Mice deficient in the erythroid transcription factor EKLF die ~dE14 from severe anemia, attributed to decreased β-globin expression. Recent reports using microarray analyses indicate that expression of numerous genes is perturbed in erythroid cells lacking EKLF. We performed flow cytometry of WT dE14 fetal liver (FL) cells with TER119 and CD71 and identified 5 previously described populations: R1+R2 composed of BFU-E and CFU-E, respectively, and R3+R4+R5 composed of more mature erythroblasts. The same analysis of dE14 EKLF-deficient FL cells showed R3+R4+R5 were absent, indicating a block in erythroid differentiation. This differentiation block introduced a bias into the previous microarray data, as EKLF-deficient FL contain only immature erythroid cells (R1+R2), while WT FL cells are predominantly more differentiated (R3+R4+R5). Thus, expression of many genes was decreased due to a loss of more mature erythroblasts, rather than due to the action of EKLF or an intermediary. To obtain more rigorous comparative data, we performed microarray analyses with EKLF-deficient FL cells and R1+R2 populations from WT FL cells. Expression of numerous genes was deregulated. Of note, many genes significantly down regulated in previous microarray analyses were not down regulated when the 2 similar populations were compared. Ingenuity Pathway Analysis of the microarray data identified a biologic network involved in cell cycle and DNA replication. At the central nodes of the network were E2F1 and E2F2, transcription factors involved in cell cycle control and differentiation. In quantitative RT-PCR, E2F1 and E2F2 expression in EKLF-deficient mRNA was decreased to 40.5±1.6 and 7.6±1.6% of WT, respectively. Western blot analysis demonstrated comparable reductions in both E2F proteins. Cell cycle analyses showed that EKLF-deficient R1 cells exhibit a significant delay exiting G0+G1 and entering S phase (p<0.001). Both R1 and R2 cells exhibited a defect exiting S and entering G2+M (p<0.004). Colony forming assays revealed that EKLF-deficient FL cells had decreased frequency of BFU-E in R1 (p<0.005), with a defect in ability to generate CFU-E, and decreased frequency of CFU-E in R2 (p<0.001) with a defect in ability to differentiate into more mature erythroblasts. Flow cytometry with annexin V staining revealed that EKLF-deficient cells were resistant to apoptosis, indicating apoptosis was not contributing to the block. Based on these data, we hypothesized that E2F1/2 were EKLF target genes. Thus we examined the chromatin at the E2F2 locus in WT and EKLF-deficient FL cells using a high throughput assay to identify DNase-I hypersensitive sites (HS). Three HS were identified in WT that were absent in EKLF-deficient chromatin, one in the promoter region, one in intron 2, and one in the 3′ region. The E2F2 promoter HS region contains 2 EKLF consensus binding sites, which were examined in gel mobility shift assays for their ability to bind EKLF. One of these probes yielded a complex that co-migrated with a control β-globin probe complex. This complex was competed by an excess of E2F2 or β-globin probe and supershifted with an anti-EKLF MoAb. These results support the hypothesis that defects in E2F1/2 expression are associated with abnormalities in cell cycle and differentiation, contributing to a failure of definitive erythropoiesis in EKLF-deficient mice.
Disclosure: No relevant conflicts of interest to declare.