PU.1 Oncogene Autoregulation Loop

Home , Yvelines

Oncogene (2010) 29, 2807–2816 & 2010 Macmillan Publishers Limited All rights reserved 0950-9232/10 $32.00 www.nature.com/onc ORIGINAL ARTICLE Subtle distinct regulations of late erythroid molecular events by PI3K/ AKT-mediated activation of Spi-1/PU.1 oncogene autoregulation loop

O Breig, O The´oleyre, A Douablin and F Baklouti mRNA Metabolism in Normal and Pathological Cells; CGMC, CNRS, Universite´ Lyon 1, Villeurbanne, France

Spi-1/PU.1 oncogene is downregulated as proerythro- balance between self-renewal and differentiation in blasts undergo terminal differentiation. Insertion of the hematopoietic progenitor cells, and functions in a Friend virus upstream of the Spi-1/PU.1 locus leads to the concentration-dependent manner to promote differen- constitutive upregulation of Spi-1/PU.1, and a subsequent tiation of the lymphoid and myeloid lineages (Back block in the differentiation of the affected erythroblasts. et al., 2004; Fisher et al., 2004 and references therein). We have shown that sustained overexpression of Spi-1/ Furthermore, Spi-1/PU.1 is expressed at low levels in PU.1 also inhibits the erythroid splicing of protein 4.1R erythroid progenitor cells and subsequently downregu- exon 16, irrespective of chemical induction of differentia- lated on terminal differentiation. It acts to maintain the tion. Here, we show a positive feedback loop that couples self-renewal capacity of immature erythroid precursors constitutive phosphatidylinositol 3-kinase (PI3K)/protein and to prevent differentiation in the erythroid lineage kinase B (AKT) signaling to high expression of Spi-1/ (Back et al., 2004; Fisher et al., 2004). PU.1 in Friend erythroleukemia cells. Inhibition of PI3K/ Spi-1/PU.1 was initially discovered as a proviral AKT results in Spi-1/PU.1 downregulation in a stepwise integration site in Friend virus-induced erythroleukemia manner and induces cell differentiation. Chromatin (reviewed in Moreau-Gachelin, 2008). Friend virus immunoprecipitation assays further supported the positive complex consists of two viruses, a helper replication- autoregulatory effect of Spi-1/PU.1. Mutational analysis competent Friend murine leukemia virus (F-MuLV) indicated that Ser41, but not Ser148, is necessary for Spi- and a replication-defective spleen focus-forming virus 1/PU.1-mediated repression of hemoglobin expression, (SFFV). Friend SFFV carries a unique env gene whereas both Ser residues are required for Spi-1/PU.1 encoding a truncated form of a retroviral envelope inhibition of the erythroid splicing event. We further show glycoprotein (gp55), which is responsible for its patho- that inhibition of the erythroid transcriptional and splicing genicity (reviewed in Moreau-Gachelin, 2008). In the events are strictly dependent on distinct Spi-1/PU.1 first stage of the disease, gp55 binds and activates the phosphorylation modifications rather than Spi-1/PU.1 ex- erythropoietin receptor (Epo-R) and a short form of the pression level per se. Our data further support the fact that receptor tyrosine kinase, resulting in constitutive activa- Spi-1/PU.1 inhibits 4.1R erythroid splicing through two tion of signal transducing molecules and the develop- different pathways, and bring new insights into the extra- ment of Epo-independent erythroid hyperplasia and cellular signal impact triggered by erythropoietin on late polycythemia, due to the polyclonal expansion and erythroid regulatory program, including pre-mRNA splicing. differentiation of erythroid cells in the absence of Epo. Oncogene (2010) 29, 2807–2816; doi:10.1038/onc.2010.29; The second stage of the disease results from the published online 1 March 2010 outgrowth of Friend SFFV-infected erythroid cells that have become transformed because of integration of the Keywords: erythroleukemia; mRNA splicing; cell virus into the Spi-1 locus. This event triggers aberrant differentiation; cell signaling overexpression of the Spi-1/PU.1 protein in erythroid cells and causes a block in their differentiation and the outgrowth of transformed mouse erythroleukemia Introduction (MEL) cells (Moreau-Gachelin et al., 1989; Paul et al., 1991; Schuetze et al., 1992). Treatment of the cells with Spi-1/PU.1 is a member of the ets family of transcription polar compounds, such as dimethylsulfoxide (DMSO) factors that is expressed specifically in hematopoietic and hexamethylene bisacetamide, causes them to reenter tissues. spi-1/pu.1 gene disruption results in lethality a differentiation program best characterized by rapid (Scott et al., 1994). It is essential for controlling the downregulation of Spi-1/PU.1 and the concomitent accumulation of hemoglobin and other membrane Correspondence: Dr F Baklouti, mRNA Metabolism in Normal and erythrocyte-specific proteins (reviewed in Moreau- Pathological Cells, CGMC; CNRS UMR 5534, Universite´Lyon 1, Gachelin, 2008). Consistently, forced expression of Baˆt. Gregor Mendel, 16, rue R. Dubois, 69622 Villeurbanne Cedex, Spi-1/PU.1 blocks MEL cell differentiation (Rao et al., France. 1997), and results in inhibition of growth and differ- E-mail: [email protected] Received 28 July 2009; revised 27 December 2009; accepted 15 January entiation and apoptotic death of other erythroleukemia 2010; published online 1 March 2010 cell lines (Yamada et al., 1997). PI3K/AKT induces PU.1 autoregulation loop O Breig et al 2808 Binding of Epo to its receptor triggers the phosphor- short hairpin RNA-mediated downregulation of Spi-1/ ylation and activation of Epo-R-bound Janus kinase 2 PU.1 release the inhibition and lead to exon 16 inclusion (JAK2) tyrosine kinase, resulting in activation of several (Blaybel et al., 2008). downstream signaling pathways that include the phos- For many years, great effort has been made to phatidylinositol 3-kinase (PI3K)/protein kinase B understand how signals received by cells impact (PKB or AKT), the signal transducer and activator of transcription factor activity. In an attempt to connect transcription 5 (STAT5)-Bcl-XL and the extracellular the splicing event and its blocking by upregulated Spi-1/ signal-regulated kinase/mitogen-activated protein ki- PU.1 with the signaling pathways activated in MEL nase (reviewed in Ghaffari et al., 2003). In human cells, we here provide the first direct evidence for an erythroid progenitors, PI3K activation protects the cells Epo-generated signal that modulates an erythroid- from apoptosis and mediates Epo-induced proliferation specific splicing. Our results show that constitutive through downregulation of p25Kip1 expression (Bouscary PI3K/AKT sustains Spi-1/PU.1 autoregulated expres- et al., 2003). Activation of AKT serine threonine kinase of sion that inhibits protein 4.1R exon 16 splicing and PKB family is crucial for Epo-induced erythropoiesis erythroid differentiation, as evidenced by impaired (Myklebust et al., 2002; Bouscary et al., 2003 and globin gene expression. Inhibition of PI3K/AKT signal- references therein). This activation is mediated by PI3K ing blocks Spi-1/PU.1 autoregulation loop, and induces through phosphorylation of Ser473 on AKT. In this cell differentiation and 4.1R erythroid splicing activa- context of Epo signaling, AKT is a major effector tion. However, mutational targeting of specific serine downstream of PI3K, having fundamental roles in the residues of Spi-1/PU.1 led to uncouple the two Spi-1/ regulation of cell cycle, survival and differentiation PU.1 inhibitory effects on 4.1R splicing and globin gene (Bao et al., 1999; Bouscary et al., 2003). Moreover, AKT expression. Taking together, these data show subtle activation has been reported in SFFV-infected erythroleu- differential effects of Spi-1/PU.1 expression on late kemic cells, in which Epo-R is constitutively activated by erythroid-regulated events. the viral gp55 (Nishigaki et al., 2000). In fact, most of the signal transduction pathways induced in erythroid cells by Epo, including the Jak-STAT pathway, are constitutively activated in SFFV-transformed cells in the absence of Epo Results (Nishigaki et al., 2006 and references therein). Proliferation of Epo-independent proerythroblastic cells (called HS2 PI3K/AKT inhibition triggers erythroleukemia cells), derived from transgenic mice overexpressing Spi-1/ cell differentiation in the absence of DMSO PU.1, requires active PI3K/AKT and mitogen-activated Phosphatidylinositol 3-kinase complex is a protein protein kinase pathways (Barnache et al., 2001). Consis- dimer that consists of a p85 regulatory subunit and a tently, Epo initiates a transcriptional program that leads to p110 catalytic subunit. Earlier works have shown that significant change of expression levels in over 580 genes, p85 expression, phosphorylation and activity decline in the majority of which are apparently regulated in a PI3K- Friend cells after DMSO exposure for 96 h (Bavelloni dependent manner (Sivertsen et al., 2006). et al., 2000; Cataldi et al., 2000). Immunoblot analysis of The inclusion of exon 16 in mature protein 4.1R p85 subunit shows a reduction of the phosphorylated mRNA is probably the best-studied example of stage- form of p85 (P-p85) when the cells are treated with specific pre-mRNA splicing that characterizes the late 10–20 mM LY294002 or DMSO (Figure 1a). Activation erythroid development. It has been well documented of PI3K triggers the phosphorylation of its downstream that exon 16 is excluded in early erythroid progenitors target AKT. As shown in Figure 1b, the phosphorylated and included in late erythroid differentiated cells (Chasis AKT (Ser473) decreases on LY294002 treatment in a et al., 1993; Baklouti et al., 1996). This regulated splicing dose-dependent manner, suggesting that LY294002 is event underlies the production of a functional 4.1R; effective on both PI3K autophosphorylation and down- inclusion of the 21 amino acid peptide encoded by exon stream AKT phosphorylation. 16, at the N-terminus of the 10 kDa spectrin/actin- We then tested the effect of PI3K/AKT inhibition on binding domain of 4.1R is indeed essential to promote MEL SFFV cells. Cells were seeded at stationary phase in high mechanical stability and high deformability of the the presence of increasing concentrations of LY294002, mature red cell membrane. We have shown that Epo- within a nontoxic range, and cell differentiation was nonresponsive MEL cells reproduce exon 16-regulated monitored 96 h after exposure to the inhibitor. This splicing on DMSO induction to erythroid differentiation experiment reproducibly revealed that LY294002 triggers from either the endogenous pre-mRNA (Baklouti et al., MEL cell differentiation in a concentration-dependent 1996) or transfected minigene transcript (Deguillien manner (Figure 1c), that is, at a 5 mM concentration, et al., 2001). More recently, we have observed that Epo- B12% of the cells were benzidine positive, whereas at a responsive erythroleukemia cells also efficiently regulate 20 mM concentration, this percentage was over 65%. In the this splicing event when induced in culture with either same sets of experiments, the proportion of differentiating DMSO or Epo (Theóleyre et al., 2004). Exon 16 splicing cells was higher, B78%, when the cells were induced to is tightly correlated with Spi-1/PU.1 expression, that is, differentiate with 1.8% DMSO. upregulated Spi-1/PU.1 in proliferative MEL cells as These results evidence that inhibition of the constitutive well as forced expression of Spi-1/PU.1 inhibit exon 16 PI3K/AKT signaling triggers Friend eythroleukemia cell inclusion (Theóleyre et al., 2004). DMSO-induced or differentiation in the absence of DMSO chemical induction.

Oncogene PI3K/AKT induces PU.1 autoregulation loop O Breig et al 2809 (–) D L10 L20 (–) L10L20 DMSO P-p85 +16 p85 –16 262224 2622 24 2622 24 26 cycles Actin

(–) L5 L10L15 L20 D L10 L20 P-AKT(Ser473) DMSO 1.8%

Actin

MEL SFFV cells 90 80 70 60 Figure 2 PI3K/AKT inhibition promotes 4.1R exon 16 inclusion 50 in Friend erythroleukemia cells. Cells were treated with 10 or 20 mM LY294002 for 4 days, and 4.1R exon 16 splicing pattern was 40 analyzed by semiquantitative RT–PCR. (a) Agarose gel electro- 30 phoresis of RT–PCR products obtained at limited number of cycles 20 (22, 24 and 26 cycles). þ 16 and À16 correspond to PCR fragments containing or missing exon 16, respectively. (À): untreated cells. 10

Benzidine positive cells, % (b) Semiquantitative analysis of exon 16 splicing expressed as a 0 percentage of ( þ 16)/( þ 16) þ (À16) ratio. Data were reproducible L5 L10 L15 L20 DMSO 1.8% through three different sets of experiments. Percentages are shown as a mean±s.d. Cells were treated with 1.8% DMSO, 10 mM Figure 1 PI3K/AKT inhibition results in differentiation of Friend LY294002 (L10) or 20 mM LY294002 (L20). erythroleukemia cells. (a) Immunoblot analysis of p85 subunit of PI3K using an anti-p85 antibody in untreated (À) SFFV erythroleukemia cells or cells treated with DMSO (D), 10 mM LY294002 (L10) or 20 mM LY294002 (L20) for 96 h. Actin constitutive PI3K/AKT signaling results in the activa- immunoblot is used as a loading control. The antibody reveals tion of 4.1R exon 16 splicing, again in an inhibitor dose- both phosphorylated (P-p85) and unphosphorylated (p85) forms dependent manner. However, in keeping with the effect of p85 subunit. (b) Immunoblot analysis of phosphorylated AKT (P-AKT(Ser473)) using an antibody specifically directed against the observed on hemoglobin production, treatment with phosphoepitope. SFFV cells were exposed to increasing concentra- LY294002 inhibitor seems to have a less conspicuous tions (5–20 mM) of LY294002 for 96 h or to 1.8% DMSO. (À): effect on exon 16 inclusion than DMSO. untreated cells. (c) Impact of LY294002 on SFFV cell differentia- Similar data were obtained using Epo-responsive tion. PI3K/AKT signaling was inhibited by adding LY294002 cells (Friend erythroleukemia SFFV SKT6 cells and compound at increasing concentrations (5–20 mM). Cells cultured without LY294002 inhibitor were induced to differentiate using Rauscher cells) (not shown), indicating that PI3K/AKT 1.8% DMSO. Differentiating cells were monitored using benzidine signaling inhibits erythroleukemia cell differentiation, reaction test. Data are the mean of four different experiments regardless of the Epo responsiveness of the cell line. (±s.d.). Collectively, these observations suggest that inhibition of the constitutive PI3K/AKT signaling alone promotes both globin gene transcription and ery- PI3K/AKT signaling inhibits protein 4.1R exon throid-specific splicing of 4.1R exon 16 in erythroleuke- 16 erythroid splicing mia cells, in the absence of DMSO or Epo induction of In a recent work, we have provided evidence that differentiation. induction of differentiation of MEL SFFV cells triggers a transcription event (globin gene activation) and a splicing event (4.1R exon 16 inclusion) through different Inhibition of PI3K/AKT signaling downregulates pathways (Blaybel et al., 2008). These events are Spi-1/PU.1 expression in a stepwise manner probably the best-characterized transcription and spli- Upregulated Spi-1/PU.1 in Friend cells inhibits globin cing events that label the late stage of erythroid gene transcription and blocks 4.1R pre-mRNA ery- differentiation. We asked whether PI3K/AKT signaling throid splicing (Theóleyre et al., 2004). We tested Spi-1/ inhibition impacts both of these regulated events. In new PU.1 expression in cells treated with PI3K/AKT sets of experiments, MEL SFFV cells were either left inhibitor LY294002. Immunoblot analysis of Spi-1/ untreated or treated with 10 or 20 mM LY294002 for 4 PU.1 expression revealed two or three isoforms with days, as described above, and exon 16 splicing was apparent molecular weights ranging within 35–37 kDa, analyzed by semiquantitative reverse transcriptase–PCR that represent different phosphorylated forms of Spi-1/ (RT–PCR). As outlined in Figure 2, inhibition of the PU.1 (Delgado et al., 1994). As shown in Figure 3a, the

Oncogene PI3K/AKT induces PU.1 autoregulation loop O Breig et al 2810 3.5 21 L0 L5L10 L15 L20 LY29, µM Spi antibody P-PU.1 3 control antibody 18 PU.1 2.5 15

Grb2 2 12

0 8 19 52 62 Benzidine, % 1.5 9

MEL SFFV cells 1 6 1.1 0.5 3 PU.1 enrichment/GAPDH 1 0 0 0.9 GAPDH 129k 153k PU.1(+20) Nip7 Figure 4 ChIP analysis of Spi-1/PU.1 autoregulation potential. 0.8 The ets site, acting as an enhancer in position þ 20 within the 0.7 spi-1/pu.1 gene (PU.1( þ 20)) was explored for Spi-1/PU.1 protein binding. DNA–protein complexes were immunoprecipitated using 0.6 either anti-Spi-1/PU.1 antibody or a nonspeciﬁc anti-p27 antibody.

norm. PU.1 mRNA Immunoprecipitated DNA fragments were amplified using primers 0.5 specific for each gene promoter/enhancer. GAPDH, 129k and 153k were used as negative controls. Nip7 served as a positive control for 0.4 Spi-1/PU.1 binding (Juban et al., 2009). Data are expressed as site L0 L5 L10 L15 L20 occupancy normalized against GAPDH background DNA–protein association. Figure 3 Inhibition of PI3K/AKT signaling downregulates Spi-1/ PU.1 expression. MEL cells were cultured in the presence of increasing concentration of LY294002 inhibitor (LY29): 0–20 mM (L0 to L20, respectively). (a) Immunoblot analysis of Spi-1/PU.1 PU.1, and that PI3K/AKT-mediated phosphorylation protein expression. Grb2 immunoblot is used as a loading control. Note that Spi-1/PU.1 phosphorylated forms (P-PU.1) decrease and of Spi-1/PU.1 is needed for its transactivation property, globin production increases in a LY294002-dose-dependent including the transactivation of spi-1/pu.1 gene itself, manner. (b) Real-time RT–PCR analysis of Spi-1/PU.1 mRNA. through a positive feedback loop. mRNA steady-state levels are normalized with respect to actin mRNA used as an internal control. Data are expressed as ratios of normalized Spi-1/PU.1 mRNA for each point to the normalized Spi-1/PU.1 mRNA in absence of the inhibitor. The experiment was Inhibition of PI3K/AKT is insufficient to activate 4.1R performed in triplicate, and data are expressed as means±s.d. It erythroid splicing in cells overexpressing Spi-1/PU.1 clearly shows that LY294002 inhibition of PI3K/AKT signaling We next asked whether PI3K/AKT inhibits cell differ- alters Spi-1/PU.1 mRNA production in a dose-dependent manner. entiation through a stricto sensu Spi-1/PU.1 upregulation or rather through a phosphorylation-activated form unphosphorylated Spi-1/PU.1 appears unaltered up to a of Spi-1/PU.1. To address this issue, wild-type Spi-1/ concentration of 15 mM of LY294002, and subsequently PU.1 was overexpressed in MEL cells, under the control declines at a higher concentration (20 mM). Most of heterologous promoters, and stably transfected clones remarkably, the phosphorylated forms of Spi-1/PU.1 were selected. One clone expressing a low level steadily decreased as the dose of LY294002 increased, (l-PU.1 þ clone) and one clone expressing a high level starting from as low as 5 mM. (h-PU.1 þ clone) of Spi-1/PU.1 were subsequently We asked whether Spi-1/PU.1 downregulation occurs analyzed for their ability to respond to PI3K/AKT at the protein level or rather at the mRNA level. To inhibition. The generated MEL cell clones were exposed address this issue, we examined Spi-1/PU.1 mRNA to DMSO or 20 mM LY294002. Untransfected cells again accumulation during LY294002 inhibition of PI3K/ showed a shut-off of Spi-1/PU.1 on exposure to DMSO AKT signaling, using real-time RT–PCR. This experi- and a concentration-dependent downregulation on ment clearly revealed a decrease in mRNA level that LY294002 exposure (Figure 5a). Constitutive low coincides with the decrease observed at the protein level expression of Spi-1/PU.1 was detected in l-PU.1 þ cells, for the same LY294002 concentrations (Figure 3b). and much stronger Spi-1/PU.1 signals were revealed in Spi-1/PU.1 has been described to activate its own h-PU.1 þ cells, cultured in the same conditions gene in myeloid cells by binding an ets sequence at þ 20 (Figure 5a). Hemoglobin synthesis was impaired in the of the transcription start site (Chen et al., 1995). We two cell lines overexpressing Spi-1/PU.1, even after examined the Spi-1/PU.1 transcription factor occupancy DMSO treatment. Interestingly, PI3K/AKT inhibition at the þ 20 ets site by chromatin immunoprecipitation induced hemoglobin synthesis in l-PU.1 þ cells. Con- (ChIP) assay, using anti-Spi-1/PU.1 antibody. DNA versely, h-PU.1 þ culture exposed to 20 mM LY294002 recovered from the immune complexes was analyzed by produced only few, if any, benzidine-positive cells. In quantitative PCR using primers surrounding the þ 20 total, a converse relationship between the level of site. As shown in Figure 4, the ChIP experiment revealed expressed Spi-1/PU.1 and the percentage of benzidine- that Spi-1/PU.1 occupancy of the þ 20 ets site is more positive cells (72% in untransfected cells, 40% in efficient than that of a nonspecific protein. l-PU.1 þ cells and 3% in h-PU.1 þ cells) is observed, These data suggest that inhibition of the PI3K/AKT suggesting a titration effect between Spi-1/PU.1 quanti- signaling triggers a stepwise downregulation of Spi-1/ tative level and its phosphorylation level (Figure 5a).

Oncogene PI3K/AKT induces PU.1 autoregulation loop O Breig et al 2811 Untransfected Mock l-PU.1+ h-PU.1+

(-) D L10 L20 (-) D L20 (-) D L20 (-) D L20

EGFP EGFP-PU.1 P-PU.1 P-PU.1 PU.1 PU.1 Grb2 Actin Actin 0 812772 07260 0040 0 0 3 Benzidine, %

45 40 35 30 25 (-) D 20 L20 15 10 Exon 16 inclusion, % 5 0 Untransf Mock l-PU.1+ h-PU.1+ Figure 5 Inhibition of hemoglobin synthesis and 4.1R erythroid splicing requires Spi-1/PU.1 phosphorylation. MEL cells were stably transfected with Spi-1/PU.1 recombinant plasmids expressing different levels of Spi-1/PU.1 (see Materials and methods section). One clone expressing low level of Spi-1/PU.1 (l-PU.1 þ ) and one clone expressing high level of Spi-1/PU.1 (h-PU.1 þ ) were investigated for their response to PI3K/AKT signaling inhibition. (a) Immunoblot analysis of Spi-1/PU.1 expression using anti-Spi-1/PU.1 antibody. Grb2 and actin immunoblots are used as controls. Left panels: Untransfected cells were treated with 10 (L10) or 20 (L20) mM LY294002. Again, LY294002 induces cell differentiation in a dose-dependent manner. Mock cells were transfected with EGFP- containing plasmid. EGFP (B27 kDa) was revealed with anti-GFP antibody and the endogenous Spi-1/PU.1 (B34–37 kDa) was analyzed by anti-Spi-1/PU.1 antibody. In mock cells, Spi-1/PU.1 is switched off in a similar manner as in untransfected cells, in the presence of either DMSO or LY294002. Right panel: Spi-1/PU.1 derived from transfected plasmids is expressed in untreated, as well as in DMSO- or LY294002-treated cells. The upper and middle parts correspond to a same membrane probed with anti-Spi-1/PU.1 antibody. They show the expression of EGFP-PU.1 (B64 kDa) and the endogenous Spi-1/PU.1, respectively. However the upper part is a 20-s exposure, whereas the middle part is a 60-s exposure. The percentage of benzidine-positive cells is indicated for one set of experiments. The experiment was reproducible at least through three cell cultures. (b) Semiquantitative analysis of exon 16 splicing. The exon inclusion is enhanced in untransfected cells and in mock cells on exposure to either DMSO or LY294002. In cells overexpressing Spi-1/PU.1, the exon is roughly excluded in all conditions, suggesting that LY294002-mediated inhibition of Spi-1/PU.1 phosphorylation is insufﬁcient to repress exon inclusion.

Moreover, western blot analysis revealed only a partial Distinct effects of phosphorylated Ser residues on inhibition of Spi-1/PU.1 phosphorylation in cells over- Spi-1/PU.1 role in globin gene expression and 4.1R expressing Spi-1/PU.1 (Figure 5a), which argues in favor erythroid splicing of a titration effect on LY294002-mediated PI3K/AKT Spi-1/PU.1 is phosphorylated on multiple serine resi- inhibition. dues in its transactivation and PEST regions (Pongubala Interestingly, exon 16 splicing was also modulated in et al., 1993). In particular, phosphorylation of Spi-1/ these cells; as expected, exon inclusion was inhibited in PU.1 on Ser148 by casein kinase II controls the DMSO-treated cells overexpressing Spi-1/PU.1, but not recruitment of B-cell-speciﬁc factors, such as NF-EM5 in untransfected or mock-transfected cells (Figure 5b). and Pip, on the kE30 and lB enhancers, leading to an However, inhibition of PI3K/AKT induced the activa- increase of the transcriptional activity of Spi-1/PU.1 tion of exon splicing in untransfected and mock-trans- (Pongubala et al., 1993; Eisenbeis et al., 1995). It has fected cells, but not in cells overexpressing Spi-1/PU.1 also been evidenced that AKT stimulation of Spi-1/PU.1 (Figure 5b). is due to phosphorylation within the transactivation Collectively, these experiments provide evidence that domain at amino acid residue Ser41. Mutation of serine Spi-1/PU.1 inhibits both late erythroid transcriptional 41 to alanine (S41A) impaired Spi-1/PU.1 induction by and splicing events in a phosphorylation-dependent AKT signal (Rieske and Pongubala, 2001). manner. Exon 16 remains massively excluded in We asked whether Spi-1/PU.1 requires either or both LY294002-treated cells. We anticipate that Spi-1/PU.1 of these sites for repression of the transcription and the inhibits exon 16 splicing in these conditions through (i) splicing events. Wild-type Spi-1/PU.1 and mutants its unphosphorylated form, expressed from the trans- containing serine to alanine mutations at positions 41 fected Spi-1/PU.1 cDNA, (ii) residual Spi-1/PU.1 and 148 were generated and transfected as enhanced phosphorylated forms, that remain active at 20 mM green ﬂuorescent protein (EGFP) fusions in MEL SFFV LY294002 concentration or/and (iii) a PI3K/AKT- cells. Cells were cultured in the presence or absence of independent phosphorylation of Spi-1/PU.1. DMSO. Immunoblot analysis using anti-Spi-1/PU.1

Oncogene PI3K/AKT induces PU.1 autoregulation loop O Breig et al 2812 Untransf.WT S41A S148A Collectively, these experiments provide evidence that DMSO –+++– – + – distinct Spi-1/PU.1 phosphorylation-mediated modi- EGFP-PU.1 ﬁcations direct speciﬁc inhibitions of late erythroid Anti-PU.1 Endo-PU.1 molecular events.

Actin Benzidine, % 00071 00 040 Discussion MEL SFFV cells 60 50 PI3K/AKT-mediated phosphorylation of Spi-1/PU.1 40 sustains a high level of expression of Spi-1/PU.1 30 through an autoregulatory loop

cells, % 20 The signaling events initiated by the binding of Epo to 10 the Epo-R induce proliferation, survival and differentia- Benzidine positive 0 tion of erythroid progenitors (Jelkmann, 2007). Most WT S41A S148A Friend erythroleukemia cell lines display an SFFV

50 provirus integrated in a reverse orientation relative to 45 the Spi-1/PU.1 promoter. This observation led to the 40 hypothesis that enhancer elements present in the long 35 terminal repeat of SFFV activate Spi-1/PU.1 transcrip- 30 tion; however, direct experimental evidence for this (-DMSO) 25 model is still lacking (Moreau-Gachelin, 2008). In (+DMSO) 20 Friend erythroleukemia, it has been anticipated that in 15 the early preleukemic stage, the Spi-1/PU.1 promoter may Exon 16 inclusion, % 10 be activated in a Spi-1/PU.1-independent manner by an 5 Epo-R/F-gp55 signal. Hereafter, cell clones with inte- 0 grated SFFV provirus in the spi-1/pu.1 gene locus expand untransf Mock WT S41A S148A and display high level of Spi-1/PU.1, involving the Figure 6 Distinct effects of Spi-1/PU.1 phosphorylation modifica- activation of spi-1/pu.1 gene transcription by an enhancer tions on hemoglobin synthesis and 4.1R splicing. MEL cells were element present in the SFFV long terminal repeat stably transfected with wild-type Spi-1/PU.1 (WT) or a mutated Spi-1/PU.1, carrying serine to alanine substitution, either at (Moreau-Gachelin et al., 1989; Okuno et al., 2005). High position 41 (S41A) or 148 (S148A). Cells were cultured in the levels of Spi-1/PU.1 protein are necessary to maintain the presence ( þ ) or absence (À) of DMSO. (a) Immunoblot analysis of transformed phenotype (Afrikanova et al., 2002). Spi-1/PU.1 expression. DMSO exposure induces the downregula- Here, we identified the PI3K/AKT pathway as a tion of the endogenous Spi-1/PU.1, but does not affect the mediator of Spi-1/PU.1 autoregulatory loop that main- expression of the transfected Spi-1/PU.1. Actin immunoblot is used as a loading control. The percentage of benzidine-positive tains Spi-1/PU.1 in a phosphorylated form capable to cells is indicated for one set of experiments. (b) Evaluation of transactivate its own gene. Data gathered from previous hemoglobin-containing cells in MEL cell clones overexpressing WT works and data presented here, altogether suggest that or mutant Spi-1/PU.1. Data are given for three independent activation of Spi-1/PU.1 in MEL cells implies a two-step experiments as means±s.d. (c) Semiquantitative analysis of exon 16 splicing in MEL cells overexpressing WT or mutant Spi-1/PU.1. regulatory mechanism, whereby SFFV gp55 glycoprotein Mock correspond to cells transfected with vector containing only first binds and activates Epo-R, which leads to constitutive the EGFP cDNA. activation of PI3K/AKT signaling. Active PI3K/AKT subsequently mediates Spi-1/PU.1 phosphorylation and spi-1/pu.1 antibody ascertained the high expression of EGFP- therefore triggers a constitutive activation of Spi-1/PU.1 fusion proteins, which is maintained after gene through a positive feedback loop. This autoregula- DMSO exposure, whereas the endogenous Spi-1/PU.1 tory mechanism sustains a high level of expression of Spi- dramatically decreases in all cell clones, on DMSO 1/PU.1, responsible for blocking erythroid differentiation. induction (Figure 6a). Expression of wild-type Spi-1/ Inihibition of PI3K/AKT blocks PI3K/AKT-mediated PU.1 and mutant S148A impaired hemoglobin synth- phosphorylation of Spi-1/PU.1 and subsequently inacti- esis, whereas that of mutant S41A resulted in about 40% vates Spi-1/PU.1-mediated autoregulation. Supportive of of hemoglobin producing cells after DMSO treatment this model is the observation that Epo-R activation of env (Figures 6a and b). These results suggest that phosphor- SFFV by the gene leads to transcriptional upregulation ylation at Ser41 residue, but not at Ser148, is required of Spi-1/PU.1 in the absence of Epo and in the absence of for Spi-1/PU.1 transcriptional activity as a globin gene an integrated SFFV provirus within the Spi-1/PU.1 locus et al. repressor. However, expression of S41A or S148A (Afrikanova , 2002). Spi-1/PU.1 mutants failed to inhibit exon inclusion (Figure 6c). These data show that Spi-1/PU.1 phosphor- Spi-1/PU.1 inhibits hemoglobin synthesis ylation-mediated modifications, both within the trans- in a PI3K/AKT-dependent manner activation domain and the PEST domain, are required Here, we show that specific inhibition of PI3K/AKT is for the inhibitory effect of Spi-1/PU.1 on exon 16 sufficient to induce hemoglobin synthesis. Nonetheless, erythroid splicing. a previous work has shown that LY294002 inhibition of

Oncogene PI3K/AKT induces PU.1 autoregulation loop O Breig et al 2813 PI3K/AKT signaling prevents differentiation of MEL P P P P cells (Bavelloni et al., 2000). This apparent contradiction S41 S41 S41 might be due to the fact that in the majority of our S148 S148 S148 experiments, we used either DMSO or 10–20 mM PU.1 PU.1 PU.1 LY294002, whereas in early works, cells were treated using 0.1–10 mM LY294002 concomitantly with DMSO (Bavelloni et al., 2000; Cataldi et al., 2000). By constitutively overexpressing Spi-1/PU.1 at different levels in MEL cells, we were able to show that Spi-1/ Hb; Ex16 Hb – PU.1 protein level can titrate the inhibitory effect of Figure 7 A model for Spi-1/PU.1 distinct effects on Spi-1/PU.1 LY294002 (Figure 5), and therefore the inhibition autoregulation and on late erythroid events. Data gathered from efficiency of the PI3K/AKT signaling depends on the previous works and new data presented here allow to draw this balance between Spi-1/PU.1 expression level and the regulatory model. Phosphorylation on serine residue 41 (S41) concentration of LY294002 inhibitor. The possibility within the transactivation domain of Spi-1/PU.1 is needed for remains that use of different MEL cell clones, 707 clone Spi-1/PU.1-positive autoregulation and for transcription repression of globin genes (Hb). It is also necessary, but insufficient, for Spi-1/ (Bavelloni et al., 2000) versus 745-A clone (this work), PU.1-mediated inhibition of 4.1R exon 16 splicing (Ex16). On the might account for the apparent conflicted results. other hand, phosphorylation on serine residue 148 (S148) within However, we did observe similar effects of PI3K/AKT the PEST domain of Spi-1/PU.1 is not required for transcription inhibition in SKT6 and Rauscher erythroleukemia cells repression of globin genes and, again, it is necessary, but (data not shown). insufficient, for inhibition of the splicing event. We also found that PI3K/AKT phosphorylation of Spi-1/PU.1 is responsible for its role in blocking High level of Spi-1/PU.1 expressed in proliferative erythroid differentiation of MEL cells and more MEL cells or forced expression of Spi-1/PU.1 in specifically in hemoglobin synthesis. As recapitulated DMSO-induced cells prone to differentiate, result in in Figure 7, the repression activity of Spi-1/PU.1 efficient block of 4.1R exon 16 splicing (Theóleyre et al., requires phosphorylation of the transactivation domain 2004). Results presented here indicated that Spi-1/PU.1 on S41 residue, but not that of the PEST domain on inhibits exon 16 splicing in a phosphorylation-depen- S148. These data are consistent with previous works dent manner. Moreover, phosphorylation of both the addressing the transcriptional regulation of erythroid transactivation and the PEST domains is important for gene targets by the functional antagonism between Spi- Spi-1/PU.1 effect on splicing (Figure 7). These data are 1/PU.1 and GATA-1. This antagonism involves direct consistent with previous observation showing the interactions between the two proteins, in which the importance of both of these domains of Spi-1/PU.1 GATA-1 C-terminal zinc finger, on one side, and both in modulating alternative splicing events within an the DNA-binding and transactivation domains of Spi-1/ exogenous pre-mRNA template (Delva et al., 2004). PU.1, on the other side, are implicated. Both of these The cumulative data on Epo-R signaling and Spi-1/ Spi-1/PU.1 domains are also needed to block erythroid PU.1 deregulation in MEL cells and the impacts on differentiation in MEL cells (Rekhtman et al., 1999; transcriptional and splicing targets, suggest a coordi- Zhang et al., 1999; Nerlov et al., 2000). nated stage-specific sequence of events that are initiated by constitutive activation of Epo-R and its downstream signaling pathways. The distinct effects of Spi-1/PU.1 Spi-1/PU.1 inhibits 4.1R erythroid splicing through phosphorylation on globin gene transcription and 4.1R PI3K/AKT-dependent and -independent pathways exon 16 splicing suggest the presence of two inhibitory A growing attention has been focused on how cellular pathways of late erythroid differentiation; one of which signals induced by growth factors, cytokines, hormones is PI3K/AKT-dependent, involves Ser41 and acts on and membrane depolarization can change the splicing Spi-1/PU.1 transcriptional activity, including the activa- pattern of several pre-mRNAs (Blaustein et al., 2007; tion of its own gene. The other pathway is PI3K/AKT- Tarn, 2007). Recent works have provided in vitro and independent, and involves Ser148. Both signaling cas- in vivo evidence showing that PI3K/AKT pathway cades, however, are required for 4.1R exon 16 splicing regulates specific alternative splicing events through inhibition in proliferative MEL cells (Figure 7). These phosphorylation of selective SR proteins (Blaustein observations further fuel the concept that we have et al., 2005; Patel et al., 2005). Here, we provide the proposed of two different Spi-1/PU.1-mediated pathways first observation that couples a functionally critical contributing to exon 16 exclusion (Blaybel et al., 2008). alternative splicing event to a signaling cascade that has an essential role in Epo-induced erythropoiesis. In total, downregulation of Spi-1/PU.1, either through DMSO treatment (Schuetze et al., 1992), short hairpin RNA- Materials and methods mediated knockdown (Blaybel et al., 2008) or PI3K/ Plasmid constructs AKT signaling inhibition (this work), remains the key We generated recombinant plasmids expressing different levels event that commits Friend erythroleukemia cells to late of Spi-1/PU.1 on MEL cell stable transfection. For low Spi-1/ erythroid differentiation and induces the erythroid PU.1 expression, we used the previously described pHOOK- splicing of 4.1R exon 16. Spi-1 construct (Theóleyre et al., 2004); the cell clone will be

Oncogene PI3K/AKT induces PU.1 autoregulation loop O Breig et al 2814 referred to as ‘l-PU.1 þ ’ clone. For high Spi-1/PU.1 expres- within exon 3: 50-CCGCACACCATGTCCACAAC-30 and a sion, we generated a new cell clone, as follows: the entire reverse primer within exon 4: 50-ATCCGGGGCATGTAGGA coding sequence of Spi-1/PU.1 was amplified by RT–PCR AAC-30, and a 193 bp fragment was generated. b-Actin using the following forward and reverse primers: PU-S1: mRNA, used as internal control, was amplified using 50-CTCTCCGGAATGTTACAGGCGTGCAAAATG-30 and previously described primers (Juban et al., 2009), and a PU-AS1: 50-TGCGAATTCTCAGTGGGGCGGGAGGCGC 230 bp fragment was generated. The reverse transcription step CG-30. For accurate cloning, BspEI and EcoRI restriction sites was performed on 1 mg of total RNA for 60 min at 37 1C using were added at the 50 ends of these primers, respectively 0.01 mg of random hexamers, 0.125 mM of dNTP mix, 1 U of (underlined region). The PCR product was inserted at the RNase inhibitor and 10 U of reverse transcriptase (Fermentas, BspEI/EcoRI site of PEFbosEGFP-C1 expression vector St-Re´my les Chevreuse, France). The PCR step was performed (Sanjuan et al., 2001) in matched open reading frame with using light cycler 480 SYBER Green I Master (Roche Applied the EGFP. The resulting construct PEFbosEGFP-C1-PU.1 Bioscience, Meylan, France) as follows: To 2.5 ml of 1/5 diluted was transfected in MEL cells, and the selected clone used in RT product, 10 mM of each primer and 5 ml of PCR Mix were this study will be referred to as ‘h-PU.1 þ ’. added in a final volume of 10 ml. A total of 45 cycles of Mutations of Spi-1/PU.1 were made by a two-step PCR amplification (20 s at 95 1C; 20 s at 60 1C; 15 s at 72 1C) were approach (Maillet et al., 1996) using the wild-type construct performed on a LightCycler MX3000P (Stratagene Agilent PEFbosEGFP-C1-PU.1 as template. The first-step PCRs were Technologies, Massy, France). performed with each of the outer primers, PU-S1 or PU-AS1, and the corresponding inner primers containing the specific Western blot mutation to be incorporated in the cDNA. The resulting PCR Erythroleukemia cells were cultured under various treat- products were then used as templates for the second-step PCR ment conditions. Whole cell lysates were prepared from these and amplification was performed using the outer primers. The cells, fractionated on sodium dodecyl sulfate polyacrylamide final PCR products were subcloned as BspEI/EcoRI-digested gel electrophoresis and blotted onto nitrocellulose mem- fragments in PEFbosEGFP-C1. The inserts were fully sequenced branes. The protocols were as previously detailed (Blaybel to ascertain the incorporation of the specific nucleotide changes et al., 2008). and the absence of other changes in sequences and reading The membranes were probed with the following primary frames. The inner primer set for Spi-1/PU.1 S41A mutation was: antibodies: anti-Spi-1/PU.1 antibodies: the first batch was a gift 0 0 S41A-S: 5 -TTCGTGGGCGCCGATGGAGA-3 ; S41A-AS: from Dr F Moreau-Gachelin, (Delgado et al., 1994) (1/1000), 0 0 5 -TCTCCATCGGCGCCCACGAA-3 . The inner primer set the second batch was an anti-Spi-1/PU.1 directed against the 0 for S148A mutation was: S148A-S: 5 -CTGGAGGTGGCTGAT C-terminal region (1/200, Santa Cruz Biotechnology, Inc., Santa 0 0 0 GGAGA-3 ; S148A-AS: 5 -TCTCCATCAGCCACCTCCAG-3 . Cruz, CA, USA, manufactured by Tebu-Bio, Le Perray-en- The nucleotide changes are underlined. Yvelines, France); anti-PI3K p85 subunit (1/1000, a gift from Dr Mouchiroud); anti-phospho-AKT(Ser473) antibody, specifically Cell culture, induction and transfection directed against the phosphorylated form of AKT on Ser residue Three erythroleukemia cell lines were used: MEL 745-A is an 473 (1/200, Santa Cruz); anti-GFP (1/1000, Roche Applied Epo-nonresponsive SFFV subclone, and SKT6 and Rauscher Bioscience); anti-b-actin monoclonal antibody directed against are Epo-responsive cell lines (Theóleyre et al., 2004). Cells were mouse b-actin (1/10 000, Chemicon International, Temecula, CA, grown in suspension in Iscove’s modified Dulbecco’s medium USA); and anti-Grb2 antibody (1/10 000, BD Transduction supplemented with 13% fetal bovine serum under standard Laboratories). Anti-b-actin and anti-Grb2 antibodies were used conditions (37 1C, 5% CO2). Induction of erythroid differenti- to standardize proteins signals. After primary antibodies, ation was performed by culturing the cells for 4 days, in a fresh appropriate peroxidase-conjugated secondary antibodies were complete medium, supplemented with 1.8% DMSO. Inhibition used. Immunoreactive bands were revealed by ECL þ enhanced of PI3K/AKT signaling was achieved by adding LY294002- chemiluminescence reagent (Amersham Biosciences AB, specific inhibitor (Sigma Aldrich, Saint Quentin Fallavier, Uppsala, Sweden). France) to cell culture at 5–20 mM final concentrations. Hemoglobin synthesis was assessed by the benzidine staining ChIP assay reaction using standard procedure (Cataldi et al., 2000). Chromatin immunoprecipitation experiments were performed For stable expression, proliferative cells (2 Â 105 cells) were essentially as previously detailed (Juban et al., 2009). Briefly, transfected with 0.5–1 mg plasmid DNA using LipofectAmine MEL cells were crosslinked with 1% formaldehyde, washed (Invitrogen, Cergy Pontoise, France) or Transfast (Promega, with PBS and lysed in sodium dodecyl sulfate buffer. DNA Charbonnie` res-Les Bains, France) compounds in 24-well tissue was then fragmented on ice by sonication to mean lengths of culture plates, according to the manufacturers procedures. about 300 bp. Complexes were immunoprecipitated overnight Transfected cells were then seeded to limited dilutions in 96- at 4 1C on a rotating wheel, using 3 mg of anti-Spi-1/PU.1 well plates with complete IMDM medium containing 800 mg antibody or anti-p27 antibody (Santa Cruz). DNA fragments G418 per ml or 400 mg zeocin per ml, to select stably were recovered by proteinase K digestion and phenol–chloro- transfected clones. Individual clones were analyzed for the form extraction. Immunoprecipitated DNA was quantified by expression of the exogenous proteins. real-time PCR using the following Spi-1/PU.1-specific primers: ChIP-S: 50-GAGACTTCCTGTAGCGCAAG-30; ChIP-AS: 0 0 Semiquantitative RT–PCR and real-time RT–PCR 5 -TGGGTCAGACGCAGGGCTCA-3 . Total RNA was isolated using TRIZOL reagent (Invitrogen) from cell lysates according to the manufacturer’s protocol. Inclusion of exon 16 in mature 4.1R mRNA was assessed by Abbreviations semiquantitative RT–PCR, as previously described (Deguillien et al., 2001; Theóleyre et al., 2004). ChIP, chromatin immunoprecipitation; DMSO, dimethylsulf- The steady-state expression of Spi-1/PU.1 mRNA was oxide; Epo, erythropoietin; MEL, mouse erythroleukemia; estimated by real-time RT–PCR, using a forward primer SFFV, spleen focus-forming virus.

Oncogene PI3K/AKT induces PU.1 autoregulation loop O Breig et al 2815 Conﬂict of interest for pEFbosEGFP_C1 plasmid, Dr F Moreau-Gachelin (Inserm U830, Institut Curie, Paris, France) for anti-Spi-1/ The authors declare no conﬂict of interest. PU.1 antibody, G Giraud for helping with the ChIP assay and Dr G Mouchiroud and Dr F Morle´(CGMC, Villeurbanne, France) for sharing reagents, including anti-p85 antibody. This Acknowledgements work was supported by grants from the ‘Ligue contre le Cancer, Comite´de la Loire’. Authors were supported by We thank Dr I Merida (Department of Immunology and the INSERM, the ‘Ligue contre le Cancer, Comite´de la Loire’, Oncology, Centro Nacional de Biotecnologia, Madrid, Spain) the ‘Ligue contre le Cancer, Comite´du Doubs’

References

Afrikanova I, Yeh E, Bartos D, Watowich SS, Longmore GD. (2002). Delgado MD, Hallier M, Meneceur P, Tavitian A, Moreau-Gachelin Oncogene cooperativity in Friend erythroleukemia: erythropoietin F. (1994). Inhibition of Friend cells proliferation by spi-1 antisense receptor activation by the env gene of SFFV leads to transcriptional oligodeoxynucleotides. Oncogene 9: 1723–1727. upregulation of PU.1, independent of SFFV proviral insertion. Delva L, Gallais I, Guillouf C, Denis N, Orvain C, Moreau-Gachelin Oncogene 21: 1272–1284. F. (2004). Multiple functional domains of the oncoproteins Spi-1/ Back J, Dierich A, Bronn C, Kastner P, Chan S. (2004). PU.1 PU.1 and TLS are involved in their opposite splicing effects in determines the self-renewal capacity of erythroid progenitor cells. erythroleukemic cells. Oncogene 23: 4389–4399. Blood 103: 3615–3623. Eisenbeis CF, Singh H, Storb U. (1995). Pip, a novel IRF family Baklouti F, Huang SC, Tang TK, Delaunay J, Marchesi VT, Benz Jr member, is a lymphoid-specific, PU.1-dependent transcriptional EJ. (1996). Asynchronous regulation of splicing events within activator. Genes Dev 9: 1377–1387. protein 4.1 pre-mRNA during erythroid differentiation. Blood 87: Fisher RC, Slayton WB, Chien C, Guthrie SM, Bray C, Scott EW. 3934–3941. (2004). PU.1 supports proliferation of immature erythroid progeni- Bao H, Jacobs-Helber SM, Lawson AE, Penta K, Wickrema A, tors. Leuk Res 28: 83–89. Sawyer ST. (1999). Protein kinase B (c-Akt), phosphatidylinositol Ghaffari S, Huang LJS, Zhang J, Lodish HF. (2003). Erythropoietin 3-kinase, and STAT5 are activated by erythropoietin (EPO) in receptor signaling processes. In: Molineux G, Foote MA and Elliott HCD57 erythroid cells but are constitutively active in an EPO- SG (eds). Erythropoietins and Erythropoiesis. Birkhauser: Basel, independent, apoptosis-resistant subclone (HCD57-SREI cells). Switzerland. Blood 93: 3757–3773. Jelkmann W. (2007). Erythropoietin after a century of research: Barnache S, Mayeux P, Payrastre B, Moreau-Gachelin F. (2001). younger than ever. Eur J Haematol 78: 183–205. Alterations of the phosphoinositide 3-kinase and mitogen-activated Juban G, Giraud G, Guyot B, Belin S, Diaz JJ, Starck J et al. (2009). protein kinase signaling pathways in the erythropoietin-independent Spi-1 and Fli-1 directly activate common target genes involved in Spi-1/PU.1 transgenic proerythroblasts. Blood 98: 2372–2381. ribosome biogenesis in Friend erythroleukemic cells. Mol Cell Biol Bavelloni A, Faenza I, Aluigi M, Ferri A, Toker A, Maraldi NM et al. 29: 2852–2864. (2000). Inhibition of phosphoinositide 3-kinase impairs pre-commit- Maillet P, Delaunay J, Baklouti F. (1996). Chimeric probe-mediated ment cell cycle traverse and prevents differentiation in erythroleu- ribonuclease protection assay for molecular diagnosis of mRNA kaemia cells. Cell Death Differ 7: 112–117. deficiencies. Hum Mutat 7: 61–64. Blaustein M, Pelisch F, Srebrow A. (2007). Signals, pathways and Moreau-Gachelin F. (2008). Multi-stage Friend murine erythroleukemia: splicing regulation. Int J Biochem Cell Biol 39: 2031–2048. molecular insights into oncogenic cooperation. Retrovirology 5: 99. Blaustein M, Pelisch F, Tanos T, Munoz MJ, Wengier D, Quadrana L Moreau-Gachelin F, Ray D, Mattei MG, Tambourin P, Tavitian A. et al. (2005). Concerted regulation of nuclear and cytoplasmic (1989). The putative oncogene Spi-1: murine chromosomal localiza- activities of SR proteins by AKT. Nat Struct Mol Biol 12: tion and transcriptional activation in murine acute erythroleuke- 1037–1044. mias. Oncogene 4: 1449–1456. Blaybel R, Theóleyre O, Douablin A, Baklouti F. (2008). Down- Myklebust JH, Blomhoff HK, Rusten LS, Stokke T, Smeland EB. regulation of the Spi-1/PU.1 oncogene induces the expression of (2002). Activation of phosphatidylinositol 3-kinase is important for TRIM10/HERF1, a key factor required for terminal erythroid cell erythropoietin-induced erythropoiesis from CD34(+) hematopoie- differentiation and survival. Cell Res 18: 834–845. tic progenitor cells. Exp Hematol 30: 990–1000. Bouscary D, Pene F, Claessens YE, Muller O, Chretien S, Fontenay- Nerlov C, Querfurth E, Kulessa H, Graf T. (2000). GATA-1 interacts Roupie M et al. (2003). Critical role for PI 3-kinase in the control of with the myeloid PU.1 transcription factor and represses PU.1- erythropoietin-induced erythroid progenitor proliferation. Blood dependent transcription. Blood 95: 2543–2551. 101: 3436–3443. Nishigaki K, Hanson C, Ohashi T, Spadaccini A, Ruscetti S. (2006). Cataldi A, Di Pietro R, Centurione L, Grilli A, Cutroneo G, Miscia S. Erythroblast transformation by the friend spleen focus-forming (2000). Phosphatidylinositol-3-kinase activation and atypical virus is associated with a block in erythropoietin-induced STAT1 protein kinase C zeta phosphorylation characterize the DMSO phosphorylation and DNA binding and correlates with high signalling in erythroleukemia cells. Cell Signal 12: 667–672. expression of the hematopoietic phosphatase SHP-1. J Virol 80: Chasis JA, Coulombel L, Conboy J, McGee S, Andrews K, Kan YW 5678–5685. et al. (1993). Differentiation-associated switches in protein 4.1 Nishigaki K, Hanson C, Ohashi T, Thompson D, Muszynski K, expression. Synthesis of multiple structural isoforms during normal Ruscetti S. (2000). Erythroid cells rendered erythropoietin indepen- human erythropoiesis. J Clin Invest 91: 329–338. dent by infection with Friend spleen focus-forming virus show Chen H, Ray-Gallet D, Zhang P, Hetherington CJ, Gonzalez DA, constitutive activation of phosphatidylinositol 3-kinase and Akt Zhang DE et al. (1995). PU.1 (Spi-1) autoregulates its expression in kinase: involvement of insulin receptor substrate-related adapter myeloid cells. Oncogene 11: 1549–1560. proteins. J Virol 74: 3037–3045. Deguillien M, Huang SC, Morinie` re M, Dreumont N, Benz Jr EJ, Okuno Y, Huang G, Rosenbauer F, Evans EK, Radomska HS, Baklouti F. (2001). Multiple cis elements regulate an alternative Iwasaki H et al. (2005). Potential autoregulation of transcription splicing event at 4.1R pre-mRNA during erythroid differentiation. factor PU.1 by an upstream regulatory element. Mol Cell Biol 25: Blood 98: 3809–3816. 2832–2845.

Oncogene PI3K/AKT induces PU.1 autoregulation loop O Breig et al 2816 Patel NA, Kaneko S, Apostolatos HS, Bae SS, Watson JE, Davidowitz Schuetze S, Paul R, Gliniak BC, Kabat D. (1992). Role of K et al. (2005). Molecular and genetic studies imply Akt-mediated the PU.1 transcription factor in controlling differentiation signaling promotes protein kinase CbetaII alternative splicing via of Friend erythroleukemia cells. Mol Cell Biol 12: phosphorylation of serine/arginine-rich splicing factor SRp40. 2967–2975. J Biol Chem 280: 14302–14309. Scott EW, Simon MC, Anastasi J, Singh H. (1994). Requirement of Paul R, Schuetze S, Kozak SL, Kozak CA, Kabat D. (1991). The Sfpi- transcription factor PU.1 in the development of multiple hemato- 1 proviral integration site of Friend erythroleukemia encodes the poietic lineages. Science 265: 1573–1577. ets-related transcription factor Pu.1. J Virol 65: 464–467. Sivertsen EA, Hystad ME, Gutzkow KB, Dosen G, Smeland EB, Pongubala JM, Van Beveren C, Nagulapalli S, Klemsz MJ, Blomhoff HK et al. (2006). PI3K/Akt-dependent Epo-induced McKercher SR, Maki RA et al. (1993). Effect of PU.1 phosphor- signalling and target genes in human early erythroid progenitor ylation on interaction with NF-EM5 and transcriptional activation. cells. Br J Haematol 135: 117–128. Science 259: 1622–1625. Tarn WY. (2007). Cellular signals modulate alternative splicing. J Rao G, Rekhtman N, Cheng G, Krasikov T, Skoultchi AI. (1997). Biomed Sci 14: 517–522. Deregulated expression of the PU.1 transcription factor blocks Theóleyre O, Deguillien M, Morinie` re M, Starck J, Moreau-Gachelin murine erythroleukemia cell terminal differentiation. Oncogene 14: F, Morle´F et al. (2004). Spi-1/PU.1 but not Fli-1 inhibits erythroid- 123–131. specific alternative splicing of 4.1R pre-mRNA in murine erythro- Rekhtman N, Radparvar F, Evans T, Skoultchi AI. (1999). Direct leukemia cells. Oncogene 23: 920–927. interaction of hematopoietic transcription factors PU.1 and GATA-1: Yamada T, Kondoh N, Matsumoto M, Yoshida M, Maekawa A, functional antagonism in erythroid cells. Genes Dev 13: 1398–1411. Oikawa T. (1997). Overexpression of PU.1 induces growth and Rieske P, Pongubala JM. (2001). AKT induces transcriptional activity differentiation inhibition and apoptotic cell death in murine of PU.1 through phosphorylation-mediated modifications within its erythroleukemia cells. Blood 89: 1383–1393. transactivation domain. J Biol Chem 276: 8460–8468. Zhang P, Behre G, Pan J, Iwama A, Wara-Aswapati N, Radomska HS Sanjuan MA, Jones DR, Izquierdo M, Merida I. (2001). Role of et al. (1999). Negative cross-talk between hematopoietic regulators: diacylglycerol kinase alpha in the attenuation of receptor signaling. GATA proteins repress PU.1. Proc Natl Acad Sci USA 96: J Cell Biol 153: 207–220. 8705–8710.

Oncogene