Degenerate DNA Binding By Mutant (E339D) KLF1 Dramatically Alters the Erythroid Transcriptome in the Nan Mouse Model

The C2H2 zinc finger transcription factor KLF1 regulates nearly all aspects of erythroid differentiation including iron trafficking, heme synthesis, globin production, membrane skeleton assembly and the cell cycle. KLF1 is primarily, but not exclusively, a transcriptional activator. Previously we showed that the inbred mouse model Nan (neonatal anemia) carries a missense mutation (E339D) in the second zinc finger of KLF1 that causes severe lifelong anemia in heterozygotes and embryonic lethality (E10-11) in homozygotes. We also showed that E339D confers an altered DNA binding specificity such that mutant Nan-KLF1 fails to properly bind numerous known KLF1 target genes. As a result, fetal liver (FL) expression of many genes including globin (Hbb, Hba), dematin (Dmtn/Epb4.9) and the cell cycle regulators E2F2 (E2f2) and p18 (Cdkn2c) are significantly decreased. To examine the Nan global erythroid transcriptome, we performed RNAseq on E14.5 FLs. Fetuses were genotyped for Y chromosome markers and 3 wild-type (WT) and 3 Nan/+ males were analyzed. Using a stringently filtered dataset (CPM > 1 in at least 3 of 6 samples, FDR 0.05, FC ≥ 2), 610 genes were upregulated and 200 downregulated in Nan vs. WT. Notably, only 18% of the upregulated and 52% of the downregulated genes overlapped the 1078 known KLF1 target genes (defined in prior studies as those differentially expressed in Klf1-/- vs +/+ FL). Moreover, just 3% of the upregulated Nan genes overlapped genes normally activated by KLF1. Thus, significant ectopic gene expression occurs in Nan. Pathway analysis (IPA, PANTHER) identified Toll receptor signaling, acute phase response signaling, and response to interferon as among the top hits in the upregulated gene set in Nan, indicating an overall inflammatory state that would be expected to exacerbate anemia. Indeed, we confirmed significantly increased levels of serum IFNβ and IRF7 in Nan. In the Nan downregulated gene set, cytoskeletal regulation by Rho GTPase and cell cycle regulation were the top hits. Previous studies identified deficiencies of membrane skeleton proteins in Nan. To analyze cell cycle status, we sorted dividing adult spleen and bone marrow (BM) erythroid precursors (proerythroblasts, basophilic and polychromatophilic erythroblasts) using CD44, Ter119 and FSC. Polychromatophilic erythroblasts in Nan BM were arrested in G1 (significantly increased percentage of cells in G0/G1, significantly decreased percentage in S phase; p < 0.05) compared to saline-injected and phenylhydrazine treated WT controls. Surprisingly, however, no other cell cycle defects were seen among other BM compartments and none were seen in the spleen. We next analyzed gene ontology (GO) terms in both the up- and downregulated Nan gene sets (MGI, Mouse Genome Informatics). Ten cell cycle associated genes were downregulated (e.g., Rgcc, Cdc25b, E2f2/4), and 14 were upregulated (e.g., Gadd45b/g, Map3k8, Map3k6, Aim1) indicating a complicated pattern of gene expression differences and physiological outcomes in Nan. Similarly, expression of genes involved in iron regulation were among those most highly affected in Nan and did not predict the observed physiological outcome. The mitochondrial iron transporter SLC25A37 (Mitoferrin) and the transporter SLC25A38, both important in heme synthesis, and the ferrireductase required for transferrin-mediated iron uptake by red cells, STEAP3, are downregulated in Nan. Hamp encoding hepcidin, a negative regulator of the iron importer, ferroportin, is upregulated in Nan >14x. Ferroportin expression is not concomitantly downregulated, however. Despite these changes, no evidence of systemic iron deficiency has been identified in Nan; serum iron levels are normal. Moreover, the zinc protoporphyrin IX to heme ratio is markedly increased in Nan, indicating that a defect in heme biosynthesis up to and including ferrochelatase is unlikely, and uptake of radiolabeled iron by Nan reticulocytes and incorporation into heme are normal. From these studies we conclude: (1) Nan-KLF1 activates and represses a large number of non-KLF1 target genes; (2) Nan-KLF1 inappropriately increases and decreases expression of subsets of normal KLF1 genes resulting in (3) unpredictable physiogical outcomes in Nan. Overall, we conclude that new homeostatic set points arise in Nan as a result of the combined contributions of a large set of differentially expressed genes.