The SCL relative LYL-1 is required for fetal and adult hematopoietic stem cell function and B-cell differentiation

Hematopoiesis is maintained by a pool of hematopoietic stem cells (HSCs) defined by their capacity to self-renew and, hence, to maintain the pool and to differentiate into all mature progenies. These properties underlie their ability to reconstitute the hematopoietic system.¹  HSCs are identified by molecular markers²  and are divided into long-term HSCs (LT-HSCs), short-term HSCs (ST-HSCs), and multilineage progenitors (MPPs).^3,4  As they differentiate, HSCs give rise to committed progenitors, such as common lymphoid progenitors (CLPs) or common myeloid progenitors (CMPs), which generate mature cells within the different hematopoietic lineages.⁵ 

The action of HSCs to self-renew or to differentiate is a tightly controlled process linked to the induction or repression of some crucial genes.⁶  Gene-targeting experiments have provided support to this hypothesis by identifying transcription factors of various families or their coregulators that are implicated in HSC biology.^7,8 

The basic helix-loop-helix (bHLH) family of transcription factors regulates a wide range of developmental events.^1,9  Some bHLHs, such as MyoD and NeuroD, display restricted tissue expression and control the differentiation of particular cellular lineages.¹⁰  They form transactivating dimers by heteromerization with ubiquitous bHLH proteins, termed E proteins, such as E2A, E12, and E47 genes.^11,12  Conversely, the latter proteins are repressed by other HLH proteins, which often precludes differentiation.^10,13,14 

The tal/scl gene (hereafter referred to as scl) encodes a bHLH protein.^15,16  It was identified through its implication in T-cell acute lymphocytic leukemia (T-ALL). Knock-out (KO) experiments^17,18  show that scl is autonomously required for primitive and definitive hematopoiesis.^19,20  Moreover, its enforced expression enhances megakaryocytopoiesis and erythropoiesis.^21,22 

The Lyl-1 gene encodes a bHLH protein closely related to scl.²³  Its transcriptional activation upon translocation is also associated with T-ALL.²⁴  SCL and LYL-1 bHLH regions show 82% of amino acid identity,²⁴  suggesting that these 2 proteins share at least some target genes and biologic functions.^25,26  However, LYL-1 and SCL diverge largely outside the bHLH region and display a distinct expression pattern in hematopoietic cells.^23,27 

The biologic functions of SCL prompted us to investigate those of LYL-1 in hematopoiesis. In contrast to scl^–/– mice, Lyl-1^–/– mice do not exhibit embryonic lethality but have a reduced number of B cells. Although the CLP compartment is normal, the immature B-cell compartments are reduced in adult mice. In addition, we show that Lyl-1 is highly expressed in stem/progenitor cells, correlating with an important role in the control of the size and function of this cell compartment. Similarly, the number of multipotent progenitor S₁₂ colony-forming units (CFU-S₁₂s) is reduced in Lyl-1^–/– animals. Overall, these defects are distinct from those revealed upon scl KO, suggesting that scl and Lyl-1 exert specific functions during hematopoiesis.

With the use of an amplified fragment of the Lyl-1 gene as a probe, we screened a phage library made with genomic DNA of embryonic stem (ES) cells (a D3 derivative provided as a gift by Drs Bettenhausen and Achim Gossler [Institute for Molecular Biology, Hannover, Germany], unpublished data, 1993) and isolated 2 overlapping phage clones representing 27 kb of the mouse Lyl-1 locus.²⁸  To generate the targeting vector, we constructed a LacZ/Neo gene cassette (pLRLacZpA/MCINeopA), which can be excised in multiple ways allowing in-frame cloning of the β-galactosidase (LacZ) reporter gene into coding regions. A 5.23-kb SmaI/SstI fragment containing the LacZ/Neo cassette replaced a 0.7-kb HpaI fragment of the Lyl-1 locus comprising the HLH domain and the entire 3′ end of the Lyl-1 gene. The 5′ arm of the targeting vector (HindIII/HpaI) spans 7.8 kb, and the 3′ arm (HpaI/EcoRI) spans 3.2 kb of the Lyl-1 locus. The linearized (HindIII) targeting vector was electroporated (Gene Pulser; Bio-Rad, Hercules, CA) into embryonic day 14.1 (E14.1) ES cells.²⁹  Transfected ES cells were grown on a single monolayer of mitomycin C–treated G418-resistant embryonic fibroblast cells in the presence of G418 (200 μg/mL). G418-resistant ES cell clones (n = 147) were expanded and analyzed by Southern blot hybridization.³⁰  Twenty-eight clones contained the targeted Lyl-1 locus. Clones 3, 4, and 10 were further characterized and used to generate chimeric mice. Mutant ES cells were injected into blastocysts of C57BL/6 mice, and the resultant male chimeras were crossed with C57Bl/6 females.

C57BL/6-Ly5.2 and C57BL/6-Ly5.1 mice (6 to 8 weeks old) were purchased from Janvier CERJ (Le Genest-St-Isle, France) and Charles River Laboratories (Les Oncins, France), respectively. For fetal liver (FL) experiments, the vaginal plug was taken as day 0 of gestation. Mice were genotyped by polymerase chain reaction (PCR) of tail DNA with primers hybridizing to exon 4 of Lyl-1 (amplifying the wild-type [wt] allele: forward, 5′-AAGCTGAGCAAGAACGAGGTGC-3′; reverse, 5′-TCTGCTCCAACTTGATGGGTCTC-3′) or to the LacZ (amplifying the mutant allele: forward, 5′-CGTCGTGACTGGGAAAACCC-3′; reverse, 5′-GGAACAAACGGCGGATTGAC-3′). PCR conditions were 94°C for 5 minutes; 35 cycles of 94°C for 30 seconds, 55°C for 1 minute, 72°C for 30 seconds; and 72°C for 7 minutes. Amplified bands were detected on a 1% agarose gel with ethidium bromide.

A sample (50 μL) of peripheral blood (PB) was harvested from the retro-orbital sinuses of anesthetized mice. Total numbers of nucleated blood cells, erythrocytes, platelets, lymphocytes, and granulocytes were automatically measured with an MS9 counter (Melet Schloesing Technologies, Cergy-Pontoise, France) calibrated for mouse blood.

Cells from bone marrow (BM) and E14 FL were harvested in medium essential medium (MEM; Gibco BRL, Paisley, United Kingdom) supplemented with 10% fetal bovine serum (FBS; Invitrogen, Paisley, United Kingdom) by repeated flushing and passage through a 70-μm nylon mesh (Falcon 2350; Becton Dickinson Labware, Franklin Lakes, NJ). Erythrocytes were lysed in ammonium chloride potassium buffer.

BM and E14 FL cells were enriched for Lineage-negative (Lin^–) cells after incubation with a cocktail of lineage antibodies (mAbs): anti–GR-1 (RB6-8C5), anti–MAC-1 (M1/70), anti-B220 (RA3-6B2), anti-CD4 (GK1.5), anti-CD8 (Lyt-1), and anti–erythroid TER-119 (Ter-119). Lin⁺ cells were depleted by the immunomagnetic bead technique (sheep anti–rat IgG; Dynal, Oslo, Norway). Then the Lin^– fraction was incubated with a phycoerythrin (PE)–conjugated goat anti–rat IgG antibody and stained with a fluorescein-isothiocyanate (FITC) anti–Sca-1 mAb (E13-161-7) and an allophycocyanin (APC) anti–c-Kit mAb (2B8). This method allowed us to determine the frequency of Sca-1 and c-Kit cells among the Lin^–, CD34^–, sca-1⁺, c-kit⁺ (LSK) cells.³¹  CD34⁺ cells among LSKs were enumerated after labeling with a cocktail of biotin-lineage antibodies, a PE-anti–Sca-1, an APC-anti–c-Kit, and an FITC-anti–CD34 (RAM 34) mAb followed by phycoerythrin-cyanin-7 (PE-Cy-7) streptavidin.^32,33  Isotype-matched antibodies were used as controls.

Side population (SP) analyses were performed with Hoechst 33342 according to Goodell et al.³⁴  Controls were performed by incubating cells in the presence of 75 μM verapamil (Sigma, St Louis, MO). Cells were subsequently centrifuged and stained with LSK antibodies, as described.

Different subsets of B-cell lineage were sorted according to their expression markers using PE–anti-CD19 (1D3), APC–anti-IgM (II/41), biotin–anti–CD43-57, and APC–Cy-7–anti-B220 mAbs for the different B-cell populations^35,36  of pro–B cells (B220⁺, CD19^–, CD43⁺, IgM^–), pre–B cells (B220⁺, CD19⁺, CD43⁺, IgM^–), and immature B cells (B220⁺, CD19⁺, CD43^–, IgM⁺). CLPs were phenotyped among the Lin^– cells as IL7R⁺ (PE-SB/199), Sca-1^+/low, and c-Kit^+/low, as described by Kondo et al.⁵  β-Galactosidase (fluorescence-activated cell sorter [FACS]–Gal) expression was studied in the different B-cell populations described and in LSK cells stained with biotinylated lineage-specific mAbs followed by streptavidin PE–Cy-7, PE–Sca-1 mAb, and APC–Kit mAb. FACS-Gal assay was performed as previously described³⁷  with fluorescein di-β-galactopyranoside (FDG; Molecular Probes, Invitrogen, Paisley, United Kingdom).

All antibodies were purchased from PharMingen (San Diego, CA), and 1 μg/mL of 7 amino actinomycin (7AAD; Sigma) was added 5 to 10 minutes before analysis to exclude dead cells. For cell cycle analysis of BM cells, Lin^– cells were sorted and stained with 10 μg/mL Hoechst 33342 and 1 μg/mL Pyronine Y (Sigma), as described by Gothot et al.³⁸  Different populations were sorted with a FACSDiva cell sorter (Becton Dickinson, Mountain View, CA) or were analyzed with an LSR II cytometer.

Total RNA was harvested using Trizol reagent (Invitrogen). Samples were first treated with DNase (Ambion, Huntington, United Kingdom), heat denatured, and reverse transcribed using superscript II RNase H reverse transcriptase (Invitrogen). Real-time TaqMan PCR was carried out in the ABI Prism GeneAmp 5700 Sequence detection system (Perkin-Elmer Applied Biosystems, Foster City, CA). Lyl-1 expression level was normalized with tubulin expression. Probes and primers were: Lyl-1 sense, 5′-CAAGCCATAGTGAGCTGGACTTG-3′; Lyl-1 antisense, 5′-AGCGCTCACGGCTGTTG-3′; Lyl-1 probe, 5′-CTGCGGGCACCAACCCCAGA-3′; GAPDH sense, 5′-CCTGCACCACCAACTGCTTA-3′; GAPDH antisense, 5′-TCATGAGCCCTTCCACAATG-3′; GAPDH probe, 5′-CCTGGCCAAGGTCATCCATGACAA-3′.

S12 colony-forming unit (CFU-S₁₂) assay was performed as described by Ohta et al.³⁹  Total BM and FL cells (5 × 10⁴ per mouse) were injected into congenic lethally irradiated (9.5 Gy) recipients. Spleens were excised 12 days later and fixed in Bouin solution, and macroscopic surface colonies were scored.

Long-term culture-initiating cell (LTC-IC) assay was performed as described by Ohta et al³⁹  and Sutherland et al.⁴⁰  Briefly, sorted LSK cells were placed on an S17 feeder layer in microwell plates and incubated at 33°C in a humidified incubator with 5% CO₂ in air. Cultures were fed weekly by half-media exchange. The cultures were recovered after 5 weeks and assayed for the presence of myeloid culture colony-forming unit (CFU-C) using a methylcellulose medium (M3234; Stem Cell Technologies, Vancouver, BC, Canada) supplemented with cytokines. After 7 to 10 days of growth at 37°C, colonies were scored with an inverted microscope based on their morphology.

C57Bl/6 (Ly 5.1) mice were used as recipients,⁴¹  whereas test BM and E14 FL cells were prepared from C57Bl/6 (Ly 5.2) donors. Recipients were lethally irradiated with two 5-Gy doses separated by a 3-hour interval and underwent cotransplantation with 5 × 10⁵, 1 × 10⁶, 5 × 10⁶Lyl-1^+/+ or Lyl-1^–/– Ly5.2 BM or E14 FL cells, together with 1 × 10⁶ Ly5.1 BM competitor cells injected through the retro-orbital sinus (ratios, 0.5:1, 1:1, and 5:1). Hematopoietic reconstitution was assessed 4 and 24 weeks after transplantation through analysis of PB collected from retro-orbital sinus and BM harvested by intrafemoral punction of anesthetized mice. Cells were stained with anti–CD45.2 FITC and were analyzed by flow cytometry Animals showing greater than 1% of Ly-5.2⁺ cells were considered positive for repopulation.

Secondary BM transplantations were also performed: 5 × 10⁵,1 × 10⁶, 5 × 10⁶Lyl-1^+/+ or Lyl-1^–/– Ly 5.2 BM sorted cells from primary mice that underwent grafting 6 months earlier, together with 1 × 10⁶ BM Ly5.1 competitor cells, were injected into secondary Ly5.1 lethally irradiated mice. Hematopoietic reconstitution was evaluated 4 and 24 weeks after the graft by flow cytometry, as described.

The Student t test was used with data obtained from at least 3 repeated experiments to determine the P value.

To evaluate the functional role of Lyl-1 in hematopoiesis, the Lyl-1 gene was disrupted by homologous recombination in ES cells. A LacZ/Neo cassette was cloned in-frame into the fourth exon of the murine Lyl-1 gene, thus deleting the entire 3′ end including the HLH domain (Figure 1A-C; see also “Materials and methods”). This deleted region was similar to that targeted in the conditional scl KO.⁴²  The β-galactosidase enzyme encoded by the LacZ gene was placed under the control of the Lyl-1 promoter. Lyl-1^LacZ mutant mice were intercrossed, and tail DNA from the offspring was analyzed by Southern blotting to confirm the deletion of the 3′ end of the Lyl-1 gene (Figure 1E-F). Homozygous Lyl-1^LacZ mice, hereafter called Lyl-1^–/–, were viable and developed normally.

Expression of the LYL-1/LacZ fusion protein was assessed throughout the hematopoietic system using the FDG assay to reveal β-galactosidase activity.³⁷  According to the percentage of β-galactosidase⁺ cells (Figure 2A) and the intensity of β-galactosidase (Figure 2B), LYL-1 was mainly expressed in BM and in E14 FL LSK cells compared with the Lin^– and total cell populations.

In contrast to scl, lyl-1 is expressed in B cells.^43,44  Therefore, we precisely determined its expression during B-cell differentiation by separating BM cells into pro–B, pre–B, and immature B cells. We found that the percentage of fluorescent cells was increased between the pro-B and pre-B stages (Figure 2C). By quantitative reverse transcription–polymerase chain reaction (RT-PCR), Lyl-1 mRNA in wt mice also increased between the pro-B and the pre-B stages (Figure 2D). These results demonstrate that Lyl-1 expression predominates in the primitive hematopoietic cell compartment and markedly increases during the first steps of B-cell maturation.

We first examined the numbers of nucleated cells, red blood cells, and platelets in the PB of 8- to 12-week-old male and female mice. No difference was noticed between wt and Lyl-1^–/– animals (Tables 1 and 2). Next, we performed flow cytometric analysis on PB, BM, and spleen cells. The numbers of myeloid, erythroid, and T cells were indistinguishable between wt and KO mice. However, a 1.5- to 3-fold decrease in the number of B220⁺ cells in the 3 analyzed tissues was observed (Tables 1 and 2).

We performed phenotypic analysis along the B-cell pathway to characterize the differentiation stage eventually affected by Lyl-1 inactivation. A marked and significant decrease in the B220⁺ CD19⁺ cell population was found in the BM of heterozygote and homozygote mice (Figure 3A). Further phenotyping of the CD19⁺ cells showed a 2-fold reduction in immature B-cell compartments in homozygote mice and an intermediate reduction in heterozygote mice (Figure 3B), suggesting a gene dosage effect. In contrast, the earlier stages of lymphoid differentiation were not affected by Lyl-1 invalidation, as shown by the percentage of CLP (Figure 3C) and pro–B cells.

Thus, the relatively high expression of Lyl-1 at the RNA and protein levels (LYL-1/LacZ fusion protein) in pre–B cells correlates with a decreased number of these cells after this stage in KO mice.

We used flow cytometric analyses to quantify the proportion of immature hematopoietic progenitors and HSCs in the BM of adult mice and E14 fetal livers. The number of LSK cells was reduced approximately 2-fold in Lyl-1^–/– compared with wt mice (0.11% ± 0.015%) in Lyl-1^–/– BM compared with 0.31% ± 0.03% in Lyl-1^+/+ BM and 0.29% ± 0.013% in E14 FL Lyl-1^–/– compared with 0.66% ± 0.027% in Lyl-1^+/+ E14 FL (Figure 4A). However, the LSK population is heterogeneous, and the less mature cells are devoid of CD34.^32,33  We thus examined the percentage of CD34^– cells within the LSK populations and found that their number was markedly decreased in Lyl-1^–/– BM (0.012% ± 0.0017% in lyl-1^–/– vs 0.074% ± 0.008% in Lyl-1^+/+ mice; Figure 4B). Importantly, this decrease (6-fold) was even more significant than that seen for the total LSK population (2-fold). In all experiments, heterozygote Lyl-1^+/–mice harbored an intermediate number of LSKs (Figure 4A-B), suggesting that the effect of Lyl-1 on the HSCs might be dose dependent.

SP cells were also characterized by their ability to efflux the Hoechst 33342 dye.^34,45  No significant difference was seen between Lyl-1^–/– and wt BM (0.8% ± 0.3% vs 1.0% ± 0.36%; Figure 4C). However, SP cells include differentiated cells^46,47  and long-term reconstitution cells (LT-RCs) that are highly enriched in the SP population with an LSK phenotype.^45,48  Thus, we quantified LSK-SP cells and found that their percentage was decreased in the BM of Lyl-1^–/– mice (8.9% ± 3.2% in Lyl-1^+/+ BM vs 3.6% ± 1.2% in Lyl-1^–/– BM; Figure 4C). Furthermore, LSK-SP cells endowed with the highest potential for long-term reconstitution incorporate the lowest level of Hoechst and, hence, are detected in the Hoechst^low population (Figure 4Ciii-iv, end of the tail).⁴⁹  Importantly, we noticed a more pronounced decrease in LSK-Hoechst^low SP cells in Lyl-1^–/– mice (1.47% ± 0.4% vs 5.66% ± 1.7% in Lyl-1^+/+ mice) that was on the same order of magnitude as that observed for CD34^– LSK cells. The same approach was not performed with E14FL cells because the presence and the significance of SP cells at this developmental stage remain controversial.^50,51  Altogether these phenotypic analyses suggest that the HSC compartment with an LT-RC phenotype is quantitatively altered in Lyl-1^–/– mice.

To determine the functional effects of Lyl-1 disruption, we performed spleen colony-forming unit (CFU-S₁₂) and LTC-IC assays, 2 cell populations roughly corresponding to short-term repopulating cells (ST-RCs). The number of CFU-S₁₂s was 1.5-fold decreased in the BM and E14FL from Lyl-1^–/– mice compared with wt mice (Table 3). This contrasted with a nonaltered number of myeloid colony-forming cells (CFCs) (data not shown). LTC-ICs were quantitatively studied by limiting dilution assays, and a 3-fold reduction was observed in BM from Lyl-1^–/– mice (Table 4). Furthermore, the number of myeloid CFCs per LTC-IC was decreased more than 7-fold in Lyl-1^–/– BM, suggesting that Lyl-1 disruption profoundly affects the more primitive LTC-ICs (Table 4). Together these results indicate that cells defining ST-RC are depleted of Lyl-1^–/– BM.

Competitive reconstitution assays were performed to compare the relative repopulating ability of cells from Lyl-1^–/– and wt mice. Different ratios (0.5:1, 1:1, 5:1) of BM cells from Ly5.2 Lyl-1^–/– or Lyl-1^+/+ mice were injected, together with a fixed number of Ly5.1 BM competitor cells, into irradiated Ly5.1 recipients. The contribution of Lyl-1^+/+ and Lyl-1^–/– Ly5.2 cells to PB leukocytes and BM was assessed after 4 and 24 weeks. A much lower contribution of Lyl-1^–/– cells compared with Lyl-1^+/+ was observed in PB (Figure 5A). In addition, the contribution of donor-derived cells increased between weeks 4 and 24 for wt but not for Lyl-1^–/– BM. Whatever the ratio used and the time point considered, the reconstitution derived from the Lyl-1^–/– graft was decreased 1.6- to 5.7-fold (Figure 5A). Strikingly, 5 × 10⁶Lyl-1^–/– BM cells were less efficient than 5 × 10⁵ wt BM cells to ensure long-term reconstitution, indicating that reconstitution capacity was more than 10-fold decreased in the absence of Lyl-1. However, analysis of donor-derived contributions to various hematopoietic lineages showed major differences between the myeloid and the lymphoid lineages, with the ability of Lyl-1^–/– BM to reconstitute the lymphoid (B and T) compartments drastically reduced, whereas reconstitution of the myeloid lineages was impaired only for a short period (Figure 5C).

Intrafemoral punctions were performed to analyze reconstitution of the BM tissue. Contributions of mutant cells were also decreased in comparison with wt cells. However, the contribution of Lyl-1^–/– cells was higher for BM than for PB leukocytes (for instance, compare the injection of 5 × 10⁶ BM cells at 24 weeks; Figure 5A-B). This was related to the major defect of Lyl-1^–/– BM cells to reconstitute B- and T-cell lineages, whereas myeloid cells were only marginally affected except in the short term (Figure 5D). Similar results were obtained with E14 FL (Figure S1, available at the Blood website; see the Supplemental Figure link at the top of the online article).

Secondary transplantations were performed to better evaluate the Lyl-1^–/– HSC functions. BM cells for primary reconstituted mice were harvested at 4 months, sorted on expression of the CD45.2 marker, and injected together with Ly5.1 BM competitor cells into secondary irradiated Ly5.1 recipients. Percentages of donor-derived PB in secondary hosts were analyzed. Whatever the number of cells injected, the proportion of Lyl-1^–/– cells was markedly altered, especially after 24 weeks (Figure 6A). The reconstitution of lymphoid compartments was even more drastically reduced than it was in primary hosts. As for primary engraftment, myeloid reconstitution was not affected by Lyl-1 deletion when 1 or 5 × 10⁶ cells were injected (Figure 6B). Furthermore, no contribution to lymphoid and myeloid hematopoiesis was observed when a low number of Lyl-1^–/– BM cells was transplanted as a secondary procedure (5 × 10⁵Lyl-1^–/–cells, whereas moderate chimerism (approximately 10% at 24 weeks) was observed for wt cells. Taken together, these results demonstrate that Lyl-1^–/– BM cells show a severe defect in their ability to reconstitute the BM and PB of both primary and secondary hosts with a markedly predominant effect on the lymphoid lineages.

To examine whether hematopoietic defects in Lyl-1^–/– BM or E14 FL reflected impaired proliferation, we studied the cell cycle status of multipotent progenitors from these 2 tissues. To this end, RNA and DNA contents of Lin^– and LSK cells were measured by pyronin-Y (PY) and Hoechst 33342 (Ho) staining, respectively, and the fraction of cells in G₀, G₁, or S/G₂-M was determined.³⁸  No difference in cell cycle distribution was observed between multipotent hematopoietic progenitor cell populations enriched from the BM of Lyl-1^–/– mice and their wt littermates (Figure 7A). Similarly, no difference in cell cycle and bromodeoxyuridine (BrdU) incorporation was seen for E14 FL LSK (Table 5 and Figure 7B).

Understanding how HSCs are generated, maintained, or fated to differentiate is of interest for their convenient clinical use.⁵²  In the present study, we studied the role of the LYL-1 bHLH transcription factor in hematopoiesis by means of gene disruption. We report that Lyl-1^–/– mice are viable and fertile. Adult Lyl-1^–/– mice show seemingly normal hematopoiesis, except for a marked reduction in the number of B cells in PB, BM, and spleen and a reduced number of phenotypically characterized stem/progenitor cells. Characterization of the B-cell defect in Lyl-1^–/– mice indicates a partial block in differentiation after the pro-B stage. Further studies indicate that the number of primitive progenitors (CFU-S₁₂) is also reduced upon Lyl-1 disruption. Most important, competitive reconstitution experiments demonstrate that BM or E14 FL cells from mutant mice display a weakened ability to reconstitute short- and long-term hematopoiesis by affecting nearly exclusively the lymphoid lineages. Secondary transplantation experiments confirm both defects. These results point out an important role for Lyl-1 in hematopoiesis.

Lyl-1 inactivation primarily affected reconstitution of the B and T lineages in the blood and in the BM of primary and secondary recipients. It is possible that the specific role of Lyl-1 in B cells contributes to the defect in lymphoid reconstitution and is responsible for the partial impairment of differentiation after the pro-B stage. In contrast, it has been reported that T cells do not express Lyl-1 mRNA.²³  Surprisingly, the production of T cells appeared to be affected to an extent comparable to that of B cells in reconstituted mice, whereas it was unaffected in Lyl-1^–/– control mice. These results suggest that the impaired T lymphopoiesis in mice engrafted with Lyl-1^–/– cells must stem from functional defects either in HSCs or later during differentiation, even if the number of phenotypically characterized CLPs appeared normal in Lyl-1^–/– mice. Conversely, the block at the pre-B or immature B stage in Lyl-1^–/– mice may reflect a distinct defect. It could come as a surprise that a defect occurred specifically in reconstituted mice but not in control mice entirely devoid of Lyl-1. However, the requirement for some genes can arise from competition effects and, hence, is most obviously (or even only) revealed in chimeric animals when mutant and wt cells are simultaneously present. This has been strikingly illustrated in Drosophila wing cell mutants for the bHLH protein Myc.⁵³  Given that our experiments showing impaired lymphoid reconstitution cells are based on a competition between wt and Lyl-1^–/– cells, it is conceivable that HSCs and CLPs lacking Lyl-1 are somehow “outcompeted” by their wt counterparts. We can hypothesize that this major defect of lymphoid reconstitution in Lyl-1^–/– mice may be related to a lack of compensation of LYL-1 function by SCL because SCL is not expressed in lymphoid cells.^43,44 

Several defects may account for impaired HSC function. Our in vivo experiments indicate that the homing of BM and E14 FL cells is not affected by Lyl-1 deletion (data not shown). In addition, pyronin-Hoechst and BrdU staining experiments showed that mutant HSCs cycled like their wt counterparts. Similarly, an impaired release of the Lyl-1^–/– HSC progeny out of the BM appears unlikely because the defect in lymphoid reconstitution was similarly observed in BM and in PB. In addition, we were unable to detect any increase in apoptosis of Lyl-1^–/– HSCs.

Three lines of evidence suggest that Lyl-1 KO does not induce a pure lymphoid differentiation defect, but it also quantitatively affects the HSC compartment. First, marked reductions in CD34^– LSKs and SP-LSKs, 2 cell populations closely related to LT-RCs, were found by immunophenotyping. Second, reduced numbers of CFU-S₁₂ and LTC-IC were demonstrated by functional assay. Third, no contribution of Lyl-1^–/– cells to hematopoiesis was observed in secondary hosts when a low number of cells (5 × 10⁵) was injected. These results demonstrate that the compartment of stem/progenitor cells was quantitatively altered in Lyl-1^–/– mice. The marked defect in the lymphoid potential of Lyl-1^–/– HSCs during reconstitution may be associated with an increased potential for myeloid differentiation. Further studies using serial transplantations and limiting dilution experiments will be necessary to precisely elucidate the HSC defect induced by Lyl-1 gene disruption.

Given the close homology between Lyl-1 and scl and their shared implication in T-ALL, a comparison of their roles in hematopoiesis is of interest.^25,26,44  Obviously, the essential role of scl in embryonic hematopoiesis is not shared by Lyl-1. Moreover, in committed cells, conditional KO experiments in mice after transplantation showed that scl was essential for the completion of megakaryocytic and erythroid differentiation,⁴²  whereas the number of differentiated erythroid cells and platelets was normal in Lyl-1^–/– mice. In fact, an inducible deletion of scl led to a complete absence of BFU-E in the BM of transgenic mice,⁵⁴  whereas our methylcellulose experiments showed that the BFU-E number was increased in Lyl-1^–/– mice, possibly as a result of the enhanced expression of scl (data not shown). Similarly, both Lyl-1 ablation and constitutive enforced scl expression impaired B-cell differentiation,⁵⁵  again suggesting opposite roles for these 2 genes.

However, the scl and Lyl-1 roles in some cell types are not always completely distinct. Indeed, the number of CFU-S₁₂ cells was also reduced 2-fold in both mutants.⁵⁴  Moreover, although scl was initially reported to be dispensable in adult HSCs,⁴²  a more recent study provided evidence that scl-deficient HSCs in BM had poor short-term reconstitutive ability. Interestingly, both studies with scl^–/– mice used KO conditional experiments, but the phenotype was revealed when scl was ablated in BM cells before their engraftment into lethally irradiated mice, an experimental setting close to our reconstitution experiments.⁵⁶  Finally, overexpression of scl in transgenic mice led to an expansion of the SP population in BM presumably endowed with HSC activity.^44,57  These results suggest that scl and Lyl-1 play a related, though not fully redundant, role in adult HSCs, whereas scl performs unique functions in embryonic hematopoiesis and erythroid and megakaryocytic differentiation. Ongoing experiments using scl/Lyl-1 double KO mice will clarify the role of Lyl-1 and scl in immature hematopoietic cells, especially with respect to the control of their number.

At a molecular level, how can we interpret the distinct but possibly overlapping functions of these 2 related transcription factors? SCL is thought to induce erythroid terminal differentiation by acting as a transcriptional activator upon the recruitment of coactivators,^25,58  although some steps of erythropoiesis can be promoted by an SCL protein unable to bind DNA.^19,59  In contrast, in undifferentiated erythroleukemia cells and presumably in T-ALL,⁶⁰ SCL is likely to dampen differentiation by acting rather as a transcriptional repressor, in part through the recruitment of corepressors⁶¹  and possibly through an Id (inhibitor of DNA binding)–like effect.⁶²  Such an ambivalent role for SCL, which makes it a molecular switch endowed with antidifferentiation and prodifferentiation activities, is not unique in its subgroup of bHLH proteins, as exemplified by MyoD.⁶³  The near-perfect similarity and functional interchangeability^25,26  between the SCL and LYL-1 bHLH regions suggests that these 2 transcription factors have at least overlapping target genes in HSCs, where they may act as transcriptional repressors. However, in certain committed cells, SCL, but not LYL-1, would switch to transcriptional activation.^58,61  Differences between the 2 proteins outside the bHLH domain may underlie this molecular specialization. The effects of LYL-1 on B-cell differentiation may be an exception to this rule. Indeed, the block at the pre-B stage we observed in Lyl-1^–/– KO mice resembled the phenotype associated with the deletion of E47, a dimerization partner of LYL-1 harboring an activation domain.⁶⁴  This suggests that one transactivating complex required for B cells to undergo differentiation beyond the pro-B stage contains the E47/LYL-1 proteins as a DNA-binding heterodimer. Future studies will further examine how Lyl-1 regulates the fate of B and hematopoietic stem cells in molecular terms.

We thank Dr M Serwe for his help in establishing the KO mice and Françoise Wendling for discussions and improvements to the manuscript, including English corrections. We thank Patrice Ardouin, Annie Rouches, and all the personnel of the animal facilities for their valuable help in maintaining the mice strains. We extend our special respects to Patrice Ardouin.