We previously identified binding of TCF to a region in the nkd
intron that corresponds to IntE (Fang et al., 2006
; Parker et al., 2008
). In this report, we find that TCF is also highly enriched on the chromatin about 10 kb upstream of the nkd
TSS (UpE) when the Wg pathway is chronically activated by Axin
RNAi (). Wg signaling also increases the binding of TCF to the region containing IntE (), but this effect occurs within a few hours of pathway stimulation (Fang et al., 2006
; Parker et al., 2008
). The increase in TCF-binding in the UpE region is not due to increased TCF, since the expression of TCF and its nuclear localization are not affected by Axin
RNAi (Chang et al., 2008
). Rather, we postulate that widespread histone acetylation at the nkd
locus upon Wg stimulation (Parker et al., 2008
) allows subsequent recruitment of TCF to the UpE region in Kc cells.
Further analysis of the UpE region with reporter gene assays revealed the presence of two overlapping stretches of DNA (UpE1 and UpE2) that confer a high degree of responsiveness to Wg signaling (). Mutagenesis of two to three predicted TCF binding sites in UpE1, UpE2 and IntE largely abolished their ability to respond to Wg signaling (). While this suggests that TCF sites within a WRE act in a redundant manner, mutation of individual sites indicates that some sites contribute more than others. For example, in UpE2, the TCF 2 and TCF 3 sites share the same sequence (Supplemental Fig. 1A
), but mutation of TCF 2 reduces Wg activation while mutation of TCF 3 does not (Supplemental Fig. 2A
). These data suggests that the exact sequence of the TCF binding site is likely not as important as the context in which they are located within the WRE.
When tested in flies, UpE1, UpE2 and IntE are all activated by Wg signaling in several tissues in patterns that partially recapitulate that of the endogenous nkd gene (, & ). In the leg and eye imaginal discs, all three WREs are active. In addition, each WRE has unique tissue-specific activities. For example, in later embryogenesis IntE is the only WRE that is active. Even though UpE1 and UpE2 share more than 400 bp of sequence (), only UpE2 is active in the embryonic epidermis. UpE1 and UpE2 are both active in the wing and antennal imaginal discs, (IntE shows no or minimal expression in these tissues), but UpE1 is expressed strongly in the notum whereas UpE2 is not. Each WRE is active in multiple tissues but also contains information that confers tissue-specific Wg responsiveness.
The basis for the tissue specificity of the various nkd
-WREs is not clear at present. It could be that different TCF sites within each WRE are utilized in different tissues. However, our data in cell culture argue that multiple TCF sites are required in each WRE in a partially redundant manner (; Supplemental Fig. 2
). In addition, the same TCF sites that are required for UpE2 and IntE activity in Kc cells are required for WRE activation in all tissues examined (; data not shown). Therefore, we favor the view that the tissue specificity for the different WREs is derived from the presence of other cis-acting elements that work with the TCF sites to allow activation by Wg signaling.
When all three WREs are placed within a single reporter construct the resulting pattern largely recapitulates that of the endogenous nkd
gene in imaginal discs at the late third larval instar stage (; data not shown). However, we have not examined the regulation of our WRE reporters at earlier larval stages, where Wg is also expressed (Williams et al., 1993
; Neumann and Cohen, 1996
). Even at the late larval stage examined, the pattern of the reporters does not completely match that of endogenous nkd
in the wing disc. Wg is expressed in a double ring pattern in the hinge region (; the proximal ring indicated by the arrow) and nkd
transcripts are also found in a double ring (; arrow). However, expression of WRE reporters in the proximal ring is weak () and often not present (). This suggests the existence of at least one other WRE for the wing imaginal disc.
In the embryo, it is even more obvious that additional regulatory information for nkd expression remains to be identified. In the embryonic epidermis the pattern of the combined WRE construct is only a subset of the endogenous nkd pattern and is equal to the sum of the IntE and UpE2 WREs (; data not shown). This suggests the presence of at least one other WRE that is active in the embryonic epidermis. Consistent with this, deletions of genomic fragments containing IntE or UpE do not affect the expression of nkd in embryos or the viability of the animals when heterozygous with a nkd deficiency (data not shown).
In the wing and leg imaginal discs, loss of UpE results in a significant decrease in nkd
transcript levels (). In contrast, the IntE deletion had nkd
expression in the normal range (). Even with the large UpE deletion, there is still significant nkd
expression in the wing disc (41% of wild type; see Results
). These data could be evidence for redundancy between IntE and UpE in these tissues. It is also possible that additional WREs exist that contribute to imaginal disc expression, which are still present in the IntE and UpE deletions.
Our data indicate that nkd does not contain a universally responding WRE that is activated by Wg signaling in all tissues. Rather there are at least several WREs that can respond to the pathway in multiple, overlapping tissues. It appears that only limited multi-tissue responsiveness can be obtained with any individual WRE. In the absence of a universal WRE, the strategy of having multiple WREs responding to Wg signaling in each tissue may be required to ensure the robustness of the Wg-Nkd feedback circuit. Whether this is the case for the regulation of other Wnt feedback antagonists or those acting in other signaling pathways remains to be determined.
The finding that the nkd locus does not contain a universal WRE raises the question of how the Wg-Nkd relationship could remain intact during animal evolution as the Wg expression pattern became more elaborate. We postulate that the existence of several WREs with broad tissue specificity could have ensured that when an enhancer evolved that expressed Wg in a new location, at least one of the existing nkd WREs would be able to respond to the pathway in that tissue. This precludes the need to have a tissue-by-tissue de novo synthesis of nkd WREs every time Wg was expressed in a new pattern. Retaining the feedback inhibition of Wg signaling by Nkd may have allowed Wg to be used more readily during the diversification of animal body plans.