Abstract
The factors that regulate the rate of production of T cells by the thymus remain incompletely defined. To test whether generation of functional T cell receptors limits the rate of thymic T cell export, we made use of a line of mice, LN3αβ, that have endogenously prerearranged TCR genes. The prerearranged TCR genes were expressed abnormally early in hemopoietic development, indicating that RAG-mediated recombination, rather than transcription factor expression, is the key determinant of the initiation of robust TCR transcription. Thymic T cell export rates were similar between wild-type (wt) and LN3αβ mice, indicating that T cell maturation rates in these mice are determined by factors other than TCR gene rearrangement. In competitive bone marrow chimeras, however, LN3αβ thymocytes were out-competed by wt cells and failed to develop beyond the double-negative 4 stage. Furthermore, wt progenitors transplanted intrathymically into LN3αβ mice proliferated excessively, suggesting that increased proliferative signals in the LN3αβ thymus compensate for faulty T cell development driven by early TCR expression.
The normal mouse thymus produces and exports millions of new T cells every day (1). Approximately 5% of thymocytes successfully mature; the other 95% of thymocytes die, either because they fail to generate a TCR that recognizes self-MHC/peptides sufficiently or because the TCR they generate has too high of an avidity for self-MHC-peptide complexes.
Because of the low frequency of successful rearrangements to generate TCRs that are positively selected, and the unpredictability of the selection outcome of individual thymocytes, it has been a challenge to follow many aspects of positive and negative selection of normal thymocytes, such as the precise developmental stages when these events occur, their intrathymic locations, the signaling pathways that are active, and which thymic epithelial populations are important in the process. It has also remained unclear whether successful TCR rearrangement is the only limiting step in T cell generation in the thymus or whether this is determined by other factors.
To facilitate the study of thymocyte selection, mouse models with higher rates of selection, such as those expressing TCR transgenes, or TCR-reactive superantigens have been used. In mice expressing superantigens, which are inherited open reading frames of mouse mammary tumor viruses, normal TCR gene rearrangement drives relatively high rates of positive and/or negative selection in mice expressing the appropriate MHC (2, 3). However, idiosyncrasies of superantigen binding and expression may drive selection events that do not occur under normal circumstances in these mice.
TCR-transgenic animals undergo even higher rates of positive and negative selection than superantigen-expressing mice, and normal peptide-MHC complexes drive selection. Nevertheless, one caveat with TCR-transgenic animals is that TCR transgenes tend to be expressed earlier and at higher levels than the wild-type (wt)3 genes during T cell development. Early expression of TCRα transgenes interferes with the TCRβ-preTα complex formation and signaling and, in the absence of this signal, double-negative (DN) 3 thymocytes fail to differentiate normally to the double-positive (DP) stage (4, 5, 6). To solve this problem, Hogquist and colleagues (7) used a Cre-lox strategy to precisely limit initiation of transgenic TCRα expression appropriately to the DP stage. They found that negative selection in these mice occurred during the DP-single positive transition, contradicting previous studies that used the same mice expressing the same TCR under control of a promoter that drove abnormally early expression (7). Thus, although TCR-transgenic mice have greatly facilitated the study of positive and negative selection, their tendency to miss-time expression of the TCR genes have underscored the importance of controlling the timing of TCR gene expression during the process of T cell development.
The use of cis-regulatory elements to control the timing of TCR expression has been only partially successful, because early expression of TCRα occurs even when genomic TCR constructs have been used (8). This might suggest that native elements do not properly regulate the initial timing of expression of prerearranged TCR genes. Alternatively, it was possible that distal regulatory elements controlled the timing of expression of TCR genes. Mice with functional TCRs knocked into their precise native chromosomal locations have not been available to address this question to this point.
Nuclear transfer (NT) techniques have enabled generation of animals whose nuclei originate from fully differentiated adult cells (9). This technique has been used to generate NT mice cloned from the nuclei of mature lymph node T and B cells, and the resulting cloned animals contained rearranged T cell and BCR genes in all cells (10). Such mice enable the study of regulation of prerearranged TCR and BCR genes in the native context and the study of thymic selection driven by endogenously regulated prerearranged TCR genes. Although mice generated by NT do have significant alterations in gene expression due to failure of transferred nuclei to fully reverse their epigenetic modifications (11), these abnormalities are repaired in the offspring of cloned mice, indicating that one passage through gametogenesis and early embryonic development is sufficient to restore normal epigenetic modifications and gene expression patterns (12, 13, 14).
In this study, we used NT to generate a mouse from a mature T cell nucleus. This mouse was bred multiple generations to the parental background to generate a line of mice, called LN3αβ, in which all cells have prerearranged TCRα and TCRβ genes at their appropriate genomic locus. Our data indicate that T cell development in these mice is aberrant and that this is due to the precocious expression of the prerearranged TCR genes. Thus, these data implicate RAG-mediated recombination of the TCR genes as the key event in initiating expression of the TCR genes and indicate that all of the transcription factors necessary for robust TCR expression are present in early hemopoietic progenitors. Our data also show that the defect caused by the early expression of the TCR genes is compensated for in the LN3αβ mice by increased levels of intrathymic proliferative factors, providing evidence that thymic output is dynamically regulated.
Materials and Methods
Mice
LN3αβ mice were generated using methods previously described (10). They were backcrossed to Thy1.1-congenic C57BL/6/Ka (Ly5.1, Thy1.1) as well as C57BL/6/RAG2−/− and C57BL/6/Ka/IAb−/−/β2m−/− mice, maintaining the offspring with the LN3αβ TCRα and TCRβ genes at each generation. Animals were kept at the Stanford University Medical Center and fed with standard mouse chow and acidified water ad libitum. For analysis of hemopoietic populations, mice that were 6–12 wk old were used. For bone marrow chimeras, recipients were 8–12 wk old at the time of transplantation. The Stanford Animal Use and Care Committee approved all studies and procedures involving the use of animals.
Abs and flow cytometry
Single-cell suspensions of thymus, spleen, or bone marrow were cleared of RBC using 0.85% ammonium chloride/0.1% potassium bicarbonate and were stained in PBS containing 2% heat-treated calf serum. Abs to CD3ε (KT31.1), CD4 (GK1.4), CD8 (53-6.7), CD25 (PC61), CD44 (IM7), c-Kit (2B8) Ly5.1 (AL1-4A2), and Ly5.2 (A20) TCRγδ (GL3) were purified from hybridoma supernatants and directly conjugated to fluorochromes: Alexa Fluor 405, 488, and 594 (Molecular Probes), PE, allophycocyanin, Cy7PE, and Cy5PE. Abs to TCRβ (H57) and CD24 (30-F1) were purchased from BD Pharmingen. FACS plots shown are gated on live cells by forward and side scatter and by exclusion of propidium iodide. Multicolor sorting and analysis were performed on a highly modified BD Biosciences FACS machine using argon (488 nm), dye (598 nm), and krypton (405 nm) lasers, and data were collected using DIVA digital electronics. Analysis of data was performed with FlowJo (Tree Star).
Quantitative analysis of TCR transcription via real-time RT-PCR
RNA was purified from double-sorted wt and LN3αβ populations using the RNeasy mini kit (Qiagen), and cDNA was generated using the First Strand cDNA synthesis kit (Invitrogen Life Technologies) according to the manufacturers’ instructions. PCR amplification of cDNA was performed using the 2× SYBR Green Master Mix (Applied Biosystems) on an Applied Biosystems Prism 7000 SDS. Relative expression levels were calculated with the standard curve method, according to Applied Biosystems instructions, using serial dilutions of cDNA harvested from whole LN3αβ thymus. Primers were used at a concentration of 400 nM. Primer sequences were as follows: β-actin.F, CCACAGCTGAGAGGGAAATC; β-actin.R, CTTCTCCAGGGAGGAAGAGG; Vβ1.F, GGAATGTGAGCAACATCTGG; Vβ1.R, GCACCGTCTCATTTCGAATC; Vα3-1.F, GGGCAGGTCTTCAGTTGCT, Vα3-1.R; GCAGCTGTGAGGTTCAGAGA; Dβ1.F, GAGAAGAATGGGGCCTTACC; Dβ1.R, CGTTGGCAGAAGAGGATTTC, Vβ10.F, GAGACGGCTGTTTTCCAGAC; and Vβ10.R, GGCCCA GAGTTTGCTTACAA.
Bone marrow chimeras
Recipient mice (C57BL/6, Ly5.2+/Ly5.1+) were lethally irradiated by giving two 475-rad doses 4 h apart with an x-ray irradiator. Donor bone marrow from both wt (Ly5.2) and LN3αβ (Ly5.1) mice was depleted of mature cells using Abs to CD3, CD4, CD8, B220, Mac-1, Gr-1, and Ter119, followed by magnetic depletion with anti-rat IgG-conjugated paramagnetic beads (Dynal). Equal numbers of lineage-depleted bone marrow cells from both wt and LN3αβ donors were injected into the retro-orbital venous sinus of recipient mice. Analysis of bone marrow chimeras was performed 12 wk after transplantation.
Intrathymic injections
Intrathymic injections were performed using previously described surgical techniques (15). For thymic output experiments, unirradiated, 4- to 6-wk-old mice were injected with 15 μl of 1 mg/ml FITC in PBS (pH 7.4). For testing of the donor cell expansion, unirradiated C57BL/6 (Ly5.2-congenic) or RAGLN3αβ mice (Ly5.2) were injected with 1 × 104 C57BL/6 or 5 × 103 LN3αβ DN1 cells (Ly5.1) that were defined as c-Kithigh, CD44high, CD25− and lineage−/low, and double sorted. Recipient mice were analyzed at day 10 after injection.
Intracellular staining for TCRβ
Cells were stained for surface Ags normally, then fixed for 20 min in PBS plus 4% paraformaldehyde. Cells were then permeabilized using 0.2% saponin in PBS for 20 min at room temperature, followed by staining with PE-conjugated anti-TCRβ (Η57; BD Pharmingen). Cells were then washed in PBS and analyzed.
Results
TCR gene rearrangements and T cell development in LN3αβ mice
Using our previous approach, we transferred the nucleus of a mature T cell from a (129/SvJae × C57BL/6)F1 male mouse into an enucleated oocyte to derive a cloned blastocyst from which we generated a T cell-derived ES cell line (10). The ES cells were injected into tetraploid blastocysts and transplanted into recipient females to generate cloned mice, LN3αβ, that contained prerearranged TCR genes in all cells. We used Southern blotting and inverted PCR to identify the TCRβ and TCRα variable regions that were rearranged in the LN3αβ mice (data not shown). The TCRβ locus had two rearranged alleles: one in-frame V-D-J rearrangement (Vβ1-D1-J1.4) and one incomplete D1-J2.2 rearrangement (Fig. 1 A). There was also one productive TCRα rearrangement (Vα3.1-J30) and one out of frame rearrangement (Vα9.1-J43). We bred this cloned animal, maintaining the two productive TCR gene rearrangements, onto the C57BL/6 (Thy1.1-congenic) background >10 generations to determine the effect of prerearranged TCR genes on T cell development.
All mice carrying LN3-derived TCR genes were heterozygous and thus maintained the wt unrearranged genes at one allele. Offspring of LN3αβ mice carried both prerearranged TCR genes (LN3αβ) or carried only either the in-frame prerearranged TCRα gene (LN3α) or carried only the TCRβ gene (LN3β). To prevent rearrangements of the wt alleles as well as secondary rearrangements of the LN3αβ TCR genes, we crossed LN3αβ with RAG2-deficient animals to generate LN3 × RAG−/− mice (LN3αβRAG).
LN3αβ mice contain a preponderance of CD8+ T cells in both the thymus and the spleen, suggesting that the LN3αβ TCR is MHC class I restricted (Fig. 1,B). The MHC restriction of the LN3αβ TCR was verified in LN3αβRAG mice, in which >99% of splenic T cells are CD8+ (Fig. 1,B, bottom panel), illustrating the fidelity of lineage-fate choice of LN3αβ thymocytes is accurate and indicating that the CD4+ T cells that do arise in the LN3αβ mice must undergo secondary TCR gene rearrangements. The thymic cellularity of the LN3αβ mice was ∼4-fold reduced as compared with wt (Table I). This reduction in cellularity could be attributed to the decrease in the thymic DP population, which makes up 83% of the normal thymus, but only 16% of the LN3αβ thymus (Fig. 1 B, left column).
. | B6 wt . | LN3αβ . | LN3α . |
---|---|---|---|
Thymus | |||
Total cellularity | 9.2 × 107 ± 5 × 106 (100) | 2.7 × 107 ± 6 × 106 (100) | 1.6 × 107 ± 5 × 106 (100) |
DN TCRβ+ | 4 × 105 ± 3 × 104 (0.4) | 6.9 × 106 ± 2 × 106 (25) | 2 × 106 ± 5 × 105 (13) |
CD4+CD3+ | 6.9 × 106 ± 6 × 105 (8) | 1.3 × 106 ± 6 × 105 (5) | 9 × 105 ± 2 × 105 (6) |
CD8+CD3+ | 1.5 × 106 ± 1 × 105 (2) | 6 × 106 ± 1 × 106 (22) | 5 × 105 ± 8 × 104 (3) |
γδ T cells | 1 × 105 ± 2 × 104 (0.1) | 1 × 104 ± 5 × 103 (0.04) | 2 × 105 ± 4 × 104 (1) |
CD4+CD8+CD3− | 7.1 × 107 ± 4 × 106 (77) | 3.3 × 106 ± 9 × 105 (12) | 8.4 × 106 ± 4 × 106 (52) |
DN1 | 3.1 × 104 ± 6 × 103 (0.03) | 1.2 × 104 ± 8 × 103 (0.04) | 1.2 × 104 ± 3 × 103 (0.08) |
DN2 | 1.9 × 104 ± 4 × 103 (0.02) | 5 × 103 ± 2 × 103 (0.02) | 8 × 103 ± 3 × 103 (0.05) |
DN3 | 9 × 105 ± 6 × 104 (1) | 6 × 104 ± 3 × 104 (0.2) | 1.5 × 106 ± 2 × 105 (9) |
DN4 | 3.9 × 106 ± 7 × 105 (4) | 2 × 105 ± 1 × 105 (0.7) | 4 × 105 ± 5 × 104 (3) |
Spleen | |||
CD4+CD3+ | 5.8 × 106 ± 2 × 106 (14) | 9 × 105 ± 6 × 105 (4) | 1.8 × 106 ± 5 × 105 (7) |
CD8+CD3+ | 3.6 × 106 ± 7 × 105 (9) | 1.1 × 107 ± 5 × 106 (44) | 1.9 × 106 ± 3 × 105 (7) |
DN TCRβ+ | 1.5 × 105 ± 1 × 104 (0.4) | 7 × 104 ± 4 × 104 (0.3) | 1.2 × 106 ± 6 × 104 (5) |
γδ T cells | 4.1 × 105 ± 7 × 104 (1) | 4 × 103 ± 2 × 103 (0.02) | 4 × 105 ± 2 × 104 (2) |
. | B6 wt . | LN3αβ . | LN3α . |
---|---|---|---|
Thymus | |||
Total cellularity | 9.2 × 107 ± 5 × 106 (100) | 2.7 × 107 ± 6 × 106 (100) | 1.6 × 107 ± 5 × 106 (100) |
DN TCRβ+ | 4 × 105 ± 3 × 104 (0.4) | 6.9 × 106 ± 2 × 106 (25) | 2 × 106 ± 5 × 105 (13) |
CD4+CD3+ | 6.9 × 106 ± 6 × 105 (8) | 1.3 × 106 ± 6 × 105 (5) | 9 × 105 ± 2 × 105 (6) |
CD8+CD3+ | 1.5 × 106 ± 1 × 105 (2) | 6 × 106 ± 1 × 106 (22) | 5 × 105 ± 8 × 104 (3) |
γδ T cells | 1 × 105 ± 2 × 104 (0.1) | 1 × 104 ± 5 × 103 (0.04) | 2 × 105 ± 4 × 104 (1) |
CD4+CD8+CD3− | 7.1 × 107 ± 4 × 106 (77) | 3.3 × 106 ± 9 × 105 (12) | 8.4 × 106 ± 4 × 106 (52) |
DN1 | 3.1 × 104 ± 6 × 103 (0.03) | 1.2 × 104 ± 8 × 103 (0.04) | 1.2 × 104 ± 3 × 103 (0.08) |
DN2 | 1.9 × 104 ± 4 × 103 (0.02) | 5 × 103 ± 2 × 103 (0.02) | 8 × 103 ± 3 × 103 (0.05) |
DN3 | 9 × 105 ± 6 × 104 (1) | 6 × 104 ± 3 × 104 (0.2) | 1.5 × 106 ± 2 × 105 (9) |
DN4 | 3.9 × 106 ± 7 × 105 (4) | 2 × 105 ± 1 × 105 (0.7) | 4 × 105 ± 5 × 104 (3) |
Spleen | |||
CD4+CD3+ | 5.8 × 106 ± 2 × 106 (14) | 9 × 105 ± 6 × 105 (4) | 1.8 × 106 ± 5 × 105 (7) |
CD8+CD3+ | 3.6 × 106 ± 7 × 105 (9) | 1.1 × 107 ± 5 × 106 (44) | 1.9 × 106 ± 3 × 105 (7) |
DN TCRβ+ | 1.5 × 105 ± 1 × 104 (0.4) | 7 × 104 ± 4 × 104 (0.3) | 1.2 × 106 ± 6 × 104 (5) |
γδ T cells | 4.1 × 105 ± 7 × 104 (1) | 4 × 103 ± 2 × 103 (0.02) | 4 × 105 ± 2 × 104 (2) |
Thymi and spleens from wt, LN3αβ, and LN3α mice were counted and analyzed by flow cytometry for T lineage and developmental markers, and the numbers of cells in each of the individual thymic and splenic populations are shown. Values represent the average number of cells in each population from three to five age-matched mice and are displayed with SDs. The percentage of total thymic or splenic cellularity represented by each population is shown in parentheses.
The dearth of DP cells in the LN3αβ thymus may be the result of LN3αβ thymocytes rapidly transiting through the DP stage because of their prerearranged receptor. Consistent with this possibility, there is a large population of CD4+CD8int cells in the LN3αβRAG thymus (Fig. 1,B, left column), a population that has been reported to contain positive selection intermediates for both CD4 and CD8 T cells (16). The high prevalence of this population in the LN3αβRAG mouse contrasts strongly with the LN3βRAG, which contains only the TCRβ chain and cannot generate TCRα chains and lacks the positive selection intermediate population (Fig. 1 B, left column). Thus, the prerearranged LN3αβ TCR drives positive selection of CD8 lineage cells, perhaps in part through a CD4+CD8int intermediate.
An almost complete absence of γδ T cells in the LN3αβ and LN3β mice (Fig. 1,B, second column) supports previous studies that showed pairing of a functional transgenic TCRβ with preTα drives DN thymocytes to develop down the αβ lineage and prevents development of γδ lineage cells (17). In contrast, γδ T cells are present in normal numbers in the LN3α mice, despite the fact that one TCRδ locus has been deleted in these animals due to the LN3 TCRα rearrangement, suggesting that the γδ developmental pathway is not altered by early expression of the LN3 TCRα gene or by the low cellularity of the LN3α thymus (Table I).
One striking abnormality in the LN3αβ thymus is the presence of a large population of DN cells that are CD3+ (Fig. 1,B, second column). CD3 surface expression is concomitant with TCR expression, because the TCR-CD3 complex can only reach the cell surface when both CD3 and TCR subunits are coexpressed. In a normal thymus, the DN CD3+ population of cells is a minor population and contains roughly 20% γδ T cells and 80% αβ T cells (Fig. 1,B, second column, and Table I), and these emigrate into the periphery as functional mature cells (18). In the LN3αβ thymus, however, the DN CD3+ population contains virtually no γδ T cells (Fig. 1,B). This population instead is entirely TCRβ positive, because it stains with the anti-TCRβ Ab H57 (data not shown). This large DN CD3+ population also predominates in LN3α mice, but not the LN3β mice, indicating that this population results from expression of the prerearranged TCRα gene (Fig. 1,B, second column, and Table I). The DN CD3+ population is also abundant in the spleens of LN3α mice, suggesting that this population of cells can exit the thymus and accumulate in the periphery similar to other mature T cell lineages (Fig. 1,B, third column, and Table I). In contrast, in the spleens of LN3αβ mice, the DN CD3+ population is rare (Fig. 1,B and Table I), indicating a qualitative difference in the nature of the DN CD3+ T cells in the thymi of LN3α and LN3αβ mice. This difference could be caused either from the different nature of the TCRs generated in the LN3α progenitors, because they require rearrangement of the TCRβ gene. The requirement for TCRβ rearrangement in the LN3α thymocytes may also result in a later developmental expression of a mature TCR, which could affect the development and maturation of the DN CD3+ cells.
The less mature, CD3−CD4−/lowCD8−/low (DN) populations in the LN3αβ mice are also dramatically altered, with many fewer thymocytes at the DN3 stage (Fig. 1 B, third column). This is consistent with the expression of a prerearranged TCRβ gene at the DN2–DN3 stage, which drives the immediate differentiation to the DN4 stage and beyond. Similarly, the DN3 population is also diminished in LN3β mice, but not in LN3α mice. Thus, expression of a prerearranged TCRβ gene drives differentiation rapidly through the DN3 stage.
Early TCR expression in LN3αβ mice
The prevalence of the DN CD3+ population in the LN3αβ mice was reminiscent of previous studies using TCR-transgenic mice that found early TCRα expression drove formation of a DN CD3+ population that shared similarities with γδ T cells (4, 5, 6). Therefore, we suspected that the LN3 TCRα gene might also be expressed earlier than normal, despite its proper genomic context. To determine the precise timing of surface TCR expression in LN3αβ mice, we first analyzed CD4−CD8− thymocyte populations for surface expression of CD3ε as a surrogate marker for expression of TCRαβ. More than 50% of LN3αβ thymocytes at the DN3 stage express high levels of surface CD3ε, indicating that both the LN3-derived TCRα and TCRβ genes are expressed at this early stage (Fig. 2). In addition, the DN4 population of LN3αβ mice contains >90% CD3+ cells. In contrast, only 3% of wt DN3 cells are CD3+ and 27% of DN4 cells express surface TCR. Thus, both the prerearranged TCRα and TCRβ genes are expressed at the DN3 and DN4 stages. Because TCRα expression normally begins at the DP stage, these data indicate that the prerearranged TCRα gene is prematurely expressed, despite its proper genomic localization.
Surface TCRαβ expression requires coassembly of both TCRα and TCRβ with several CD3 subunits, some of which may not be expressed until the DN3 stage. Therefore, TCRα and TCRβ genes could be expressed earlier than the DN3 stage without the TCR protein being detected at the cell surface. To determine how early in hemopoietic development the LN3-derived TCRβ was expressed, we performed intracellular staining for TCRβic on thymic progenitors. Strikingly, in the thymus, LN3αβ TCRβic was expressed at high levels at the DN1 stage, much earlier than the DN3 stage observed in wt thymocytes (Fig. 3 A).
DN1 is the earliest well-defined thymocyte stage, and the finding that these cells in LN3αβ mice all expressed TCRβic suggested that the LN3 TCRβ gene might be expressed in prethymic progenitors in the bone marrow, possibly providing a clue to which cells from the bone marrow are destined to go to the thymus. Therefore, we analyzed TCRβic expression in the earliest lymphoid-committed and prelymphoid bone marrow progenitors. The common lymphoid progenitor population (CLP) is the earliest well-defined lymphoid-committed progenitor and is most accurately defined as lineageneg, c-Kitint, Sca-1int, IL-7R+, Flk2+ (Fig. 3,B, top right panel) (Refs. 19 and 20 and H. Karsunky, unpublished observations). We found 75% of this population in the LN3αβ bone marrow-expressed TCRβic (Fig. 3,B, bottom panel). CLP develop from the c-Kit+, Sca-1+, lineage− (KLS) population, which includes long-term hemopoietic stem cells (LT-HSC), as well as short-term HSC (ST-HSC) and multipotent progenitors (MPP), which can be defined by their expression of Thy1.1 and Flk-2 (Fig. 3 B, top left panel) (21, 22). Although each of these populations (LT-HSC, ST-HSC, and MPP) can give rise to all hemopoietic lineages, subsets within the MPP population defined by the highest Flk2 expression, and/or the lack of VCAM-1 expression, or by expression of a RAG1 reporter may be primed to develop toward the lymphoid/macrophage/granulocyte lineages (22, 23, 24, 25). We found that 35% of the MPP express TCRβic, suggesting that these early cells contain all of the transcription factors necessary to transcribe the TCRβ gene and consistent with these cells being primed for T cell differentiation. In addition, 12% of ST-HSC and 5% of LT-HSC expressed detectable levels of TCRβic, suggesting that the transcription factors necessary for TCR expression are present in subsets of these populations. These data show that CLP as well as subpopulations of the KLS fraction of bone marrow of LN3αβ mice express TCRβic, likely reflecting that these populations primed, though not committed, to differentiate into the T lineage.
It was possible that the precocious expression of the LN3αβ TCRβ gene was a manifestation of normally occurring sterile transcription of a TCR Vβ1 region gene, which, in the case of the LN3αβ mice, would encode a full-length TCRβ protein. Alternatively, it was possible that early expression of the LN3αβ Vβ1 gene was the result of an abnormally active Vβ1 promoter.
To distinguish these possibilities, we compared the levels of transcripts of the Vβ1 and Vα3-1 gene segments in the LN3αβ and wt mice. Both prerearranged TCR genes in LN3αβ mice were detected far earlier in development and at much higher levels than their unrearranged wt counterparts (Fig. 3 C, note differences in axis scales between wt and LN3αβ). Consistent with the intracellular staining of TCRβ protein, transcripts of Vβ1 in LN3αβ mice were first detected at the MPP stage and then at higher levels in the CLP and DN1 populations. In contrast, there was little or no detectable transcription of the unrearranged Vβ1 locus in wt mice either in MPP, CLP, or DN1 progenitors, indicating that sterile transcripts from this locus are rare. Wild-type Vβ1 transcripts were detected beginning at the DN3 stage, but at a much lower level than was observed in DN3 cells from LN3αβ. This indicated that prerearranged TCR genes are not properly regulated, being more actively transcribed earlier in development than their unrearranged counterparts.
Transcripts of the prerearranged Vα3-1 gene were also detected much earlier than in the wt progenitors; HSC, MPP, and CLP from LN3αβ mice all show robust transcription of the Vα3-1 gene (Fig. 3,C, note difference in the axis scales between wt and LN3αβ). Transcript levels of Vα3-1 appear to decrease in the DN1 and DN2 stages and then ramp back up at the DN3 stage, suggesting complex regulation of this promoter during these early stages. In contrast to the LN3αβ Vα3-1 locus, transcripts of the wt Vα3-1 locus are practically undetectable in bone marrow progenitors, and are robustly detected beginning at the DP stage, when the Vα genes are being actively rearranged (Fig. 3 C).
Because the LN3αβ TCR transcripts encode full-length TCRs and the unrearranged counterparts encode only incomplete sterile transcripts, it was possible that the differences in transcript levels we detected were due to differential stability of the two transcripts, rather than the actual rates of transcription. To evaluate whether the prerearranged locus is more transcriptionally active, we measured transcript levels of the Vβ10 gene, which is 1.7 kb distal to the Vβ1 gene in the TCRβ locus, and unrearranged in both the wt and LN3αβ progenitors. We found that transcript levels Vβ10 are increased in DN1 and DN2 cells from LN3αβ mice, relative to their wt counterparts (Fig. 3 C, bottom right panel). Because Vβ10 genes are unrearrranged in both wt and LN3αβ progenitors, the increased Vβ10 transcript levels in LN3αβ progenitors likely reflects an increased rate of transcription. Thus, the LN3αβ TCR genes are transcribed abnormally early in hemopoietic development.
TCRβ gene rearrangement moves the Vβ regions several hundred kb closer to the 3′ Eβ enhancer, which is essential for controlling TCRβ expression and rearrangement (26, 27). Therefore, it was possible that transcription of the prerearranged Vβ1 gene was under greater influence of the now proximal Eβ enhancer, because it had moved from 490 kb away to only ∼20 kb away in the LN3αβ mice, and that the timing of activation of the Eβ enhancer drove the early expression of the LN3 TCRβ gene. To address whether timing of activation by the Eβ enhancer might cause the early TCRβ expression in LN3 mice, we measured levels of germline transcription from the pDβ1 promoter, which is located ∼20 kb from the Eβ enhancer, and requires Eβ for its activity (28). We found that while wt DN1 cells contained high levels of Dβ1 transcripts, CLP contained very low levels, and Dβ1 transcripts were absent in MPP (Fig. 3 C). Thus, because prerearranged TCRβ transcripts are present at relatively high levels in both CLP and MPP in LN3 mice, transcriptional timing driven by Eβ may partially account for the early transcription of the LN3αβ TRVβ1 gene, but other factors, including the strength of the Vβ1 promoter, as well as synergistic effects between 5′ and 3′ elements, probably also contribute.
Thymic output in LN3αβ mice
It was clear from this analysis that the prerearranged LN3αβ TCR genes had profound effects on T cell development, partly due to the nature of the positively selecting receptor and partly due to early TCR expression. We next addressed whether expression of the prerearranged TCR genes affected T cell production in the LN3αβ thymus. TCR-transgenic mice have been used previously to measure the efficiency of T cell production, and results have been mixed, possibly reflecting variability in the nature of the TCRs, abundance of selecting ligands, or possibly reflecting the variations in expression due to idiosyncrasies of transgenic insertions (29, 30, 31). The LN3αβ mice provided a unique model to measure the effect of prerearranged TCR genes on T cell production, because expression of the TCR was uniform and at normal levels in cells undergoing selection, although the early timing of expression may strongly influence T cell production.
To quantify thymic output of the LN3αβ mice, we used the method of intrathymic injection FITC, which covalently labels >95% of thymocytes and has been used to quantify the rate of mature T cell export from the thymus by counting FITC-labeled cells that appear in the periphery in the days following injection (1). The toxicity of this treatment is mostly limited to a fraction of the DP thymocytes, and because the contribution of DP cells to the single-positive T cell pool takes several days, this assay closely approximates the thymic output of the unmanipulated thymus over short time periods (32). We measured thymic output of the LN3αβ mice over a 3-day period and found that rates of T cell output from LN3αβ and wt thymi were roughly equivalent at ∼1–2 × 106 cells exported per day (Fig. 4). Expression of CD24, a marker of recent thymic emigrants, on the FITC-labeled T cells in the periphery confirmed that the FITC-labeled cells were recent emigrants to the periphery (Fig. 4 and Ref. 33). Thus, despite having prerearranged TCRα and TCRβ genes, the thymic output of LN3αβ mice is similar to that of wt.
Failure of LN3αβ thymocytes to effectively compete in mixed bone marrow chimeras
The fact that LN3αβ mice and wt mice produced similar numbers of mature T cells suggested that the rate of thymic T cell production is controlled by factors other than the efficiency of TCR gene rearrangement. Therefore, to compare the efficiency of T cell development between wt and LN3 mice, we transplanted lethally irradiated recipient mice with equal numbers of lineage-depleted bone marrow cells from both LN3αβ and wt donors and measured the contribution of each donor to various hemopoietic populations in both spleen and thymus tissues by flow cytometry 12 wk after transplantation, using congenic Ly5 markers that differed between host (Ly5.1+Ly5.2+), LN3αβ donor (Ly5.1+), and competitor (Ly5.2+). The spleens of chimeric mice contained an excess of B cells and granulocytes from the LN3αβ donor, most likely reflecting a slightly greater contribution at the level of HSC, perhaps due to a greater number of HSC in the depleted bone marrow preparations (Fig. 5 A). Despite the greater overall chimerism from the LN3αβ donor, 80% of splenic T cells were derived from the wt donor. Wild-type donor cells contributed 50% of CD8+ T cells and >90% of the CD4+ T cells in the recipient mice. The increased contribution of the CD4+ T cells from the wt donor might be expected because the LN3αβ TCR drives cells to the CD8+ lineage. However, the fact that the wt cells contributed 50% of the CD8+ cells as well suggested that the LN3αβ-derived T lineage cells had a disadvantage in the competitive setting. The diminished LN3αβ T cell contribution occurred similarly at low (5%) chimerism levels, suggesting that limiting levels of positively selecting ligands had little to do with the failure of LN3αβ to compete with wt cells (data not shown).
The relative scarcity of LN3αβ-derived T cells in the periphery of the mixed chimeric mice suggested that there might be a block in the development of the LN3αβ precursors in the thymus. Analysis of the thymi of mixed chimeras revealed a profound deficiency in the number of LN3αβ thymocytes compared with wt competitors, with the LN3αβ donor contributing only 2% of thymocytes (Fig. 5,B). The block in development of the LN3αβ thymocytes was almost complete at the DN stage, with only a small fraction of cells progressing to the CD8+ intermediate single-positive stage or beyond. This block occurred in recipients that received bone marrow at ratios of 1:1, 1:20, and 20:1 and contrasted sharply with irradiated mice that received LN3αβ bone marrow alone and that underwent thymic development similar to that observed in unmanipulated LN3αβ mice (Fig. 5,B, right panel, cf with Fig. 1 B).
We further analyzed the mixed chimeras at the specific DN (CD3−, CD4−, CD8−) stages to determine the precise stage at which LN3αβ thymocytes were blocked. The mixed chimeras contained roughly equivalent numbers of wt and LN3αβ cells at the DN1 and DN2 stages, indicating that thymic seeding, early proliferation, and differentiation from these two donors was equal (Fig. 5,C). The DN3 stage was significantly underrepresented by the LN3αβ donor cells. This is consistent with the rapid progression of LN3αβ thymocytes from the DN3 stage to DN4 and is also observed in the unmanipulated LN3αβ mice (Fig. 1,B). There also appeared to be normal numbers of LN3αβ-derived DN4 cells (Fig. 5,C). However, as mentioned above, there is a striking absence of LN3αβ-derived CD8 intermediate single-positive and DP cells as well as single-positive cells (Fig. 5 B). Thus, competition with wt thymocytes reveals a striking deficiency in the development of LN3αβ thymocytes that was not apparent in the absence of competition and manifests as a failure of LN3αβ thymocytes to progress normally beyond the DN4 stage.
Increased intrathymic growth-promoting signals in LN3αβ mice
The normal thymic output of LN3αβ mice in combination with their inability to develop efficiently in a competitive setting suggested that the LN3αβ thymus might have higher levels of compensatory growth-promoting signals than wt mice to drive the production of LN3αβ thymocytes. To test whether such a compensatory mechanism was active in the LN3αβ mice, we intrathymically injected wt thymic progenitors into thymi of unirradiated LN3αβ and wt mice and measured the proliferation and development of the donor cells after 10 days. We found that wt DN1 cells, when injected into unirradiated wt recipients, proliferated 8.5-fold over 10 days and differentiated to the DP stage (Fig. 6). In contrast, wt cells injected into LN3αβRAG thymi proliferated 140-fold, a 16-fold increase in total proliferation, while developing at a similar rate. These data support the hypothesis that LN3αβ mice compensate for defective thymocytes by having higher levels of growth-promoting signals. In contrast, LN3αβ DN1 cells failed to proliferate when intrathymically injected into wt mice, consistent with their developmental deficiency observed in the competitive bone marrow transplantation experiments and consistent with their dependence on a higher availability of growth-promoting signals to develop. Thus, LN3αβ thymocytes are developmentally deficient in a competitive setting, and this deficiency is modulated in the noncompetitive setting by increased availability of growth-promoting signals in the LN3αβ thymus.
Discussion
We initiated the study of T cell development in the LN3αβ mice with the idea that the prerearranged TCR genes in their natural chromosomal context would drive developmentally appropriate timing and levels of expression of the TCR genes, and that this would enable a more physiological study of positive selection than is observed in TCR-transgenic animals, where the timing and levels of expression of TCR genes are not regulated by their native elements. Consistent with these expectations, LN3αβ thymocytes showed skewing of several thymocyte populations that suggested rapid and efficient development. These included diminished DN3 and DP populations, both of which represent preselection intermediate populations that LN3αβ thymocytes would be expected to pass through rapidly. Similarly, the LN3αβ thymus contained a very predominant CD8+CD3+ population and when crossed to the RAG-deficient background to prevent secondary TCR gene rearrangements, LN3αβRAG mice underwent essentially perfect CD8 vs CD4 lineage determination. These properties were all suggestive of rapid and efficient positive selection of LN3αβ precursors. However, the LN3αβ thymus also contained an abnormally large number of DN TCRαβ+ cells, a population previously shown to be overabundant in TCR-transgenic animals because they expressed TCRα too early in T cell development (6). Upon further investigation, the LN3αβ mice were found to express both prerearranged TCRβ and TCRα genes abnormally early and exhibited several developmental defects as a result of this precocious expression.
The LN3-derived TCRα gene is prematurely expressed in HSC, the most primitive stage of hemopoietic development, as well as its immediate downstream progeny, while the LN3-derived TCRβ gene is expressed prematurely in a subset of MPP, possibly revealing one of the very earliest lymphoid-primed progenitors. This contrasts strongly with the normal developmental timing of robust transcription of these same TCRα and TCRβ genes in wt mice, which occurs at the DP and DN3 stages in the thymus, respectively.
What is the cause of this early TCR expression? It is formally possible that the active epigenetic status of TCRβ and TCRα loci have been maintained from the original T cell nucleus that was cloned and that this is driving the early expression of the TCR genes. Studies of epigenetically active genes in cloned mice, sheep, and pigs have shown that although cloned animals can maintain the epigenetic status of genes from the original transferred nucleus, epigenetic states became restored to normal in the offspring of cloned animals, likely through the epigenetic reprogramming that occurs during gametogenesis and early embryogenesis (12, 13, 14). This strongly suggests that, because the LN3αβ mice used in this study were at least seven generations removed from the original cloned mice, epigenetic mechanisms are unlikely to account for the early TCR gene expression.
Rather, we think that it is most likely that the genetic rearrangements of the LN3 TCR loci are responsible for the early expression of these genes. Recent evidence for the formation of a holocomplex containing the Eβ enhancer, the Dβ1 promoter, and associated factors suggest a cooperative mechanism for control of TCR transcription (34). This type of control would likely be strongly influenced by the proximity of the prerearranged TCR V region to the enhancer. In the TCRα locus of LN3αβ mice, the entire TCRδ locus that normally separates the Vα and Jα segments is deleted, eliminating any insulating effect of this large region on Vα transcription. Similarly, in the TCRβ locus, the Vβ1segment is normally ∼490 kb away from the Eβ enhancer, but in the LN3 mouse, this prerearranged gene is only 20 kb away. This relocation of the V subunits closer to the enhancer elements may account for the severalfold increase in levels of mRNA of the prerearranged LN3 TCRβ gene compared with the unrearranged sterile transcripts in hemopoietic progenitors. Consistent with this model, LN3αβ progenitors also contain increased levels of Vβ10 transcripts, which is the unrearranged segment 1.7 kb distal to the prerearranged Vβ1 gene (Fig. 3 C). Thus, our data implicate the timing of RAG-mediated recombination as the key regulatory step controlling the initiation of robust transcription of TCR genes. Our data also indicate that all transcription factors necessary for high levels of TCR transcription are present and active even at very early stages of hematopoiesis, and these may poise cells for rapid and productive TCR rearrangement and expression upon commitment to the T lineage. Furthermore, these data offer an explanation as to why mice that contain TCR transgenes with endogenous TCR promoters consistently express the TCR too early during T cell development: proper timing of expression of such transgenes is simply not feasible because robust expression from these promoters is limited only by RAG-mediated recombination.
Despite the early expression of TCR genes, unmanipulated LN3αβ mice produced and exported normal numbers of T cells from the thymus. In contrast, in mixed chimeras containing both LN3αβ and wt precursors, LN3αβ thymocytes failed to effectively compete and were largely developmentally blocked at the DN4 stage. Such a block at the DN4 stage in thymic development is unusual, because most TCR-related defects in mice either cause a failure at the preTCRα checkpoint leading to a block at the DN3 stage (e.g., RAG−/−, lck−/−, or preTα−/− mice) (35, 36, 37) or impair positive selection leading to a block at the DP stage of development (e.g., TCRα−/−) (38). It is possible that these cells may only resemble DN4 by surface phenotype and instead may represent a developmental intermediate of the DN TCRβ+ population, which is common in the LN3αβ mice. In either case, the LN3αβ thymocytes fail to develop to the DP stage or beyond when forced to compete with wt thymocytes.
The failure to efficiently advance beyond the DN4 stage and the apparent failure of the DN-DP proliferative burst both suggest that LN3αβ thymocytes fail to compete effectively for the normal thymic proliferative and differentiation signals. Correlating with the inability of LN3αβ thymocytes to efficiently respond to thymic proliferative signals, there is an increase in availability of such proliferative signals within the LN3αβ thymus, as demonstrated by the excessive proliferation of wt DN1 cells when injected into LN3αβRAG thymi. The molecular identity of the factor or factors driving this proliferation is not currently known, but available space, soluble, matrix, or cell-bound factors are all possibilities. The short time course of the intrathymic injection experiments (10 days) only allowed donor cells to develop to the DP stage, suggesting that most, if not all, of the proliferation occurred before positive selection. Therefore, it is unlikely that MHC-bound ligands drive the hyperproliferation of wt cells in the LN3αβ environment. One candidate for this homeostatic factor could be Notch ligands, because previous work has shown that Notch signaling, in addition to preTCRα signaling, is required for the DN-DP transition (39, 40). In addition, it has been shown recently that Notch ligands can be subject to competition within the thymus and that this competition may regulate the size of the DP population (41). Another possible reason for the heightened proliferative signals may be the lack of DN3 cells in the LN3αβ thymus, because previous studies have shown that DN3 cells are key regulators of thymocyte expansion (42). It will be important to know the identity of the factor or factors that contribute to the increased proliferative environment in the LN3αβ thymus. This study suggests that the factor is available in limited quantities, it is subject to competition, it requires cells to have normal timing of TCR gene expression to efficiently compete for it, and it can act strongly to drive proliferation and differentiation at the DN-DP stages. These traits do not exclude soluble factors or simply physical space, but they do suggest that non-cell autonomous growth-promoting signals in the thymus are an essential component at the DN-DP transition.
The hyperproliferative environment of LN3αβ thymus provides a mechanism that compensates for the defects of the LN3αβ thymocytes. Given the importance of a functional immunocompetence for survival, homeostatic regulators of thymocyte development might be important for consistently generating a useful T cell repertoire in individuals whose thymocytes have moderate differences in proliferative capacities. However, a large increase in such thymic proliferative signals, while beneficial in terms of generation of a T cell repertoire, may have some costs. Approximately 50% of LN3αβ mice develop thymic T cell lymphomas (T. Serwold, unpublished observations). Similarly, in TCR-transgenic mice, which are likely to have similar increases in thymic proliferative signals, a high incidence of lymphomas is also observed (43). It is possible that the increased availability of growth-promoting signals within a setting of defective T cell development may be an important component of T cell lymphomagenesis.
Acknowledgments
We thank Holger Karsunky and Emmanuelle Passegue for personal communications and helpful discussions. We thank Amy Wagers and Lauren Ehrlich for critical comments on this manuscript. We thank Christina Richter for Ab conjugations. We thank Libuse Jerabek for laboratory management and training in the animal protocols.
Disclosures
The authors have no financial conflict of interest.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by Ruth L. Kirschstein National Research Service Award 1F 32 AI 58521 (to T.S.), by National Institutes of Health Grants R01-AI047458 and R01-AI047457 (to I.W.), and by National Institutes of Health Grants R37-CA84198, R01-HD0445022, and R01-CA87869 (to R.J.).
Abbreviations used in this paper: wt, wild type; DN, double negative; DP, double positive; NT, nuclear transfer; int, intermediate; CLP, common lymphoid progenitor; LT-HSC, long-term hemopoietic stem cell; ST-HSC, short-term HSC; MPP, multipotent progenitor.