xFoxJ1 and xFoxK1, two novel members of the fork head/winged helix family of transcription factors in Xenopus, have been identified by sequence homology to their mammalian orthologues. We here report the complete amino acid sequences of xFoxJ1, its pseudo-allelic version xFoxJ1′, and of xFoxK1, the latter also comprising a fork-head-associated (FHA) domain. We have further analysed the temporal and spatial expression of both genes. xFoxJ1 transcripts were detected during gastrulation in the dorsal blastopore lip and in the animal half. A predominant feature of xFoxJ1 expression until the tailbud stage is its spotty pattern within the epidermis. xFoxK1 is maternally expressed and localized within the animal half during early cleavage stages. Zygotic transcription starts during neurulation and is initially restricted to neuroectoderm. As development proceeds, xFoxK1 is expressed in ectoderm- and mesoderm-derived tissues, like the branchial arches, brain, eye, otic vesicle, pronephros, somites and abdominal muscle precursors.
Fork head/winged helix proteins represent transcription factors that have been found in a wide variety of eukaryotic organisms ranging from yeast to mammals. An evolutionarily conserved DNA-binding domain (Forkhead box: Fox) defines the members of this family. Fox proteins are not only known as important regulators during early development but are also involved in many differentiation events at later stages of ontogenesis.
With the re-naming of this transcription factor family to the unified Fox nomenclature (Kaestner et al. 2000), the entire family was divided into subgroups now containing the 17 subgroups designated A to Q. Members of all 17 subgroups have been identified in mammals, representing altogether about 50 different genes, while from the South African clawed frog, Xenopus laevis, only about 20 genes belonging to just 10 subclasses are known so far.
The sequence information available for human and, in part, also for mouse enabled us to search for missing members of the Fox family in Xenopus by querying the Xenopus EST databases with the sequences of the mouse or human homologues. The EST sequences obtained were amplified via PCR and used as probes to screen cDNA libraries of X. laevis embryos or to generate primers for RT-PCR.
Screening a X. laevis tadpole stage cDNA library led to the isolation of xFoxJ1 and the closely related pseudo-allelic variant xFoxJ1′ (sequences are deposited under EMBL accession numbers AJ609390 and AJ609391). The derived amino acid sequences are shown in Fig. 1. The sequences are 88.4% identical at the DNA and 94% at the amino acid level, respectively. xFoxJ1 and xFoxJ1′ differ inside their fork head domain by only one amino acid. A very short PCR fragment encoding part of the fork head domain of xFoxJ1 has already been described as XFKH5 (Dirksen and Jamrich 1995), but the published amino acid sequence accounts for only a small part of the complete protein. Orthologues of FoxJ1 are known from rat (L36388, Hackett et al. 1995), mouse (L13204, Clevidence et al. 1993) and human (X99349, Murphy et al. 1997; U69537, Pelletier et al. 1998). The complete proteins share 58.2% identity between Xenopus and rat, 61.2% between Xenopus and mouse, and the two human sequences FOXJ1a and FOXJ1b are 58.6% and 61.6% identical to xFoxJ1, respectively.
xFoxK1 has been amplified from tadpole stage RNA by RT-PCR using primers designed from the BC046369 IMAGE clone, because its isolation from X. laevis cDNA libraries had failed. The derived amino acid sequence is shown in Fig. 2. Homologues of xFoxK1 are known from different species, but the nomenclature is not yet clear. The isoforms of human ILF (interleukin-2 enhancer binding factor, X60787; Li et al. 1991; NM_004514; Nirula et al. 1997) seem to be the orthologues of Xenopus FoxK1. Two mouse ILFs are only distantly related to the human sequences, whereas two additional genes in mouse are quite similar to xFoxK1. The first is most closely related to xFoxK1 and probably represents the mouse orthologue, but the only information available is its sequence (accession number XM_126489). The second homologue, which occurs in two splice variants, was termed MNF (myocyte nuclear factor; NM_010812, U95016, L26507; Bassel-Duby et al. 1994) and is expressed in myocyte precursor cells in adult mice. This second homologue has recently been termed foxk1 (Hawke et al. 2003). However, based on a comparison with the full-length sequences from Xenopus and human, it obviously does not represent a FoxK1 orthologue. As shown in Fig. 2, xFoxK1 protein shares the highest identities with the human ILF sequences (79%) and the mouse XM_126489 sequence (75.6%), whereas the mouse MNF shows only 58.2% identity to Xenopus FoxK1. Moreover, the fork head domain, which is virtually identical among Xenopus, mouse and human FoxK1 (only one amino acid is divergent in the human sequence), exhibits ten amino acid substitutions in comparison to mouse MNF. Interestingly, six of these ten amino acid exchanges are shared by a gene of the ascidian Ciona intestinalis (cieg034p17; Satou et al. 2002; only the fork head domain is shown in Fig. 2). This suggests that the mouse MNF and the Ciona gene may compose a second class of FoxK.
The N-terminal portion of the FoxK1 proteins exhibits a conserved FHA-domain (fork-head-associated). This domain, which is not only present in Fox proteins, contains phosphoprotein-binding modules found in diverse signalling proteins that bind partners phosphorylated on threonine or serine. It is suggested to be involved in DNA repair, cell cycle regulation, or pre-mRNA processing (e.g. Lee et al. 2003; Li et al. 1999).
The temporal expression patterns of xFoxJ1 and xFoxK1 during embryonic development were analysed from stage 2 to stage 39 using RT-PCR (Fig. 3). While no maternal transcripts of xFoxJ1 can be detected, zygotic expression begins during gastrulation, increases continuously until stage 26 and then becomes slightly down-regulated during the following stages. xFoxK1 is expressed throughout all developmental stages analysed. Maternal transcripts persist during early cleavage stages, but their amount decreases with the onset of gastrulation. With the start of neurulation, the level of xFoxK1 expression increases and is then maintained throughout all stages analysed.
The spatial expression of FoxJ1 was analysed using whole-mount in situ hybridisation (Harland 1991) with the complete cDNA as a probe (Fig. 4). Our results support previous findings, which were obtained with a short PCR fragment (Dirksen and Jamrich 1995). Briefly summarised, expression starts in the dorsal lip at stage 11, and shortly after continues in the animal cap in a spotty pattern (Fig. 4A, B). Due to the movements of gastrulation, the stained cells of the blastopore lip become located in the dorsal midline (Fig. 4C). A transverse section shows that they are restricted to the neuroectoderm (Fig. 4D). The epidermal cells expressing xFoxJ1 continue to cover the complete surface; however, the total number of stained cells appears to remain constant, such that as the embryo grows the spotted pattern becomes more dispersed (Fig. 4E, F). The spots were previously suggested to be due to expression in ciliated cells of the epidermis. During gastrulation in the mouse, foxj1/Hfh-4 is expressed transiently in the node. Hfh-4 null-mice show severe defects in ciliogenesis and the determination of the left-right axis (Brody et al. 2000). Furthermore, Hfh-4-null mice were missing the 9+2-type cilia typical of epithelial cells, but not the 9+0-type of cilia typical of neuroepithelial cells.
Consistent with its temporal expression profile, xFoxK1 RNA is already detected at early cleavage stages by whole-mount in situ hybridisations (Fig. 5A). The amount of transcripts decreases during gastrula stages. In the neurula (Fig. 5B), xFoxK1 transcripts are localised in a distinct stripe at the dorsal midline. The transverse section shows that staining is restricted to the future floor plate (Fig. 5C). Faint expression is visible in the neural field encompassing the neural crest progenitor cells. With ongoing development, overall staining becomes stronger in the neural field, and as they migrate out of the neural field it is apparent that the neural crest cells express xFoxK1 (Fig. 5D). At stage 24, xFoxK1 is expressed in the eye, brain, branchial arches and in the presomitic mesoderm in the posterior part of the embryo. With further differentiation of the embryo, additional staining is found in the pronephric tubules (stage 29, Fig. 5F). The lateral muscle precursors of the abdomen start to migrate ventrally across the gut at stage 35, and these cells express xFoxK1 during migration (Fig. 5G). Additionally, the developing proctodeum and structures in the head continue to express xFoxK1. Head structures expressing xFoxK1 include the branchial arches, eyes and otic vesicles (Fig. 5H, I). xFoxK1 expression also persists in the nephros.
The spatial expression of the mouse and rat FoxK1 orthologues is largely unknown. On the other hand, expression of the mouse MNF gene is well characterised and has been shown to play a crucial role in the differentiation of myogenic stem cells. In a knockout mouse, myogenic progenitor cells are reduced and regeneration following injury is severely impaired (Hawke et al. 2003). Since the related xFoxK1 gene is also expressed in the somitic and abdominal muscle precursors, it will be interesting to analyse whether a morpholino knock-down of xFoxK1 will also effect cell cycle regulation of myogenic progenitor cells in the frog.
We thank G. Schlosser and K. Astrahantseff for helpful discussions of the manuscript and K. Botzenhart and C. Donow for skilful technical assistance. This investigation was aided by grants from the Deutsche Forschungsgemeinschaft (SFB 497/A 3) and by Fonds der Chemischen Industrie.