FKLF, a novel Krüppel-like factor that activates human embryonic and fetal beta-like globin genes

Abstract

A cDNA encoding a novel Krüppel-type zinc finger protein, FKLF, was cloned from fetal globin-expressing human fetal erythroid cells. The deduced polypeptide sequence composed of 512 amino acids revealed that, like Sp1 and EKLF, FKLF has three contiguous zinc fingers at the near carboxyl-terminal end. A long amino-terminal domain is characterized by the presence of two acidic and two proline-rich regions. Reverse transcription (RT)-PCR assays using various cell lines demonstrated that the FKLF mRNA is expressed predominantly in erythroid cells. FKLF message is detectable by RT-PCR in fetal liver but not in adult bone marrow cells. As predicted from its structural features, FKLF is a transcriptional activator. In luciferase assays FKLF activated the gamma- and epsilon-globin gene promoters, and, to a much lower degree, the beta-globin promoter. Studies of HS2-gamma gene reporter constructs carrying CACCC box deletions revealed that the CACCC box sequence of the gamma gene promoter mediates the activation of the gamma gene by FKLF. Other erythroid promoters (GATA-1, glycophorin B, ferrochelatase, porphobilinogen deaminase, and 5-aminolevulinate synthase) containing CACCC elements or GC-rich potential Sp1-binding sites were activated minimally, if at all, by FKLF, indicating that FKLF is not a general activator of genes carrying the CACCC motifs. Transfection of K562 cells with FKLF cDNA enhanced the expression of the endogenous epsilon- and gamma-globin genes, suggesting an in vivo role of FKLF in fetal and embryonic globin gene expression. Our results indicate that the protein potentially encoded by the FKLF cDNA acts as a transcriptional activator of embryonic and fetal beta-like globin genes.

FIG. 1

(A) Deduced polypeptide sequence of FKLF. Proline residues are shown in boldface. The sequence of the three zinc fingers is underlined, and the conserved polypeptide regions used for the preparation of degenerate primers are shaded. (B) Schematic representation of structure of deduced FKLF protein. Acidic and proline-rich regions are indicated by shaded and hatched boxes, respectively. Three zinc fingers are represented by striped boxes. Numbers above the boxes represent amino acid positions from the first methionine.

FIG. 2

Percent amino acid identities of zinc fingers among Sp1- or EKLF-related proteins. Identities with FKLF zinc fingers are shown in boldface. Two shaded areas indicate high homology groups, i.e., the Sp1 family, including Sp1 (18), Sp2 (19), Sp3 (14, 19), and Sp4 (14), and the EKLF family, including EKLF (24), BTEB2 (37), LKLF (1), BKLF (7), and GKLF/EZF (11, 36). The identity between these two groups, indicated by striped area, is relatively low. Note that BTEB (17), TIEG (41), and FKLF show identities similar to both the Sp1 and the EKLF family proteins.

FIG. 3

FKLF mRNA expression in K562 and Jurkat cells by Northern blotting. Four micrograms of poly(A)⁺RNA was run on each lane. After standard capillary transfer to a nylon membrane, the RNA was blotted with a specific FKLF probe (upper panel). Subsequently, the FKLF probe was stripped off, and the membrane was reblotted with a murine GAPDH probe (lower panel). 28S and 18S rRNA positions are indicated.

FIG. 4

FKLF mRNA expression in primary cells and established cell lines. First-strand cDNA was transcribed from total RNA by using random hexamers. The cDNA solution was diluted to give similar band intensities in the same cycles of amplification of 28S rRNA. (A) FKLF expression in adult human bone marrow and human fetal liver. Note that the amplification of FKLF message is less efficient in the cDNA of the adult bone marrow than in that of the fetal liver, while these cDNAs gave similar amplification of 28S rRNA. (B) FKLF expression in various cell lines. Total RNA used for the RT-PCR assays was prepared from erythroid lines (K562, CHRF, MB-02, HEL, and MEG-O1), a T-cell line (CEM), an Epstein-Barr virus-transformed B-cell line (Russell-Hardy-2), myeloid lines (KG-1 and HL-60), a fibroblastic line (82-6), a neuroblastoma line (SK-N-SH), an endothelial line (CRL1998), a smooth muscle line (CRL1999), a kidney epithelial line (HH-39), and a hepatoma line (Hu-H7). Three bands in each picture show the results of amplification in different cycles, i.e., from left to right, 35, 33, and 31 cycles for FKLF, and 22, 20, and 18 cycles for 28S rRNA. Note that the FKLF gene was amplified from RNA of all erythroid lines but not from RNA of lymphoid or myeloid lines. FKLF was also amplified in the endothelial line, which is noteworthy in view of the fact that endothelial cells express other hemopoietic lineage characters such as the erythropoietin receptor, kit and kit ligand, CD34, etc. Amplification of FKLF cDNA in the other nonhemopoietic cell lines was less efficient.

FIG. 5

trans activation of β- and γ-globin gene promoters by FKLF compared with EKLF. Reporter constructs containing a luciferase gene driven by HS2 and either the β or the γ gene promoter were transfected into K562 cells with or without the activator plasmid, pSG5/FKLF or pSG5/EKLF. Luciferase activities were corrected by protein concentrations and expressed as relative percentages of luciferase activity of pHS2γLuc which were not cotransfected by transactivator plasmid (100%). Data are expressed as means (columns) ± standard deviations (error bars) derived from multiple transfections with two different plasmid sets. Note that FKLF activates γ- and β (at a lower level)-globin gene promoters, while EKLF activates only the β gene promoter.

FIG. 6

Structures of γ gene reporter constructs with deletion of a CACCC box or a whole HS2 sequence. Deletions of 9-bp CACCC sequences of HS2 (HS-CAC2 and HS-CAC3) and the γ gene promoter are indicated.

FIG. 7

γ gene activation by FKLF with reporter constructs with deletions of a CACCC sequence or an HS2. Reporter constructs depicted in Fig. 7 were transiently transfected into K562 cells with or without pSG5/FKLF. Luciferase activities were corrected by protein concentrations and expressed as relative percentages of luciferase activity of pHS2γLuc which were not cotransfected by transactivator plasmid (100%). P values given by statistical comparison of data of two groups, i.e., the presence and the absence of FKLF, are shown above the columns. Notice that FKLF activates the γ gene promoter of the constructs pHS2^ΔCAC2γLuc, pHS2^ΔCAC3γLuc, and pγLuc lacking the HS2 element but failed to activate the γ gene promoter in the construct pHS2γ^ΔCACLuc.

FIG. 8

Reporter constructs containing a luciferase gene driven by the HS2 and a promoter of a gene expressed in erythroid cells. CACCC sequences and GC-rich potential Sp1-binding sites in the promoters are illustrated with solid rectangles and open ellipses, respectively. Numbers above the promoters are base pair distances from the cap site (γ, GPB, FC, PBGD, and ALAS) or from the end of exon 1 (GATA-1 [GATA]) and indicate the upstream and the downstream ends of the promoter sequences cloned and the positions of the CACCC or the GC-rich sequences.

FIG. 9

trans activation of promoters of various genes expressed in erythroid cells by FKLF. Reporter constructs depicted in Fig. 9 were transiently transfected into K562 cells with or without pSG5/FKLF. Luciferase activities were corrected by protein concentrations and expressed as relative percentages of luciferase activity of pHS2γLuc which was not cotransfected by transactivator plasmid (100%). P values given by statistical comparison of data of two groups, i.e., the presence and the absence of FKLF, are shown above the columns. Notice that FKLF activates the γ gene promoter as expected, but minimal (if any) activation of other erythroid promoters by FKLF was detected in spite of the fact that those promoters contained a CACCC or a GC-rich sequence.

FIG. 10

Comparison of activities of FKLF on ɛ-, γ-, and β-globin gene promoters. Reporter constructs containing a luciferase gene driven by HS2 and either the ɛ, the γ, or the β gene promoter were transfected into K562 cells with or without the activator plasmid, pSG5/FKLF. Luciferase activities were corrected by protein concentrations and expressed as relative percentages of luciferase activity of pHS2γLuc which was not cotransfected by transactivator plasmid (100%). Note that FKLF activates the ɛ gene promoter more strongly than the γ gene promoter.

FIG. 11

Activation of the endogenous ɛ- and γ-globin genes by FKLF. FKLF cDNA was transiently transfected into K562 cells, and the expression of ɛ- and γ-globin genes was analyzed at day 3 by RT-PCR. Amplifications of the ɛ- and the γ-globin cDNAs with and without exogenous FKLF cDNA were compared under conditions which give rise to similar band patterns of 28S rRNA. Three bands in each picture show the results of amplification in different cycles; i.e., from left to right, 28, 26, and 24 cycles for the ɛ gene; 27, 25, and 23 cycles for the γ gene; and 21, 19, and 17 cycles for the 28S rRNA. The specificity of amplifications of the ɛ- and the γ-globin genes was confirmed by NcoI digestion of the PCR products, resulting in 177- and 60-bp bands for the ɛ gene and 140- and 97-bp bands for the γ gene (data not shown). Note that both the ɛ- and the γ-globin genes were more efficiently amplified in the presence of exogenous FKLF than in its absence, indicating that FKLF up-regulated the transcription of these genes.