Overexpression of Liar inhibits erythroid differentiation. (A) Myc-tagged Liar (MSCV-Liar) and LiarΔNLS (MSCV-LiarΔNLS) overexpression inhibit BFU-E and CFU-E formation. Methylcellulose was used for erythroid colony–forming assays using fetal liver cells infected with retroviruses expressing Liar (MSCV-Liar), LiarΔNLS (MSCV-LiarΔNLS), or vector only (MSCV). The number of Epo-responsive benzidine-positive colonies (CFU-E and BFU-E) is shown. The mean ± standard deviation (SD) is presented (n = 3). **Statistically significant (2-way ANOVA) values (P ≤ .01). (B) J2E cells were infected with retroviruses expressing myc-tagged Liar (MSCV-Liar), LiarΔNLS (MSCV-LiarΔNLS), or vector only (MSCV) and analyzed for Liar subcellular localization by confocal microscopy and immunofluorescence staining using anti-myc antibodies. (C) Effects of Liar/LiarΔNLS expression on Epo-induced hemoglobin production detected by benzidine staining. (D) Effects of Liar/LiarΔNLS expression on J2E proliferation measured by a MTT-based assay after Epo stimulation. The mean (± SD) is presented (n = 3). *Statistically significant (2-way ANOVA) values (P ≤ .05). (E) Liar/LiarΔNLS expression reduce globin protein levels. Expression of exogenous Liar/LiarΔNLS (anti-myc), GATA-1, globin, and Erk2 were analyzed by Western blot analysis. GATA-1 and globin levels are expressed relative to Erk2 expression. (F) Western blot analysis of J2E cells overexpressing Liar (MSCV-Liar), LiarΔNLS (MSCV-LiarΔNLS), and control cells (MSCV) for phospo-Akt levels relative to total Akt. (G) Effect on viability of reducing the serum levels in media for the cell lines as in panel F.