The β-globin locus is a paradigm for the study of developmental gene regulation (reviewed in Hardison et al.,1997, and Li et al.,2002). The regulation of human, murine, and avian β-like globin genes suggests that they are controlled by similar factors (Reitman et al.,1990; Glauber et al.,1991; Foley et al.,1994; Mason et al.,1995; Basu et al.,2004). In experiments both in a mouse erythroleukemia (MEL) cell line and in transgenic mice, the globin genes in a construct with the entire chicken β-globin locus were developmentally regulated and exhibited a pattern of expression mirroring that seen in the chicken (Glauber et al.,1991; Mason et al.,1995). Another study demonstrated that a construct containing the chicken ε-globin gene and the β/ε enhancer exhibited autonomous, high-level stage- and tissue-specific expression in transgenic mice (Foley et al.,1994). When the human γ- and β-globin genes are transiently transfected into primitive chicken blood cells, the γ-globin gene is expressed and β-globin transcription is suppressed; that is, the genes are correctly regulated (Basu et al.,2004). These findings suggest that common factor(s) direct developmental expression of the globin genes in the chicken, in mouse, and in human cells.
Among the large number of mammalian genes that exhibit sequence homology to the DNA-binding domain of Drosophila melanogaster Krüppel, the Krüppel-like factors (KLFs) are a particularly closely related family. The KLFs have three Cys2His2 (C2/H2) zinc finger domains, and specific KLFs share conserved residues within and between these zinc fingers. There are 17 KLFs that have been identified in mammals, and they are related to the SP1-like family of transcription factors, with one overlap in nomenclature (SP6/KLF14; Philipsen and Suske,1999; Turner and Crossley,1999; Bieker,2001; Black et al.,2001; Kaczynski et al.,2003). We previously have identified chicken homologues of the human KLF2, 3, 4, 5, 9, 11, 12, 13, and 15 genes, and evolutionary conservation of these genes suggests functional conservation (Basu et al.,2004).
Erythroid Krüppel-like Factor (EKLF or KLF1) is a positive regulator of adult β-globin gene expression. It was identified by subtractive hybridization of DS-19 MEL and J774 monocyte–macrophage cell lines as part of a strategy to identify novel erythroid-specific genes (Miller and Bieker,1993). EKLF binds to the consensus sequence 5′-NCNCNCCCN-3′. Sites of this nature are found in the promoters of several erythroid-specific genes, including the adult β-globin gene. Consistent with its binding specificity, EKLF activates β-globin gene reporter constructs containing an intact CACCC box (CCACACCCT) but does not activate those containing a CACCC box mutated to sequences present in the β-CACCC box of certain patients with β-thalassemia (Feng et al.,1994). EKLF is not required for expression of the embryonic globin genes (Nuez et al.,1995; Perkins et al.,1995). However, it is expressed starting early during mouse development at embryonic day 7 (E7) primitive streak stage (Southwood et al.,1996). EKLF−/− mice develop fatal anemia during definitive erythropoiesis and die by day 16 during gestation. Recently, EKLF has been shown to play a role in primitive as well definitive erythropoiesis (Drissen et al.,2005; Hodge et al.,2005).
Phylogenetic studies indicate that EKLF is most closely related to KLF2 and KLF4. Recently, we reported that KLF2/LKLF is essential for primitive erythropoiesis and regulates the embryonic globin genes in vivo (Basu et al.,2005). These mice also have a defect in blood vessel formation and die between E12 and E14 (Kuo et al.,1997; Wani et al.,1998). KLF4−/− mice die shortly after birth due to defect in skin barrier function, but no erythroid defects have been demonstrated so far (Segre et al.,1999).
EKLF orthologues have been described in nonmammalian systems. The Xenopus laevis Neptune KLF is the orthologue of human EKLF based on sequence similarity. Neptune is expressed in sites of primitive erythropoiesis in Xenopus, and it enhances globin induction in animal cap explants in conjunction with XGATA-1 (Huber et al.,2001). There is no unique orthologue of EKLF in zebrafish based on sequence similarity. Possible candidates are Klf2a and 2b, Klf4 (also called biklf), and Klfd. However, the expression pattern of Klfd during definitive erythropoiesis marks this gene as the most likely orthologue of EKLF (Oates et al.,2001).
The chicken presents a valuable model system in which to study the regulation and switching of the globin genes. Previous work in our laboratory identified many KLFs in the chicken with homology to the human KLFs (Basu et al.,2004), but interestingly, no homologue of EKLF was found. In this study, an erythroid-specific gene in the chicken with homology to EKLF has been identified. This finding will allow for the study of the individual and combined effects of the KLFs on globin gene regulation in a model system with nucleated definitive red blood cells.
Identification of Chicken EKLF
The mRNA sequences of the murine EKLF gene and the Xenopus Neptune gene were used to search the NCBI, UCSC Genome Browser, and BBSRC ChickEST databases. Although no matches were found in the expressed sequence tag (EST) database, sequences from both the NCBI and UCSC databases matched the query sequences. Sequences that produced the most significant matches to the murine and Xenopus sequences were examined to eliminate those that were more closely related to KLFs other than mammalian EKLF and Neptune. The resulting chicken sequences were aligned (PRETTY, GCG Wisconsin package) to create a contiguous sequence (contig). Primers were designed to amplify regions within the contig and polymerase chain reaction (PCR) was performed using genomic DNA and cDNA generated from 14-day blood as templates. These PCR products were sequenced, and the sequence data was aligned, to determine the cDNA sequence and exon–exon junctions for chicken EKLF (cEKLF). A contiguous sequence is now available at http://genome.ucsc.edu at the cEKLF chromosomal location on chromosome 8 (20699K-20702K; Table 1). We have determined the chromosomal location of the chicken KLFs, and they are presented in Table 1. Although the EKLF and KLF2 genes are located in close proximity on the same chromosome in mouse and human (Basu et al.,2005), this finding is not the case in chicken. Chicken EKLF has three exons, as do its mammalian homologues. Virtual translation of the chicken EKLF mRNA sequence (TRANSLATE, GCG Wisconsin package) predicts a 405 amino acid protein. cEKLF has a proline-rich amino terminus and three C2/H2 zinc fingers in the carboxy terminus, like the mammalian EKLF proteins (Fig. 1). The 3′ nuclear localization signal (NLS) is also conserved with the mammalian EKLFs and Neptune (boxed sequence, Fig. 1). The mammalian NLS sequence (KPKRGRRSWPRKR) is present only in EKLF, KLF2, and KLF4. The cEKLF NLS sequence (KPKRGRRSWARKR) has one amino acid different than the mammalian sequence, and is 100% identical to Neptune.
Table 1. Chromosomal Location of the Chicken KLF Genes and Their Mammalian Homologuesa
Chromosomal location in chicken (genomic location in kb)
Evolutionary Conservation of the EKLF Gene Across Species
Phylogenetic analysis of the zinc finger regions of the mammalian and avian KLF1, KLF2, and KLF4 proteins and of the KLF2b, KLF2a, and KLFd proteins from zebrafish shows that chicken EKLF clusters with the human and murine EKLF proteins, along with zebrafish Klfd, as expected (Fig. 2). Figure 3A shows the alignment of the amino acid sequence of the zinc finger region of chicken EKLF with the amino acid sequence of the zinc finger region of the EKLF proteins from mammals (human, mouse, dog) and related proteins from nonmammals (zebrafish Klfd and Xenopus Neptune). These proteins, both mammalian and nonmammalian, are remarkably similar to the chicken EKLF protein in the zinc finger region. Chicken EKLF is most similar to the Xenopus protein Neptune (98.8% similar and 97.5% identical) and KLFd from zebrafish (95.1% similar and 87.7% identical). The chicken EKLF zinc finger region is 91.4% similar and 87.7% identical to human EKLF and 87.7% similar and 82.7% identical to murine EKLF. One of the two residues in the chicken EKLF zinc finger domain that differs from Xenopus matches the human protein instead (position 50; Fig. 3A). The C and H residues that chelate zinc ions, and the residues that contact DNA, are conserved among all of these proteins (Oates et al.,2001). Taken together, these data suggest that chicken EKLF has a similar function to its mammalian and nonmammalian counterparts.
Expression of Chicken EKLF mRNA During Development
Because EKLF is expressed during primitive erythropoiesis in the mouse (Southwood et al.,1996), the expression pattern of chicken EKLF in early embryogenesis was determined for comparison. Primitive hematopoietic cells are derived from the posterior of the epiblast in chicken. After ingression of the epiblast at the primitive streak stage, these mesodermal cells migrate toward the extraembryonic yolk sac area and form blood islands starting at stage 4. Cells in the blood islands differentiate into the erythroid and endothelial lineages between 16 (stage 4 by Hamburger and Hamilton,1951) and 50 (stage 13) hours of development.
To explore the spatiotemporal distribution of EKLF mRNA in the chicken embryo, whole-mount in situ hybridization was performed on embryos between stage 4 and 13. The expression of EKLF was detectable already throughout the epiblast at stage 4. Strong expression of EKLF was seen in the posterior primitive streak (indicated by white arrowheads in Fig. 3B), which gives rise to hematopoietic cells, through at least stage 8. Between stages 5 and 6, EKLF expression in the blood islands begins before the expression of hemoglobin at stage 7 (Minko et al.,2003), and continues through stage 13 (embryonic day 2 of development). EKLF was also expressed in the circulating blood cells (indicated by yellow arrowheads in stage 13 of Fig. 3B). The expression of EKLF was also easily detectable in the neural fold and neural plate (stage 7 and 8) and weakly detected in the lateral plate between stages 7 and 13, which was surprising. Otherwise, the early expression pattern of chicken EKLF was similar to murine EKLF (Southwood et al.,1996).
To study expression of chicken EKLF mRNA in the later stages of development, cEKLF mRNA was quantified by real-time PCR during erythroid and brain development, at 5, 7, and 14 days. Expression of the embryonic ρ- and ε-globin genes occurs between 3 and 7 days before switching to the adult βA-globin gene by 8 days, so these time points correspond to primitive (5-day), transitional (7-day), and definitive (14-day) erythropoiesis (Landes et al.,1982). Levels of expression of ρ-globin and βA-globin mRNA (Fig. 4A) were measured in blood at 5, 7, and 14 days of development. Expression of ρ-globinmRNA decreased sharply, whereas expression of βA-globin mRNA increased sharply after 5 days of development. There was no detectable expression of cEKLF in the brain at these stages of development. In the blood, expression of chicken EKLF mRNA increased slightly between 5 and 7 days but showed a significant 8- to 10-fold increase between 7 and 14 days of development (P < 0.05; Fig. 4B).
A chicken homologue of EKLF/KLF1 could not be identified in the EST database previously studied. However, it was expected that the chicken have an EKLF gene, because the mammalian globin genes are correctly regulated in chicken cells, and vice versa. In fact, a complex binding specifically to globin CACCC boxes was observed with chicken adult blood nuclear extracts in gel-shift assays (Basu et al.,2004). In the current study, we have identified the chicken homologue of the mammalian EKLF gene and examined its tissue specificity and expression pattern during development.
EKLF was the first of the KLFs to be identified (Miller and Bieker,1993). A search of the chicken genome revealed a gene with similarity to both the human and murine EKLF genes, but unique from chicken KLF2 and KLF4. Phylogenetic analyses placed this gene in a cluster with the human and murine EKLF, along with zebrafish Klfd, confirming that it was indeed chicken EKLF. The cEKLF is expressed in mesodermally derived erythroid cells, and its expression level rises significantly during definitive erythropoiesis, similar to the murine expression pattern (Southwood et al.,1996). An interesting aspect of the phylogenetic analyses was that the zebrafish Klf2a and Klf2b genes cluster with the EKLF subgroup rather than the KLF2 subgroup. This finding, along with that of a previous study (Oates et al.,2001), seems to suggest that assignment of nomenclature for the closely related zebrafish KLFs (namely Klfd, Klf2a, and Klf2b) with numbers for their mammalian counterparts may not reflect proper phylogeny. On the other hand, these zebrafish genes may have undergone several mutations from the common ancestral form and the nomenclature reflects orthology but does not reflect functional conservation.
Between human and mouse EKLF, the zinc finger domain is 90.1% similar. Interestingly, the chicken EKLF zinc finger domain is more similar to human EKLF (91.4%) than mouse is to human. This stronger similarity in the zinc finger domains of chicken and human KLFs was also observed for KLF2, KLF5, and KLF12 (Basu et al.,2004), and suggests similar functional activity.
It was initially thought that EKLF is not required for primitive erythropoiesis in the yolk sac but is required only during fetal liver erythropoiesis, for maturation of definitive erythroid cells (Nuez et al.,1995; Perkins et al,1995; Wijgerde et al.,1996). It was intriguing, therefore, that EKLF is expressed during mouse primitive erythropoiesis, starting early in embryonic development (Southwood et al.,1996). Recently, two separate studies have shown that EKLF plays an essential role in hemoglobin metabolism and membrane stability in both primitive and definitive erythroid cells (Drissen et al.,2005; Hodge et al.,2005). With a view to confirm that this tissue-specific expression pattern was conserved across species, we examined the expression pattern of chicken EKLF mRNA during early development. Southwood et al. (1996) showed that murine EKLF mRNA is first expressed during the neural plate stage (embryonic day 7.5), within primitive erythroid cells at the beginning of blood island formation in the yolk sac. Similarly, abundant cEKLF expression is observed in the blood islands, beginning at stage 5 and continuing through at least stage 13. cEKLF was also expressed in the circulating blood cells in early development and at days 5–14, as expected. There are, however, two differences in the observed expression patterns of EKLF in chicken and mouse embryos. First, Southwood et al. (1996) did not observe expression of EKLF before the formation of yolk sac blood islands in the mouse. In contrast, we observed expression of EKLF at stage 4 in the chicken, suggesting that EKLF may be present in very early mesodermal cells, before their migration to the yolk sac blood islands. This may not represent an actual distinction between mouse and chicken EKLF expression but, rather, a difference in the sensitivity of the procedures used. Second, cEKLF is expressed in the neural fold and neural plate at stages 7 and 8, and no neural expression was observed in mouse. This finding could possibly be indicative of an additional functional role of EKLF in chicken compared with mouse. However, this expression is transient, because cEKLF expression was not detected later in the developing chicken brain, and was restricted to the erythroid cells.
The similar expression pattern of cEKLF suggests that it has a function similar to mammalian EKLF. Our studies demonstrate that cEKLF is a marker for the primitive erythroid lineage in chicken. Finally, the expression pattern of cEKLF further supports a role for EKLF in primitive erythropoiesis.
RNA and DNA Isolation and Preparation of cDNA
Total RNA was isolated from chicken blood cells extracted from 5-, 7-, and 14-day embryos. RNA was isolated from the blood cells using TRIzol reagent (Invitrogen) according to the manufacturer's instructions, except that the extraction step was repeated two additional times. RNA was also isolated from brain samples taken from 5-, 7-, and 14-day chicken embryos. Solution D was added to the brain samples (Chomczynski and Sacchi,1987), vortexed until the tissues dissolved, and then total RNA was isolated using TRIzol reagent. The blood and brain samples were then subjected to DNAse I treatment using the DNA-free kit (Ambion). cDNA was made using the iScript cDNA Synthesis Kit (Bio-Rad). Genomic DNA was extracted from the same chicken embryos that were used to obtain blood and brain samples.
Data Mining and Bioinformatics
The mRNA sequences for the erythroid-specific proteins EKLF (murine) and Neptune (Xenopus) were used to search chicken genomic (NCBI, http://www.ncbi.nih.gov/blast; UCSC Genome Browser, http://www.genome.ucsc.edu) and EST (BBSRC ChickEST Database, http://chick.umist.ac.uk) databases. The sequences obtained were screened to ensure that they shared greater homology to EKLF and Neptune than to other KLFs. Sequences representing different parts of the same gene were aligned to build contiguous sequences.
PCR and Sequencing
Primer sets were generated based on the contiguous sequences (PRIME, GCG Wisconsin package), and PCR reactions were carried out for each template (genomic DNA or cDNA from 14-day blood RNA) using standard reaction conditions. PCR products from multiple reactions with each of 10 primer combinations were pooled, then purified and concentrated using the QIAquick PCR Purification Kit (Qiagen) per the manufacturer's instructions. Amplicons of the cDNA and genomic DNA were sequenced from both ends using the primers generated for PCR (available on request) in an ABI Prism 3700 DNA Analyzer (Applied Biosystems). After sequencing, the genomic and cDNA sequences were compared and the exon–exon junctions were identified.
Real-time PCR was used to examine the expression of ρ- and βA-globin and chicken EKLF at 5, 7, and 14 days of development in both blood and brain samples. For chicken EKLF mRNA, reverse transcriptase (RT) -PCR Mix with SYBR Green (Applied Biosystems) was used per the manufacturer's instructions, except the total reaction volume was 25 μl. Amplification was performed on an ABI Prism 7900HT (Applied Biosystems) using standard cycle parameters. Melting points of the PCR products were calculated during a final dissociation step, which verified amplicon specificity. Primer sequences are as follows (f and r denote position of forward and reverse, respectively, primers on sequenced genomic DNA contig for cEKLF or mRNA sequence for cGAPDH): cEKLF-1 (f33/r178), 5′-CTCCATCTCCACCTTCACCAAC-3′ and 5′-GGATGAAGTCCAAATCCACGAA-3′; cEKLF-2 (f1365/r2252), 5′-GGCAAGACGTACACCAAGAG-3′ and 5′- ATGTGCCGCTTCATGTGCAG-3′; cGAPDH (f180/r537), 5′-GATTCTACACACGGACACTTC-3′ and 5′-TAAGACCCTCCACAATGCC-3′. For measurement of ρ- and βA-globin mRNA, reactions were performed as for chicken EKLF mRNA. The primer sets used were as follows: ρ-globin (f83/r197), 5′-CCAGCGTCTGGAGCAAAGTC-3′ and 5′-CTGGAGAGGTTCCCGAAGTTATC-3′; βA-globin (f332/r429), 5′-CTTCTCCCAACTGTCCGAACTG-3′ and 5′-GGGCGGCCAGGACAAT-3′. Values for each time point–tissue combination were normalized to chicken GAPDH. The amount of expression of cEKLF was calculated relative to expression in 14-day blood samples, which was set equal to 1. For the RT-PCR experiments, each data series represents the average of 12 measurements (2 primer sets, 3 repeats each, 2 trials). For the chicken globin experiments, total β-like globin mRNA was set to 100%, and ρ- or βA-globin mRNA is expressed as percentage of the total. Statistical analysis was performed using the Student's t-test, and the results were considered significant at P < 0.05.
In Situ Hybridization Studies
A 794-bp fragment of cEKLF was generated by PCR using 14-day blood cDNA with the following primers: cEKLF-(f591/r1384), 5′-CCCAATTACCCTCTGCCTGA-3′ and 5′-CTCTTGGTGTACGTCTTGCC-3′. The PCR product was digested with NotI and PstI, and a 544-bp fragment, excluding the zinc finger domain, was cloned. Whole-mount in situ hybridization using digoxigenin-labeled RNA probes was performed as described previously (Streit and Stern,2001). Hybridization and posthybridization washes were done at 68–70°C.
We thank Mohua Basu and Megan Smith for excellent technical assistance, and Dr. Thanh Giang Sargent for technical advice on dissecting chicken embryos. We also thank Dr. Jack L. Haar and Dr. Gordon D. Ginder for critical evaluation of our work and invaluable advice.