During thymocyte development, progression from T cell receptor (TCR)β to TCRα rearrangement is mediated by a CD3-associated pre-TCR composed of the TCRβ chain paired with pre-TCRα (pTα). A major issue is how surface expression of the pre-TCR is regulated during normal thymocyte development to control transition through this checkpoint. Here, we show that developmental expression of pTα is time- and stage-specific, and is confined in vivo to a limited subset of large cycling human pre-T cells that coexpress low density CD3. This restricted expression pattern allowed the identification of a novel subset of small CD3− thymocytes lacking surface pTα, but expressing cytoplasmic TCRβ, that represent late noncycling pre-T cells in which recombination activating gene reexpression and downregulation of T early α transcription are coincident events associated with cell cycle arrest, and immediately preceding TCRα gene expression. Importantly, thymocytes at this late pre-T cell stage are shown to be functional intermediates between large pTα+ pre-T cells and TCRα/β+ thymocytes. The results support a developmental model in which pre-TCR–expressing pre-T cells are brought into cycle, rapidly downregulate surface pre-TCR, and finally become small resting pre-T cells, before the onset of TCRα gene expression.
Early in T cell development, thymocytes that have succeeded in productive V-D-J rearrangements at the TCRβ locus are selected for cellular expansion and further maturation before the TCRα gene is expressed (1–3). This process, termed “β-selection,” is regulated by the pre-TCR, which comprises the CD3 complex in association with the TCRβ chain and the invariant pre-TCRα (pTα)1 chain (4–6). A key question has been whether surface expression is essential for the pre-TCR complex to exert its regulatory function. Recent studies support that this may be the case, since pre-TCR–induced thymocyte maturation involves both the extracellular constant region and the transmembrane region of TCRβ (7), and requires exit of the pre-TCR from the endoplasmic reticulum/cis-Golgi compartment (8). The question remains as to whether pre-TCR signaling is triggered by binding to an extracellular ligand or, alternatively, as proposed recently (9), whether pre-TCR complexes become constitutively active as soon as they reach the plasma membrane, where signaling molecules are available. In this latter situation, pre-TCR activity might be regulated by control of membrane expression. However, extremely low levels of the pre-TCR complex (∼100-fold lower than those of the TCRα/β on mature T cells) appear to reach the plasma membrane of immature thymocytes (10), a fact that has hindered the development of monospecific anti–pre-TCR reagents and, hence, the study of pre-TCR expression patterns on normal thymocytes.
Current data support the notion that one of the first consequences of pre-TCR expression is the induction of a cell cycle progression that results in the greatest expansion in cell numbers that occurs in the developing thymus (1, 11). In mice, this process is associated with differentiation of CD44loCD25+ into CD44loCD25− double negative (DN) thymocytes, suggesting that the pre-TCR is first expressed on the cell surface at this developmental transition (11). Accordingly, CD44loCD25− thymocytes from normal mice are large-sized cells expressing trace but distinguishable levels of TCRβ and CD3 (12). Similarly, the fraction of large thymocytes present in TCRα-deficient mice as well as in TCRβ transgenic recombination activating gene (RAG)-1 mutant mice expresses low but stoichiometric levels of TCRβ and CD3 (13, 14). A highly analogous checkpoint may occur in T cell development in humans during the transition from CD4+CD8− TCRα/β− precursors to CD4+CD8+ TCRα/β− double positive (DP) thymocytes, as the latter cells are mostly large cycling cells, in which TCRβ is part of a complex that is distinct from the mature TCRα/β, and could be the pre-TCR (15).
The pre-TCR–induced cell cycle transition is, in turn, associated with the downregulation of RAG-1 and RAG-2 gene transcription (16) and RAG-2 protein expression (11), which is likely to be an important component of the process of allelic exclusion at the TCRβ locus (3, 11). However, RAG genes have to be reexpressed at a later stage to allow rearrangements at the TCRα locus (16). Likewise, TCRα germline transcription and, hence, expression of sterile T early α (TEA) transcripts, is later induced as an obligatory early event in the opening of the TCRα locus for subsequent VJα rearrangement (17). This terminal program is rapidly triggered in mice during, or immediately after, the transition from CD44loCD25− cycling thymocytes to CD4+CD8+ DP resting thymocytes (16–18). Accordingly, surface expression of the mature CD3–TCRα/β complex is first detectable on small nonproliferating DP thymocytes (19, 20). However, progress towards a more precise definition of the stages involved in the transition from TCRβ to TCRα locus rearrangement has been hampered thus far, both in mice and in humans, because previous attempts to demonstrate surface expression of the pre-TCR complex throughout normal thymocyte development have been unsuccessful.
In this study, analysis performed with a polyclonal rabbit Ab that recognizes an exposed epitope of the native human pTα protein revealed a restricted pattern of surface pTα expression during normal human pre-T cell development. On the basis of surface CD3/pTα expression and cell size, we have identified a novel subset of small pre-T cells which lack surface CD3/pTα expression and are mostly in a noncycling state. Transition to this developmental stage is shown to be associated with the induction of specific developmental events that precede expression of the TCRα gene. Interestingly, such small pTα− noncycling pre-T cells are shown to be functional intermediates between large pTα-bearing pre-T cells and the first thymocytes expressing the mature α/β TCR.
Materials And Methods
Isolation of Thymocyte Subsets.
Postnatal thymocytes isolated from thymus samples removed during corrective cardiac surgery of patients aged 1 mo to 3 yr were fractionated by centrifugation on stepwise Percoll density gradients (LKB, Uppsala, Sweden), as described elsewhere (21). Thymocytes from the 1.068 and 1.08 density layers were designated as large and small thymocytes, respectively. Large thymocytes were depleted (>99% purity) of mature T cells (α/β and γ/δ), B cells, NK cells, and myeloid cells by two rounds of treatment with magnetic beads (Dynabeads; Dynal A.S., Oslo, Norway) coupled to the following mAbs: anti-TCRα/β (BMA031 [reference 22]; provided by Dr. R. Kurrle, Behringwerke AG, Marburg, Germany); anti-TCRγ/δ (TCRδ1 [reference 23]; provided by Dr. M. Brenner, Brigham and Women's Hospital, Boston, MA); anti-CD19 (Dynal A.S.), anti-CD56 (Leu-19; Becton Dickinson, San Jose, CA), and anti-CD14 (3C10, TIB 228; American Type Culture Collection, Rockville, MD). CD4+CD8+ thymocytes were magnetically sorted from the recovered pool with anti-CD8–coated beads (Dynal A.S.), and CD4+CD8−CD3− thymocytes were then sorted from the CD8-depleted pool with anti-CD4–coated beads (Dynal A.S.), as described (15). Cells in the former subset (referred to as large TCRα/β− DP thymocytes) were either CD3− or CD3low. Isolation of the CD3− cells (referred to as large CD3− DP thymocytes) was performed by depletion of CD3low cells with anti-CD3–coated magnetic beads (Dynal A.S.). To avoid cell death induced by treatment with anti-CD3, large CD3low DP thymocytes were isolated from the 1.068 fraction recovered after two rounds of fractionation on Percoll gradients, followed by depletion of T, B, NK, and myeloid cells, and anti-CD8 sorting, as described above. Such large TCRα/β− DP thymocytes consisted almost entirely (95–99%) of CD3low cells.
Small thymocytes recovered from the 1.08 density layer were depleted of TCRα/β+ thymocytes (either CD3int or CD3bright) by anti-TCRα/β magnetic bead depletion as described above for large cells. CD3− cells (termed small CD3− DP thymocytes) were then isolated from the recovered population (>99% CD4+CD8+) by anti-CD3 bead depletion (Dynal A.S.). Mature TCRα/β+ single positive (SP) thymocytes were isolated as described previously (15).
Flow Cytometry Analysis.
Directly labeled mAbs against CD3 (Leu4-PE) and CD8 (Leu2a-FITC) were obtained from Becton Dickinson; anti-CD4 (CD4-PE-Cy5) mAbs were purchased from Caltag Laboratories, Inc. (San Francisco, CA). A PE-labeled mAb against the human TCR Vβ1 family (24) was obtained from Serotec Ltd. (Kidlington, Oxford, UK). Unlabeled mAbs against monomorphic determinants of either TCRα/β (BMA031) or TCRγ/δ (TCRδ1), or against the human TCR Vα12.1 (25; provided by Dr. M. Brenner), were used in combination with goat anti–mouse FITC- or PE-coupled F(ab′)2 Ig (Caltag Laboratories, Inc.). Isotype-matched irrelevant mAbs (Caltag Laboratories, Inc.) were used as negative controls.
For detection of cytoplasmic TCRβ, cells were stained with the anti-TCRβ chain mAb βF1 (26; provided by Dr. M. Brenner), as described elsewhere (15). Surface expression of pTα chain was determined by sequential staining with a rabbit polyclonal Ab (ED-1) derived in this study (see below), plus FITC-conjugated goat anti–rabbit F(ab′)2 Ig (Southern Biotechnology Associates, Inc., Birmingham, AL). Preimmune rabbit serum was used as negative control. For peptide competition, peptide (100 μg/ml) was first incubated with the anti-pTα serum for 1 h at room temperature. Stained cells were analyzed in a flow cytometer (EPICS XL; Coulter Corp., Hialeah, FL) as described previously (15). Cell cycle analyses were performed by flow cytometry using a doublet discrimination function in cells treated with 0.05% digitonin (Sigma Chemical Co., St. Louis, MO), washed, and stained with 50 μg/ml of propidium iodide (PI; Sigma Chemical Co.), as described elsewhere (15).
Generation of Polyclonal Anti–human pTα Abs.
A synthetic peptide corresponding to the human pTα sequence 61–82 (15), with an additional Cys in the COOH-terminal region, was coupled to Maleimide-activated KLH following the manufacturer's instructions (Pierce Chemical Co., Rockford, IL). Rabbits were immunized with 1 mg of the peptide–KLH conjugate in CFA (Difco Laboratories Inc., Detroit, MI) and boosted 30 and 50 d later with 0.5 mg of immunogen in IFA (Difco Laboratories Inc.). The animals were bled 10 d after the last booster injection, and the sera were purified by affinity chromatography (Sulfolink Coupling gel; Pierce Chemical Co.) and tested for antipeptide reactivity with a horseradish peroxidase–labeled polyclonal goat anti–rabbit IgG (Nycomed Amersham plc, Little Chalfont, Bucks, UK) plus o-phenylenediamine dihydrochloride (OPD; Sigma Chemical Co.).
Cell Transfections and Immunofluorescence Assays.
C-myc tagging was performed by PCR amplification of a complete pTα cDNA contained in the Bluescript-KS II plasmid (Stratagene Inc., La Jolla, CA) with the sense 5′-GGG CCC GGA TCC ATA TGG CCG GTA CAT GGC TG-3′ and antisense 5′-GGG GGA TCC TCA CAG GTC TTC CTC GGA GAT CAG CTT CTG CTC TCC GCC GGC AGC TCC AGC CTG CAG-3′ primers, which contained the sequence coding for the 9E10 mAb– reacting c-myc peptide. The PCR product was subsequently BamHI digested and ligated into the pSRα vector (pSR-pTαmyc [reference 27]; provided by Dr. B. Alarcón, Centro de Biología Molecular “Severo Ochoa”). The pTα–green fluorescent protein (GFP) fusion was carried out by PCR amplification of a complete pTα cDNA with the sense 5′-GGG CCC GGA TCC ATA TGG CCG GTA CAT GGC TG-3′ and antisense 5′-GGG GGA TCC CCG GCA GCT CCA GCC TGC AG-3′ primers followed by digestion and ligation into the BamHI site of the pEGFP-N1 plasmid vector (Clontech, Palo Alto, CA). An EcoRI-NotI restriction fragment from the pPTα-EGFP vector was subsequently ligated into the pcDNA3 plasmid vector (Invitrogen Corp., Carlsbad, CA). The TCRα (Vα12.1) full-length cDNA (AV12S1) was the gift of Dr. J.A. López de Castro (Centro de Biología Molecular “Severo Ochoa”). AV12S1 cDNA (28) was cloned into the BamHI site of the pcDNA3 plasmid vector. COS cells were transfected by electroporation with 5 μg of pSR-pTαmyc plasmid plus 20 μg of carrier pUC-19 plasmid DNA (Promega Corp., Madison, WI), at 250V, 960 μF in a Gene Pulser (Bio-Rad Laboratories, Richmond, CA) as described elsewhere (27). Electroporated cells were then plated on coverslips. SUP-T1 cells were electroporated with 30 μg of the pTα-GFP or AV12S1-containing vectors as described above, and 24–48 h afterwards transfected cells were plated into 96-well round-bottomed plates at 105 cells/ml, grown in the presence of 1 mg/ml G418 for 2 wk, and analyzed by flow cytometry.
For immunofluorescence detection, 24–48 h after transfection, COS cells were fixed with 2% paraformaldehyde, washed in PBS, and permeabilized with 0.1% saponin in PBS containing 1% BSA as blocking agent. The coverslips were then sequentially incubated with the 9E10 anti–c-myc mAb (CRL 1729; American Type Culture Collection), the ED-1 anti-pTα rabbit antiserum, FITC-conjugated goat anti–rabbit Ig, and Texas red–conjugated goat anti–mouse F(ab′)2 IgG (Southern Biotechnology Associates, Inc.). The coverslips were visualized on a Zeiss Axioskop microscope.
Northern Blot Analysis.
Preparations of total RNA (10 μg) isolated as described previously (15) were run on 1% agarose/ formaldehyde gels, transferred to nylon membranes, and hybridized with 32P-labeled cDNA probes corresponding to the TCR Cα (PY1.4 ) or Cβ (Jurβ2 ) regions (provided by Dr. T.W. Mak, The Ontario Cancer Institute, Toronto, Ontario, Canada). The RAG-1 and RAG-2 cDNA probes (31) were the gift of Dr. L.A. Turka (The Howard Hughes Medical Institute, Ann Arbor, MI), and the human pTα cDNA probe was derived in our laboratory (15). The TEA probe was generated by PCR amplification (sense primer 5′-TGG ATG GAT AGA GAC AAG TGC-3′ and antisense primer 5′- CCT GCC CTG GGG AAT AAT AGG-3′) of the K562 erythroleukemia genomic DNA and cloning in a pMOS Blue-T vector (Nycomed Amersham plc). The fragment obtained by HindIII-EcoRI restriction was used for Northern blotting. The same blot was subsequently stripped and hybridized with a β-actin probe (15).
Reverse Transcription PCR Analysis.
Total RNA (1 μg) was reverse-transcribed into cDNA according to the manufacturer's protocol (Boehringer Mannheim, Mannheim, Germany). Equivalent amounts of cDNA among different samples was estimated by reverse transcription (RT)-PCR carried out for 18, 21, and 25 cycles with β-actin primers as described previously (15). Titration of cycle number allowed us to perform densitometric analyses (Bio-imaging BAS 1500; Fujifilm, Kanagawa, Japan) under nonsaturating conditions. Vα degenerate primers used in combination with Cα primers enabled the amplification of all known human Vα segments, as described (32). Specific amplifications were detected by Southern blotting with a Cα probe (29).
Hybrid Human/Mouse Fetal Thymic Organ Cultures.
The in vitro generation of mature TCRα/β+ human T cells was analyzed using a modification of the previously described hybrid human/ mouse fetal thymic organ culture (hu/mo FTOC ). In brief, thymi removed from 15-d-old embryos of Swiss mice were precultured for 5–6 d in the presence of 1.35 mM dGuo (Sigma Chemical Co.). The thymic lobes were then washed and cocultured in hanging drops in Terasaki plates (Nunc, Inc., Roskilde, Denmark) with either large CD3low (105 cells/lobe) or small CD3− (106 cells/lobe) human pre-T cells. After 2 d, lobes were transferred to filters (Millipore Corp., Bedford, MA), which were layered over gelfoam rafts and cultured in IMDM supplemented with 2% human AB serum and 5% FCS (GIBCO BRL, Paisley, UK). Surface staining of human cells was performed at the indicated culture periods, and flow cytometric analyses were then performed on electronically gated CD45+ human cells.
Surface CD3 Expression and Cell Size Define Distinct Subsets of Human TCRα/β− DP Thymocytes.
We have previously identified a subset of large cycling CD4+CD8+ human thymocytes in which the TCRβ chain is expressed as part of a complex distinct from the mature α/β TCR, likely the pre-TCR (15). According to their size, such TCRα/β− DP thymocytes could be selectively isolated from the fraction of large cells recovered from Percoll density gradients (approximately one third of total unfractionated thymocytes), whereas conventional TCRα/β+ DP thymocytes were more common with the small-sized cell fraction (around two thirds of total thymocytes ). Since human thymocytes typically coexpress CD3 and the α/β TCR in stoichiometric amounts, CD3 expression studies similarly defined a differential distribution of cell subsets among Percoll-fractionated thymocytes (Table 1). However, analysis of the correlated expression of CD3 versus TCRα/β revealed that, although most large TCRα/β− thymocytes were CD3−, a distinct proportion of them expressed low but detectable levels of CD3 (Fig. 1). These large CD3low TCRα/β− thymocytes did not represent γ/δ T cells, as expression of the γ/δ TCR was exclusively detected on large TCRα/β− thymocytes with a CD3bright phenotype (Fig. 1, and data not shown). Unexpectedly, CD3− and CD3low cells were also recovered from the small cell fraction (Fig. 1). Forward scatter (FSC) analyses ruled out the possibility that such cells represented large-sized contaminants (Table 1). Moreover, in contrast to large CD3low thymocytes, essentially all CD3low small cells (15– 20% of total small thymocytes) coexpressed the α/β TCR. However, small CD3− thymocytes were phenotypically similar to CD3− large cells in that they expressed neither the α/β nor the γ/δ TCR (Fig. 1, and data not shown). As both large and small CD3low thymocytes displayed a homogeneous CD4+CD8+ DP phenotype (see below), they were phenotypically indistinguishable except for the expression of α/β TCR on small DP thymocytes, but not on large DP thymocytes.
Surface Expression of pTα Chain Can Be Detected with an Anti-pTα Ab.
The above results prompted us to investigate whether large thymocytes with the CD3low TCRα/β− phenotype do represent pre-T cells expressing the pre-TCR. However, this issue was difficult to approach because no appropriate reagents such as anti-pTα Abs or Abs able to recognize the human TCRβ chain on the cell surface were available. Consequently, Abs were raised in rabbits against a synthetic peptide contained in the extracellular Ig-like domain of the human pTα molecule (15). The specificity of the affinity-purified antisera was then assayed by immunofluorescence microscopy of COS cells transfected with a pTα cDNA, tagged with a c-myc epitope that is recognized by the specific 9E10 mAb. Results in Fig. 2 A show that one of these anti-pTα antisera (ED-1) was reactive against all c-myc+ transfectants (top panels), and that both anti–c-myc and anti-pTα reagents displayed an identical intracellular recognition pattern (bottom panels), thus confirming the anti-pTα specificity of the ED-1 antiserum.
To determine whether the anti-pTα antiserum was also able to recognize the pTα chain when expressed on the cell surface, we next derived pTα stable transfectants from the human T cell line SUP-T1, which expresses TCRβ (Vβ1.1) in the absence of a functional TCRα chain and, hence, lacks surface TCRα/β heterodimers (34). As a pTα-GFP chimeric protein was used in these studies, reactivity of the anti-pTα antiserum could be analyzed by flow cytometry on stable transfectants traced by their GFP expression. As shown in Fig. 2,B, such GFP+ transfectants expressed low but detectable levels of CD3, but were unreactive with the BMA031 mAb which recognizes a common epitope of the TCRα/β dimer (22). In contrast, both TCRα/β and CD3 were detected on SUP-T1 clones stably transfected with a TCRα chain (Vα12.1), whose expression could be followed with the anti-Vα12.1 mAb 6D6 (25). Interestingly, a reciprocal expression pattern was observed when surface staining was performed with the anti-pTα Ab plus anti-CD3. Thus, pTα (GFP+) transfectants, but not TCRα (Vα12.1+) transfectants, were reactive with the anti-pTα antiserum and coexpressed CD3 in stoichiometric amounts (Fig. 2,B). It is worth noting that expression of the endogenous TCRβ could be specifically detected with an anti-Vβ1 mAb (24) on both cell types. Strikingly, levels of TCRβ expressed on pTα+ transfectants were consistently lower than those on TCRα-expressing clones, although in both cases, either pTα or TCRα was coexpressed with TCRβ in stoichiometric amounts (Fig. 2 B). As a whole, these data suggest that the ED-1 antiserum was able to specifically detect pTα-containing surface complexes which likely comprise CD3-associated TCRβ-pTα heterodimers, the hallmark of the pre-TCR complex.
Surface pTα Expression Is Restricted In Vivo to Large-sized CD3low TCRα/β− DP Thymocytes.
Having established that the anti-pTα antiserum recognized specifically pTα-containing surface complexes, we wished to examine whether pTα was actually expressed on the surface of TCRα/β− primary thymocytes. To this end, Percoll-fractionated large and small DP thymocytes depleted of CD3int and CD3bright cells (including both TCRα/β+ and TCRγ/δ+ cells) were analyzed by flow cytometry for their reactivity with the anti-pTα Ab. As expected, both isolated DP cell subsets were exclusively composed of CD3− and CD3low cells (Fig. 3,A). CD3low thymocytes made up ∼50 and 30% of the large- and small-sized DP thymocytes, respectively. Of these, only small CD3low thymocytes coexpressed the α/β TCR (see above), albeit at low levels, suggesting that they were representative of the developmental onset of TCRα/β expression. As shown in Fig. 3,A, such cells were unreactive with the anti-pTα Ab. Expression of pTα was negative as well on CD3− DP thymocytes, regardless of their cellular size. In contrast, essentially all large CD3low cells displayed a low but detectable reactivity with the anti-pTα Ab, thus providing direct evidence that pTα-containing complexes are expressed in vivo on the surface of normal pre-T cells. That this low level staining is specific was demonstrated by showing that it could be completely inhibited by the specific pTα peptide (Fig. 3,B). It is worth noting that pTα and CD3 were coexpressed on large CD3low thymocytes in a stoichiometric-like fashion similar to that observed on SUP-T1 pTα transfectants (Fig. 2 B), suggesting that the pTα-containing complex expressed on the former cells did correspond to the CD3-associated pre-TCR.
Developmental Status of Subsets of TCRα/β− DP Thymocytes.
The above results allowed a novel subdivision of the TCRα/β− DP compartment into three individual subsets of thymocytes defined as large DP CD3−, large DP CD3low, and small DP CD3−. Because pTα expression was restricted to large DP CD3low thymocytes, we wanted to investigate further the developmental status of the distinct pTα+ and pTα− populations in order to improve definition of their precursor–product relationships. To this end, the three cell subsets were independently isolated and examined for their respective patterns of TCRβ, TCRα, and pTα gene expression. Northern blot analysis shown in Fig. 4,A revealed that both the 1.3-kb mature and the 1.0-kb immature TCRβ transcripts were expressed in the three subsets of TCRα/β− DP thymocytes, regardless of their cellular size and CD3 phenotype. In contrast, TCRα transcription was undetectable in all of them, but occurred at high levels in mature SP thymocytes included as control. As expected, CD4+CD8−CD3− cells, which represent upstream precursors of the TCRα/β− DP thymocyte pool as a whole (15), lacked both TCRβ and TCRα mature transcripts, but expressed 1.0-kb TCRβ mRNA. Therefore, we concluded that the three TCRα/β− DP subsets identified in this study include cells that have already completed TCRβ, but not TCRα, gene rearrangement and transcription, indicating that they represent discrete pre-T cell stages along the pathway of T cell development. As expected of pre-T cells, all three cell types expressed pTα mRNA, with higher levels in the large CD3− subset. Maximal pTα expression was found in the more immature CD4+CD8−CD3− thymocytes (Fig. 4,A). A more sensitive RT-PCR analysis of these very same populations confirmed the patterns of TCRβ and pTα expression obtained by Northern blotting (not shown). However, it revealed that TCRα transcription had occurred, albeit at low levels, in small CD3− thymocytes, whereas it was completely absent from both the CD3− and the CD3low subsets of large pre-T cells (Fig. 4 B). Based on these results, we concluded that small DP CD3− thymocytes represented the particular stage at which TCRα gene rearrangement and transcription are initiated during human T cell development; therefore, this subset was placed at the latest pre-T cell stage, downstream of both subsets of large pre-T cells.
Additional support for the proposed model came from studies aimed at investigating the TEA and RAG gene transcription patterns displayed by the three pre-T cell subsets. TEA is a TCRα germline transcript whose expression seems to be an obligatory early event in the opening of the TCRα locus for subsequent rearrangement (17). As Vα to Jα recombination necessarily involves deletion of the TEA region (17), we reasoned that the onset of TCRα gene expression should necessarily be accompanied by a reciprocal shutdown of TEA transcription. Northern blot analysis performed with a specific TEA probe generated in this study revealed that TEA transcription had not been induced in early CD4+CD8−CD3− thymocytes, but occurred at high levels in large CD3− DP thymocytes, and was maximal in pTα+ pre-T cells (Fig. 4,A). However, it was sharply downregulated in small CD3− thymocytes, thus supporting the notion that the onset of TCRα gene expression concurs with a decrease in TEA transcription, these being coincident events in the transition from large to small pre-T cells. A reciprocal pattern of RAG gene expression was observed at this developmental point after hybridization with RAG-1– and RAG-2–specific probes (Fig. 4 A). Thus, although RAG-1 and RAG-2 transcripts were detected at similarly high levels in CD4+CD8−CD3− and large DP CD3− thymocytes, their expression was five- to eightfold lower in pTα+ pre-T cells. Interestingly, maximal transcription levels of both genes corresponded to small CD3− pre-T cells. These results suggest that RAG gene expression is selectively turned down in pre-TCR– expressing pre-T cells, and is later regained in small pre-T cells, allowing rearrangements at the TCRα locus to occur.
TCRβ Chain Expression and Cell Cycle Analysis of Subsets of Pre-T Cells.
Formation and expression of the pre-TCR is claimed to immediately promote a cell cycle transition, which results in expansion and selection (β-selection) of the pool of pre-T cells deemed useful by virtue of successful TCRβ chain expression (1, 11). Therefore, the prediction would be that all cells downstream of the pre-TCR–expressing pre-T cell stage should show evidence of β-selection. To address this issue, pre-T cell subsets were independently analyzed by flow cytometry for their DNA content as well as expression of cytoplasmic TCRβ protein. Results shown in Fig. 5 A revealed that essentially all (>90%) large pTα+ as well as small CD3− pre-T cells expressed cytoplasmic TCRβ; therefore, both cell subsets comprise β-selected pre-T cells. Unexpectedly, however, only 50–80% of large CD3− DP thymocytes (50% in this particular experiment) expressed cytoplasmic TCRβ, whereas the remaining 30– 50% were TCRβ−. Such a differential expression of cytoplasmic TCRβ defined two distinct cell subsets of large CD3− pre-T cells which could thus be placed on either side of the β-selection process.
Formal support for this notion came from additional flow cytometric studies that addressed directly the cell cycle status of either the TCRβ+ or the TCRβ− subsets of large CD3− pre-T cells. As shown in Fig. 5,B, double staining with anti-TCRβ and PI demonstrated that essentially all (>90%) large CD3− pre-T cells lacking TCRβ were arrested in the G0/G1 phase of the cell cycle, whereas, as expected of β-selected thymocytes, TCRβ+ CD3− pre-T cells featured a high proportion (up to 55%) of cells in S/ G2/M. This is consistent with 30% of bulk CD3− pre-T cells being in S/G2/M (Fig. 5,A). Therefore, TCRβ− large pre-T cells are strong candidates for cells immediately before β-selection, and most likely immediately downstream of the CD4+CD8−CD3− precursor stage, which was essentially composed of TCRβ− thymocytes (>95%) displaying only a background level (<10%) of cells in S/G2/M (Fig. 5,A). As expected of β-selected thymocytes, large pTα+ pre-T cells were highly enriched in cycling cells (∼55% in S/G2/M). However, TCRβ expression could not be associated with an active cycling state in small CD3− pre-T cells. Rather, these cells typically displayed only background levels of cells in S/G2/M (<15%), with a substantial fraction of them (>50%) in the G2/M phase (Fig. 5 A). As an additional indicator of their resting state, small CD3− pre-T cells were shown to display exclusively the fast hypophosphorylated form of retinoblastoma (not shown). We thus concluded that most, if not all, small CD3− DP thymocytes are noncycling pre-T cells that have already passed through β-selection. This, in turn, suggests that β-selected large pre-T cells may normally lose surface pre-TCR expression and return to slow cycle conditions before the onset of TCRα gene expression. As a whole, these data provide strong evidence that small resting pre-T cells represent the latest pre-T cell stage in human thymocyte development, immediately upstream of conventional DP TCRα/β+ resting thymocytes.
Small CD3− Pre-T Cells Are Functional Intermediates between Large CD3low Pre-T Cells and TCRα/β+ DP Thymocytes.
To seek direct evidence that small CD3− DP thymocytes represent the normal progeny of large pre-TCR–expressing pre-T cells in the pathway of T cell differentiation, highly purified large CD3low pre-T cells (>98% pure) were analyzed for their developmental fate in a hybrid hu/mo FTOC system. The pattern of differentiation from several experiments was identical (Fig. 6,A): the rapid appearance of CD3− DP cells (up to 60% by day 5 in this experiment) with minimal differentiation into TCRα/β+ cells (>5%), followed by the generation of a major population of conventional DP thymocytes that coexpressed CD3 and the α/β TCR at low to intermediate levels (85% by day 17), and the later appearance of small numbers of mature SP thymocytes (not shown). FSC analysis of the cells harvested on day 5 in the experiment shown in Fig. 6 A revealed that, by this stage, the cells that remained CD3low had kept their original size, whereas the CD3− cells generated in the lobes were significantly smaller (mean FSC: 450 vs. 410, respectively). However, by day 17, essentially all large cells had reverted to small cells, and thus, all TCRα/β+ progeny generated by this time (85%) were similar in size to the remaining (15%) CD3− DP cells (mean FSC: 330, 335, and 329, for CD3−, TCRα/βlow, and TCRα/βint cells, respectively). Interestingly, total yields of viable human cells increased progressively during the initial phase of culture, resulting in a 15–20-fold increase of absolute cell numbers by days 5–7, but cellular recoveries then stabilized or increased modestly (up to 2–3 times) through the next 10–12 d, and declined steadily thereafter.
The above data indicate that cell division in thymus lobes reconstituted with large CD3low pre-T cells is extensive and skewed to the early stages of culture. Therefore, the high yields of TCRα/β+ DP progeny in FTOC are mostly a reflection of cellular expansion of blast precursors, presumably before transition to small CD3− pre-T cells. This in turn suggests that differentiation into TCRα/β+ DP cells can occur in the absence of cell division from small noncycling CD3− pre-T cells. To provide direct evidence of precursor activity, we tested the capacity of highly purified (>98%) populations of small CD3− DP thymocytes to produce TCRα/β+ progeny in the FTOC system. As shown in Fig. 6 B, a high proportion of both TCRα/βlow (20%) and TCRα/βint (50%) progeny was already seen in the thymic lobes at day 1, the earliest sampling time. However, the number of TCRα/β+ progeny did not increase in absolute terms thereafter, an expected finding considering that all cells recovered by day 1 were small-sized cells (mean FSC: 300, 293, and 290, for CD3−, TCRα/βlow, and TCRα/βint cells, respectively). Thus, although kinetics of TCRα/β+ cell generation were more pronounced with small CD3− than with large CD3low pre-T cells, total cell yields were substantially lower with the small CD3− pre-T cell fraction. Based on the above results, we concluded that small CD3− DP thymocytes represent functional intermediates between large pre-T cells and TCRα/β+ DP thymocytes.
Considerable progress has recently been made in defining the role that preantigen receptors, namely the pre-B cell receptor and the pre-TCR, play in lymphocyte development. It is now established that both receptors direct in an analogous way the survival, expansion, and clonality of pre-B and pre-T lymphocytes by triggering cell cycle activation and the simultaneous downregulation of RAG genes (10, 11). However, less is known about the mechanisms that control terminal differentiation of lymphocyte precursors thus selected, especially in the T cell lineage. This can be partly attributed in both mice and humans to the lack of experimental data concerning regulation of pre-TCR expression on the surface of primary thymocytes, a fact that has hampered the definition of the developmental stages involved in the transition from TCRβ to TCRα rearrangement. In this study, analysis performed with a polyclonal rabbit Ab that recognizes an exposed epitope of the native human pTα protein has provided evidence for a restricted pattern of surface pTα expression during normal T cell development in humans. Surface pTα versus CD3 expression, together with cell cycle analyses, enabled a novel subdivision of the whole compartment of TCRβ-expressing pre-T cells into three distinct subsets of increasing maturity, and allowed the identification of a late stage of small noncycling pre-T cells representing the immediate precursors of TCRα/β-bearing thymocytes. The definition of the precursor–product relationships between such pre-T cell subsets, together with the characterization of the stage-specific events associated with the developmental onset of TCRα gene expression, namely exit from cell cycle, reexpression of RAG genes, and downregulation of TCRα germline transcription, collectively support the developmental scheme depicted in Fig. 7.
The distinct pre-T cell stages defined in our model are all included within a subset of CD4+CD8+ DP thymocytes that lack the mature α/β TCR and represent, as a whole, the downstream progeny of CD4+CD8−CD3− thymocyte precursors (15). About one third of such DP TCRα/β− thymocytes are larger in size than the remaining two thirds and, thus, the two cell types have been defined, respectively, as large and small pre-T cells. Although pTα transcription is common to all pre-T cell stages, surface expression of the pTα protein is shown to be restricted to a limited fraction (50%) of large-sized pre-T cells that coexpress small but stoichiometric amounts of CD3. As neither pTα nor CD3 is detectable on the rest of the large and small pre-T cells, the coexpression of both molecules seems to define the particular subset of primary pre-T cells in which the pTα chain is paired with TCRβ and associates with CD3 to form the pre-TCR. However, attempts to demonstrate coexpression of surface TCRβ in vivo were unsuccessful, essentially because neither anti-TCRβ mAbs useful for flow cytometry nor anti-pTα reagents suitable for biochemical studies are yet available. Despite this, the possibility that TCRβ-pTα heterodimers associated with CD3 are indeed expressed on pTα+ pre-T cells is strongly supported by several independent findings: (a) low but stoichiometric amounts of pTα and CD3 were specifically coexpressed with endogenous TCRβ on pTα transfectants derived from a TCRα-deficient cell line; (b) we have previously shown that heterodimeric complexes containing TCRβ without TCRα could be immunoprecipitated from unfractionated large DP TCRα/β− thymocytes (15); (c) others have noticed that large thymocytes from TCRα- deficient and TCRβ transgenic RAG-1 mutant mice express low but stoichiometric amounts of surface TCRβ and CD3 (13), similar to what has been reported for mouse thymocytes from which a CD3-associated pTα-TCRβ heterodimeric complex has recently been characterized (14); and (d) to date, no surface pTα expression has been described without association with TCRβ and CD3. Therefore, expression of surface pre-TCR complexes is proposed in our model to be confined to the minor subset of large pTα+ CD3low pre-T cells, whereas large pre-T cells lacking detectable amounts of surface pTα and CD3 (∼50% of all large DP TCRα/β− thymocytes) are proposed to be homogeneously negative for pre-TCR expression (Fig. 7). However, the latter cells represent a heterogeneous population in which a major fraction (50–80%) have already passed β-selection, as indicated by their high expression levels of intracellular TCRβ (11), whereas the remaining cells (20–50%) still lack cytoplasmic TCRβ and may thus represent intermediates between CD4+CD8−CD3− thymocytes and the first β-selected pre-T cells. It is highly likely that such intermediates include the pool of precursor thymocytes undergoing rearrangements at the TCRβ locus, although they may also include cells carrying nonproductive Vβ-Dβ-Jβ joints on both TCRβ loci, which may thus be destined to die. Both possibilities, illustrated in Fig. 7, are compatible with the hypothesis that the human pre-TCR does not participate, as does its murine counterpart, in the transition to the DP stage (2, 3, 18). Rather, expression of CD8 appears to precede pre-TCR expression during human T cell development.
An important aspect of our study was the observation that virtually all large pre-T cells with cytoplasmic TCRβ, whether or not they display surface pTα chain expression, were actively engaged in cell cycle, a characteristic previously associated with the process of β-selection (11). Conversely, DP thymocytes lacking TCRβ protein were nondividing cells arrested at G0/G1. The finding that up to 40% of cycling, β-selected pre-T cells did not express the putative pre-TCR is apparently difficult to reconcile with the current idea that cell cycle activation involves signaling mediated through the pre-TCR (2, 3, 11). However, the possibility that undetectable, but functional, amounts of the pre-TCR are expressed on the surface of such cycling pTα− pre-T cells cannot be formally excluded. Alternatively, it is likely that, as proposed for pre-B cells at the equivalent developmental point (35), β-selected pre-T cells rapidly downregulate expression of the pre-TCR from the cell surface while they are still in cycle. In this latter situation, it could be expected that the maintained expression of surface pre-TCR in a short developmental window is both necessary and sufficient to provide a sustained proliferation signal that would allow pre-T cells to undergo a great cellular expansion before turning back to slow cycle conditions. Supporting this hypothesis, results from a recent study have provided evidence that such a proliferation phase corresponds in mice to nine rapid cell divisions that last for ∼4 d and end at the small resting DP thymocyte stage (36). This concurs with our finding that a major fraction (about two thirds) of β-selected pre-TCR− pre-T cells in humans are small-sized, nondividing cells. Interestingly, although such small pre-T cells do not yet express the mature α/β TCR, they already transcribe low levels of the TCRα gene. Thus, they are proposed to define the developmental point at which onset of TCRα gene rearrangement and transcription occurs, and are placed in our model at the latest pre-T cell stage, immediately upstream of the first TCRα/β-expressing DP thymocytes (Fig. 7).
Consistent with the above proposal, we found that indicators of Vα-Jα recombinase activity, such as RAG gene reexpression and downregulation of TEA transcription, are coincident and stage-specific events induced after entry of late pre-T cells into the pool of small, resting cells. Thus, it was observed that expression of RAG genes, which is turned down after pre-TCR signaling (11, 18), is regained in small pre-T cells, allowing rearrangements at the TCRα locus to be initiated at this stage. Further, the demonstration that germline transcription of TCRα spans all cycling pre-T cell stages but drops significantly in resting pre-T cells also supports the concept that these cells are actively rearranging their Vα genes. Similarly, Ig L chain gene rearrangement is restricted to small, resting pre-B cells that represent the equivalent precursor stage along the B cell pathway (35). In contrast to the proposal that small resting TCR− DP thymocytes are functional intermediates in the T cell differentiation pathway, it is currently assumed that these cells represent the large pool of end-stage products of failed rearrangement attempts. However, recently published data have shown that small noncycling TCR− DP thymocytes in the mouse are actually the physiological targets of the multiple rearrangements that occur at the TCRα locus (20), and are subject to positive selection (21, 37). Direct evidence of the physiological relevance of small resting CD3− pre-T cells in humans was further provided by the demonstration that these cells are functional intermediates between large pTα+ pre-T cells and TCRα/β+ DP thymocytes. Accordingly, as shown previously in mice (19, 20), surface expression of the mature CD3–TCRα/β complex can be first detectable on small nonproliferating DP thymocytes.
As a whole, our results suggest that, after pre-TCR–mediated cellular expansion, β-selected large pre-T cells may normally downregulate surface pre-TCR expression and return to slow cycle conditions before the onset of TCRα gene expression. The proposed pattern of pre-TCR expression differs from previous hypothetical models in mice postulating that mature TCRα/β and pre-TCR complexes are coexpressed on the cell surface of late pre-T cells (10). However, it is still possible that some of these cells are cotranscribing pTα and TCRα genes. It is tempting to speculate that, in that situation, both molecules compete with each other for dimerization with TCRβ, the affinity of TCRα being higher than that of pTα. Alternatively, as proposed in mice, another still unknown component of the pre-TCR (i.e., the hypothetical VpreT) might be already shut off at the earliest TCRα+ stages, hence preventing surface expression of the whole pre-TCR complex (3). Finally, it must be stressed that the restricted pattern of surface pTα expression shown in this study closely resembles that of the surrogate light chain of the pre-B cell receptor (35, 38). This emphasizes the similarities of early developmental events associated with the transient expression of both preantigen receptors during T and B cell development.
We wish to thank Drs. B. Alarcón, M.A. Alonso, M. Brenner, R. Kurrle, J.A. López de Castro, T.W. Mak, and L.A. Turka for the generous gift of Abs and cDNAs, Dr. S.G. Copín for invaluable advice on the FTOC, and the Pediatric Cardiosurgery Units from the Centro Especial Ramón y Cajal and Ciudad Sanitaria La Paz (Madrid) for the thymus samples.
This work was supported in part by Glaxo Wellcome S.A., and by grants SAF95-0006 and SAF97-0161 from Comisión Interministerial de Ciencia y Tecnología (CICYT) and CAM083/013/97 from Comunidad de Madrid. We would also like to thank the Fundación Ramón Areces for an Institutional Grant to the Centro de Biología Molecular “Severo Ochoa.” C. Trigueros and A.R. Ramiro are fellows of Fondo de Investigaciones Sanitarias and the Ministerio de Educación y Ciencia, respectively.
Abbreviations used in this paper
C. Trigueros and A.R. Ramiro contributed equally to this work.
Address correspondence to María L. Toribio, Centro de Biología Molecular “Severo Ochoa,” CSIC-UAM, Facultad de Biología, Universidad Autónoma de Madrid, Cantoblanco 28049, Madrid, Spain. Phone: 34-1-3978076; Fax: 34-1-3978087; E-mail: firstname.lastname@example.org