Assessing the roles of different Roc proteins in Arm regulation
A substantial amount of data support the idea that a Cullin1-based SCF complex with Slimb as the F-box protein regulates the targeted degradation of Arm/ßcat 
. Flies have three Roc proteins—Roc1a associates with Cullins 1–4, Roc1b binds Cullin3, and Roc2 binds Cullin5 
. However, although Roc1a is a canonical component of the Cullin1-based SCF complex 
mutant clones in larval wing discs do not accumulate Arm above wild-type levels, but do accumulate a different SCF substrate, the Hedgehog effector Cubitus Interruptus 
. Given these data, we set out to determine whether a different Roc protein in Drosophila
acts in the SCF complex, or if the three Rocs function redundantly in this process.
We first tested these hypotheses by analyzing Arm accumulation in embryos and larval tissues lacking Roc1b or Roc2. We examined null alleles of Roc1b
, a coding sequence deletion that is homozygous viable but male sterile 
) and of Roc2
, generated by P-element insertion, which is homozygous viable and fertile; 
). We verified the presence of both mutations by PCR (data not shown). Given the essential role of Wnt signaling, the viability of Roc1b
mutants suggests that neither is an essential part of the E3 complex targeting Arm, or alternately suggests that the Roc proteins act redundantly.
To directly assess whether loss of either Roc1b or Roc2 affect Arm levels, we immunostained three tissues from Roc1bdc3
mutants (since both are viable, we could examine whole animals rather than clones of mutant cells). As an internal control, we stained wild-type animals marked with Histone-GFP together with each mutant, and imaged them on the same slides using the same confocal settings. In wild-type embryos, Arm is found at the plasma membrane of all epithelial cells, as part of the cadherin-catenin complex. In cells not receiving Wnt signal, there is little Arm inside cells, as it is targeted for destruction (, arrowhead). Stripes of cells in each segment receive Wnt signals and accumulate Arm in the cytoplasm and nuclei (, arrow). In contrast, embryos lacking the destruction complex proteins APC1 and APC2 accumulate Arm at very high levels, much higher than even wild-type cells receiving Wnt signal (; 
). When compared to wild-type, neither Roc1bdc3
mutant embryos () or Roc2KG
embryos () showed elevated Arm accumulation. We also did not see elevated Arm accumulation in imaginal discs mutant for either Roc1bdc3
() or Roc2KG
(), or in larval brains mutant for either gene (). We also assessed Arm accumulation by immunoblot of protein from stage 9 embryos; Roc1bdc3
mutants have the same amount of Arm protein as wild-type (). Together, these data suggest that neither Roc1b nor Roc2 is essential for regulating Arm degradation.
An RNAi screen reveals SCF components regulating Arm stability in cultured Drosophila S2 cells
Together with the earlier work on Roc1a in imaginal discs 
, these data suggest that none of the three Rocs are individually essential for Arm degradation, even though they are thought to be the key RING finger proteins in Cullin-based E3 ubiquitin ligases. We thus broadened our search for proteins regulating Arm stability, using an RNAi screen in Drosophila
S2 cells. These cells are superb for this purpose: rather than having to design shRNAs and transfect them into cells, one simply adds ~500 bp double-standed RNAs (dsRNAs) to the medium, and the cells take these up and process them into siRNAs. In parallel with a directed RNAi screen for SCF components that regulate centrosome number in cultured Drosophila
S2 cells 
, we carried out a similar screen for proteins whose knockdown stabilized Arm. We examined the six fly Cullins, the seven fly Skp proteins, all three fly Rocs and a set of 42 F-box proteins. Cells were treated for 7 days with double-stranded RNA to each target protein in a multiwell format, and then fixed and stained for Arm, and also with Hoechst to label DNA to automate detection of individual cells. Plates then were scanned with an Array Scan V (Cellomics) automated microscope. Software was used to partition the field into cells, and images of 5000 cells per well were acquired and analyzed using vHCS View (Cellomics). This allowed us to quantitate Arm levels using average integrated fluorescence intensity (; several treatments reduced Arm levels—we did not pursue these further).
The results of our RNAi screen for SCF and E3 ligase components that alter Arm levels in Drosophila S2 cells.
Several genes scored positive for increased Arm levels. To follow-up these findings, these were examined more closely, by RNAi followed by immunoblotting for Arm. Cullin1, a core SCF complex component, was the only Cullin to score positive in the initial screen (). To followup, we repeated Arm immunoblots on cells treated with dsRNA to each of the five fly Cullins. Once again, Cullin1 was the only Cullin to score positive in the follow-up Western analysis, with knockdown elevating Arm levels (). For three of the Cullins, Cullins1, 4 and 5, we were able to use available antibodies to verify knockdown in the same samples used to assess effects on Arm levels (; these same controls were used to verify Cullin knockdown in our parallel screen for regulators of centrosome number 
). This result is consistent with previous work in vivo suggesting a role for Cullin1 
, but suggests that Cullin4, which has been reported to negatively regulate Arm/ßcat 
in other contexts, is not a key regulator in Drosophila
S2 cells. Among Skp proteins, only SkpA scored positive in the initial screen (). We did follow-up immunoblots for SkpA and SkpB; both scored positive for elevated Arm levels in this assay (). However, due to sequence similarity between the two, SkpB knockdown also reduced SkpA levels (). We suspect SkpA is the key player in vivo, as it is expressed at much higher levels than any of the other fly Skps 
. Alternately, SkpA and SkpB may regulate Arm levels redundantly. Together, these data add further support to the model in which the primary E3 ligase targeting Arm for destruction is a canonical SCF complex using Cullin1 and SkpA.
A canonical SCF complex including Roc1a regulates Arm levels in Drosophila S2 cells.
The canonical SCF complex also uses the RING finger protein Roc1, but previous analysis in imaginal discs suggested the fly Roc1 ortholog (Roc1a) does not play a role in Arm regulation in that tissue 
. However, in S2 cells our RNAi screen suggested Roc1a does play a role. RNAi of Roc1a
substantially elevated Arm levels in the screen (). Roc1a
RNAi also elevated Arm levels in the follow-up immunoblots (). In contrast, neither RNAi of Roc1b
alone elevated Arm levels in either assay (; ; Roc1b
RNAi reduced Arm levels as assessed in the screen, perhaps due to subtle effects on cell cycle progression). Triple RNAi of all three Rocs also elevated Arm levels to approximately the same levels as Roc1a
RNAi alone (). Because of the discrepancy with earlier experiments on Roc1a in vivo, we carried out an additional experiment to ensure that the elevation of Arm levels in response to Roc1a
RNAi was not due to an off-target effect of our original Roc1a
dsRNA. We designed several different dsRNAs to Roc1a
, including a pair of non-overlapping dsRNAs representing the 5′ and 3′ halves of the mRNA (). Each of these led to elevated Arm levels relative to the SK RNAi control (), consistent with our original result. Thus in S2 cells, Roc1a appears to be essential for Arm regulation, consistent with its known role in the SCF complex.
We tried several approaches to test whether Roc1a is essential for Arm degradation in the animal. One cannot make embryos maternally mutant for Roc1a as Roc1a is required for proliferation of germline stem cells 
. We generated clones of Roc1a
mutant cells in imaginal discs, but as was seen by Noureddine et al. (2002) 
, clones were infrequent and only comprised a few cells, and thus we could not effectively analyze Arm levels. We also tried using lines that were designed to allow in vivo Roc1a
RNAi. We tested both a line from the Vienna RNAi collection 
, expressing it in imaginal discs, and a line from the Valium 20 collection 
, expressing it maternally using the matGAL4 driver. Neither effort produced either a change in Arm levels or any apparent phenotype (data not shown), suggesting that neither significantly depleted Roc1a—we have observed this with other RNAi lines from these collections. In the future additional RNAi lines may prove more effective, allowing our hypothesis to be tested in vivo.
The striking difference between the clear role we found for Roc1a in Arm destruction in S2 cells, and the failure to find such a role in imaginal discs 
is consistent with two possibilities: 1) Roc1a may play a cell type specific role in Arm regulation, or 2) since loss of Roc1a is predicted to inactivate all
SCF E3 ligases, it may be that when clones of Roc1a
mutant cells are generated in imaginal discs 
, cells arrest due to effects on other target proteins before Roc1a levels drop severely enough to affect Arm regulation. Further work is needed to distinguish between these possibilities.
We next investigated which F-box proteins regulate Arm stability in S2 cells. Several F-box proteins scored at least marginally positive in our initial screen ()—we followed up each of these by repeating the RNAi and immunoblotting for Arm. The only F-box protein to score positive in both assays was the known Arm regulator Slimb (; all lanes except SkpA are fly F-box proteins; most remain genetically uncharacterized and thus are only known by their CG numbers). It is also worth noting that we saw no effect on Arm levels in this cell type in either the screen or the follow-up immunoblots with RNAi against Ebi (; ), an F-box protein previously implicated in ßcat stability in other cell types 
. Of course Ebi and other F-box proteins may play roles in Arm/ßcat stability in a cell type specific manner, but they do not seem to play a critical role in Arm regulation in S2 cells.
Is Armadillo regulation different in embryos and larvae?
Another issue in the current literature about machinery regulating Arm levels during normal fly development concerns whether all tissues use the same machinery. This issue was raised by apparent differences between accumulation levels of Arm in embryos and larval tissues after inactivation of destruction complex or E3 ligase proteins. Arm accumulates to very high levels in fly embryos lacking both APC2 and APC1 (APC2g10 APC1Q8
maternal zygotic mutants; (; 
). In contrast, we previously found that clones of APC2 APC1
double null mutant cells in the optic lobes of third instar larval brains only accumulate modest levels of Arm (, arrows vs. arrowheads, 
). We first tested the hypothesis that this was a brain-specific difference, by examining Arm levels in clones of cells double mutant for null alleles of both APC2
in third instar wing imaginal discs, relative to adjacent wild-type cells. As in the larval brain, apparent elevation of Arm levels was modest (, arrows; in this experiment and most of those below mutant cells are marked with GFP) relative to Arm elevation in double mutant embryos (). As was previously observed 
, the activation of Wnt signaling in APC2 APC1
double mutant cells also triggers a dramatic cell shape change. Cells apically constrict and invaginate to form cysts, particularly in regions surrounding the wing blade (, arrowhead; 
; activating Wnt signaling downstream of APC has similar effects 
). These data suggest Arm levels are embryonic and imaginal cells are differentially sensitive to elimination of APC function.
Arm accumulates to similar levels in wing imaginal disc cells mutant for different destruction complex or SCF proteins.
Previous work demonstrated that wing imaginal disc cells mutant for Axin
accumulated elevated levels of Arm, helping demonstrate that these destruction complex or E3 ligase components are part of the machinery required to regulate Arm levels 
. The differential effect of loss of APC family proteins on relative Arm levels in embryos and imaginal discs led us to explore the hypothesis that there might be APC –dependent and APC-independent means of regulating Arm levels. To test this, we generated wing disc clones mutant for other destruction complex or E3 ligase proteins, including Axin
, and directly compared Arm levels to those seen in APC2 APC1
double mutant cells. As previously reported, immunostaining of wing discs revealed that clones mutant for Axin
(, arrows) or slimb
(, arrows) accumulate elevated levels of Arm. However, as we observed in APC2 APC1
double mutant cells, (, arrows), the elevation of Arm levels in Axin
mutant cells was not as extreme as that previously seen in embryos lacking destruction complex proteins. Loss of Slimb in clones of cells in the larval brain optic lobe also only resulted in modest elevation of Arm levels (), qualitatively similar to what we observed in cells double mutant for both APCs (; 
). We also saw elevated Arm levels in the few wing imaginal disc clones mutant for Cullin1
we obtained (, insets—note that here mutant cells are those lacking GFP). Cullin1 clones were very small and rare, probably due to effects on other SCF targets important for cell viability or cell cycle progression; similar clone size and rarity were previously seen in clones mutant for Roc1a 
. We also noted in passing that cells mutant for Axin
(, arrowhead) or slimb
(, arrowhead) also invaginated, forming cysts like those seen with APC2 APC1
double mutants 
. Thus, disruption of different components of the destruction complex or the E3 ligase in larval tissues led to similar modest elevation of Arm levels, reducing the likelihood of an APC-independent mechanism of Arm regulation.
In previous work 
and in our own data, a subset of clones mutant for slimb
, or double mutant for APC2 APC1
did appear to accumulate highly elevated levels of Arm (e.g., , arrowhead). We thus explored the reason for this apparent discrepancy. As noted above, in addition to affecting Arm levels and activating Wnt target genes, activating Wnt signaling in clones of cells in imaginal discs has drastic consequences for cell morphology—cells with activated Wnt signaling apically constrict, distorting the epithelial sheet 
. This can be clearly seen in some clonal patches, where co-staining with actin reveals groups of mutant cells with strongly constricted apical ends (, yellow arrowhead in boxed region, yellow arrowhead in inset). Both actin and Arm are strongly enriched in cell-cell adherens junctions 
, which are in the apical-most region of the lateral cell membrane. We thus hypothesized that the apparent high level of accumulation in mutant clones such as these might be due to differences in the plane of focus between wild-type cells and adjacent mutant neighbors, due to changes in the folding of the epithelial sheet. Images taken at the apical-most end of even a wild-type cell will show a higher level of Arm than a more basal section, because the apical-most section will pass through the adherens junction (, top). Consistent with the hypothesis that differences in apparent Arm accumulation could be caused by differences in cell morphology, Arm staining was relatively brighter in APC2 APC1
double mutant clones which have apically constricted (e.g., , blue arrows are non-apically constricted cells versus yellow arrowhead showing apically constricted cells, as revealed by the bright actin staining of the constricted cells). To further test this hypothesis, we examined different sections through clones mutant for slimb
. In fact, sections through the same clone revealed apparently very high levels of Arm in mutant clones in very apical sections (, top inset), while a more basal section of the same clone has more modest elevation of Arm (, bottom inset)— likely because more apical sections pass through adherens junctions of apically constricted mutant cells and more basal regions of neighboring wild-type cells (, bottom). Thus together, our data support the idea that the same machinery regulates Arm levels in embryonic and larval tissues. However, the consequences of removing this machinery on Arm levels differ between the tissues.
We next addressed the question of why we observed such a striking difference in Arm accumulation after destruction complex inactivation when comparing embryos and larval tissues. We hypothesized that in embryos the known transcriptional up-regulation of arm
after the midblastula transition 
might program the translation of more Arm protein, but that this newly synthesized protein might be rapidly turned over by the destruction complex. In this hypothesis, since cells in stage 9 embryos would have higher levels of arm
mRNA than cells in larval tissues, they would respond to inactivating the destruction complex by accumulating Arm protein more rapidly.
This hypothesis predicts that the ratio of arm
mRNA to protein would be higher in stage 9 embryos than in larval tissues. To test this hypothesis, we first compared Arm protein levels () of stage 9 embryos (when Wnt signaling is maximal), wing discs and brains of third instar larvae, and, as a control, stage 17 embryos (after most Wnt signaling in embryos is done and when we expected Arm protein levels to be low; 
). Arm protein accumulation increases in stage 9 embryos as segment identities are defined 
. We found that the amount of Arm was not significantly different in larval tissues than in stage 9 embryos, when normalized to tubulin (; quantified in ). Next, we looked at arm
mRNA levels, comparing mRNA levels from wild-type animals from all three stages by Northern blot, using the ribosomal protein gene rp49
as a loading control (). arm
mRNA levels in stage 9 embryos were roughly two times higher than in 3rd
instar larval brains and imaginal discs, when normalized to the rp49
mRNA levels were even lower in stage 17 embryos, as expected 
). To confirm this and deal with the issue that our Northern analysis combined both imaginal discs and brains, we used RNAseq data from hand-dissected imaginal discs (). Using the same normalization to rp49
, we found that arm
transcripts were 2.6 fold more abundant in stage 9 embryos than in 3rd
instar wing imaginal discs. Together, these data suggest that there is more arm
mRNA in embryos than in larval tissues, despite similar levels of protein. Thus if levels of translation are equivalent, the destruction complex would have to destroy more newly synthesized Arm in stage 9 embryos than in larval tissues. This model further predicts that if the destruction complex were inactivated, Arm levels would increase more dramatically in embryos than in imaginal tissues, which is in fact what we observed.
While Arm protein levels are similar in larval imaginal tissues and stage 9 embryos, arm mRNA is more abundant in stage 9 embryos.
RNAseq transcript numbers for arm normalized to rp49.
This hypothesis is also consistent with previous work on APC2
alleles of different strengths. Both null and hypomorphic alleles cause significant effects on cell fate in the embryo 
, though they differ in the strength of these effects. In contrast, null and hypomorphic APC2
alleles have very different effects in the imaginal discs. In clones of cells double mutant for null alleles of APC2
, Wnt target genes are activated, and cells apically constrict and invaginate, and those that do not apoptose ultimately exhibit fate changes in the adult wing, taking on wing margin fates 
. In contrast, in clones of cells double mutant for hypomorphic APC2
alleles and a null allele of APC1
, all these phenotypes are reduced or eliminated 
. These data suggest that cells in larval wing imaginal discs require less APC2 function to regulate the Wnt pathway than do cells in stage 9 embryos, consistent with the different levels of destruction complex activity predicted to be required from the higher levels of arm
mRNA in embryos than in larval tissues.
APC233 is hypomorphic and retains residual function in embryos, imaginal discs and the larval brain
This difference in phenotype between null and hypomorphic alleles in wing imaginal discs also allowed us to further characterize two interesting alleles of APC2. In 2008, Takacs et al. 
described a series of experiments suggesting that APC2 in the developing Drosophila
eye had paradoxical effects—reducing levels of APC2 suppressed
the effects of inappropriate Wnt activation caused by loss of APC1, suggesting APC2 might have positive as well as negative roles in Wnt signaling 
During their analysis, they found that different APC2
alleles they tested differed in whether they suppressed loss of APC1. Surprisingly, the allele we standardly use as a null allele, APC2g10
, did not suppress effects of APC1 loss, although deletion of the genomic region including APC2
did so 
. This was surprising, as APC2g10
has a stop codon about one-third of the way through the coding sequence (in the seventh Arm repeat; ), and we could not detect a truncated protein with an N-terminal antibody 
, although we could detect a truncated protein in an allele with a slightly later stop codon 
. In contrast, effects of loss of APC1 were suppressed by two new alleles of APC2
that were generated by mobilizing transposable elements in the 5′ flanking region or 5′ UTR 
. Both deleted part APC2's
coding sequence— APC219–33
deletes the translation start and most of the coding sequence, including all the Arm repeats, the 15 amino acid repeats, and the first two 20 amino acid repeats (), while APC233
deletes the transcription and translation starts and coding sequence extending into the 5th Arm repeat (). Based on differences in Arm accumulation in imaginal discs between cells double mutant for APC233
and a null allele of APC1
versus cells double mutant for definitive null alleles of both APC2
, they suggested that APC233
might encode an N-terminally truncated APC2 protein lacking most of the Arm repeats, but retaining the 15 and 20 amino acid repeats that bind Arm/ßcat and the SAMP repeats that bind Axin, and also retaining some function in negatively regulating Wnt signaling (it is worth noting that they could not detect this protein by immunoblotting 
, so its levels must be very low). Consistent with this, recent work revealed that remnant mobile elements like those remaining at the site of deletion in both alleles 
can contain promoters driving expression of adjacent genes 
. Since N-terminally truncated fragments of human APC can rescue ßcat degradation in human colon cancer cells 
, it is not inconceivable that APC233
or even APC219–33
might encode very low levels of an N-terminally truncated APC2 protein that nonetheless retained some function in Wnt regulation. Takacs et al. thus suggested that our allele APC2g10
produced very low levels of a C-terminally truncated APC2 that retained some residual activity in negatively regulated signaling, and also retained the postulated positive effect of APC2 on Wnt signaling, while the putative N-terminally truncated APC2 protein produced by APC233
lacked this positive effect of APC2 on Wnt signaling.
Mutations in APC2g10, APC233 and APC219-3, and structure of APC2ΔArmrepeats and APC2Armrepeatsonly.
We used imaginal discs to directly compare the effects on Wnt regulation of three different APC2
alleles, which had distinct effects in the assays of Takacs et al. 
. To do so, we assessed Arm levels and cell behavior in clones of cells double mutant for each of these different alleles APC2
and also mutant for a definitive null allele of APC1
(with a stop codon in Arm repeat 4; 
). Cells double mutant for APC219-3
() resembled cells double mutant for our standard null allele APC2g10
(). In both cases mutant cells in the wing pouch accumulated elevated levels of Arm (, arrows), and cells around the margin of the wing pouch also apically constricted and invaginated ( arrowheads). In contrast, as reported by Takacs et al., cells double mutant for APC233
did not accumulate detectably elevated levels of Arm (, arrows), nor did they invaginate from the imaginal disc epithelium ( arrowheads). In contrast, cells triple mutant for APC233
, and Axin
did accumulate Arm (, arrows), showing that there was not a suppressor of this phenotype on the chromosome. In its properties APC233
resembles other previously characterized hypomorphic APC2
. These data are thus consistent with the possibility that APC233
produces an N-terminally truncated protein retaining some function in negatively regulating Wnt signaling, while suggesting that APC219-3
is a functional null allele.
APC233 has a hypomorphic phenotype.
We saw similar differences between APC233
and the other two APC2
alleles when we examined clones of APC2 APC1
double mutant cells in the larval brain. As we previously observed 
, clones of cells in the medullar region of the brain that are double mutant for our standard null allele APC2g10
accumulate modestly elevated levels of Arm, and segregate from their neighbors (, arrow versus arrowhead); when clones are generated in medullar neurons, their axons do not extend to the medullar neuropil and instead form knots in the center of the clones. Cells double mutant for APC219-3
behaved similarly, accumulating elevated Arm levels and segregating from their neighbors (, arrow vs. arrowhead). In contrast, APC233 APC1Q8
double mutant cells exhibited a weaker phenotype—while double mutant medullar neurepithelial cells sometimes segregated from their neighbors (, arrow), Arm accumulation was less obvious. Further, while APC2g10 APC1Q8
double mutant neurons send out axons into a knot in the center of the clone (, arrow; 
), APC233 APC1Q8
double mutant neurons did not form axon knots, but instead sent axons to the medullar neuropil (, arrows) as do wild-type neurons 
. In these ways APC233
behaved similarly to other hypomorphic APC2
. Finally, we examined embryos maternally and zygotically APC233
mutant, using cuticle preparations to assess the strength of defects in Wnt signaling by the numerical scale of McCartney et al 
, where 0 is a wild-type embryo and 6 indicates the most severe defects. APC233
maternal/zygotic mutants had an average cuticle score of 3.2 (n
251). This is less severe than APC2g10
, and is in the range of other hypomorphic mutants 
. Together these data further support the hypothesis of Takacs et al. that APC233
is hypomorphic and not null for negative regulation of Wnt signaling. They also reinforce the idea there is not a one-to-one correspondence between the negative regulatory effects of a given APC2
allele on Wnt signaling and its ability to suppress loss of APC1—both APC2g10
have stronger effects on Wnt regulation than APC233
, yet only APC219-3
suppress the loss of APC1.
An APC2 protein lacking the Arm repeats retains residual activity in Wnt regulation
These data and those of Takacs et al. suggested the hypothesis that APC2 proteins lacking the Arm repeats might retain some function in Wnt regulation. However, this was based on the hypothetical N-terminally protein encoded by APC233
, which Takacs et al. could not detect by immunoblotting 
. To directly explore the function of such an N-terminally truncated APC2 protein, we generated a GFP-tagged mutant of APC2 largely matching the protein that might be produced by APC233
. We expressed it using its own ATG codon and from the endogenous APC2
promoter and verified accumulation levels were near normal, relative to wild-type GFP-APC (). This mutant, APC2ΔArmRepeats, lacks the Arm repeats but retains the 15 and 20 amino acid repeats and SAMP repeats (). In parallel, we generated a mutant encoding only the Arm repeats of APC2 (APC2Armrepeatsonly; ; ), which should largely mimic hypothetical predicted protein made by APC2g10
APC2 lacking its Arm repeats cannot downregulate ßcat levels but retains some ability to blunt Wnt signaling.
We then tested whether these two proteins could negatively regulate Wnt signaling, using transgenic flies in which the mutant proteins were expressed at normal levels under control of the endogenous promoter 
. We explored their ability to rescue Wnt signaling in the embryonic epidermis, using the cuticle as a measure. Anterior cells in wild-type embryos secrete hair-like denticles (, arrows), while posterior cells secrete naked cuticle (, arrowheads). We first tested APC2ΔArmRepeats in embryos maternally and zygotically null for APC2
. These embryos have strong Wnt pathway activation, but retain a small amount of Wnt regulation due to the low levels of APC1 remaining 
. As a result almost all cells are converted to posterior fates and only a few denticles remain (). When we expressed APC2ΔArmRepeats in the APC2g10
maternal/zygotic mutant, it significantly rescued Wnt signaling in the embryonic epidermis (, quantified in 7F), largely but not completely restoring anterior cell fates and thus denticle belts to the cuticle. In contrast, APC2Armrepeatsonly had only a modest rescuing effect (). We next tested APC2ΔArmRepeats in maternal and zygotic APC2 APC1
double mutant embryos. In these embryos all cell fates are converted to naked cuticle (; 
). This is a more stringent test of the activity of the mutant protein 
. In this background, APC2ΔArmRepeats provided only very weak rescuing activity (; quantified in 8G), contrasting with its stronger rescuing ability in the single APC2
mutant. Based on comparison with other mutants we have analyzed 
, this suggests that APC2ΔArmRepeats cannot rescue Arm degradation, but may be able to blunt Wnt signaling by sequestering Arm.
To test this directly, we assessed both mutants in cultured human SW480 colon cancer cells, which carry a truncated version of human APC, and thus accumulate very high levels of ßcat in the cytoplasm and nucleus 
. We previously found that Drosophila
APC2 effectively rescues Wnt regulation in these cells, reducing both ßcat levels and Wnt-regulated transcription 
. We thus transfected SW480 cells with GFP-tagged Drosophila
APC2, APC2ΔArmRepeats, or APC2Armrepeatsonly. We confirmed expression of stable proteins both by immunoblotting cell extracts with anti-GFP antibody (; tubulin was the loading control), and by GFP-fluorescence in transfected cells (). Wild-type fly APC2 reduces ßcat levels in these cells 
, as assessed by immunofluorescence () or by automated quantitation of hundreds of cells (). In contrast, neither APC2ΔArmRepeats nor APC2Armrepeatsonly down-regulated ßcat levels by either assay (, transfected cells are marked with GFP). However, APC2ΔArmRepeats (but not APC2Armrepeatsonly) could reduce expression of the Wnt-responsive reporter TOPFLASH (). When we compare these results to those we saw with a series of other mutants in APC2 we tested 
, the phenotypes of APC2ΔArmRepeats fit best with mutant proteins that cannot not rescue Arm/ßcat destruction, but, because they retain ßcat binding sites, can sequester ßcat in the cytoplasm and thus reduce downstream Wnt signaling. Our immunofluorescence images of APC2ΔArmRepeats are consistent with this hypothesis—expression of this mutant somewhat reduced relative ßcat levels in the nucleus (, compare arrowheads). Together, these data suggest that an APC2 protein lacking the Arm repeats can blunt Wnt signaling somewhat, and are consistent with the idea that the hypothetical truncated APC233
protein might act similarly, helping explain its hypomorphic phenotype in imaginal discs.