IQGAP1 Functions as a Modulator of Dishevelled Nuclear Localization in Wnt Signaling

Dishevelled (DVL) is a central factor in the Wnt signaling pathway, which is highly conserved among various organisms. DVL plays important roles in transcriptional activation in the nucleus, but the molecular mechanisms underlying their nuclear localization remain unclear. In the present study, we identified IQGAP1 as a regulator of DVL function. In Xenopus embryos, depletion of IQGAP1 reduced Wnt-induced nuclear accumulation of DVL, and expression of Wnt target genes during early embryogenesis. The domains in DVL and IQGAP1 that mediated their interaction are also required for their nuclear localization. Endogenous expression of Wnt target genes was reduced by depletion of IQGAP1 during early embryogenesis, but notably not by depletion of other IQGAP family genes. Moreover, expression of Wnt target genes caused by depletion of endogenous IQGAP1 could be rescued by expression of wild-type IQGAP1, but not IQGAP1 deleting DVL binding region. These results provide the first evidence that IQGAP1 functions as a modulator in the canonical Wnt signaling pathway.


Introduction
Wnt signaling plays important roles in multiple developmental events during embryogenesis [1], [2]. Canonical Wnt signaling is initiated by binding of the Wnt ligand to the cell-surface Frizzled and transmembrane LRP complex. This leads to the membrane recruitment and activation of Dishevelled (DVL), which inactivates the APC/Axin/GSK-3 complex in the cytoplasm, responsible for the degradation of ß-catenin [3], [4]. As a result, ß-catenin accumulates in the cytoplasm, translocates to the nucleus and associates with Tcf transcription factors, which activate the Wnt target genes [5], [6]. In Xenopus, Wnt signaling accompanied by ßcatenin nuclear localization at the dorsal side is an important for axis formation during early embryogenesis [7]. Ventral overexpression of Xwnt-8, ß-catenin and DVL2 induces a secondary axis and promotes expression of Wnt target genes, such as Siamois, Xnr3 and Xtwn [8]- [14].
There are three DVL isoforms, DVL1, DVL2 and DVL3 [15], [16], which are well-conserved among various organisms. Each isoform plays a similar role in the canonical Wnt pathway, but have different sensitivities to Wnt stimulation [17]. DVL contains three conserved regions known as the DIX, PDZ and DEP domains [18], [19]. Both the DIX and PDZ domains are necessary for canonical Wnt inactivation of ß-catenin degradation [20], [21]. In contrast, the DEP domain does not affect canonical signaling, but is involved in the planar cell polarity (PCP) pathway [21]- [23]. DVL plays an additional role in the Wnt signaling pathway, by localizing to the nucleus and binding a complex containing ß-catenin and Tcf, which in turn activates Wnt target genes in the nucleus [24]. The subcellular localization of DVL, either on the cell membrane or in the nucleus, is important for understanding its function in Wnt signaling. IQGAP1 contains multiple protein-interacting domains: the CH (calponin homology) domain binds to F-actin, the WW domain binds to ERK2, the IQ repeat motifs bind to calmodulin and myosin light chain, and the Ras GAP-like domain binds to Cdc42 and Rac1 [25]- [32]. IQGAP1 is also known to bind to Ecadherin and ß-catenin, and is involved in cytoskeltal reorganization and cell adhesion [33], [34]. On the other hand, IQGAP1 stimulates ß-catenin-mediated transcriptional activation 34 . The subcellular localization of IQGAP1 varies in several cultured cells, and analysis of its domains indicates that IQGAP1 may be localized in the cytoplasm, cell membrane and nucleus [35]. These subcellular localizations are presumably linked to its cellular functions. There are also three isoforms of IQGAP: IQGAP1, IQGAP2 and IQGAP3. Their subcellular localizations suggest both similarities and differences in function [36]. Each isoform has a different role, and in some cases IQGAP1 has an opposite function to IQGAP2 [36], [37]. The Xenopus xIQGAP1 and xIQGAP2 genes have been isolated [38] and shown to be involved in cadherin-mediated cell adhesion [38], [39]. We also isolated xIQGAP3 and have generated antisense morpholino oligonucleotides based on these sequences.
In the present study, we identified IQGAP1 as a novel DVLbinding protein. Binding between IQGAP1 and DVL2 mutually contributed to their nuclear localization. The depletion of endogenous IQGAP1 in Xenopus embryos suppressed secondary axis induction and expression of Wnt target genes. These results reveal a novel role for IQGAP1 in modulating the subcellular localization and transcriptional activation of components of the Wnt signaling pathway.

Ethics statement
All animal experiments were performed under the ethical guidelines of Tokyo Medical and Dental University, and animal protocols were reviewed and approved by the animal welfare committee of the Tokyo Medical and Dental University.

Primers
The sequences of the primer pairs were as follows. In Figure S1  (xIQGAP2-MO), 59-CCTCCGGCCTTACACTCCATTCCTG-39 (xIQ-GAP3-MO). The specificity of each MO was confirmed by its ability to inhibit the translation of FLAG-tagged mRNAs containing the targeted site with or without 5-mismatched sequences. MO (10 ng) and FLAG-tagged mRNAs (100 pg) were co-injected with ß-globin-FLAG mRNA (100 pg) as loading control into the animal poles of 4-cell stage embryos, and the injected animal caps were dissected at stage 10. Lysates from the animal caps were subjected to Western blotting with anti-FLAG antibody (M2, Sigma) (Fig. S1C).

Embryo handling and morpholino oligonucleotides
MOs and mRNAs were injected into four animal blastomeres at the 4-cell stage for dissection of animal caps or into two dorsal or ventral blastomeres at the 4-cell-stage for quantitative RT-PCR analysis and observation of embryo phenotypes. Animal cap explants of the injected (10 pg mRNA of each GFP fused construct) embryos were dissected at the early gastrula stage (st.10), and fixed for DAPI staining as previously reported [40]. We counted the number of cell that has fluorescence signals. When the fluorescence signal overlapped with DAPI staining was similar and brighter than un-overlapped fluorescence signal in counted cells, we defined such cells as nuclear localized cells. If nuclear fluorescence signals were not clear, we used ImageJ software (NIH) and measured the strength of brightness of fluorescence signals to define nuclear localized signals or not. The ratio of nuclear localized cells in total counted cells was computed for every explant and the average of ratio was taken with six explants in 3 independent experiments. Dorsal or ventral sectors of the injected embryos were dissected at st.10, and total RNA was extracted for RT-PCR analysis. The cytoplasmic and nuclear fractions were prepared as described with modifications [41].

RT-PCR analysis
Total RNA was prepared using TRIzol (Invitrogen). cDNA synthesis was carried out using Moloney murine leukemia virus reverse transcriptase (Invitrogen). Quantitative PCR was performed with an Applied Biosystems 7300 Real-Time PCR Cycler (ABI) using THUNDERBIRD SYBR qPCR Mix (TOYOBO). The sequences of the primer pairs were as follows.

Protein identification by LC-MS/MS analysis
FLAG-human DVL1 was expressed in HEK 293 cells, and DVL1 and associated proteins were recovered from cell extracts by immunoprecipitation with anti-FLAG antibody. The DVL1associated complexes were digested with Axhromobacter protease I, and the resulting peptides were analyzed using a nanoscale LC-MS/MS system, as described previously [44].

IQGAP associates with DVL
To identify novel proteins that may bind to DVL, we performed a high-throughput analysis of proteins that co-immunoprecipitated with mouse DVL1 in HEK 293 cells using direct nanoflow liquid chromatography-coupled tandem mass spectrometry [44]. We identified several known DVL-binding proteins, such as CK1 [45], CK2 [46], Strabismus [47], Par1 [48], Axin [49] and PP2C [50]. In addition, we identified IQGAP1 as a candidate protein that may physically interact with DVL1. An interaction between ectopically expressed IQGAP1 and DVL1 was confirmed in HEK 293 cells (Fig. 1A). Immunoprecipitation analysis using each protein antibody also confirmed the existence of an endogenous IQGAP1 and DVL1 complex in HEK 293T cells, and their interaction was increased by Wnt stimulation (Fig. 1B). In vertebrates, three isoforms of IQGAP and DVL have been identified: IQGAP1, IQGAP2 and IQGAP3, and DVL1, DVL2 and DVL3. We confirmed that each IQGAP isoform also bound to each DVL isoform ( Fig. 1C-1E). To determine the region in IQGAP1 responsible for binding to DVL1, several truncated mutants of IQGAP1 were examined in co-precipitation assays. We found that the region between the C-terminal IQ repeat domain and the N-terminal Ras GAP-like domain of IQGAP1 (termed DBR; Dishevelled Binding Region) was responsible for binding to DVL1 ( Fig. 1F and 1G) Conversely, the C-terminus of DVL1 (termed IBR; IQGAP Binding Region) is necessary for binding to IQGAP1 ( Fig. 1H and 1I). Both the IBR and DBR are wellconserved among the three DVL and IQGAP isoforms, respectively. The amino acid sequence similarities of IBR and DBR among the three DVL and IQGAP isoforms were as follows, the similarity of IBR between DVL1 and DVL2 is 47.1%; the similarity of IBR between DVL1 and DVL3 is 55.7%; the similarity of IBR between DVL2 and DVL3 is 53.7%; the similarity of DBR between IQGAP1 and IQGAP2 is 82.4%; the similarity of DBR between IQGAP1 and IQGAP3 is 76.8%; the similarity of DBR between IQGAP2 and IQGAP3 is 72.8%. We conclude that IQGAP1 can associate with DVL in mammalian cells.

IQGAP1 determines the nuclear localization of DVL
Whereas membrane-localized DVL functions to inhibit degradation of cytoplasmic ß-catenin in the canonical Wnt pathway [20], [21], nuclear-localized DVL is required, together with nuclear ß-catenin, for transactivation of the downstream targets of Wnt signaling [24]. To analyze DVL and IQGAP1 functions, we used the systems of Xenopus embryos. All isoforms of DVL and IQGAP are conserved well among vertebrates. The transcripts of their Xenopus homologues were expressed during early embryonic stages and were equivalently expressed at early gastrula stages (Fig. S1A and S1B). We subcloned their cDNA and generated their antisense morpholino oligo nucleotides (Fig. S1C). To examine how IQGAP affects DVL localization in Wnt signaling, we investigated the subcellular distribution of DVL fused to green fluorescent protein (GFP) in Xenopus embryonic cells. Fluorescence produced by the DVL2 (xDVL2)-GFP fusion appeared as a punctate pattern in the cytoplasm ( Fig. 2A, left panel). DVL has been reported to be recruited to the plasma membrane by the Frizzled receptors in the Wnt pathway [21]. We confirmed that xDVL2-GFP accumulated in the plasma membrane when coexpressed with Xenopus frizzled 7 (Xfz7) ( Fig. 2A, center panel). Depletion of xIQGAP1 by antisense morpholino oligonucleotides (xIQGAP1-MO) did not affect the membrane localization of xDVL2-GFP ( Fig. 2A, right panel). These results suggest that xIQGAP1 is not involved in the plasma membrane localization of xDVL2.
Stimulation by Wnt ligands is known to increase the nuclear localization of DVL [40]. When xDVL2-GFP was co-expressed with Xwnt-8 in animal cap cells, nuclear GFP fluorescence was increased (Fig. 2B, third and fourth panels, 2D, lanes 1, 3), whereas GFP was mainly localized in the cytoplasm with or without Xwnt-8 (Fig. 2B, first and second panels, 2C). However, injection of xIQGAP1-MO decreased nuclear fluorescence generated by co-expression of xDVL2-GFP and Xwnt-8 (Fig. 2B, fifth and sixth panels, 2D, lanes 3, 4). We also confirmed that the amounts of xDVL2-MYC protein in the nuclear fractions of animal cap cells were reduced by depletion of xIQGAP1 (Fig. 2E). Expression of xIQGAP1-GFP resulted in fluorescence localized mainly to the cytoplasm, but nuclear fluorescence was increased by co-expression of Xwnt-8 (Fig. 3A, first and second panels, 3B, lanes 1,3). Depletion of xDVL2 led to a decrease in nuclear fluorescence generated by the co-expression of xIQGAP1-GFP and Xwnt-8 (Fig. 3B, lanes 3, 4). Moreover, we found that depletion of all three xDVLs (xDVL1, xDVL2 and xDVL3) reduced severely the nuclear localization of xIQGAP1-GFP in Wnt-stimulated cells (Fig. 3A, third and fourth panels, 3C, lanes 3, 4 and 3D). These results suggest that xIQGAP1 and xDVL2 play a crucial role in each other nuclear accumulation, depending on Wnt signaling.
We next examined whether a physical interaction between xIQGAP1 and xDVL2 is required for their nuclear localization. We generated a fusion of GFP to xDVL2-DIBR, a truncated version of xDVL2 lacking the IBR domain, and observed a punctate fluorescence pattern in the cytoplasm, similar to that seen with xDVL2-GFP (Fig. S2A). The proportion of fluorescence found in the nucleus was also similar to xDVL2-GFP (Fig. S2B). However, co-expression of Xwnt-8 did not alter the proportion of GFP fluorescence found in the nucleus, (Fig. S2B), suggesting that the ability of xIQGAP1 to promote nuclear localization of xDVL2 requires the IBR domain in xDVL2. Consistent with this, a fusion of just the IBR domain to GFP (xDVL2-IBR-GFP) was localized predominantly in the nucleus ( Fig. S2A and S2C). We next fused GFP to xIQGAP1-DDBR, which deletes the DBR domain of xIQGAP1. This xIQGAP1-DDBR-GFP fusion was also localized mainly in the cytoplasm (Fig. S2A), and the proportion of fluorescence found in the nucleus was less than that observed with full-length xIQGAP1-GFP (Fig. S2D). Co-expression of Xwnt-8 also did not affect the nuclear localization of xIQGAP1-DDBR-GFP (Fig. S2D). A fusion of just the DBR domain to GFP (xIQGAP1-DBR-GFP) was localized mainly in the nucleus ( Fig.  S2A and S2E). We further investigated the effects of overexpression of xIQGAP1 or xDVL2 on the nuclear localization of xDVL2-GFP or xIQGAP1-GFP, respectively. Nuclear localization of xDVL2-GFP was increased by the expression of xIQGAP1 and xIQGAP1-DBR (Fig. S3A). Meanwhile, expression of xIQGAP1-DDBR suppressed the nuclear localization of xDVL2-GFP induced by co-expression of Xwnt-8 (Fig. S3A). In contrast, the nuclear localization of xIQGAP1-GFP was not affected by the expression  of xDVL2 or xDVL2-IBR (Fig. S3B). However, xDVL2-DIBR suppressed the nuclear localization of xDVL2-GFP induced by coexpression of Xwnt-8 (Fig. S3B). Moreover, immunoprecipitation experiments showed that Wnt stimulation increased the interaction between xDVL2 and xIQGAP1 in HEK 293T cells (Fig. 3E). Taken together, these results suggest that a physical interaction between xDVL2 and xIQGAP1 is required for their nuclear localization induced by the canonical Wnt signaling pathway.

xIQGAP1 is necessary for the canonical Wnt pathway
To determine whether xIQGAP1 is also involved in the canonical Wnt pathway during early development, we investigated the effects of xIQGAP1 on the transactivation of Wnt target genes and the secondary axis induction. Dorsal injection of an antisense morpholino oligonucleotide against xIQGAP1 (xIQGAP1-MO) reduced endogenous transcripts of the Wnt signal target genes Siamois, Xnr3 and Xtwn (Fig. 4A). When xDVL2, Xwnt-8 or ß-catenin mRNA was injected into the ventral sides of four-cell embryos, a secondary axis was formed and Wnt signal target genes were induced. The induction of the partial secondary axis and Wnt target genes induced by Xwnt-8 was also suppressed by the depletion of xIQGAP1 (Fig. 4B-4D). Dorsal overexpression of xIQGAP1 mRNA increased expression of Wnt target genes (Fig. 4E). On the other hand, the overexpression or depletion of xIQGAP2 showed opposite effects for Wnt target gene expression (Fig. 4A, 4D and 4E). xIQGAP3 showed ambiguous effects on Wnt target gene expression, especially expression of Siamois (Fig. 4A, 4D and 4E). However, depletions of either xIQGAP2 or xIQGAP3 did not alter the partial secondary axis induction by Xwnt-8 ( Fig. 4B  and 4C). These results suggest that xIQGAP1 is necessary for Wnt-related early embryogenesis in a subtype-specific manner.
We also found that the suppressions of induction of Wnt target genes and partial secondary axis by xIQGAP1-MO were rescued by expression of wild-type xIQGAP1, but not by either xIQGAP1-DDBR or -DBR (Fig. 4F, 4G, S4A-S4D). However, depletion of xIQGAP1 did not affect the secondary axis formation induced by Simaois, which is one of the Wnt signal target genes (Fig. 4H). Moreover, we also observed the reduction of endogenous IQGAP1 by the siRNA (siIQGAP1) suppressed the expression of Wnt target genes induced by Wnt3A stimulation in cultured cells (Fig. S5A). These results suggest that xIQGAP1 functions as an intermediate molecule in the canonical Wnt signaling pathway in early development promoting the nuclear localization of xDVL2.

The IQGAP binding region of xDVL2 is important for canonical Wnt signaling
To further confirm whether the binding between IQGAP1 and DVL is critical in the canonical Wnt pathway during early development, we investigated the effects of xDVL mutants on the transactivation of Wnt target genes and secondary axis induction. Similar to our previous observation that injection of xDVL2-MO did not affect severely nuclear localization of xIQGAP1 (Fig. 3B), we also observed no reduction in Wnt target gene expression induced at the ventral side by Xwnt-8 when xDVL2-MO was coinjected (Fig. 5A). However, depletion of all three xDVLs reduced nuclear localization of xIQGAP1, expression of the Wnt target genes and suppressed formation of the secondary axis induced by Xwnt-8 or ß-catenin (Fig. 3C, 3D, 5B, and 5C). These results suggest that three xDVL genes act redundantly in the canonical Wnt signal pathway. Suppression of secondary axis formation and Wnt target gene expression caused by depletion of all three xDVLs could be rescued by co-expression of wild-type xDVL2, but only weakly by xDVL2-DIBR and barely by xDVL2-IBR (Fig. 5B, and  5C). Moreover, co-expression of xIQGAP1-DBR reduced the expression of Wnt target genes induced by Xwnt-8, xDVL2 or ßcatenin in Xenopus embryos (Fig. 5D, S6A and S6B), and the expression of xIQGAP1-DBR in cultured cells reduced the expression of Wnt target gene induced by Wnt3A (Fig. S5B). These results support the idea that binding between xDVL2 and xIQGAP1 plays important roles for canonical Wnt signaling.

Discussion
In the present studies, we show that IQGAP1 is necessary for the nuclear localization of DVL in the canonical Wnt signaling pathway. Previous studies have shown that nuclear localization of DVL is necessary for the sequential activation of Wnt target genes [24], [40]. It has also been shown that the nuclear localization signal (NLS) located between the PDZ and DEP domain of xDVL2, and the nuclear export signal (NES) located at the Cterminus are important for the nuclear localization and transcriptional activation of Wnt target genes [40]. Interestingly, we observed that xDVL2-IBR, a truncated protein consisting of just the IQGAP-binding region from the C-terminus of xDVL2, localized predominantly in the nucleus, even though this region contains an NES and not an NLS. Moreover, the nuclear localization of xDVL2-DIBR-GFP, which contains an NLS, but not an NES, did not increase with Wnt-8 stimulation any longer. Although our findings suggest a new molecular mechanism mediating xIQGAP1-dependent nuclear localization of xDVL2, we could not positively state that our findings is independent of the NLS or NES motifs within DVL2. Further studies need to clarify the inconsistencies using same mutants.
In the canonical Wnt pathway, DVL is necessary for both the inactivation of ß-catenin degradation in cytoplasm 21 and the activation of Wnt target genes by forming a complex containing ßcatenin and Tcf in nuclei [24]. In the present study, we have shown that IQGAP1 interacted with DVL and that the depletion of IQGAP1 reduced the nuclear localization of DVL, while IQGAP1 did not affect on the membrane localization of DVL required for ß-catenin stability in cytoplasm. These results suggest that IQGAP1 plays a role in the nuclear translocation of DVL in the canonical Wnt pathway.
We showed that the nuclear localization of xIQGAP1 and xDVL2 were increased by Wnt stimulation. In contrast, the nuclear localization of xDVL2-DIBR-GFP and xIQGAP1-DDBR-GFP did not increase with Wnt stimulation, while over-expression of xDVL2-DIBR or xIQGAP1-DDBR interfered with the nuclear localization of xIQGAP1-GFP or xDVL2-GFP induced by Wnt stimulation, respectively. Reduced expression of Wnt target genes due to depletion of endogenous xIQGAP1 or xDVLs was barely or weakly rescued by expression of xIQGAP1-DDBR or xDVL2-DIBR, respectively. Conversely, xDVL2-IBR-GFP and xIQGAP1-DBR- . (E) Cytoplasmic and nuclear distribution of xDVL2, xIQGAP1 and ß-catenin in animal cap cells. MYC-tagged xDVL2 mRNA (100 pg) was injected into the animal poles of 4-cell stage embryos, and the injected animal caps were dissected at stage 10. Lysates from the animal caps were fractionated and subjected to Western blotting with indicated antibodies. Each relative intensity was measured by ImageJ, and its relative ratio was calculated against Input with beta-tubulin for cytoplasm or with Histone H3 for nuclear. Error bars represent standard deviation of the mean in three experiments. GFP were mainly localized in the nuclei regardless of Wnt stimulation. Moreover, xIQGAP1-DBR reduced expression of the Wnt target genes induced by xDVL2, Xwnt-8 and ß-catenin. Taken together, these results suggest that the domains mediating binding between xIQGAP1 and xDVL2 play important roles in both their nuclear localization and their Wnt-stimulated activities.
In vertebrates, DVL1, DVL2 and DVL3 have redundant function in part [51], [52]. Depletion of all three xDVLs; xDVL1, xDVL2 and xDVL3, did reduce severely the nuclear localization of xIQGAP1 rather than only xDVL2 depletion. Moreover, induction of Wnt target genes and formation of the secondary axis by Xwnt-8 or ß-catenin was suppressed by the depletion of all three xDVLs, but not by the depletion of xDVL2 alone. However, xDVL2 expression could rescue suppression of Wnt target genes by the depletion of all three xDVLs. These results suggest that xDVL1, xDVL2 and xDVL3 also function redundantly in Wnt signaling involving xIQGAP1. On the other hand, we showed that all IQGAP isoforms bound to each DVL isoform, nevertheless only IQGAP1 was necessary for Wnt signaling. Previous report also showed the functional differences, their subcellular localization and the interaction with binding proteins among IQGAP isoforms in many different cellular processes [36]. Therefore, unidentified binding molecules might cause the functional  The ratio of injected embryos that exhibited a partial differences among IQGAP isoforms. Further molecular analyses will be needed to clarify the different roles of IQGAP isoforms. Figure S1 Expression of Xenopus DVL and IQGAP1 isoforms and confirmation of the morpholino specificity. Reverse transcription-polymerase chain reaction analysis was performed using total RNA extracted from Xenopus embryos at different stages of development and from different regions.