Glass is a zinc-finger transcription factor that is required for the differentiation of all photoreceptor cells. It is expressed in all cells behind the morphogenetic furrow; however, glass-dependent gene transcription is restricted to only the R cells. Since onecut probably participates in late differentiation events and is expressed exclusively in all R cells, it was of interest to investigate if its expression in the developing eye is dependent on glass. Immunostaining of third instar larval eye discs from a glass mutant, with Onecut antibodies reveals that onecut expression is not affected. It is interesting to note that the expression of several other photoreceptor-specific genes that are required for proper differentiation of R cells, for example, Orthodenticle (Otd), a homeodomain protein, and Calphotin, a calcium-channel protein, are also independent of Glass. This observation suggests that glass may regulate only some aspects of neuronal differentiation in the eye. Indeed, in null glass mutants, the expression of some neural-specific antigens such as those recognized by monoclonal antibody 22C10 and anti-HRP antibody is not affected. Thus, onecut is not downstream of glass, but may act in a parallel regulatory pathway in the control of photoreceptor cell differentiation (Nguyen, 2000).
Electrophoretic mobility gel shift assays (EMSAs) were carried out to address the DNA-binding properties of Onecut. For EMSA, the putative DNA-binding domain of Onecut synthesized and purified as the carboxyl half of the protein (aa 731-1081), which includes the cut domain and homeodomain fused to GST (GST-CHD). Binding was tested to the RCSI elements, a sequence element found in all rhodopsin promoters, from the rh2, rh3, and rh4 genes, whose sequences are similar but not identical (RCSIs from rh1 and rh4 are the same) (Fortini, 1991). All three RCSI probes bind strongly to GST-CHD. Increasing the amount of unlabeled double-stranded oligonucleotides effectively competes and reduces these binding interactions, demonstrating that Onecut specifically recognizes these RCSI promoter elements (Nguyen, 2000).
There are two lines of evidence to suggest that Onecut may have similar DNA-binding properties as Onecut proteins from other species: (1) the amino acid sequences of the cut domain and homeodomain are highly conserved; (2) although much more divergent than the mammalian and Drosophila Onecut proteins, both of the C. elegans proteins, Ceh-39 and Ceh-21, have been shown to bind to mammalian target DNA sequences (Lannoy, 1998). In light of these findings, the ability of Drosophila Onecut to bind to previously described HNF-6 probes (Lannoy, 1998; Lemaigre, 1996; Rausa, 1997; Samadani, 1996) was tested. The sequences are derived from the promoters of the TTR (transthyretin), HNF-3beta, and PFK-2 (6-phosphofructo-2-kinase) genes. GST-CHD fusion protein binds equally well to all three mammalian promoter sequences and can be competed off with the corresponding unlabeled oligonucleotides. As a further demonstration of the specificity of this interaction, a mutant HNF-3beta probe was found to be unable to form any DNA-protein complexes with GST-CHD, nor was it able to compete with the wild type probe in the binding reaction. Thus, in vitro, the cut-homeobox DNA-binding domain of Drosophila Onecut behaves very similar to other Onecut proteins in recognizing the same set of target binding sites (Nguyen, 2000).
This raises an interesting question: is there a common recognition sequence within these cis-acting sequences that confers DNA-binding specificity to Drosophila Onecut and other Onecut proteins? Previous studies on HNF-6 have suggested a consensus sequence from an alignment of all oligonucleotide sequences that bound (Lannoy, 1998; Lemaigre, 1996; Samadani, 1996). An obvious core motif ATTG is shared by both the Drosophila and mammalian sequences used in this study. Two lines of evidence lend support to this finding: (1) oligonucleotide sequences that do not bind to Drosophila Onecut do not contain the tetranucleotide sequence, as is the case for the RUS4A promoter element from the rh4 gene; (2) when changes are made to nucleotides within the ATTG motif, as in the mutant HNF-3beta probe, DNA-binding activity is completely abolished (i.e. mutant probe cannot compete with wild type probe for binding. It is interesting to note that some of the oligonucleotide sequences also contain multiple ATTG sequences and/or the general homeobox recognition sequence ATTA (Nguyen, 2000).
rh gene expression is regulated by multiple cis-acting regulatory elements (Fortini, 1991). Upstream of the Rh1 RCSI sequence lies a proximal enhancer Rh1PE that is bound by Glass, a zinc-finger transcription factor required for rh1 gene expression in photoreceptor cells (R1-6). Within the Rh1PE sequence there is an evolutionarily conserved ATTG tetra-nucleotide immediately downstream of a sequence protected by Glass binding. It was of interest to determine if Onecut binds to this rh1 enhancer element. Onecut (GST-CHD) does indeed bind strongly to the Rh1PE sequence, which includes a single ATTG repeat. This interaction is specific, as demonstrated by competition experiments with an unlabeled ds-oligonucleotide. However, when the ATTG core sequence is mutated (Rh1PE m1) DNA-binding is not completely abolished. This is in contrast to the ATTG mutation in the mutant HNF-3beta probe, which results in a complete loss of binding. This conflicting result suggests that there may be additional sequences required to potentiate the function of the ATTG in the Rh1PE. To substantiate this idea, a mutant probe was generated in the 5' flanking tetranucleotide (Rh1PE m2) and a double mutant probe encompassing the entire eight base pairs (Rh1PE m3). Surprisingly, mutations in the 5' flanking region reduce binding to a greater extent than mutations in the ATTG sequence. The double mutation greatly abolishes DNA-binding activity of both GST-CHD and GST-CD. Thus, the results point to additional flanking sequences within Rh1PE that are required for Onecut binding, and suggest that the sequence context surrounding the ATTG motif render a significant contribution to high DNA-binding affinity and specificity (Nguyen, 2000).
To test if the individual cut domain and homeodomain of Onecut can bind to DNA, fusion proteins GST-CD and GST-HD were used for EMSA analysis. Surprisingly, the homeodomain alone is unable to bind to either the Drosophila Rh1PE element or to the mammalian TTR, HNF-3beta, and PFK-2 binding sites. In contrast, evidence is provided that the cut domain alone can indeed bind to some target sequences with high affinity, such as the Rh1PE element, and the promoter elements from the TTR and HNF-3beta genes. However, the cut domain fails to form a complex with a binding site derived from the PFK-2 promoter. Interestingly, however, the GST-CHD (cut domain and homeodomain together) does bind to this site. This suggests that the DNA-binding activity of the cut domain, in this sequence context, is dependent on the presence of the homeodomain. The addition of the individual domains together in the binding reaction does not promote the formation of any ternary DNA-protein complexes. Taken together, these results imply the existence of cooperative cis-interaction between the homeodomain and the cut domain to effect DNA-binding (Nguyen, 2000).
The role of the homeodomain can be further illustrated by its influence on the behavior of the cut domain interaction with some target sequences. When mutations (m1) are introduced within the ATTG core sequence, the cut domain alone, surprisingly, binds much stronger in comparison to the intact cut-homeodomain fusion protein. The affinities of the cut domain for the TTR and HNF-3beta probes are decreased when compared to binding by GST-CHD, particularly for the HNF-3beta probe. Thus, the presence of the homeodomain clearly regulates some aspects of the cut domain DNA-binding affinity and specificity (Nguyen, 2000).
Thus, an alignment of the oligonucleotide sequences used in electrophoretic mobility gel shift assay (EMSA) analysis reveals an ATTG core motif that is common to all probes that bind to Onecut. This motif is included in a consensus sequence, WTATTGATTW (where W is A/T), previously defined for HNF-6 binding (Lannoy, 1998; Lemaigre, 1996; Samadani, 1996). In some cases, there is a strong requirement for this tetranucleotide since binding is completely abolished when this sequence is mutated. However, in another sequence context there appears to be additional sequence requirement in addition to the ATTG core motif. For example, mutant Rhodopsin1 promoter (RH1PE) probes with an intact ATTG show a dramatic reduction in binding to GST-CHD, but are still highly effective in binding to the cut domain alone. Only when both the flanking sequences and the ATTG core are changed is the DNA-binding abolished. This suggests that in addition to the ATTG core, the flanking sequences in Rh1PE also contribute significantly to the DNA-binding activity of Onecut. Indeed, a mutation in a flanking nucleotide of the ATTG motif has been identified in the HNF-6 binding site within the Type I protein C promoter that reduces HNF-6 binding and abolishes transactivation of a reporter gene in tissue culture (Spek, 1998). Likewise, the cut repeats of the Cut proteins could also bind to other sequences, as well as the major CCAAT binding site (Nguyen, 2000).
The observations regarding binding characteristics of the mutant Rh1PE probes may suggest an interesting possibility that Onecut binding may require an initial docking event after which the cut domain would scan for the ATTG motif. The homeodomain may play a role in regulating these processes and stabilizing the interaction. In this view, mutations in the flanking sequences but not in the ATTG could affect the initial docking of Onecut. Mutations in the ATTG motif but not in the flanking sequences would still allow for docking (Nguyen, 2000).
The presence of two DNA-binding motifs in the Onecut protein and its nuclear localization provide strong evidence that Onecut transcriptionally regulates gene expression. Since the mammalian Onecut proteins, HNF-6 and OC-2, have been shown to be transcriptional activators (Jacquemin, 1999; Lannoy, 1998; Lemaigre, 1996; Rausa, 1997), it was of interest to determine the transcriptional property of Onecut in the Drosophila Schneider S2 cell line by transient transfection assays. A onecut expression vector was constructed under the control of the copia LTR (pDOC-Copia), which provides constitutive expression; also prepared was a luciferase reporter construct driven by a Drosophila minimal hs43 promoter (pLuc-hs43) with or without six tandem copies of the Rh1PE enhancer element (pLuc-6XRh1PE). During the course of the experiment, it was found that the S2 cell line appears to express a low level of endogenous Onecut, since transfection of the pLuc- 6XRh1PE reporter vector alone gives about a 3-fold induction of luciferase activity over the empty pLuc-hs43 vector. Cotransfection of the onecut expression vector pDOC-Copia with either pLuc-hs43 or the pLuc-6XRh1PE reporter vector gives an approximately 2.5-fold or 5-fold induction, respectively, of reporter gene activity over the basal level. This level of induction is comparable to those reported for HNF-6, which can induce a 3-4-fold level of reporter gene activity under the control of the PFK- 2 L promoter in hepatoma FTO-2B cells (Lemaigre, 1996). It is interesting to note that the pDOC-Copia vector carrying the minimal hs43 promoter also generates some induction of luciferase activity. A closer inspection of the hs43 promoter reveals sequences containing the ATTG motif that may favor low affinity binding to Onecut. Nevertheless, under the transfection assay condition used, Onecut appears to function as a moderate transcriptional activator (Nguyen, 2000).
Home page: The Interactive Fly © 1997 Thomas B. Brody, Ph.D.
The Interactive Fly resides on the
Society for Developmental Biology's Web server.