 40 40 +/+ + – (Notch1 ) from Tg or Tg donors were injected 20 20 intrathymically (alone or as a 50:50 mixture) 0 into wild-type hosts. Left, relative contributions 0 Tg+ Tg+ Tg– Tg– Tg+ + Tg– Tg– Tg+ of Tg+, Tg– and host cells in individual thymic lobes 3 weeks later. Right, numbers of DP 120 thymocytes derived from Tg– donors (open 100 100 Tg+ circles) versus Tg+ donors (filled diamonds) in P < 0.005 80 Tg+ Notch1+/– 80 individual thymic lobes. (b) Notch1 60 Tg– heterozygosity reduces the competitive fitness of 60 P < 0.0001 40 Host Tg+ progenitors in mixed thymic chimeras. Mixed 40 20 thymic chimeras were generated as described in 0 20 a with equal numbers of Lin– bone marrow progenitors from Tg– donors, Tg+ donors or Tg+ 0 + +/+ – +/+ + +/– – +/+ Tg Notch1 + Tg Notch1 Tg Notch1 + Tg Notch1 Notch1+/– donors. Left, the contribution of each donor to the total thymocyte pool in individual Donor genotype thymic lobes. Right, numbers of DP thymocytes Donor genotype derived from Tg– donors (open circles) versus Tg+ donors (filled diamonds) or Tg+ Notch1+/– donors (open diamonds) in individual thymic lobes. Similar results were obtained in two or three independent experiments for each type of chimera. The average numbers and range of DP thymocytes produced in each type of chimera are summarized in Supplementary Table 2. Donor (%)
Donor DP per lobe (×106)
a
636
VOLUME 7
NUMBER 6
JUNE 2006
+
Tg + N Tg – otch N 1 +/– ot ch 1 +/
+
+/
Tg + N Tg – otch N 1 ot ch 1
+/
+
Donor DP per lobe (×106)
b
Donor (%)
© 2006 Nature Publishing Group http://www.nature.com/natureimmunology
% of max
20
Figure 3 Increased binding of Dll1 by Tg+ thymocytes. Dll1-Fc–anti-Fc (solid lines) versus Jag1-Fc–anti-Fc (dashed lines) staining on DN3 (CD117loCD25+), DN4 (CD117loCD25–), and DP (CD4+ CD8+) thymocytes from nontransgenic mice (WT; top) or Tg+ mice (bottom). Total thymocytes were stained with anti-CD4 and anti-CD8, followed by predetermined optimal concentrations of Dll1-Fc or Jag1-Fc immune complexes. DN thymocytes were obtained by lineage depletion and were stained with anti-CD117 versus anti-CD25, followed by Dll1-Fc or Jag1-Fc immune complexes. Control Fc–anti-Fc immune complexes did not stain these thymocyte subsets (data not shown). Numbers above bracketed lines indicate the percentage of cells staining above background for Dll1-Fc. Similar results were obtained in three independent experiments.
NATURE IMMUNOLOGY
ARTICLES
a
OP9-GFP
OP9-DII1
b
OP9-DII1
26
62
87
0
7
35
8
4
13
0
4
28
62
89
0
15
OP9-DII1 0.3
30
52
17
51
32
0.5
29
© 2006 Nature Publishing Group http://www.nature.com/natureimmunology
Lfng–/– 5
3
11
CD19
0
4
48
WT
CD4
CD117
WT
CD25
Lfng –/– 26
44
CD8
Figure 5 Lfng–/– progenitors generate T cells on OP9-Dll1 stromal cells. (a) Flow cytometry of Lin–c-Kit+Sca-1+ HSCs (6 102 cells) sorted from wild-type or Lfng–/– fetal liver cells (E13.5–E15.5) that were cultured together with OP9-GFP or OP9-Dll1 stromal cells for 7 d, then collected and stained with anti-CD117 versus anti-CD19 or anti-CD25. Plots are gated on live GFP– cells with lymphoid forward and side scatter properties. (b) Flow cytometry of wild-type or Lfng–/– fetal liver HSCs cultured on OP9-Dll1 cells, collected after 14 d and stained with anti-CD4 versus anti-CD8. Numbers in quadrants indicate the percentage of cells in each. Similar results were obtained in three independent experiments.
higher frequency of HSCs and early lymphoid progenitors in E13.5– E15.5 fetal liver45. Thus, 75–95% of the bone marrow was derived from fetal liver cells in both sets of mixed chimeras. However, only 3– 11% of thymocytes were derived from Lfng–/– fetal liver donor cells, whereas 67–76% were derived from wild-type fetal liver cells in control chimeras. The normalized contribution of fetal liver HSC–derived donor cells to the thymus (relative to their bone marrow contribution), ranged from 0.78 to 0.94 for control chimeras but was only 0.04–0.15 for Lfng–/– fetal liver chimeras (Fig. 2a). That low degree of thymus chimerism reflected the generation of many fewer DP thymocytes from Lfng–/– than from wild-type fetal liver cell progenitors (Fig. 2b and Supplementary Table 1 online). Lfng–/– fetal liver donor cells also produced about tenfold more intrathymic B cells than did wild-type fetal liver cells. Thus, intrathymic T cell generation by Lfng–/– fetal liver HSCs was profoundly defective in competitive conditions. Transgenic Lfng enhances thymocyte binding of Dll1 Fringe proteins can modify Notch receptors in nonhematopoietic cells to enhance their binding to Dll ligands35,46. Therefore, we developed an assay to determine whether Lfng could alter thymocyte binding to Dll versus Jag Notch ligands. We made immune complexes of Fc fusion proteins containing the extracellular portions of Dll1 or Jag1 with fluorescence-labeled antibody to Fc (anti-Fc) and showed that C2C12 myoblasts, which respond equally well to Dll1 and Jag1 (ref. 47), bound robustly to both Dll1-Fc and Jag1-Fc (J.S.Y. and C.J.G., unpublished data). Furthermore, overexpression of Lfng enhanced binding of C2C12 myoblasts to Dll1-Fc but did not affect Jag1-Fc binding, documenting that this assay can detect Fringe-induced alterations in Notch receptors. In contrast to results obtained with C2C12 cells, we detected no binding of Jag1-Fc to wild-type DN3, DN4 and DP thymocytes, whereas those thymocyte subsets bound small amounts of Dll1-Fc (Fig. 3). Lfng greatly enhanced the binding of Dll1-Fc by Tg+ DN3, DN4 and DP thymocytes (four- to eightfold) without affecting the binding of Jag1-Fc (Fig. 3). These data demonstrate that Lfng overexpression in a ‘preferential’ and cell-autonomous way enhances binding of Dll ligands to late DN and DP thymocytes. ‘Supercompetitive’ activity of Tg+ thymocytes Both Lfng deficiency (Fig. 2) and Notch1 heterozygosity8,29 substantially reduced the ability of T cell progenitors to generate large
NATURE IMMUNOLOGY
VOLUME 7
NUMBER 6
JUNE 2006
numbers of DP thymocytes in competitive situations. However, neither genetic modification causes a global defect in Notch1 activation, as the generation of DP thymocytes is relatively normal in noncompetitive conditions. Thus, the reduced competitive fitness of Lfng–/– and Notch1+/– T cell progenitors most likely reflects ineffective competition for limited intrathymic resources. It has been suggested that the number of DP thymocytes and hence thymus size, is homeostatically regulated by competition among DN3 thymocytes for access to stromal niches28. If one or more Dll Notch ligands are limiting in these niches, then the enhanced ability of Tg+ progenitors to bind Dll ligands (Fig. 3) should provide them with a competitive advantage for generating DP thymocytes relative to Lfng trangene–negative (Tg–) progenitors. To test that hypothesis, we set up competitive repopulation experiments in which we injected equal numbers of Lin– bone marrow progenitors from Tg+ (B6 CD45.2) and Tg– (B6 CD45.1;CD45.2) mice together into the thymic lobes of sublethally irradiated wild-type B6 CD45.1 mice. We analyzed DP thymocyte production from each donor in individual thymic lobes 3 weeks later (Fig. 4a). When injected separately, both Tg+ and Tg– donor cells repopulated host thymic lobes to similar extents. However, thymic lobes injected with Tg+ donor cells were abnormally small, containing 45% fewer DP thymocytes than those injected with Tg– donor cells (Supplementary Table 2 online). When Tg+ and Tg– progenitors were injected together, Tg+ progenitors generated the same average number of DP thymocytes as in single chimeras. However, Tg– progenitors generated 80-fold more DP thymocytes (on average) in single chimeras than in mixed chimeras (Supplementary Table 2). Thus, most donor thymocytes (90–98%) were derived from Tg+ progenitors in mixed chimeras (Fig. 4a), giving rise to Tg+/Tg– thymocyte ratios of 8:1 to 25:1. These experiments demonstrated that Tg+ progenitors profoundly block Tg– progenitors from contributing to the DP thymocyte pool. The competitive advantage of Tg+ progenitors reflected modification of Notch1 by Lfng, as Tg+ Notch1+/– progenitors generated only about half the number of DP thymocytes generated by Tg+ Notch1+/+ progenitors (Fig. 4b). However, decreased production of Tg+ Notch1+/––derived DP cells was balanced by increased production of DP cells from Tg– Notch1+/+ donors in the same thymic lobes. Thus, the size of the total DP pool generated from mixtures of Tg+ Notch1+/+ and Tg– Notch1+/+ versus Tg+ Notch1+/– and Tg– Notch1+/+ donor cells was constant, but the pool contained very different proportions of DP thymocytes derived from the Tg– Notch1+/+ donor. These data indicated that Lfng modifies Notch1 to convert transgenic thymocytes into ‘supercompetitors’.
Table 1 Lfng does not profoundly affect competition for Dll1 in vitro Progenitor type
Competitive mixture
Fetal liver HSCs
Lfng–/– + WT
7
1.3
Tg+ + WT
14 9
0.8 2.1
DN3 thymocytes
Time (d)
CD45.2/CD45.1
Progeny ratios, determined by staining for CD45.1. Top row, fetal liver HSCs (5 102) from Lfng–/– mice (B6 CD45.2) and wild-type mice (B6 CD45.1) were seeded together onto OP9-Dll1 monolayers as described for Figure 5, then cells were collected (time, column 3) and stained with CD45.1 to determine the proportion of HSC progeny derived from each donor. After 7 d, 80% of the progeny had progressed to the DN2 and DN3 stages (based on CD117 and CD25 staining). Bottom row, DN3 thymocytes (1 103) from Tg+ mice (CD45.2) and wild-type mice (B6 CD45.1) were seeded together onto OP9-Dll1 monolayers; after 9 d, the frequencies of each donor were determined. At that time, 85% of the progeny were CD4–CD8+ or CD4+CD8+.
637
ARTICLES Figure 6 Lfng-Notch1 interactions regulate Notch1+/+ host access of T cell progenitors to DN3 niches. Total Notch1+/– host 3 10 bone marrow cells isolated from Tg– (B6 CD45.1) Tg+ donor 100 Host Tg– donor or Tg+ (B6 CD45.2) donors were injected Tg+ donor 80 2 intravenously (3 107 cells/mouse) into 10 Tg– donor unconditioned Notch1+/+ or Notch1+/– host B6 60 CD45.1;CD45.2 mice (four to five mice/group). 40 101 The percentage and absolute numbers of donor 20 and host cells in thymus and bone marrow were 0 determined 7 weeks after injection as described8. 100 Notch1+/+ Notch1+/– Host: Notch1+/+ Notch1+/– (a) Engraftment of Tg– or Tg+ donor cells in the Tg– Tg– Tg+ Tg+ Donor: 102 thymi of unconditioned Notch1+/+ or Notch1+/– hosts. Data represent the percentage of 101 thymocytes derived from Tg– or Tg+ donor cells 100 Notch1+/– host in individual Notch1+/+ or Notch1+/– hosts. 80 Tg+ donor (b) Absolute numbers of DP thymocytes and 60 100 – + intrathymic B cells derived from Tg or Tg donor 40 +/+ +/– cells in Notch1 or Notch1 hosts. The 20 10–1 average numbers and range of DP thymocytes 0 and thymic B cells produced by donor and host Notch1+/+ Notch1+/– Host: Notch1+/+ Notch1+/– ETP + DN2 DN3 DP Tg– Tg– Tg+ Tg+ Donor: cells in each type of chimera are summarized in Supplementary Table 3. B cells of donor origin were undetectable in all chimeras. (c) Chimerism of DN and DP thymocyte subsets in unconditioned Notch1+/– hosts injected with Tg+ donor cells. The contribution of Notch1+/– host-derived versus Tg+ donor-derived cells to the DN and DP thymocyte pools was determined for the five chimeras in a. Chimerism in DN subsets and DP thymocytes was assessed by staining of Lin– thymocytes as described8. Each bar represents a single chimera. Similar results were obtained in three independent experiments.
b
Chimerism (%)
B cells/thymus (×106)
c
Lack of Lfng-regulated competition for Dll1 in vitro Our experiments reported above demonstrated that Lfng-Notch1 interactions homeostatically regulate the size of the DP thymocyte pool and suggested that Lfng enhances T cell progenitor competition for access to limiting Dll ligands in cortical niches. That model would ‘predict’ that Lfng expression should have less profound effects on the competitive fitness of T cell progenitors when Dll ligands are abundant. We used the OP9-Dll1 stromal cell system to test that prediction, as these cells have high expression of Dll1 protein and can support DP thymocyte development from HSCs in vitro32,48. We first demonstrated that Lfng–/– fetal liver HSCs responded to Dll1 in this culture system, as B cell production was efficiently suppressed and
B cells 103
P < 0.005
b
–
Tg Notch1
+/+
+
Tg Notch1
they generated DN2, DN3 and DP thymocytes with frequencies similar to those generated by wild-type fetal liver HSCs (Fig. 5). However, the total cell yield from Lfng–/– fetal liver HSCs was 45–50% of that from wild-type fetal liver HSCs after 7 and 14 d of culture. Thus, Lfng–/– fetal liver HSCs respond less efficiently than do wild-type fetal liver HSCs to Dll1, indicating that Lfng regulates the sensitivity of T cell progenitors to Dll ligands. When equal numbers of Lfng–/– and wild-type fetal liver HSCs were cultured together on OP9-Dll1 cells, the ratio of their progeny was close to 1 (Table 1), suggesting near-normal competitive fitness in vitro. That observation is in contrast to the profoundly reduced competitive fitness of Lfng–/– progenitors in vivo (Fig. 2). When equal
+/+
+
Tg Notch1
+/–
100 Chimerism (%)
102
10–1 Notch1 Tg
+/+ –
+/– –
+/+ +
+/– +