Title: Evidence implicating Gfi-1 and Pim-1 in pre-T-cell differentiation steps associated with β-selection
Abstract: Article15 September 1998free access Evidence implicating Gfi-1 and Pim-1 in pre-T-cell differentiation steps associated with β-selection Thorsten Schmidt Thorsten Schmidt Institut für Zellbiologie (Tumorforschung), IFZ, Universitätsklinikum Essen, Virchowstrasse 173, D-45122 Essen, Germany Search for more papers by this author Holger Karsunky Holger Karsunky Institut für Zellbiologie (Tumorforschung), IFZ, Universitätsklinikum Essen, Virchowstrasse 173, D-45122 Essen, Germany Search for more papers by this author Bernd Rödel Bernd Rödel Institut für Zellbiologie (Tumorforschung), IFZ, Universitätsklinikum Essen, Virchowstrasse 173, D-45122 Essen, Germany Search for more papers by this author Branko Zevnik Branko Zevnik Institut für Zellbiologie (Tumorforschung), IFZ, Universitätsklinikum Essen, Virchowstrasse 173, D-45122 Essen, Germany Search for more papers by this author Hans-Peter Elsässer Hans-Peter Elsässer Institut für Zytobiologie und Zytopathologie, Philipps Universität Marburg, Robert-Koch-Strasse 5, D-35033 Marburg, Germany Search for more papers by this author Tarik Möröy Corresponding Author Tarik Möröy Institut für Zellbiologie (Tumorforschung), IFZ, Universitätsklinikum Essen, Virchowstrasse 173, D-45122 Essen, Germany Search for more papers by this author Thorsten Schmidt Thorsten Schmidt Institut für Zellbiologie (Tumorforschung), IFZ, Universitätsklinikum Essen, Virchowstrasse 173, D-45122 Essen, Germany Search for more papers by this author Holger Karsunky Holger Karsunky Institut für Zellbiologie (Tumorforschung), IFZ, Universitätsklinikum Essen, Virchowstrasse 173, D-45122 Essen, Germany Search for more papers by this author Bernd Rödel Bernd Rödel Institut für Zellbiologie (Tumorforschung), IFZ, Universitätsklinikum Essen, Virchowstrasse 173, D-45122 Essen, Germany Search for more papers by this author Branko Zevnik Branko Zevnik Institut für Zellbiologie (Tumorforschung), IFZ, Universitätsklinikum Essen, Virchowstrasse 173, D-45122 Essen, Germany Search for more papers by this author Hans-Peter Elsässer Hans-Peter Elsässer Institut für Zytobiologie und Zytopathologie, Philipps Universität Marburg, Robert-Koch-Strasse 5, D-35033 Marburg, Germany Search for more papers by this author Tarik Möröy Corresponding Author Tarik Möröy Institut für Zellbiologie (Tumorforschung), IFZ, Universitätsklinikum Essen, Virchowstrasse 173, D-45122 Essen, Germany Search for more papers by this author Author Information Thorsten Schmidt1, Holger Karsunky1, Bernd Rödel1, Branko Zevnik1, Hans-Peter Elsässer2 and Tarik Möröy 1 1Institut für Zellbiologie (Tumorforschung), IFZ, Universitätsklinikum Essen, Virchowstrasse 173, D-45122 Essen, Germany 2Institut für Zytobiologie und Zytopathologie, Philipps Universität Marburg, Robert-Koch-Strasse 5, D-35033 Marburg, Germany *Corresponding author. E-mail: [email protected] The EMBO Journal (1998)17:5349-5359https://doi.org/10.1093/emboj/17.18.5349 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info After rearrangement of the T-cell receptor (TCR) β-locus, early CD4−/CD8− double negative (DN) thymic T-cells undergo a process termed 'β-selection' that allows the preferential expansion of cells with a functional TCR β-chain. This process leads to the formation of a rapidly cycling subset of DN cells that subsequently develop into CD4+/CD8+ double positive (DP) cells. Using transgenic mice that constitutively express the zinc finger protein Gfi-1 and the serine/threonine kinase Pim-1, we found that the levels of both proteins are important for the correct development of DP cells from DN precursors at the stage where 'β-selection' occurs. Analysis of the CD25+/CD44−,lo DN subpopulation from these animals revealed that Gfi-1 inhibits and Pim-1 promotes the development of larger β-selected cycling cells ('L subset') from smaller resting cells ('E subset') within this subpopulation. We conclude from our data that both proteins, Pim-1 and Gfi-1, participate in the regulation of β-selection-associated pre-T-cell differentiation in opposite directions and that the ratio of both proteins is important for pre-T-cells to pass the 'E' to 'L' transition correctly during β-selection. Introduction T-cells develop from poorly defined precursor cells that infiltrate the thymus from the bone marrow and undergo a series of successive phenotypic transitions that are accompanied by several selection steps, the appearance of specific surface markers and DNA rearrangements of the antigen receptor loci (for a review see Fehling and von Boehmer, 1997). The most mature TCR α/β+CD4+ and TCR α/β+CD8+ cells (single positive cells, SP) emerge by a process called 'positive selection' from a pool of CD4+/CD8+ double positive (DP) cells that constitute the major part of the thymus (reviewed in Benoist and Mathis, 1997). DP cells develop from the CD4−/CD8− double negative (DN) population that consists of bone marrow-derived precursor cells. These precursors undergo several differentiation and proliferative expansion steps regulated by members of the haematopoietin family of cytokines, particularly interleukin-7 (IL-7) (Murray et al., 1989; Peschon et al., 1994; von Freeden-Jeffrey et al., 1995, 1997; Maraskovsky et al., 1997) and the signalling through the pre-T-cell receptor (TCR) complex (Groettrup and von Boehmer, 1993; Fehling et al., 1995) to give rise to CD4+/CD8+ DP cells. According to their differential expression of the surface molecules CD25 (IL-2 receptor α chain) and CD44 (Pgp-1), immature DN cells can be subdivided into four subpopulations (Pearse et al., 1989; Godfrey and Zlotnik. 1993; Godfrey et al., 1993). The two major DN subpopulations consist of CD25+/CD44−,lo cells that give rise to CD25−/CD44−,lo cells which then quickly upregulate CD4 and CD8 and constitute the DP population. The rearrangement of the TCR β-chain genes takes place in cells of the CD25+/CD44−,lo DN population. DP cells and CD25−/CD44−,lo DN cells emerging from this pool express functional TCR β-chains and are termed 'β-selected' (Mallick et al., 1993; Dudley et al., 1994). The process of β-selection is restricted to the CD25+/CD44−,lo DN cells and appears to be governed by the assembly of a TCR β-chain, a pTα molecule and CD3 (Fehling and von Boehmer, 1997). During β-selection, CD25+/CD44− DN cells that productively rearrange their TCR β locus and are able to form a pre-TCR complex start to proliferate and enter the next developmental stage; they downregulate CD25 expression and start to upregulate CD4 and CD8. Cells from the CD25+/CD44−,lo DN population that fail to rearrange both TCR β alleles productively remain quiescent and die unless they have the potential to become γ/δ cells. A closer analysis of the CD25+/CD44−,lo DN cells revealed that this population can be divided further into two subsets: one that contains larger, cycling cells (∼15%, termed 'L' cells) with a high proportion of in-frame β-rearrangements and a second subset that represents the major pre-selected population. These cells are termed 'E' cells for 'expected size' and are smaller than the 'L' cells. They are resting cells without an enrichment of in-frame β-rearrangements and represent ∼85% of the whole CD25+/CD44−,lo DN population (Hoffmann et al., 1996). The Gfi-1 zinc finger protein is expressed at readily detectable levels almost exclusively in thymocytes, but its role in pre-T-cell development has not been investigated so far. The gfi-1 gene was first discovered as an integration site for Moloney murine leukaemia virus (MoMuLV) in virally infected cells that were selected for IL-2 independence (Gilks et al., 1993). Other studies involving MoMuLV-infected transgenic mice already carrying the trans-oncogenes pim-1 and L-myc in their germline showed that the gfi-1 gene is a frequent target for the integration of proviral DNA and is most likely involved in the accelerated progression of lymphoid malignancies in MoMuLV-infected lymphoid cells. These findings provided evidence that Gfi-1 can act synergistically at least in the process of lymphomagenesis with Myc, a helix–loop–helix zinc finger (HLH-LZ) transcription factor, and Pim-1, a cytoplasmic serine/threonine kinase (Schmidt et al., 1996; Zörnig et al., 1996; Scheijen et al., 1997). The pim-1 gene was itself first identified as a MoMuLV proviral insertion site, and studies in pim-1 transgenic mice demonstrated a low oncogenic potential for Pim-1 (Selten et al., 1985; van Lohuizen et al., 1989). Biochemical studies suggested that Gfi-1 functions as a transcriptional repressor that mediates its activity by sequence-specific DNA binding in a position-independent manner (Grimes et al., 1996a; Zweidler-McKay et al., 1996). Although the biological role of Gfi-1 remains to be clarified, several in vitro studies indicated that a constitutive expression can relieve peripheral mature T-cells from a requirement for IL-2 to overcome a G1 arrest (Grimes et al., 1996a) or, in general, could help to sustain cell proliferation of IL-2 dependent cells in the absence of the cytokine (Zörnig et al., 1996). Surprisingly, Gfi-1 expression is highest in thymocytes (Gilks et al., 1993) where potential signalling via the IL-2 receptor is restricted to the more mature cells that are programmed to leave the thymus to constitute the peripheral immune response. Experimental results that could shed some light on the role of Gfi-1 in pre-T-cell development in the thymus do not exist at present. Similarly, Pim-1 is expressed in thymic pre-T-cells and its function in pre-T-cell development is not clear either, but MoMuLV infection experiments provided strong evidence that Gfi-1 and Pim-1 cooperate efficiently in T-cell tumorigenesis (see above). This suggests that both proteins are acting in complementary signal transduction pathways that have not yet been identified. Therefore, both proteins are likely to be of considerable interest with regard to a potential role in T-cell differentiation. To be able to gain first insight into the functions of Gfi-1 and Pim-1 in T-cell development and into the nature of their synergistic effects, we analysed gain-of-function mouse mutants that overexpress either Pim-1 or Gfi-1 specifically in T-cells. Analysis of these mouse models showed that high levels of Gfi-1 can inhibit the development of DP thymic T-cells from DN precursors. More precise inspection of the DN subpopulations revealed that the β-selection-associated development of larger cycling L cells from smaller resting E cells within the CD25+/CD44−,lo subset was disturbed by Gfi-1, suggesting that Gfi-1 interferes directly with β-selection-associated processes. In contrast, we find that high levels of Pim-1 can promote pre-T-cell development through β-selection and that Pim-1 can relieve the Gfi-1-imposed block in Gfi-1/Pim-1 double transgenic mice. We propose a model in which Gfi-1 and Pim-1 can regulate the process of β-selection and in particular the E to L cell transition within the CD25+/CD44−,lo DN pre-T-cell subset in opposite directions. Our findings infer that both proteins are likely candidates for factors that participate in the regulation of β-selection at a critical point in pre-T-cell differentiation. Results Thymic cellularity is reduced in mice overexpressing an lck-driven gfi-1 transgene In order to investigate a possible function of Gfi-1 in T-cell development, we generated transgenic mice that constitutively express the gfi-1 gene at a higher than endogenous level in the thymus. The murine gfi-1 cDNA (Zörnig et al., 1996) was placed between the proximal lck promoter and genomic sequences derived from the human growth hormone (hGH) gene (Figure 1A). This construct had been used successfully in earlier experiments (Zörnig et al., 1996). The portion of the construct consisting of the lck promoter, the gfi-1 cDNA and the hGH 3′-untranslated region (Figure 1A) was microinjected into fertilized oocytes according to standard procedures (see Materials and methods) and several founder animals were obtained. Two founders were chosen to establish transgenic lines by continuous backcrossing with C57/Bl6 animals (lines lck-gfi-1 124 and lck-gfi-1 223). Compared with wild-type animals, mice of both lck-gfi-1 transgenic lines expressed the gfi-1 transgene at elevated levels in the thymus; line 223 at slightly higher protein levels than line 124 (Figure 1A). The size and cellularity of the thymus were found to be reduced drastically in animals of both lck-gfi-1 transgenic lines. Cell numbers dropped to ∼5–20% of normal control littermates (Figure 1B). Accordingly, the size of transgenic thymi was significantly smaller but thymic architecture per se was not significantly altered (Figure 1C and D). The boundaries between thymic cortex and medulla, clearly visible in the normal control, are less pronounced in some areas but still present in thymi from transgenic mice (Figure 1C and D). Forced co-expression of the cytoplasmic protein kinase Pim-1 by crossing Eμ pim-1 transgenic mice that express the Pim-1 serine/threonine kinase in both T- and B-lymphoid compartments (van Lohuizen et al., 1989) with lck-gfi-1 animals could partially restore thymic cellularity in lck-gfi-1 transgenic mice (Figure 1B). These findings suggested that overexpression of Gfi-1 impairs normal thymopoiesis and that Pim-1 can to some extent antagonize this activity of Gfi-1. Figure 1.Transgenic gfi-1 construct, transgene expression and histological analysis of the thymus of lck-gfi-1 transgenic mice. (A) Schematic representation of the construct used to generate lck-gfi-1 transgenic mice and detection of endogenous and transgenic Gfi-1-specific mRNA (left panel) and Gfi-1 protein of two lck-gfi-1 transgenic lines 124 and 223 in comparison with control thymus from non-transgenic littermates. Protein expression from the gfi-1 transgene is slightly higher in line 223 than in line 124. (B) Total number of cells per thymus for non-transgenic control animals (n = 10) and for transgenic animals from the lines lck-gfi-1 124 (n = 6) and lck-gfi-1 223 (n = 7), and from doubly transgenic mice (n = 4) resulting from cross-breeding of the lines Eμ pim-1 (van Lohuizen et al., 1989; Zörnig et al., 1996) and lck-gfi-1 124. All animals were 4–8 weeks old. (C and D) H/E staining of the histological sections demonstrates the size reduction of the transgenic thymus compared with a normal thymus. Shown are photomicrographs of stained histological sections through thymi from a normal control animal (C) and a lck-gfi-1 transgenic mouse (D) representative of several individual animals of both transgenic lines (enlargement was 320-fold). Download figure Download PowerPoint Differentiation from DN to DP cells is inhibited by high levels of Gfi-1 but can be rescued by co-expression of Pim-1 To identify the differentiation status and thereby the mechanism responsible for the observed loss in thymic cellularity in lck-gfi-1 transgenics, we analysed the frequencies of CD4 and CD8 subsets by flow cytometry. Although all CD4/CD8 thymocyte subsets are present, both transgenic lines displayed drastic alterations in relative percentages and in absolute cell numbers of the different CD4/CD8 subpopulations (Figure 2A, D and E). In particular, a significant reduction in absolute cell numbers of the SP and the DP subsets was observed (Figure 2E), whereas the absolute cell numbers of the DN population remained largely unchanged compared with non-transgenic littermates (Figure 2E). In all cases, the transgenic line with the higher level of transgene expression (lck-gfi-1 223) consistently had a more profound phenotype (Figure 2A). Figure 2.Quantification of the major CD4- and CD8-bearing subpopulations in thymus from transgenic and normal mice. (A–C) Thymocytes from 4- to 8-week-old lck-gfi-1 transgenic mice of both lines 124 and 223, Eμ pim-1 animals, Eμ bcl-2 animals, normal control mice and double transgenic mice from a cross between lck-gfi-1 lines and Eμ pim-1 or Eμ bcl-2 animals were isolated, stained with fluorescein isothiocyanate (FITC)-labelled anti-CD4 and phycoerythrin (PE)-labelled anti-CD8 antibodies and analysed by flow cytometry. Documented are dot plots representative of several animals, with the relative percentage of cells bearing CD4 or CD8 falling into the respective quadrants [lck-gfi-1 124 (n = 7), lck-gfi-1 223 (n = 10), normal control littermates (n = 9), Eμ bcl-2 (n = 5), Eμ bcl-2/lck-gfi-1 124 (n = 3), Eμ bcl-2. lck-gfi-1 223 (n = 2), Eμ pim-1 (n = 3), Eμ pim-1/lck-gfi-1 124 (n = 6)]. (D) Relative percentages of cells that fall into the four CD4/CD8 subpopulations are shown for both lck-gfi-1 lines, for double transgenic Eμ-pim/lck gfi-1 124 animals as well as for age-matched littermate controls. Shown are average values with standard deviations for several mice representing the two different transgenic lines lck-gfi-1 124 (n = 7) and lck-gfi-1 223 (n = 10), normal control littermates (n = 9) or Eμ pim-1. lck-gfi-1 124 double transgenics (n = 6). (E) Absolute numbers of cells of each CD4/CD8 subpopulation were counted from several 4- to 8-week-old mice of the lck-gfi-1 line 124 (n = 4), the lck-gfi-1 line 223 (n = 6), double Eμ pim-1/lck-gfi-1 124 mice and normal control littermates (n = 8) and are shown as average values with standard deviation. Download figure Download PowerPoint One possible reason for the apparent loss of CD4+/CD8+ DP cells could be that high level Gfi-1 expression renders these cells more susceptible to programmed cell death. However, thymocytes explanted from lck-gfi-1 mice were no more prone to cell death than their counterparts from normal littermates when left untreated in medium or when apoptosis was induced with DNA-damaging agents or steroids (data not shown). It has been reported that Gfi-1 directly represses the expression of the Bax and Bak proteins (Grimes et al., 1996b). However, no significant changes in the RNA expression levels of Bcl-2 family members, Bax, Bak, Bcl-x or Bcl-2, could be detected in thymi from lck-gfi-1 transgenic mice (data not shown), indicating that the observed loss of DP cells in lck-gfi-1 animals is not due to a higher susceptibility of these cells to undergo cell death. To rule out conclusively that apoptosis is responsible for the loss of DP cells, we crossed the lck-gfi-1 transgenic mice of both lines (124 and 223) to animals that express the Eμ bcl-2 transgene that has been shown previously to be active and expressed in B- and T-lymphoid cells and in particular to be able to rescue DP cells from apoptosis induced by a variety of stimuli (Sentman et al., 1991). However, double lck-gfi-1. Eμ bcl-2 transgenics do not show an altered phenotype compared with single lck-gfi-1 transgenics, i.e. a similar degree of DP cell loss was observed in gfi-1/bcl-2 double transgenics (Figure 2B). This inability of bcl-2 to rescue the phenotype seen in lck-gfi-1 mice clearly confirmed that the loss of cellularity and in particular the loss of DP cells seen in lck-gfi-1 mice is not due to programmed cell death. In a subsequent experiment, we analysed the double transgenic mice that resulted from a cross between lck-gfi-1 transgenics from the line 124 and Eμ pim-1 animals. In contrast to the gfi-1/bcl-2 mice, we observed that the loss of DP and SP thymocyte subpopulations was indeed restored to almost normal numbers in double pim-1/gfi-1 transgenics (Figure 2C), suggesting that Pim-1 is able to counteract the effect of Gfi-1 overexpression on thymic pre-T-cell subsets. One consequence of this combination of Pim-1 and Gfi-1 expression was the restoration of normal numbers of CD4/CD8 DP cells (see Figure 2) but also the rapid development of thymic lymphomas in double transgenic mice (Schmidt et al., 1998). When the lck-gfi-1 line 223 was used for crossings into the Pim-1 transgenics, the synergistic effect of the combination of Pim-1 and Gfi-1 was so strong that no tumour-free animals were obtained for analysis at 4–6 weeks of age (not shown). V(D)J recombination or lineage decisions do not appear to be affected by high levels of Gfi-1 Examination of the expression of TCR α/β, TCR γ/δ, CD3 and CD25 by flow cytometry showed that both types of TCRs and CD3 molecules as well as the IL-2 receptor α-chain (CD25) are expressed on thymocytes of lck-gfi-1 transgenics, including the major subtypes of TCR β- and α-chains (Vα8 subtype) (Figure 3). However, the relative numbers of TCR γ/δ-bearing cells as well as the relative numbers of CD25+ and CD3hi cells are altered in thymocytes from both lck-gfi-1 transgenics lines compared with normal control cells (Figure 3A and B), which is very probably due to the higher relative proportion of CD4−/CD8− DN cells in transgenic thymi (see above; Figure 2). Absolute numbers of TCR α/β cells were drastically reduced in the transgenic mice down to 10% of control levels (Figure 3C), as was expected considering the loss of DP cells in the lck-gfi-1 transgenics. Also, the absolute number of TCR γ/δ-bearing cells was reduced in lck-gfi-1 animals albeit only to ∼50% of the levels in non-transgenic controls (Figure 3D). The findings further confirm the notion that Gfi-1 overexpression blocks the development of DP from DN precursors, but demonstrate that this is very unlikely to be due to an altered V(D)J recombination programme, TCR assembly or CD3 expression. Our results rather suggest that high Gfi-1 levels mainly block the replenishment of the thymus with TCR α/β DP and SP cells by inhibiting differentiation of DP cells from the DN precursor pool. Figure 3.Expression of TCR, CD25 and CD3 molecules on thymocytes from lck-gfi-1 transgenics. (A and B) lck-gfi-1 transgenic mice show normal expression levels of CD25, CD3 and of both types of TCR. Thymocytes from both transgenic lines and a non-transgenic control were isolated and stained with FITC- or PE-conjugated antibodies against CD3, CD25, the TCR β-chain common epitope (α/β TCR, common β-chain epitope, H57-597) or TCR γ/δ. Shown are values representative of each lck-gfi-1 line and normal control mice. (C) Absolute numbers of TCR α/β-bearing cells are reduced in both lines of lck-gfi-1 transgenic mice down to 10% of numbers in controls (line 223). A number of Vβ and Vα subtypes of variable chains can be detected on thymocytes of both lck-gfi-1 transgenic lines. (D) The absolute numbers of TCR γ/δ cells are decreased in transgenic thymi to ∼50% of the numbers found in normal littermate controls. For the absolute numbers of TCR α/β and TCR γ/δ cells, average values are shown with standard deviation from normal control animals (n = 6), from lck-gfi-1 124 mice (n = 2) and from the lck-gfi-1 223 line (n = 4). Download figure Download PowerPoint Gfi-1 and Pim-1 overexpression can influence the frequencies and proliferation of CD4−/CD8− (DN) subpopulations in the thymus Next, we tested the effect of Gfi-1 and Pim-1 overexpression on the different subsets within the DN subpopulation. The analysis of the expression of CD25 had already shown that the relative number of cells bearing this surface marker was increased in the lck-gfi-1 transgenic mice (Figure 3A). The DN thymocyte subpopulation can be subdivided further into several cellular subsets according to the expression of CD25 and CD44 surface markers. As the absolute number of DN cells is not significantly altered in lck-gfi-1 transgenics compared with normal control mice (Figure 2C), the CD25/CD44-expressing subpopulations could be compared directly between transgenic and non-transgenic mice. Therefore, we prepared DN cells of lck-gfi-1 and control animals by depleting them of CD4 and CD8 positive cells with antibody-coupled magnetic beads. The depleted cells were first checked by flow cytometry to be >99% CD4−/CD8− and then stained for CD25 and CD44 surface markers. The results depicted in Figure 4A illustrate that expression of the gfi-1 transgene correlated with a strong increase of CD25+/CD44−,lo cells (75 and 83% for gate R1) at the expense of CD25−/CD44− and CD25−/CD44+ cells within the DN population in both transgenic mouse lines compared with values obtained from normal controls (54% for gate R1). This suggested that overexpression of Gfi-1 can provoke a developmental block in pre-T-cell differentiation at transition from CD25+/CD44−,lo cells to CD25−/CD44−,lo cells, leading to a situation where almost all cells in the DN population are of the CD25+/CD44−,lo type. Figure 4.Characterization of the CD4−/CD8− DN subpopulations in normal and transgenic mice according to their expression of CD25 and CD44 surface markers. Gate R1: CD25+/CD44−,lo cells. Thymocytes were depleted of CD4/CD8 positive cells with magnetic beads coupled with specific antibodies (Dynal), stained with FITC-labelled anti-CD25 and PE-labelled anti-CD44 antibodies and analysed for CD25 and CD44 expression on dual parameter dot plots. (A) Within the CD4−/CD8− (DN) population, CD25+/CD44−,lo cells (gate R1) are present at higher frequencies in lck-gfi-1 transgenic mice at the expense of CD25−/CD44−,lo cells compared with normal controls. This effect is more pronounced in the lck-gfi-1 223 line that has a higher transgene expression level. By contrast, Eμ pim-1 mice show a lower percentage of CD25+/CD44−,lo cells compared with normal controls. Doubly Pim/Gfi-1 animals have similar numbers of CD25+/CD44−,lo cells compared with Eμ pim-1 single transgenics. Data are representative of several lck-gfi-1 transgenic mice (line 124, n = 5, line 223, n = 8), Eμ-pim-1 (n = 6), Eμ-pim-1/lck-gfi-1 124 (n = 3) and non-transgenic controls (n = 8). (B) DN CD25+/CD44−,lo cells were stained with the DNA dye Hoechst 33382 and analysed for cells in G1/G0 phase and in S/G2/M phase according to their DNA content. Cells from the CD25+/CD44−,lo subset appeared to be arrested in G1/G0 in both gfi-1 transgenic lines because they show only 3–5% of cells in S/G2/M phase compared with controls or Eμ pim-1 mice which showed more cells in S/G2/M (27%) compared with normal cells (16%). Download figure Download PowerPoint By contrast, in Eμ pim-1 transgenics, the relative numbers for the CD25+/CD44−,lo cells dropped to 28% compared with normal controls (54% for gate R1; Figure 4A). Given that DN to DP transition is not disturbed in Eμ pim-1 mice, this could indicate that high levels of Pim-1 can promote the development of CD25+/CD44−,lo pre-T-cells towards CD25−/CD44−,lo cells. Forced co-expression of a Gfi-1 transgene in double Pim-1/Gfi-1 transgenics did not alter this low percentage of CD25+/CD44−,lo cells (29%; Figure 4A), indicating that the presence and expression of Pim-1 can overrule the effect of the Gfi-1 transgene. This is in agreement with the finding that the loss of CD4+/CD8+ DP cell numbers seen in lck-gfi-1 mice is restored in pim-1/gfi-1 double transgenic mice. However, it also suggests that high level expression of Pim-1 and of Gfi-1 disturbs pre-T-cell differentiation. We also observed that in Eμ pim-1 transgenics the CD25−/ CD44−,lo DN cell population is underrepresented in comparison with non-transgenic mice (Figure 4A). Given that in Eμ pim-1 transgenics a more rapid transition from the DN to the DP stage occurs, this could be interpreted as the result of a more rapid formation of DP cells from CD25−/ CD44−,lo DN cells leading to a smaller size of this particular DN subset. To measure a potential influence of high levels of Gfi-1 and Pim-1 on cell cycle progression of CD25+/CD44−,lo pre-T-cells undergoing β-selection, DN cells were prepared, stained with the DNA dye Hoechst 33382 as well as for CD25 and CD44 surface markers and then gated as described in Figure 4A. In both transgenic lines, Gfi-1 alone appeared to arrest CD25+/CD44−,lo cells in the G1/G0 phase of the cell cycle (Figure 4B), pointing to an inhibitory effect of Gfi-1 on cell cycle progression. Results obtained with Eμ pim-1 mice showed that high levels of Pim-1 alone promoted cell cycle progression in CD25+/CD44−,lo cells evidenced by almost 2-fold more cells inS/G2/M phase in transgenics compared with normal controls (Figure 4B). Co-expression of both Pim-1 and Gfi-1 in double transgenic animals restored a normal proliferation rate in CD25+/CD44−,lo DN cells (Figure 4B). The findings suggested that elevated levels of Gfi-1 negatively affects and Pim-1 positively affects cell cycle progression in CD25+/CD44−,lo DN cells. Gfi-1 and Pim-1 have antagonistic effects on the transition from E to L cells during β-selection During β-selection, proliferation of DN thymocyte subsets and in particular the CD25+/CD44−,lo DN cells occurs upon pre-TCR-mediated signaling in the L subset of the CD25+/CD44−,lo DN subpopulation. To obtain a more precise picture and a possible explanation for why CD25+/CD44−,lo cells in lck-gfi-1 transgenic mice proliferate at a slower rate as seen in Figure 4, we gated the CD25+/CD44−,lo DN subset on CD4/CD8-depleted cells as described in Figure 4A (gate R1) and analysed their size by analysing cell counts against forward scatter. The boundary between E and L type cells according to their size was done as described (Hoffmann et al., 1996; Figure 5A, dashed line) and by comparing them with CD25+/CD44−,lo DN pre-T-cells from Rag-2-deficient mice that only have E type cells (not shown). We found that the relative proportions of E-type smaller cells versus L-type cells within the CD25+/CD44−,lo DN subpopulation were drastically altered by high levels of Gfi-1 (Figure 5). In both lines of gfi-1 transgenic mice, the CD25+/CD44−,lo DN L-subset is almost non-existent or significantly reduced (4–5% versus 15% in normal mice; Figure 5A) and cells of the E