The Wingless signaling pathway controls various developmental processes in both vertebrates and invertebrates. Here I probe the requirement for nuclear localization of APC2 and Axin in the Wg signal transduction pathway during embryonic development of Drosophila melanogaster. I find that nuclear localization of APC2 appears to be required, but Axin can block signaling when tethered to the membrane. These results support the model where Axin regulates Armadillo localization and activity in the cytoplasm.
THE Wnt/Wingless (Wg) signaling pathway plays essential roles in the development of animals. Aspects of signaling regulate cell proliferation, differentiation, polarity, and survival. Importantly, Wg pathway components have been found to affect stem cell maintenance and tumor progression. The basic step in signal transduction through this pathway is the regulation of Armadillo/β-catenin (Arm/β-cat) levels. In the absence of extracellular ligand, a degradation complex consisting of the scaffold proteins Axin and APC and the kinases CKI and Zw3 (Shaggy, GSK3) forms. This complex mediates the phosphorylation of Arm, tagging it for proteasome-mediated degradation. Upon ligand binding, the receptor complex of Frizzled (Fz) and Arrow (LRP5/6, Arr) activates Disheveled (Dsh), which in turn inhibits the degradation complex leading to increased Arm protein levels. Arm in turn translocates to the nucleus where it activates transcription in conjunction with the transcription factor TCF (Logan and Nusse 2004).
Previous results suggested a model where ligand-mediated receptor activation recruited Axin to the membrane where it bound Arr in a larger complex with Dsh and Fz. It was further shown that Axin was the rate-limiting component, and its levels were regulated through proteasomal degradation in a signal-dependent manner (Lee et al. 2003; Tolwinski et al. 2003; Tolwinski and Wieschaus 2004b). Other approaches showed that the membrane localization of the degradation complex was a key step in signaling as GSK3 and CKI potentiated signaling at the membrane by phosporylating LRP (Davidson et al. 2005; Zeng et al. 2005). Other, somewhat contradictory findings were that Axin was required in the nucleus for pathway activation (Cong and Varmus 2004; Wiechens et al. 2004). Another important finding was that APC, a component of the degradation complex, was shown to function in nuclear export of Arm (Rosin-Arbesfeld et al. 2000), and in opposing transcriptional activation by the Arm/TCF complex (Sierra et al. 2006).
Here I report that membrane-tethered Axin is sufficient for the proper transduction of Wg signal in the patterning of the Drosophila embryo. In contrast, membrane-tethered APC2 does not rescue signaling.
MATERIALS AND METHODS
Crosses for the third chromosome (Ch.) mutants:
(Ch. 3-left arm) da-GAL4, (Ch. 3-right arm) FRT 82B, axinS044230/TM3
(3L) da-GAL4, (3R) FRT 82B, apc1Q8, apc2d40/TM3
(3L) da-GAL4, (3R) FRT 82B, apc1Q8, apc2d40, axinS044230/TM3.
These lines were crossed to FRT 82B, OvoD males resulting in females that made only maternally mutant eggs of which 50% have da-GAL4. This is because the recombination only occurs on 3R (Xu and Rubin 1993; Chou and Perrimon 1996). The resulting females were crossed to males:
(Ch. 2) UAS-axin; (Ch. 3) mutant*/TM3 balancer
(Ch. 2) UAS-myr-axin; (Ch. 3) mutant*/TM3 balancer
(Ch. 2) UAS-apc2; (Ch. 3) mutant*/TM3 balancer
(Ch. 2) UAS-myr-apc2; (Ch. 3) mutant*/TM3 balancer.
Crosses for the first or X chromosome mutants:
y, zw3M11-1 FRT 101/FM6 (Ch. 1); arm-GAL4 (Ch. 2).
This line was crossed to OvoD, FRT101 males resulting in females that made only maternally mutant eggs of which 50% have arm-GAL4. These females were crossed to homozygous UAS transgenes as above. In all cases, the GAL4/UAS system was used to express transgenes (Brand and Perrimon 1993).
Genotypes were identified unambiguously as follows. The daughterless-GAL4 insertion is located on the left arm of chromosome 3. Therefore, when germline clones are made for the right arm of the third chromosome, all embryos are maternally mutant but only 50% express GAL4. Similarly for Ch. 1, the driver assorts independently of the mutant chromosome as it is on Ch. 2. Maternal and zygotic (M/Z) mutants for zw3, axin and the triple mutant are completely naked, whereas maternal (M) only mutants retain a small number of distinct denticles making them readily identifiable. Also, the UAS transgene was zygotically crossed in as homozygous. Combining these facts, 50% of embryos express the transgene, and I observe that 50% show an effect depending on the transgene used irrespective of whether they are zygotically mutant (see numbers in Table 1). The remaining 50% can be classified according to phenotype into half M/Z and half M depending on whether small numbers of denticles are present or absent. Having these two classes each present at 25% shows that the embryonic patterning defects are mostly due to loss of the maternal contribution of these genes rather than on the zygotic contribution, an effect also observed in Hamada et al. (1999) and Peterson-Nedry et al. (2008).
The apc double mutant differs only in that, although M/Z mutants are completely naked similar to the others discussed above, the M only embryos are fully, paternally rescued and hatch. Still, as in the above case, 50% express the transgene, and therefore it is simply a matter of counting the four classes that result (Ahmed et al. 2002). Embryos were collected at 25° before being dechorionated and mounted in Hoyer's media. The results of one representative experiment that was repeated multiple times are quantified in Table 1.
Myristoylated constructs were made by adding a sequence identical to the NH2 terminus of src (MGNKCCSKRQGTMAGNI) to the NH2 teminus of both axin and apc2 by PCR. This sequence has proven to be very effective for membrane targeting of arm (Zecca et al. 1996; Tolwinski and Wieschaus 2001; Tolwinski and Wieschaus 2004a). The PCR products were then transferred by Gateway cloning (Invitrogen) into a pUASt with COOH-terminal 3XFLAG tag vector (http://www.ciwemb.edu/labs/murphy/Gateway%20vectors.html), and injected by standard methods. I used full-length axin constructs kindly provided by Karl Willert and Roel Nusse (Willert et al. 1999) and UAS-apc2 constructs that were kindly provided by Mariann Bienz.
RESULTS AND DISCUSSION
Mutations in axin result in constitutive activation of the Wg pathway, or the complete absence of patterning of the ventral embryonic epidermis—the naked phenotype (Hamada et al. 1999; Willert et al. 1999). In contrast, overexpression of axin leads to complete lack of Wg signaling and loss of patterning or the uniform denticle—the wg phenotype (Hamada et al. 1999; Willert et al. 1999). In axin mutant embryos, I assayed the ability of axin that was tethered to the membrane to rescue the naked phenotype. As shown in Figure 1B, this membrane-bound form can rescue patterning to some extent. In contrast, an unmodified axin, leads to complete loss of patterning and the wg phenotype Figure 1A.
There are two genes that encode apc in Drosophila, apc 1 and 2. They are largely redundant, and only double mutants lead to strong pathway activation in embryos (Ahmed et al. 2002; McCartney et al. 2006). In apc1 apc2 double mutants, I assayed the ability of membrane-tethered apc2 to block signaling activation. As shown in Figure 1H, there was no real effect on the patterning of the cuticle. In contrast, expression of untethered apc2 rescued to a wild-type cuticle pattern (Figure 1G). I further used the apc1 apc2 double mutants to test the activity of both membrane-tethered and -untethered axin. Interestingly, expression of axin abolished signaling brought about by loss of both apc genes leading to a completely denticle-covered cuticle (Figure 1E). The tethered axin, however, was unable to overcome the loss of apc and signaling was not blocked (naked phenotype, Figure 1F).
To further assay the activity of the membrane-tethered and -untethered apc2 and axin, I expressed all four in a triple mutant for axin apc1 apc2. In the triple mutant, only untethered Axin was able to block signaling (Figure 1I), as both membrane-tethered axin and apc2 and the untethered apc2 failed to restore patterning (Figure 1, J–L). This finding suggests that in the absence of APC, Axin at the membrane is insufficient to block signaling, perhaps because APC is not available to export Arm from the nucleus. Arm that is made in the cytoplasm may therefore elude the membrane bound Axin and activate signaling when APC isn't present to export it out of the nucleus.
The triple mutant could only be rescued by the expression of untethered axin, and further, expression both apc2 and tethered apc2 failed to rescue the signaling activation brought about by the loss of axin alone (Figure 1, C and D). These results taken together imply that in the complete absence of Axin, signaling cannot be blocked by APC in any form. This is supported by the fact that in zw3 mutant embryos expression of untethered axin blocked signaling, and tethered axin could do so minimally (Figure 1, M and N). Expression of either form of apc2 had no effect in zw3 mutants (Figure 1, O and P). All these combinations taken together suggest that only wild-type Axin can block signaling in all the other destruction complex mutants. Membrane bound Axin could only rescue axin mutants and to a much lesser extent zw3 mutants. Expression of apc2 only rescued apc mutants, whereas membrane bound apc2 could not rescue any of the destruction complex mutants or combinations.
Therefore, as predicted from the mutant studies above, expression of myr-axin in otherwise wild-type embryos is sufficient to block Wg signaling (Figure 2A), but myr-apc2 has minor effects (Figure 2B). Both constructs appear to be localized primarily to the membrane (Figure 2, C–H), but unfortunately I cannot exclude the possibility that our membrane-tethering system is at some low level leaky. However, on the basis of the fact that I do not observe either APC or Axin in the nucleus, that the constructs are functionally distinct, and my previous observations of the efficacy of this membrane-tethering sequence (Tolwinski and Wieschaus 2004a), it is likely that the function of these alleles is largely at the membrane.
Overall, these results suggest that the different members of the degradation complex perform different roles in signal transduction. Specifically, since UAS-axin blocks signaling in apc and zw3 mutants it appears to be epistatic to the other destruction complex components. From a biochemical perspective, Axin is thought to be the rate-limiting component as it appears to be present at ∼5000 fold lower levels in extracts from Xenopus oocytes (Lee et al. 2003), and its levels can be modulated in response to Wg signaling (Tolwinski et al. 2003). A recent study, however, showed that in fly eye development APC1 also exists at threshold levels and that this is required for proper graded responses to Wg signal (Benchabane et al. 2008). This is unlikely to be true for embryonic Wg signaling, as loss of APC1 has no effect on embryogenesis and appears to mainly affect eye development (Ahmed et al. 1998).
An important finding shown here is that Axin when tethered to the membrane cannot block signaling in the absence of APC. This suggests that either APC is required for nuclear export when Axin is absent from the cytoplasm and nucleus or Axin is required in the cytoplasm and not just the membrane for its Arm anchoring role (Tolwinski and Wieschaus 2001; Krieghoff et al. 2006). However, our finding raises a major conundrum in that expression of Axin alone can rescue the loss of Zw3. As phosphorylation of Arm by Zw3 is required for Arm proteasome-mediated degradation, loss of Zw3 leads to very high levels of Arm (Siegfried et al. 1994). In contrast, Zw3 phosphorylation has the opposite effect on Axin, stabilizing its levels and preventing its degradation (Yamamoto et al. 1999; Lee et al. 2003). Therefore, it is unclear how unregulated levels of Arm protein can be blocked from entering the nucleus by Axin expression unless the expression of Axin is at enormously high levels. This is unlikely since Axin levels should be lower when Zw3 is absent, leading to the conclusion that there must be another pathway that may target Arm for degradation, or the Arm protein present under these conditions somehow lacks activity or activation by an as yet unidentified component or components.
Finally, although these experiments may raise more new questions than they answer, it is important to note that they support much of what is believed to be the major form of pathway activation. Upon Wg binding to the receptor Fz, a complex involving Fz, Dsh, Arr, and the destruction complex forms. At this point Dsh somehow inactivates the phosphorylation of Arm and allows it to enter the nucleus where it is involved in transcription. APC's role appears to be twofold, as it is involved in the destruction complex and in the export of Arm from the nucleus. That this robust complex forms at the plasma membrane seems certain (Cliffe et al. 2003; Davidson et al. 2005; Zeng et al. 2005; Peterson-Nedry et al. 2008), but some unanswered questions remain such as how does Dsh inactivate the destruction complex and whether there is an as yet uncharacterized activation step for Arm.
Communicating editor: T. Schüpbach
- Received November 2, 2008.
- Accepted January 2, 2009.
- Copyright © 2009 by the Genetics Society of America