Translational control of SCL-isoform expression in hematopoietic lineage choice

Abstract

We investigated the translational regulation of SCL protein expression and its role in hematopoietic lineage choice. We show that the expression of different SCL protein isoforms is regulated by signal transduction pathways that modulate translation initiation factor (eIF) function. A conserved small upstream open reading frame (uORF) in SCL transcripts acts as a cis-regulatory element for isoform expression. At the onset of erythroid differentiation, truncated SCL protein isoforms arise by alternative translation initiation and favor the erythroid lineage. In comparison, full-length SCL proteins are more efficient at enhancing the megakaryocyte lineage. Together, our studies unravel translational control as a novel mechanism regulating hematopoietic outcome.

Figure 1

Translational controlled differential expression of SCL protein isoforms during erythroid differentiation. (A) Erythroid differentiation in HD3 erythroblasts was induced by temperature shift (37°C–42°C for 12 h) to inactivate ts-v-erbB. (Left) Rapamycin was added 6 h after temperature shift where indicated. (Right) Expression of individual SCL isoforms from sequential translation initiation codons revealed by 5′ deletion of chicken SCL cDNA (termed A–D) in COS-1 cells. SCL-isoform expression was quantified by densitometry of X-ray films and normalized values are depicted below. For immunodetection, specific antiserum against C terminus of chicken SCL was used. (B) Human TF-1 or HEL cells were treated for 4–48 h with Epo or DMSO, respectively, to induce erythroid differentiation. 2-Amino purine (2AP) was added where indicated. Immunodetection of SCL in total cell extracts of TF-1 (data shown for 0 and 4 h postinduction) or HEL cells using a human SCL C terminus-specific antiserum is shown. In HEL cells, the SCL-D isoform was obscured by a nonspecific band and was therefore omitted. (C) Transiently transfected COS-1 cells with HA-tagged human SCL cDNA (left panel), and N-terminal-SCL deletion constructs (A–D, right panel). Cells were treated with rapamycin (Rap), or mock (− −) treated for 24 h where indicated. For immunodetection of SCL, an HA-immunotag-specific antiserum was used. SCL protein bands (▸), an alternative CUG (▹), and nonspecific/unidentified bands (*) are indicated. (D) Representation of the conserved vertebrate scl structure and translation initiation sites. Numbering of exons follows the nomenclature for the human scl gene. The SCL coding region is depicted in gray, the upstream open reading frame (uORF) is depicted in black. Potential translation initiation AUG-sites are designated A–D for SCL and Φ for the uORF. Shading in the initiation sites comparison indicates critical nucleotides at position −3 and +4 according to the optimal Kozak translation initiation consensus sequence (Kozak 1991).

Figure 2

The SCL protein isoform ratio is modulated by translation initiation factor activity. (A) HA-tagged SCL was cotransfected with PKRΔ6 (to activate eIF2α), eIF4E, or control empty vector (−) into HEK293A. SCL expression was analyzed by immunoblotting using a HA-epitope-specific antiserum. The schematic representations of the SCL mRNA on the left indicate SCL initiation sites in relation to the protein bands. An alternative CUG initiation codon is depicted with an arrowhead (▹). (B) HD3 erythroblasts were infected with retrovirus encoding PKRΔ6, eIF4E, or a control vector (−) and endogenous SCL was examined by immunoblotting using chicken SCL-C terminus-specific antiserum. (C) HEL cells were infected with PKRΔ6 or control virus (−) and endogenous SCL expression was examined by immunoblotting using human SCL-C terminus-specific antiserum. (Right) Expression of eIF4E transgene was controlled by immunoblotting. eIF2α-specific and eIF2α phosphorylation-specific antibodies (eIF2α-P) were used to determine the effect of PKRΔ6 on eIF2α. Because the antibody raised against human eIF2α fails to detect the chicken homolog, a nonspecific band (*) detected by the eIF2α-P antibody serves as an internal control for the loading of HD3 extracts. Northern blots of polyA⁺ RNA revealed no alterations in scl RNA splicing pattern between control (−) and cells expressing PKRΔ6 or eIF2α.

Figure 3

The uORF of SCL controls translation initiation at alternative initiation sites. COS-1 cells were transiently transfected with SCL constructs and analyzed by immunoblotting using HA-epitope-specific antiserum. (A) The wild-type uORF initiation site (wtΦ) was compared to an optimized initiation context (Φ⁺), a weakened context (Φ⁻), or a noninitiation site (ΔΦ). (B) Mutation of alternative translation initiation sites affects SCL-isoform expression. Null (Δ) mutations of the individual initiation sites (A, B, or C) are shown. X indicates an additional initiation site engineered between sites B and C with the uORF initiation codon in wild-type context (X) or removed (XΔΦ). Schematic representations of the SCL mRNA with initiation sites indicated in relation to the protein bands is shown at the left. An alternative CUG initiation codon is depicted with an arrowhead (▹).

Figure 4

Functional differences between SCL protein isoforms in hematopoietic lineage outcome. (A) HEL cells were infected with the empty pMSCVpuro vector (−), or vectors encoding the SCL-A or SCL-D isoform. After puromycin selection cells were analyzed for expression of the megakaryocytic surface marker CD41a by flow cytometry. Data shown are the mean ± S.D. of two independent infections with analysis performed in duplicates. All HEL cells expressed similar amounts of recombinant proteins (data not shown). (B) Mouse bone marrow cells were infected with the empty pMSCV vector (−), or vectors encoding SCL-A (p42) or SCL-D (p22) isoforms. Recombinant SCL was detected in nuclear extracts by immunoblotting with a monoclonal antibody against human SCL. The blot was stripped and reprobed with an anti-E12 antibody as a control. (C) Increase of megakaryocyte colonies by full-length SCL-A. Mouse bone marrow cells were infected with the indicated retroviral constructs, antibiotic (G418)-selected, and plated in methylcellulose culture under selective pressure. Colonies were scored on day 9 of culture. Three independent experiments performed in duplicates are shown for each: control (open circles), SCL-A (gray circles), and SCL-D (black circles); the mean is indicated by a bar. The differences between control, SCL-A, and SCL-D are significant (p < 0.05). (D) Increased numbers of erythroid progenitors in multipotent colonies by SCL-D. Clonal analysis of erythroid development from multipotent progenitors. Mouse bone marrow cells were infected with the indicated retroviral constructs, antibiotic (G418)-selected, and plated in methylcellulose culture supplemented with erythropoietin. On day 10, pluripotent colonies were individually picked and assayed for their contents in erythroid progenitors (CFU-Es). The pluripotent colonies were ranked according to their capacities to give rise to increasing numbers of CFU-Es as shown. Note that SCL-D colonies show the highest incident of CFU-Es. Data shown are typical of two independent experiments.

Figure 5

Model of the SCL uORF as a ribosome relay. Translation initiation at site A occurs by leaky-scanning over the uORF initiation site Φ (path 1, blue). Initiation at the uORF relays ribosomes over the initiation site A. Subsequent reinitiation occurs at internal initiation sites B, C, and D (path 2, red, broken arrows). Efficiency of initiation at Φ and efficiency of reinitiation at the downstream sites are regulated by translation initiation factors. Accordingly, the activities of eIFs in conjunction with the uORF determine initiation site preference and thus the expression of various SCL protein isoforms.