A Pentamer Transcriptional Complex Including tal-1 and Retinoblastoma Protein Downmodulates c-kit Expression in Normal Erythroblasts

Abstract

Human proerythroblasts and early erythroblasts, generated in vitro by normal adult progenitors, contain a pentamer protein complex comprising the tal-1 transcription factor heterodimerized with the ubiquitous E2A protein and linked to Lmo2, Ldb1, and retinoblastoma protein (pRb). The pentamer can assemble on a consensus tal-1 binding site. In the pRb⁻ SAOS-2 cell line transiently transfected with a reporter plasmid containing six tal-1 binding site, pRb enhances the transcriptional activity of tal-1–E12–Lmo2 and tal-1–E12–Lmo2–Ldb1 complexes but not that of a tal-1–E12 heterodimer. We explored the functional significance of the pentamer in erythropoiesis, specifically, its transcriptional effect on the c-kit receptor, a tal-1 target gene stimulating early hematopoietic proliferation downmodulated in erythroblasts. In TF1 cells, the pentamer decreased the activity of the reporter plasmid containing the c-kit proximal promoter with two inverted E box-2 type motifs. In SAOS-2 cells the pentamer negatively regulates (i) the activity of the reporter plasmid containing the proximal human c-kit promoter and (ii) endogenous c-kit expression. In both cases pRb significantly potentiates the inhibitory effect of the tal-1–E12–Lmo2–Ldb1 tetramer. These data indicate that this pentameric complex assembled in maturing erythroblasts plays an important regulatory role in c-kit downmodulation; hypothetically, the complex may regulate the expression of other critical erythroid genes.

The role of pRb in ontogenetic development of the hematopoietic system is still a matter of debate. In fact, pRb⁻ mice die in early gestation due to gross defects of both central nervous and hematopoietic systems (9, 24, 33). The latter abnormalities involve reduced embryonic liver erythropoiesis due to hampered differentiation of late erythroid (E) progenitors (CFU-E) (9, 24, 33). On the other hand, other studies have suggested that the effect of pRb on E differentiation might not be cell autonomous (35).

We have investigated the expression and function of pRb in normal human adult hematopoiesis, as revealed by analysis of purified hematopoietic progenitor cells (HPCs) differentiating selectively through the E or granulopoietic (G) pathway (11). During the initial HPC differentiation stages, the RB gene is gradually induced at mRNA and protein levels in both E and G cultures. During late HPC differentiation and then precursor maturation, pRb expression is sustained in the E lineage, whereas it is downmodulated in the G series. In agreement with this expression pattern, CFU-E treatment with an antisense oligomer targeting Rb mRNA causes a dose-dependent inhibition of colony formation. In line with our studies, RB gene transfer favors terminal differentiation of a mouse erythroleukemic (MEL) cell line (45); furthermore, Rb^−/− fetal liver progenitor cells transplanted in vivo show a selective maturation defect in the erythroblast series (23).

In normal erythropoiesis, dephosphorylated pRb may be present in sufficient amounts to capture other transcription factors (TFs) in addition to the E2F products (28, 62); hypothetically, it may associate with and potentiate the activity of erythrocyte-specific TFs (11). Furthermore, pRb can inhibit cell cycle progression and promote differentiation in SAOS-2 osteosarcoma cells; using pRb mutants unable to bind E2F, it was possible to dissociate these two functions (47).

The TAL-1 gene (also known as SCL or TLC-5), identified by analysis of t(1;14) (p32;q11) translocations in human T-lymphocytic leukemia (T-ALL), codes for the tal-1 protein belonging to the family of basic helix-loop-helix (bHLH) domain TFs (reviewed in reference 4). The TAL-1 gene, although silent in normal adult T lymphocytes (5, 57), is constitutively activated in >60% of T-ALLs (3); in transgenic mice constitutive tal-1 expression in T cells causes T-lymphocytic neoplasias (13, 27).

In vitro the tal-1 protein heterodimerizes with products of the E2A gene: the heterodimer preferentially binds to the E box consensus motif CAGATG (E box-1 type [see Table 1]) with a strict requirement for the adjacent bases (20–22). Recent casting experiments have defined an extended tal-1–E2A binding site sequence associated with the GATA site comprising the E box consensus CAGGTG (E box-2 type [see Table 1]) with little requirement for adjacent bases (10, 60).

TABLE 1.

Oligonucleotides used in the DNA-binding studies

Oligonucleotide	Sequencea (5′–3′)
E box-1 type	ACCTGAACAGATGGTCGGCT
E box-1 type mutant	ACCTGAACcGATtGTCGGCT
E box-2 type	TCGGCGCCAGGTGCTGCGTC-3′
c-kit (−383 to −369 nt)	AGCACCTGGCAGGTGGCGG
c-kit E-1 mutant	AGaACCTGGCAGGTgGCGG
c-kit E-2 mutant	AGCACCTGGCAGGTgGCGG-3′
c-kit double mutant	AGaACCTGGCAGGTgGCGG

^aE boxes are underlined. Mutations are shown in lowercase. 

In ontogenetic development, the absence of tal-1 determines a block of early blood cell formation (46, 49) involving all hematopoietic and lymphoid lineages (42). In normal or leukemic adult hematopoiesis, tal-1 expression is restricted to CD34⁺ HPCs and E, megakaryocytic, and mastocytic lineages (38, 43). We have investigated the expression and function of tal-1 in purified HPCs channeled into unilineage E and G differentiation and maturation (12). The expression pattern of the TAL-1 gene is similar to that of RB. (i) tal-1 mRNA is induced and sustainedly expressed in E differentiation and maturation, while it is only transiently induced in the first week of G differentiation. (ii) The expression pattern of the tal-1–E2A heterodimer was consistent with mRNA assay results, and, more importantly, treatment of HPCs with an antisense oligomer targeting tal-1 mRNA causes a selective, dose-related inhibitory effect on CFU-E colony formation. (iii) Finally, enforced tal-1 expression stimulates primitive, E, and megakaryocytic HPCs but blocks the G differentiation program (55).

Growing evidence indicates that tal-1 biochemically interacts with not only E2A but also other transcriptional proteins, particularly Lmo2, Ldb1, and GATA-1.

Lmo2, a nuclear protein with two cysteine-rich LIM domains essential for E differentiation (6), is complexed with tal-1 in differentiating MEL cells (53); Lmo2 expression in an HPC unilineage E culture is similar to that of tal-1 (our unpublished results). Lmo2 and tal-1 synergize to induce T-cell tumors in transgenic mice (31), thus suggesting their functional interaction. A partner of Lmo2 called Ldb1, NL1, or Clim-2 has been identified (1, 25). In MEL cells, Ldb1 and Lmo2 proteins form a stable complex (58, 60), while forced expression of the Ldb1 or the LMO2 gene inhibits E cell maturation in the G1ER proerythroblast cell line (58). Furthermore, a multiprotein complex composed of tal-1–E2A, Lmo2, and Ldb1 can assemble on the E box-1 type in mouse fetal liver erythroblasts (58). Finally, GATA-1 has been reported to physically interact with Lmo2 (40). tal-1–E2A, Lbd1, Lmo2, and GATA-1 can assemble on and transcriptionally activate a promoter containing a bipartite binding motif containing an E box-2 type followed by a GATA site in MEL cells (60).

Recent evidence suggests that a putative TAL-1 target gene is c-kit (29), i.e., a transmembrane receptor for a hematopoietic growth factor (HGF), termed c-kit ligand (KL) or stem cell factor, which plays a key role in early hematopoietic proliferation. In an HPC unilineage E differentiation culture, c-kit expression is characterized by a progressive decline, starting from the CFU-E–early erythroblast stage through terminal E cells (16); interestingly, this declining pattern is inversely related to that of Rb and tal-1 expression, which peaks at the CFU-E level and is sustainedly expressed through the erythroblast series (11, 12).

We report the biochemical interaction of pRb with a tal-1–E2A–Lmo2–Ldb1 tetramer complex in human adult proerythroblasts and erythroblasts. The c-kit promoter region containing two inverted E box-2 type motifs binds the Rb–tal-1–E2A–Lmo2–Ldb1 complex; this pentamer negatively regulates the c-kit promoter activity both in hematopoietic (TF1) and nonhematopoietic (SAOS-2) cells. Furthermore, the pentameric complex inhibits expression of endogenous c-kit in transiently transfected pRb⁻ SAOS-2 cells.

MATERIALS AND METHODS

Recombinant human HGFs and culture medium.

Interleukin-3 (IL-3), IL-6, and granulocyte-macrophage colony-stimulating factor (GM-CSF) were supplied by Genetics Institute (Cambridge, Mass.), KL and flt3 ligand (FL) were supplied by Immunex (Seattle, Wash.), and Epo and bovine basic fibroblast growth factor (bFGF) were supplied by Amgen (Thousand Oaks, Calif.). Recombinant human granulocyte-CSF (G-CSF) and macrophage-CSF (M-CSF) were purchased from R&D Systems (Minneapolis, Minn.), and thrombopoietin (Tpo) was purchased from Peprotech (London, England). Iscove's modified Dulbecco's medium (IMDM) (GIBCO-BRL, Grand Island, N.Y.) was freshly prepared weekly.

HPC purification and unilineage E differentiation and maturation in liquid-phase culture.

Adult peripheral blood was obtained from 20- to 40-year-old healthy male donors after informed consent, and a buffy coat was prepared by centrifugation (11). The CD34⁺ HPCs were purified according to a modification (11, 67) of a previously reported method (17). The clonogenic assay of purified HPCs in fetal calf serum-positive (FCS⁺) culture was modified from the previously reported procedure (30) by addition of not only KL (10 ng/ml), IL-3 (100 U), GM-CSF (10 ng), and Epo (3 U), but also bFGF (100 ng), FL (100 U), IL-6 (10 ng), M-CSF (250 U), G-CSF (500 U), and Tpo (100 ng) (67).

(i) Unilineage E culture.

Step IIIP HPCs grown in FCS⁻ liquid culture (5 × 10⁴ cells/ml in IMDM, supplemented as indicated in reference 54) were induced to specific E differentiation by appropriate HGF combinations (a saturating Epo dose and low IL-3 and GM-CSF doses [3 U/ml, 0.01 U, and 0.001 ng, respectively]) (11).

(ii) Morphology analysis.

Cells were harvested on different days, smeared on glass slides by cytospin centrifugation, and stained with May-Grünwald Giemsa.

(iii) Membrane phenotype analysis.

Cells were incubated for 30 min at 4°C phycoerythrin- or fluorescein isothiocyanate-labeled anti-CD34, anti-glycophorin A (Immunotech, Marseilles, France), and anti-CD11b (Becton Dickinson) monoclonal antibodies (MAb) and analyzed as described above.

The clonogenetic assay of step IIIP HPCs differentiating in E culture was performed as described previously (30) upon addition of the HGF combination used in the unilineage E liquid-phase culture.

Nuclear extracts and IP.

Nuclear extracts were precleared with 25 μl of protein A- or G-Sepharose (Sigma) in 300 μl of immunoprecipitation (IP) buffer containing 375 mM NaCl, 20 mM HEPES, 1.5% Triton X-100, 2.5 mM EDTA, 1.5 mM MgCl₂, and 2 mg of bovine serum albumin (BSA)/ml for 30 min at 4°C on a rolling platform. Each cleared supernatant was incubated overnight at 4°C with polyclonal antisera anti-Tal-1 7742 (14), anti-Rb SC-15 (Santa Cruz Biotechnology, Santa Cruz, Calif.), anti-Lmo2/RbTN2 (40), anti-Ldb1 (25), and anti-E2A E526, -E12 (H-208, V18), and -E47 (N-649) (Santa Cruz Biotechnology) or with MAb anti-Rb (XZ55, XZ104, and XZ133) (Pharmingen), anti-tal-1 (BTL73 and 2TL75) (43), and anti-E12 and -E47 (Pharmingen, San Diego, Calif.). The same amount of irrelevant anti-CD3 MAb or normal rabbit or mouse serum was used as the negative control. After incubation for 2 h with protein A- or G-Sepharose (polyclonal antibody or MAb, respectively) immunoprecipitates were collected by centrifugation and washed three times in IP buffer and once with IP buffer without BSA. Samples were denatured in sodium dodecyl sulfate loading buffer and separated by SDS-polyacrylamide gel electrophoresis (PAGE).

Western blotting and EMSA.

The protein concentrations of nuclear extracts were determined by the Bradford assay (Bio-Rad). Western blot analysis was performed with anti-tal-1 rabbit serum 1080 (21), anti-Rb rabbit serum SC-15 (Santa Cruz Biotechnology), pRb MAb XZ55 (Pharmingen), anti-E2A rabbit serum 526 (21), and H-208 (Santa Cruz Biotechnology) by enhanced chemiluminescence, according to the manufacturer's protocol (Amersham).

An electrophoretic mobility shift assay (EMSA) was performed as previously reported (12). Each reaction mixture contained, in 20 μl, 10 to 25 μg of nuclear extract, 10 mM HEPES (pH 7.9), 50 mM KCl, 2 mM MgCl₂, 4% Ficoll, 1 mM EDTA, 1 mM dithiothreitol, 0.5 μg of poly(dI-dC), and 1 to 3 pmol of a ³²P-labeled, double-stranded oligonucleotide probe containing the E box-1 type or E box-2 type sequence (Table 1). After 15 min of incubation at room temperature, the assay mixture was loaded onto a 15-cm 4% polyacrylamide gel containing 0.25× Tris-borate-EDTA electrophoresis buffer and electrophoresed at 180 V at 4°C for 2 to 3 h. In some binding reactions, the extracts were preincubated for 10 min at room temperature with 1 μl of one of the following reagents: anti-tal-1 rabbit antiserum 370 (21), anti-E2A rabbit antiserum 526, anti-E12 (V-18) rabbit antiserum (Santa Cruz Biotechnology), anti-E-47 (N-649) rabbit antiserum (Santa Cruz Biotechnology), anti-LMO2 rabbit antiserum (kindly provided by T. H. Rabbits), anti-Ldb1 (25), anti-pRB MAb XZ55 and XZ77 (kindly provided by A. Felsani), anti-E2F-1 MAb (Santa Cruz Biotechnology), anti-GATA-1 MAb (Santa Cruz Biotechnology), an irrelevant anti-CD3 MAb, or normal rabbit or mouse immunoglobulin (negative control). A 100-fold molar excess of E box-1 type unlabeled competitor or mutated oligonucleotides that bear two nucleotide substitutions in the E box core and an unlabeled c-kit oligonucleotide competitor were included in some binding reaction mixtures (Table 1).

Plasmid clone.

Expression plasmids encoding human Lmo2, tal-1, Ldb1, E12, pRb, p107, and p130 were constructed by subcloning the relevant cDNA sequences (12, 14) with CMV-neo BamHI (2) or pCDNA3 vectors (Pharmacia).

pCMV-SEΔ encodes the Rb C terminus from a point corresponding to the Ssp 1 site to the end with an internal deletion (amino acids [aa] 785 to 806). pCMV-A/B was constructed by excising wild-type (wt) RB from the EcoRI site in exon 13 to the Mun 1 site in exon 24 (encoding aa 834).

The pXP2-c-kit reporter was constructed by subcloning a fragment of the c-kit promoter from nucleotide (nt) −443 to nt −1 (64) in the SmaI/BglII site of pXP2 luciferase vector (39). For the Dual-luciferase reporter assay system (Promega, Madison, Wis.) in TF1 cells, the same fragment of the c-kit proximal promoter was cloned in the KpnI/SacI site of the firefly luciferase reporter vector pGL3-Basic (Promega) to generate the plasmid pGL3-c-kit reporter.

Transient transfection and luciferase assays.

The human osteosarcoma cell lines SAOS-2 and TF1 were obtained from the American Type Culture Collection. All cells were grown in IMDM with 10% FCS (GIBCO-BRL). In the TF1 cells the medium was also supplemented with 5 ng of human GM-CSF/ml.

Transfection in SAOS-2 cells was performed by the standard calcium phosphate method (14). SAOS-2 cells (at ∼70% confluence in a 100-mm-diameter dish) were transfected with 32 to 36 μg of input plasmid DNA containing one to seven DNA constructs which expressed β-galactosidase, tal-1, E12, Lmo2, Ldb1, pRb, p107, p130, and the pRb mutants SEΔ and A/B.

The plasmid DNA sample included 2.5 μg of pSV-Bgal (Promega); 5 μg of pE1b-LUCE6, pXP2-c-kit reporter, pXP2, and pGL2; 2 μg of pCMV-E12; 8 μg of pCMV-Tal-1, pCMV-Lmo2, and pCMV-Ldb1; increasing amounts of pCMV-Rb (ranging from 0.5 μg to 5 μg), and 5 μg of pCMV 107, pCMV-130, pCMV-SEΔ, and pCMV-A/B. This was supplemented with carrier DNA (pCMV vector) to provide a constant amount of DNA per sample. Rous sarcoma virus-driven luciferase (RSV-luc) was used as an external control for all transfection and pXP2 or pGL2 (Promega) was used as a negative control. Forty hours after transfection, cells were collected and lysed and the β-galactosidase or luciferase assay was performed according to the manufacturer's protocol (Promega). Each sample was assayed in an Optocomp luminometer for light emission during the 60 s immediately following injection of 100 μl of luciferin (150 μg/ml). The luciferase activity of each transfected SAOS-2 lysate was normalized with respect to β-galactosidase activity in order to evaluate the variation in DNA uptake.

The transfection protocol for TF1 cells by electroporation and the luciferase assay using β-galactosidase as an internal control were performed as previously reported by Krosl et al. (29). In addition the transcriptional activity of the c-kit proximal promoter cloned in the pGL3 vector was also tested using the Dual-luciferase reporter assay system (Promega). In this experiment the pSV-Bgal was replaced by 50 ng of pRL-TK control vector for normalization in the input DNA. Quantification of the luminescence signals from each of the two luciferase reporter enzymes was performed according to the manufacturer's protocol (Promega).

RESULTS

Biochemical studies of protein interaction were performed on ≥97% pure erythroblasts generated by ≥90 to 95% purified HPCs in a unilineage E culture (66); in the second week of culture E cells gradually mature from proerythroblasts-basophilic erythroblasts through orthochromatic erythroblasts (30). Representative results for erythroblasts from a culture on days 4, 8, and 9 are shown here. However, in all cases, results similar to those for the day 8 or 9 culture were obtained for day 7, 10, or 12 erythroblasts (data not presented).

EMSA of tal-1–E2A–Lmo2–Ldb1 complex in HPC unilineage E culture.

To evaluate the DNA-binding activity of tal-1 in HPCs differentiating in unilineage E cultures, EMSA was performed on the nuclear extracts from a day 8 E culture with a radiolabeled oligonucleotide probe containing the preferred tal-1 E box-1 type consensus sequence (21).

As shown in Fig. 1A, incubation of this probe with the E nuclear extracts (lane 3) generated several protein-DNA complexes; two of these complexes specifically competed by an excess of wt but not mutated cold tal-1 E box-1 type oligonucleotide (lanes 4 and 5, respectively) were further analyzed. As previously reported, the high-mobility complex represents the tal-1–E2A heterodimer (12). The low-mobility complex has a higher molecular weight in the E lineage (Fig. 1A, lane 3) than in Jurkat cells (Fig. 1A, lane 2). In the latter cells the low-mobility band contains the E2A homodimers (12, 21). In day 8 erythroblasts, the low-mobility band (Fig. 1B, lane 3) was supershifted or partially abrogated by incubation with anti-LMO2 antiserum (lane 5), anti-tal-1 antiserum (lane 7), anti-E2A antiserum (lane 9), and anti-Ldb1 antiserum (Fig. 1C, lane 4) but not with the corresponding preimmune antisera (Fig. 1B, lanes 4, 6, and 8) or normal rabbit serum (Fig. 1C, lane 3); this is in line with recent studies indicating tal-1–E2A–Lmo2–Ldb1 complex formation (10, 58, 60). These results indicate that a tal-1–E2A–Lmo2–Ldb1 complex is present in the E low-mobility band. This low-mobility complex is detected from the stage of CFU-E proerythroblasts (i.e., day 7 and 8 unilineage E culture) to maturing erythroblasts (i.e., through day 14) in the erythropoietic lineage but not in the G series (12), thus suggesting a stage- and lineage-specific role for this complex in E differentiation.

FIG. 1.

EMSA of tal-1–E2A–Lmo2–Ldb1 complexes in day 8 erythroblasts. A double-stranded ³²P-labeled oligonucleotide probe containing the tal-1 E box-1 type consensus sequence (upper strand: ACCTGAACAGATGGTCGGCT [the E-box is underlined]) was incubated with nuclear extracts from Jurkat (A and B, lane 2; C, lane 1; D, lane 2) and E culture cells (A, lanes 3 through 5; B, lanes 3 through 9; C, lanes 2 through 4; D, lanes 3 through 5). Incubation of this probe with the nuclear extracts from Jurkat cells generated several protein-DNA complexes. As previously reported (12), one of these complexes, represents the tal-1–E2A heterodimer (A, B, and D, lane 2; C, lane 1). The E cells exhibit a band representing the tal-1–E2A heterodimer and a low-mobility complex (arrow) not present in Jurkat cells (A, B, and D, lane 3; C, lane 2). This low-mobility complex is self-competed by a cold tal-1 E box-1 type oligonucleotide (A, lane 4) but is not abrogated by a cold tal-1 E box-1 type mutant (A, lane 5) and is sharply supershifted by incubation with anti-LMO2, anti-tal-1, and anti-E2A serum (B, lanes 5, 7, and 9 [asterisks, supershifted bands]), anti-Ldb1 serum (C, lane 4), and two anti-pRb MAb, XZ55 and XZ77 (D, lanes 4 and 5, respectively [asterisk, supershifted bands]), while it is not affected by the corresponding preimmune serum (B, lanes 4, 6, and 8), normal rabbit serum (C, lane 3), or an irrelevant anti-CD3 MAb (D, lane 3). mt, mutant.

Interaction of pRb with the tal-1–E2A–Lmo2–Ldb1 complex by EMSA.

The expression and function of pRb in normal adult erythropoiesis (11) are similar to those observed for tal-1, E2A, and Lmo2 (12, 66), suggesting that pRb may interact with the last three proteins in E development. In an attempt to provide direct evidence for a multiprotein complex comprising pRb and tal-1–E2A–Lmo2–Ldb1, we investigated the low-mobility complex from day 8 erythroblasts by supershift analysis. Thus, two anti-pRb MAb, XZ55 and XZ77, were separately added to the EMSA mixture (Fig. 1D). Each of these antibodies supershifted the low-mobility band (lanes 4 and 5), compared with the irrelevant anti-CD3 antibodies (lane 3), thus suggesting that pRb is present in the tal-1–E2A–Lmo2–Ldb1 complex. The low-mobility band was not supershifted by antibodies to E2F-1 or GATA-1 (data not shown).

pRb and the tal-1–E2A–Lmo2–Ldb1 complex interact in differentiating E precursors: reciprocal IP and Western blot analysis. (i) pRb is complexed with tal-1.

In an attempt to explore the “in vivo” biochemical interaction between Rb and tal-1 in purified HPCs induced to unilineage E differentiation, nuclear extracts from a day 8 erythroblast culture were immunoprecipitated by either a pRb MAb (XZ104) or anti-tal-1 MAb BTL73 and 2TL75.

The resulting immunoprecipitates were resolved by SDS-PAGE and subjected to Western blotting analysis using antibodies specific for pRB (SC-15) or tal-1 (tal-1 rabbit antiserum 1080) (Fig. 2A and B, respectively). This allowed precipitation of the pRb protein with an anti-TAL-1 antiserum (Fig. 2A, lane 1). The reciprocal experiment using the pRb MAb XZ104 identified a band corresponding to the tal-1 protein (Fig. 2B, lane 3). These IP results were not due to nonspecific antibody cross-reaction; in fact, antibodies against pRb epitopes or anti-tal-1 antibodies did not recognize the tal-1 protein or pRb, respectively (data not shown).

FIG. 2.

pRb is complexed with tal-1 in differentiating day 8 erythroblasts: reciprocal IP and Western blot (WB) analysis. Nuclear extracts from day 8 erythroblasts or SAOS-2 cells were immunoprecipitated by anti-pRb MAb (XZ133), anti-tal-1 MAb BTL73 and 2TL75, or anti-CD3 MAb. The resulting immunoprecipitates were resolved by SDS-PAGE and subjected to WB analysis using antibodies specific for pRb (SC-15) (A) or tal-1 (no. 1080) (B). (A) pRb was coprecipitated by a mixture of anti-tal-1 MAb BTL73 and 2TL75 (lane 1) but was not detected in the anti-CD3 immunoprecipitate (lane 3). pRb was also absent in the anti-tal-1, anti-pRb, and anti-CD3 immunoprecipitates from SAOS-2 cells (lanes 4 to 6). The anti-pRb immunoprecipitate from erythroblasts and nuclear extracts from Jurkat cells were used as positive controls (lanes 2 and 7, respectively). (B) In the reciprocal experiment, the tal-1 protein was detected in the anti-pRb (XZ133) immunoprecipitates but was absent from the anti-CD3 immunoprecipitates (lanes 3 and 4, respectively). The Jurkat cell line and anti-tal-1 immunoprecipitates from erythroblasts were used as positive controls (lanes 1 and 2, respectively).

(ii) pRb is complexed with E2A.

To further investigate the interaction between Rb and the tal-1–E2A–Lmo2–Ldb1 complex, nuclear extracts from day 8 erythroblasts were immunoprecipitated by either pRb antibodies (SC-15 in Fig. 3A, lane 5; XZ104 in Fig. 3B, lane 4) or E2A antibodies (E12 [H-208] [Fig. 3A, lane 3]; E12 and E47 MAb [Fig. 3B, lane 3]). The resulting immunoprecipitates were resolved by SDS-PAGE and analyzed by Western blotting using antibodies specific for pRb (XZ55) or E2A (rabbit antiserum 526) in Fig. 3A and B, respectively. We were again able to coprecipitate pRb protein by an anti-E2A antibody (Fig. 3A, lane 3), and, in the reciprocal experiment, we used anti-pRb MAb to coprecipitate a band corresponding to the E2A protein (Fig. 3B, lane 4). Once more, these IP results were not due to specific antibody cross-reaction: antibodies against pRb or anti-E2A did not recognize E2A or pRb, respectively (data not shown).

FIG. 3.

pRb is complexed with E2A in differentiating day 8 erythroblasts: reciprocal IP and Western blot (WB) analysis. (A) Nuclear extracts from day 8 erythroblasts were immunoprecipitated by anti-E2A rabbit serum (E12 [H-208]) (lane 3), normal rabbit serum (lane 4), or anti-pRb antibody (SC-15) (lane 5). The resulting immunoprecipitates were resolved by SDS-PAGE and subjected to WB analysis using an anti-pRb MAb (XZ55). pRb was coprecipitated by an anti-E2A rabbit serum (lane 3) and was not detected by normal rabbit serum (lane 4). (B) The lysates from a day 8 E culture were immunoprecipitated by anti-E2A (E12 and E47) (lane 3), anti-pRb (XZ104) (lane 4), or anti-CD3 (lane 2) MAb. The resulting immunoprecipitates were resolved by SDS-PAGE and subjected to WB analysis using anti-E2A (no. 526) serum. The E2A protein was detected in the anti-pRb immunoprecipitates (lane 4) but was absent in the anti-CD3 immunoprecipitates (lane 2). Nuclear extract from Jurkat cells (A and B, lane 1) or anti-pRB immunoprecipitate from erythroblasts (A, lane 5, and B, lane 4) were used as positive controls. n.r.s., normal rabbit serum.

(iii) pRb is complexed with Lmo2.

Nuclear extracts from a day 8 E culture were immunoprecipitated by anti-Lmo2 rabbit antiserum or pRb antibody (SC-15). The resulting immunoprecipitates were resolved by SDS-PAGE and subjected to Western blotting analysis using antibodies specific for pRb (XZ55). A band corresponding to pRb was coprecipitated by anti-LMO2 serum (Fig. 4A, lane 3).

FIG. 4.

(A) pRb is complexed with Lmo2 in differentiating day 8 erythroblasts: IP and Western blot (WB) analysis. Nuclear extracts from day 8 erythroblasts were immunoprecipitated by anti-LMO2 rabbit serum (lane 3), normal rabbit serum (lane 4), or anti-pRb antibody (SC-15) (lane 5). The resulting immunoprecipitates were resolved by SDS-PAGE and analyzed by Western blotting using a pRb MAb (XZ55). pRb was coprecipitated by anti-LMO2 rabbit serum (lane 3) but was not detected by the normal rabbit serum (lane 4). Nuclear extracts from SAOS-2 (lane 1) and Jurkat (lane 2) cell lines or anti-pRB immunoprecipitates from erythroblasts (lane 5) were used as negative and positive controls, respectively. (B) pRb is complexed with Ldb1 in differentiating day 8 erythroblasts: IP and Western blot analysis. Nuclear extracts from day 8 erythroblasts or the SAOS-2 cell line were immunoprecipitated by anti-Ldb1 rabbit serum (lanes 2 and 3, respectively) or anti-pRb antibody (SC-15) (lane 1). The resulting immunoprecipitates were resolved by SDS-PAGE and analyzed by Western blotting using an anti-pRb MAb (XZ55). pRb was coprecipitated by anti-Ldb1 rabbit serum in E cells (lane 2) but not in SAOS cells (lane 3). n.r.s., normal rabbit serum.

(iv) pRb is complexed with Ldb1.

Nuclear extracts from a day 8 E culture were immunoprecipitated by anti-Ldb1 rabbit antiserum or pRb antibody (SC-15). The resulting immunoprecipitates were resolved by SDS-PAGE and analyzed by Western blotting using antibodies specific for pRb (XZ55). We were able to coprecipitate pRb protein by anti-Ldb1 serum (Fig. 4B, lane 2).

Altogether, the IP and Western blot experiments in paragraphs i to iv indicate a biochemical interaction between pRb and the tal-1–E2A–Lmo2–Ldb1 complex.

The human c-kit receptor contains two inverted E box-2 type motifs in the promoter regions that bind a complex comprising tal-1, E2A, Lmo2, Ldb1, and pRb.

Recent evidence (29) suggests that c-kit is a possible tal-1 downstream target gene. Therefore, we investigated the c-kit promoter region containing a putative tal-1 E box sequence to assess the biological expression and function of the tal-1–E2A–Lmo2–Ldb1–pRb complex in HPCs.

We analyzed a proximal promoter region containing two inverted E box-2 type motifs at positions −383 to −369 nt upstream of the initiating methionine (Table 1) separated by 1 nt. EMSA was performed on the nuclear extracts from a day-8 E culture with a radiolabeled oligonucleotide probe containing the inverted E box-2 type motifs found in the c-kit promoter or mutant sequences where the upstream (E1 mutant) or downstream (E2 mutant) E box-2 type or both sites (double mutant) were mutated (Table 1).

As shown in Fig. 5A, incubation of a c-kit oligonucleotide probe with the E nuclear extracts (lane 2) generated several protein-DNA complexes; one of these complexes has a molecular weight similar to that of the low-mobility complex detected with the E box-1 type consensus sequence (21) (lane 1) and is subtotally abrogated by incubation with a 100-fold excess of the unlabeled c-kit oligonucleotide (lane 3). However, this particular low-mobility complex was not observed when the E nuclear extracts were incubated with the double-mutant c-kit oligonucleotide (lane 4) but can still bind to the E1 mutant c-kit oligonucleotide or, to a lesser extent, to the E2 mutant c-kit oligonucleotide (lanes 5 and 6, respectively). This observation indicates that recognition by the low-mobility complex of the c-kit oligonucleotide motif requires binding of an individual E box site.

FIG. 5.

(A) EMSA of a specific protein complex that binds two inverted E box-2 type motifs contained in the c-kit promoter in E nuclear extract at day 8. A series of double-stranded ³²P-labeled oligonucleotide probes encoding the E box-1 type oligonucleotide (lane 1), the c-kit oligonucleotide (lanes 2 and 3), a double mutant c-kit oligonucleotide (lane 4), a c-kit mutant E1 oligonucleotide (lane 5), and a c-kit mutant E2 oligonucleotide (lane 6) were incubated with nuclear extract from day 8 erythroblasts. The low-mobility protein-DNA complexes (arrow) are self-competed by a cold c-kit oligonucleotide. This low-mobility complex is detected by the E1 mutant or, to a lesser extent, by the E2 mutant c-kit oligonucleotide (lanes 5 and 6, respectively) but does not bind the double c-kit mutant oligonucleotide (lane 4). (B to D) Identification of proteins which recognize the c-kit E box-2 type motif in E cells. A double-stranded ³²P-labeled oligonucleotide probe containing the E box-1 type consensus sequence (B and D, lanes 1 and 2) or the c-kit oligonucleotide (B, lanes 3 through 8; C, lanes 1 through 4; D, lanes 3 through 8) was incubated with nuclear extract from Jurkat (B and D, lane 1) and E culture cells (B, lanes 2 through 8; C, lanes 1 through 4; D, lanes 3 through 8). The low-mobility complex is sharply supershifted or partially abrogated with anti-Lmo2 (B, lane 5), anti-TAL-1 (B, lane 7), anti-E47 or anti-E2A (C, lanes 3 and 4, respectively), anti-pRb (D, lane 5), or anti-Ldb1 (D, lane 7), while it is not affected by the corresponding preimmune serum (B, lanes 4 and 6), normal rabbit serum (C, lane 2; D, lane 6), or irrelevant antibodies (D, lane 4). Arrow and dash, bands corresponding to tal-1–Lmo2–E2A–pRb–Ldb1 or tal-1–E2A complexes, respectively. Asterisk, high-mobility complex that birds to the first E box-2 type motifs (A) or supershifted band (D).

As shown in Fig. 5B and C, this band was supershifted or partially abrogated by incubation with anti-Lmo2 antiserum (Fig. 5B, lane 5), anti-tal-1 antiserum (Fig. 5B, lane 7), and anti-E47 or anti-E2A antiserum (Fig. 5C, lanes 3 and 4, respectively) but not with the corresponding preimmune antisera (Fig. 5B, lanes 4 and 6) or the normal rabbit serum (Fig. 5C, lane 2).

Interestingly, this low-mobility complex was supershifted by incubation with an anti-pRb MAb (XZ55) and anti-Ldb1 antiserum (Fig. 5D, lanes 5 and 7, respectively), compared with results for the irrelevant anti-CD3 antibodies (lane 4) or the normal rabbit serum (lane 6).

Altogether, the EMSA results show that a multicomponent complex composed of tal-1, E2A, Lmo2, Ldb1, and pRb binds to inverted E box-2 type motifs but not to the tal-1–E2A heterodimer or the E2A-E2A homodimers.

Interestingly, a high-mobility complex is also detected by incubation of the c-kit oligonucleotide with the E nuclear extract (Fig. 5A). This complex, which does not contain tal-1, E2A, Lmo2, Ldb1, or pRb, binds to the first E box-2 type motif (Fig. 5A, compare lanes 5 and 6). Further studies are ongoing in our laboratory to identify the composition of this high-mobility complex.

pRb enhances the transcriptional activation by the tal-1–E12–Lmo2 complex in transient transfection assay.

To assess the functional significance of the interaction between pRb and the tal-1–E12–Lmo2 complex, the effect of pRb on the transcriptional activity by tal-1–E12–Lmo2 was examined in transiently transfected Rb⁻ SAOS-2 cells using an artificial reporter plasmid (E1b Luc-E6) containing six E box-1 type binding sites for tal-1–E2A heterodimers (see Materials and Methods).

We observed (Fig. 6A and results not shown) that tal-1, E12, and Lmo2, alone or in combination (particularly tal-1–E2A; also tal-1–Lmo2 or E2A–Lmo2 [data not shown]), did not significantly increase luciferase activity (lanes 2 to 6), compared with vector E1b Luc-E6 alone (lane 1). However, coexpression of pRb enhances up to five- to sixfold the transcriptional activity of the tal-1–E12–Lmo2 complex but not that of the tal-1–E12 heterodimer (lanes 6 and 5, respectively). It is noteworthy that coexpression of pRb increased luciferase activity in a dose-dependent manner up to 5 μg of pRb (Fig. 6B).

FIG. 6.

(A) pRb activates the tal-1–E12–Lmo2 complex for transcription driven by a promoter containing six tal-1 E box-1 type consensus elements. Rb⁻ SAOS-2 cells were transfected with one or more of six DNA constructs which express β-galactosidase, luciferase, tal-1, E12, Lmo2, and pRb. The amounts of each DNA construct used in the transfections and the procedures for transient transfection by the calcium phosphate method, cell lysis, and β-galactosidase and luciferase assays are detailed in Materials and Methods. RLU, relative light units. The histograms represent means ± standard errors of the means (SEM) of luciferase activity from 10 independent experiments. ○ ○, P < 0.01 by Student's t test. (B) pRb enhances transcriptional activation by a tal-1–E12–Lmo2 complex in a dose-dependent manner. The amounts of each DNA construct used in the transfections are detailed in Materials and Methods. The histograms represent means ± SEM of luciferase activity from three separate transfections.

In our preliminary experiments, the addition of Ldb1 to the tal-1–E12–Lmo2 complex only slightly modified the effect of pRb on the transcriptional activity of the E1b promoter containing six E box-1 type binding sites (data not shown).

The synthesis of tal-1, E12, Lmo2, Ldb1, and pRb in a transiently transfected SAOS-2 cell line was confirmed by IP or Western blot analysis.

These results strongly suggest that the presence of pRb within a complex comprising tal-1, E12, and Lmo2 or tal-1, E12, Lmo2, and Ldb1 is not merely a structural association but can exert a strong positive effect on transcriptional activity.

The pRb–tal-1–E12–Lmo2–Ldb1 complex negatively regulates the activity of the c-kit promoter in a transient transfection assay.

To explore the functional significance of the pentameric complex in an E context, the effect of pRb, tal-1, E12, Lmo2, and Ldb1 on the transcriptional activity of a putative direct Tal-1 target gene, the c-kit receptor, was examined in transiently transfected Rb⁺ hematopoietic TF1 cells, a multipotent leukemic cell line grossly comparable to CFU-GEMM, which can be induced in more mature E cells or macrophage-like cells by appropriate stimuli. An artificial reporter plasmid containing the 0.5-kb c-kit proximal promoter region with two inverted E box-2 type motifs (see details in Materials and Methods) was used. Similar results were obtained with TF1 cells using the Dual-luciferase reporter assay system with the reporter vector pGL3-c-kit (Fig. 7A) or the luciferase and β-galactosidase assays with the reporter vector pXP2-c-kit (data not shown).

FIG. 7.

(A) The pRb–tal-1–E12–Lmo2–Ldb1 complex negatively regulates the activity of the c-kit proximal promoter in transiently transfected Rb⁺ hematopoietic TF1 cells. Rb⁺ TF1 cells were transfected with one or more of six DNA constructs which express firefly luciferase, tal-1, E12, Lmo2, Ldb1, and pRb together with a Renilla luciferase control reporter (pRL-TK). The amounts of each DNA construct used in the transfection, the procedure for transfection protocols, and the Dual-luciferase assay are detailed in Materials and Methods. The histograms represent means ± standard errors of the means (SEM) of firefly luciferase activity normalized with respect to a Renilla luciferase control report from three independent experiments. ○ ○, P ≤ 0.01 by Student's t test. (B) The pRb–tal-1–E12–Lmo2–Ldb1 complex negatively regulates transcription driven by a 0.5-kb c-kit promoter containing the inverted E box-2 type motifs. Rb⁻ SAOS-2 cells were transfected with one or more of seven DNA constructs which express α-galactosidase, luciferase, tal-1, E12, Lmo2, Ldb1, pRb, p107, and p130. The amounts of each DNA construct used in the transfection and the procedures for transient transfection by the calcium phosphate method, cell lysis, and β-galactosidase and luciferase assays are detailed in Materials and Methods. The histogram represents means ± SEM of luciferase activity from three independent experiments. ○, P ≤ 0.05; ○ ○, P ≤ 0.01 by Student's t test.

As shown in Fig. 7A, no significant difference was observed with either the reporter vector pGL3-c-kit alone or pGL3-c-kit and tal-1 (lanes 2 and 3, respectively). Indeed, a positive effect of Tal-1 on the transcriptional activity of the c-kit proximal promoter has been observed by Krosl et al. in TF1 cells expressing a dominant-negative (dn) tal-1 (29).

Interestingly, coexpression of tal-1, E12, Lmo2, and Ldb1 decreased luciferase activity of the reporter plasmid by threefold (lane 4). However, coexpression of pRb with tal-1, E12, Lmo2, and Ldb1 did not significantly decrease the luciferase activity compared to that due to the tetramer, an effect most probably due to the presence of endogenous pRb in the TF1 cell line (lane 5).

Furthermore, to better clarify the Rb role in the tal-1 pentameric complex, we extended the functional experiments in an Rb null cell line. As shown in Fig. 7B, in transiently transfected Rb⁻ and c-kit⁺ SAOS-2 cells, no significant difference in luciferase activities was observed with either the reporter vector pXP2-c-kit alone or pXP2-c-kit and Rb (lanes 2 and 3, respectively). Coexpression of tal-1, E12, Lmo2, and Ldb1 decreased the transcriptional activity of the c-kit promoter by twofold (lane 4). Interestingly, coexpression of pRb with the tal-1–E12–Lmo2–Ldb1 complex decreased luciferase activity by an additional 40% compared with pXP2-c-kit together with this complex (lane 5). On the other hand, the Rb-like proteins p107 and p130 were not able to downmodulate the luciferase activity of the tal-1–E12–Lmo2–Ldb1 complex on this “natural” promoter region. Similar results were obtained with the coexpression of pRb with tal-1, E12, and Lmo2 and without Ldb1 (data not shown).

In conclusion, these data suggest that the cis regulatory acting sequence on the c-kit promoter can be negatively regulated by the pRb–Tal-1–E12–Lmo2–Ldb1 complex both in hematopoietic (TF1) and nonhematopoietic (SAOS-2) cell lines.

Gel shift analysis of parental TF1 cells and an HPC unilineage culture at different stages of erythropoiesis using the c-kit E box double-site oligonucleotide.

The suggestion that the tal-1–E2A–Lmo2–Ldb1–pRb complex negatively regulates the c-kit promoter is in contrast with recent data (29) suggesting that tal-1 upmodulates c-kit expression in transiently transfected TF1 cells expressing a dn tal-1. This discrepancy could be due to the different developmental stage, i.e., tal-1 may be in a different multiprotein complex in TF-1 cells or at an early stage of erythropoiesis (day 3 or 4) compared with maturing erythroblasts (days 8 to 12). For this reason, EMSA was performed on TF1 nuclear extract and the results were compared with those from day 4 (CD34⁺ erythroblast cells) and day 9 (basophilic-orthochromatophilic-polychromatophilic erythroblast) unilineage E cultures using a radiolabeled oligonucleotide probe containing the c-kit E-box double site.

As shown in Fig. 8A, the c-kit oligonucleotide generated diverse protein DNA complexes. Interestingly, the lowest-mobility complex comprising tal-1, E2A, Lmo2, Ldb1, and pRb is detected only with nuclear extracts from cells in a day 9 E culture with a cell number ranging from 2 × 10⁵ to 8 × 10⁵ (lanes 1 to 3), not at day 4 (8 × 10⁵ cells; lane 4) or from TF1 cells (8 × 10⁵ cells; lane 5), thus suggesting a stage-specific role of the pentamer complex in E differentiation and maturation. On the other hand, the two higher-migrating complexes have molecular weights in a day 4 E culture (lane 4) similar to those in TF1 cells (lane 5). Furthermore, the two lowest-mobility complexes generated with the c-kit oligonucleotide probe in TF1 cells were not partially abrogated by incubation with anti-tal-1 antiserum (Fig. 8B, lane 3) or supershifted with anti-pRb MAb (data not shown), similar to the corresponding low-molecular-weight complex in a day 8 E culture (cf. Fig. 5B, lane 7, and Fig. 5D, lane 5), suggesting that these complexes did not contain the corresponding protein.

FIG. 8.

(A) EMSA in parental TF1 cells and in and HPC unilineage culture at different stages of erythropoiesis using a c-kit E box double-site oligonucleotide. A double-stranded ³²P-labeled c-kit E box double-site oligonucleotide was incubated with nuclear extracts from a day 9 (lanes 1 through 3) or day 4 (lane 4) unilineage E culture and parental TF1 cells (lane 5). The cell numbers used for preparing nuclear extracts are shown. Arrow, lowest-mobility complex comprising tal-1, E2A, Lmo2, Ldb1, and pRb. (B) The mobility complex generated by a c-kit oligonucleotide probe in TF1 cells does not contain tal-1. The two lowest-mobility complexes generated with the c-kit oligonucleotide probe in TF1 cells were not partially abrogated by incubation with anti-tal-1 antiserum (lane 3), suggesting that these complexes did not contain the corresponding protein.

In conclusion our results provide evidence that tal-1 is in a different protein complex in the early, versus the late, stage of E differentiation and maturation.

pRb domain requirements for interaction with tal-1–E12–Lmo2–Ldb1 complex.

To define the precise domains by which pRb binds and modulates the tal-1–E12–Lmo2–Ldb1 complex, pRb mutants (see details in Materials and Methods) were coexpressed with tal-1, E12, Lmo2, and Ldb1 in pRb⁻ SAOS cells. Briefly, the most efficient pRb mutant mimicking the inhibitory effect of wt pRb is the pCMV A/B mutant, which contains the A/B pocket and therefore binds LXCXE but which does not exhibit high-affinity E2F binding or bind c-Abl (Fig. 9, lanes 4 and 6).

FIG. 9.

pRb domain requirement for interaction with a tal-1–E12–Lmo2–Ldb1 complex. To define the precise domain by which pRb binds and modulates the tal-1–E12–Lmo2–Ldb1 complex, wt pRb (lane 4) and mutants (pCMV SEΔ [lane 5] and pCMV A/B [lane 6]) were coexpressed with tal-1, E12, Lmo2, and Ldb1 in pRb⁻ SAOS-2 cells. The amounts of the DNA constructs used in the transfection and the procedure for transient transfection are detailed in Materials and Methods. The histograms represent means ± standard errors of the means of luciferase activity from four independent experiments. ○ ○, P ≤ 0.01 between bars 3 and 2; ○, P ≤ 0.05 between bars 3 and 4 or bars 3 and 6 by Student's t test.

Conversely, pCMV SEΔ, a C-terminal mutant that still binds to c-Abl, does not have an additional inhibitory effect on the tal-1–E12–Lmo2–Ldb1 complex (lanes 3 and 5).

Thus, transcriptional studies with mutated forms of pRb in transiently transfected Rb⁻ SAOS-2 cells suggest that the A/B region of pRb containing a binding site for the LXCXE motif is necessary and sufficient for interaction with the tal-1–E12–Lmo2–Ldb1 complex and for enhancing specific transcriptional inhibition.

In vivo effect of the pRb–tal-1–E12–Lmo2–Ldb1 complex: downmodulation of the endogenous c-kit receptor.

Recent studies indicated that c-kit receptors are weakly expressed by SAOS-2 osteoblast cells and that they may be implicated in cell contact-dependent interaction among specialized bone cell populations (18). To better assess the biological function of the tal-1–E2A–Lmo2–Ldb1–pRb complex, we have analyzed the expression of the endogenous c-kit receptor by Western blotting in transiently transfected SAOS-2 cells with the tetramer and pentamer complexes. Two representative experiments (no. 1 and 2) are shown in Fig. 10. Coexpression of tal-1, E12, Lmo2, and Ldb1 downmodulated the expression of the endogenous c-kit receptor (no. 1, lane 6; no. 2, lane 4). As described above for the luciferase assay experiments, coexpression of pRb with the tal-1–E12–Lmo2–Ldb1 complex additionally decreased the amount of c-kit, compared with expression of this complex alone (no. 1, lane 5; no. 2, lane 3). As controls, the pCMV vector alone (no. 1, lane 4; no. 2, lane 2) and the Rb protein (no. 1, lane 7; no. 2, lane 5) moderately downmodulated the endogenous c-kit, compared to results for the untransfected SAOS-2 cells (no. 1, lane 3; no. 2, lane 1). Altogether, these data suggest that a pRb–tal-1–E12–Lmo2–Ldb1 multiprotein transcription complex has an important role in the regulation of c-kit receptor expression.

FIG. 10.

Expression of the endogenous c-kit receptor in transiently transfected SAOS-2 cells by Western blot analysis. Two representative experiments (Expt. 1 and 2) are shown. (A) Coexpression of tal-1, E12, Lmo2, and Ldb1 downmodulates the activity of the endogenous c-kit receptor (expt. 1, lane 6; expt. 2, lane 4). The addition of pRb to this complex additionally decreased the amount of c-kit compared with results for the complex alone (expt. 1, lane 5; expt. 2, lane 3). Controls were SAOS-2 cells untransfected (expt. 1, lane 3; expt. 2, lane 1) or transfected with pCMV vector alone (expt. 1, lane 4; expt. 2, lane 2) or the Rb protein (expt. 1, lane 7; expt. 2, lane 5). Jurkat cells or the TF1 cell line was used as negative or positive controls for Western blot analysis (expt. 1, lanes 1 and 2; expt. 2, lanes 7 and 6, respectively). (B) β-Actin was used as an internal standard to control for cell lysate loading.

DISCUSSION

Recent studies indicate a complex network of biochemical and functional interactions involving tal-1, E2A, Lmo2, Ldb1, and possibly other nuclear protein(s) in MEL cells (60) and murine fetal liver erythroblasts (58). However, the biochemical interaction of these proteins in normal adult erythropoiesis, particularly in humans, has not been elucidated. More importantly, the functional significance of these transcriptional complexes is unknown, particularly as related to their action on target genes. Finally, no information on the biochemical and functional interaction of Rb with these protein complexes is available.

Our studies provide novel information on several of these aspects. (i) pRb biochemically interacts with the tal-1–E2A–Lmo2–Ldb1 tetramer complex in normal human adult E precursors, specifically at the CFU-E–proerythroblast stage and then in early maturing erythroblasts. (ii) Multiprotein complex formation involving tal-1–E2A–Lmo2–Ldb1–pRb can readily occur on the E box-1 type motif; similarly, the pentamer complex can assemble on two inverted E box-2 type motifs in the human c-kit promoter, specifically in maturing erythroblasts but not in early undifferentiated E progenitors. (iii) Transcriptional assays in a hematological context (i.e., transiently transfected Rb⁺ TF1 cells) indicate that the tetramer and pentamer negatively regulate the cis-acting regulatory sequence on the c-kit promoter. (iv) Transcriptional assays with transiently transfected Rb⁻ SAOS-2 cells indicate that the tal-1–E2A–Lmo2 and tal-1–E2A–Lmo2–Ldb1 complexes require the presence of pRb to activate a promoter containing a concatemer of six E box-1 type motifs and that, conversely, the tal-1–E2A–Lmo2–Ldb1 tetramer negatively regulates a c-kit promoter region containing two inverted E box-2 type elements and, more importantly, inhibits expression of the endogenous c-kit. In both cases, the presence of pRb significantly potentiates the inhibitory activity of the tetramer.

These studies shed light on the mechanism of action of tal-1 and Rb in erythropoiesis. Potential key target genes of tal-1 in erythropoiesis include the erythropoietin receptor (EpoR) (37), GATA-1 (68), and c-kit (29). The c-kit receptor has an important proliferative function in early hematopoiesis, which is predominantly restricted to the HSC and HPC compartments, particularly at the BFU-E-to-CFU-E transition (reviewed in reference 34). In most lineages, particularly the E lineage, c-kit is downmodulated after early commitment (26). In advanced erythropoiesis, i.e., from CFU-E through erythroblasts, the gradual decline of c-kit expression (16) contrasts with peak and then sustained expression of Rb and tal-1 (11, 12). Studies of the human c-kit promoter (56, 64) revealed that (i) c-kit expression is controlled at the transcriptional level and (ii) the regulation of transcription is complex and involves several activators and repressors (the cis-acting sequences in the promoter include putative binding sites for Sp1, AP-2, bHLH, Ets-like proteins, GATA-1, and c-Myb). Myb and Ets proteins may act cooperatively as positive c-kit regulators (44). In addition, selective Sp1 binding is critical for c-kit core promoter activity (41).

We have identified an ∼0.5-kb human c-kit promoter region containing two inverted E box-2 type motifs. A sequence comparison with the mouse c-kit promoter shows that these two motifs are highly conserved (52), suggesting an important functional role for this region. This promoter region binds to the tal-1–E2A–Lmo2–Ldb1–pRb pentamer; more importantly, it is negatively regulated by this complex, as indicated by a transient transcriptional assay of both hematopoietic (Rb⁺ TF1) and nonhematopoietic (Rb⁻ SAOS) cells. Interestingly, the pentamer negatively modulates the endogenous c-kit in pRb⁻ SAOS-2 cells.

Taken together, the present studies on c-kit, tal-1, and Rb expression in erythropoiesis (11, 12, 16) and c-kit promoter modulation in SAOS-2 cells suggest that the tal-1–E2A–Lmo2–Ldb1–pRb complex may play a key role in downmodulation of c-kit expression in maturing erythroblasts.

Although two E-box sites on DNA are required for optimal binding of the tal-1–E2A–Lmo2–Ldb1–pRB multiprotein complex, our results showed that the pentameric complex also binds, to a lesser extent, a single E-box site located on the c-kit promoter. Similar results have been obtained by Visvader et al. (58). Orkin's group showed that a multiprotein complex composed of at least tal-1, E2A, Lmo2, and Ldb1 (lacking GATA-1) can assemble on a single consensus tal-1 binding site.

Recent data (29) indicate that tal-1 upmodulates c-kit expression in transiently transfected TF1 cells expressing a dn tal-1. However, the apparent discrepancy with our results may be reconciled in terms of the different cell context. Indeed, the tal-1–Rb pentamer complex binds to the two inverted E box-2 type motifs in maturing erythroblasts but not in TF1 cells and early erythropoiesis, suggesting that tal-1 is in a different DNA-binding complex in an early stage than in a late stage of E differentiation and maturation (see Results).

Altogether, the hypothesis that the tal-1–E2A heterodimer, assembled in a multiprotein complex, may positively or negatively regulate key E genes in relation to the different developmental stage may be considered. Depending on the tal-1–E2A transcriptional partners, the heterodimer might act positively or negatively even on the same gene (e.g., c-kit) at different stages of erythropoiesis. In particular, tal-1 upmodulates c-kit transcription in early undifferentiated hematopoietic cells (29); while complexed with Lmo2–Ldb1–pRb it downmodulates c-kit transcription in erythroblasts (our results). Furthermore, depending on the target E box and adjacent sequence, the tal-1–E2A heterodimer in complex with Lmo2–Ldb1–pRb exerts either a stimulatory effect (i.e., on a concatermerized E box-1 type motif) or an inhibitory action (i.e., on the two inverted E box-2 type motifs in ∼0.5 kb of the c-kit promoter in erythroblasts). In conclusion, our studies suggest a dynamic change of tal-1 transcription factor complexes during E differentiation according to the “cocktail party” model (50).

The role of pRb in erythropoiesis deserves discussion. As previously mentioned, the present studies relate to previous observations on Rb function in knockout mice and other model systems. E development is fully inhibited in the Rb^−/− embryo via a differentiation blockade at the CFU-E level (9, 24, 33), in agreement with our results on human adult erythropoiesis (11) (present observations) and studies on MEL cells (45). Analysis of chimeric mice partially composed of Rb^−/− cells demonstrates an unexpectedly widespread contribution of Rb^−/− transplanted fetal liver cells to maturing erythroblasts (35, 63). However, long-term effects were not monitored due to premature death caused by metastatic tumors. In this regard, long-term transplantation studies indicate that Rb^−/− transplanted fetal liver progenitor cells give rise to erythroblasts with defective maturation (23), in line with the studies on the Rb^−/− embryo (9, 24, 33) and in vitro erythropoiesis (11) (present results). Furthermore, CFU-E assays in vivo and in vitro of cells from recipients of Rb^−/− cells from sibling mice demonstrate an increasing proliferation of Rb^−/− erythrocytes (23).

In the E lineage pRb expression peaks at the CFU-E/proerythroblast level, when proliferation is exponential and the E differentiation program becomes fully expressed (11). In this developmental stage, elevated pRb may counterbalance the seemingly elevated E2F-1 levels, thus impeding apoptosis (15, 48). Our observations suggest a further unexplored function of pRb: in addition to controlling the accumulation of E2F, pRb may regulate the activity of the tal-1–E2A–Lmo2–Ldb1 TF complex at the early-to-intermediate erythroblast stage.

The results suggest that pRb acts as a protein-assembling multicomponent TF according to the “matchmaker” model (61). The precise domains or other conformational changes by which pRb binds and activates this complex have been investigated; results for pCMV-SEΔ and pCMV-A/B pRb mutant transfection in SAOS cells suggest that the A/B region of pRb, which contains a binding site for the LXCXE motif, is necessary and sufficient for interaction with the tal-1–E2A–Lmo2–Ldb1 complex and for enhancement of specific transcription inhibition.

Altogether, we suggest that pRb may (i) participate in E differentiation, modulating the transcriptional rate of c-kit and hypothetically other tal-1 target genes, and (ii) regulate proliferative and differentiative events crucial in erythropoiesis by linkage of cell cycle and transcriptional machinery.

In line with the model proposed herein, recent studies indicate a differentiative role for Rb in other cell systems (51). Thus, pRb may be involved in myelopoiesis, as suggested by its binding to the NF–IL-6 TF during granulocytic-monocytic differentiation of the U937 cell line (7). The Rb function in the muscle system is indicated by the biochemical and functional interaction with the bHLH MyoD in skeletal muscle cell maturation (19, 32) and the failure of myogenesis in transgenic mice that express low levels of pRb (65). Additionally, pRb positively regulates the adipogenesis differentiation program via interaction with the C/EBPs family of TFs crucial for adipogenesis (8). Finally, pRb favors terminal differentiation and blocks apoptosis in the nervous system, as shown by defects in cell maturation and apoptosis associated with the Rb^−/− phenotype (32).

Moreover, it is noteworthy that Lmo2 does not modulate per se transcription by the tal-1–E2A heterodimer but is strictly required for the potentiating effect of Rb on the transcriptional activity of the tal-1–E2A complex. It has been suggested that Lmo2 may act as a physical bridge for a bHLH protein(s) (59). In our experiments, Lmo2 may bridge tal-1–E2A to pRb to form a multicomponent transcriptional complex. It is also possible that Lmo2 might interact with pRb through other pRb-binding proteins, e.g., RBP2 (36).

In conclusion, these results suggest that interaction of the tal-1–E2A–Lmo2–Ldb1 complex with pRb on E box sequences in the c-kit promoter and possibly other key E genes may represent a fundamental regulatory mechanism underlying erythropoiesis.